Align compiler and AST behavior with CPython (#8138)

* Align compiler and AST behavior with CPython

Update the compiler pipeline, AST module support, and related VM compile/eval paths toward CPython 3.14 behavior.

Use upstream Ruff parser crates directly and keep RustPython-specific syntax preflight handling in the compile path.

Refresh opcode metadata, snapshots, and targeted tests for the aligned bytecode and AST behavior.

* fix

* Fix PR 8138 CI failures

* Fix match mapping wildcard scan complexity

* fixes

* Use Ruff runtime AST fields

* Minimize residual patch diff

* refactor error handling

* Remove _ast conversion context wrappers

* Clean up _ast conversion glue and compiler warning escalation

Rename the Node conversion parameter `ctx` to `vm` throughout the _ast
module.

Remove the unused compile_program, compile_program_single,
compile_block_expression, and compile_expression forwarders; compile_top
stays as the entry point.

Stash an escalated compiler SyntaxWarning so a non-SyntaxWarning category
propagates unchanged instead of being rewritten to SyntaxError, matching
PyErr_ExceptionMatches(SyntaxWarning) in compiler_warn. Drop the now-unused
CompileWarningError::into_codegen_error.

Replace the PositionalArguments two-variant enum with a struct holding a
shared range and a PositionalArgumentsKind enum.

Assisted-by: Claude

* Represent compiler flags as a CompilerFlags bitflags

Replace the loose PY_CF_*/CO_* i32 constants in compile_mode with a
CompilerFlags bitflags covering the PyCF_* flags and __future__
CO_FUTURE_* bits. The Python-visible ast.PyCF_* attribute values are
derived from it via bits(); parser input modes stay plain consts.

builtins::compile and the compile_string flag handling test flags
through CompilerFlags::contains instead of raw bit math.

Assisted-by: Claude

* Fix PR 8138 test regressions

* Fix PR 8138 CI failures

* Avoid shell interpolation in Linux deps action

* Fix review findings in compiler/AST changes

- frame.rs: compute the MATCH_CLASS class name lazily so a successful
  match no longer takes the type name lock or allocates; the
  __match_args__-not-a-tuple branch reuses it instead of recomputing.
- _ast/constant.rs: add the missing leading space to the with_recursion
  label ("maximum recursion depth exceeded during compilation").
- sys.set_int_max_str_digits: restore the "maxdigits must be 0 or larger
  than {n}" error text.
- codegen: set CO_MAXBLOCKS to 20 in both ir.rs and compile.rs.
- _ast/exception.rs: extract the duplicated excepthandler node guard
  into ensure_excepthandler_node.
- codegen: correct the validate_keywords doc comment.
- Drop redundant "CPython" qualifiers from comments added in this branch.

Assisted-by: Claude

.cspell.json

@@ -66,8 +66,11 @@
     "deoptimize",
     "emscripten",
     "excs",
+    "flufl",
     "fnfe",
+    "fsdefault",
     "ifexp",
+    "implicits",
     "interps",
     "jitted",
     "jitting",

.github/actions/install-linux-deps/action.yml

@@ -39,11 +39,36 @@ runs:
     - name: Install Linux dependencies
       shell: bash
       if: ${{ runner.os == 'Linux' }}
-      run: >
-        sudo apt-get update
+      env:
+        GCC_MULTILIB: ${{ inputs.gcc-multilib }}
+        MUSL_TOOLS: ${{ inputs.musl-tools }}
+        CLANG: ${{ inputs.clang }}
+        GCC_AARCH64_LINUX_GNU: ${{ inputs.gcc-aarch64-linux-gnu }}
+      run: |
+        if ! sudo apt-get update; then
+          echo "::warning::apt-get update failed; disabling nonessential Microsoft apt sources and retrying"
+          for source in /etc/apt/sources.list.d/*microsoft* /etc/apt/sources.list.d/*azure-cli*; do
+            if [ -e "$source" ]; then
+              sudo mv "$source" "$source.disabled"
+            fi
+          done
+          sudo apt-get update
+        fi
 
-        sudo apt-get install --no-install-recommends
-        ${{ fromJSON(inputs.gcc-multilib) && 'gcc-multilib' || '' }}
-        ${{ fromJSON(inputs.musl-tools) && 'musl-tools' || '' }}
-        ${{ fromJSON(inputs.clang) && 'clang' || '' }}
-        ${{ fromJSON(inputs.gcc-aarch64-linux-gnu) && 'gcc-aarch64-linux-gnu linux-libc-dev-arm64-cross libc6-dev-arm64-cross' || '' }}
+        packages=()
+        if [[ "$GCC_MULTILIB" == "true" ]]; then
+          packages+=(gcc-multilib)
+        fi
+        if [[ "$MUSL_TOOLS" == "true" ]]; then
+          packages+=(musl-tools)
+        fi
+        if [[ "$CLANG" == "true" ]]; then
+          packages+=(clang)
+        fi
+        if [[ "$GCC_AARCH64_LINUX_GNU" == "true" ]]; then
+          packages+=(gcc-aarch64-linux-gnu linux-libc-dev-arm64-cross libc6-dev-arm64-cross)
+        fi
+
+        if ((${#packages[@]})); then
+          sudo apt-get install --no-install-recommends "${packages[@]}"
+        fi

Cargo.lock

@@ -3631,9 +3631,8 @@ dependencies = [
 
 [[package]]
 name = "rustpython-ruff_python_ast"
-version = "0.15.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f021ff72cabf5e2cd6d8ec8813d376a8445a228dc610ab56c27bd9054cda70d4"
+version = "0.15.9"
+source = "git+https://github.com/RustPython/ruff.git?tag=0.15.19-rustpython#3c1ab2dd9987f190f3df94d28452ed5e65e106af"
 dependencies = [
  "aho-corasick",
  "bitflags 2.13.0",
@@ -3650,9 +3649,8 @@ dependencies = [
 
 [[package]]
 name = "rustpython-ruff_python_parser"
-version = "0.15.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "01e6ee78bd9671fb5766664b2695fe1f2a92a961f4d9101646c570d8acdb1e0b"
+version = "0.15.9"
+source = "git+https://github.com/RustPython/ruff.git?tag=0.15.19-rustpython#3c1ab2dd9987f190f3df94d28452ed5e65e106af"
 dependencies = [
  "bitflags 2.13.0",
  "bstr",
@@ -3671,9 +3669,8 @@ dependencies = [
 
 [[package]]
 name = "rustpython-ruff_python_trivia"
-version = "0.15.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "79e7cfd1056f3a02ff0d2d0e4474286ca963260782f878b7b81c1dd87432e682"
+version = "0.15.9"
+source = "git+https://github.com/RustPython/ruff.git?tag=0.15.19-rustpython#3c1ab2dd9987f190f3df94d28452ed5e65e106af"
 dependencies = [
  "itertools 0.14.0",
  "rustpython-ruff_source_file",
@@ -3683,9 +3680,8 @@ dependencies = [
 
 [[package]]
 name = "rustpython-ruff_source_file"
-version = "0.15.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "948107aad62ddb12a11fc7bf68a49e52a0b0a3737d415a2505e54f5a9edac737"
+version = "0.15.9"
+source = "git+https://github.com/RustPython/ruff.git?tag=0.15.19-rustpython#3c1ab2dd9987f190f3df94d28452ed5e65e106af"
 dependencies = [
  "memchr",
  "rustpython-ruff_text_size",
@@ -3693,9 +3689,8 @@ dependencies = [
 
 [[package]]
 name = "rustpython-ruff_text_size"
-version = "0.15.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8291ee0f5a779e54ccd4e0151a0c426f8b49a123f99b5b6545db17ccdd4277aa"
+version = "0.15.9"
+source = "git+https://github.com/RustPython/ruff.git?tag=0.15.19-rustpython#3c1ab2dd9987f190f3df94d28452ed5e65e106af"
 dependencies = [
  "get-size2",
 ]
@@ -3855,6 +3850,7 @@ dependencies = [
  "static_assertions",
  "strum",
  "strum_macros",
+ "thin-vec",
  "thiserror",
  "timsort",
  "wasm-bindgen",
@@ -3881,6 +3877,7 @@ dependencies = [
  "js-sys",
  "rustpython-common",
  "rustpython-pylib",
+ "rustpython-ruff_text_size",
  "rustpython-stdlib",
  "rustpython-vm",
  "serde-wasm-bindgen",
@@ -4363,6 +4360,12 @@ version = "0.16.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c13547615a44dc9c452a8a534638acdf07120d4b6847c8178705da06306a3057"
 
+[[package]]
+name = "thin-vec"
+version = "0.2.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b0f7e269b48f0a7dd0146680fa24b50cc67fc0373f086a5b2f99bd084639b482"
+
 [[package]]
 name = "thiserror"
 version = "2.0.18"

Cargo.toml

@@ -183,19 +183,11 @@ rustpython-sre_engine = { path = "crates/sre_engine", version = "0.5.0" }
 rustpython-wtf8 = { path = "crates/wtf8", version = "0.5.0" }
 rustpython-doc = { path = "crates/doc", version = "0.5.0" }
 
-# Use RustPython-packaged Ruff crates from the published fork while keeping
-# existing crate names in the codebase.
-ruff_python_parser = { package = "rustpython-ruff_python_parser", version = "0.15.8" }
-ruff_python_ast = { package = "rustpython-ruff_python_ast", version = "0.15.8" }
-ruff_text_size = { package = "rustpython-ruff_text_size", version = "0.15.8" }
-ruff_source_file = { package = "rustpython-ruff_source_file", version = "0.15.8" }
-# To update ruff crates, comment out the above lines and uncomment the following lines to pull directly from the Ruff repository at the specified commit hash.
-# Ruff tag 0.15.8 is based on commit c2a8815842f9dc5d24ec19385eae0f1a7188b0d9
-# at the time of this capture. We use the commit hash to ensure reproducible builds.
-# ruff_python_parser = { git = "https://github.com/astral-sh/ruff.git", rev = "c2a8815842f9dc5d24ec19385eae0f1a7188b0d9" }
-# ruff_python_ast = { git = "https://github.com/astral-sh/ruff.git", rev = "c2a8815842f9dc5d24ec19385eae0f1a7188b0d9" }
-# ruff_text_size = { git = "https://github.com/astral-sh/ruff.git", rev = "c2a8815842f9dc5d24ec19385eae0f1a7188b0d9" }
-# ruff_source_file = { git = "https://github.com/astral-sh/ruff.git", rev = "c2a8815842f9dc5d24ec19385eae0f1a7188b0d9" }
+# Use the RustPython Ruff fork for RustPython public `_ast` metadata.
+ruff_python_parser = { package = "rustpython-ruff_python_parser", git = "https://github.com/RustPython/ruff.git", tag = "0.15.19-rustpython" }
+ruff_python_ast = { package = "rustpython-ruff_python_ast", git = "https://github.com/RustPython/ruff.git", tag = "0.15.19-rustpython" }
+ruff_text_size = { package = "rustpython-ruff_text_size", git = "https://github.com/RustPython/ruff.git", tag = "0.15.19-rustpython" }
+ruff_source_file = { package = "rustpython-ruff_source_file", git = "https://github.com/RustPython/ruff.git", tag = "0.15.19-rustpython" }
 
 der = { version = "0.8", features = ["alloc", "oid", "pem", "zeroize"] }
 phf = { version = "0.13.1", default-features = false, features = ["macros"]}
@@ -311,6 +303,7 @@ tcl-sys = { git = "https://github.com/arihant2math/tkinter.git", tag = "v0.2.0"
 textwrap = { version = "0.16.2", default-features = false }
 termios = "0.3.3"
 thiserror = "2.0"
+thin-vec = "0.2.14"
 timsort = "0.1.2"
 tk-sys = { git = "https://github.com/arihant2math/tkinter.git", tag = "v0.2.0" }
 icu_casemap = "2"

Lib/test/test_ast/test_ast.py

@@ -150,7 +150,6 @@ class AST_Tests(unittest.TestCase):
             self.assertRaises(TypeError, ast.parse, ast.Constant(42),
                               optimize=optval)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ValueError: compile() unrecognized flags
     def test_optimization_levels__debug__(self):
         cases = [(-1, '__debug__'), (0, '__debug__'), (1, False), (2, False)]
         for (optval, expected) in cases:
@@ -586,7 +585,6 @@ class AST_Tests(unittest.TestCase):
             compile(m, "<test>", "exec")
         self.assertIn("but got expr()", str(cm.exception))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ValueError: expected str for name
     def test_invalid_identifier(self):
         m = ast.Module([ast.Expr(ast.Name(42, ast.Load()))], [])
         ast.fix_missing_locations(m)
@@ -1365,7 +1363,6 @@ class CopyTests(unittest.TestCase):
         self.assertIs(repl.ctx, context)
         self.assertRaises(AttributeError, getattr, repl, 'extra')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "Name\.__replace__\ missing\ 1\ keyword\ argument:\ 'id'\." does not match "replace() does not support Name objects"
     def test_replace_reject_missing_field(self):
         # case: warn if deleted field is not replaced
         node = ast.parse('x').body[0].value
@@ -1700,7 +1697,6 @@ Module(
             full="Module(body=[Import(names=[alias(name='_ast', asname='ast')]), ImportFrom(module='module', names=[alias(name='sub')], level=0)], type_ignores=[])",
         )
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ?                                              ^^^^^^^^^                                                                                  ^^^^^^^^^
     def test_copy_location(self):
         src = ast.parse('1 + 1', mode='eval')
         src.body.right = ast.copy_location(ast.Constant(2), src.body.right)
@@ -1737,7 +1733,6 @@ Module(
             "end_col_offset=0), lineno=1, col_offset=0, end_lineno=1, end_col_offset=0)])"
         )
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ?                                              ^^^^^^^^^                                                                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
     def test_increment_lineno(self):
         src = ast.parse('1 + 1', mode='eval')
         self.assertEqual(ast.increment_lineno(src, n=3), src)
@@ -1959,7 +1954,6 @@ Module(
                 (\
             \ ''')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; TypeError: required field "lineno" missing from alias
     def test_bad_integer(self):
         # issue13436: Bad error message with invalid numeric values
         body = [ast.ImportFrom(module='time',
@@ -3259,7 +3253,6 @@ class ASTConstructorTests(unittest.TestCase):
                                    r"MyAttrs.__init__ got an unexpected keyword argument 'c'."):
             obj = MyAttrs(c=3)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; DeprecationWarning not triggered
     def test_fields_and_types_no_default(self):
         class FieldsAndTypesNoDefault(ast.AST):
             _fields = ('a',)
@@ -3273,7 +3266,6 @@ class ASTConstructorTests(unittest.TestCase):
         obj = FieldsAndTypesNoDefault(a=1)
         self.assertEqual(obj.a, 1)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; DeprecationWarning not triggered
     def test_incomplete_field_types(self):
         class MoreFieldsThanTypes(ast.AST):
             _fields = ('a', 'b')
@@ -3293,7 +3285,6 @@ class ASTConstructorTests(unittest.TestCase):
         self.assertEqual(obj.a, 1)
         self.assertEqual(obj.b, 2)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; TypeError: Expected type 'str' but 'bytes' found.
     def test_malformed_fields_with_bytes(self):
         class BadFields(ast.AST):
             _fields = (b'\xff'*64,)
@@ -3713,7 +3704,6 @@ class ASTOptimizationTests(unittest.TestCase):
             f"{ast.dump(optimized_tree)}",
         )
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ValueError: compile() unrecognized flags
     def test_folding_format(self):
         code = "'%s' % (a,)"
 

Lib/test/test_audit.py

@@ -77,7 +77,6 @@ class AuditTest(unittest.TestCase):
     def test_open(self):
         self.do_test("test_open", os_helper.TESTFN)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_cantrace(self):
         self.do_test("test_cantrace")
 

Lib/test/test_builtin.py

@@ -486,7 +486,6 @@ class BuiltinTest(ComplexesAreIdenticalMixin, unittest.TestCase):
                                 msg=f"source={source} mode={mode}")
 
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_compile_top_level_await(self):
         """Test whether code with top level await can be compiled.
 
@@ -627,7 +626,6 @@ class BuiltinTest(ComplexesAreIdenticalMixin, unittest.TestCase):
         exec(co, glob)
         self.assertEqual(type(glob['ticker']()), AsyncGeneratorType)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: <_ast.Name object at 0xb40000731e3d1360> is not an instance of <class '_ast.Constant'>
     def test_compile_ast(self):
         args = ("a*__debug__", "f.py", "exec")
         raw = compile(*args, flags = ast.PyCF_ONLY_AST).body[0]
@@ -1020,7 +1018,6 @@ class BuiltinTest(ComplexesAreIdenticalMixin, unittest.TestCase):
         finally:
             sys.stdout = savestdout
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; TypeError: Unexpected keyword argument closure
     def test_exec_closure(self):
         def function_without_closures():
             return 3 * 5

Lib/test/test_cmd_line_script.py

@@ -645,7 +645,6 @@ class CmdLineTest(unittest.TestCase):
             self.assertNotIn("\f", text)
             self.assertIn("\n    1 + 1 = 2\n    ^^^^^\n", text)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_syntaxerror_multi_line_fstring(self):
         script = 'foo = f"""{}\nfoo"""\n'
         with os_helper.temp_dir() as script_dir:

Lib/test/test_codeop.py

@@ -279,7 +279,6 @@ class CodeopTests(unittest.TestCase):
         self.assertNotEqual(compile_command("a = 1\n", "abc").co_filename,
                             compile("a = 1\n", "def", 'single').co_filename)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 0 != 2
     def test_warning(self):
         # Test that the warning is only returned once.
         with warnings_helper.check_warnings(

Lib/test/test_compile.py

@@ -209,7 +209,6 @@ class TestSpecifics(unittest.TestCase):
         self.assertEqual(eval("0o777"), 511)
         self.assertEqual(eval("-0o0000010"), -8)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised
     def test_int_literals_too_long(self):
         n = 3000
         source = f"a = 1\nb = 2\nc = {'3'*n}\nd = 4"
@@ -283,7 +282,6 @@ class TestSpecifics(unittest.TestCase):
             self.assertRaises(SyntaxError, compile, stmt, 'tmp', 'single')
             self.assertRaises(SyntaxError, compile, stmt, 'tmp', 'exec')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised by compile
     def test_import(self):
         succeed = [
             'import sys',
@@ -348,7 +346,6 @@ class TestSpecifics(unittest.TestCase):
         l = lambda: "this is the only const"
         self.assertEqual(l.__code__.co_consts, ("this is the only const",))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised by compile
     def test_encoding(self):
         code = b'# -*- coding: badencoding -*-\npass\n'
         self.assertRaises(SyntaxError, compile, code, 'tmp', 'exec')
@@ -465,7 +462,6 @@ class TestSpecifics(unittest.TestCase):
         # See gh-113054
         compile('if (5 if 5 else T): 0', '<eval>', 'exec')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_condition_expression_with_redundant_comparisons_compiles(self):
         # See gh-113054, gh-114083
         exprs = [
@@ -580,7 +576,6 @@ class TestSpecifics(unittest.TestCase):
 
         compile(ast.fix_missing_locations(tree), "<file>", "exec")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: <code object <module> at 0xb77555080 file "<assign>1", line 1> != <code object <module> at 0xb77554f00 file "<assign>3", line 1>
     def test_compile_ast(self):
         fname = __file__
         if fname.lower().endswith('pyc'):
@@ -696,7 +691,6 @@ class TestSpecifics(unittest.TestCase):
         self.compile_single("class T:\n   pass")
         self.compile_single("c = '''\na=1\nb=2\nc=3\n'''")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised by compile_single
     def test_bad_single_statement(self):
         self.assertInvalidSingle('1\n2')
         self.assertInvalidSingle('def f(): pass')
@@ -708,7 +702,6 @@ class TestSpecifics(unittest.TestCase):
         self.assertInvalidSingle('x = 5 # comment\nx = 6\n')
         self.assertInvalidSingle("c = '''\nd=1\n'''\na = 1\n\nb = 2\n")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: b'source code cannot contain null bytes' not found in b'OSError: stream did not contain valid UTF-8\n'
     def test_particularly_evil_undecodable(self):
         # Issue 24022
         src = b'0000\x00\n00000000000\n\x00\n\x9e\n'
@@ -719,7 +712,6 @@ class TestSpecifics(unittest.TestCase):
             res = script_helper.run_python_until_end(fn)[0]
         self.assertIn(b"source code cannot contain null bytes", res.err)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: b'source code cannot contain null bytes' not found in b'OSError: stream did not contain valid UTF-8\n'
     def test_yet_more_evil_still_undecodable(self):
         # Issue #25388
         src = b"#\x00\n#\xfd\n"
@@ -756,7 +748,6 @@ class TestSpecifics(unittest.TestCase):
         # check_limit("a", " if a else a")
         # check_limit("if a: pass", "\nelif a: pass", mode="exec")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "cannot contain null" does not match "invalid syntax (<dummy>, line 1)"
     def test_null_terminated(self):
         # The source code is null-terminated internally, but bytes-like
         # objects are accepted, which could be not terminated.
@@ -1673,7 +1664,6 @@ class TestSpecifics(unittest.TestCase):
 
         self.assertRaises(NameError, ns['foo'])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; + [3, 5, 3, 5]
     def test_compile_warnings(self):
         # Each invocation of compile() emits compiler warnings, even if they
         # have the same message and line number.
@@ -1691,7 +1681,6 @@ class TestSpecifics(unittest.TestCase):
 
         self.assertEqual([wm.lineno for wm in caught], [3, 5] * 2)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; + [5, 9]
     def test_compile_warning_in_finally(self):
         # Ensure that warnings inside finally blocks are
         # only emitted once despite the block being
@@ -1742,7 +1731,6 @@ class TestSpecifics(unittest.TestCase):
             self.assertEqual(wm.category, SyntaxWarning)
             self.assertIn("\"is\" with 'int' literal", str(wm.message))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     @support.subTests('src', [
         textwrap.dedent("""
             def f():

Lib/test/test_exceptions.py

@@ -231,7 +231,6 @@ class ExceptionTests(unittest.TestCase):
                     line = line.removeprefix('\ufeff')
                 self.assertIn(line, cm.exception.text)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_error_offset_continuation_characters(self):
         check = self.check
         check('"\\\n"(1 for c in I,\\\n\\', 2, 2)

Lib/test/test_fstring.py

@@ -701,7 +701,6 @@ x = (
                             ["f'{ {{}} }'", # dict in a set
                              ])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_compile_time_concat(self):
         x = 'def'
         self.assertEqual('abc' f'## {x}ghi', 'abc## defghi')
@@ -816,7 +815,6 @@ x = (
         s = "f'{1}' 'x' 'y'" * 1024
         self.assertEqual(eval(s), '1xy' * 1024)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_format_specifier_expressions(self):
         width = 10
         precision = 4
@@ -947,7 +945,6 @@ x = (
                             ["f'{3)+(4}'",
                              ])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_newlines_before_syntax_error(self):
         self.assertAllRaise(SyntaxError,
                             "f-string: expecting a valid expression after '{'",
@@ -1031,7 +1028,6 @@ x = (
                              r"'\N{GREEK CAPITAL LETTER DELTA'",
                              ])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_backslashes_in_expression_part(self):
         self.assertEqual(f"{(
                         1 +
@@ -1732,7 +1728,6 @@ x = (
         with self.assertRaisesRegex(ValueError, error_msg):
             f'{1:_,}'
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "f-string: expecting a valid expression after '{'" does not match "invalid syntax (?, line 1)"
     def test_syntax_error_for_starred_expressions(self):
         with self.assertRaisesRegex(SyntaxError, "can't use starred expression here"):
             compile("f'{*a}'", "?", "exec")

Lib/test/test_future_stmt/test_future.py

@@ -81,7 +81,6 @@ class FutureTest(unittest.TestCase):
         ):
             from test.test_future_stmt import test_future_multiple_features  # noqa: F401
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 1 != 24
     def test_unknown_future_flag(self):
         code = """
             from __future__ import nested_scopes
@@ -135,14 +134,12 @@ class FutureTest(unittest.TestCase):
         """
         self.assertSyntaxError(code, offset=54)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 1 != 24
     def test_future_import_star(self):
         code = """
             from __future__ import *
         """
         self.assertSyntaxError(code, message='future feature * is not defined', offset=24)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_future_import_braces(self):
         code = """
             from __future__ import braces
@@ -188,7 +185,6 @@ class FutureTest(unittest.TestCase):
         out = kill_python(p)
         self.assertNotIn(b'SyntaxError: invalid syntax', out)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_future_dotted_import(self):
         with self.assertRaises(ImportError):
             exec("from .__future__ import spam")
@@ -480,7 +476,6 @@ class AnnotationsFutureTestCase(unittest.TestCase):
         self.assertEqual(foo.__code__.co_cellvars, ())
         self.assertEqual(foo().__code__.co_freevars, ())
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised
     def test_annotations_forbidden(self):
         with self.assertRaises(SyntaxError):
             self._exec_future("test: (yield)")

Lib/test/test_genexps.py

@@ -159,7 +159,7 @@ Verify that syntax error's are raised for genexps used as lvalues
        ...
     SyntaxError: cannot assign to generator expression
 
-    >>> (y for y in (1,2)) += 10  # TODO: RUSTPYTHON # doctest: +EXPECTED_FAILURE
+    >>> (y for y in (1,2)) += 10
     Traceback (most recent call last):
        ...
     SyntaxError: 'generator expression' is an illegal expression for augmented assignment

Lib/test/test_global.py

@@ -28,7 +28,6 @@ class GlobalTests(unittest.TestCase):
     ### Syntax error cases as covered in Python/symtable.c
     ######################################################
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 12 != 5
     def test_name_param(self):
         prog_text = """\
 def fn(name_param):
@@ -36,7 +35,6 @@ def fn(name_param):
 """
         check_syntax_error(self, prog_text, lineno=2, offset=5)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 12 != 5
     def test_name_after_assign(self):
         prog_text = """\
 def fn():
@@ -45,7 +43,6 @@ def fn():
 """
         check_syntax_error(self, prog_text, lineno=3, offset=5)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 12 != 5
     def test_name_after_use(self):
         prog_text = """\
 def fn():
@@ -54,7 +51,6 @@ def fn():
 """
         check_syntax_error(self, prog_text, lineno=3, offset=5)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 12 != 5
     def test_name_annot(self):
         prog_text_3 = """\
 def fn():

Lib/test/test_grammar.py

@@ -114,7 +114,6 @@ class TokenTests(unittest.TestCase):
         # Sanity check: no literal begins with an underscore
         self.assertRaises(NameError, eval, "_0")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_bad_numerical_literals(self):
         check = self.check_syntax_error
         check("0b12", "invalid digit '2' in binary literal")
@@ -137,7 +136,6 @@ class TokenTests(unittest.TestCase):
         check("1e2_", "invalid decimal literal")
         check("1e+", "invalid decimal literal")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_end_of_numerical_literals(self):
         def check(test, error=False):
             with self.subTest(expr=test):
@@ -251,7 +249,6 @@ the \'lazy\' dog.\n\
                 compile(s, "<test>", "exec")
             self.assertIn("was never closed", str(cm.exception))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised
     @skip_wasi_stack_overflow()
     def test_max_level(self):
         # Macro defined in Parser/lexer/state.h
@@ -298,7 +295,6 @@ class GrammarTests(unittest.TestCase):
         my_lst[one()-1]: int = 5
         self.assertEqual(my_lst, [5])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_var_annot_syntax_errors(self):
         # parser pass
         check_syntax_error(self, "def f: int")
@@ -751,7 +747,6 @@ class GrammarTests(unittest.TestCase):
 
     # Check the heuristic for print & exec covers significant cases
     # As well as placing some limits on false positives
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_former_statements_refer_to_builtins(self):
         keywords = "print", "exec"
         # Cases where we want the custom error
@@ -1165,7 +1160,6 @@ class GrammarTests(unittest.TestCase):
             """,
             True)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_yield(self):
         # Allowed as standalone statement
         def g(): yield 1
@@ -1205,7 +1199,6 @@ class GrammarTests(unittest.TestCase):
         # Check annotation refleak on SyntaxError
         check_syntax_error(self, "def g(a:(yield)): pass")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_yield_in_comprehensions(self):
         # Check yield in comprehensions
         def g(): [x for x in [(yield 1)]]
@@ -1302,7 +1295,6 @@ class GrammarTests(unittest.TestCase):
         else:
             self.fail("AssertionError not raised by 'assert False'")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_assert_syntax_warnings(self):
         # Ensure that we warn users if they provide a non-zero length tuple as
         # the assertion test.
@@ -1317,7 +1309,6 @@ class GrammarTests(unittest.TestCase):
             compile('assert x, "msg"', '<testcase>', 'exec')
             compile('assert False, "msg"', '<testcase>', 'exec')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_assert_warning_promotes_to_syntax_error(self):
         # If SyntaxWarning is configured to be an error, it actually raises a
         # SyntaxError.
@@ -1496,7 +1487,6 @@ class GrammarTests(unittest.TestCase):
         if 1 not in (): pass
         if 1 < 1 > 1 == 1 >= 1 <= 1 != 1 in 1 not in x is x is not x: pass
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_comparison_is_literal(self):
         def check(test, msg):
             self.check_syntax_warning(test, msg)
@@ -1526,7 +1516,6 @@ class GrammarTests(unittest.TestCase):
             compile('True is x', '<testcase>', 'exec')
             compile('... is x', '<testcase>', 'exec')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_warn_missed_comma(self):
         def check(test):
             self.check_syntax_warning(test, msg)

Lib/test/test_listcomps.py

@@ -219,7 +219,6 @@ class ListComprehensionTest(unittest.TestCase):
         """
         self._check_in_scopes(code, raises=NameError)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; SyntaxError: compiler_make_closure: cannot find '__conditional_annotations__' in parent vars
     def test_references___conditional_annotations___nested(self):
         code = """
             class i: [lambda: __conditional_annotations__ for x in y]

Lib/test/test_named_expressions.py

@@ -4,35 +4,30 @@ GLOBAL_VAR = None
 
 class NamedExpressionInvalidTest(unittest.TestCase):
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; wrong error message
     def test_named_expression_invalid_01(self):
         code = """x := 0"""
 
         with self.assertRaisesRegex(SyntaxError, "invalid syntax"):
             exec(code, {}, {})
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; wrong error message
     def test_named_expression_invalid_02(self):
         code = """x = y := 0"""
 
         with self.assertRaisesRegex(SyntaxError, "invalid syntax"):
             exec(code, {}, {})
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; wrong error message
     def test_named_expression_invalid_03(self):
         code = """y := f(x)"""
 
         with self.assertRaisesRegex(SyntaxError, "invalid syntax"):
             exec(code, {}, {})
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; wrong error message
     def test_named_expression_invalid_04(self):
         code = """y0 = y1 := f(x)"""
 
         with self.assertRaisesRegex(SyntaxError, "invalid syntax"):
             exec(code, {}, {})
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; wrong error message
     def test_named_expression_invalid_06(self):
         code = """((a, b) := (1, 2))"""
 

Lib/test/test_patma.py

@@ -82,7 +82,6 @@ class TestInheritance(unittest.TestCase):
         self.assertEqual(self.check_mapping_then_sequence(S3()), "seq")
         self.assertEqual(self.check_mapping_then_sequence(S4()), "seq")
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_late_registration_mapping(self):
         class Parent:
             pass
@@ -106,7 +105,6 @@ class TestInheritance(unittest.TestCase):
         self.assertEqual(self.check_mapping_then_sequence(ChildPost()), "map")
         self.assertEqual(self.check_mapping_then_sequence(GrandchildPost()), "map")
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_late_registration_sequence(self):
         class Parent:
             pass
@@ -2246,7 +2244,6 @@ class TestPatma(unittest.TestCase):
         self.assertEqual(f(None), {})
         self.assertEqual(f((1, 2)), {})
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_patma_210(self):
         def f(w):
             match w:
@@ -2955,7 +2952,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_invalid_syntax_3(self):
         self.assert_syntax_error("""
         match ...:
@@ -3075,7 +3071,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_patterns_may_only_match_literals_and_attribute_lookups_0(self):
         self.assert_syntax_error("""
         match ...:
@@ -3083,7 +3078,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_patterns_may_only_match_literals_and_attribute_lookups_1(self):
         self.assert_syntax_error("""
         match ...:
@@ -3126,7 +3120,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_real_number_wrong_ops(self):
         for op in ["*", "/", "@", "**", "%", "//"]:
             with self.subTest(op=op):
@@ -3202,7 +3195,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_mapping_pattern_duplicate_key_edge_case0(self):
         self.assert_syntax_error("""
         match ...:
@@ -3210,7 +3202,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_mapping_pattern_duplicate_key_edge_case1(self):
         self.assert_syntax_error("""
         match ...:
@@ -3225,8 +3216,6 @@ class TestSyntaxErrors(unittest.TestCase):
                 pass
         """)
 
-
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_mapping_pattern_duplicate_key_edge_case3(self):
         self.assert_syntax_error("""
         match ...:
@@ -3258,7 +3247,6 @@ class TestTypeErrors(unittest.TestCase):
         self.assertEqual(x, range(10))
         self.assertIs(y, None)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_got_multiple_subpatterns_for_attribute_0(self):
         class Class:
             __match_args__ = ("a", "a")
@@ -3273,7 +3261,6 @@ class TestTypeErrors(unittest.TestCase):
         self.assertIs(y, None)
         self.assertIs(z, None)
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_got_multiple_subpatterns_for_attribute_1(self):
         class Class:
             __match_args__ = ("a",)
@@ -3379,7 +3366,6 @@ class TestTypeErrors(unittest.TestCase):
 
 class TestValueErrors(unittest.TestCase):
 
-    @unittest.expectedFailure   # TODO: RUSTPYTHON
     def test_mapping_pattern_checks_duplicate_key_1(self):
         class Keys:
             KEY = "a"

Lib/test/test_pdb.py

@@ -2165,7 +2165,7 @@ if not SKIP_CORO_TESTS:
             >>> def test_function():
             ...     asyncio.run(main(), loop_factory=asyncio.EventLoop)
 
-            >>> with PdbTestInput([  # TODO: RUSTPYTHON # doctest: +ELLIPSIS +EXPECTED_FAILURE
+            >>> with PdbTestInput([  # doctest: +ELLIPSIS
             ...     'x = await task',
             ...     'p x',
             ...     'x = await test()',
@@ -2280,7 +2280,7 @@ if not SKIP_CORO_TESTS:
             >>> def test_function():
             ...     asyncio.run(main(), loop_factory=asyncio.EventLoop)
 
-            >>> with PdbTestInput([  # TODO: RUSTPYTHON # doctest: +EXPECTED_FAILURE
+            >>> with PdbTestInput([
             ...     'p var.get()',
             ...     'print(await get_var())',
             ...     'print(await asyncio.create_task(set_var(100)))',
@@ -2768,7 +2768,7 @@ def test_pdb_multiline_statement():
     >>> def test_function():
     ...     import pdb; pdb.Pdb(nosigint=True, readrc=False).set_trace()
 
-    >>> with PdbTestInput([  # TODO: RUSTPYTHON # doctest: +NORMALIZE_WHITESPACE +EXPECTED_FAILURE
+    >>> with PdbTestInput([  # doctest: +NORMALIZE_WHITESPACE
     ...     'def f(x):',
     ...     '  return x * 2',
     ...     '',

Lib/test/test_peepholer.py

@@ -157,7 +157,6 @@ class TestTranforms(BytecodeTestCase):
                 self.assertNotInBytecode(code, 'UNPACK_SEQUENCE')
                 self.check_lnotab(code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; LOAD_CONST count mismatch in long-tuple branch
     def test_constant_folding_tuples_of_constants(self):
         for line, elem in (
             ('a = 1,2,3', (1, 2, 3)),

Lib/test/test_pep646_syntax.py

@@ -312,7 +312,7 @@ is *not* valid syntax.)
     >>> f4.__annotations__
     {'args': StarredB, 'arg1': <class 'int'>}
 
-    >>> def f5(*args: *b = (1,)): pass  # TODO: RUSTPYTHON # doctest: +EXPECTED_FAILURE
+    >>> def f5(*args: *b = (1,)): pass
     Traceback (most recent call last):
         ...
     SyntaxError: invalid syntax

Lib/test/test_pydoc/test_pydoc.py

@@ -932,7 +932,6 @@ class PydocDocTest(unittest.TestCase):
             synopsis = pydoc.synopsis(TESTFN, {})
             self.assertEqual(synopsis, 'line 1: h\xe9')
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_source_synopsis(self):
         def check(source, expected, encoding=None):
             if isinstance(source, str):

Lib/test/test_pyrepl/test_interact.py

@@ -117,7 +117,6 @@ class TestSimpleInteract(unittest.TestCase):
 SyntaxError: duplicate argument 'x' in function definition"""
         self.assertIn(r, f.getvalue())
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_runsource_shows_syntax_error_for_failed_compilation(self):
         console = InteractiveColoredConsole()
         source = "print('Hello, world!'"
@@ -133,7 +132,6 @@ SyntaxError: duplicate argument 'x' in function definition"""
             console.runsource(source)
             mock_showsyntaxerror.assert_called_once()
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_runsource_survives_null_bytes(self):
         console = InteractiveColoredConsole()
         source = "\x00\n"
@@ -155,7 +153,6 @@ SyntaxError: duplicate argument 'x' in function definition"""
         self.assertFalse(result)
         self.assertEqual(f.getvalue(), "{'x': <class 'int'>}\n")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_future_annotations(self):
         console = InteractiveColoredConsole()
         source = dedent("""\
@@ -210,7 +207,6 @@ class TestMoreLines(unittest.TestCase):
         console = InteractiveColoredConsole(namespace, filename="<stdin>")
         self.assertFalse(_more_lines(console, code))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_multiline_single_block(self):
         namespace = {}
         code = dedent("""\
@@ -227,7 +223,6 @@ class TestMoreLines(unittest.TestCase):
         console = InteractiveColoredConsole(namespace, filename="<stdin>")
         self.assertFalse(_more_lines(console, code))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_multiple_statements(self):
         namespace = {}
         code = dedent("""\
@@ -237,7 +232,6 @@ class TestMoreLines(unittest.TestCase):
         console = InteractiveColoredConsole(namespace, filename="<stdin>")
         self.assertTrue(_more_lines(console, code))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_multiple_blocks(self):
         namespace = {}
         code = dedent("""\
@@ -285,7 +279,6 @@ class TestMoreLines(unittest.TestCase):
 
 
 class TestWarnings(unittest.TestCase):
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_pep_765_warning(self):
         """
         Test that a SyntaxWarning emitted from the

Lib/test/test_pyrepl/test_pyrepl.py

@@ -466,7 +466,6 @@ class TestPyReplAutoindent(TestCase):
         reader = ReadlineAlikeReader(console=console, config=config)
         return reader
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_default(self):
         # fmt: off
         input_code = (
@@ -486,7 +485,6 @@ class TestPyReplAutoindent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, output_code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_continuation(self):
         # auto indenting according to previous user indentation
         # fmt: off
@@ -514,7 +512,6 @@ class TestPyReplAutoindent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, output_code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_prev_block(self):
         # auto indenting according to indentation in different block
         # fmt: off
@@ -546,7 +543,6 @@ class TestPyReplAutoindent(TestCase):
         output2 = multiline_input(reader)
         self.assertEqual(output2, output_code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_multiline(self):
         # fmt: off
         events = itertools.chain(
@@ -586,7 +582,6 @@ class TestPyReplAutoindent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, output_code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_with_comment(self):
         # fmt: off
         events = code_to_events(
@@ -605,7 +600,6 @@ class TestPyReplAutoindent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, output_code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_auto_indent_with_multicomment(self):
         # fmt: off
         events = code_to_events(
@@ -680,7 +674,6 @@ class TestPyReplOutput(ScreenEqualMixin, TestCase):
         wrapper = _ReadlineWrapper(f_in=None, f_out=None, reader=reader)
         self.assertIs(type(wrapper.get_line_buffer()), str)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_multiline_edit(self):
         events = itertools.chain(
             code_to_events("def f():\n...\n\n"),
@@ -744,7 +737,6 @@ class TestPyReplOutput(ScreenEqualMixin, TestCase):
         self.assertEqual(output, "1+1")
         self.assert_screen_equal(reader, "1+1", clean=True)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_history_with_multiline_entries(self):
         code = "def foo():\nx = 1\ny = 2\nz = 3\n\ndef bar():\nreturn 42\n\n"
         events = list(itertools.chain(
@@ -1426,7 +1418,6 @@ class TestPasteEvent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, code)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_paste_mid_newlines_not_in_paste_mode(self):
         # fmt: off
         code = (
@@ -1448,7 +1439,6 @@ class TestPasteEvent(TestCase):
         output = multiline_input(reader)
         self.assertEqual(output, expected)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_paste_not_in_paste_mode(self):
         # fmt: off
         input_code = (

Lib/test/test_pyrepl/test_reader.py

@@ -180,7 +180,6 @@ class TestReader(ScreenEqualMixin, TestCase):
         reader, _ = handle_all_events(events)
         self.assert_screen_equal(reader, "")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: Lists differ
     def test_newline_within_block_trailing_whitespace(self):
         # fmt: off
         code = (

Lib/test/test_repl.py

@@ -437,7 +437,6 @@ class TestAsyncioREPL(unittest.TestCase):
         expected = "toplevel contextvar test: ok"
         self.assertIn(expected, output, expected)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 101 != 0
     def test_toplevel_contextvars_async(self):
         user_input = dedent("""\
         from contextvars import ContextVar

Lib/test/test_symtable.py

@@ -198,7 +198,6 @@ class SymtableTest(unittest.TestCase):
     T = find_block(GenericMine, "T")
     U = find_block(GenericMine, "U")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: <SymbolTableType.ANNOTATION: 'annotation'> != 'type alias'
     def test_type(self):
         self.assertEqual(self.top.get_type(), "module")
         self.assertEqual(self.Mine.get_type(), "class")

Lib/test/test_sys_setprofile.py

@@ -169,7 +169,6 @@ class ProfileHookTestCase(TestCaseBase):
                               (1, 'return', g_ident),
                               ])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_exception_propagation(self):
         def f(p):
             1/0

Lib/test/test_type_comments.py

@@ -252,7 +252,6 @@ class TypeCommentTests(unittest.TestCase):
     def classic_parse(self, source):
         return ast.parse(source)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AttributeError: 'FunctionDef' object has no attribute 'type_comment'
     def test_funcdef(self):
         for tree in self.parse_all(funcdef):
             self.assertEqual(tree.body[0].type_comment, "() -> int")
@@ -261,7 +260,6 @@ class TypeCommentTests(unittest.TestCase):
         self.assertEqual(tree.body[0].type_comment, None)
         self.assertEqual(tree.body[1].type_comment, None)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_asyncdef(self):
         for tree in self.parse_all(asyncdef, minver=5):
             self.assertEqual(tree.body[0].type_comment, "() -> int")
@@ -274,12 +272,10 @@ class TypeCommentTests(unittest.TestCase):
         with self.assertRaises(SyntaxError):
             self.classic_parse(asyncvar)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_asynccomp(self):
         for tree in self.parse_all(asynccomp, minver=6):
             pass
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_matmul(self):
         for tree in self.parse_all(matmul, minver=5):
             pass
@@ -288,37 +284,31 @@ class TypeCommentTests(unittest.TestCase):
         for tree in self.parse_all(fstring):
             pass
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_underscorednumber(self):
         for tree in self.parse_all(underscorednumber, minver=6):
             pass
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_redundantdef(self):
         for tree in self.parse_all(redundantdef, maxver=0,
                                 expected_regex="^Cannot have two type comments on def"):
             pass
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AttributeError: 'FunctionDef' object has no attribute 'type_comment'
     def test_nonasciidef(self):
         for tree in self.parse_all(nonasciidef):
             self.assertEqual(tree.body[0].type_comment, "() -> àçčéñt")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AttributeError: 'For' object has no attribute 'type_comment'
     def test_forstmt(self):
         for tree in self.parse_all(forstmt):
             self.assertEqual(tree.body[0].type_comment, "int")
         tree = self.classic_parse(forstmt)
         self.assertEqual(tree.body[0].type_comment, None)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AttributeError: 'With' object has no attribute 'type_comment'
     def test_withstmt(self):
         for tree in self.parse_all(withstmt):
             self.assertEqual(tree.body[0].type_comment, "int")
         tree = self.classic_parse(withstmt)
         self.assertEqual(tree.body[0].type_comment, None)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AttributeError: 'With' object has no attribute 'type_comment'
     def test_parenthesized_withstmt(self):
         for tree in self.parse_all(parenthesized_withstmt):
             self.assertEqual(tree.body[0].type_comment, "int")
@@ -327,14 +317,12 @@ class TypeCommentTests(unittest.TestCase):
         self.assertEqual(tree.body[0].type_comment, None)
         self.assertEqual(tree.body[1].type_comment, None)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: None != 'int'
     def test_vardecl(self):
         for tree in self.parse_all(vardecl):
             self.assertEqual(tree.body[0].type_comment, "int")
         tree = self.classic_parse(vardecl)
         self.assertEqual(tree.body[0].type_comment, None)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; +  (11, ' whatever')]
     def test_ignores(self):
         for tree in self.parse_all(ignores):
             self.assertEqual(
@@ -350,7 +338,6 @@ class TypeCommentTests(unittest.TestCase):
         tree = self.classic_parse(ignores)
         self.assertEqual(tree.type_ignores, [])
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: SyntaxError not raised : feature_version=(3, 4)
     def test_longargs(self):
         for tree in self.parse_all(longargs, minver=8):
             for t in tree.body:
@@ -381,7 +368,6 @@ class TypeCommentTests(unittest.TestCase):
                     self.assertIsNone(arg.type_comment, "%s(%s:%r)" %
                                       (t.name, arg.arg, arg.type_comment))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; Tests for inappropriately-placed type comments.
     def test_inappropriate_type_comments(self):
         """Tests for inappropriately-placed type comments.
 
@@ -416,7 +402,6 @@ class TypeCommentTests(unittest.TestCase):
             _testcapi.Py_CompileStringExFlags(
                 b"def a(f=8, #type: \x80\n\x80", "<test>", 256, flags)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; ValueError: mode must be "exec", "eval", "ipython", or "single"
     def test_func_type_input(self):
 
         def parse_func_type_input(source):

Lib/test/test_type_params.py

@@ -683,7 +683,6 @@ class TypeParamsClassScopeTest(unittest.TestCase):
         self.assertIs(X.foo.__type_params__[0].__bound__, float)
         self.assertIs(X.Alias.__value__, float)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; + global
     def test_binding_uses_global(self):
         ns = run_code("""
             x = "global"
@@ -1076,7 +1075,6 @@ class TypeParamsTypeVarTest(unittest.TestCase):
 
 
 class TypeParamsTypeVarTupleTest(unittest.TestCase):
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "cannot use bound with TypeVarTuple" does not match "invalid syntax (<test string>, line 1)"
     def test_typevartuple_01(self):
         code = """def func1[*A: str](): pass"""
         check_syntax_error(self, code, "cannot use bound with TypeVarTuple")
@@ -1100,7 +1098,6 @@ class TypeParamsTypeVarTupleTest(unittest.TestCase):
 
 
 class TypeParamsTypeVarParamSpecTest(unittest.TestCase):
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "cannot use bound with ParamSpec" does not match "invalid syntax (<test string>, line 1)"
     def test_paramspec_01(self):
         code = """def func1[**A: str](): pass"""
         check_syntax_error(self, code, "cannot use bound with ParamSpec")

Lib/test/test_unpack_ex.py

@@ -168,7 +168,7 @@ List comprehension element unpacking
     ...
     SyntaxError: iterable unpacking cannot be used in comprehension
 
-    >>> {**{} for a in [1]} # TODO: RUSTPYTHON # doctest: +EXPECTED_FAILURE
+    >>> {**{} for a in [1]}
     Traceback (most recent call last):
     ...
     SyntaxError: dict unpacking cannot be used in dict comprehension
@@ -356,7 +356,7 @@ Now some general starred expressions (all fail).
       ...
     SyntaxError: can't use starred expression here
 
-    >>> (*x),y = 1, 2 # TODO: RUSTPYTHON # doctest:+ELLIPSIS +EXPECTED_FAILURE
+    >>> (*x),y = 1, 2
     Traceback (most recent call last):
       ...
     SyntaxError: cannot use starred expression here
@@ -366,12 +366,12 @@ Now some general starred expressions (all fail).
       ...
     SyntaxError: cannot use starred expression here
 
-    >>> z,(*x),y = 1, 2, 4 # TODO: RUSTPYTHON # doctest:+ELLIPSIS +EXPECTED_FAILURE
+    >>> z,(*x),y = 1, 2, 4
     Traceback (most recent call last):
       ...
     SyntaxError: cannot use starred expression here
 
-    >>> z,(*x) = 1, 2 # TODO: RUSTPYTHON # doctest:+ELLIPSIS +EXPECTED_FAILURE
+    >>> z,(*x) = 1, 2
     Traceback (most recent call last):
       ...
     SyntaxError: cannot use starred expression here

crates/capi/src/ceval.rs

@@ -1,17 +1,13 @@
 use crate::pystate::with_vm;
+use crate::unicodeobject::decode_fsdefault_and_size;
 use core::ffi::{CStr, c_char, c_int};
 use core::ptr::NonNull;
 use rustpython_vm::builtins::{PyCode, PyDict};
-use rustpython_vm::compiler::Mode;
 use rustpython_vm::function::ArgMapping;
 use rustpython_vm::scope::Scope;
+use rustpython_vm::version;
 use rustpython_vm::{AsObject, PyObject, TryFromObject};
 
-const PY_SINGLE_INPUT: c_int = 256;
-const PY_FILE_INPUT: c_int = 257;
-const PY_EVAL_INPUT: c_int = 258;
-const PY_FUNC_TYPE_INPUT: c_int = 345;
-
 #[unsafe(no_mangle)]
 pub unsafe extern "C" fn Py_CompileString(
     code: *const c_char,
@@ -19,27 +15,11 @@ pub unsafe extern "C" fn Py_CompileString(
     start: c_int,
 ) -> *mut PyObject {
     with_vm(|vm| {
-        let code = unsafe { CStr::from_ptr(code) }.to_str().map_err(|_| {
-            vm.new_system_error("Py_CompileString called with non UTF-8 code string")
-        })?;
-        let filename = unsafe { CStr::from_ptr(filename) }
-            .to_str()
-            .map_err(|_| vm.new_system_error("Py_CompileString called with non UTF-8 filename"))?;
-
-        let mode = match start {
-            PY_SINGLE_INPUT => Mode::Single,
-            PY_FILE_INPUT => Mode::Exec,
-            PY_EVAL_INPUT => Mode::Eval,
-            PY_FUNC_TYPE_INPUT => Mode::BlockExpr,
-            _ => {
-                return Err(
-                    vm.new_system_error("Invalid start argument passed to Py_CompileString")
-                );
-            }
-        };
-
-        vm.compile(code, mode, filename)
-            .map_err(|err| vm.new_syntax_error(&err, Some(code)))
+        let code = unsafe { CStr::from_ptr(code) }.to_bytes();
+        let filename_size = unsafe { CStr::from_ptr(filename) }.to_bytes().len();
+        let filename = decode_fsdefault_and_size(vm, filename, filename_size)?;
+        let filename = filename.to_string_lossy();
+        vm.compile_string_object_with_flags(code, &filename, start, 0, version::MINOR as c_int, -1)
     })
 }
 

crates/capi/src/unicodeobject.rs

@@ -4,8 +4,8 @@ use core::ffi::{CStr, c_char, c_int};
 use core::ptr::NonNull;
 use core::slice;
 use core::str;
-use rustpython_vm::PyObjectRef;
-use rustpython_vm::builtins::PyStr;
+use rustpython_vm::builtins::{PyStr, PyStrRef};
+use rustpython_vm::{PyObjectRef, PyResult, VirtualMachine};
 
 define_py_check!(fn PyUnicode_Check, types.str_type);
 define_py_check!(exact fn PyUnicode_CheckExact, types.str_type);
@@ -113,26 +113,42 @@ pub unsafe extern "C" fn PyUnicode_DecodeFSDefaultAndSize(
             .try_into()
             .map_err(|_| vm.new_system_error("size must be non-negative"))?;
 
-        let bytes = if s.is_null() {
-            if size != 0 {
-                return Err(vm.new_system_error(
-                    "PyUnicode_DecodeFSDefaultAndSize called with null data and non-zero size",
-                ));
-            }
-            &[][..]
-        } else {
-            unsafe { slice::from_raw_parts(s.cast::<u8>(), size) }
-        };
-
-        vm.state.codec_registry.decode_text(
-            vm.ctx.new_bytes(bytes.to_vec()).into(),
-            vm.fs_encoding().as_str(),
-            Some(vm.fs_encode_errors().to_owned()),
-            vm,
-        )
+        decode_fsdefault_and_size(vm, s, size)
     })
 }
 
+#[unsafe(no_mangle)]
+pub unsafe extern "C" fn PyUnicode_DecodeFSDefault(s: *const c_char) -> *mut PyObject {
+    with_vm(|vm| {
+        let size = unsafe { CStr::from_ptr(s) }.to_bytes().len();
+        decode_fsdefault_and_size(vm, s, size)
+    })
+}
+
+pub(crate) fn decode_fsdefault_and_size(
+    vm: &VirtualMachine,
+    s: *const c_char,
+    size: usize,
+) -> PyResult<PyStrRef> {
+    let bytes = if s.is_null() {
+        if size != 0 {
+            return Err(vm.new_system_error(
+                "PyUnicode_DecodeFSDefaultAndSize called with null data and non-zero size",
+            ));
+        }
+        &[][..]
+    } else {
+        unsafe { slice::from_raw_parts(s.cast::<u8>(), size) }
+    };
+
+    vm.state.codec_registry.decode_text(
+        vm.ctx.new_bytes(bytes.to_vec()).into(),
+        vm.fs_encoding().as_str(),
+        Some(vm.fs_encode_errors().to_owned()),
+        vm,
+    )
+}
+
 #[unsafe(no_mangle)]
 pub unsafe extern "C" fn PyUnicode_EncodeFSDefault(unicode: *mut PyObject) -> *mut PyObject {
     with_vm(|vm| {

crates/codegen/src/error.rs

@@ -3,21 +3,6 @@ use core::fmt::Display;
 use rustpython_compiler_core::SourceLocation;
 use thiserror::Error;
 
-#[derive(Clone, Copy, Debug)]
-pub enum PatternUnreachableReason {
-    NameCapture,
-    Wildcard,
-}
-
-impl Display for PatternUnreachableReason {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match self {
-            Self::NameCapture => write!(f, "name capture"),
-            Self::Wildcard => write!(f, "wildcard"),
-        }
-    }
-}
-
 // pub type CodegenError = rustpython_parser_core::source_code::LocatedError<CodegenErrorType>;
 
 #[derive(Error, Debug)]
@@ -70,6 +55,8 @@ pub enum CodegenErrorType {
     SyntaxError(String),
     /// Multiple `*` detected
     MultipleStarArgs,
+    MultipleStarredExpressionsInSequencePattern,
+    MultipleStarredNamesInSequencePattern,
     /// Misplaced `*` expression
     InvalidStarExpr,
     /// Break statement outside of loop.
@@ -87,14 +74,17 @@ pub enum CodegenErrorType {
     AsyncReturnValue,
     InvalidFuturePlacement,
     InvalidFutureFeature(String),
-    FunctionImportStar,
+    InvalidFutureBraces,
+    RecursionError,
     TooManyStarUnpack,
+    TooManyExpressionsInStarUnpackingSequencePattern,
     EmptyWithItems,
     EmptyWithBody,
     ForbiddenName,
     DuplicateStore(String),
-    UnreachablePattern(PatternUnreachableReason),
-    RepeatedAttributePattern,
+    UnreachableWildcardPattern,
+    UnreachableNameCapturePattern(String),
+    RepeatedAttributePattern(String),
     ConflictingNameBindPattern,
     /// break/continue/return inside except* block
     BreakContinueReturnInExceptStar,
@@ -112,6 +102,12 @@ impl fmt::Display for CodegenErrorType {
             Self::MultipleStarArgs => {
                 write!(f, "multiple starred expressions in assignment")
             }
+            Self::MultipleStarredExpressionsInSequencePattern => {
+                write!(f, "multiple starred expressions in sequence pattern")
+            }
+            Self::MultipleStarredNamesInSequencePattern => {
+                write!(f, "multiple starred names in sequence pattern")
+            }
             Self::InvalidStarExpr => write!(f, "can't use starred expression here"),
             Self::InvalidBreak => write!(f, "'break' outside loop"),
             Self::InvalidContinue => write!(f, "'continue' not properly in loop"),
@@ -128,9 +124,7 @@ impl fmt::Display for CodegenErrorType {
                 )
             }
             Self::AsyncYieldFrom => write!(f, "'yield from' inside async function"),
-            Self::AsyncReturnValue => {
-                write!(f, "'return' with value inside async generator")
-            }
+            Self::AsyncReturnValue => write!(f, "'return' with value in async generator"),
             Self::InvalidFuturePlacement => write!(
                 f,
                 "from __future__ imports must occur at the beginning of the file"
@@ -138,12 +132,16 @@ impl fmt::Display for CodegenErrorType {
             Self::InvalidFutureFeature(feat) => {
                 write!(f, "future feature {feat} is not defined")
             }
-            Self::FunctionImportStar => {
-                write!(f, "import * only allowed at module level")
+            Self::InvalidFutureBraces => write!(f, "not a chance"),
+            Self::RecursionError => {
+                write!(f, "maximum recursion depth exceeded during compilation")
             }
             Self::TooManyStarUnpack => {
                 write!(f, "too many expressions in star-unpacking assignment")
             }
+            Self::TooManyExpressionsInStarUnpackingSequencePattern => {
+                write!(f, "too many expressions in star-unpacking sequence pattern")
+            }
             Self::EmptyWithItems => {
                 write!(f, "empty items on With")
             }
@@ -153,14 +151,18 @@ impl fmt::Display for CodegenErrorType {
             Self::ForbiddenName => {
                 write!(f, "forbidden attribute name")
             }
-            Self::DuplicateStore(s) => {
-                write!(f, "duplicate store {s}")
+            Self::DuplicateStore(s) => write!(f, "multiple assignments to name '{s}' in pattern"),
+            Self::UnreachableWildcardPattern => {
+                write!(f, "wildcard makes remaining patterns unreachable")
             }
-            Self::UnreachablePattern(reason) => {
-                write!(f, "{reason} makes remaining patterns unreachable")
+            Self::UnreachableNameCapturePattern(name) => {
+                write!(
+                    f,
+                    "name capture '{name}' makes remaining patterns unreachable"
+                )
             }
-            Self::RepeatedAttributePattern => {
-                write!(f, "attribute name repeated in class pattern")
+            Self::RepeatedAttributePattern(name) => {
+                write!(f, "attribute name repeated in class pattern: {name}")
             }
             Self::ConflictingNameBindPattern => {
                 write!(f, "alternative patterns bind different names")

crates/codegen/src/ir.rs

@@ -78,15 +78,108 @@ impl ConstantPool {
         }
     }
 
-    pub fn insert_full(&mut self, constant: ConstantData) -> (usize, bool) {
-        // CPython's _PyCode_ConstantKey() keeps NaN-bearing constants distinct
-        // because Python-level NaN keys do not compare equal.
-        if !Self::constant_contains_nan(&constant)
-            && let Some(idx) = self
-                .constants
+    fn frozenset_key_contains(elements: &[ConstantData], needle: &ConstantData) -> bool {
+        if Self::constant_contains_nan(needle) {
+            return false;
+        }
+        elements.iter().any(|element| {
+            !Self::constant_contains_nan(element) && Self::constant_key_eq(element, needle)
+        })
+    }
+
+    fn frozenset_key_eq(left: &[ConstantData], right: &[ConstantData]) -> bool {
+        left.iter()
+            .all(|element| Self::frozenset_key_contains(right, element))
+            && right
                 .iter()
-                .position(|existing| existing == &constant)
-        {
+                .all(|element| Self::frozenset_key_contains(left, element))
+    }
+
+    fn constant_key_eq(left: &ConstantData, right: &ConstantData) -> bool {
+        match (left, right) {
+            (ConstantData::Tuple { elements: left }, ConstantData::Tuple { elements: right }) => {
+                left.len() == right.len()
+                    && left
+                        .iter()
+                        .zip(right.iter())
+                        .all(|(left, right)| Self::constant_key_eq(left, right))
+            }
+            (
+                ConstantData::Frozenset { elements: left },
+                ConstantData::Frozenset { elements: right },
+            ) => Self::frozenset_key_eq(left, right),
+            (ConstantData::Slice { elements: left }, ConstantData::Slice { elements: right }) => {
+                left.iter()
+                    .zip(right.iter())
+                    .all(|(left, right)| Self::constant_key_eq(left, right))
+            }
+            _ => left == right,
+        }
+    }
+
+    fn canonicalize_constant_key(constant: ConstantData) -> crate::InternalResult<ConstantData> {
+        match constant {
+            ConstantData::Tuple { elements } => {
+                let mut canonical = Vec::new();
+                canonical
+                    .try_reserve_exact(elements.len())
+                    .map_err(|_| InternalError::MalformedControlFlowGraph)?;
+                for element in elements {
+                    canonical.push(Self::canonicalize_constant_key(element)?);
+                }
+                Ok(ConstantData::Tuple {
+                    elements: canonical,
+                })
+            }
+            ConstantData::Slice { elements } => {
+                let [start, stop, step] = *elements;
+                Ok(ConstantData::Slice {
+                    elements: Box::new([
+                        Self::canonicalize_constant_key(start)?,
+                        Self::canonicalize_constant_key(stop)?,
+                        Self::canonicalize_constant_key(step)?,
+                    ]),
+                })
+            }
+            ConstantData::Frozenset { elements } => {
+                let mut canonical = Vec::new();
+                canonical
+                    .try_reserve_exact(elements.len())
+                    .map_err(|_| InternalError::MalformedControlFlowGraph)?;
+                for element in elements {
+                    let element = Self::canonicalize_constant_key(element)?;
+                    if !Self::frozenset_key_contains(&canonical, &element) {
+                        canonical.push(element);
+                    }
+                }
+                Ok(ConstantData::Frozenset {
+                    elements: canonical,
+                })
+            }
+            other => Ok(other),
+        }
+    }
+
+    fn canonicalize_constant_key_infallible(constant: ConstantData) -> ConstantData {
+        Self::canonicalize_constant_key(constant)
+            .expect("constant key canonicalization only fails on allocation error")
+    }
+
+    /// Index of an already-stored constant equal to `constant`, if any.
+    /// _PyCode_ConstantKey() keeps NaN-bearing constants distinct because
+    /// Python-level NaN keys do not compare equal.
+    fn find_existing(&self, constant: &ConstantData) -> Option<usize> {
+        if Self::constant_contains_nan(constant) {
+            return None;
+        }
+        self.constants
+            .iter()
+            .position(|existing| Self::constant_key_eq(existing, constant))
+    }
+
+    pub fn insert_full(&mut self, constant: ConstantData) -> (usize, bool) {
+        let constant = Self::canonicalize_constant_key_infallible(constant);
+        if let Some(idx) = self.find_existing(&constant) {
             return (idx, false);
         }
         let idx = self.constants.len();
@@ -95,14 +188,8 @@ impl ConstantPool {
     }
 
     fn try_insert_full(&mut self, constant: ConstantData) -> crate::InternalResult<(usize, bool)> {
-        // CPython's _PyCode_ConstantKey() keeps NaN-bearing constants distinct
-        // because Python-level NaN keys do not compare equal.
-        if !Self::constant_contains_nan(&constant)
-            && let Some(idx) = self
-                .constants
-                .iter()
-                .position(|existing| existing == &constant)
-        {
+        let constant = Self::canonicalize_constant_key(constant)?;
+        if let Some(idx) = self.find_existing(&constant) {
             return Ok((idx, false));
         }
         self.constants
@@ -1485,6 +1572,7 @@ impl Blocks {
                     | PseudoInstruction::JumpIfTrue { .. }),
                 ) => {
                     let opcode = pseudo.into();
+                    let opcode_is_false = matches!(pseudo, PseudoInstruction::JumpIfFalse { .. });
                     match target.instr.pseudo().map(Into::into) {
                         Some(PseudoOpcode::Jump)
                             if self.jump_thread(block_idx, i, &target, opcode)? =>
@@ -1492,22 +1580,25 @@ impl Blocks {
                             continue;
                         }
                         Some(PseudoOpcode::JumpIfFalse)
-                            if matches!(
-                                opcode,
-                                AnyInstruction::Pseudo(PseudoInstruction::JumpIfFalse { .. })
-                            ) && self.jump_thread(block_idx, i, &target, opcode)? =>
+                            if opcode_is_false
+                                && self.jump_thread(block_idx, i, &target, opcode)? =>
                         {
                             continue;
                         }
                         Some(PseudoOpcode::JumpIfTrue)
-                            if matches!(
-                                opcode,
-                                AnyInstruction::Pseudo(PseudoInstruction::JumpIfTrue { .. })
-                            ) && self.jump_thread(block_idx, i, &target, opcode)? =>
+                            if !opcode_is_false
+                                && self.jump_thread(block_idx, i, &target, opcode)? =>
                         {
                             continue;
                         }
-                        Some(PseudoOpcode::JumpIfFalse | PseudoOpcode::JumpIfTrue) => {
+                        Some(PseudoOpcode::JumpIfTrue) if opcode_is_false => {
+                            let next = self[inst.target].next;
+                            debug_assert!(next != BlockIdx::NULL);
+                            debug_assert!(next != inst.target);
+                            self[block_idx].instructions[i].target = next;
+                            continue;
+                        }
+                        Some(PseudoOpcode::JumpIfFalse) if !opcode_is_false => {
                             let next = self[inst.target].next;
                             debug_assert!(next != BlockIdx::NULL);
                             debug_assert!(next != inst.target);
@@ -2335,10 +2426,10 @@ impl Blocks {
         Ok(())
     }
 
-    /// flowgraph.c mark_cold (two-pass to match CPython).
+    /// flowgraph.c mark_cold (two-pass).
     ///
     /// Phase 1 (mark_warm): propagate "warm" from entry via fall-through and
-    /// jump targets. CPython asserts while visiting warm blocks that they are not
+    /// jump targets. The pass asserts while visiting warm blocks that they are not
     /// exception handlers.
     ///
     /// Phase 2 (mark_cold): propagate "cold" from except_handler blocks via
@@ -2349,7 +2440,7 @@ impl Blocks {
     /// empty unreachable placeholders left by remove_unreachable; they stay in
     /// their original chain position (e.g. between entry and the post-try
     /// continuation for a nested try/except whose inner_end was emptied by
-    /// optimize_cfg). This matches CPython's behavior and is necessary for
+    /// optimize_cfg). This is necessary for
     /// optimize_load_fast to terminate fall-through at those placeholders.
     /// flowgraph.c mark_warm
     fn mark_warm(&mut self) -> crate::InternalResult<()> {
@@ -2981,42 +3072,12 @@ impl Blocks {
     }
 }
 
-impl From<Vec<Block>> for Blocks {
-    fn from(value: Vec<Block>) -> Self {
-        Self(value)
-    }
-}
-
-impl From<Box<[Block]>> for Blocks {
-    fn from(value: Box<[Block]>) -> Self {
-        Self(value.into())
-    }
-}
-
-impl From<&[Block]> for Blocks {
-    fn from(value: &[Block]) -> Self {
-        Self(value.to_vec())
-    }
-}
-
-impl From<&mut [Block]> for Blocks {
-    fn from(value: &mut [Block]) -> Self {
-        Self(value.to_vec())
-    }
-}
-
 impl<const N: usize> From<[Block; N]> for Blocks {
     fn from(value: [Block; N]) -> Self {
         Self(value.into())
     }
 }
 
-impl<const N: usize> From<&[Block; N]> for Blocks {
-    fn from(value: &[Block; N]) -> Self {
-        Self(value.to_vec())
-    }
-}
-
 impl Deref for Blocks {
     type Target = [Block];
 
@@ -3060,7 +3121,7 @@ impl IndexMut<BlockIdx> for Blocks {
 }
 
 pub(crate) const START_DEPTH_UNSET: i32 = i32::MIN;
-const CO_MAXBLOCKS: usize = 20;
+const CO_MAXBLOCKS: usize = 21;
 
 /// flowgraph.c struct _PyCfgExceptStack
 #[derive(Clone, Debug)]
@@ -3099,7 +3160,7 @@ impl CfgTraversalStack {
 #[derive(Clone, Debug)]
 pub(crate) struct InstructionSequenceLabelMap {
     block_labels: Vec<InstructionSequenceLabel>,
-    /// Codegen-side shadow of CPython's instruction-sequence label map.
+    /// Codegen-side shadow of the instruction-sequence label map.
     ///
     /// `_PyInstructionSequence_UseLabel()` can map multiple labels to the same
     /// instruction offset before `_PyCfg_FromInstructionSequence()` materializes
@@ -3305,7 +3366,7 @@ pub struct CodeInfo {
 
     // Reference to the symbol table for this scope
     pub symbol_table_index: usize,
-    // CPython compile.c uses PyList_GET_SIZE(u->u_ste->ste_varnames)
+    // compile.c uses PyList_GET_SIZE(u->u_ste->ste_varnames)
     // when calling flowgraph.c _PyCfg_OptimizeCodeUnit().
     pub nparams: usize,
 
@@ -3488,7 +3549,7 @@ impl CodeInfo {
     }
 
     fn prepare_cfg_from_codegen(&mut self) -> crate::InternalResult<InstructionSequence> {
-        // CPython compile.c optimize_and_assemble_code_unit passes
+        // compile.c optimize_and_assemble_code_unit passes
         // u_instr_sequence directly into flowgraph.c _PyCfg_FromInstructionSequence().
         self.take_recorded_instr_sequence()
     }
@@ -3509,12 +3570,12 @@ fn optimize_code_unit(
     optimize_cfg(metadata, blocks, metadata.firstlineno)?;
     blocks.remove_unused_consts(&mut metadata.consts)?;
     add_checks_for_loads_of_uninitialized_variables(blocks, nlocals, nparams)?;
-    // CPython inserts superinstructions in _PyCfg_OptimizeCodeUnit, before
+    // Superinstructions are inserted in _PyCfg_OptimizeCodeUnit, before
     // later jump normalization / block reordering can create adjacencies
     // that never exist at this stage in flowgraph.c.
     blocks.insert_superinstructions()?;
     blocks.push_cold_blocks_to_end()?;
-    // CPython resolves line numbers again after cold-block extraction.
+    // Line numbers are resolved again after cold-block extraction.
     blocks.resolve_line_numbers(metadata.firstlineno)?;
     Ok(())
 }
@@ -3525,20 +3586,20 @@ fn optimize_cfg(
     firstlineno: OneIndexed,
 ) -> crate::InternalResult<()> {
     // flowgraph.c optimize_cfg
-    // CPython optimize_cfg() starts with check_cfg() and raises
+    // optimize_cfg() starts with check_cfg() and raises
     // SystemError if a jump or scope exit is not the last instruction in
     // its block.
     blocks.check_cfg()?;
     blocks.inline_small_or_no_lineno_blocks()?;
-    // CPython does not re-run instruction-sequence label-map/CFG conversion
+    // The instruction-sequence label-map/CFG conversion is not re-run
     // after this point. Unreferenced label blocks left by jump inlining
     // remain block boundaries and can preserve line-marker NOPs.
     blocks.remove_unreachable()?;
-    // CPython optimize_cfg resolves line numbers before local checks and
+    // optimize_cfg resolves line numbers before local checks and
     // superinstruction insertion, so fusion decisions see propagated
     // source locations.
     blocks.resolve_line_numbers(firstlineno)?;
-    // CPython optimize_cfg() runs optimize_load_const() and then
+    // optimize_cfg() runs optimize_load_const() and then
     // optimize_basic_block() after line numbers are resolved.
     optimize_load_const(metadata, blocks)?;
     let mut block_idx = BlockIdx(0);
@@ -3548,7 +3609,7 @@ fn optimize_cfg(
         block_idx = next_block;
     }
     blocks.remove_redundant_nops_and_pairs()?;
-    // CPython optimize_cfg() removes newly-unreachable blocks and
+    // optimize_cfg() removes newly-unreachable blocks and
     // redundant NOP/jump chains before _PyCfg_OptimizeCodeUnit() prunes
     // unused constants.
     blocks.remove_unreachable()?;
@@ -3568,7 +3629,7 @@ fn optimized_cfg_to_instruction_sequence(
     let max_stackdepth = blocks.calculate_stackdepth()?;
     debug_assert!(!is_generator(flags) || max_stackdepth != 0);
     let nlocalsplus = prepare_localsplus(metadata, blocks, flags)?;
-    // Match CPython order: pseudo ops are lowered after stackdepth and
+    // Pseudo ops are lowered after stackdepth and
     // localsplus preparation, before normalize_jumps.
     convert_pseudo_ops(blocks)?;
     blocks.normalize_jumps()?;
@@ -3636,6 +3697,9 @@ impl CodeInfo {
             kwonlyargcount: kwonlyarg_count,
             firstlineno: first_line_number,
         } = metadata;
+        let code_arg_count = posonlyarg_count
+            .checked_add(arg_count)
+            .ok_or(InternalError::MalformedControlFlowGraph)?;
 
         resolve_unconditional_jumps(&mut instr_sequence)?;
         resolve_jump_offsets(&mut instr_sequence)?;
@@ -3653,7 +3717,7 @@ impl CodeInfo {
         Ok(CodeObject {
             flags,
             posonlyarg_count,
-            arg_count,
+            arg_count: code_arg_count,
             kwonlyarg_count,
             source_path,
             first_line_number: Some(first_line_number),
@@ -4109,8 +4173,13 @@ fn const_folding_safe_multiply(left: &ConstantData, right: &ConstantData) -> Opt
             const_folding_safe_multiply(right, left)
         }
         (ConstantData::Tuple { elements }, ConstantData::Integer { value: n }) => {
+            if elements.is_empty() {
+                return Some(ConstantData::Tuple {
+                    elements: Vec::new(),
+                });
+            }
             let n = n.to_usize()?;
-            if n != 0 && !elements.is_empty() {
+            if n != 0 {
                 if n > MAX_COLLECTION_SIZE / elements.len() {
                     return None;
                 }
@@ -4265,7 +4334,7 @@ fn eval_const_complex_binop(
         BinOp::Add => left + right,
         BinOp::Subtract => {
             let re = left.re - right.re;
-            // Preserve CPython's signed-zero behavior for real-zero
+            // Preserve signed-zero behavior for real-zero
             // minus zero-complex expressions such as `0 - 0j`.
             let im = if left.re == 0.0
                 && left.im == 0.0
@@ -4829,7 +4898,7 @@ fn fold_constant_intrinsic_list_to_tuple(
     Ok(false)
 }
 
-/// Port of CPython's flowgraph.c optimize_lists_and_sets().
+/// Port of flowgraph.c optimize_lists_and_sets().
 fn optimize_lists_and_sets(
     metadata: &mut CodeUnitMetadata,
     block: &mut Block,
@@ -4889,8 +4958,7 @@ fn optimize_lists_and_sets(
 
     if !contains_or_iter {
         debug_assert!(i >= 2);
-        let folded_loc = block.instructions[i].location;
-        let end_loc = block.instructions[i].end_location;
+        let folded_loc = instr_location(&block.instructions[i]);
 
         nop_out(block, &operand_indices);
 
@@ -4901,9 +4969,7 @@ fn optimize_lists_and_sets(
         }
         .into();
         instr_set_op1(&mut block.instructions[i - 2], build_instr, OpArg::new(0));
-        block.instructions[i - 2].location = folded_loc;
-        block.instructions[i - 2].end_location = end_loc;
-        block.instructions[i - 2].lineno_override = None;
+        instr_set_location(&mut block.instructions[i - 2], folded_loc);
 
         instr_set_op1(
             &mut block.instructions[i - 1],
@@ -5152,7 +5218,7 @@ fn basicblock_optimize_load_const(
     block: &mut Block,
 ) -> crate::InternalResult<()> {
     let mut i = 0;
-    let mut effective_opcode = None;
+    let mut effective_opcode = Instruction::Nop.into();
     let mut effective_oparg = OpArg::new(0);
     while i < block.instruction_used {
         if matches!(
@@ -5166,21 +5232,19 @@ fn basicblock_optimize_load_const(
         let curr = block.instructions[i];
         let curr_arg = curr.arg;
 
-        // Only combine if the source is a real instruction.
-        let Some(curr_instr) = curr.instr.real() else {
-            i += 1;
-            continue;
-        };
-
         let is_copy_of_load_const = matches!(
-            (effective_opcode, curr_instr),
-            (Some(Instruction::LoadConst { .. }), Instruction::Copy { i }) if i.get(curr_arg) == 1
+            (effective_opcode, curr.instr.real()),
+            (AnyInstruction::Real(Instruction::LoadConst { .. }), Some(Instruction::Copy { i }))
+                if i.get(curr_arg) == 1
         );
         if !is_copy_of_load_const {
-            effective_opcode = Some(curr_instr);
+            effective_opcode = curr.instr;
             effective_oparg = curr_arg;
         }
-        let Some(const_instr) = effective_opcode else {
+        debug_assert!(!effective_opcode.is_assembler());
+        let Some(const_instr @ (Instruction::LoadConst { .. } | Instruction::LoadSmallInt { .. })) =
+            effective_opcode.real()
+        else {
             i += 1;
             continue;
         };
@@ -5273,7 +5337,7 @@ fn basicblock_optimize_load_const(
                     Opcode::PopJumpIfNone
                 }
                 .into();
-                i = jump_idx;
+                i += 1;
                 continue;
             }
         }
@@ -6557,21 +6621,19 @@ fn get_max_label(blocks: &Blocks) -> i32 {
 }
 
 /// flowgraph.c make_except_stack
-#[allow(clippy::unnecessary_wraps)]
-fn make_except_stack() -> crate::InternalResult<CfgExceptStack> {
+fn make_except_stack() -> CfgExceptStack {
     let handlers = [BlockIdx::NULL; CO_MAXBLOCKS + 2];
     debug_assert_eq!(handlers[0], BlockIdx::NULL);
-    Ok(CfgExceptStack { handlers, depth: 0 })
+    CfgExceptStack { handlers, depth: 0 }
 }
 
 /// flowgraph.c copy_except_stack
-#[allow(clippy::unnecessary_wraps)]
-fn copy_except_stack(stack: &CfgExceptStack) -> crate::InternalResult<CfgExceptStack> {
+fn copy_except_stack(stack: &CfgExceptStack) -> CfgExceptStack {
     debug_assert!(stack.depth <= CO_MAXBLOCKS + 1);
-    Ok(CfgExceptStack {
+    CfgExceptStack {
         handlers: stack.handlers,
         depth: stack.depth,
-    })
+    }
 }
 
 /// flowgraph.c except_stack_top
@@ -6623,7 +6685,7 @@ pub(crate) fn label_exception_targets(blocks: &mut Blocks) -> crate::InternalRes
 
     todo.push(BlockIdx(0));
     blocks[0].visited = true;
-    blocks[0].except_stack = Some(make_except_stack()?);
+    blocks[0].except_stack = Some(make_except_stack());
 
     while let Some(block_idx) = todo.pop() {
         let bi = block_idx.idx();
@@ -6650,7 +6712,7 @@ pub(crate) fn label_exception_targets(blocks: &mut Blocks) -> crate::InternalRes
                 if !blocks[target].visited {
                     blocks[target].except_stack = Some(copy_except_stack(
                         stack.as_ref().expect("active exception stack"),
-                    )?);
+                    ));
                     todo.push(target);
                     blocks[target].visited = true;
                 }
@@ -6674,7 +6736,7 @@ pub(crate) fn label_exception_targets(blocks: &mut Blocks) -> crate::InternalRes
                     if bb_has_fallthrough(&blocks[bi]) {
                         blocks[target].except_stack = Some(copy_except_stack(
                             stack.as_ref().expect("active exception stack"),
-                        )?);
+                        ));
                     } else {
                         blocks[target].except_stack = stack.take();
                         stack_transferred = true;
@@ -6852,6 +6914,58 @@ pub(crate) fn fix_cell_offsets(
 #[cfg(test)]
 mod tests {
     use super::*;
+    use rustpython_compiler_core::bytecode::Arg;
+
+    fn int_const(value: i32) -> ConstantData {
+        ConstantData::Integer {
+            value: BigInt::from(value),
+        }
+    }
+
+    fn nan_const() -> ConstantData {
+        ConstantData::Float { value: f64::NAN }
+    }
+
+    #[test]
+    fn constant_pool_frozenset_key_ignores_order_and_duplicates_like_cpython() {
+        let mut pool = ConstantPool::default();
+        let (first, inserted) = pool.insert_full(ConstantData::Frozenset {
+            elements: vec![int_const(1), int_const(2)],
+        });
+        assert_eq!(first, 0);
+        assert!(inserted);
+
+        let (second, inserted) = pool.insert_full(ConstantData::Frozenset {
+            elements: vec![int_const(2), int_const(1), int_const(1)],
+        });
+        assert_eq!(
+            second, first,
+            "CPython _PyCode_ConstantKey uses frozenset item keys, not insertion order"
+        );
+        assert!(!inserted);
+        assert!(matches!(
+            &pool.constants[first],
+            ConstantData::Frozenset { elements } if elements.len() == 2
+        ));
+    }
+
+    #[test]
+    fn constant_pool_frozenset_key_preserves_nan_duplicates_like_cpython() {
+        let mut pool = ConstantPool::default();
+        let (idx, inserted) = pool.insert_full(ConstantData::Frozenset {
+            elements: vec![nan_const(), nan_const()],
+        });
+
+        assert_eq!(idx, 0);
+        assert!(inserted);
+        assert!(matches!(
+            &pool.constants[idx],
+            ConstantData::Frozenset { elements }
+                if elements.iter().filter(|constant| {
+                    matches!(constant, ConstantData::Float { value } if value.is_nan())
+                }).count() == 2
+        ));
+    }
 
     fn test_location(line: u32) -> SourceLocation {
         SourceLocation {
@@ -6886,6 +7000,13 @@ mod tests {
         instr
     }
 
+    fn test_true_cond_jump(target: BlockIdx, line: u32) -> InstructionInfo {
+        let mut instr = test_instr(Instruction::Nop, line);
+        instr.instr = PseudoOpcode::JumpIfTrue.into();
+        instr.target = target;
+        instr
+    }
+
     fn test_block_push(block: &mut Block, info: InstructionInfo) {
         let off = basicblock_next_instr(block).expect("test block instruction slot");
         block.instructions[off] = info;
@@ -6970,7 +7091,7 @@ mod tests {
 
     #[test]
     fn except_stack_tracks_cpython_depth_and_handler_slots() {
-        let mut stack = make_except_stack().unwrap();
+        let mut stack = make_except_stack();
         assert_eq!(stack.depth, 0);
         assert_eq!(stack.handlers.len(), CO_MAXBLOCKS + 2);
         assert_eq!(stack.handlers[0], BlockIdx::NULL);
@@ -6994,7 +7115,7 @@ mod tests {
         assert!(handler.preserve_lasti);
         assert!(blocks[1].preserve_lasti);
 
-        let copy = copy_except_stack(&stack).unwrap();
+        let copy = copy_except_stack(&stack);
         assert_eq!(copy.depth, stack.depth);
         assert_eq!(copy.handlers, stack.handlers);
 
@@ -7399,6 +7520,52 @@ mod tests {
         );
     }
 
+    #[test]
+    fn optimize_load_const_pseudo_opcode_breaks_effective_load_const() {
+        let mut block = Block::default();
+        test_block_push(
+            &mut block,
+            test_instr(
+                Instruction::LoadConst {
+                    consti: Arg::marker(),
+                },
+                90,
+            ),
+        );
+        test_block_push(&mut block, test_true_cond_jump(BlockIdx::new(0), 90));
+        let mut copy = test_instr(Instruction::Copy { i: Arg::marker() }, 90);
+        copy.arg = OpArg::new(1);
+        test_block_push(&mut block, copy);
+        test_block_push(&mut block, test_instr(Instruction::ToBool, 90));
+
+        let mut code = test_code_info(block);
+        let (const_idx, _) = code.metadata.consts.insert_full(ConstantData::Tuple {
+            elements: vec![ConstantData::Integer {
+                value: BigInt::from(1),
+            }],
+        });
+        code.blocks[0].instructions[0].arg = OpArg::new(const_idx as u32);
+
+        optimize_load_const(&mut code.metadata, &mut code.blocks)
+            .expect("optimize_load_const succeeds");
+
+        // `basicblock_optimize_load_const()` assigns the current
+        // pseudo opcode to its effective opcode slot, so the following COPY 1
+        // is not treated as a copy of the earlier LOAD_CONST.
+        assert!(matches!(
+            code.blocks[0].instructions[1].instr.pseudo(),
+            Some(PseudoInstruction::Jump { .. })
+        ));
+        assert!(matches!(
+            code.blocks[0].instructions[2].instr.real(),
+            Some(Instruction::Copy { .. })
+        ));
+        assert!(matches!(
+            code.blocks[0].instructions[3].instr.real(),
+            Some(Instruction::ToBool)
+        ));
+    }
+
     #[test]
     fn optimize_load_fast_records_no_input_opcode_ref_at_cpython_produced_index() {
         let mut block = Block::default();
@@ -7488,6 +7655,24 @@ mod tests {
         ));
     }
 
+    #[test]
+    fn empty_tuple_repeat_folds_negative_count_like_cpython() {
+        let folded = const_folding_safe_multiply(
+            &ConstantData::Tuple {
+                elements: Vec::new(),
+            },
+            &ConstantData::Integer {
+                value: BigInt::from(-1),
+            },
+        )
+        .expect("CPython skips repeat-count checks for empty tuples");
+
+        assert!(matches!(
+            folded,
+            ConstantData::Tuple { elements } if elements.is_empty()
+        ));
+    }
+
     #[test]
     fn resolve_line_numbers_duplicates_exit_blocks_like_cpython() {
         let exit = BlockIdx::new(2);
@@ -7621,4 +7806,51 @@ mod tests {
         assert_eq!(threaded.target, BlockIdx::new(3));
         assert_eq!(u32::from(threaded.arg), 3);
     }
+
+    #[test]
+    fn same_direction_pseudo_conditional_jump_thread_false_keeps_target() {
+        let mut blocks = Blocks::from([Block::default(), Block::default(), Block::default()]);
+        for (i, block) in blocks.iter_mut().enumerate() {
+            block.cpython_label = InstructionSequenceLabel::from_index(i as i32);
+        }
+        blocks[0].next = BlockIdx::new(1);
+        blocks[1].next = BlockIdx::new(2);
+        test_block_push(&mut blocks[0], test_cond_jump(BlockIdx::new(1), 10));
+        test_block_push(&mut blocks[1], test_cond_jump(BlockIdx::new(1), 20));
+        test_block_push(&mut blocks[2], test_instr(Instruction::ReturnValue, 30));
+
+        let mut metadata = test_code_info(Block::default()).metadata;
+        blocks
+            .optimize_basic_block(&mut metadata, BlockIdx::new(0))
+            .expect("valid conditional jump chain");
+
+        // Only rewrite JUMP_IF_FALSE -> JUMP_IF_TRUE through
+        // target->b_next. For same-direction jumps, a failed jump_thread()
+        // leaves the original target unchanged.
+        assert_eq!(blocks[0].instructions[0].target, BlockIdx::new(1));
+        assert!(matches!(
+            blocks[0].instructions[0].instr.pseudo(),
+            Some(PseudoInstruction::JumpIfFalse { .. })
+        ));
+    }
+
+    #[test]
+    fn opposite_direction_pseudo_conditional_uses_target_fallthrough() {
+        let mut blocks = Blocks::from([Block::default(), Block::default(), Block::default()]);
+        for (i, block) in blocks.iter_mut().enumerate() {
+            block.cpython_label = InstructionSequenceLabel::from_index(i as i32);
+        }
+        blocks[0].next = BlockIdx::new(1);
+        blocks[1].next = BlockIdx::new(2);
+        test_block_push(&mut blocks[0], test_cond_jump(BlockIdx::new(1), 10));
+        test_block_push(&mut blocks[1], test_true_cond_jump(BlockIdx::new(2), 20));
+        test_block_push(&mut blocks[2], test_instr(Instruction::ReturnValue, 30));
+
+        let mut metadata = test_code_info(Block::default()).metadata;
+        blocks
+            .optimize_basic_block(&mut metadata, BlockIdx::new(0))
+            .expect("valid conditional jump chain");
+
+        assert_eq!(blocks[0].instructions[0].target, BlockIdx::new(2));
+    }
 }

crates/codegen/src/lib.rs

@@ -8,13 +8,15 @@ extern crate log;
 
 extern crate alloc;
 
+use rustpython_compiler_core::bytecode::ConstantData;
+
 type IndexMap<K, V> = indexmap::IndexMap<K, V, rapidhash::quality::RandomState>;
 type IndexSet<T> = indexmap::IndexSet<T, rapidhash::quality::RandomState>;
 
 pub mod compile;
 pub mod error;
 pub mod ir;
-mod preprocess;
+pub mod preprocess;
 mod string_parser;
 pub mod symboltable;
 mod unparse;
@@ -24,6 +26,73 @@ use ruff_python_ast as ast;
 
 pub(crate) use compile::InternalResult;
 
+#[cfg(test)]
+pub(crate) fn constant_data_to_ast_constant_value(value: ConstantData) -> ast::ConstantValue {
+    match value {
+        ConstantData::None => ast::ConstantValue::None,
+        ConstantData::Boolean { value } => ast::ConstantValue::Boolean(value),
+        ConstantData::Str { value } => ast::ConstantValue::Str(value.to_string().into_boxed_str()),
+        ConstantData::Bytes { value } => ast::ConstantValue::Bytes(value.into_boxed_slice()),
+        ConstantData::Integer { value } => ast::ConstantValue::Integer(value.to_string().into()),
+        ConstantData::Tuple { elements } => ast::ConstantValue::Tuple(
+            elements
+                .into_iter()
+                .map(constant_data_to_ast_constant_value)
+                .collect(),
+        ),
+        ConstantData::Frozenset { elements } => ast::ConstantValue::Frozenset(
+            elements
+                .into_iter()
+                .map(constant_data_to_ast_constant_value)
+                .collect(),
+        ),
+        ConstantData::Float { value } => ast::ConstantValue::Float(value),
+        ConstantData::Complex { value } => ast::ConstantValue::Complex {
+            real: value.re,
+            imag: value.im,
+        },
+        ConstantData::Ellipsis => ast::ConstantValue::Ellipsis,
+        ConstantData::Code { .. } | ConstantData::Slice { .. } => {
+            unreachable!("ast.Constant values cannot contain code objects or slices")
+        }
+    }
+}
+
+pub(crate) fn ast_constant_value_to_constant_data(value: ast::ConstantValue) -> ConstantData {
+    match value {
+        ast::ConstantValue::None => ConstantData::None,
+        ast::ConstantValue::Boolean(value) => ConstantData::Boolean { value },
+        ast::ConstantValue::Str(value) => ConstantData::Str {
+            value: value.to_string().into(),
+        },
+        ast::ConstantValue::Bytes(value) => ConstantData::Bytes {
+            value: value.into_vec(),
+        },
+        ast::ConstantValue::Integer(value) => ConstantData::Integer {
+            value: value
+                .parse()
+                .expect("RustPython ast.Constant integer values are decimal integers"),
+        },
+        ast::ConstantValue::Tuple(elements) => ConstantData::Tuple {
+            elements: elements
+                .into_iter()
+                .map(ast_constant_value_to_constant_data)
+                .collect(),
+        },
+        ast::ConstantValue::Frozenset(elements) => ConstantData::Frozenset {
+            elements: elements
+                .into_iter()
+                .map(ast_constant_value_to_constant_data)
+                .collect(),
+        },
+        ast::ConstantValue::Float(value) => ConstantData::Float { value },
+        ast::ConstantValue::Complex { real, imag } => ConstantData::Complex {
+            value: num_complex::Complex::new(real, imag),
+        },
+        ast::ConstantValue::Ellipsis => ConstantData::Ellipsis,
+    }
+}
+
 pub trait ToPythonName {
     /// Returns a short name for the node suitable for use in error messages.
     fn python_name(&self) -> &'static str;
@@ -48,6 +117,19 @@ impl ToPythonName for ast::Expr {
             }
             Self::EllipsisLiteral(_) => "ellipsis",
             Self::NoneLiteral(_) => "None",
+            Self::Constant(expr) => match &expr.value {
+                ast::ConstantValue::None => "None",
+                ast::ConstantValue::Boolean(true) => "True",
+                ast::ConstantValue::Boolean(false) => "False",
+                ast::ConstantValue::Ellipsis => "ellipsis",
+                ast::ConstantValue::Tuple(_) => "tuple",
+                ast::ConstantValue::Frozenset(_) => "literal",
+                ast::ConstantValue::Str(_)
+                | ast::ConstantValue::Bytes(_)
+                | ast::ConstantValue::Integer(_)
+                | ast::ConstantValue::Float(_)
+                | ast::ConstantValue::Complex { .. } => "literal",
+            },
             Self::NumberLiteral(_) | Self::BytesLiteral(_) | Self::StringLiteral(_) => "literal",
             Self::Tuple(_) => "tuple",
             Self::List { .. } => "list",
@@ -65,7 +147,7 @@ impl ToPythonName for ast::Expr {
             Self::Lambda { .. } => "lambda",
             Self::If { .. } => "conditional expression",
             Self::Named { .. } => "named expression",
-            Self::IpyEscapeCommand(_) => todo!(),
+            Self::IpyEscapeCommand(_) => "expression",
         }
     }
 }

crates/codegen/src/preprocess.rs

@@ -8,29 +8,401 @@ use ruff_python_ast::{
     visitor::transformer::{self, Transformer},
 };
 use ruff_text_size::{Ranged, TextRange};
+use rustpython_compiler_core::bytecode;
 
 const MAXDIGITS: usize = 3;
 const F_LJUST: u8 = 1;
 
-pub(crate) fn preprocess_mod(module: &mut ast::Mod) {
-    let preprocessor = AstPreprocessor;
+/// ast_preprocess.c ControlFlowInFinallyContext
+#[derive(Clone, Copy)]
+struct ControlFlowInFinallyContext {
+    in_finally: bool,
+    in_funcdef: bool,
+    in_loop: bool,
+}
+
+/// ast_preprocess.c before_return
+fn before_return<E>(
+    contexts: &[ControlFlowInFinallyContext],
+    range: TextRange,
+    warn: &mut impl FnMut(TextRange, String) -> Result<(), E>,
+) -> Result<(), E> {
+    if let Some(ctx) = contexts.last()
+        && ctx.in_finally
+        && !ctx.in_funcdef
+    {
+        warn(range, "'return' in a 'finally' block".to_owned())?;
+    }
+    Ok(())
+}
+
+/// ast_preprocess.c before_loop_exit
+fn before_loop_exit<E>(
+    contexts: &[ControlFlowInFinallyContext],
+    range: TextRange,
+    kw: &str,
+    warn: &mut impl FnMut(TextRange, String) -> Result<(), E>,
+) -> Result<(), E> {
+    if let Some(ctx) = contexts.last()
+        && ctx.in_finally
+        && !ctx.in_loop
+    {
+        warn(range, format!("'{kw}' in a 'finally' block"))?;
+    }
+    Ok(())
+}
+
+fn visit_body_with_control_flow_context<E>(
+    body: &[ast::Stmt],
+    contexts: &mut Vec<ControlFlowInFinallyContext>,
+    warn: &mut impl FnMut(TextRange, String) -> Result<(), E>,
+    in_finally: bool,
+    in_funcdef: bool,
+    in_loop: bool,
+) -> Result<(), E> {
+    contexts.push(ControlFlowInFinallyContext {
+        in_finally,
+        in_funcdef,
+        in_loop,
+    });
+    visit_body_for_control_flow_in_finally(body, contexts, warn)?;
+    contexts.pop();
+    Ok(())
+}
+
+fn visit_body_for_control_flow_in_finally<E>(
+    body: &[ast::Stmt],
+    contexts: &mut Vec<ControlFlowInFinallyContext>,
+    warn: &mut impl FnMut(TextRange, String) -> Result<(), E>,
+) -> Result<(), E> {
+    for stmt in body {
+        visit_stmt_for_control_flow_in_finally(stmt, contexts, warn)?;
+    }
+    Ok(())
+}
+
+/// ast_preprocess.c astfold_stmt control-flow warning traversal.
+fn visit_stmt_for_control_flow_in_finally<E>(
+    stmt: &ast::Stmt,
+    contexts: &mut Vec<ControlFlowInFinallyContext>,
+    warn: &mut impl FnMut(TextRange, String) -> Result<(), E>,
+) -> Result<(), E> {
+    match stmt {
+        ast::Stmt::FunctionDef(function) => {
+            visit_body_with_control_flow_context(
+                &function.body,
+                contexts,
+                warn,
+                false,
+                true,
+                false,
+            )?;
+        }
+        ast::Stmt::ClassDef(class) => {
+            visit_body_for_control_flow_in_finally(&class.body, contexts, warn)?;
+        }
+        ast::Stmt::Return(return_stmt) => {
+            before_return(contexts, return_stmt.range, warn)?;
+        }
+        ast::Stmt::For(for_stmt) => {
+            visit_body_with_control_flow_context(
+                &for_stmt.body,
+                contexts,
+                warn,
+                false,
+                false,
+                true,
+            )?;
+            visit_body_for_control_flow_in_finally(&for_stmt.orelse, contexts, warn)?;
+        }
+        ast::Stmt::While(while_stmt) => {
+            visit_body_with_control_flow_context(
+                &while_stmt.body,
+                contexts,
+                warn,
+                false,
+                false,
+                true,
+            )?;
+            visit_body_for_control_flow_in_finally(&while_stmt.orelse, contexts, warn)?;
+        }
+        ast::Stmt::If(if_stmt) => {
+            visit_body_for_control_flow_in_finally(&if_stmt.body, contexts, warn)?;
+            for clause in &if_stmt.elif_else_clauses {
+                visit_body_for_control_flow_in_finally(&clause.body, contexts, warn)?;
+            }
+        }
+        ast::Stmt::Try(try_stmt) => {
+            visit_body_for_control_flow_in_finally(&try_stmt.body, contexts, warn)?;
+            for handler in &try_stmt.handlers {
+                match handler {
+                    ast::ExceptHandler::ExceptHandler(handler) => {
+                        visit_body_for_control_flow_in_finally(&handler.body, contexts, warn)?;
+                    }
+                }
+            }
+            visit_body_for_control_flow_in_finally(&try_stmt.orelse, contexts, warn)?;
+            visit_body_with_control_flow_context(
+                &try_stmt.finalbody,
+                contexts,
+                warn,
+                true,
+                false,
+                false,
+            )?;
+        }
+        ast::Stmt::With(with_stmt) => {
+            visit_body_for_control_flow_in_finally(&with_stmt.body, contexts, warn)?;
+        }
+        ast::Stmt::Match(match_stmt) => {
+            for case in &match_stmt.cases {
+                visit_body_for_control_flow_in_finally(&case.body, contexts, warn)?;
+            }
+        }
+        ast::Stmt::Break(break_stmt) => {
+            before_loop_exit(contexts, break_stmt.range, "break", warn)?;
+        }
+        ast::Stmt::Continue(continue_stmt) => {
+            before_loop_exit(contexts, continue_stmt.range, "continue", warn)?;
+        }
+        _ => {}
+    }
+    Ok(())
+}
+
+/// ast_preprocess.c control_flow_in_finally_warning
+pub fn warn_control_flow_in_finally<E>(
+    module: &ast::Mod,
+    mut warn: impl FnMut(TextRange, String) -> Result<(), E>,
+) -> Result<(), E> {
+    let mut contexts = Vec::new();
     match module {
-        ast::Mod::Module(module) => preprocessor.visit_body(&mut module.body),
+        ast::Mod::Module(module) => {
+            visit_body_for_control_flow_in_finally(&module.body, &mut contexts, &mut warn)?;
+        }
+        ast::Mod::Expression(_) => {}
+    }
+    Ok(())
+}
+
+pub fn has_future_annotations(module: &ast::Mod) -> bool {
+    future_features(module).contains(bytecode::CodeFlags::FUTURE_ANNOTATIONS)
+}
+
+pub fn future_features(module: &ast::Mod) -> bytecode::CodeFlags {
+    checked_future_features(module).unwrap_or_else(|err| err.features)
+}
+
+pub struct FutureFeatureError {
+    pub features: bytecode::CodeFlags,
+    pub range: TextRange,
+    pub kind: FutureFeatureErrorKind,
+}
+
+pub enum FutureFeatureErrorKind {
+    InvalidFeature(String),
+    InvalidBraces,
+}
+
+pub fn checked_future_features(
+    module: &ast::Mod,
+) -> Result<bytecode::CodeFlags, FutureFeatureError> {
+    let ast::Mod::Module(module) = module else {
+        return Ok(bytecode::CodeFlags::empty());
+    };
+    checked_future_features_in_body(&module.body)
+}
+
+pub fn checked_future_features_in_body(
+    body: &[ast::Stmt],
+) -> Result<bytecode::CodeFlags, FutureFeatureError> {
+    let mut future_features = bytecode::CodeFlags::empty();
+    let mut statements = body.iter();
+    if let Some(ast::Stmt::Expr(ast::StmtExpr { value, .. })) = statements.clone().next()
+        && string_literal_expr_value(value).is_some()
+    {
+        statements.next();
+    }
+    for statement in statements {
+        match statement {
+            ast::Stmt::ImportFrom(ast::StmtImportFrom {
+                module,
+                names,
+                level,
+                ..
+            }) if *level == 0 && module.as_ref().map(|id| id.as_str()) == Some("__future__") => {
+                for alias in names {
+                    match alias.name.as_str() {
+                        "nested_scopes" | "generators" | "division" | "absolute_import"
+                        | "with_statement" | "print_function" | "unicode_literals"
+                        | "generator_stop" => {}
+                        "annotations" => {
+                            future_features.insert(bytecode::CodeFlags::FUTURE_ANNOTATIONS);
+                        }
+                        // Accept the future feature name, but leave it
+                        // as a RustPython no-op.
+                        "barry_as_FLUFL" => {}
+                        "braces" => {
+                            return Err(FutureFeatureError {
+                                features: future_features,
+                                range: alias.range,
+                                kind: FutureFeatureErrorKind::InvalidBraces,
+                            });
+                        }
+                        other => {
+                            return Err(FutureFeatureError {
+                                features: future_features,
+                                range: alias.range,
+                                kind: FutureFeatureErrorKind::InvalidFeature(other.to_owned()),
+                            });
+                        }
+                    }
+                }
+            }
+            _ => return Ok(future_features),
+        }
+    }
+    Ok(future_features)
+}
+
+pub fn preprocess_statements(
+    body: &mut [ast::Stmt],
+    optimize: u8,
+    future_annotations: bool,
+    syntax_check_only: bool,
+) {
+    let preprocessor = AstPreprocessor {
+        optimize,
+        future_annotations,
+        constant_folding: !syntax_check_only,
+    };
+    for stmt in body {
+        preprocessor.visit_stmt(stmt);
+    }
+}
+
+pub fn preprocess_mod(
+    module: &mut ast::Mod,
+    optimize: u8,
+    future_annotations: bool,
+    syntax_check_only: bool,
+) {
+    let preprocessor = AstPreprocessor {
+        optimize,
+        future_annotations,
+        constant_folding: !syntax_check_only,
+    };
+    match module {
+        ast::Mod::Module(module) => preprocessor.visit_astfold_body(&mut module.body),
         ast::Mod::Expression(expr) => preprocessor.visit_expr(&mut expr.body),
     }
 }
 
-struct AstPreprocessor;
+struct AstPreprocessor {
+    optimize: u8,
+    future_annotations: bool,
+    constant_folding: bool,
+}
 
-impl Transformer for AstPreprocessor {
-    fn visit_expr(&self, expr: &mut Expr) {
-        transformer::walk_expr(self, expr);
-        if let Some(optimized) = optimize_format(expr) {
-            *expr = optimized;
+impl AstPreprocessor {
+    fn visit_astfold_body(&self, body: &mut ast::Suite) {
+        let mut docstring = body_starts_with_docstring(body);
+        if docstring && self.optimize >= 2 {
+            remove_docstring_from_body(body);
+            docstring = false;
+        }
+
+        for stmt in body.iter_mut() {
+            self.visit_stmt(stmt);
+        }
+
+        if !docstring && body_starts_with_docstring(body) {
+            wrap_first_docstring_as_fstring(body);
         }
     }
 }
 
+impl Transformer for AstPreprocessor {
+    fn visit_stmt(&self, stmt: &mut ast::Stmt) {
+        match stmt {
+            ast::Stmt::FunctionDef(function) => {
+                if let Some(type_params) = &mut function.type_params {
+                    self.visit_type_params(type_params);
+                }
+                self.visit_parameters(&mut function.parameters);
+                self.visit_astfold_body(&mut function.body);
+                for decorator in &mut function.decorator_list {
+                    self.visit_decorator(decorator);
+                }
+                if let Some(returns) = &mut function.returns {
+                    self.visit_annotation(returns);
+                }
+            }
+            ast::Stmt::ClassDef(class) => {
+                if let Some(type_params) = &mut class.type_params {
+                    self.visit_type_params(type_params);
+                }
+                if let Some(arguments) = &mut class.arguments {
+                    self.visit_arguments(arguments);
+                }
+                self.visit_astfold_body(&mut class.body);
+                for decorator in &mut class.decorator_list {
+                    self.visit_decorator(decorator);
+                }
+            }
+            _ => transformer::walk_stmt(self, stmt),
+        }
+    }
+
+    fn visit_annotation(&self, expr: &mut Expr) {
+        if !self.future_annotations {
+            transformer::walk_annotation(self, expr);
+        }
+    }
+
+    fn visit_pattern(&self, pattern: &mut ast::Pattern) {
+        transformer::walk_pattern(self, pattern);
+        if !self.constant_folding {
+            return;
+        }
+        match pattern {
+            ast::Pattern::MatchValue(value) => fold_match_value_constant_expr(&mut value.value),
+            ast::Pattern::MatchMapping(mapping) => {
+                for key in &mut mapping.keys {
+                    fold_match_value_constant_expr(key);
+                }
+            }
+            _ => {}
+        }
+    }
+
+    fn visit_expr(&self, expr: &mut Expr) {
+        transformer::walk_expr(self, expr);
+        if self.constant_folding {
+            if let Some(optimized) = optimize_format(expr) {
+                *expr = optimized;
+            } else if let Some(optimized) = fold_debug_constant(expr, self.optimize) {
+                *expr = optimized;
+            }
+        }
+    }
+}
+
+fn fold_debug_constant(expr: &Expr, optimize: u8) -> Option<Expr> {
+    let Expr::Name(name) = expr else {
+        return None;
+    };
+    if !matches!(name.ctx, ast::ExprContext::Load) || name.id.as_str() != "__debug__" {
+        return None;
+    }
+
+    Some(Expr::BooleanLiteral(ast::ExprBooleanLiteral {
+        node_index: name.node_index.clone(),
+        range: name.range,
+        value: optimize == 0,
+    }))
+}
+
 fn optimize_format(expr: &Expr) -> Option<Expr> {
     let Expr::BinOp(binop) = expr else {
         return None;
@@ -38,9 +410,7 @@ fn optimize_format(expr: &Expr) -> Option<Expr> {
     if !matches!(binop.op, Operator::Mod) {
         return None;
     }
-    let Expr::StringLiteral(format) = binop.left.as_ref() else {
-        return None;
-    };
+    let (format, _) = string_literal_expr_value(&binop.left)?;
     let Expr::Tuple(tuple) = binop.right.as_ref() else {
         return None;
     };
@@ -52,7 +422,7 @@ fn optimize_format(expr: &Expr) -> Option<Expr> {
         return None;
     }
 
-    let elements = parse_format(format.value.to_str(), &tuple.elts)?;
+    let elements = parse_format(format, &tuple.elts)?;
     Some(Expr::FString(ExprFString {
         node_index: binop.node_index.clone(),
         range: binop.range,
@@ -62,6 +432,8 @@ fn optimize_format(expr: &Expr) -> Option<Expr> {
             elements: InterpolatedStringElements::from(elements),
             flags: FStringFlags::empty(),
         }),
+        runtime_joined_str: None,
+        runtime_values: None,
     }))
 }
 
@@ -163,6 +535,9 @@ fn parse_format_arg(chars: &[char], pos: &mut usize, arg: Expr) -> Option<Interp
         debug_text: None,
         conversion,
         format_spec,
+        runtime_str: None,
+        runtime_interpolation_format_spec: None,
+        runtime_formatted_value_format_spec: None,
     })
 }
 
@@ -223,3 +598,396 @@ fn generated_literal(value: String) -> InterpolatedStringLiteralElement {
         value: value.into_boxed_str(),
     }
 }
+
+fn remove_docstring_from_body(body: &mut ast::Suite) {
+    if let Some(range) = take_docstring(body) {
+        if !body.is_empty() {
+            return;
+        }
+        let start = range.start();
+        let pass_range = TextRange::new(start, start + ruff_text_size::TextSize::from(4));
+        body.push(ast::Stmt::Pass(ast::StmtPass {
+            node_index: Default::default(),
+            range: pass_range,
+        }));
+    }
+}
+
+fn take_docstring(body: &mut ast::Suite) -> Option<TextRange> {
+    let ast::Stmt::Expr(expr_stmt) = body.first()? else {
+        return None;
+    };
+    if let Some((_, range)) = string_literal_expr_value(&expr_stmt.value) {
+        body.remove(0);
+        return Some(range);
+    }
+    None
+}
+
+fn body_starts_with_docstring(body: &[ast::Stmt]) -> bool {
+    let Some(ast::Stmt::Expr(expr_stmt)) = body.first() else {
+        return false;
+    };
+    string_literal_expr_value(&expr_stmt.value).is_some()
+}
+
+fn wrap_first_docstring_as_fstring(body: &mut [ast::Stmt]) {
+    let Some(ast::Stmt::Expr(expr_stmt)) = body.first_mut() else {
+        return;
+    };
+    let Some((value, range)) = string_literal_expr_value(&expr_stmt.value) else {
+        return;
+    };
+    let value = value.to_string();
+    *expr_stmt.value = ast::Expr::FString(ast::ExprFString {
+        node_index: AtomicNodeIndex::NONE,
+        range,
+        value: FStringValue::single(FString {
+            range,
+            node_index: AtomicNodeIndex::NONE,
+            elements: InterpolatedStringElements::from(vec![InterpolatedStringElement::Literal(
+                InterpolatedStringLiteralElement {
+                    range,
+                    node_index: AtomicNodeIndex::NONE,
+                    value: value.into_boxed_str(),
+                },
+            )]),
+            flags: FStringFlags::empty(),
+        }),
+        runtime_joined_str: None,
+        runtime_values: None,
+    });
+}
+
+fn string_literal_expr_value(expr: &Expr) -> Option<(&str, TextRange)> {
+    match expr {
+        Expr::StringLiteral(string) => Some((string.value.to_str(), expr.range())),
+        Expr::Constant(ast::ExprConstant {
+            value: ast::ConstantValue::Str(value),
+            ..
+        }) => Some((value.as_ref(), expr.range())),
+        _ => None,
+    }
+}
+
+fn fold_match_value_constant_expr(expr: &mut ast::Expr) {
+    match expr {
+        ast::Expr::UnaryOp(unary)
+            if matches!(unary.op, ast::UnaryOp::USub)
+                && matches!(unary.operand.as_ref(), ast::Expr::NumberLiteral(_)) =>
+        {
+            if let Some(number) = negate_match_number(&unary.operand) {
+                *expr = ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
+                    node_index: unary.node_index.clone(),
+                    range: unary.range,
+                    value: number,
+                });
+            }
+        }
+        ast::Expr::BinOp(binop) if matches!(binop.op, ast::Operator::Add | ast::Operator::Sub) => {
+            fold_match_value_constant_expr(&mut binop.left);
+            if let Some(number) = fold_match_number_binop(&binop.left, binop.op, &binop.right) {
+                *expr = ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
+                    node_index: binop.node_index.clone(),
+                    range: binop.range,
+                    value: number,
+                });
+            }
+        }
+        _ => {}
+    }
+}
+
+fn negate_match_number(expr: &ast::Expr) -> Option<ast::Number> {
+    let ast::Expr::NumberLiteral(number) = expr else {
+        return None;
+    };
+    Some(match &number.value {
+        ast::Number::Int(value) => {
+            if *value == ast::Int::ZERO {
+                ast::Number::Int(ast::Int::ZERO)
+            } else {
+                return None;
+            }
+        }
+        ast::Number::Float(value) => ast::Number::Float(-value),
+        ast::Number::Complex { real, imag } => ast::Number::Complex {
+            real: -real,
+            imag: -imag,
+        },
+    })
+}
+
+fn fold_match_number_binop(
+    left: &ast::Expr,
+    op: ast::Operator,
+    right: &ast::Expr,
+) -> Option<ast::Number> {
+    let ast::Expr::NumberLiteral(left) = left else {
+        return None;
+    };
+    let ast::Expr::NumberLiteral(right) = right else {
+        return None;
+    };
+    let right = match right.value {
+        ast::Number::Complex { real, imag } => (real, imag),
+        _ => return None,
+    };
+    enum MatchNumberLeft {
+        Real(f64),
+        Complex { real: f64, imag: f64 },
+    }
+    let left = match &left.value {
+        ast::Number::Int(value) => MatchNumberLeft::Real(value.as_i64()? as f64),
+        ast::Number::Float(value) => MatchNumberLeft::Real(*value),
+        ast::Number::Complex { real, imag } => MatchNumberLeft::Complex {
+            real: *real,
+            imag: *imag,
+        },
+    };
+    let (real, imag) = match (left, op) {
+        (MatchNumberLeft::Real(left), ast::Operator::Add) => (left + right.0, right.1),
+        (MatchNumberLeft::Real(left), ast::Operator::Sub) => (left - right.0, -right.1),
+        (MatchNumberLeft::Complex { real, imag }, ast::Operator::Add) => {
+            (real + right.0, imag + right.1)
+        }
+        (MatchNumberLeft::Complex { real, imag }, ast::Operator::Sub) => {
+            (real - right.0, imag - right.1)
+        }
+        _ => return None,
+    };
+    Some(ast::Number::Complex { real, imag })
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn first_match_value(source: &str) -> ast::Expr {
+        let parsed = ruff_python_parser::parse(source, ruff_python_parser::Mode::Module.into())
+            .unwrap()
+            .into_syntax();
+        let mut module = parsed;
+        let future_annotations = has_future_annotations(&module);
+        preprocess_mod(&mut module, 0, future_annotations, false);
+        let ast::Mod::Module(module) = module else {
+            panic!("expected module");
+        };
+        let [ast::Stmt::Match(match_stmt)] = &module.body[..] else {
+            panic!("expected a single match statement");
+        };
+        let ast::Pattern::MatchValue(value) = &match_stmt.cases[0].pattern else {
+            panic!("expected a value pattern");
+        };
+        *value.value.clone()
+    }
+
+    fn preprocess_source(source: &str) -> ast::Mod {
+        let mut module = ruff_python_parser::parse(source, ruff_python_parser::Mode::Module.into())
+            .unwrap()
+            .into_syntax();
+        let future_annotations = has_future_annotations(&module);
+        preprocess_mod(&mut module, 0, future_annotations, false);
+        module
+    }
+
+    fn preprocess_source_with_optimize(source: &str, optimize: u8) -> ast::Mod {
+        let mut module = ruff_python_parser::parse(source, ruff_python_parser::Mode::Module.into())
+            .unwrap()
+            .into_syntax();
+        let future_annotations = has_future_annotations(&module);
+        preprocess_mod(&mut module, optimize, future_annotations, false);
+        module
+    }
+
+    fn preprocess_source_syntax_check_only(source: &str, optimize: u8) -> ast::Mod {
+        let mut module = ruff_python_parser::parse(source, ruff_python_parser::Mode::Module.into())
+            .unwrap()
+            .into_syntax();
+        let future_annotations = has_future_annotations(&module);
+        preprocess_mod(&mut module, optimize, future_annotations, true);
+        module
+    }
+
+    #[test]
+    fn folds_match_value_negative_float_in_preprocess() {
+        let value = first_match_value(
+            "\
+match value:
+    case -1.5:
+        pass
+",
+        );
+        let ast::Expr::NumberLiteral(number) = value else {
+            panic!("expected folded number literal, got {value:?}");
+        };
+        assert!(matches!(number.value, ast::Number::Float(value) if value == -1.5));
+    }
+
+    #[test]
+    fn folds_match_value_complex_binop_in_preprocess() {
+        let value = first_match_value(
+            "\
+match value:
+    case 1 + 2j:
+        pass
+",
+        );
+        let ast::Expr::NumberLiteral(number) = value else {
+            panic!("expected folded number literal, got {value:?}");
+        };
+        assert!(
+            matches!(number.value, ast::Number::Complex { real, imag } if real == 1.0 && imag == 2.0)
+        );
+    }
+
+    #[test]
+    fn folds_match_value_complex_complex_binop_in_preprocess() {
+        let left = ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
+            node_index: AtomicNodeIndex::NONE,
+            range: TextRange::default(),
+            value: ast::Number::Complex {
+                real: 0.0,
+                imag: 1.0,
+            },
+        });
+        let right = ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
+            node_index: AtomicNodeIndex::NONE,
+            range: TextRange::default(),
+            value: ast::Number::Complex {
+                real: 0.0,
+                imag: 2.0,
+            },
+        });
+        let number = fold_match_number_binop(&left, ast::Operator::Add, &right)
+            .expect("CPython fold_const_match_patterns() uses PyNumber_Add");
+        assert!(
+            matches!(number, ast::Number::Complex { real, imag } if real == 0.0 && imag == 3.0)
+        );
+    }
+
+    #[test]
+    fn folds_match_value_real_minus_zero_complex_preserves_negative_zero_in_preprocess() {
+        let value = first_match_value(
+            "\
+match value:
+    case 0 - 0j:
+        pass
+",
+        );
+        let ast::Expr::NumberLiteral(number) = value else {
+            panic!("expected folded number literal, got {value:?}");
+        };
+        assert!(matches!(number.value, ast::Number::Complex { real, imag }
+                if real == 0.0 && imag == 0.0 && imag.is_sign_negative()));
+    }
+
+    #[test]
+    fn future_annotations_skip_annotation_preprocess_like_cpython() {
+        let module = preprocess_source(
+            "\
+from __future__ import annotations
+def f(x: __debug__) -> __debug__:
+    pass
+y: __debug__
+z = __debug__
+",
+        );
+        let ast::Mod::Module(module) = module else {
+            panic!("expected module");
+        };
+        let ast::Stmt::FunctionDef(function) = &module.body[1] else {
+            panic!("expected function");
+        };
+        let annotation = function.parameters.args[0]
+            .parameter
+            .annotation
+            .as_deref()
+            .expect("missing parameter annotation");
+        assert!(
+            matches!(annotation, ast::Expr::Name(name) if name.id.as_str() == "__debug__"),
+            "future annotations should skip parameter annotation folding, got {annotation:?}"
+        );
+        let returns = function
+            .returns
+            .as_deref()
+            .expect("missing return annotation");
+        assert!(
+            matches!(returns, ast::Expr::Name(name) if name.id.as_str() == "__debug__"),
+            "future annotations should skip return annotation folding, got {returns:?}"
+        );
+        let ast::Stmt::AnnAssign(ann_assign) = &module.body[2] else {
+            panic!("expected annotated assignment");
+        };
+        assert!(
+            matches!(ann_assign.annotation.as_ref(), ast::Expr::Name(name) if name.id.as_str() == "__debug__"),
+            "future annotations should skip annotated assignment annotation folding, got {:?}",
+            ann_assign.annotation
+        );
+        let ast::Stmt::Assign(assign) = &module.body[3] else {
+            panic!("expected assignment");
+        };
+        assert!(
+            matches!(assign.value.as_ref(), ast::Expr::BooleanLiteral(boolean) if boolean.value),
+            "non-annotation expression should still fold __debug__, got {:?}",
+            assign.value
+        );
+    }
+
+    #[test]
+    fn late_future_annotations_do_not_affect_preprocess_like_cpython() {
+        let module = preprocess_source(
+            "\
+x = 1
+from __future__ import annotations
+y: __debug__
+",
+        );
+        let ast::Mod::Module(module) = module else {
+            panic!("expected module");
+        };
+        let ast::Stmt::AnnAssign(ann_assign) = &module.body[2] else {
+            panic!("expected annotated assignment");
+        };
+        assert!(
+            matches!(ann_assign.annotation.as_ref(), ast::Expr::BooleanLiteral(boolean) if boolean.value),
+            "late future import should not disable annotation folding, got {:?}",
+            ann_assign.annotation
+        );
+    }
+
+    #[test]
+    fn optimize_two_wraps_new_docstring_after_removing_original() {
+        let module = preprocess_source_with_optimize("\"first\"\n\"second\"\n", 2);
+        let ast::Mod::Module(module) = module else {
+            panic!("expected module");
+        };
+        let [ast::Stmt::Expr(expr)] = &module.body[..] else {
+            panic!("expected only the second statement to remain");
+        };
+        assert!(
+            matches!(expr.value.as_ref(), ast::Expr::FString(_)),
+            "CPython wraps the new leading string as JoinedStr so it is not a docstring"
+        );
+    }
+
+    #[test]
+    fn syntax_check_only_disables_constant_folding_but_keeps_docstring_strip() {
+        let module = preprocess_source_syntax_check_only("\"doc\"\nvalue = __debug__\n", 2);
+        let ast::Mod::Module(module) = module else {
+            panic!("expected module");
+        };
+        assert!(
+            matches!(module.body[0], ast::Stmt::Assign(_)),
+            "optimize=2 should still strip docstrings in syntax_check_only mode"
+        );
+        let ast::Stmt::Assign(assign) = &module.body[0] else {
+            panic!("expected assignment");
+        };
+        assert!(
+            matches!(assign.value.as_ref(), ast::Expr::Name(name) if name.id.as_str() == "__debug__"),
+            "syntax_check_only should skip __debug__ folding, got {:?}",
+            assign.value
+        );
+    }
+}

crates/codegen/src/unparse.rs

@@ -58,6 +58,63 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
         self.f.write_fmt(f)
     }
 
+    fn unparse_float(&mut self, value: f64) -> fmt::Result {
+        #[allow(clippy::correctness, clippy::assertions_on_constants)]
+        const {
+            assert!(f64::MAX_10_EXP == 308)
+        };
+
+        if value.is_infinite() {
+            self.p("1e309")
+        } else {
+            self.p(&rustpython_literal::float::to_string(value))
+        }
+    }
+
+    fn unparse_complex(&mut self, real: f64, imag: f64) -> fmt::Result {
+        self.p(&rustpython_literal::complex::to_string(real, imag).replace("inf", "1e309"))
+    }
+
+    fn unparse_constant_value(&mut self, value: &ast::ConstantValue) -> fmt::Result {
+        match value {
+            ast::ConstantValue::None => self.p("None"),
+            ast::ConstantValue::Boolean(value) => self.p(if *value { "True" } else { "False" }),
+            ast::ConstantValue::Str(value) => UnicodeEscape::new_repr(value.as_ref().into())
+                .str_repr()
+                .fmt(self.f),
+            ast::ConstantValue::Bytes(value) => AsciiEscape::new_repr(value.as_ref())
+                .bytes_repr()
+                .fmt(self.f),
+            ast::ConstantValue::Integer(value) => self.p(value.as_ref()),
+            ast::ConstantValue::Tuple(elements) => {
+                self.p("(")?;
+                let mut first = true;
+                for element in elements {
+                    self.p_delim(&mut first, ", ")?;
+                    self.unparse_constant_value(element)?;
+                }
+                self.p_if(elements.len() == 1, ",")?;
+                self.p(")")
+            }
+            ast::ConstantValue::Frozenset(elements) => {
+                if elements.is_empty() {
+                    self.p("frozenset()")
+                } else {
+                    self.p("frozenset({")?;
+                    let mut first = true;
+                    for element in elements {
+                        self.p_delim(&mut first, ", ")?;
+                        self.unparse_constant_value(element)?;
+                    }
+                    self.p("})")
+                }
+            }
+            ast::ConstantValue::Float(value) => self.unparse_float(*value),
+            ast::ConstantValue::Complex { real, imag } => self.unparse_complex(*real, *imag),
+            ast::ConstantValue::Ellipsis => self.p("..."),
+        }
+    }
+
     fn unparse_expr(&mut self, ast: &ast::Expr, level: u8) -> fmt::Result {
         macro_rules! op_prec {
             ($op_ty:ident, $x:expr, $enu:path, $($var:ident($op:literal, $prec:ident)),*$(,)?) => {
@@ -87,6 +144,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 values,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 let (op, prec) = op_prec!(bin, op, ast::BoolOp, And("and", AND), Or("or", OR));
                 group_if!(prec, {
@@ -102,6 +160,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 value,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 group_if!(precedence::TUPLE, {
                     self.unparse_expr(target, precedence::ATOM)?;
@@ -115,6 +174,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 right,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 let right_associative = matches!(op, ast::Operator::Pow);
                 let (op, prec) = op_prec!(
@@ -146,6 +206,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 operand,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 let (op, prec) = op_prec!(
                     un,
@@ -166,6 +227,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 body,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 group_if!(precedence::TEST, {
                     if let Some(parameters) = parameters {
@@ -183,6 +245,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 orelse,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 group_if!(precedence::TEST, {
                     self.unparse_expr(body, precedence::TEST + 1)?;
@@ -196,6 +259,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 items,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("{")?;
                 let mut first = true;
@@ -214,6 +278,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 elts,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("{")?;
                 let mut first = true;
@@ -228,6 +293,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 generators,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("[")?;
                 self.unparse_expr(elt, precedence::TEST)?;
@@ -239,6 +305,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 generators,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("{")?;
                 self.unparse_expr(elt, precedence::TEST)?;
@@ -251,6 +318,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 generators,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("{")?;
                 self.unparse_expr(key, precedence::TEST)?;
@@ -265,6 +333,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 generators,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.p("(")?;
                 self.unparse_expr(elt, precedence::TEST)?;
@@ -275,6 +344,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 value,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 group_if!(precedence::AWAIT, {
                     self.p("await ")?;
@@ -285,6 +355,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 value,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 if let Some(value) = value {
                     write!(self, "(yield {})", UnparseExpr::new(value, self.source))?;
@@ -296,6 +367,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 value,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 write!(
                     self,
@@ -309,6 +381,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 comparators,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 group_if!(precedence::CMP, {
                     let new_lvl = precedence::CMP + 1;
@@ -326,6 +399,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 arguments: ast::Arguments { args, keywords, .. },
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 self.unparse_expr(func, precedence::ATOM)?;
                 self.p("(")?;
@@ -379,26 +453,13 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                     .bytes_repr()
                     .fmt(self.f)?
             }
-            ast::Expr::NumberLiteral(ast::ExprNumberLiteral { value, .. }) => {
-                #[allow(clippy::correctness, clippy::assertions_on_constants)]
-                const {
-                    assert!(f64::MAX_10_EXP == 308)
-                };
-
-                let inf_str = "1e309";
-                match value {
-                    ast::Number::Int(int) => int.fmt(self.f)?,
-                    &ast::Number::Float(fp) => {
-                        if fp.is_infinite() {
-                            self.p(inf_str)?
-                        } else {
-                            self.p(&rustpython_literal::float::to_string(fp))?
-                        }
-                    }
-                    &ast::Number::Complex { real, imag } => self
-                        .p(&rustpython_literal::complex::to_string(real, imag)
-                            .replace("inf", inf_str))?,
-                }
+            ast::Expr::NumberLiteral(ast::ExprNumberLiteral { value, .. }) => match value {
+                ast::Number::Int(int) => int.fmt(self.f)?,
+                &ast::Number::Float(fp) => self.unparse_float(fp)?,
+                &ast::Number::Complex { real, imag } => self.unparse_complex(real, imag)?,
+            },
+            ast::Expr::Constant(ast::ExprConstant { value, .. }) => {
+                self.unparse_constant_value(value)?
             }
             ast::Expr::BooleanLiteral(ast::ExprBooleanLiteral { value, .. }) => {
                 self.p(if *value { "True" } else { "False" })?
@@ -460,6 +521,7 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
                 step,
                 node_index: _,
                 range: _range,
+                ..
             }) => {
                 if let Some(lower) = lower {
                     self.unparse_expr(lower, precedence::TEST)?;
@@ -554,7 +616,9 @@ impl<'a, 'b, 'c> Unparser<'a, 'b, 'c> {
         let buffered =
             fmt::from_fn(|f| Unparser::new(f, self.source).unparse_expr(val, precedence::TEST + 1))
                 .to_string();
-        if let Some(ast::DebugText { leading, trailing }) = debug_text {
+        if let Some(debug_text) = debug_text {
+            let leading = debug_text.leading.as_str();
+            let trailing = debug_text.trailing.as_str();
             self.p(leading)?;
             self.p(self.source.slice(val.range()))?;
             self.p(trailing)?;

crates/compiler-core/src/bytecode.rs

@@ -468,6 +468,13 @@ bitflags! {
         const COROUTINE = 0x0080;
         const ITERABLE_COROUTINE = 0x0100;
         const ASYNC_GENERATOR = 0x0200;
+        const FUTURE_DIVISION = 0x20000;
+        const FUTURE_ABSOLUTE_IMPORT = 0x40000;
+        const FUTURE_WITH_STATEMENT = 0x80000;
+        const FUTURE_PRINT_FUNCTION = 0x100000;
+        const FUTURE_UNICODE_LITERALS = 0x200000;
+        const FUTURE_BARRY_AS_BDFL = 0x400000;
+        const FUTURE_GENERATOR_STOP = 0x800000;
         const FUTURE_ANNOTATIONS = 0x1000000;
         /// If a code object represents a function and has a docstring,
         /// this bit is set and the first item in co_consts is the docstring.

crates/compiler-core/src/bytecode/instruction.rs

@@ -754,9 +754,8 @@ impl Opcode {
     /// Stack effect when the instruction takes its branch (jump=true).
     ///
     /// CPython equivalent: `stack_effect(opcode, oparg, jump=True)`.
-    /// For most instructions this equals the fallthrough effect.
-    /// Override for instructions where branch and fallthrough differ
-    /// (e.g. [`Self::ForIter`]: fallthrough = +1, branch = −1).
+    /// Current opcode metadata has the same real-opcode stack effect
+    /// for jump and fallthrough stack-depth calculation.
     #[must_use]
     pub fn stack_effect_jump(&self, oparg: u32) -> i32 {
         self.stack_effect(oparg)
@@ -1415,6 +1414,26 @@ mod tests {
         assert!(!AnyInstruction::from(PseudoOpcode::Jump).has_const());
     }
 
+    #[test]
+    fn stack_effects_match_cpython_opcode_metadata() {
+        assert_eq!(Opcode::ForIter.stack_effect_info(0).popped(), 1);
+        assert_eq!(Opcode::ForIter.stack_effect_info(0).pushed(), 2);
+        assert_eq!(Opcode::ForIter.stack_effect(0), 1);
+        assert_eq!(Opcode::ForIter.stack_effect_jump(0), 1);
+
+        assert_eq!(Opcode::EndAsyncFor.stack_effect_info(0).popped(), 2);
+        assert_eq!(Opcode::EndAsyncFor.stack_effect_info(0).pushed(), 0);
+        assert_eq!(Opcode::PopJumpIfFalse.stack_effect(0), -1);
+        assert_eq!(Opcode::PopJumpIfFalse.stack_effect_jump(0), -1);
+
+        assert_eq!(PseudoOpcode::SetupFinally.stack_effect_info(0).pushed(), 1);
+        assert_eq!(PseudoOpcode::SetupFinally.stack_effect(0), 0);
+        assert_eq!(PseudoOpcode::SetupFinally.stack_effect_jump(0), 1);
+        assert_eq!(PseudoOpcode::SetupCleanup.stack_effect_info(0).pushed(), 2);
+        assert_eq!(PseudoOpcode::SetupCleanup.stack_effect(0), 0);
+        assert_eq!(PseudoOpcode::SetupCleanup.stack_effect_jump(0), 2);
+    }
+
     #[test]
     fn no_fallthrough_flags_match_cpython_basicblock_nofallthrough() {
         assert!(Opcode::JumpForward.is_no_fallthrough());

crates/compiler-core/src/bytecode/oparg.rs

@@ -777,19 +777,20 @@ oparg_enum!(
 #[derive(Copy, Clone)]
 pub struct UnpackExArgs {
     pub before: u8,
-    pub after: u8,
+    pub after: u32,
 }
 
 impl From<u32> for UnpackExArgs {
     fn from(value: u32) -> Self {
-        let [before, after, ..] = value.to_le_bytes();
+        let before = (value & 0xFF) as u8;
+        let after = value >> 8;
         Self { before, after }
     }
 }
 
 impl From<UnpackExArgs> for u32 {
     fn from(value: UnpackExArgs) -> Self {
-        Self::from_le_bytes([value.before, value.after, 0, 0])
+        Self::from(value.before) | (value.after << 8)
     }
 }
 

crates/stdlib/src/_opcode.rs

@@ -205,7 +205,7 @@ mod tests {
             let scope = vm.new_scope_with_builtins();
             let code_obj = vm
                 .compile(source.trim(), Mode::Exec, FNAME)
-                .map_err(|err| vm.new_syntax_error(&err, Some(source)))
+                .map_err(|err| err.into_pyexception(vm, Some(source)))
                 .unwrap();
             scope.globals.set_item("code", code_obj.into(), vm).unwrap();
 
@@ -228,7 +228,7 @@ output = re.sub(r'(<code object \w+ at )0x[0-9a-fA-F]+', r'\g<1>0xdeadbeef', tmp
 
             let py_code_obj = vm
                 .compile(py_source, Mode::Exec, FNAME)
-                .map_err(|err| vm.new_syntax_error(&err, Some(py_source)))
+                .map_err(|err| err.into_pyexception(vm, Some(py_source)))
                 .unwrap();
 
             vm.run_code_obj(py_code_obj, scope.clone()).unwrap();

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__bare_function_annotations_check_attribute_and_subscript_expressions.snap

@@ -1,5 +1,6 @@
 ---
 source: crates/stdlib/src/_opcode.rs
+assertion_line: 318
 expression: "dis(r#\"\ndef f(one: int):\n    int.new_attr: int\n    [list][0].new_attr: [int, str]\n    my_lst = [1]\n    my_lst[one]: int\n    return my_lst\n\"#)"
 ---
   0           RESUME                   0
@@ -15,7 +16,7 @@ expression: "dis(r#\"\ndef f(one: int):\n    int.new_attr: int\n    [list][0].ne
 
 Disassembly of <code object __annotate__ at 0xdeadbeef file "<?>", line 1>:
   1           RESUME                   0
-              LOAD_FAST_BORROW         0 (format)
+              LOAD_FAST_CHECK          0 (format)
               LOAD_SMALL_INT           2
               COMPARE_OP             132 (>)
               POP_JUMP_IF_FALSE        3 (to L1)

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__const_no_op.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: x = not True
+assertion_line: 281
+expression: "dis(r#\"\nx = not True\n\"#)"
 ---
   0           RESUME                   0
 

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__constant_true_if_pass_keeps_line_anchor_nop.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: "if 1:\n  pass"
+assertion_line: 290
+expression: "dis(r#\"\nif 1:\n    pass\n\"#)"
 ---
   0           RESUME                   0
 

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__if_ands.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: "if True and False and False:\n  pass"
+assertion_line: 252
+expression: "dis(r#\"\nif True and False and False:\n    pass\n\"#)"
 ---
   0           RESUME                   0
 

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__if_mixed.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: "if (True and False) or (False and True):\n  pass"
+assertion_line: 262
+expression: "dis(r#\"\nif (True and False) or (False and True):\n    pass\n\"#)"
 ---
   0           RESUME                   0
 

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__if_ors.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: "if True or False or False:\n  pass"
+assertion_line: 242
+expression: "dis(r#\"\nif True or False or False:\n    pass\n\"#)"
 ---
   0           RESUME                   0
 

crates/stdlib/src/snapshots/rustpython_stdlib___opcode__tests__nested_bool_op.snap

@@ -1,6 +1,7 @@
 ---
 source: crates/stdlib/src/_opcode.rs
-expression: x = Test() and False or False
+assertion_line: 272
+expression: "dis(r#\"\nx = Test() and False or False\n\"#)"
 ---
   0           RESUME                   0
 

crates/vm/Cargo.toml

@@ -70,6 +70,7 @@ static_assertions = { workspace = true }
 strum = { workspace = true }
 strum_macros = { workspace = true }
 thiserror = { workspace = true }
+thin-vec = { workspace = true }
 memchr = { workspace = true }
 
 flamer = { workspace = true, optional = true }

crates/vm/src/builtins/type.rs

@@ -34,7 +34,7 @@ use core::{
     ops::Deref,
     pin::Pin,
     ptr::NonNull,
-    sync::atomic::{AtomicBool, AtomicPtr, AtomicU32, Ordering},
+    sync::atomic::{AtomicBool, AtomicPtr, AtomicU32, AtomicU64, Ordering},
 };
 use indexmap::{IndexMap, map::Entry};
 use itertools::Itertools;
@@ -53,6 +53,7 @@ pub struct PyType {
     pub heaptype_ext: Option<Pin<Box<HeapTypeExt>>>,
     /// Type version tag for inline caching. 0 means unassigned/invalidated.
     pub tp_version_tag: AtomicU32,
+    pub abc_tpflags: AtomicU64,
 }
 
 /// Monotonic counter for type version tags. Once it reaches `u32::MAX`,
@@ -590,7 +591,9 @@ impl PyType {
 
         // Check each base in order and inherit the first collection flag found
         for base in bases {
-            let base_flags = base.slots.flags & COLLECTION_FLAGS;
+            let base_flags = (base.slots.flags
+                | PyTypeFlags::from_bits_truncate(base.abc_tpflags.load(Ordering::Acquire)))
+                & COLLECTION_FLAGS;
             if !base_flags.is_empty() {
                 slots.flags |= base_flags;
                 return;
@@ -598,6 +601,58 @@ impl PyType {
         }
     }
 
+    fn inherited_abc_tpflags(bases: &[PyRef<Self>]) -> u64 {
+        const COLLECTION_FLAGS: PyTypeFlags = PyTypeFlags::from_bits_truncate(
+            PyTypeFlags::SEQUENCE.bits() | PyTypeFlags::MAPPING.bits(),
+        );
+        for base in bases {
+            let base_flags =
+                PyTypeFlags::from_bits_truncate(base.abc_tpflags.load(Ordering::Acquire))
+                    & COLLECTION_FLAGS;
+            if !base_flags.is_empty() {
+                return base_flags.bits();
+            }
+        }
+        0
+    }
+
+    pub fn has_patma_collection_flag(&self, flag: PyTypeFlags) -> bool {
+        debug_assert!(matches!(flag, PyTypeFlags::SEQUENCE | PyTypeFlags::MAPPING));
+        const COLLECTION_FLAGS: PyTypeFlags = PyTypeFlags::from_bits_truncate(
+            PyTypeFlags::SEQUENCE.bits() | PyTypeFlags::MAPPING.bits(),
+        );
+        let slot_flags = self.slots.flags & COLLECTION_FLAGS;
+        if !slot_flags.is_empty() {
+            return slot_flags.contains(flag);
+        }
+        PyTypeFlags::from_bits_truncate(self.abc_tpflags.load(Ordering::Acquire)).contains(flag)
+    }
+
+    pub fn set_abc_collection_flags_recursive(&self, flags: PyTypeFlags) {
+        const COLLECTION_FLAGS: PyTypeFlags = PyTypeFlags::from_bits_truncate(
+            PyTypeFlags::SEQUENCE.bits() | PyTypeFlags::MAPPING.bits(),
+        );
+        let flags = flags & COLLECTION_FLAGS;
+        if flags.is_empty() {
+            return;
+        }
+        let collection_bits = COLLECTION_FLAGS.bits();
+        let flags_bits = flags.bits();
+        let _ = self
+            .abc_tpflags
+            .fetch_update(Ordering::AcqRel, Ordering::Acquire, |old| {
+                Some((old & !collection_bits) | flags_bits)
+            });
+        self.modified();
+        for weak_ref in self.subclasses.read().iter() {
+            if let Some(subclass) = weak_ref.upgrade()
+                && let Some(subclass) = subclass.downcast_ref::<Self>()
+            {
+                subclass.set_abc_collection_flags_recursive(flags);
+            }
+        }
+    }
+
     /// Check for __abc_tpflags__ and set the appropriate flags
     /// This checks in attrs and all base classes for __abc_tpflags__
     fn check_abc_tpflags(
@@ -626,21 +681,19 @@ impl PyType {
                         .to_owned(),
                 );
             }
-            // Don't override flags already inherited from a base class.
-            if !slots.flags.intersects(COLLECTION_FLAGS) {
-                slots.flags |= masked;
-            }
+            slots.flags.remove(COLLECTION_FLAGS);
+            slots.flags |= masked;
             return Ok(());
         }
 
-        // No __abc_tpflags__ on this class — inheritance already happened
-        // in inherit_patma_flags, so nothing more to do if those bits are set.
+        // No __abc_tpflags__ on this class. Inheritance already happened in
+        // inherit_patma_flags, using base order and including ABC markers.
         if slots.flags.intersects(COLLECTION_FLAGS) {
             return Ok(());
         }
 
-        // Then check in base classes (legacy path for cases that bypass
-        // inherit_patma_flags).
+        // Then check in base classes for legacy paths that bypassed
+        // inherit_patma_flags.
         for base in bases {
             if let Some(abc_tpflags_obj) = base.find_name_in_mro(abc_tpflags_name)
                 && let Some(int_obj) = abc_tpflags_obj.downcast_ref::<crate::builtins::int::PyInt>()
@@ -654,6 +707,7 @@ impl PyType {
                             .to_owned(),
                     );
                 }
+                slots.flags.remove(COLLECTION_FLAGS);
                 slots.flags |= masked;
                 return Ok(());
             }
@@ -716,6 +770,7 @@ impl PyType {
             ));
         }
 
+        let inherited_abc_tpflags = Self::inherited_abc_tpflags(&bases);
         let new_type = PyRef::new_ref(
             Self {
                 base: Some(base),
@@ -726,6 +781,7 @@ impl PyType {
                 slots,
                 heaptype_ext: Some(Pin::new(Box::new(heaptype_ext))),
                 tp_version_tag: AtomicU32::new(0),
+                abc_tpflags: AtomicU64::new(inherited_abc_tpflags),
             },
             metaclass,
             None,
@@ -775,6 +831,7 @@ impl PyType {
             slots.flags |= PyTypeFlags::MANAGED_WEAKREF;
         }
 
+        let inherited_abc_tpflags = Self::inherited_abc_tpflags(core::slice::from_ref(&base));
         let bases = PyRwLock::new(vec![base.clone()]);
         let mro = base.mro_map_collect(|x| x.to_owned());
 
@@ -788,6 +845,7 @@ impl PyType {
                 slots,
                 heaptype_ext: None,
                 tp_version_tag: AtomicU32::new(0),
+                abc_tpflags: AtomicU64::new(inherited_abc_tpflags),
             },
             metaclass,
             None,

crates/vm/src/eval.rs

@@ -6,7 +6,7 @@ pub fn eval(vm: &VirtualMachine, source: &str, scope: Scope, source_path: &str)
             debug!("Code object: {bytecode:?}");
             vm.run_code_obj(bytecode, scope)
         }
-        Err(err) => Err(vm.new_syntax_error(&err, Some(source))),
+        Err(err) => Err(err.into_pyexception(vm, Some(source))),
     }
 }
 

crates/vm/src/exceptions.rs

@@ -2480,12 +2480,18 @@ pub(super) mod types {
             let maybe_lineno = zelf
                 .as_object()
                 .get_attr("lineno", vm)
-                .and_then(|obj| obj.str_utf8(vm))
-                .ok();
-            let maybe_filename = zelf.as_object().get_attr("filename", vm).ok().map(|obj| {
-                obj.str(vm)
-                    .unwrap_or_else(|_| vm.ctx.new_str("<filename str() failed>"))
-            });
+                .ok()
+                .filter(|obj| !vm.is_none(obj))
+                .and_then(|obj| obj.str_utf8(vm).ok());
+            let maybe_filename = zelf
+                .as_object()
+                .get_attr("filename", vm)
+                .ok()
+                .filter(|obj| !vm.is_none(obj))
+                .map(|obj| {
+                    obj.str(vm)
+                        .unwrap_or_else(|_| vm.ctx.new_str("<filename str() failed>"))
+                });
 
             let msg = match zelf.as_object().get_attr("msg", vm) {
                 Ok(obj) => obj

crates/vm/src/frame.rs

@@ -3050,9 +3050,17 @@ impl ExecutingFrame<'_> {
                 let subject = self.pop_value();
                 let nargs_val = nargs.get(arg) as usize;
 
+                let Some(cls_type) = cls.downcast_ref::<PyType>() else {
+                    return Err(vm.new_type_error("called match pattern must be a class"));
+                };
+                // Only the error paths need the class name; compute it lazily so a
+                // successful match does not take the name lock or allocate.
+                let type_name = || cls_type.name().to_string();
+
                 // Check if subject is an instance of cls
                 if subject.is_instance(cls.as_ref(), vm)? {
                     let mut extracted = vec![];
+                    let seen_attrs = PySet::default().into_ref(&vm.ctx);
 
                     // Get __match_args__ for positional arguments if nargs > 0
                     if nargs_val > 0 {
@@ -3066,12 +3074,7 @@ impl ExecutingFrame<'_> {
                                 Ok(tuple) => tuple,
                                 Err(match_args) => {
                                     // __match_args__ must be a tuple
-                                    // Get type names for error message
-                                    let type_name = cls
-                                        .downcast::<crate::builtins::PyType>()
-                                        .ok()
-                                        .and_then(|t| t.__name__(vm).to_str().map(str::to_owned))
-                                        .unwrap_or_else(|| String::from("?"));
+                                    let type_name = type_name();
                                     let match_args_type_name = match_args.class().__name__(vm);
                                     return Err(vm.new_type_error(format!(
                                         "{type_name}.__match_args__ must be a tuple (got {match_args_type_name})"
@@ -3081,9 +3084,12 @@ impl ExecutingFrame<'_> {
 
                             // Check if we have enough match args
                             if match_args.len() < nargs_val {
+                                let type_name = type_name();
+                                let plural = if match_args.len() == 1 { "" } else { "s" };
                                 return Err(vm.new_type_error(format!(
-                                    "class pattern accepts at most {} positional sub-patterns ({} given)",
+                                    "{type_name}() accepts {} positional sub-pattern{} ({} given)",
                                     match_args.len(),
+                                    plural,
                                     nargs_val
                                 )));
                             }
@@ -3094,11 +3100,20 @@ impl ExecutingFrame<'_> {
                                 let attr_name_str = match attr_name.downcast_ref::<PyStr>() {
                                     Some(s) => s,
                                     None => {
-                                        return Err(vm.new_type_error(
-                                            "__match_args__ elements must be strings",
-                                        ));
+                                        let attr_type_name = attr_name.class().name();
+                                        return Err(vm.new_type_error(format!(
+                                            "__match_args__ elements must be strings (got {attr_type_name})"
+                                        )));
                                     }
                                 };
+                                if seen_attrs.__contains__(attr_name.as_object(), vm)? {
+                                    let type_name = type_name();
+                                    let attr_repr = attr_name.as_object().repr(vm)?;
+                                    return Err(vm.new_type_error(format!(
+                                        "{type_name}() got multiple sub-patterns for attribute {attr_repr}"
+                                    )));
+                                }
+                                seen_attrs.add(attr_name.clone(), vm)?;
                                 match subject.get_attr(attr_name_str, vm) {
                                     Ok(value) => extracted.push(value),
                                     Err(e)
@@ -3115,9 +3130,8 @@ impl ExecutingFrame<'_> {
                             // No __match_args__, check if this is a type with MATCH_SELF behavior
                             // For built-in types like bool, int, str, list, tuple, dict, etc.
                             // they match the subject itself as the single positional argument
-                            let is_match_self_type = cls
-                                .downcast::<PyType>()
-                                .is_ok_and(|t| t.slots.flags.contains(PyTypeFlags::_MATCH_SELF));
+                            let is_match_self_type =
+                                cls_type.slots.flags.contains(PyTypeFlags::_MATCH_SELF);
 
                             if is_match_self_type {
                                 if nargs_val == 1 {
@@ -3125,16 +3139,18 @@ impl ExecutingFrame<'_> {
                                     extracted.push(subject.clone());
                                 } else if nargs_val > 1 {
                                     // Too many positional arguments for MATCH_SELF
-                                    return Err(vm.new_type_error(
-                                        "class pattern accepts at most 1 positional sub-pattern for MATCH_SELF types",
-                                    ));
+                                    let type_name = type_name();
+                                    return Err(vm.new_type_error(format!(
+                                        "{type_name}() accepts 1 positional sub-pattern ({nargs_val} given)"
+                                    )));
                                 }
                             } else {
                                 // No __match_args__ and not a MATCH_SELF type
                                 if nargs_val > 0 {
-                                    return Err(vm.new_type_error(
-                                        "class pattern defines no positional sub-patterns (__match_args__ missing)",
-                                    ));
+                                    let type_name = type_name();
+                                    return Err(vm.new_type_error(format!(
+                                        "{type_name}() accepts 0 positional sub-patterns ({nargs_val} given)"
+                                    )));
                                 }
                             }
                         }
@@ -3143,6 +3159,14 @@ impl ExecutingFrame<'_> {
                     // Extract keyword attributes
                     for name in kwd_attrs {
                         let name_str = name.downcast_ref::<PyStr>().unwrap();
+                        if seen_attrs.__contains__(name_str.as_object(), vm)? {
+                            let type_name = type_name();
+                            let attr_repr = name.as_object().repr(vm)?;
+                            return Err(vm.new_type_error(format!(
+                                "{type_name}() got multiple sub-patterns for attribute {attr_repr}"
+                            )));
+                        }
+                        seen_attrs.add(name.clone(), vm)?;
                         match subject.get_attr(name_str, vm) {
                             Ok(value) => extracted.push(value),
                             Err(e) if e.fast_isinstance(vm.ctx.exceptions.attribute_error) => {
@@ -3166,10 +3190,14 @@ impl ExecutingFrame<'_> {
                 let subject = self.nth_value(1); // stack[-2]
 
                 // Check if subject is a mapping and extract values for keys
-                if subject.class().slots.flags.contains(PyTypeFlags::MAPPING) {
+                if subject
+                    .class()
+                    .has_patma_collection_flag(PyTypeFlags::MAPPING)
+                {
                     let keys = keys_tuple.downcast_ref::<PyTuple>().unwrap();
                     let mut values = Vec::new();
                     let mut all_match = true;
+                    let seen_keys = PySet::default().into_ref(&vm.ctx);
 
                     // We use the two argument form of map.get(key, default) for two reasons:
                     // - Atomically check for a key and get its value without error handling.
@@ -3186,6 +3214,13 @@ impl ExecutingFrame<'_> {
                             .new_base_object(vm.ctx.types.object_type.to_owned(), None);
 
                         for key in keys {
+                            if seen_keys.__contains__(key.as_object(), vm)? {
+                                return Err(vm.new_value_error(format!(
+                                    "mapping pattern checks duplicate key ({})",
+                                    key.as_object().repr(vm)?
+                                )));
+                            }
+                            seen_keys.add(key.as_object().to_owned(), vm)?;
                             // value = map.get(key, dummy)
                             match get_method.call((key.as_object(), dummy.clone()), vm) {
                                 Ok(value) => {
@@ -3202,6 +3237,13 @@ impl ExecutingFrame<'_> {
                     } else {
                         // Fallback if .get() method is not available (shouldn't happen for mappings)
                         for key in keys {
+                            if seen_keys.__contains__(key.as_object(), vm)? {
+                                return Err(vm.new_value_error(format!(
+                                    "mapping pattern checks duplicate key ({})",
+                                    key.as_object().repr(vm)?
+                                )));
+                            }
+                            seen_keys.add(key.as_object().to_owned(), vm)?;
                             match subject.get_item(key.as_object(), vm) {
                                 Ok(value) => values.push(value),
                                 Err(e) if e.fast_isinstance(vm.ctx.exceptions.key_error) => {
@@ -3231,7 +3273,9 @@ impl ExecutingFrame<'_> {
                 let subject = self.pop_value();
 
                 // Check if the type has the MAPPING flag
-                let is_mapping = subject.class().slots.flags.contains(PyTypeFlags::MAPPING);
+                let is_mapping = subject
+                    .class()
+                    .has_patma_collection_flag(PyTypeFlags::MAPPING);
 
                 self.push_value(subject);
                 self.push_value(vm.ctx.new_bool(is_mapping).into());
@@ -3242,7 +3286,9 @@ impl ExecutingFrame<'_> {
                 let subject = self.pop_value();
 
                 // Check if the type has the SEQUENCE flag
-                let is_sequence = subject.class().slots.flags.contains(PyTypeFlags::SEQUENCE);
+                let is_sequence = subject
+                    .class()
+                    .has_patma_collection_flag(PyTypeFlags::SEQUENCE);
 
                 self.push_value(subject);
                 self.push_value(vm.ctx.new_bool(is_sequence).into());
@@ -6845,7 +6891,7 @@ impl ExecutingFrame<'_> {
         }
     }
 
-    fn execute_unpack_ex(&mut self, vm: &VirtualMachine, before: u8, after: u8) -> FrameResult {
+    fn execute_unpack_ex(&mut self, vm: &VirtualMachine, before: u8, after: u32) -> FrameResult {
         let (before, after) = (before as usize, after as usize);
         let value = self.pop_value();
         let not_iterable = value.class().slots.iter.load().is_none()

crates/vm/src/import.rs

@@ -141,7 +141,7 @@ pub fn import_file(
             file_path,
             vm.compile_opts(),
         )
-        .map_err(|err| vm.new_syntax_error(&err, Some(content)))?;
+        .map_err(|err| err.into_pyexception(vm, Some(content)))?;
     import_code_obj(vm, module_name, code, true)
 }
 
@@ -154,7 +154,7 @@ pub fn import_source(vm: &VirtualMachine, module_name: &str, content: &str) -> P
             "<source>",
             vm.compile_opts(),
         )
-        .map_err(|err| vm.new_syntax_error(&err, Some(content)))?;
+        .map_err(|err| err.into_pyexception(vm, Some(content)))?;
     import_code_obj(vm, module_name, code, false)
 }
 

crates/vm/src/object/core.rs

@@ -2479,6 +2479,7 @@ pub(crate) fn init_type_hierarchy() -> (PyTypeRef, PyTypeRef, PyTypeRef) {
             slots: PyType::make_slots(),
             heaptype_ext: None,
             tp_version_tag: core::sync::atomic::AtomicU32::new(0),
+            abc_tpflags: core::sync::atomic::AtomicU64::new(0),
         };
         let object_payload = PyType {
             base: None,
@@ -2489,6 +2490,7 @@ pub(crate) fn init_type_hierarchy() -> (PyTypeRef, PyTypeRef, PyTypeRef) {
             slots: object::PyBaseObject::make_slots(),
             heaptype_ext: None,
             tp_version_tag: core::sync::atomic::AtomicU32::new(0),
+            abc_tpflags: core::sync::atomic::AtomicU64::new(0),
         };
         // Both type_type and object_type are instances of `type`, which has
         // HAS_DICT and HAS_WEAKREF, so they need both ObjExt and WeakRefList prefixes.
@@ -2585,6 +2587,7 @@ pub(crate) fn init_type_hierarchy() -> (PyTypeRef, PyTypeRef, PyTypeRef) {
         slots: PyWeak::make_slots(),
         heaptype_ext: None,
         tp_version_tag: core::sync::atomic::AtomicU32::new(0),
+        abc_tpflags: core::sync::atomic::AtomicU64::new(0),
     };
     let weakref_type = PyRef::new_ref(weakref_type, type_type.clone(), None);
     // Static type: untrack from GC (was tracked by new_ref because PyType has HAS_TRAVERSE)

crates/vm/src/protocol/callable.rs

@@ -207,7 +207,7 @@ impl VirtualMachine {
         event: TraceEvent,
         arg: Option<PyObjectRef>,
     ) -> PyResult<Option<PyObjectRef>> {
-        if self.use_tracing.get() {
+        if self.use_tracing.get() && !self.tracing_is_suppressed() {
             self._trace_event_inner(event, arg)
         } else {
             Ok(None)
@@ -247,7 +247,9 @@ impl VirtualMachine {
         // tracing function itself.
         if is_trace_event && !self.is_none(&trace_func) {
             self.use_tracing.set(false);
+            self.enter_tracing();
             let res = trace_func.call(args.clone(), self);
+            self.leave_tracing();
             self.use_tracing.set(true);
             match res {
                 Ok(result) => {
@@ -268,7 +270,9 @@ impl VirtualMachine {
 
         if is_profile_event && !self.is_none(&profile_func) {
             self.use_tracing.set(false);
+            self.enter_tracing();
             let res = profile_func.call(args, self);
+            self.leave_tracing();
             self.use_tracing.set(true);
             if res.is_err() {
                 *self.profile_func.borrow_mut() = self.ctx.none();

crates/vm/src/stdlib/_abc.rs

@@ -9,11 +9,11 @@ pub(crate) use _abc::module_def;
 mod _abc {
     use crate::{
         AsObject, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, VirtualMachine,
-        builtins::{PyFrozenSet, PyList, PySet, PyStr, PyTupleRef, PyTypeRef, PyWeak},
+        builtins::{PyFrozenSet, PyList, PySet, PyStr, PyTupleRef, PyType, PyTypeRef, PyWeak},
         common::lock::PyRwLock,
         convert::ToPyObject,
         protocol::PyIterReturn,
-        types::Constructor,
+        types::{Constructor, PyTypeFlags},
     };
     use core::sync::atomic::{AtomicU64, Ordering};
 
@@ -238,6 +238,23 @@ mod _abc {
         // Invalidate negative cache
         increment_invalidation_counter();
 
+        if let Some(cls_type) = cls.downcast_ref::<PyType>()
+            && let Some(subclass_type) = subclass.downcast_ref::<PyType>()
+        {
+            // _abc_register propagates Py_TPFLAGS_SEQUENCE/MAPPING
+            // recursively so MATCH_SEQUENCE/MATCH_MAPPING see ABC registration.
+            let collection_mask = PyTypeFlags::SEQUENCE | PyTypeFlags::MAPPING;
+            let collection_flags = (cls_type.slots.flags
+                | PyTypeFlags::from_bits_truncate(cls_type.abc_tpflags.load(Ordering::Acquire)))
+                & collection_mask;
+            if !subclass_type.is(vm.ctx.types.str_type)
+                && !subclass_type.is(vm.ctx.types.bytes_type)
+                && !subclass_type.is(vm.ctx.types.bytearray_type)
+            {
+                subclass_type.set_abc_collection_flags_recursive(collection_flags);
+            }
+        }
+
         Ok(subclass)
     }
 

crates/vm/src/stdlib/_ast/argument.rs

@@ -2,14 +2,78 @@ use super::*;
 use rustpython_compiler_core::SourceFile;
 
 pub(super) struct PositionalArguments {
-    pub range: TextRange,
-    pub args: Box<[ast::Expr]>,
+    range: TextRange,
+    kind: PositionalArgumentsKind,
+}
+
+enum PositionalArgumentsKind {
+    Args(Box<[ast::Expr]>),
+    RuntimeValues(Vec<Option<ast::Expr>>),
+}
+
+impl PositionalArguments {
+    pub(super) fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        let values: Vec<Option<ast::Expr>> =
+            get_node_list_field(vm, source_file, object, field, typ)?;
+        Ok(Self::from_values(TextRange::default(), values))
+    }
+
+    fn from_args(range: TextRange, args: Box<[ast::Expr]>) -> Self {
+        Self {
+            range,
+            kind: PositionalArgumentsKind::Args(args),
+        }
+    }
+
+    fn from_runtime_values(range: TextRange, values: Vec<Option<ast::Expr>>) -> Self {
+        Self {
+            range,
+            kind: PositionalArgumentsKind::RuntimeValues(values),
+        }
+    }
+
+    fn from_values(range: TextRange, values: Vec<Option<ast::Expr>>) -> Self {
+        if values.iter().any(Option::is_none) {
+            Self::from_runtime_values(range, values)
+        } else {
+            Self::from_args(
+                range,
+                values
+                    .into_iter()
+                    .flatten()
+                    .collect::<Vec<_>>()
+                    .into_boxed_slice(),
+            )
+        }
+    }
+
+    fn range(&self) -> TextRange {
+        self.range
+    }
+
+    fn into_args_and_runtime_values(self) -> (Box<[ast::Expr]>, Option<Vec<Option<ast::Expr>>>) {
+        match self.kind {
+            PositionalArgumentsKind::Args(args) => (args, None),
+            PositionalArgumentsKind::RuntimeValues(values) => (
+                lower_runtime_expr_list(values.clone()).into_boxed_slice(),
+                Some(values),
+            ),
+        }
+    }
 }
 
 impl Node for PositionalArguments {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { args, range: _ } = self;
-        BoxedSlice(args).ast_to_object(vm, source_file)
+        match self.kind {
+            PositionalArgumentsKind::Args(args) => BoxedSlice(args).ast_to_object(vm, source_file),
+            PositionalArgumentsKind::RuntimeValues(values) => values.ast_to_object(vm, source_file),
+        }
     }
 
     fn ast_from_object(
@@ -18,10 +82,7 @@ impl Node for PositionalArguments {
         object: PyObjectRef,
     ) -> PyResult<Self> {
         let args: BoxedSlice<_> = Node::ast_from_object(vm, source_file, object)?;
-        Ok(Self {
-            args: args.0,
-            range: TextRange::default(), // TODO
-        })
+        Ok(Self::from_args(TextRange::default(), args.0))
     }
 }
 
@@ -30,6 +91,21 @@ pub(super) struct KeywordArguments {
     pub keywords: Box<[ast::Keyword]>,
 }
 
+impl KeywordArguments {
+    pub(super) fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        Ok(Self {
+            keywords: get_node_boxed_slice_field(vm, source_file, object, field, typ)?,
+            range: TextRange::default(),
+        })
+    }
+}
+
 impl Node for KeywordArguments {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self { keywords, range: _ } = self;
@@ -45,7 +121,7 @@ impl Node for KeywordArguments {
         let keywords: BoxedSlice<_> = Node::ast_from_object(vm, source_file, object)?;
         Ok(Self {
             keywords: keywords.0,
-            range: TextRange::default(), // TODO
+            range: TextRange::default(),
         })
     }
 }
@@ -54,13 +130,16 @@ pub(super) fn merge_function_call_arguments(
     pos_args: PositionalArguments,
     key_args: KeywordArguments,
 ) -> ast::Arguments {
-    let range = pos_args.range.cover(key_args.range);
+    let range = pos_args.range().cover(key_args.range);
+    let (args, runtime_args) = pos_args.into_args_and_runtime_values();
 
     ast::Arguments {
         node_index: Default::default(),
         range,
-        args: pos_args.args,
+        args,
         keywords: key_args.keywords,
+        runtime_args,
+        runtime_bases: None,
     }
 }
 
@@ -68,10 +147,12 @@ pub(super) fn split_function_call_arguments(
     args: ast::Arguments,
 ) -> (PositionalArguments, KeywordArguments) {
     let ast::Arguments {
-        node_index: _,
         range: _,
         args,
         keywords,
+        runtime_args,
+        runtime_bases: _,
+        ..
     } = args;
 
     let positional_arguments_range = args
@@ -80,10 +161,10 @@ pub(super) fn split_function_call_arguments(
         .reduce(|acc, next| acc.cover(next))
         .unwrap_or_default();
     // debug_assert!(range.contains_range(positional_arguments_range));
-    let positional_arguments = PositionalArguments {
-        range: positional_arguments_range,
-        args,
-    };
+    let positional_arguments = runtime_args.map_or_else(
+        || PositionalArguments::from_args(positional_arguments_range, args),
+        |values| PositionalArguments::from_runtime_values(positional_arguments_range, values),
+    );
 
     let keyword_arguments_range = keywords
         .iter()
@@ -107,10 +188,12 @@ pub(super) fn split_class_def_args(
         Some(args) => *args,
     };
     let ast::Arguments {
-        node_index: _,
         range: _,
         args,
         keywords,
+        runtime_args: _,
+        runtime_bases,
+        ..
     } = args;
 
     let positional_arguments_range = args
@@ -119,10 +202,10 @@ pub(super) fn split_class_def_args(
         .reduce(|acc, next| acc.cover(next))
         .unwrap_or_default();
     // debug_assert!(range.contains_range(positional_arguments_range));
-    let positional_arguments = PositionalArguments {
-        range: positional_arguments_range,
-        args,
-    };
+    let positional_arguments = runtime_bases.map_or_else(
+        || PositionalArguments::from_args(positional_arguments_range, args),
+        |values| PositionalArguments::from_runtime_values(positional_arguments_range, values),
+    );
 
     let keyword_arguments_range = keywords
         .iter()
@@ -146,10 +229,10 @@ pub(super) fn merge_class_def_args(
         return None;
     }
 
-    let args = if let Some(positional_arguments) = positional_arguments {
-        positional_arguments.args
+    let (args, runtime_bases) = if let Some(positional_arguments) = positional_arguments {
+        positional_arguments.into_args_and_runtime_values()
     } else {
-        vec![].into_boxed_slice()
+        (vec![].into_boxed_slice(), None)
     };
     let keywords = if let Some(keyword_arguments) = keyword_arguments {
         keyword_arguments.keywords
@@ -162,5 +245,7 @@ pub(super) fn merge_class_def_args(
         range: Default::default(), // TODO
         args,
         keywords,
+        runtime_args: None,
+        runtime_bases,
     }))
 }

crates/vm/src/stdlib/_ast/basic.rs

@@ -1,4 +1,5 @@
 use super::*;
+use crate::builtins::PyIntRef;
 use rustpython_codegen::compile::ruff_int_to_bigint;
 use rustpython_compiler_core::SourceFile;
 
@@ -13,7 +14,11 @@ impl Node for ast::Identifier {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let py_str = PyUtf8StrRef::try_from_object(vm, object)?;
+        if !object.class().is(vm.ctx.types.str_type) {
+            return Err(vm.new_type_error("AST identifier must be of type str"));
+        }
+        let py_str = PyUtf8StrRef::try_from_object(vm, object)
+            .map_err(|_| vm.new_type_error("AST identifier must be of type str"))?;
         Ok(Self::new(py_str.as_str(), TextRange::default()))
     }
 }
@@ -28,7 +33,6 @@ impl Node for ast::Int {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        // FIXME: performance
         let value: PyIntRef = object.try_into_value(vm)?;
         let value = value.as_bigint().to_string();
         Ok(value.parse().unwrap())
@@ -45,6 +49,6 @@ impl Node for bool {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        i32::try_from_object(vm, object).map(|i| i != 0)
+        node_object_to_i32(vm, object).map(|i| i != 0)
     }
 }

crates/vm/src/stdlib/_ast/constant.rs

@@ -1,12 +1,15 @@
 use super::*;
 use crate::builtins::{PyComplex, PyFrozenSet, PyTuple};
 use ast::str_prefix::StringLiteralPrefix;
-use rustpython_compiler_core::SourceFile;
+use rustpython_codegen::compile::ruff_int_to_bigint;
+use rustpython_compiler_core::{SourceFile, bytecode::ConstantData};
 
 #[derive(Debug)]
 pub(super) struct Constant {
     pub(super) range: TextRange,
     pub(super) value: ConstantLiteral,
+    kind: Option<Box<str>>,
+    invalid_type: Option<String>,
 }
 
 impl Constant {
@@ -19,6 +22,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Str { value, prefix },
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -26,6 +31,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Int(value),
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -33,6 +40,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Float(value),
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -40,6 +49,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Complex { real, imag },
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -47,6 +58,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Bytes(value),
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -54,6 +67,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Bool(value),
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -61,6 +76,8 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::None,
+            kind: None,
+            invalid_type: None,
         }
     }
 
@@ -68,15 +85,29 @@ impl Constant {
         Self {
             range,
             value: ConstantLiteral::Ellipsis,
+            kind: None,
+            invalid_type: None,
         }
     }
 
     pub(crate) fn into_expr(self) -> ast::Expr {
-        constant_to_ruff_expr(self)
+        let Self {
+            range,
+            value,
+            kind,
+            invalid_type,
+        } = self;
+        ast::Expr::Constant(ast::ExprConstant {
+            node_index: Default::default(),
+            range,
+            value: constant_data_to_ast_constant_value(constant_literal_to_constant_data(&value)),
+            kind: kind.or_else(|| constant_literal_kind(&value)),
+            invalid_type: invalid_type.map(String::into_boxed_str),
+        })
     }
 }
 
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub(crate) enum ConstantLiteral {
     None,
     Bool(bool),
@@ -96,20 +127,361 @@ pub(crate) enum ConstantLiteral {
     Ellipsis,
 }
 
+pub(super) fn invalid_constant_type(expr: &ast::Expr) -> Option<Box<str>> {
+    match expr {
+        ast::Expr::Constant(expr) => expr.invalid_type.clone(),
+        _ => None,
+    }
+}
+
+pub(super) fn runtime_string_from_pyobject(
+    vm: &VirtualMachine,
+    object: PyObjectRef,
+) -> (Option<Box<str>>, Option<Vec<u8>>) {
+    runtime_string_from_object(vm, object)
+}
+
+pub(super) fn runtime_string_object(
+    vm: &VirtualMachine,
+    value: Option<Box<str>>,
+    bytes: Option<Vec<u8>>,
+) -> Option<PyObjectRef> {
+    runtime_string_to_object(vm, value, bytes)
+}
+
+pub(super) fn expr_constant_to_object(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    expr: ast::ExprConstant,
+) -> PyObjectRef {
+    let ast::ExprConstant {
+        node_index: _,
+        range,
+        value,
+        kind,
+        invalid_type: _,
+    } = expr;
+    let constant = ast_constant_value_to_constant_data(value);
+    let node = NodeAst
+        .into_ref_with_type(vm, pyast::NodeExprConstant::static_type().to_owned())
+        .unwrap();
+    let dict = node.as_object().dict().unwrap();
+    dict.set_item("value", constant_data_to_object(vm, constant), vm)
+        .unwrap();
+    let kind = kind.map_or_else(|| vm.ctx.none(), |kind| vm.ctx.new_str(kind).into());
+    dict.set_item("kind", kind, vm).unwrap();
+    node_add_location(&dict, range, vm, source_file);
+    node.into()
+}
+
+pub(super) fn runtime_interpolation_object(
+    vm: &VirtualMachine,
+    str: Option<ast::ConstantValue>,
+    format_spec: Option<Box<ast::Expr>>,
+) -> Option<(PyObjectRef, Option<Box<ast::Expr>>)> {
+    let str = str?;
+    Some((
+        constant_data_to_object(vm, ast_constant_value_to_constant_data(str)),
+        format_spec,
+    ))
+}
+
+pub(super) fn runtime_stmt_type_comment_object(
+    vm: &VirtualMachine,
+    value: Option<Box<str>>,
+    bytes: Option<Vec<u8>>,
+) -> Option<PyObjectRef> {
+    runtime_string_object(vm, value, bytes)
+}
+
+fn constant_literal_to_constant_data(value: &ConstantLiteral) -> ConstantData {
+    match value {
+        ConstantLiteral::None => ConstantData::None,
+        ConstantLiteral::Bool(value) => ConstantData::Boolean { value: *value },
+        ConstantLiteral::Str { value, .. } => ConstantData::Str {
+            value: value.as_ref().into(),
+        },
+        ConstantLiteral::Bytes(value) => ConstantData::Bytes {
+            value: value.to_vec(),
+        },
+        ConstantLiteral::Int(value) => ConstantData::Integer {
+            value: ruff_int_to_bigint(value).unwrap(),
+        },
+        ConstantLiteral::Tuple(value) => ConstantData::Tuple {
+            elements: value
+                .iter()
+                .map(constant_literal_to_constant_data)
+                .collect(),
+        },
+        ConstantLiteral::FrozenSet(value) => ConstantData::Frozenset {
+            elements: value
+                .iter()
+                .map(constant_literal_to_constant_data)
+                .collect(),
+        },
+        ConstantLiteral::Float(value) => ConstantData::Float { value: *value },
+        ConstantLiteral::Complex { real, imag } => ConstantData::Complex {
+            value: num_complex::Complex::new(*real, *imag),
+        },
+        ConstantLiteral::Ellipsis => ConstantData::Ellipsis,
+    }
+}
+
+fn constant_literal_kind(value: &ConstantLiteral) -> Option<Box<str>> {
+    match value {
+        ConstantLiteral::Str {
+            prefix: StringLiteralPrefix::Unicode,
+            ..
+        } => Some("u".into()),
+        _ => None,
+    }
+}
+
+pub(super) fn constant_data_to_ast_constant_value(value: ConstantData) -> ast::ConstantValue {
+    match value {
+        ConstantData::None => ast::ConstantValue::None,
+        ConstantData::Boolean { value } => ast::ConstantValue::Boolean(value),
+        ConstantData::Str { value } => ast::ConstantValue::Str(value.to_string().into_boxed_str()),
+        ConstantData::Bytes { value } => ast::ConstantValue::Bytes(value.into_boxed_slice()),
+        ConstantData::Integer { value } => ast::ConstantValue::Integer(value.to_string().into()),
+        ConstantData::Tuple { elements } => ast::ConstantValue::Tuple(
+            elements
+                .into_iter()
+                .map(constant_data_to_ast_constant_value)
+                .collect(),
+        ),
+        ConstantData::Frozenset { elements } => ast::ConstantValue::Frozenset(
+            elements
+                .into_iter()
+                .map(constant_data_to_ast_constant_value)
+                .collect(),
+        ),
+        ConstantData::Float { value } => ast::ConstantValue::Float(value),
+        ConstantData::Complex { value } => ast::ConstantValue::Complex {
+            real: value.re,
+            imag: value.im,
+        },
+        ConstantData::Ellipsis => ast::ConstantValue::Ellipsis,
+        ConstantData::Code { .. } | ConstantData::Slice { .. } => {
+            unreachable!("ast.Constant values cannot contain code objects or slices")
+        }
+    }
+}
+
+pub(super) fn ast_constant_value_to_constant_data(value: ast::ConstantValue) -> ConstantData {
+    match value {
+        ast::ConstantValue::None => ConstantData::None,
+        ast::ConstantValue::Boolean(value) => ConstantData::Boolean { value },
+        ast::ConstantValue::Str(value) => ConstantData::Str {
+            value: value.to_string().into(),
+        },
+        ast::ConstantValue::Bytes(value) => ConstantData::Bytes {
+            value: value.into_vec(),
+        },
+        ast::ConstantValue::Integer(value) => ConstantData::Integer {
+            value: value
+                .parse()
+                .expect("RustPython ast.Constant integer values are decimal integers"),
+        },
+        ast::ConstantValue::Tuple(elements) => ConstantData::Tuple {
+            elements: elements
+                .into_iter()
+                .map(ast_constant_value_to_constant_data)
+                .collect(),
+        },
+        ast::ConstantValue::Frozenset(elements) => ConstantData::Frozenset {
+            elements: elements
+                .into_iter()
+                .map(ast_constant_value_to_constant_data)
+                .collect(),
+        },
+        ast::ConstantValue::Float(value) => ConstantData::Float { value },
+        ast::ConstantValue::Complex { real, imag } => ConstantData::Complex {
+            value: num_complex::Complex::new(real, imag),
+        },
+        ast::ConstantValue::Ellipsis => ConstantData::Ellipsis,
+    }
+}
+
+pub(super) fn constant_object_to_constant_data(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    value_object: PyObjectRef,
+) -> PyResult<ConstantData> {
+    let value = ConstantLiteral::ast_from_object(vm, source_file, value_object)?;
+    Ok(constant_literal_to_constant_data(&value))
+}
+
+fn runtime_string_from_object(
+    vm: &VirtualMachine,
+    object: PyObjectRef,
+) -> (Option<Box<str>>, Option<Vec<u8>>) {
+    if object.class().is(vm.ctx.types.str_type) {
+        (
+            Some(
+                object
+                    .try_to_value::<String>(vm)
+                    .expect("AST string field was validated as str")
+                    .into_boxed_str(),
+            ),
+            None,
+        )
+    } else {
+        (
+            None,
+            Some(
+                object
+                    .try_to_value::<Vec<u8>>(vm)
+                    .expect("AST string field was validated as bytes"),
+            ),
+        )
+    }
+}
+
+fn runtime_string_to_object(
+    vm: &VirtualMachine,
+    value: Option<Box<str>>,
+    bytes: Option<Vec<u8>>,
+) -> Option<PyObjectRef> {
+    if let Some(bytes) = bytes {
+        Some(vm.ctx.new_bytes(bytes).into())
+    } else {
+        value.map(|value| vm.ctx.new_str(value).into())
+    }
+}
+
+fn first_invalid_constant_type(vm: &VirtualMachine, value_object: PyObjectRef) -> PyResult<String> {
+    let cls = value_object.class();
+    let class_name = cls.name().to_owned();
+    if cls.is(vm.ctx.types.tuple_type) {
+        vm.with_recursion(" during compilation", || {
+            let tuple = value_object.clone().downcast::<PyTuple>().map_err(|obj| {
+                vm.new_type_error(format!(
+                    "Expected type {}, not {}",
+                    PyTuple::static_type().name(),
+                    obj.class().name()
+                ))
+            })?;
+            for item in tuple.iter() {
+                if let Some(invalid_type) = first_invalid_constant_type_opt(vm, item.clone())? {
+                    return Ok(invalid_type);
+                }
+            }
+            Ok(class_name)
+        })
+    } else if cls.is(vm.ctx.types.frozenset_type) {
+        vm.with_recursion(" during compilation", || {
+            let set = value_object.clone().downcast::<PyFrozenSet>().unwrap();
+            for item in set.elements() {
+                if let Some(invalid_type) = first_invalid_constant_type_opt(vm, item)? {
+                    return Ok(invalid_type);
+                }
+            }
+            Ok(class_name)
+        })
+    } else {
+        Ok(class_name)
+    }
+}
+
+fn first_invalid_constant_type_opt(
+    vm: &VirtualMachine,
+    value_object: PyObjectRef,
+) -> PyResult<Option<String>> {
+    let cls = value_object.class();
+    if cls.is(vm.ctx.types.none_type)
+        || cls.is(vm.ctx.types.bool_type)
+        || cls.is(vm.ctx.types.str_type)
+        || cls.is(vm.ctx.types.bytes_type)
+        || cls.is(vm.ctx.types.int_type)
+        || cls.is(vm.ctx.types.float_type)
+        || cls.is(vm.ctx.types.complex_type)
+        || cls.is(vm.ctx.types.ellipsis_type)
+    {
+        return Ok(None);
+    }
+    if cls.is(vm.ctx.types.tuple_type) || cls.is(vm.ctx.types.frozenset_type) {
+        return first_invalid_constant_type(vm, value_object).map(Some);
+    }
+    Ok(Some(cls.name().to_owned()))
+}
+
+fn constant_data_to_object(vm: &VirtualMachine, constant: ConstantData) -> PyObjectRef {
+    match constant {
+        ConstantData::None => vm.ctx.none(),
+        ConstantData::Boolean { value } => vm.ctx.new_bool(value).to_pyobject(vm),
+        ConstantData::Str { value } => vm.ctx.new_str(value.to_string()).to_pyobject(vm),
+        ConstantData::Bytes { value } => vm.ctx.new_bytes(value).to_pyobject(vm),
+        ConstantData::Integer { value } => vm.ctx.new_int(value).into(),
+        ConstantData::Tuple { elements } => {
+            let value = elements
+                .into_iter()
+                .map(|c| constant_data_to_object(vm, c))
+                .collect();
+            vm.ctx.new_tuple(value).to_pyobject(vm)
+        }
+        ConstantData::Frozenset { elements } => PyFrozenSet::from_iter(
+            vm,
+            elements.into_iter().map(|c| constant_data_to_object(vm, c)),
+        )
+        .unwrap()
+        .into_pyobject(vm),
+        ConstantData::Float { value } => vm.ctx.new_float(value).into_pyobject(vm),
+        ConstantData::Complex { value } => vm.ctx.new_complex(value).into_pyobject(vm),
+        ConstantData::Ellipsis => vm.ctx.ellipsis.clone().into(),
+        ConstantData::Code { .. } | ConstantData::Slice { .. } => {
+            unreachable!("ast.Constant values cannot contain code objects or slices")
+        }
+    }
+}
+
 // constructor
+pub(super) fn constant_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<Constant> {
+    let value_object = get_node_field(vm, &object, "value", "Constant")?;
+    let (value, invalid_type) =
+        match ConstantLiteral::ast_from_object(vm, source_file, value_object.clone()) {
+            Ok(value) => (value, None),
+            Err(_) => (
+                ConstantLiteral::None,
+                Some(first_invalid_constant_type(vm, value_object)?),
+            ),
+        };
+    let kind = get_node_field_opt(vm, &object, "kind")?
+        .map(|object| {
+            if !object.class().is(vm.ctx.types.str_type) {
+                return Err(vm.new_type_error("AST string must be of type str"));
+            }
+            Ok(object.try_to_value::<String>(vm)?.into_boxed_str())
+        })
+        .transpose()?;
+
+    Ok(Constant {
+        range,
+        value,
+        kind,
+        invalid_type,
+    })
+}
+
 impl Node for Constant {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { range, value } = self;
+        let Self {
+            range,
+            value,
+            kind,
+            invalid_type: _,
+        } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeExprConstant::static_type().to_owned())
             .unwrap();
-        let kind = match &value {
-            ConstantLiteral::Str {
-                prefix: StringLiteralPrefix::Unicode,
-                ..
-            } => vm.ctx.new_str("u").into(),
-            _ => vm.ctx.none(),
-        };
+        let kind = kind
+            .or_else(|| constant_literal_kind(&value))
+            .map_or_else(|| vm.ctx.none(), |kind| vm.ctx.new_str(kind).into());
         let value = value.ast_to_object(vm, source_file);
         let dict = node.as_object().dict().unwrap();
         dict.set_item("value", value, vm).unwrap();
@@ -123,16 +495,8 @@ impl Node for Constant {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let value_object = get_node_field(vm, &object, "value", "Constant")?;
-        let value = Node::ast_from_object(vm, source_file, value_object)?;
-
-        Ok(Self {
-            value,
-            // kind: get_node_field_opt(_vm, &_object, "kind")?
-            //     .map(|obj| Node::ast_from_object(_vm, obj))
-            //     .transpose()?,
-            range: range_from_object(vm, source_file, object, "Constant")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "Constant")?;
+        constant_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -201,8 +565,12 @@ impl Node for ConstantLiteral {
             })?;
             let tuple = tuple
                 .into_iter()
-                .cloned()
-                .map(|object| Node::ast_from_object(vm, source_file, object))
+                .map(|object| {
+                    let object = object.clone();
+                    vm.with_recursion(" during compilation", || {
+                        Node::ast_from_object(vm, source_file, object)
+                    })
+                })
                 .collect::<PyResult<_>>()?;
             Self::Tuple(tuple)
         } else if cls.is(vm.ctx.types.frozenset_type) {
@@ -210,7 +578,11 @@ impl Node for ConstantLiteral {
             let elements = set
                 .elements()
                 .into_iter()
-                .map(|object| Node::ast_from_object(vm, source_file, object))
+                .map(|object| {
+                    vm.with_recursion(" during compilation", || {
+                        Node::ast_from_object(vm, source_file, object)
+                    })
+                })
                 .collect::<PyResult<_>>()?;
             Self::FrozenSet(elements)
         } else if cls.is(vm.ctx.types.float_type) {
@@ -244,117 +616,6 @@ impl Node for ConstantLiteral {
     }
 }
 
-fn constant_to_ruff_expr(value: Constant) -> ast::Expr {
-    let Constant { value, range } = value;
-    match value {
-        ConstantLiteral::None => ast::Expr::NoneLiteral(ast::ExprNoneLiteral {
-            node_index: Default::default(),
-            range,
-        }),
-        ConstantLiteral::Bool(value) => ast::Expr::BooleanLiteral(ast::ExprBooleanLiteral {
-            node_index: Default::default(),
-            range,
-            value,
-        }),
-        ConstantLiteral::Str { value, prefix } => {
-            ast::Expr::StringLiteral(ast::ExprStringLiteral {
-                node_index: Default::default(),
-                range,
-                value: ast::StringLiteralValue::single(ast::StringLiteral {
-                    node_index: Default::default(),
-                    range,
-                    value,
-                    flags: ast::StringLiteralFlags::empty().with_prefix(prefix),
-                }),
-            })
-        }
-        ConstantLiteral::Bytes(value) => {
-            ast::Expr::BytesLiteral(ast::ExprBytesLiteral {
-                node_index: Default::default(),
-                range,
-                value: ast::BytesLiteralValue::single(ast::BytesLiteral {
-                    node_index: Default::default(),
-                    range,
-                    value,
-                    flags: ast::BytesLiteralFlags::empty(), // TODO
-                }),
-            })
-        }
-        ConstantLiteral::Int(value) => ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
-            node_index: Default::default(),
-            range,
-            value: ast::Number::Int(value),
-        }),
-        ConstantLiteral::Tuple(value) => ast::Expr::Tuple(ast::ExprTuple {
-            node_index: Default::default(),
-            range,
-            elts: value
-                .into_iter()
-                .map(|value| {
-                    constant_to_ruff_expr(Constant {
-                        range: TextRange::default(),
-                        value,
-                    })
-                })
-                .collect(),
-            ctx: ast::ExprContext::Load,
-            // TODO: Does this matter?
-            parenthesized: true,
-        }),
-        ConstantLiteral::FrozenSet(value) => {
-            let args = if value.is_empty() {
-                Vec::new()
-            } else {
-                vec![ast::Expr::Set(ast::ExprSet {
-                    node_index: Default::default(),
-                    range: TextRange::default(),
-                    elts: value
-                        .into_iter()
-                        .map(|value| {
-                            constant_to_ruff_expr(Constant {
-                                range: TextRange::default(),
-                                value,
-                            })
-                        })
-                        .collect(),
-                })]
-            };
-            ast::Expr::Call(ast::ExprCall {
-                node_index: Default::default(),
-                range,
-                func: Box::new(ast::Expr::Name(ast::ExprName {
-                    node_index: Default::default(),
-                    range: TextRange::default(),
-                    id: ast::name::Name::new_static("frozenset"),
-                    ctx: ast::ExprContext::Load,
-                })),
-                arguments: ast::Arguments {
-                    node_index: Default::default(),
-                    range,
-                    args: args.into(),
-                    keywords: Box::default(),
-                },
-            })
-        }
-        ConstantLiteral::Float(value) => ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
-            node_index: Default::default(),
-            range,
-            value: ast::Number::Float(value),
-        }),
-        ConstantLiteral::Complex { real, imag } => {
-            ast::Expr::NumberLiteral(ast::ExprNumberLiteral {
-                node_index: Default::default(),
-                range,
-                value: ast::Number::Complex { real, imag },
-            })
-        }
-        ConstantLiteral::Ellipsis => ast::Expr::EllipsisLiteral(ast::ExprEllipsisLiteral {
-            node_index: Default::default(),
-            range,
-        }),
-    }
-}
-
 pub(super) fn number_literal_to_object(
     vm: &VirtualMachine,
     source_file: &SourceFile,
@@ -364,6 +625,7 @@ pub(super) fn number_literal_to_object(
         node_index: _,
         range,
         value,
+        ..
     } = constant;
     let c = match value {
         ast::Number::Int(n) => Constant::new_int(n, range),
@@ -382,6 +644,7 @@ pub(super) fn string_literal_to_object(
         node_index: _,
         range,
         value,
+        ..
     } = constant;
     let prefix = value
         .iter()
@@ -400,6 +663,7 @@ pub(super) fn bytes_literal_to_object(
         node_index: _,
         range,
         value,
+        ..
     } = constant;
     let bytes = value.as_slice().iter().flat_map(|b| b.value.iter());
     let c = Constant::new_bytes(bytes.copied().collect(), range);
@@ -415,6 +679,7 @@ pub(super) fn boolean_literal_to_object(
         node_index: _,
         range,
         value,
+        ..
     } = constant;
     let c = Constant::new_bool(value, range);
     c.ast_to_object(vm, source_file)
@@ -428,6 +693,7 @@ pub(super) fn none_literal_to_object(
     let ast::ExprNoneLiteral {
         node_index: _,
         range,
+        ..
     } = constant;
     let c = Constant::new_none(range);
     c.ast_to_object(vm, source_file)
@@ -441,6 +707,7 @@ pub(super) fn ellipsis_literal_to_object(
     let ast::ExprEllipsisLiteral {
         node_index: _,
         range,
+        ..
     } = constant;
     let c = Constant::new_ellipsis(range);
     c.ast_to_object(vm, source_file)

crates/vm/src/stdlib/_ast/elif_else_clause.rs

@@ -12,10 +12,15 @@ pub(super) fn ast_to_object(
         range,
         test,
         body,
+        runtime_body,
+        runtime_orelse,
     } = clause;
     let Some(test) = test else {
         assert!(rest.len() == 0);
-        return body.ast_to_object(vm, source_file);
+        return runtime_body.map_or_else(
+            || body.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
     };
     let node = NodeAst
         .into_ref_with_type(vm, pyast::NodeStmtIf::static_type().to_owned())
@@ -24,10 +29,15 @@ pub(super) fn ast_to_object(
 
     dict.set_item("test", test.ast_to_object(vm, source_file), vm)
         .unwrap();
-    dict.set_item("body", body.ast_to_object(vm, source_file), vm)
-        .unwrap();
+    let body = runtime_body.map_or_else(
+        || body.ast_to_object(vm, source_file),
+        |values| values.ast_to_object(vm, source_file),
+    );
+    dict.set_item("body", body, vm).unwrap();
 
-    let orelse = if let Some(next) = rest.next() {
+    let orelse = if let Some(values) = runtime_orelse {
+        values.ast_to_object(vm, source_file)
+    } else if let Some(next) = rest.next() {
         if next.test.is_some() {
             let next = ast::ElifElseClause {
                 range: TextRange::new(next.range.start(), range.end()),
@@ -37,7 +47,10 @@ pub(super) fn ast_to_object(
                 .new_list(vec![ast_to_object(next, rest, vm, source_file)])
                 .into()
         } else {
-            next.body.ast_to_object(vm, source_file)
+            next.runtime_body.map_or_else(
+                || next.body.ast_to_object(vm, source_file),
+                |values| values.ast_to_object(vm, source_file),
+            )
         }
     } else {
         vm.ctx.new_list(vec![]).into()
@@ -48,40 +61,45 @@ pub(super) fn ast_to_object(
     node.into()
 }
 
-pub(super) fn ast_from_object(
+pub(super) fn ast_from_object_with_range(
     vm: &VirtualMachine,
     source_file: &SourceFile,
     object: PyObjectRef,
+    range: TextRange,
 ) -> PyResult<ast::StmtIf> {
-    let test = Node::ast_from_object(vm, source_file, get_node_field(vm, &object, "test", "If")?)?;
-    let body = Node::ast_from_object(vm, source_file, get_node_field(vm, &object, "body", "If")?)?;
-    let orelse: Vec<ast::Stmt> = Node::ast_from_object(
-        vm,
-        source_file,
-        get_node_field(vm, &object, "orelse", "If")?,
-    )?;
-    let range = range_from_object(vm, source_file, object, "If")?;
+    let test = get_required_node_field(vm, source_file, &object, "test", "If")?;
+    let body: Vec<Option<ast::Stmt>> = get_node_list_field(vm, source_file, &object, "body", "If")?;
+    let orelse: Vec<Option<ast::Stmt>> =
+        get_node_list_field(vm, source_file, &object, "orelse", "If")?;
+    let runtime_body = runtime_stmt_list_metadata(&body);
+    let runtime_orelse = runtime_stmt_list_metadata(&orelse);
+    let body = lower_runtime_stmt_list(body);
+    let orelse = lower_runtime_stmt_list(orelse);
 
     let elif_else_clauses = if orelse.is_empty() {
         vec![]
     } else if let [ast::Stmt::If(_)] = &*orelse {
         let Some(ast::Stmt::If(ast::StmtIf {
-            node_index: _,
+            node_index,
             range,
             test,
             body,
             mut elif_else_clauses,
+            runtime_body,
         })) = orelse.into_iter().next()
         else {
             unreachable!()
         };
+        debug_assert!(runtime_orelse.is_none());
         elif_else_clauses.insert(
             0,
             ast::ElifElseClause {
-                node_index: Default::default(),
+                node_index,
                 range,
                 test: Some(*test),
                 body,
+                runtime_body,
+                runtime_orelse: None,
             },
         );
         elif_else_clauses
@@ -91,6 +109,8 @@ pub(super) fn ast_from_object(
             range,
             test: None,
             body: orelse,
+            runtime_body: runtime_orelse,
+            runtime_orelse: None,
         }]
     };
 
@@ -100,5 +120,6 @@ pub(super) fn ast_from_object(
         body,
         elif_else_clauses,
         range,
+        runtime_body,
     })
 }

crates/vm/src/stdlib/_ast/exception.rs

@@ -1,6 +1,22 @@
 use super::*;
 use rustpython_compiler_core::SourceFile;
 
+fn ensure_excepthandler_node(vm: &VirtualMachine, object: &PyObjectRef) -> PyResult<()> {
+    if vm.is_none(object)
+        || !is_node_instance(
+            vm,
+            object,
+            pyast::NodeExceptHandlerExceptHandler::static_type(),
+        )?
+    {
+        return Err(vm.new_type_error(format!(
+            "expected some sort of excepthandler, but got {}",
+            object.repr(vm)?
+        )));
+    }
+    Ok(())
+}
+
 // sum
 impl Node for ast::ExceptHandler {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
@@ -13,25 +29,53 @@ impl Node for ast::ExceptHandler {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(
-            if cls.is(pyast::NodeExceptHandlerExceptHandler::static_type()) {
-                Self::ExceptHandler(ast::ExceptHandlerExceptHandler::ast_from_object(
-                    vm,
-                    source_file,
-                    object,
-                )?)
-            } else {
-                return Err(vm.new_type_error(format!(
-                    "expected some sort of excepthandler, but got {}",
-                    object.repr(vm)?
-                )));
-            },
-        )
+        ensure_excepthandler_node(vm, &object)?;
+        let range = excepthandler_range_from_object(vm, source_file, object.clone())?;
+        Ok(Self::ExceptHandler(except_handler_from_object_with_range(
+            vm,
+            source_file,
+            object,
+            range,
+        )?))
     }
 }
 
 // constructor
+fn except_handler_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::ExceptHandlerExceptHandler> {
+    let body: Vec<Option<ast::Stmt>> =
+        get_node_list_field(vm, source_file, &object, "body", "ExceptHandler")?;
+    let (runtime_body, body) = runtime_stmt_list_from_values(body);
+    Ok(ast::ExceptHandlerExceptHandler {
+        node_index: Default::default(),
+        type_: get_node_field_opt(vm, &object, "type")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        name: get_node_field_opt(vm, &object, "name")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        body,
+        range,
+        runtime_body,
+    })
+}
+
+pub(super) fn except_handler_from_object_unvalidated_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+) -> PyResult<ast::ExceptHandler> {
+    ensure_excepthandler_node(vm, &object)?;
+    let range = excepthandler_range_from_object_unvalidated(vm, source_file, object.clone())?;
+    Ok(ast::ExceptHandler::ExceptHandler(
+        except_handler_from_object_with_range(vm, source_file, object, range)?,
+    ))
+}
+
 impl Node for ast::ExceptHandlerExceptHandler {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -40,6 +84,7 @@ impl Node for ast::ExceptHandlerExceptHandler {
             name,
             body,
             range,
+            runtime_body,
         } = self;
         let node = NodeAst
             .into_ref_with_type(
@@ -52,8 +97,11 @@ impl Node for ast::ExceptHandlerExceptHandler {
             .unwrap();
         dict.set_item("name", name.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("body", body.ast_to_object(vm, source_file), vm)
-            .unwrap();
+        let body = runtime_body.map_or_else(
+            || body.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("body", body, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -63,20 +111,7 @@ impl Node for ast::ExceptHandlerExceptHandler {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            type_: get_node_field_opt(vm, &object, "type")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            name: get_node_field_opt(vm, &object, "name")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            body: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "body", "ExceptHandler")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "ExceptHandler")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "ExceptHandler")?;
+        except_handler_from_object_with_range(vm, source_file, object, range)
     }
 }

crates/vm/src/stdlib/_ast/module.rs

@@ -18,7 +18,7 @@ use rustpython_compiler_core::SourceFile;
 /// - `FunctionType`: A function signature with argument and return type
 ///   annotations, representing the type hints of a function (e.g., `def add(x: int, y: int) -> int`).
 pub(super) enum Mod {
-    Module(ast::ModModule),
+    Module(ModModule),
     Interactive(ModInteractive),
     Expression(ast::ModExpression),
     FunctionType(ModFunctionType),
@@ -40,53 +40,66 @@ impl Node for Mod {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeModModule::static_type()) {
-            Self::Module(ast::ModModule::ast_from_object(vm, source_file, object)?)
-        } else if cls.is(pyast::NodeModInteractive::static_type()) {
-            Self::Interactive(ModInteractive::ast_from_object(vm, source_file, object)?)
-        } else if cls.is(pyast::NodeModExpression::static_type()) {
-            Self::Expression(ast::ModExpression::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodeModFunctionType::static_type()) {
-            Self::FunctionType(ModFunctionType::ast_from_object(vm, source_file, object)?)
-        } else {
-            return Err(vm.new_type_error(format!(
-                "expected some sort of mod, but got {}",
-                object.repr(vm)?
-            )));
-        })
+        Ok(
+            if object.is_instance(pyast::NodeModModule::static_type().as_object(), vm)? {
+                Self::Module(ModModule::ast_from_object(vm, source_file, object)?)
+            } else if object
+                .is_instance(pyast::NodeModInteractive::static_type().as_object(), vm)?
+            {
+                Self::Interactive(ModInteractive::ast_from_object(vm, source_file, object)?)
+            } else if object.is_instance(pyast::NodeModExpression::static_type().as_object(), vm)? {
+                Self::Expression(ast::ModExpression::ast_from_object(
+                    vm,
+                    source_file,
+                    object,
+                )?)
+            } else if object
+                .is_instance(pyast::NodeModFunctionType::static_type().as_object(), vm)?
+            {
+                Self::FunctionType(ModFunctionType::ast_from_object(vm, source_file, object)?)
+            } else {
+                return Err(vm.new_type_error(format!(
+                    "expected some sort of mod, but got {}",
+                    object.repr(vm)?
+                )));
+            },
+        )
     }
 }
 
+pub(super) struct ModModule {
+    pub(crate) module: ast::ModModule,
+    pub(crate) type_ignores: Vec<TypeIgnore>,
+}
+
 // constructor
-impl Node for ast::ModModule {
+impl Node for ModModule {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
+            module,
+            type_ignores,
+        } = self;
+        let ast::ModModule {
             node_index: _,
             body,
-            // type_ignores,
-            range,
-        } = self;
+            range: _,
+            runtime_body,
+        } = module;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeModModule::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("body", body.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        // TODO: Improve ruff API
-        // ruff ignores type_ignore comments currently.
-        let type_ignores: Vec<TypeIgnore> = vec![];
+        let body = runtime_body.map_or_else(
+            || body.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("body", body, vm).unwrap();
         dict.set_item(
             "type_ignores",
             type_ignores.ast_to_object(vm, source_file),
             vm,
         )
         .unwrap();
-        let _ = range;
         node.into()
     }
 
@@ -95,38 +108,45 @@ impl Node for ast::ModModule {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
+        let body: Vec<Option<ast::Stmt>> =
+            get_node_list_field(vm, source_file, &object, "body", "Module")?;
+        let (runtime_body, body) = runtime_stmt_list_from_values(body);
+        let type_ignores = get_node_list_field(vm, source_file, &object, "type_ignores", "Module")?;
         Ok(Self {
-            node_index: Default::default(),
-            body: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "body", "Module")?,
-            )?,
-            // type_ignores: Node::ast_from_object(
-            //     _vm,
-            //     get_node_field(_vm, &_object, "type_ignores", "Module")?,
-            // )?,
-            range: Default::default(),
+            module: ast::ModModule {
+                node_index: Default::default(),
+                body,
+                range: Default::default(),
+                runtime_body,
+            },
+            type_ignores,
         })
     }
 }
 
 pub(super) struct ModInteractive {
     pub(crate) range: TextRange,
-    pub(crate) body: Vec<ast::Stmt>,
+    pub(crate) body: ast::Suite,
+    pub(crate) runtime_body: Option<Vec<Option<ast::Stmt>>>,
 }
 
 // constructor
 impl Node for ModInteractive {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { body, range } = self;
+        let Self {
+            body,
+            range: _,
+            runtime_body,
+        } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeModInteractive::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("body", body.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        let _ = range;
+        let body = runtime_body.map_or_else(
+            || body.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("body", body, vm).unwrap();
         node.into()
     }
 
@@ -135,13 +155,13 @@ impl Node for ModInteractive {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
+        let body: Vec<Option<ast::Stmt>> =
+            get_node_list_field(vm, source_file, &object, "body", "Interactive")?;
+        let (runtime_body, body) = runtime_stmt_list_from_values(body);
         Ok(Self {
-            body: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "body", "Interactive")?,
-            )?,
+            body,
             range: Default::default(),
+            runtime_body,
         })
     }
 }
@@ -152,7 +172,7 @@ impl Node for ast::ModExpression {
         let Self {
             node_index: _,
             body,
-            range,
+            range: _,
         } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeModExpression::static_type().to_owned())
@@ -160,7 +180,6 @@ impl Node for ast::ModExpression {
         let dict = node.as_object().dict().unwrap();
         dict.set_item("body", body.ast_to_object(vm, source_file), vm)
             .unwrap();
-        let _ = range;
         node.into()
     }
 
@@ -171,11 +190,7 @@ impl Node for ast::ModExpression {
     ) -> PyResult<Self> {
         Ok(Self {
             node_index: Default::default(),
-            body: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "body", "Expression")?,
-            )?,
+            body: get_required_node_field(vm, source_file, &object, "body", "Expression")?,
             range: Default::default(),
         })
     }
@@ -184,7 +199,7 @@ impl Node for ast::ModExpression {
 pub(super) struct ModFunctionType {
     pub(crate) argtypes: Box<[ast::Expr]>,
     pub(crate) returns: ast::Expr,
-    pub(crate) range: TextRange,
+    pub(crate) runtime_argtypes: Option<Vec<Option<ast::Expr>>>,
 }
 
 // constructor
@@ -193,21 +208,19 @@ impl Node for ModFunctionType {
         let Self {
             argtypes,
             returns,
-            range,
+            runtime_argtypes,
         } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeModFunctionType::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item(
-            "argtypes",
-            BoxedSlice(argtypes).ast_to_object(vm, source_file),
-            vm,
-        )
-        .unwrap();
+        let argtypes = runtime_argtypes.map_or_else(
+            || BoxedSlice(argtypes).ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("argtypes", argtypes, vm).unwrap();
         dict.set_item("returns", returns.ast_to_object(vm, source_file), vm)
             .unwrap();
-        let _ = range;
         node.into()
     }
 
@@ -216,21 +229,13 @@ impl Node for ModFunctionType {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
+        let argtypes: Vec<Option<ast::Expr>> =
+            get_node_list_field(vm, source_file, &object, "argtypes", "FunctionType")?;
+        let (runtime_argtypes, argtypes) = runtime_expr_list_from_values(argtypes);
         Ok(Self {
-            argtypes: {
-                let argtypes: BoxedSlice<_> = Node::ast_from_object(
-                    vm,
-                    source_file,
-                    get_node_field(vm, &object, "argtypes", "FunctionType")?,
-                )?;
-                argtypes.0
-            },
-            returns: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "returns", "FunctionType")?,
-            )?,
-            range: Default::default(),
+            argtypes: argtypes.into_boxed_slice(),
+            returns: get_required_node_field(vm, source_file, &object, "returns", "FunctionType")?,
+            runtime_argtypes,
         })
     }
 }

crates/vm/src/stdlib/_ast/node.rs

@@ -1,5 +1,6 @@
-use crate::{PyObjectRef, PyResult, VirtualMachine};
+use crate::{PyObjectRef, PyResult, VirtualMachine, builtins::PyList};
 use rustpython_compiler_core::SourceFile;
+use thin_vec::ThinVec;
 
 pub(crate) trait Node: Sized {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef;
@@ -31,14 +32,52 @@ impl<T: Node> Node for Vec<T> {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        // Recursion guard for each element: prevents stack overflow when a
-        // sequence element transitively references the sequence itself
-        // (e.g. `l = ast.List(...); l.elts = [l]`). See issue #4862.
-        vm.extract_elements_with(&object, |obj| {
-            vm.with_recursion("while traversing AST node", || {
-                Node::ast_from_object(vm, source_file, obj)
-            })
-        })
+        let list = object.downcast_ref::<PyList>().ok_or_else(|| {
+            vm.new_type_error(format!(
+                "AST list field must be a list, not a {}",
+                object.class().name()
+            ))
+        })?;
+        let len = list.borrow_vec().len();
+        let mut result = Self::with_capacity(len);
+        for i in 0..len {
+            let item = {
+                let items = list.borrow_vec();
+                if items.len() != len {
+                    return Err(
+                        vm.new_runtime_error("AST list field changed size during iteration")
+                    );
+                }
+                items[i].clone()
+            };
+            result.push(vm.with_recursion("while traversing AST node", || {
+                Node::ast_from_object(vm, source_file, item)
+            })?);
+            if list.borrow_vec().len() != len {
+                return Err(vm.new_runtime_error("AST list field changed size during iteration"));
+            }
+        }
+        Ok(result)
+    }
+}
+
+impl<T: Node> Node for ThinVec<T> {
+    fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
+        vm.ctx
+            .new_list(
+                self.into_iter()
+                    .map(|node| node.ast_to_object(vm, source_file))
+                    .collect(),
+            )
+            .into()
+    }
+
+    fn ast_from_object(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: PyObjectRef,
+    ) -> PyResult<Self> {
+        Vec::<T>::ast_from_object(vm, source_file, object).map(Into::into)
     }
 }
 
@@ -52,10 +91,6 @@ impl<T: Node> Node for Box<T> {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        // Recursion guard: every descent through a Box<AstNode> increments the
-        // VM's recursion depth so cyclic or pathologically deep ASTs raise
-        // RecursionError instead of overflowing the native stack.
-        // See issue #4862.
         vm.with_recursion("while traversing AST node", || {
             T::ast_from_object(vm, source_file, object).map(Self::new)
         })

crates/vm/src/stdlib/_ast/operator.rs

@@ -16,17 +16,18 @@ impl Node for ast::BoolOp {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeBoolOpAnd::static_type()) {
-            Self::And
-        } else if cls.is(pyast::NodeBoolOpOr::static_type()) {
-            Self::Or
-        } else {
-            return Err(vm.new_type_error(format!(
-                "expected some sort of boolop, but got {}",
-                object.repr(vm)?
-            )));
-        })
+        Ok(
+            if is_node_instance(vm, &object, pyast::NodeBoolOpAnd::static_type())? {
+                Self::And
+            } else if is_node_instance(vm, &object, pyast::NodeBoolOpOr::static_type())? {
+                Self::Or
+            } else {
+                return Err(vm.new_type_error(format!(
+                    "expected some sort of boolop, but got {}",
+                    object.repr(vm)?
+                )));
+            },
+        )
     }
 }
 
@@ -56,39 +57,40 @@ impl Node for ast::Operator {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeOperatorAdd::static_type()) {
-            Self::Add
-        } else if cls.is(pyast::NodeOperatorSub::static_type()) {
-            Self::Sub
-        } else if cls.is(pyast::NodeOperatorMult::static_type()) {
-            Self::Mult
-        } else if cls.is(pyast::NodeOperatorMatMult::static_type()) {
-            Self::MatMult
-        } else if cls.is(pyast::NodeOperatorDiv::static_type()) {
-            Self::Div
-        } else if cls.is(pyast::NodeOperatorMod::static_type()) {
-            Self::Mod
-        } else if cls.is(pyast::NodeOperatorPow::static_type()) {
-            Self::Pow
-        } else if cls.is(pyast::NodeOperatorLShift::static_type()) {
-            Self::LShift
-        } else if cls.is(pyast::NodeOperatorRShift::static_type()) {
-            Self::RShift
-        } else if cls.is(pyast::NodeOperatorBitOr::static_type()) {
-            Self::BitOr
-        } else if cls.is(pyast::NodeOperatorBitXor::static_type()) {
-            Self::BitXor
-        } else if cls.is(pyast::NodeOperatorBitAnd::static_type()) {
-            Self::BitAnd
-        } else if cls.is(pyast::NodeOperatorFloorDiv::static_type()) {
-            Self::FloorDiv
-        } else {
-            return Err(vm.new_type_error(format!(
-                "expected some sort of operator, but got {}",
-                object.repr(vm)?
-            )));
-        })
+        Ok(
+            if is_node_instance(vm, &object, pyast::NodeOperatorAdd::static_type())? {
+                Self::Add
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorSub::static_type())? {
+                Self::Sub
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorMult::static_type())? {
+                Self::Mult
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorMatMult::static_type())? {
+                Self::MatMult
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorDiv::static_type())? {
+                Self::Div
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorMod::static_type())? {
+                Self::Mod
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorPow::static_type())? {
+                Self::Pow
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorLShift::static_type())? {
+                Self::LShift
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorRShift::static_type())? {
+                Self::RShift
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorBitOr::static_type())? {
+                Self::BitOr
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorBitXor::static_type())? {
+                Self::BitXor
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorBitAnd::static_type())? {
+                Self::BitAnd
+            } else if is_node_instance(vm, &object, pyast::NodeOperatorFloorDiv::static_type())? {
+                Self::FloorDiv
+            } else {
+                return Err(vm.new_type_error(format!(
+                    "expected some sort of operator, but got {}",
+                    object.repr(vm)?
+                )));
+            },
+        )
     }
 }
 
@@ -109,21 +111,22 @@ impl Node for ast::UnaryOp {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeUnaryOpInvert::static_type()) {
-            Self::Invert
-        } else if cls.is(pyast::NodeUnaryOpNot::static_type()) {
-            Self::Not
-        } else if cls.is(pyast::NodeUnaryOpUAdd::static_type()) {
-            Self::UAdd
-        } else if cls.is(pyast::NodeUnaryOpUSub::static_type()) {
-            Self::USub
-        } else {
-            return Err(vm.new_type_error(format!(
-                "expected some sort of unaryop, but got {}",
-                object.repr(vm)?
-            )));
-        })
+        Ok(
+            if is_node_instance(vm, &object, pyast::NodeUnaryOpInvert::static_type())? {
+                Self::Invert
+            } else if is_node_instance(vm, &object, pyast::NodeUnaryOpNot::static_type())? {
+                Self::Not
+            } else if is_node_instance(vm, &object, pyast::NodeUnaryOpUAdd::static_type())? {
+                Self::UAdd
+            } else if is_node_instance(vm, &object, pyast::NodeUnaryOpUSub::static_type())? {
+                Self::USub
+            } else {
+                return Err(vm.new_type_error(format!(
+                    "expected some sort of unaryop, but got {}",
+                    object.repr(vm)?
+                )));
+            },
+        )
     }
 }
 
@@ -150,32 +153,33 @@ impl Node for ast::CmpOp {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeCmpOpEq::static_type()) {
-            Self::Eq
-        } else if cls.is(pyast::NodeCmpOpNotEq::static_type()) {
-            Self::NotEq
-        } else if cls.is(pyast::NodeCmpOpLt::static_type()) {
-            Self::Lt
-        } else if cls.is(pyast::NodeCmpOpLtE::static_type()) {
-            Self::LtE
-        } else if cls.is(pyast::NodeCmpOpGt::static_type()) {
-            Self::Gt
-        } else if cls.is(pyast::NodeCmpOpGtE::static_type()) {
-            Self::GtE
-        } else if cls.is(pyast::NodeCmpOpIs::static_type()) {
-            Self::Is
-        } else if cls.is(pyast::NodeCmpOpIsNot::static_type()) {
-            Self::IsNot
-        } else if cls.is(pyast::NodeCmpOpIn::static_type()) {
-            Self::In
-        } else if cls.is(pyast::NodeCmpOpNotIn::static_type()) {
-            Self::NotIn
-        } else {
-            return Err(vm.new_type_error(format!(
-                "expected some sort of cmpop, but got {}",
-                object.repr(vm)?
-            )));
-        })
+        Ok(
+            if is_node_instance(vm, &object, pyast::NodeCmpOpEq::static_type())? {
+                Self::Eq
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpNotEq::static_type())? {
+                Self::NotEq
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpLt::static_type())? {
+                Self::Lt
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpLtE::static_type())? {
+                Self::LtE
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpGt::static_type())? {
+                Self::Gt
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpGtE::static_type())? {
+                Self::GtE
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpIs::static_type())? {
+                Self::Is
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpIsNot::static_type())? {
+                Self::IsNot
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpIn::static_type())? {
+                Self::In
+            } else if is_node_instance(vm, &object, pyast::NodeCmpOpNotIn::static_type())? {
+                Self::NotIn
+            } else {
+                return Err(vm.new_type_error(format!(
+                    "expected some sort of cmpop, but got {}",
+                    object.repr(vm)?
+                )));
+            },
+        )
     }
 }

crates/vm/src/stdlib/_ast/other.rs

@@ -11,14 +11,12 @@ impl Node for ast::ConversionFlag {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        // Python's AST uses ASCII codes: 's', 'r', 'a', -1=None
-        // Note: 255 is -1i8 as u8 (ruff's ConversionFlag::None)
-        match i32::try_from_object(vm, object)? {
-            -1 | 255 => Ok(Self::None),
+        match node_object_to_i32(vm, object)? {
+            -1 => Ok(Self::None),
             x if x == b's' as i32 => Ok(Self::Str),
             x if x == b'r' as i32 => Ok(Self::Repr),
             x if x == b'a' as i32 => Ok(Self::Ascii),
-            _ => Err(vm.new_value_error("invalid conversion flag")),
+            x => Err(vm.new_system_error(format!("Unrecognized conversion character {x}"))),
         }
     }
 }
@@ -34,10 +32,13 @@ impl Node for ast::name::Name {
         _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        match object.downcast::<PyStr>() {
-            Ok(name) => Ok(Self::new(name)),
-            Err(_) => Err(vm.new_value_error("expected str for name")),
+        if !object.class().is(vm.ctx.types.str_type) {
+            return Err(vm.new_type_error("AST identifier must be of type str"));
         }
+        object
+            .downcast::<PyStr>()
+            .map(Self::new)
+            .map_err(|_| vm.new_type_error("AST identifier must be of type str"))
     }
 }
 
@@ -89,11 +90,7 @@ impl Node for ast::Alias {
     ) -> PyResult<Self> {
         Ok(Self {
             node_index: Default::default(),
-            name: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field_required(vm, &object, "name", "alias")?,
-            )?,
+            name: get_required_identifier_field(vm, source_file, &object, "name", "alias")?,
             asname: get_node_field_opt(vm, &object, "asname")?
                 .map(|obj| Node::ast_from_object(vm, source_file, obj))
                 .transpose()?,
@@ -137,10 +134,12 @@ impl Node for ast::WithItem {
     ) -> PyResult<Self> {
         Ok(Self {
             node_index: Default::default(),
-            context_expr: Node::ast_from_object(
+            context_expr: get_required_node_field(
                 vm,
                 source_file,
-                get_node_field_required(vm, &object, "context_expr", "withitem")?,
+                &object,
+                "context_expr",
+                "withitem",
             )?,
             optional_vars: get_node_field_opt(vm, &object, "optional_vars")?
                 .map(|obj| Node::ast_from_object(vm, source_file, obj))

crates/vm/src/stdlib/_ast/parameter.rs

@@ -11,10 +11,12 @@ impl Node for ast::Parameters {
             vararg,
             kwonlyargs,
             kwarg,
-            range,
+            range: _,
+            runtime_defaults,
         } = self;
-        let (posonlyargs, args, defaults) =
+        let (posonlyargs, args, mut defaults) =
             extract_positional_parameter_defaults(posonlyargs, args);
+        defaults.runtime_defaults = runtime_defaults;
         let (kwonlyargs, kw_defaults) = extract_keyword_parameter_defaults(kwonlyargs);
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeArguments::static_type().to_owned())
@@ -40,9 +42,12 @@ impl Node for ast::Parameters {
         .unwrap();
         dict.set_item("kwarg", kwarg.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("defaults", defaults.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        let _ = range;
+        let runtime_defaults = defaults.runtime_defaults.take();
+        let defaults = runtime_defaults.map_or_else(
+            || defaults.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("defaults", defaults, vm).unwrap();
         node.into()
     }
 
@@ -51,33 +56,44 @@ impl Node for ast::Parameters {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let kwonlyargs = Node::ast_from_object(
+        let posonlyargs = PositionalParameters::ast_from_field(
             vm,
             source_file,
-            get_node_field(vm, &object, "kwonlyargs", "arguments")?,
+            &object,
+            "posonlyargs",
+            "arguments",
         )?;
-        let kw_defaults = Node::ast_from_object(
+        let args =
+            PositionalParameters::ast_from_field(vm, source_file, &object, "args", "arguments")?;
+        let vararg = get_node_field_opt(vm, &object, "vararg")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?;
+        let kwonlyargs =
+            KeywordParameters::ast_from_field(vm, source_file, &object, "kwonlyargs", "arguments")?;
+        let kw_defaults = ParameterDefaults::ast_from_field(
             vm,
             source_file,
-            get_node_field(vm, &object, "kw_defaults", "arguments")?,
+            &object,
+            "kw_defaults",
+            "arguments",
+        )?;
+        let kwarg = get_node_field_opt(vm, &object, "kwarg")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?;
+        let defaults = ParameterDefaults::ast_from_field_preserve_none(
+            vm,
+            source_file,
+            &object,
+            "defaults",
+            "arguments",
         )?;
-        let kwonlyargs = merge_keyword_parameter_defaults(vm, kwonlyargs, kw_defaults)?;
 
-        let posonlyargs = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "posonlyargs", "arguments")?,
-        )?;
-        let args = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "args", "arguments")?,
-        )?;
-        let defaults = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "defaults", "arguments")?,
-        )?;
+        let ParameterDefaults {
+            runtime_defaults,
+            defaults,
+            _range: _,
+        } = defaults;
+        let kwonlyargs = merge_keyword_parameter_defaults(vm, kwonlyargs, kw_defaults)?;
         let (posonlyargs, args) =
             merge_positional_parameter_defaults(vm, posonlyargs, args, defaults)?;
 
@@ -85,14 +101,11 @@ impl Node for ast::Parameters {
             node_index: Default::default(),
             posonlyargs,
             args,
-            vararg: get_node_field_opt(vm, &object, "vararg")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
+            vararg,
             kwonlyargs,
-            kwarg: get_node_field_opt(vm, &object, "kwarg")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
+            kwarg,
             range: Default::default(),
+            runtime_defaults,
         })
     }
 
@@ -103,62 +116,68 @@ impl Node for ast::Parameters {
 
 // product
 impl Node for ast::Parameter {
-    fn ast_to_object(self, _vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
+    fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
             node_index: _,
             name,
             annotation,
             // type_comment,
             range,
+            runtime_type_comment,
+            runtime_type_comment_bytes,
         } = self;
 
         // ruff covers the ** in range but python expects it to start at the ident
         let range = TextRange::new(name.start(), range.end());
 
         let node = NodeAst
-            .into_ref_with_type(_vm, pyast::NodeArg::static_type().to_owned())
+            .into_ref_with_type(vm, pyast::NodeArg::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("arg", name.ast_to_object(_vm, source_file), _vm)
+        dict.set_item("arg", name.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item(
-            "annotation",
-            annotation.ast_to_object(_vm, source_file),
-            _vm,
+        dict.set_item("annotation", annotation.ast_to_object(vm, source_file), vm)
+            .unwrap();
+        let type_comment = super::constant::runtime_string_object(
+            vm,
+            runtime_type_comment,
+            runtime_type_comment_bytes,
         )
-        .unwrap();
-        // Ruff AST doesn't track type_comment, so always set to None
-        dict.set_item("type_comment", _vm.ctx.none(), _vm).unwrap();
-        node_add_location(&dict, range, _vm, source_file);
+        .unwrap_or_else(|| vm.ctx.none());
+        dict.set_item("type_comment", type_comment, vm).unwrap();
+        node_add_location(&dict, range, vm, source_file);
         node.into()
     }
 
     fn ast_from_object(
-        _vm: &VirtualMachine,
+        vm: &VirtualMachine,
         source_file: &SourceFile,
         _object: PyObjectRef,
     ) -> PyResult<Self> {
+        let name = get_required_identifier_field(vm, source_file, &_object, "arg", "arg")?;
+        let annotation = get_node_field_opt(vm, &_object, "annotation")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?;
+        let type_comment = get_ast_string_field_opt(vm, &_object, "type_comment")?;
+        let (runtime_type_comment, runtime_type_comment_bytes) = type_comment
+            .map_or((None, None), |type_comment| {
+                super::constant::runtime_string_from_pyobject(vm, type_comment)
+            });
+        let range = range_from_object(vm, source_file, _object, "arg")?;
         Ok(Self {
             node_index: Default::default(),
-            name: Node::ast_from_object(
-                _vm,
-                source_file,
-                get_node_field_required(_vm, &_object, "arg", "arg")?,
-            )?,
-            annotation: get_node_field_opt(_vm, &_object, "annotation")?
-                .map(|obj| Node::ast_from_object(_vm, source_file, obj))
-                .transpose()?,
-            // type_comment: get_node_field_opt(_vm, &_object, "type_comment")?
-            //     .map(|obj| Node::ast_from_object(_vm, obj))
-            //     .transpose()?,
-            range: range_from_object(_vm, source_file, _object, "arg")?,
+            name,
+            annotation,
+            range,
+            runtime_type_comment,
+            runtime_type_comment_bytes,
         })
     }
 }
 
 // product
 impl Node for ast::Keyword {
-    fn ast_to_object(self, _vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
+    fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
             node_index: _,
             arg,
@@ -166,32 +185,28 @@ impl Node for ast::Keyword {
             range: _range,
         } = self;
         let node = NodeAst
-            .into_ref_with_type(_vm, pyast::NodeKeyword::static_type().to_owned())
+            .into_ref_with_type(vm, pyast::NodeKeyword::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("arg", arg.ast_to_object(_vm, source_file), _vm)
+        dict.set_item("arg", arg.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("value", value.ast_to_object(_vm, source_file), _vm)
+        dict.set_item("value", value.ast_to_object(vm, source_file), vm)
             .unwrap();
-        node_add_location(&dict, _range, _vm, source_file);
+        node_add_location(&dict, _range, vm, source_file);
         node.into()
     }
     fn ast_from_object(
-        _vm: &VirtualMachine,
+        vm: &VirtualMachine,
         source_file: &SourceFile,
         _object: PyObjectRef,
     ) -> PyResult<Self> {
         Ok(Self {
             node_index: Default::default(),
-            arg: get_node_field_opt(_vm, &_object, "arg")?
-                .map(|obj| Node::ast_from_object(_vm, source_file, obj))
+            arg: get_node_field_opt(vm, &_object, "arg")?
+                .map(|obj| Node::ast_from_object(vm, source_file, obj))
                 .transpose()?,
-            value: Node::ast_from_object(
-                _vm,
-                source_file,
-                get_node_field_required(_vm, &_object, "value", "keyword")?,
-            )?,
-            range: range_from_object(_vm, source_file, _object, "keyword")?,
+            value: get_required_node_field(vm, source_file, &_object, "value", "keyword")?,
+            range: range_from_object(vm, source_file, _object, "keyword")?,
         })
     }
 }
@@ -201,6 +216,21 @@ struct PositionalParameters {
     pub args: Box<[ast::Parameter]>,
 }
 
+impl PositionalParameters {
+    fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        Ok(Self {
+            args: get_node_boxed_slice_field(vm, source_file, object, field, typ)?,
+            _range: TextRange::default(),
+        })
+    }
+}
+
 impl Node for PositionalParameters {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         BoxedSlice(self.args).ast_to_object(vm, source_file)
@@ -214,7 +244,7 @@ impl Node for PositionalParameters {
         let args: BoxedSlice<_> = Node::ast_from_object(vm, source_file, object)?;
         Ok(Self {
             args: args.0,
-            _range: TextRange::default(), // TODO
+            _range: TextRange::default(),
         })
     }
 }
@@ -224,6 +254,21 @@ struct KeywordParameters {
     pub keywords: Box<[ast::Parameter]>,
 }
 
+impl KeywordParameters {
+    fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        Ok(Self {
+            keywords: get_node_boxed_slice_field(vm, source_file, object, field, typ)?,
+            _range: TextRange::default(),
+        })
+    }
+}
+
 impl Node for KeywordParameters {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         BoxedSlice(self.keywords).ast_to_object(vm, source_file)
@@ -237,16 +282,55 @@ impl Node for KeywordParameters {
         let keywords: BoxedSlice<_> = Node::ast_from_object(vm, source_file, object)?;
         Ok(Self {
             keywords: keywords.0,
-            _range: TextRange::default(), // TODO
+            _range: TextRange::default(),
         })
     }
 }
 
 struct ParameterDefaults {
     pub _range: TextRange, // TODO: Use this
+    runtime_defaults: Option<Vec<Option<ast::Expr>>>,
     defaults: Box<[Option<Box<ast::Expr>>]>,
 }
 
+impl ParameterDefaults {
+    fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        Ok(Self {
+            defaults: get_node_boxed_slice_field(vm, source_file, object, field, typ)?,
+            runtime_defaults: None,
+            _range: TextRange::default(),
+        })
+    }
+
+    fn ast_from_field_preserve_none(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        let defaults: Vec<Option<Box<ast::Expr>>> =
+            get_node_list_field(vm, source_file, object, field, typ)?;
+        let runtime_defaults = defaults.iter().any(Option::is_none).then(|| {
+            defaults
+                .iter()
+                .map(|default| default.as_deref().cloned())
+                .collect()
+        });
+        Ok(Self {
+            defaults: defaults.into_boxed_slice(),
+            runtime_defaults,
+            _range: TextRange::default(),
+        })
+    }
+}
+
 impl Node for ParameterDefaults {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         BoxedSlice(self.defaults).ast_to_object(vm, source_file)
@@ -260,7 +344,8 @@ impl Node for ParameterDefaults {
         let defaults: BoxedSlice<_> = Node::ast_from_object(vm, source_file, object)?;
         Ok(Self {
             defaults: defaults.0,
-            _range: TextRange::default(), // TODO
+            runtime_defaults: None,
+            _range: TextRange::default(),
         })
     }
 }
@@ -287,6 +372,7 @@ fn extract_positional_parameter_defaults(
             .map(|item| item.range())
             .reduce(|acc, next| acc.cover(next))
             .unwrap_or_default(),
+        runtime_defaults: None,
         defaults: defaults.into_boxed_slice(),
     };
 
@@ -325,14 +411,13 @@ fn merge_positional_parameter_defaults(
     vm: &VirtualMachine,
     posonlyargs: PositionalParameters,
     args: PositionalParameters,
-    defaults: ParameterDefaults,
+    defaults: Box<[Option<Box<ast::Expr>>]>,
 ) -> PyResult<(
     Vec<ast::ParameterWithDefault>,
     Vec<ast::ParameterWithDefault>,
 )> {
     let posonlyargs = posonlyargs.args;
     let args = args.args;
-    let defaults = defaults.defaults;
 
     let mut posonlyargs: Vec<_> = <Box<[_]> as IntoIterator>::into_iter(posonlyargs)
         .map(|parameter| ast::ParameterWithDefault {
@@ -383,6 +468,7 @@ fn extract_keyword_parameter_defaults(
             .map(|item| item.range())
             .reduce(|acc, next| acc.cover(next))
             .unwrap_or_default(),
+        runtime_defaults: None,
         defaults: defaults.into_boxed_slice(),
     };
 
@@ -422,5 +508,5 @@ fn merge_keyword_parameter_defaults(
             default,
             range: Default::default(),
         })
-        .collect())
+        .collect::<Vec<_>>())
 }

crates/vm/src/stdlib/_ast/pattern.rs

@@ -10,6 +10,7 @@ impl Node for ast::MatchCase {
             guard,
             body,
             range: _,
+            runtime_body,
         } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeMatchCase::static_type().to_owned())
@@ -19,8 +20,11 @@ impl Node for ast::MatchCase {
             .unwrap();
         dict.set_item("guard", guard.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("body", body.ast_to_object(vm, source_file), vm)
-            .unwrap();
+        let body = runtime_body.map_or_else(
+            || body.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("body", body, vm).unwrap();
         node.into()
     }
 
@@ -29,22 +33,18 @@ impl Node for ast::MatchCase {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
+        let body: Vec<Option<ast::Stmt>> =
+            get_node_list_field(vm, source_file, &object, "body", "match_case")?;
+        let (runtime_body, body) = runtime_stmt_list_from_values(body);
         Ok(Self {
             node_index: Default::default(),
-            pattern: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field_required(vm, &object, "pattern", "match_case")?,
-            )?,
+            pattern: get_required_node_field(vm, source_file, &object, "pattern", "match_case")?,
             guard: get_node_field_opt(vm, &object, "guard")?
                 .map(|obj| Node::ast_from_object(vm, source_file, obj))
                 .transpose()?,
-            body: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "body", "match_case")?,
-            )?,
+            body,
             range: Default::default(),
+            runtime_body,
         })
     }
 }
@@ -68,64 +68,152 @@ impl Node for ast::Pattern {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodePatternMatchValue::static_type()) {
-            Self::MatchValue(ast::PatternMatchValue::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchSingleton::static_type()) {
-            Self::MatchSingleton(ast::PatternMatchSingleton::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchSequence::static_type()) {
-            Self::MatchSequence(ast::PatternMatchSequence::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchMapping::static_type()) {
-            Self::MatchMapping(ast::PatternMatchMapping::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchClass::static_type()) {
-            Self::MatchClass(ast::PatternMatchClass::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchStar::static_type()) {
-            Self::MatchStar(ast::PatternMatchStar::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchAs::static_type()) {
-            Self::MatchAs(ast::PatternMatchAs::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodePatternMatchOr::static_type()) {
-            Self::MatchOr(ast::PatternMatchOr::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
+        if vm.is_none(&object) {
+            return Err(vm.new_type_error(format!(
+                "expected some sort of pattern, but got {}",
+                object.repr(vm)?
+            )));
+        }
+        enum PatternKind {
+            Value,
+            Singleton,
+            Sequence,
+            Mapping,
+            Class,
+            Star,
+            As,
+            Or,
+        }
+        let kind = if is_node_instance(vm, &object, pyast::NodePatternMatchValue::static_type())? {
+            PatternKind::Value
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchSingleton::static_type())? {
+            PatternKind::Singleton
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchSequence::static_type())? {
+            PatternKind::Sequence
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchMapping::static_type())? {
+            PatternKind::Mapping
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchClass::static_type())? {
+            PatternKind::Class
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchStar::static_type())? {
+            PatternKind::Star
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchAs::static_type())? {
+            PatternKind::As
+        } else if is_node_instance(vm, &object, pyast::NodePatternMatchOr::static_type())? {
+            PatternKind::Or
         } else {
             return Err(vm.new_type_error(format!(
                 "expected some sort of pattern, but got {}",
                 object.repr(vm)?
             )));
+        };
+        let range = pattern_range_from_object(vm, source_file, object.clone())?;
+        Ok(match kind {
+            PatternKind::Value => Self::MatchValue(pattern_match_value_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            PatternKind::Singleton => Self::MatchSingleton(
+                pattern_match_singleton_from_object_with_range(vm, source_file, object, range)?,
+            ),
+            PatternKind::Sequence => Self::MatchSequence(
+                pattern_match_sequence_from_object_with_range(vm, source_file, object, range)?,
+            ),
+            PatternKind::Mapping => Self::MatchMapping(
+                pattern_match_mapping_from_object_with_range(vm, source_file, object, range)?,
+            ),
+            PatternKind::Class => Self::MatchClass(pattern_match_class_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            PatternKind::Star => Self::MatchStar(pattern_match_star_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            PatternKind::As => Self::MatchAs(pattern_match_as_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            PatternKind::Or => Self::MatchOr(pattern_match_or_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
         })
     }
 }
+
+fn null_pattern_placeholder(range: TextRange) -> ast::Pattern {
+    ast::Pattern::MatchAs(ast::PatternMatchAs {
+        node_index: Default::default(),
+        range,
+        pattern: None,
+        name: None,
+    })
+}
+
+fn lower_nullable_patterns(values: &[Option<ast::Pattern>], range: TextRange) -> Vec<ast::Pattern> {
+    values
+        .iter()
+        .cloned()
+        .map(|value| value.unwrap_or_else(|| null_pattern_placeholder(range)))
+        .collect()
+}
+
+fn null_expr_placeholder(range: TextRange) -> ast::Expr {
+    ast::Expr::NoneLiteral(ast::ExprNoneLiteral {
+        node_index: Default::default(),
+        range,
+    })
+}
+
+fn lower_nullable_exprs(values: &[Option<ast::Expr>], range: TextRange) -> Vec<ast::Expr> {
+    values
+        .iter()
+        .cloned()
+        .map(|value| value.unwrap_or_else(|| null_expr_placeholder(range)))
+        .collect()
+}
+
+type RuntimePatternList = Option<Vec<Option<ast::Pattern>>>;
+type PatternListField = (RuntimePatternList, Vec<ast::Pattern>);
+
+fn pattern_list_from_field(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: &PyObject,
+    field: &'static str,
+    typ: &str,
+    range: TextRange,
+) -> PyResult<PatternListField> {
+    let values: Vec<Option<ast::Pattern>> =
+        get_node_list_field(vm, source_file, object, field, typ)?;
+    let runtime_patterns = values.iter().any(Option::is_none).then(|| values.clone());
+    Ok((runtime_patterns, lower_nullable_patterns(&values, range)))
+}
+
 // constructor
+fn pattern_match_value_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchValue> {
+    Ok(ast::PatternMatchValue {
+        node_index: Default::default(),
+        value: get_required_node_field(vm, source_file, &object, "value", "MatchValue")?,
+        range,
+    })
+}
+
 impl Node for ast::PatternMatchValue {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -148,19 +236,29 @@ impl Node for ast::PatternMatchValue {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            value: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "value", "MatchValue")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "MatchValue")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchValue")?;
+        pattern_match_value_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn pattern_match_singleton_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchSingleton> {
+    Ok(ast::PatternMatchSingleton {
+        node_index: Default::default(),
+        value: Node::ast_from_object(
+            vm,
+            source_file,
+            get_node_field(vm, &object, "value", "MatchSingleton")?,
+        )?,
+        range,
+    })
+}
+
 impl Node for ast::PatternMatchSingleton {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -186,15 +284,8 @@ impl Node for ast::PatternMatchSingleton {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            value: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "value", "MatchSingleton")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "MatchSingleton")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchSingleton")?;
+        pattern_match_singleton_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -219,21 +310,35 @@ impl Node for ast::Singleton {
         } else if object.is(&vm.ctx.false_value) {
             Ok(Self::False)
         } else {
-            Err(vm.new_value_error(format!(
-                "Expected None, True, or False, got {:?}",
-                object.class().name()
-            )))
+            Err(vm.new_value_error("MatchSingleton can only contain True, False and None"))
         }
     }
 }
 
 // constructor
+fn pattern_match_sequence_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchSequence> {
+    let (runtime_patterns, patterns) =
+        pattern_list_from_field(vm, source_file, &object, "patterns", "MatchSequence", range)?;
+    Ok(ast::PatternMatchSequence {
+        node_index: Default::default(),
+        patterns: patterns.to_vec(),
+        range,
+        runtime_patterns,
+    })
+}
+
 impl Node for ast::PatternMatchSequence {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
             node_index: _,
             patterns,
             range,
+            runtime_patterns,
         } = self;
         let node = NodeAst
             .into_ref_with_type(
@@ -242,8 +347,11 @@ impl Node for ast::PatternMatchSequence {
             )
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("patterns", patterns.ast_to_object(vm, source_file), vm)
-            .unwrap();
+        let patterns = runtime_patterns.map_or_else(
+            || patterns.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("patterns", patterns, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -253,19 +361,40 @@ impl Node for ast::PatternMatchSequence {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            patterns: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "patterns", "MatchSequence")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "MatchSequence")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchSequence")?;
+        pattern_match_sequence_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn pattern_match_mapping_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchMapping> {
+    let keys: Vec<Option<ast::Expr>> =
+        get_node_list_field(vm, source_file, &object, "keys", "MatchMapping")?;
+    let patterns: Vec<Option<ast::Pattern>> =
+        get_node_list_field(vm, source_file, &object, "patterns", "MatchMapping")?;
+    let runtime_keys = keys.iter().any(Option::is_none).then(|| keys.clone());
+    let runtime_patterns = patterns
+        .iter()
+        .any(Option::is_none)
+        .then(|| patterns.clone());
+    Ok(ast::PatternMatchMapping {
+        node_index: Default::default(),
+        keys: lower_nullable_exprs(&keys, range),
+        patterns: lower_nullable_patterns(&patterns, range),
+        rest: get_node_field_opt(vm, &object, "rest")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+        runtime_keys,
+        runtime_patterns,
+    })
+}
+
 impl Node for ast::PatternMatchMapping {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -274,15 +403,23 @@ impl Node for ast::PatternMatchMapping {
             patterns,
             rest,
             range,
+            runtime_keys,
+            runtime_patterns,
         } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodePatternMatchMapping::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("keys", keys.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        dict.set_item("patterns", patterns.ast_to_object(vm, source_file), vm)
-            .unwrap();
+        let keys = runtime_keys.map_or_else(
+            || keys.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("keys", keys, vm).unwrap();
+        let patterns = runtime_patterns.map_or_else(
+            || patterns.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("patterns", patterns, vm).unwrap();
         dict.set_item("rest", rest.ast_to_object(vm, source_file), vm)
             .unwrap();
         node_add_location(&dict, range, vm, source_file);
@@ -294,27 +431,57 @@ impl Node for ast::PatternMatchMapping {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            keys: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "keys", "MatchMapping")?,
-            )?,
-            patterns: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "patterns", "MatchMapping")?,
-            )?,
-            rest: get_node_field_opt(vm, &object, "rest")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            range: range_from_object(vm, source_file, object, "MatchMapping")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchMapping")?;
+        pattern_match_mapping_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn pattern_match_class_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchClass> {
+    let cls = get_required_node_field(vm, source_file, &object, "cls", "MatchClass")?;
+    let patterns: Vec<Option<ast::Pattern>> =
+        get_node_list_field(vm, source_file, &object, "patterns", "MatchClass")?;
+    let kwd_attrs = PatternMatchClassKeywordAttributes::ast_from_field(
+        vm,
+        source_file,
+        &object,
+        "kwd_attrs",
+        "MatchClass",
+    )?;
+    let kwd_patterns: Vec<Option<ast::Pattern>> =
+        get_node_list_field(vm, source_file, &object, "kwd_patterns", "MatchClass")?;
+    let has_runtime_shape = kwd_attrs.0.len() != kwd_patterns.len()
+        || patterns.iter().any(Option::is_none)
+        || kwd_patterns.iter().any(Option::is_none);
+    let runtime_patterns = has_runtime_shape.then(|| patterns.clone());
+    let runtime_kwd_attrs = has_runtime_shape.then(|| kwd_attrs.0.clone());
+    let runtime_kwd_patterns = has_runtime_shape.then(|| kwd_patterns.clone());
+    let patterns = PatternMatchClassPatterns(lower_nullable_patterns(&patterns, range));
+    let kwd_patterns =
+        PatternMatchClassKeywordPatterns(lower_nullable_patterns(&kwd_patterns, range));
+    let (patterns, keywords) = merge_pattern_match_class(patterns, kwd_attrs, kwd_patterns);
+
+    Ok(ast::PatternMatchClass {
+        node_index: Default::default(),
+        cls,
+        range,
+        arguments: ast::PatternArguments {
+            node_index: Default::default(),
+            range: Default::default(),
+            patterns,
+            keywords,
+        },
+        runtime_patterns,
+        runtime_kwd_attrs,
+        runtime_kwd_patterns,
+    })
+}
+
 impl Node for ast::PatternMatchClass {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -322,24 +489,36 @@ impl Node for ast::PatternMatchClass {
             cls,
             arguments,
             range,
+            runtime_patterns,
+            runtime_kwd_attrs,
+            runtime_kwd_patterns,
         } = self;
-        let (patterns, kwd_attrs, kwd_patterns) = split_pattern_match_class(arguments);
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodePatternMatchClass::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
         dict.set_item("cls", cls.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("patterns", patterns.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        dict.set_item("kwd_attrs", kwd_attrs.ast_to_object(vm, source_file), vm)
-            .unwrap();
-        dict.set_item(
-            "kwd_patterns",
-            kwd_patterns.ast_to_object(vm, source_file),
-            vm,
-        )
-        .unwrap();
+        let (patterns, kwd_attrs, kwd_patterns) =
+            if let (Some(patterns), Some(kwd_attrs), Some(kwd_patterns)) =
+                (runtime_patterns, runtime_kwd_attrs, runtime_kwd_patterns)
+            {
+                (
+                    patterns.ast_to_object(vm, source_file),
+                    kwd_attrs.ast_to_object(vm, source_file),
+                    kwd_patterns.ast_to_object(vm, source_file),
+                )
+            } else {
+                let (patterns, kwd_attrs, kwd_patterns) = split_pattern_match_class(arguments);
+                (
+                    patterns.ast_to_object(vm, source_file),
+                    kwd_attrs.ast_to_object(vm, source_file),
+                    kwd_patterns.ast_to_object(vm, source_file),
+                )
+            };
+        dict.set_item("patterns", patterns, vm).unwrap();
+        dict.set_item("kwd_attrs", kwd_attrs, vm).unwrap();
+        dict.set_item("kwd_patterns", kwd_patterns, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -349,41 +528,8 @@ impl Node for ast::PatternMatchClass {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let patterns = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "patterns", "MatchClass")?,
-        )?;
-        let kwd_attrs: PatternMatchClassKeywordAttributes = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "kwd_attrs", "MatchClass")?,
-        )?;
-        let kwd_patterns: PatternMatchClassKeywordPatterns = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "kwd_patterns", "MatchClass")?,
-        )?;
-        if kwd_attrs.0.len() != kwd_patterns.0.len() {
-            return Err(vm.new_value_error("MatchClass has mismatched kwd_attrs and kwd_patterns"));
-        }
-        let (patterns, keywords) = merge_pattern_match_class(patterns, kwd_attrs, kwd_patterns);
-
-        Ok(Self {
-            node_index: Default::default(),
-            cls: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "cls", "MatchClass")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "MatchClass")?,
-            arguments: ast::PatternArguments {
-                node_index: Default::default(),
-                range: Default::default(),
-                patterns,
-                keywords,
-            },
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchClass")?;
+        pattern_match_class_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -405,6 +551,24 @@ impl Node for PatternMatchClassPatterns {
 
 struct PatternMatchClassKeywordAttributes(Vec<ast::Identifier>);
 
+impl PatternMatchClassKeywordAttributes {
+    fn ast_from_field(
+        vm: &VirtualMachine,
+        source_file: &SourceFile,
+        object: &PyObject,
+        field: &'static str,
+        typ: &str,
+    ) -> PyResult<Self> {
+        Ok(Self(get_node_list_field(
+            vm,
+            source_file,
+            object,
+            field,
+            typ,
+        )?))
+    }
+}
+
 impl Node for PatternMatchClassKeywordAttributes {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         self.0.ast_to_object(vm, source_file)
@@ -435,6 +599,21 @@ impl Node for PatternMatchClassKeywordPatterns {
     }
 }
 // constructor
+fn pattern_match_star_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchStar> {
+    Ok(ast::PatternMatchStar {
+        node_index: Default::default(),
+        name: get_node_field_opt(vm, &object, "name")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for ast::PatternMatchStar {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -457,17 +636,30 @@ impl Node for ast::PatternMatchStar {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            name: get_node_field_opt(vm, &object, "name")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            range: range_from_object(vm, source_file, object, "MatchStar")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchStar")?;
+        pattern_match_star_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn pattern_match_as_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchAs> {
+    Ok(ast::PatternMatchAs {
+        node_index: Default::default(),
+        pattern: get_node_field_opt(vm, &object, "pattern")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        name: get_node_field_opt(vm, &object, "name")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for ast::PatternMatchAs {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -493,33 +685,45 @@ impl Node for ast::PatternMatchAs {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            pattern: get_node_field_opt(vm, &object, "pattern")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            name: get_node_field_opt(vm, &object, "name")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            range: range_from_object(vm, source_file, object, "MatchAs")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchAs")?;
+        pattern_match_as_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn pattern_match_or_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::PatternMatchOr> {
+    let (runtime_patterns, patterns) =
+        pattern_list_from_field(vm, source_file, &object, "patterns", "MatchOr", range)?;
+    Ok(ast::PatternMatchOr {
+        node_index: Default::default(),
+        patterns: patterns.to_vec(),
+        range,
+        runtime_patterns,
+    })
+}
+
 impl Node for ast::PatternMatchOr {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
             node_index: _,
             patterns,
             range,
+            runtime_patterns,
         } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodePatternMatchOr::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("patterns", patterns.ast_to_object(vm, source_file), vm)
-            .unwrap();
+        let patterns = runtime_patterns.map_or_else(
+            || patterns.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        );
+        dict.set_item("patterns", patterns, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -528,15 +732,8 @@ impl Node for ast::PatternMatchOr {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            patterns: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "patterns", "MatchOr")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "MatchOr")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "MatchOr")?;
+        pattern_match_or_from_object_with_range(vm, source_file, object, range)
     }
 }
 

crates/vm/src/stdlib/_ast/pyast.rs

@@ -3,7 +3,6 @@ use crate::builtins::{PyGenericAlias, PyTuple, PyTupleRef, PyTypeRef, make_union
 use crate::common::ascii;
 use crate::convert::ToPyObject;
 use crate::function::FuncArgs;
-use crate::types::Initializer;
 
 macro_rules! impl_node {
     (
@@ -61,6 +60,12 @@ macro_rules! impl_node {
 macro_rules! impl_base_node {
     // Base node without fields/attributes (e.g. NodeMod, NodeExpr)
     ($name:ident) => {
+        impl_base_node!($name, attributes: []);
+    };
+    ($name:ident, attributes: [$($attr:expr),* $(,)?]) => {
+        impl_base_node!($name, attributes: [$($attr),*], optional_end_location: false);
+    };
+    ($name:ident, attributes: [$($attr:expr),* $(,)?], optional_end_location: $optional_end_location:expr) => {
         #[pyclass(flags(HAS_DICT, BASETYPE))]
         impl $name {
             #[pymethod]
@@ -83,9 +88,24 @@ macro_rules! impl_base_node {
                     (*flags).remove(crate::types::PyTypeFlags::IMMUTABLETYPE);
                 }
                 class.set_attr(
-                    identifier!(ctx, _attributes),
+                    identifier!(ctx, _fields),
                     ctx.empty_tuple.clone().into(),
                 );
+                class.set_str_attr("__match_args__", ctx.empty_tuple.clone(), ctx);
+                class.set_attr(
+                    identifier!(ctx, _attributes),
+                    ctx.new_tuple(vec![
+                        $(
+                            ctx.new_str(ascii!($attr)).into()
+                        ),*
+                    ])
+                    .into(),
+                );
+                if $optional_end_location {
+                    let none = ctx.none();
+                    class.set_str_attr("end_lineno", none.clone(), ctx);
+                    class.set_str_attr("end_col_offset", none, ctx);
+                }
             }
         }
     };
@@ -179,7 +199,11 @@ impl_node!(
 #[repr(transparent)]
 pub(crate) struct NodeStmt(NodeAst);
 
-impl_base_node!(NodeStmt);
+impl_base_node!(
+    NodeStmt,
+    attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"],
+    optional_end_location: true
+);
 
 impl_node!(
     #[pyclass(module = "_ast", name = "FunctionType", base = NodeMod)]
@@ -378,7 +402,11 @@ impl_node!(
 #[repr(transparent)]
 pub(crate) struct NodeExpr(NodeAst);
 
-impl_base_node!(NodeExpr);
+impl_base_node!(
+    NodeExpr,
+    attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"],
+    optional_end_location: true
+);
 
 impl_node!(
     #[pyclass(module = "_ast", name = "Continue", base = NodeStmt)]
@@ -533,12 +561,11 @@ impl_node!(
     attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"],
 );
 
-// NodeExprConstant needs custom Initializer to default kind to None
 #[pyclass(module = "_ast", name = "Constant", base = NodeExpr)]
 #[repr(transparent)]
 pub(crate) struct NodeExprConstant(NodeExpr);
 
-#[pyclass(flags(HAS_DICT, BASETYPE), with(Initializer))]
+#[pyclass(flags(HAS_DICT, BASETYPE))]
 impl NodeExprConstant {
     #[extend_class]
     fn extend_class_with_fields(ctx: &Context, class: &'static Py<PyType>) {
@@ -580,24 +607,6 @@ impl NodeExprConstant {
     }
 }
 
-impl Initializer for NodeExprConstant {
-    type Args = FuncArgs;
-
-    fn slot_init(zelf: PyObjectRef, args: FuncArgs, vm: &VirtualMachine) -> PyResult<()> {
-        <NodeAst as Initializer>::slot_init(zelf.clone(), args, vm)?;
-        // kind defaults to None if not provided
-        let dict = zelf.as_object().dict().unwrap();
-        if !dict.contains_key("kind", vm) {
-            dict.set_item("kind", vm.ctx.none(), vm)?;
-        }
-        Ok(())
-    }
-
-    fn init(_zelf: PyRef<Self>, _args: Self::Args, _vm: &VirtualMachine) -> PyResult<()> {
-        unreachable!("slot_init is defined")
-    }
-}
-
 impl_node!(
     #[pyclass(module = "_ast", name = "Attribute", base = NodeExpr)]
     pub(crate) struct NodeExprAttribute,
@@ -841,7 +850,11 @@ impl_node!(
 #[repr(transparent)]
 pub(crate) struct NodeExceptHandler(NodeAst);
 
-impl_base_node!(NodeExceptHandler);
+impl_base_node!(
+    NodeExceptHandler,
+    attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"],
+    optional_end_location: true
+);
 
 impl_node!(
     #[pyclass(module = "_ast", name = "comprehension", base = NodeAst)]
@@ -893,7 +906,10 @@ impl_node!(
 #[repr(transparent)]
 pub(crate) struct NodePattern(NodeAst);
 
-impl_base_node!(NodePattern);
+impl_base_node!(
+    NodePattern,
+    attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"]
+);
 
 impl_node!(
     #[pyclass(module = "_ast", name = "match_case", base = NodeAst)]
@@ -967,7 +983,10 @@ impl_node!(
 #[repr(transparent)]
 pub(crate) struct NodeTypeParam(NodeAst);
 
-impl_base_node!(NodeTypeParam);
+impl_base_node!(
+    NodeTypeParam,
+    attributes: ["lineno", "col_offset", "end_lineno", "end_col_offset"]
+);
 
 impl_node!(
     #[pyclass(module = "_ast", name = "TypeIgnore", base = NodeTypeIgnore)]
@@ -1681,7 +1700,6 @@ fn populate_field_types(vm: &VirtualMachine, module: &Py<PyModule>) {
 
     let field_types_attr = vm.ctx.intern_str("_field_types");
     let annotations_attr = vm.ctx.intern_str("__annotations__");
-    let empty_dict: PyObjectRef = vm.ctx.new_dict().into();
 
     for &(class_name, fields) in FIELD_TYPES {
         if fields.is_empty() {
@@ -1752,36 +1770,6 @@ fn populate_field_types(vm: &VirtualMachine, module: &Py<PyModule>) {
             type_obj.set_attr(annotations_attr, field_types);
         }
     }
-
-    // Base AST classes (e.g., expr, stmt) should still expose __annotations__.
-    const BASE_AST_TYPES: &[&str] = &[
-        "mod",
-        "stmt",
-        "expr",
-        "expr_context",
-        "boolop",
-        "operator",
-        "unaryop",
-        "cmpop",
-        "excepthandler",
-        "pattern",
-        "type_ignore",
-        "type_param",
-    ];
-    for &class_name in BASE_AST_TYPES {
-        let class = module
-            .get_attr(class_name, vm)
-            .unwrap_or_else(|_| panic!("AST class '{class_name}' not found in module"));
-        let Some(type_obj) = class.downcast_ref::<PyType>() else {
-            continue;
-        };
-        if type_obj.get_attr(field_types_attr).is_none() {
-            type_obj.set_attr(field_types_attr, empty_dict.clone());
-        }
-        if type_obj.get_attr(annotations_attr).is_none() {
-            type_obj.set_attr(annotations_attr, empty_dict.clone());
-        }
-    }
 }
 
 fn populate_singletons(vm: &VirtualMachine, module: &Py<PyModule>) {

crates/vm/src/stdlib/_ast/python.rs

@@ -7,13 +7,10 @@ use super::{
 #[pymodule]
 pub(crate) mod _ast {
     use crate::{
-        AsObject, Context, Py, PyObjectRef, PyPayload, PyRef, PyResult, VirtualMachine,
-        builtins::{
-            PyDictRef, PyStr, PyStrRef, PyTupleRef, PyType, PyTypeRef, PyUtf8Str, PyUtf8StrRef,
-        },
+        AsObject, Context, Py, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, VirtualMachine,
+        builtins::{PyDictRef, PySet, PyStr, PyTupleRef, PyType, PyTypeRef, PyUtf8Str},
         class::{PyClassImpl, StaticType},
-        common::wtf8::Wtf8,
-        function::{FuncArgs, KwArgs, PyMethodDef, PyMethodFlags},
+        function::{ArgIterable, FuncArgs, KwArgs, PyMethodDef, PyMethodFlags},
         stdlib::_ast::repr,
         types::{Constructor, Initializer},
         warn,
@@ -117,10 +114,12 @@ pub(crate) mod _ast {
 
         let fields = cls.get_attr(vm.ctx.intern_str("_fields"));
         if let Some(fields) = fields {
-            let fields: Vec<PyStrRef> = fields.try_to_value(vm)?;
+            let fields = fields.sequence_unchecked();
+            let numfields = fields.length(vm)?;
             let mut positional: Vec<PyObjectRef> = Vec::new();
-            for field in fields {
-                if dict.get_item_opt::<Wtf8>(field.as_wtf8(), vm)?.is_some() {
+            for i in 0..numfields {
+                let field = fields.get_item(i as isize, vm)?;
+                if dict.get_item_opt(&*field, vm)?.is_some() {
                     positional.push(vm.ctx.none());
                 } else {
                     break;
@@ -136,6 +135,85 @@ pub(crate) mod _ast {
             .new_tuple(vec![type_obj, vm.ctx.new_tuple(vec![]).into(), dict.into()]))
     }
 
+    fn ast_replace_update_payload(
+        payload: &PyDictRef,
+        keys: Option<&PyObjectRef>,
+        dict: &PyDictRef,
+        vm: &VirtualMachine,
+    ) -> PyResult<()> {
+        let Some(keys) = keys else {
+            return Ok(());
+        };
+        let keys = keys.sequence_unchecked();
+        let num_keys = keys.length(vm)?;
+        for i in 0..num_keys {
+            let key = keys.get_item(i as isize, vm)?;
+            if let Some(value) = dict.get_item_opt(&*key, vm)? {
+                payload.set_item(&*key, value, vm)?;
+            }
+        }
+        Ok(())
+    }
+
+    fn ast_replace_set_update(
+        expecting: &PyRef<PySet>,
+        iterable: Option<&PyObjectRef>,
+        vm: &VirtualMachine,
+    ) -> PyResult<()> {
+        let Some(iterable) = iterable else {
+            return Ok(());
+        };
+        let iterable = iterable.clone().try_into_value::<ArgIterable>(vm)?;
+        for item in iterable.iter(vm)? {
+            expecting.add(item?, vm)?;
+        }
+        Ok(())
+    }
+
+    fn ast_replace_set_discard(
+        expecting: &PyRef<PySet>,
+        key: &PyObject,
+        vm: &VirtualMachine,
+    ) -> PyResult<bool> {
+        let contained = expecting
+            .as_object()
+            .sequence_unchecked()
+            .contains(key, vm)?;
+        if contained {
+            vm.call_method(expecting.as_object(), "discard", (key.to_owned(),))?;
+        }
+        Ok(contained)
+    }
+
+    fn ast_replace_set_difference_update(
+        expecting: &PyRef<PySet>,
+        iterable: Option<&PyObjectRef>,
+        vm: &VirtualMachine,
+    ) -> PyResult<()> {
+        let Some(iterable) = iterable else {
+            return Ok(());
+        };
+        let iterable = iterable.clone().try_into_value::<ArgIterable>(vm)?;
+        for item in iterable.iter(vm)? {
+            let item = item?;
+            ast_replace_set_discard(expecting, &item, vm)?;
+        }
+        Ok(())
+    }
+
+    fn ast_set_attr(
+        obj: &PyObject,
+        name: &PyObject,
+        value: impl Into<PyObjectRef>,
+        vm: &VirtualMachine,
+    ) -> PyResult<()> {
+        let name = name
+            .to_owned()
+            .downcast::<PyStr>()
+            .map_err(|_| vm.new_type_error("attribute name must be string"))?;
+        obj.set_attr(&name, value, vm)
+    }
+
     pub(crate) fn ast_replace(zelf: PyObjectRef, args: FuncArgs, vm: &VirtualMachine) -> PyResult {
         if !args.args.is_empty() {
             return Err(vm.new_type_error("__replace__() takes no positional arguments"));
@@ -146,22 +224,13 @@ pub(crate) mod _ast {
         let attributes = cls.get_attr(vm.ctx.intern_str("_attributes"));
         let dict = zelf.as_object().dict();
 
-        let mut expecting: std::collections::HashSet<String> = std::collections::HashSet::new();
-        if let Some(fields) = fields.clone() {
-            let fields: Vec<PyUtf8StrRef> = fields.try_to_value(vm)?;
-            for field in fields {
-                expecting.insert(field.as_str().to_owned());
-            }
-        }
-        if let Some(attributes) = attributes.clone() {
-            let attributes: Vec<PyUtf8StrRef> = attributes.try_to_value(vm)?;
-            for attr in attributes {
-                expecting.insert(attr.as_str().to_owned());
-            }
-        }
+        let expecting = PySet::default().into_ref(&vm.ctx);
+        ast_replace_set_update(&expecting, fields.as_ref(), vm)?;
+        ast_replace_set_update(&expecting, attributes.as_ref(), vm)?;
 
         for (key, _value) in &args.kwargs {
-            if !expecting.remove(key) {
+            let key_obj: PyObjectRef = vm.ctx.new_str(key.as_str()).into();
+            if !ast_replace_set_discard(&expecting, &key_obj, vm)? {
                 return Err(vm.new_type_error(format!(
                     "{}.__replace__ got an unexpected keyword argument '{}'.",
                     cls.name(),
@@ -172,16 +241,9 @@ pub(crate) mod _ast {
 
         if let Some(dict) = dict.as_ref() {
             for (key, _value) in dict.items_vec() {
-                if let Ok(key) = key.downcast::<PyUtf8Str>() {
-                    expecting.remove(key.as_str());
-                }
-            }
-            if let Some(attributes) = attributes.clone() {
-                let attributes: Vec<PyUtf8StrRef> = attributes.try_to_value(vm)?;
-                for attr in attributes {
-                    expecting.remove(attr.as_str());
-                }
+                ast_replace_set_discard(&expecting, &key, vm)?;
             }
+            ast_replace_set_difference_update(&expecting, attributes.as_ref(), vm)?;
         }
 
         // Discard optional fields (T | None).
@@ -189,20 +251,18 @@ pub(crate) mod _ast {
             && let Ok(field_types) = field_types.downcast::<crate::builtins::PyDict>()
         {
             for (key, value) in field_types.items_vec() {
-                let Ok(key) = key.downcast::<PyUtf8Str>() else {
-                    continue;
-                };
                 if value.fast_isinstance(vm.ctx.types.union_type) {
-                    expecting.remove(key.as_str());
+                    ast_replace_set_discard(&expecting, &key, vm)?;
                 }
             }
         }
 
-        if !expecting.is_empty() {
-            let mut names: Vec<String> = expecting
-                .into_iter()
-                .map(|name| format!("{name:?}"))
-                .collect();
+        let remaining = expecting.elements();
+        if !remaining.is_empty() {
+            let mut names = Vec::with_capacity(remaining.len());
+            for name in &remaining {
+                names.push(name.repr(vm)?.to_string());
+            }
             names.sort();
             let missing = names.join(", ");
             let count = names.len();
@@ -217,22 +277,8 @@ pub(crate) mod _ast {
 
         let payload = vm.ctx.new_dict();
         if let Some(dict) = dict {
-            if let Some(fields) = fields {
-                let fields: Vec<PyStrRef> = fields.try_to_value(vm)?;
-                for field in fields {
-                    if let Some(value) = dict.get_item_opt::<Wtf8>(field.as_wtf8(), vm)? {
-                        payload.set_item(field.as_object(), value, vm)?;
-                    }
-                }
-            }
-            if let Some(attributes) = attributes {
-                let attributes: Vec<PyStrRef> = attributes.try_to_value(vm)?;
-                for attr in attributes {
-                    if let Some(value) = dict.get_item_opt::<Wtf8>(attr.as_wtf8(), vm)? {
-                        payload.set_item(attr.as_object(), value, vm)?;
-                    }
-                }
-            }
+            ast_replace_update_payload(&payload, fields.as_ref(), &dict, vm)?;
+            ast_replace_update_payload(&payload, attributes.as_ref(), &dict, vm)?;
         }
         for (key, value) in args.kwargs {
             payload.set_item(vm.ctx.intern_str(key), value, vm)?;
@@ -327,7 +373,7 @@ pub(crate) mod _ast {
         }
 
         fn py_new(_cls: &Py<PyType>, _args: Self::Args, _vm: &VirtualMachine) -> PyResult<Self> {
-            unimplemented!("use slot_new")
+            unreachable!("NodeAst construction is handled by slot_new")
         }
     }
 
@@ -350,52 +396,55 @@ pub(crate) mod _ast {
                         zelf.class().name()
                     ))
                 })?;
-            let fields: Vec<PyUtf8StrRef> = fields.try_to_value(vm)?;
+            let fields_seq = fields.sequence_unchecked();
+            let numfields = fields_seq.length(vm)?;
+            let remaining_fields = PySet::default().into_ref(&vm.ctx);
+            ast_replace_set_update(&remaining_fields, Some(&fields), vm)?;
             let n_args = args.args.len();
-            if n_args > fields.len() {
+            if n_args > numfields {
                 return Err(vm.new_type_error(format!(
                     "{} constructor takes at most {} positional argument{}",
                     zelf.class().name(),
-                    fields.len(),
-                    if fields.len() == 1 { "" } else { "s" },
+                    numfields,
+                    if numfields == 1 { "" } else { "s" },
                 )));
             }
 
-            // Track which fields were set
-            let mut set_fields = std::collections::HashSet::new();
-            let mut attributes: Option<Vec<PyStrRef>> = None;
+            let mut attributes: Option<PyObjectRef> = None;
 
-            for (name, arg) in fields.iter().zip(args.args) {
-                zelf.set_attr(name, arg, vm)?;
-                set_fields.insert(name.as_str().to_owned());
+            for (i, arg) in args.args.into_iter().enumerate() {
+                let name = fields_seq.get_item(i as isize, vm)?;
+                ast_set_attr(&zelf, &name, arg, vm)?;
+                ast_replace_set_discard(&remaining_fields, &name, vm)?;
             }
             for (key, value) in args.kwargs {
-                if let Some(pos) = fields.iter().position(|f| f.as_bytes() == key.as_bytes())
-                    && pos < n_args
-                {
-                    return Err(vm.new_type_error(format!(
-                        "{} got multiple values for argument '{}'",
-                        zelf.class().name(),
-                        key
-                    )));
-                }
-
-                if fields
-                    .iter()
-                    .all(|field| field.as_bytes() != key.as_bytes())
-                {
-                    let attrs = if let Some(attrs) = &attributes {
-                        attrs
+                let key_obj: PyObjectRef = vm.ctx.new_str(key.as_str()).into();
+                let contains = fields_seq.contains(&key_obj, vm)?;
+                if contains {
+                    if !ast_replace_set_discard(&remaining_fields, &key_obj, vm)? {
+                        return Err(vm.new_type_error(format!(
+                            "{} got multiple values for argument '{}'",
+                            zelf.class().name(),
+                            key
+                        )));
+                    }
+                } else {
+                    let attrs = if let Some(attributes) = &attributes {
+                        attributes
                     } else {
                         let attrs = zelf
                             .class()
                             .get_attr(vm.ctx.intern_str("_attributes"))
-                            .and_then(|attr| attr.try_to_value::<Vec<PyStrRef>>(vm).ok())
-                            .unwrap_or_default();
+                            .ok_or_else(|| {
+                                vm.new_attribute_error(format!(
+                                    "type object '{}' has no attribute '_attributes'",
+                                    zelf.class().name()
+                                ))
+                            })?;
                         attributes = Some(attrs);
                         attributes.as_ref().unwrap()
                     };
-                    if attrs.iter().all(|attr| attr.as_bytes() != key.as_bytes()) {
+                    if !attrs.sequence_unchecked().contains(&key_obj, vm)? {
                         let message = vm.ctx.new_str(format!(
                             "{}.__init__ got an unexpected keyword argument '{}'. \
 Support for arbitrary keyword arguments is deprecated and will be removed in Python 3.15.",
@@ -412,7 +461,6 @@ Support for arbitrary keyword arguments is deprecated and will be removed in Pyt
                     }
                 }
 
-                set_fields.insert(key.clone());
                 zelf.set_attr(vm.ctx.intern_str(key), value, vm)?;
             }
 
@@ -425,17 +473,14 @@ Support for arbitrary keyword arguments is deprecated and will be removed in Pyt
                 let expr_ctx_type: PyObjectRef =
                     super::super::pyast::NodeExprContext::make_static_type().into();
 
-                for field in &fields {
-                    if set_fields.contains(field.as_str()) {
-                        continue;
-                    }
-                    if let Some(ftype) = ft_dict.get_item_opt::<Wtf8>(field.as_wtf8(), vm)? {
+                for field in remaining_fields.elements() {
+                    if let Some(ftype) = ft_dict.get_item_opt(&*field, vm)? {
                         if ftype.fast_isinstance(vm.ctx.types.union_type) {
                             // Optional field (T | None) — no default
                         } else if ftype.fast_isinstance(vm.ctx.types.generic_alias_type) {
                             // List field (list[T]) — default to []
                             let empty_list: PyObjectRef = vm.ctx.new_list(vec![]).into();
-                            zelf.set_attr(vm.ctx.intern_str(field.as_wtf8()), empty_list, vm)?;
+                            ast_set_attr(&zelf, &field, empty_list, vm)?;
                         } else if ftype.is(&expr_ctx_type) {
                             // expr_context — default to Load()
                             let load_type =
@@ -445,13 +490,15 @@ Support for arbitrary keyword arguments is deprecated and will be removed in Pyt
                                 .unwrap_or_else(|| {
                                     vm.ctx.new_base_object(load_type, Some(vm.ctx.new_dict()))
                                 });
-                            zelf.set_attr(vm.ctx.intern_str(field.as_wtf8()), load_instance, vm)?;
+                            ast_set_attr(&zelf, &field, load_instance, vm)?;
                         } else {
                             // Required field missing: emit DeprecationWarning.
+                            let field_repr = field.repr(vm)?;
                             let message = vm.ctx.new_str(format!(
-                                "{}.__init__ missing 1 required positional argument: '{}'",
+                                "{}.__init__ missing 1 required positional argument: {}. \
+This will become an error in Python 3.15.",
                                 zelf.class().name(),
-                                field.as_wtf8()
+                                field_repr
                             ));
                             warn::warn(
                                 message.into(),
@@ -461,6 +508,21 @@ Support for arbitrary keyword arguments is deprecated and will be removed in Pyt
                                 vm,
                             )?;
                         }
+                    } else {
+                        let field_repr = field.repr(vm)?;
+                        let message = vm.ctx.new_str(format!(
+                            "Field {} is missing from {}._field_types. \
+This will become an error in Python 3.15.",
+                            field_repr,
+                            zelf.class().name()
+                        ));
+                        warn::warn(
+                            message.into(),
+                            Some(vm.ctx.exceptions.deprecation_warning.to_owned()),
+                            1,
+                            None,
+                            vm,
+                        )?;
                     }
                 }
             }
@@ -510,9 +572,29 @@ Support for arbitrary keyword arguments is deprecated and will be removed in Pyt
             .map_err(|_| vm.new_type_error("AST is not a type"))?;
         let ctx = &vm.ctx;
         let empty_tuple = ctx.empty_tuple.clone();
+        let set_empty_annotations = |typ: &Py<PyType>| {
+            typ.set_str_attr("__annotations__", ctx.new_dict(), ctx);
+        };
+        set_empty_annotations(&ast_type);
         ast_type.set_str_attr("_fields", empty_tuple.clone(), ctx);
         ast_type.set_str_attr("_attributes", empty_tuple.clone(), ctx);
         ast_type.set_str_attr("__match_args__", empty_tuple, ctx);
+        for typ in [
+            super::super::pyast::NodeMod::static_type(),
+            super::super::pyast::NodeStmt::static_type(),
+            super::super::pyast::NodeExpr::static_type(),
+            super::super::pyast::NodeExprContext::static_type(),
+            super::super::pyast::NodeBoolOp::static_type(),
+            super::super::pyast::NodeOperator::static_type(),
+            super::super::pyast::NodeUnaryOp::static_type(),
+            super::super::pyast::NodeCmpOp::static_type(),
+            super::super::pyast::NodeExceptHandler::static_type(),
+            super::super::pyast::NodePattern::static_type(),
+            super::super::pyast::NodeTypeIgnore::static_type(),
+            super::super::pyast::NodeTypeParam::static_type(),
+        ] {
+            set_empty_annotations(typ);
+        }
 
         const AST_REDUCE: PyMethodDef = PyMethodDef::new_const(
             "__reduce__",

crates/vm/src/stdlib/_ast/repr.rs

@@ -1,7 +1,8 @@
 use crate::{
     AsObject, PyObjectRef, PyResult, VirtualMachine,
-    builtins::{PyList, PyTuple},
+    builtins::{PyList, PyStr, PyTuple},
     class::PyClassImpl,
+    recursion::ReprGuard,
     stdlib::_ast::NodeAst,
 };
 use rustpython_common::wtf8::Wtf8Buf;
@@ -33,9 +34,7 @@ fn repr_ast_list(vm: &VirtualMachine, items: Vec<PyObjectRef>, depth: usize) ->
         rendered.push_wtf8(&parts[0]);
     }
     if items.len() > 2 {
-        if !parts[0].is_empty() {
-            rendered.push_wtf8(", ...".as_ref());
-        }
+        rendered.push_wtf8(", ...".as_ref());
         if parts.len() > 1 {
             rendered.push_wtf8(", ".as_ref());
             rendered.push_wtf8(&parts[1]);
@@ -75,9 +74,7 @@ fn repr_ast_tuple(vm: &VirtualMachine, items: Vec<PyObjectRef>, depth: usize) ->
         rendered.push_wtf8(&parts[0]);
     }
     if items.len() > 2 {
-        if !parts[0].is_empty() {
-            rendered.push_wtf8(", ...".as_ref());
-        }
+        rendered.push_wtf8(", ...".as_ref());
         if parts.len() > 1 {
             rendered.push_wtf8(", ".as_ref());
             rendered.push_wtf8(&parts[1]);
@@ -86,9 +83,6 @@ fn repr_ast_tuple(vm: &VirtualMachine, items: Vec<PyObjectRef>, depth: usize) ->
         rendered.push_wtf8(", ".as_ref());
         rendered.push_wtf8(&parts[1]);
     }
-    if items.len() == 1 {
-        rendered.push_wtf8(",".as_ref());
-    }
     rendered.push_wtf8(")".as_ref());
     Ok(rendered)
 }
@@ -104,18 +98,24 @@ pub(crate) fn repr_ast_node(
         s.push_wtf8("(...)".as_ref());
         return Ok(s);
     }
+    let Some(_guard) = ReprGuard::enter(vm, obj.as_object()) else {
+        let mut s = Wtf8Buf::from(&*cls.name());
+        s.push_wtf8("(...)".as_ref());
+        return Ok(s);
+    };
 
-    let fields = cls.get_attr(vm.ctx.intern_str("_fields"));
-    let fields = match fields {
-        Some(fields) => fields.try_to_value::<Vec<crate::builtins::PyStrRef>>(vm)?,
+    let fields = match cls.get_attr(vm.ctx.intern_str("_fields")) {
+        Some(fields) => fields,
         None => {
             let mut s = Wtf8Buf::from(&*cls.name());
             s.push_wtf8("(...)".as_ref());
             return Ok(s);
         }
     };
+    let fields = fields.sequence_unchecked();
+    let numfields = fields.length(vm)?;
 
-    if fields.is_empty() {
+    if numfields == 0 {
         let mut s = Wtf8Buf::from(&*cls.name());
         s.push_wtf8("()".as_ref());
         return Ok(s);
@@ -124,8 +124,12 @@ pub(crate) fn repr_ast_node(
     let mut rendered = Wtf8Buf::from(&*cls.name());
     rendered.push_wtf8("(".as_ref());
 
-    for (idx, field) in fields.iter().enumerate() {
-        let value = obj.get_attr(field, vm)?;
+    for idx in 0..numfields {
+        let field = fields.get_item(idx as isize, vm)?;
+        let field = field
+            .downcast::<PyStr>()
+            .map_err(|_| vm.new_type_error("attribute name must be string"))?;
+        let value = obj.get_attr(&field, vm)?;
         let value_repr = if value.fast_isinstance(vm.ctx.types.list_type) {
             let list = value
                 .downcast::<PyList>()

crates/vm/src/stdlib/_ast/string.rs

@@ -19,6 +19,7 @@ fn ruff_fstring_element_into_iter(
 }
 
 fn ruff_fstring_element_to_joined_str_part(
+    vm: &VirtualMachine,
     element: ast::InterpolatedStringElement,
 ) -> JoinedStrPart {
     match element {
@@ -38,12 +39,21 @@ fn ruff_fstring_element_to_joined_str_part(
             conversion,
             format_spec,
             node_index: _,
-        }) => JoinedStrPart::FormattedValue(FormattedValue {
-            value: expression,
-            conversion,
-            format_spec: ruff_format_spec_to_joined_str(format_spec),
-            range,
-        }),
+            runtime_str: _,
+            runtime_interpolation_format_spec: _,
+            runtime_formatted_value_format_spec,
+        }) => {
+            let runtime_format_spec = runtime_formatted_value_format_spec.or_else(|| {
+                ruff_format_spec_to_joined_str(vm, format_spec)
+                    .map(|joined_str| Box::new(joined_str.into_expr(false)))
+            });
+            JoinedStrPart::FormattedValue(FormattedValue {
+                value: expression,
+                conversion,
+                format_spec: runtime_format_spec,
+                range,
+            })
+        }
     }
 }
 
@@ -235,6 +245,7 @@ fn warn_invalid_escape_sequences_in_format_spec(
 }
 
 fn ruff_format_spec_to_joined_str(
+    vm: &VirtualMachine,
     format_spec: Option<Box<ast::InterpolatedStringFormatSpec>>,
 ) -> Option<Box<JoinedStr>> {
     match format_spec {
@@ -254,10 +265,14 @@ fn ruff_format_spec_to_joined_str(
                 range
             };
             let values: Vec<_> = ruff_fstring_element_into_iter(elements)
-                .map(ruff_fstring_element_to_joined_str_part)
+                .map(|element| ruff_fstring_element_to_joined_str_part(vm, element))
                 .collect();
             let values = normalize_joined_str_parts(values).into_boxed_slice();
-            Some(Box::new(JoinedStr { range, values }))
+            Some(Box::new(JoinedStr {
+                range,
+                values,
+                runtime_values: None,
+            }))
         }
     }
 }
@@ -290,40 +305,98 @@ fn ruff_fstring_element_to_ruff_fstring_part(
     }
 }
 
-fn joined_str_to_ruff_format_spec(
-    joined_str: Option<Box<JoinedStr>>,
+fn format_spec_expr_to_ruff_format_spec(
+    format_spec: Option<Box<ast::Expr>>,
 ) -> Option<Box<ast::InterpolatedStringFormatSpec>> {
-    match joined_str {
-        None => None,
-        Some(joined_str) => {
-            let JoinedStr { range, values } = *joined_str;
-            let elements: Vec<_> = Box::into_iter(values)
-                .map(joined_str_part_to_ruff_fstring_element)
-                .collect();
-            let format_spec = ast::InterpolatedStringFormatSpec {
-                node_index: Default::default(),
+    let format_spec = format_spec?;
+    let ast::Expr::FString(mut fstring) = *format_spec else {
+        return None;
+    };
+    let ast::ExprFString {
+        range,
+        ref mut value,
+        node_index: _,
+        runtime_joined_str: _,
+        runtime_values: _,
+    } = fstring;
+    let default_part = ast::FStringPart::FString(ast::FString {
+        node_index: Default::default(),
+        range: Default::default(),
+        elements: Default::default(),
+        flags: ast::FStringFlags::empty(),
+    });
+    let mut elements = Vec::new();
+    for i in 0..value.as_slice().len() {
+        let part = core::mem::replace(value.iter_mut().nth(i).unwrap(), default_part.clone());
+        match part {
+            ast::FStringPart::Literal(ast::StringLiteral {
                 range,
-                elements: elements.into(),
-            };
-            Some(Box::new(format_spec))
+                value,
+                node_index: _,
+                flags: _,
+            }) => elements.push(ast::InterpolatedStringElement::Literal(
+                ast::InterpolatedStringLiteralElement {
+                    node_index: Default::default(),
+                    range,
+                    value,
+                },
+            )),
+            ast::FStringPart::FString(ast::FString {
+                elements: fstring_elements,
+                ..
+            }) => {
+                elements.extend(ruff_fstring_element_into_iter(fstring_elements));
+            }
         }
     }
+    Some(Box::new(ast::InterpolatedStringFormatSpec {
+        node_index: Default::default(),
+        range,
+        elements: elements.into(),
+    }))
 }
 
 #[derive(Debug)]
 pub(super) struct JoinedStr {
     pub(super) range: TextRange,
     pub(super) values: Box<[JoinedStrPart]>,
+    pub(super) runtime_values: Option<Vec<Option<ast::Expr>>>,
 }
 
 impl JoinedStr {
-    pub(super) fn into_expr(self) -> ast::Expr {
-        let Self { range, values } = self;
+    pub(super) fn into_expr(self, from_ast_object: bool) -> ast::Expr {
+        let Self {
+            range,
+            values,
+            runtime_values: mut raw_runtime_values,
+        } = self;
+        let values = if values.iter().any(joined_str_part_requires_runtime_values) {
+            if raw_runtime_values.is_none() {
+                raw_runtime_values = Some(
+                    values
+                        .into_vec()
+                        .into_iter()
+                        .map(|part| joined_str_part_to_expr(from_ast_object, part))
+                        .map(Some)
+                        .collect(),
+                );
+            }
+            Vec::new().into_boxed_slice()
+        } else {
+            values
+        };
+        let (runtime_joined_str, runtime_values) =
+            raw_runtime_values.take().map_or((None, None), |values| {
+                if values.iter().any(Option::is_none) {
+                    (None, Some(values))
+                } else {
+                    (Some(values.into_iter().flatten().collect()), None)
+                }
+            });
         ast::Expr::FString(ast::ExprFString {
             node_index: Default::default(),
-            range: Default::default(),
+            range,
             value: match values.len() {
-                // ruff represents an empty fstring like this:
                 0 => ast::FStringValue::single(ast::FString {
                     node_index: Default::default(),
                     range,
@@ -332,7 +405,8 @@ impl JoinedStr {
                 }),
                 1 => ast::FStringValue::single(
                     Box::<[_]>::into_iter(values)
-                        .map(joined_str_part_to_ruff_fstring_element)
+                        .map(|part| joined_str_part_to_ruff_fstring_element(from_ast_object, part))
+                        .map(Option::unwrap)
                         .map(|element| ast::FString {
                             node_index: Default::default(),
                             range,
@@ -344,54 +418,108 @@ impl JoinedStr {
                 ),
                 _ => ast::FStringValue::concatenated(
                     Box::<[_]>::into_iter(values)
-                        .map(joined_str_part_to_ruff_fstring_element)
+                        .map(|part| joined_str_part_to_ruff_fstring_element(from_ast_object, part))
+                        .map(Option::unwrap)
                         .map(ruff_fstring_element_to_ruff_fstring_part)
                         .collect(),
                 ),
             },
+            runtime_joined_str,
+            runtime_values,
         })
     }
 }
 
-fn joined_str_part_to_ruff_fstring_element(part: JoinedStrPart) -> ast::InterpolatedStringElement {
+fn joined_str_part_requires_runtime_values(part: &JoinedStrPart) -> bool {
+    matches!(
+        part,
+        JoinedStrPart::Constant(Constant {
+            value,
+            ..
+        }) if !matches!(value, ConstantLiteral::Str { .. })
+    )
+}
+
+fn joined_str_part_to_expr(from_ast_object: bool, part: JoinedStrPart) -> ast::Expr {
+    match part {
+        JoinedStrPart::FormattedValue(value) => formatted_value_to_expr(from_ast_object, value),
+        JoinedStrPart::Constant(value) => value.into_expr(),
+    }
+}
+
+fn joined_str_part_to_ruff_fstring_element(
+    from_ast_object: bool,
+    part: JoinedStrPart,
+) -> Option<ast::InterpolatedStringElement> {
     match part {
         JoinedStrPart::FormattedValue(value) => {
-            ast::InterpolatedStringElement::Interpolation(ast::InterpolatedElement {
-                node_index: Default::default(),
-                range: value.range,
-                expression: value.value.clone(),
-                debug_text: None, // TODO: What is this?
-                conversion: value.conversion,
-                format_spec: joined_str_to_ruff_format_spec(value.format_spec),
-            })
+            let format_spec = value.format_spec.clone();
+            let runtime_formatted_value_format_spec = (from_ast_object && format_spec.is_some())
+                .then_some(format_spec.clone())
+                .flatten();
+            Some(ast::InterpolatedStringElement::Interpolation(
+                ast::InterpolatedElement {
+                    node_index: Default::default(),
+                    range: value.range,
+                    expression: value.value.clone(),
+                    debug_text: None,
+                    conversion: value.conversion,
+                    format_spec: format_spec_expr_to_ruff_format_spec(format_spec),
+                    runtime_str: None,
+                    runtime_interpolation_format_spec: None,
+                    runtime_formatted_value_format_spec,
+                },
+            ))
         }
         JoinedStrPart::Constant(value) => {
-            ast::InterpolatedStringElement::Literal(ast::InterpolatedStringLiteralElement {
-                node_index: Default::default(),
-                range: value.range,
-                value: match value.value {
-                    ConstantLiteral::Str { value, .. } => value,
-                    _ => todo!(),
+            let Constant { range, value, .. } = value;
+            let ConstantLiteral::Str { value, .. } = value else {
+                return None;
+            };
+            Some(ast::InterpolatedStringElement::Literal(
+                ast::InterpolatedStringLiteralElement {
+                    node_index: Default::default(),
+                    range,
+                    value,
                 },
-            })
+            ))
         }
     }
 }
 
 // constructor
+pub(super) fn joined_str_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<JoinedStr> {
+    let values: Vec<Option<ast::Expr>> =
+        get_node_list_field(vm, source_file, &object, "values", "JoinedStr")?;
+    Ok(JoinedStr {
+        values: Vec::new().into_boxed_slice(),
+        runtime_values: Some(values),
+        range,
+    })
+}
+
 impl Node for JoinedStr {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { values, range } = self;
+        let Self {
+            values,
+            runtime_values,
+            range,
+        } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeExprJoinedStr::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item(
-            "values",
-            BoxedSlice(values).ast_to_object(vm, source_file),
-            vm,
-        )
-        .unwrap();
+        let values = if let Some(runtime_values) = runtime_values {
+            BoxedSlice(runtime_values.into_boxed_slice()).ast_to_object(vm, source_file)
+        } else {
+            BoxedSlice(values).ast_to_object(vm, source_file)
+        };
+        dict.set_item("values", values, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -400,15 +528,8 @@ impl Node for JoinedStr {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let values: BoxedSlice<_> = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "values", "JoinedStr")?,
-        )?;
-        Ok(Self {
-            values: values.0,
-            range: range_from_object(vm, source_file, object, "JoinedStr")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "JoinedStr")?;
+        joined_str_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -431,8 +552,7 @@ impl Node for JoinedStrPart {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        if cls.is(pyast::NodeExprFormattedValue::static_type()) {
+        if is_node_instance(vm, &object, pyast::NodeExprFormattedValue::static_type())? {
             Ok(Self::FormattedValue(Node::ast_from_object(
                 vm,
                 source_file,
@@ -452,11 +572,31 @@ impl Node for JoinedStrPart {
 pub(super) struct FormattedValue {
     value: Box<ast::Expr>,
     conversion: ast::ConversionFlag,
-    format_spec: Option<Box<JoinedStr>>,
+    format_spec: Option<Box<ast::Expr>>,
     range: TextRange,
 }
 
 // constructor
+pub(super) fn formatted_value_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<FormattedValue> {
+    Ok(FormattedValue {
+        value: get_required_node_field(vm, source_file, &object, "value", "FormattedValue")?,
+        conversion: Node::ast_from_object(
+            vm,
+            source_file,
+            get_node_field(vm, &object, "conversion", "FormattedValue")?,
+        )?,
+        format_spec: get_node_field_opt(vm, &object, "format_spec")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for FormattedValue {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -487,25 +627,24 @@ impl Node for FormattedValue {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            value: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "value", "FormattedValue")?,
-            )?,
-            conversion: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "conversion", "FormattedValue")?,
-            )?,
-            format_spec: get_node_field_opt(vm, &object, "format_spec")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            range: range_from_object(vm, source_file, object, "FormattedValue")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "FormattedValue")?;
+        formatted_value_from_object_with_range(vm, source_file, object, range)
     }
 }
 
+pub(super) fn formatted_value_to_expr(
+    from_ast_object: bool,
+    formatted: FormattedValue,
+) -> ast::Expr {
+    let range = formatted.range;
+    JoinedStr {
+        range,
+        values: vec![JoinedStrPart::FormattedValue(formatted)].into_boxed_slice(),
+        runtime_values: None,
+    }
+    .into_expr(from_ast_object)
+}
+
 pub(super) fn fstring_to_object(
     vm: &VirtualMachine,
     source_file: &SourceFile,
@@ -515,7 +654,27 @@ pub(super) fn fstring_to_object(
         range,
         mut value,
         node_index: _,
+        runtime_joined_str,
+        runtime_values,
     } = expression;
+    if let Some(joined_str) = runtime_joined_str {
+        return JoinedStr {
+            range,
+            values: Vec::new().into_boxed_slice(),
+            runtime_values: Some(joined_str.into_iter().map(Some).collect()),
+        }
+        .ast_to_object(vm, source_file);
+    }
+
+    if let Some(values) = runtime_values {
+        return JoinedStr {
+            range,
+            values: Vec::new().into_boxed_slice(),
+            runtime_values: Some(values),
+        }
+        .ast_to_object(vm, source_file);
+    }
+
     let default_part = ast::FStringPart::FString(ast::FString {
         node_index: Default::default(),
         range: Default::default(),
@@ -545,7 +704,7 @@ pub(super) fn fstring_to_object(
                 node_index: _,
             }) => {
                 for element in ruff_fstring_element_into_iter(elements) {
-                    values.push(ruff_fstring_element_to_joined_str_part(element));
+                    values.push(ruff_fstring_element_to_joined_str_part(vm, element));
                 }
             }
         }
@@ -555,12 +714,13 @@ pub(super) fn fstring_to_object(
         if let JoinedStrPart::FormattedValue(value) = part
             && let Some(format_spec) = &value.format_spec
         {
-            warn_invalid_escape_sequences_in_format_spec(vm, source_file, format_spec.range);
+            warn_invalid_escape_sequences_in_format_spec(vm, source_file, format_spec.range());
         }
     }
     let c = JoinedStr {
         range,
         values: values.into_boxed_slice(),
+        runtime_values: None,
     };
     c.ast_to_object(vm, source_file)
 }
@@ -568,8 +728,9 @@ pub(super) fn fstring_to_object(
 // ===== TString (Template String) Support =====
 
 fn ruff_tstring_element_to_template_str_part(
-    element: ast::InterpolatedStringElement,
+    vm: &VirtualMachine,
     source_file: &SourceFile,
+    element: ast::InterpolatedStringElement,
 ) -> TemplateStrPart {
     match element {
         ast::InterpolatedStringElement::Literal(ast::InterpolatedStringLiteralElement {
@@ -588,6 +749,9 @@ fn ruff_tstring_element_to_template_str_part(
             conversion,
             format_spec,
             node_index: _,
+            runtime_str,
+            runtime_interpolation_format_spec,
+            runtime_formatted_value_format_spec: _,
         }) => {
             let expr_range =
                 extend_expr_range_with_wrapping_parens(source_file, range, expression.range())
@@ -604,11 +768,23 @@ fn ruff_tstring_element_to_template_str_part(
             } else {
                 tstring_interpolation_expr_str(source_file, range, expr_range)
             };
+            let runtime_interpolation = super::constant::runtime_interpolation_object(
+                vm,
+                runtime_str,
+                runtime_interpolation_format_spec,
+            );
             TemplateStrPart::Interpolation(TStringInterpolation {
                 value: expression,
-                str: expr_str,
+                str: runtime_interpolation
+                    .as_ref()
+                    .map_or_else(|| vm.ctx.new_str(expr_str).into(), |(str, _)| str.clone()),
                 conversion,
-                format_spec: ruff_format_spec_to_joined_str(format_spec),
+                format_spec: runtime_interpolation
+                    .and_then(|(_, format_spec)| format_spec)
+                    .or_else(|| {
+                        ruff_format_spec_to_joined_str(vm, format_spec)
+                            .map(|joined_str| Box::new(joined_str.into_expr(false)))
+                    }),
                 range,
             })
         }
@@ -695,34 +871,51 @@ fn strip_interpolation_expr(expr_source: &str) -> String {
 pub(super) struct TemplateStr {
     pub(super) range: TextRange,
     pub(super) values: Box<[TemplateStrPart]>,
+    pub(super) runtime_values: Option<Vec<Option<ast::Expr>>>,
 }
 
 pub(super) fn template_str_to_expr(
     vm: &VirtualMachine,
+    source_file: &SourceFile,
     template: TemplateStr,
 ) -> PyResult<ast::Expr> {
-    let TemplateStr { range, values } = template;
-    let elements = template_parts_to_elements(vm, values)?;
+    let TemplateStr {
+        range,
+        values,
+        runtime_values: raw_runtime_values,
+    } = template;
+    let elements = template_parts_to_elements(vm, source_file, values)?;
     let tstring = ast::TString {
         range,
         node_index: Default::default(),
         elements,
         flags: ast::TStringFlags::empty(),
     };
+    let (runtime_template_str, runtime_values) =
+        raw_runtime_values.map_or((None, None), |values| {
+            if values.iter().any(Option::is_none) {
+                (None, Some(values))
+            } else {
+                (Some(values.into_iter().flatten().collect()), None)
+            }
+        });
     Ok(ast::Expr::TString(ast::ExprTString {
         node_index: Default::default(),
         range,
         value: ast::TStringValue::single(tstring),
+        runtime_template_str,
+        runtime_values,
     }))
 }
 
 pub(super) fn interpolation_to_expr(
     vm: &VirtualMachine,
+    source_file: &SourceFile,
     interpolation: TStringInterpolation,
 ) -> PyResult<ast::Expr> {
+    let range = interpolation.range;
     let part = TemplateStrPart::Interpolation(interpolation);
-    let elements = template_parts_to_elements(vm, vec![part].into_boxed_slice())?;
-    let range = TextRange::default();
+    let elements = template_parts_to_elements(vm, source_file, vec![part].into_boxed_slice())?;
     let tstring = ast::TString {
         range,
         node_index: Default::default(),
@@ -733,22 +926,26 @@ pub(super) fn interpolation_to_expr(
         node_index: Default::default(),
         range,
         value: ast::TStringValue::single(tstring),
+        runtime_template_str: None,
+        runtime_values: None,
     }))
 }
 
 fn template_parts_to_elements(
     vm: &VirtualMachine,
+    source_file: &SourceFile,
     values: Box<[TemplateStrPart]>,
 ) -> PyResult<ast::InterpolatedStringElements> {
     let mut elements = Vec::with_capacity(values.len());
     for value in values.into_vec() {
-        elements.push(template_part_to_element(vm, value)?);
+        elements.push(template_part_to_element(vm, source_file, value)?);
     }
     Ok(ast::InterpolatedStringElements::from(elements))
 }
 
 fn template_part_to_element(
     vm: &VirtualMachine,
+    source_file: &SourceFile,
     part: TemplateStrPart,
 ) -> PyResult<ast::InterpolatedStringElement> {
     match part {
@@ -767,12 +964,18 @@ fn template_part_to_element(
         TemplateStrPart::Interpolation(interpolation) => {
             let TStringInterpolation {
                 value,
+                str,
                 conversion,
                 format_spec,
                 range,
-                ..
             } = interpolation;
-            let format_spec = joined_str_to_ruff_format_spec(format_spec);
+            let str_constant =
+                super::constant::constant_object_to_constant_data(vm, source_file, str)?;
+            let runtime_str = Some(super::constant::constant_data_to_ast_constant_value(
+                str_constant,
+            ));
+            let runtime_interpolation_format_spec = format_spec.clone();
+            let format_spec = format_spec_expr_to_ruff_format_spec(format_spec);
             Ok(ast::InterpolatedStringElement::Interpolation(
                 ast::InterpolatedElement {
                     range,
@@ -781,6 +984,9 @@ fn template_part_to_element(
                     debug_text: None,
                     conversion,
                     format_spec,
+                    runtime_str,
+                    runtime_interpolation_format_spec,
+                    runtime_formatted_value_format_spec: None,
                 },
             ))
         }
@@ -788,19 +994,38 @@ fn template_part_to_element(
 }
 
 // constructor
+pub(super) fn template_str_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<TemplateStr> {
+    let values: Vec<Option<ast::Expr>> =
+        get_node_list_field(vm, source_file, &object, "values", "TemplateStr")?;
+    Ok(TemplateStr {
+        values: Vec::new().into_boxed_slice(),
+        runtime_values: Some(values),
+        range,
+    })
+}
+
 impl Node for TemplateStr {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { values, range } = self;
+        let Self {
+            values,
+            runtime_values,
+            range,
+        } = self;
         let node = NodeAst
             .into_ref_with_type(vm, pyast::NodeExprTemplateStr::static_type().to_owned())
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item(
-            "values",
-            BoxedSlice(values).ast_to_object(vm, source_file),
-            vm,
-        )
-        .unwrap();
+        let values = if let Some(runtime_values) = runtime_values {
+            BoxedSlice(runtime_values.into_boxed_slice()).ast_to_object(vm, source_file)
+        } else {
+            BoxedSlice(values).ast_to_object(vm, source_file)
+        };
+        dict.set_item("values", values, vm).unwrap();
         node_add_location(&dict, range, vm, source_file);
         node.into()
     }
@@ -809,15 +1034,8 @@ impl Node for TemplateStr {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let values: BoxedSlice<_> = Node::ast_from_object(
-            vm,
-            source_file,
-            get_node_field(vm, &object, "values", "TemplateStr")?,
-        )?;
-        Ok(Self {
-            values: values.0,
-            range: range_from_object(vm, source_file, object, "TemplateStr")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "TemplateStr")?;
+        template_str_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -840,8 +1058,7 @@ impl Node for TemplateStrPart {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        if cls.is(pyast::NodeExprInterpolation::static_type()) {
+        if is_node_instance(vm, &object, pyast::NodeExprInterpolation::static_type())? {
             Ok(Self::Interpolation(Node::ast_from_object(
                 vm,
                 source_file,
@@ -860,13 +1077,38 @@ impl Node for TemplateStrPart {
 #[derive(Debug)]
 pub(super) struct TStringInterpolation {
     value: Box<ast::Expr>,
-    str: String,
+    str: PyObjectRef,
     conversion: ast::ConversionFlag,
-    format_spec: Option<Box<JoinedStr>>,
+    format_spec: Option<Box<ast::Expr>>,
     range: TextRange,
 }
 
 // constructor
+pub(super) fn tstring_interpolation_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<TStringInterpolation> {
+    let value = get_required_node_field(vm, source_file, &object, "value", "Interpolation")?;
+    let str = get_node_field(vm, &object, "str", "Interpolation")?;
+    let conversion = Node::ast_from_object(
+        vm,
+        source_file,
+        get_node_field(vm, &object, "conversion", "Interpolation")?,
+    )?;
+    let format_spec: Option<Box<ast::Expr>> = get_node_field_opt(vm, &object, "format_spec")?
+        .map(|obj| Node::ast_from_object(vm, source_file, obj))
+        .transpose()?;
+    Ok(TStringInterpolation {
+        value,
+        str,
+        conversion,
+        format_spec,
+        range,
+    })
+}
+
 impl Node for TStringInterpolation {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -882,8 +1124,7 @@ impl Node for TStringInterpolation {
         let dict = node.as_object().dict().unwrap();
         dict.set_item("value", value.ast_to_object(vm, source_file), vm)
             .unwrap();
-        dict.set_item("str", vm.ctx.new_str(str).into(), vm)
-            .unwrap();
+        dict.set_item("str", str, vm).unwrap();
         dict.set_item("conversion", conversion.ast_to_object(vm, source_file), vm)
             .unwrap();
         dict.set_item(
@@ -900,25 +1141,8 @@ impl Node for TStringInterpolation {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let str_obj = get_node_field(vm, &object, "str", "Interpolation")?;
-        let str_val: String = str_obj.try_into_value(vm)?;
-        Ok(Self {
-            value: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "value", "Interpolation")?,
-            )?,
-            str: str_val,
-            conversion: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "conversion", "Interpolation")?,
-            )?,
-            format_spec: get_node_field_opt(vm, &object, "format_spec")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            range: range_from_object(vm, source_file, object, "Interpolation")?,
-        })
+        let range = range_from_object(vm, source_file, object.clone(), "Interpolation")?;
+        tstring_interpolation_from_object_with_range(vm, source_file, object, range)
     }
 }
 
@@ -931,7 +1155,42 @@ pub(super) fn tstring_to_object(
         range,
         mut value,
         node_index: _,
+        runtime_template_str,
+        runtime_values,
     } = expression;
+    if let Some(template_str) = runtime_template_str {
+        return TemplateStr {
+            range,
+            values: Vec::new().into_boxed_slice(),
+            runtime_values: Some(template_str.into_iter().map(Some).collect()),
+        }
+        .ast_to_object(vm, source_file);
+    }
+
+    if let Some(values) = runtime_values {
+        return TemplateStr {
+            range,
+            values: Vec::new().into_boxed_slice(),
+            runtime_values: Some(values),
+        }
+        .ast_to_object(vm, source_file);
+    }
+
+    if let [tstring] = value.as_slice()
+        && let Some(ast::InterpolatedStringElement::Interpolation(interp)) =
+            tstring.elements.iter().next()
+        && tstring.elements.get(1).is_none()
+        && let Some((str, format_spec)) = super::constant::runtime_interpolation_object(
+            vm,
+            interp.runtime_str.clone(),
+            interp.runtime_interpolation_format_spec.clone(),
+        )
+        && let Some(interpolation) =
+            standalone_tstring_interpolation_to_object(vm, source_file, &value, str, format_spec)
+    {
+        return interpolation;
+    }
+
     let default_tstring = ast::TString {
         node_index: Default::default(),
         range: Default::default(),
@@ -943,8 +1202,9 @@ pub(super) fn tstring_to_object(
         let tstring = core::mem::replace(value.iter_mut().nth(i).unwrap(), default_tstring.clone());
         for element in ruff_fstring_element_into_iter(tstring.elements) {
             values.push(ruff_tstring_element_to_template_str_part(
-                element,
+                vm,
                 source_file,
+                element,
             ));
         }
     }
@@ -952,6 +1212,37 @@ pub(super) fn tstring_to_object(
     let c = TemplateStr {
         range,
         values: values.into_boxed_slice(),
+        runtime_values: None,
     };
     c.ast_to_object(vm, source_file)
 }
+
+fn standalone_tstring_interpolation_to_object(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    value: &ast::TStringValue,
+    str: PyObjectRef,
+    format_spec: Option<Box<ast::Expr>>,
+) -> Option<PyObjectRef> {
+    let [tstring] = value.as_slice() else {
+        return None;
+    };
+    let mut elements = tstring.elements.iter();
+    let ast::InterpolatedStringElement::Interpolation(interp) = elements.next()? else {
+        return None;
+    };
+    if elements.next().is_some() {
+        return None;
+    }
+    let interpolation = TStringInterpolation {
+        value: interp.expression.clone(),
+        str,
+        conversion: interp.conversion,
+        format_spec: format_spec.or_else(|| {
+            ruff_format_spec_to_joined_str(vm, interp.format_spec.clone())
+                .map(|joined_str| Box::new(joined_str.into_expr(false)))
+        }),
+        range: interp.range,
+    };
+    Some(interpolation.ast_to_object(vm, source_file))
+}

crates/vm/src/stdlib/_ast/type_ignore.rs

@@ -2,6 +2,7 @@ use super::*;
 use rustpython_compiler_core::SourceFile;
 
 pub(super) enum TypeIgnore {
+    None,
     TypeIgnore(TypeIgnoreTypeIgnore),
 }
 
@@ -9,6 +10,7 @@ pub(super) enum TypeIgnore {
 impl Node for TypeIgnore {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         match self {
+            Self::None => vm.ctx.none(),
             Self::TypeIgnore(cons) => cons.ast_to_object(vm, source_file),
         }
     }
@@ -17,8 +19,9 @@ impl Node for TypeIgnore {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeTypeIgnoreTypeIgnore::static_type()) {
+        Ok(if vm.is_none(&object) {
+            Self::None
+        } else if is_node_instance(vm, &object, pyast::NodeTypeIgnoreTypeIgnore::static_type())? {
             Self::TypeIgnore(TypeIgnoreTypeIgnore::ast_from_object(
                 vm,
                 source_file,
@@ -34,15 +37,14 @@ impl Node for TypeIgnore {
 }
 
 pub(super) struct TypeIgnoreTypeIgnore {
-    range: TextRange,
-    lineno: PyRefExact<PyInt>,
-    tag: PyRefExact<PyStr>,
+    lineno: i32,
+    tag: PyObjectRef,
 }
 
 // constructor
 impl Node for TypeIgnoreTypeIgnore {
-    fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        let Self { lineno, tag, range } = self;
+    fn ast_to_object(self, vm: &VirtualMachine, _source_file: &SourceFile) -> PyObjectRef {
+        let Self { lineno, tag } = self;
         let node = NodeAst
             .into_ref_with_type(
                 vm,
@@ -50,25 +52,20 @@ impl Node for TypeIgnoreTypeIgnore {
             )
             .unwrap();
         let dict = node.as_object().dict().unwrap();
-        dict.set_item("lineno", lineno.to_pyobject(vm), vm).unwrap();
-        dict.set_item("tag", tag.to_pyobject(vm), vm).unwrap();
-        node_add_location(&dict, range, vm, source_file);
+        dict.set_item("lineno", vm.ctx.new_int(lineno).into(), vm)
+            .unwrap();
+        dict.set_item("tag", tag, vm).unwrap();
         node.into()
     }
 
     fn ast_from_object(
         vm: &VirtualMachine,
-        source_file: &SourceFile,
+        _source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
         Ok(Self {
-            lineno: get_node_field(vm, &object, "lineno", "TypeIgnore")?
-                .downcast_exact(vm)
-                .unwrap(),
-            tag: get_node_field(vm, &object, "tag", "TypeIgnore")?
-                .downcast_exact(vm)
-                .unwrap(),
-            range: range_from_object(vm, source_file, object, "TypeIgnore")?,
+            lineno: get_int_field(vm, &object, "lineno", "TypeIgnore")?,
+            tag: node_object_to_ast_string(vm, get_node_field(vm, &object, "tag", "TypeIgnore")?)?,
         })
     }
 }

crates/vm/src/stdlib/_ast/type_parameters.rs

@@ -3,7 +3,10 @@ use rustpython_compiler_core::SourceFile;
 
 impl Node for ast::TypeParams {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
-        self.type_params.ast_to_object(vm, source_file)
+        self.runtime_type_params.map_or_else(
+            || self.type_params.ast_to_object(vm, source_file),
+            |values| values.ast_to_object(vm, source_file),
+        )
     }
 
     fn ast_from_object(
@@ -11,22 +14,51 @@ impl Node for ast::TypeParams {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let type_params: Vec<ast::TypeParam> = Node::ast_from_object(vm, source_file, object)?;
-        let range = Option::zip(type_params.first(), type_params.last())
-            .map(|(first, last)| first.range().cover(last.range()))
-            .unwrap_or_default();
-        Ok(Self {
-            node_index: Default::default(),
-            type_params,
-            range,
-        })
+        Ok(type_params_from_values(
+            vm,
+            Node::ast_from_object(vm, source_file, object)?,
+        ))
     }
 
     fn is_none(&self) -> bool {
-        self.type_params.is_empty()
+        self.type_params.is_empty() && self.runtime_type_params.is_none()
     }
 }
 
+pub(super) fn type_params_from_field(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: &PyObject,
+    field: &'static str,
+    typ: &str,
+) -> PyResult<Option<Box<ast::TypeParams>>> {
+    let type_params: Vec<Option<ast::TypeParam>> =
+        get_node_list_field(vm, source_file, object, field, typ)?;
+    let type_params = type_params_from_values(vm, type_params);
+    Ok((!type_params.is_none()).then_some(Box::new(type_params)))
+}
+
+fn type_params_from_values(
+    _vm: &VirtualMachine,
+    values: Vec<Option<ast::TypeParam>>,
+) -> ast::TypeParams {
+    let runtime_type_params = values.iter().any(Option::is_none).then(|| values.clone());
+    let type_params = lower_nullable_type_params(&values);
+    let range = Option::zip(type_params.first(), type_params.last())
+        .map(|(first, last)| first.range().cover(last.range()))
+        .unwrap_or_default();
+    ast::TypeParams {
+        node_index: Default::default(),
+        type_params,
+        range,
+        runtime_type_params,
+    }
+}
+
+fn lower_nullable_type_params(values: &[Option<ast::TypeParam>]) -> Vec<ast::TypeParam> {
+    values.iter().filter_map(Clone::clone).collect()
+}
+
 // sum
 impl Node for ast::TypeParam {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
@@ -42,35 +74,70 @@ impl Node for ast::TypeParam {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        let cls = object.class();
-        Ok(if cls.is(pyast::NodeTypeParamTypeVar::static_type()) {
-            Self::TypeVar(ast::TypeParamTypeVar::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodeTypeParamParamSpec::static_type()) {
-            Self::ParamSpec(ast::TypeParamParamSpec::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
-        } else if cls.is(pyast::NodeTypeParamTypeVarTuple::static_type()) {
-            Self::TypeVarTuple(ast::TypeParamTypeVarTuple::ast_from_object(
-                vm,
-                source_file,
-                object,
-            )?)
+        if vm.is_none(&object) {
+            return Err(vm.new_type_error(format!(
+                "expected some sort of type_param, but got {}",
+                object.repr(vm)?
+            )));
+        }
+        enum TypeParamKind {
+            TypeVar,
+            ParamSpec,
+            TypeVarTuple,
+        }
+        let kind = if is_node_instance(vm, &object, pyast::NodeTypeParamTypeVar::static_type())? {
+            TypeParamKind::TypeVar
+        } else if is_node_instance(vm, &object, pyast::NodeTypeParamParamSpec::static_type())? {
+            TypeParamKind::ParamSpec
+        } else if is_node_instance(vm, &object, pyast::NodeTypeParamTypeVarTuple::static_type())? {
+            TypeParamKind::TypeVarTuple
         } else {
             return Err(vm.new_type_error(format!(
                 "expected some sort of type_param, but got {}",
                 object.repr(vm)?
             )));
+        };
+        let range = type_param_range_from_object(vm, source_file, object.clone())?;
+        Ok(match kind {
+            TypeParamKind::TypeVar => Self::TypeVar(type_var_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            TypeParamKind::ParamSpec => Self::ParamSpec(param_spec_from_object_with_range(
+                vm,
+                source_file,
+                object,
+                range,
+            )?),
+            TypeParamKind::TypeVarTuple => Self::TypeVarTuple(
+                type_var_tuple_from_object_with_range(vm, source_file, object, range)?,
+            ),
         })
     }
 }
 
 // constructor
+fn type_var_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::TypeParamTypeVar> {
+    Ok(ast::TypeParamTypeVar {
+        node_index: Default::default(),
+        name: get_required_identifier_field(vm, source_file, &object, "name", "TypeVar")?,
+        bound: get_node_field_opt(vm, &object, "bound")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        default: get_node_field_opt(vm, &object, "default_value")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for ast::TypeParamTypeVar {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -99,27 +166,28 @@ impl Node for ast::TypeParamTypeVar {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            name: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "name", "TypeVar")?,
-            )?,
-            bound: get_node_field_opt(vm, &object, "bound")?
-                .map(|obj| Node::ast_from_object(vm, source_file, obj))
-                .transpose()?,
-            default: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "default_value", "TypeVar")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "TypeVar")?,
-        })
+        let range = type_param_range_from_object(vm, source_file, object.clone())?;
+        type_var_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn param_spec_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::TypeParamParamSpec> {
+    Ok(ast::TypeParamParamSpec {
+        node_index: Default::default(),
+        name: get_required_identifier_field(vm, source_file, &object, "name", "ParamSpec")?,
+        default: get_node_field_opt(vm, &object, "default_value")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for ast::TypeParamParamSpec {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -145,24 +213,28 @@ impl Node for ast::TypeParamParamSpec {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            name: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "name", "ParamSpec")?,
-            )?,
-            default: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "default_value", "ParamSpec")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "ParamSpec")?,
-        })
+        let range = type_param_range_from_object(vm, source_file, object.clone())?;
+        param_spec_from_object_with_range(vm, source_file, object, range)
     }
 }
 
 // constructor
+fn type_var_tuple_from_object_with_range(
+    vm: &VirtualMachine,
+    source_file: &SourceFile,
+    object: PyObjectRef,
+    range: TextRange,
+) -> PyResult<ast::TypeParamTypeVarTuple> {
+    Ok(ast::TypeParamTypeVarTuple {
+        node_index: Default::default(),
+        name: get_required_identifier_field(vm, source_file, &object, "name", "TypeVarTuple")?,
+        default: get_node_field_opt(vm, &object, "default_value")?
+            .map(|obj| Node::ast_from_object(vm, source_file, obj))
+            .transpose()?,
+        range,
+    })
+}
+
 impl Node for ast::TypeParamTypeVarTuple {
     fn ast_to_object(self, vm: &VirtualMachine, source_file: &SourceFile) -> PyObjectRef {
         let Self {
@@ -191,19 +263,7 @@ impl Node for ast::TypeParamTypeVarTuple {
         source_file: &SourceFile,
         object: PyObjectRef,
     ) -> PyResult<Self> {
-        Ok(Self {
-            node_index: Default::default(),
-            name: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "name", "TypeVarTuple")?,
-            )?,
-            default: Node::ast_from_object(
-                vm,
-                source_file,
-                get_node_field(vm, &object, "default_value", "TypeVarTuple")?,
-            )?,
-            range: range_from_object(vm, source_file, object, "TypeVarTuple")?,
-        })
+        let range = type_param_range_from_object(vm, source_file, object.clone())?;
+        type_var_tuple_from_object_with_range(vm, source_file, object, range)
     }
 }

crates/vm/src/stdlib/_ast/validate.rs

@@ -1,8 +1,10 @@
 // spell-checker: ignore assignlist ifexp
 
 use super::module::Mod;
-use crate::{PyResult, VirtualMachine};
+use crate::{PyResult, VirtualMachine, compiler::CompileError};
 use ruff_python_ast as ast;
+use rustpython_codegen::error::{CodegenError, CodegenErrorType};
+use rustpython_compiler_core::bytecode::ConstantData;
 
 fn expr_context_name(ctx: ast::ExprContext) -> &'static str {
     match ctx {
@@ -13,6 +15,17 @@ fn expr_context_name(ctx: ast::ExprContext) -> &'static str {
     }
 }
 
+fn invalid_syntax_error(vm: &VirtualMachine) -> crate::builtins::PyBaseExceptionRef {
+    vm.new_syntax_error(
+        &CompileError::Codegen(CodegenError {
+            location: None,
+            error: CodegenErrorType::SyntaxError("invalid syntax".to_owned()),
+            source_path: "<unknown>".to_owned(),
+        }),
+        None,
+    )
+}
+
 fn validate_name(vm: &VirtualMachine, name: &ast::name::Name) -> PyResult<()> {
     match name.as_str() {
         "None" | "True" | "False" => Err(vm.new_value_error(format!(
@@ -30,6 +43,7 @@ fn validate_comprehension(vm: &VirtualMachine, gens: &[ast::Comprehension]) -> P
     for comp in gens {
         validate_expr(vm, &comp.target, ast::ExprContext::Store)?;
         validate_expr(vm, &comp.iter, ast::ExprContext::Load)?;
+        validate_runtime_expr_list_slots(vm, comp.runtime_ifs.as_ref(), ast::ExprContext::Load)?;
         validate_exprs(vm, &comp.ifs, ast::ExprContext::Load, false)?;
     }
     Ok(())
@@ -42,30 +56,49 @@ fn validate_keywords(vm: &VirtualMachine, keywords: &[ast::Keyword]) -> PyResult
     Ok(())
 }
 
+fn validate_parameter_annotation(vm: &VirtualMachine, parameter: &ast::Parameter) -> PyResult<()> {
+    if let Some(annotation) = &parameter.annotation {
+        validate_expr(vm, annotation, ast::ExprContext::Load)?;
+    }
+    Ok(())
+}
+
 fn validate_parameters(vm: &VirtualMachine, params: &ast::Parameters) -> PyResult<()> {
-    for param in params
-        .posonlyargs
-        .iter()
-        .chain(&params.args)
-        .chain(&params.kwonlyargs)
-    {
-        if let Some(annotation) = &param.parameter.annotation {
-            validate_expr(vm, annotation, ast::ExprContext::Load)?;
-        }
-        if let Some(default) = &param.default {
-            validate_expr(vm, default, ast::ExprContext::Load)?;
-        }
+    for param in params.posonlyargs.iter().chain(&params.args) {
+        validate_parameter_annotation(vm, &param.parameter)?;
     }
     if let Some(vararg) = &params.vararg
         && let Some(annotation) = &vararg.annotation
     {
         validate_expr(vm, annotation, ast::ExprContext::Load)?;
     }
+    for param in &params.kwonlyargs {
+        validate_parameter_annotation(vm, &param.parameter)?;
+    }
     if let Some(kwarg) = &params.kwarg
         && let Some(annotation) = &kwarg.annotation
     {
         validate_expr(vm, annotation, ast::ExprContext::Load)?;
     }
+    if let Some(defaults) = params.runtime_defaults.as_ref() {
+        for default in defaults {
+            let Some(default) = default else {
+                return Err(vm.new_value_error("None disallowed in expression list"));
+            };
+            validate_expr(vm, default, ast::ExprContext::Load)?;
+        }
+    } else {
+        for param in params.posonlyargs.iter().chain(&params.args) {
+            if let Some(default) = &param.default {
+                validate_expr(vm, default, ast::ExprContext::Load)?;
+            }
+        }
+    }
+    for param in &params.kwonlyargs {
+        if let Some(default) = &param.default {
+            validate_expr(vm, default, ast::ExprContext::Load)?;
+        }
+    }
     Ok(())
 }
 
@@ -99,8 +132,14 @@ fn validate_assignlist(
     validate_exprs(vm, targets, ctx, false)
 }
 
-fn validate_body(vm: &VirtualMachine, body: &[ast::Stmt], owner: &'static str) -> PyResult<()> {
+fn validate_body(
+    vm: &VirtualMachine,
+    body: &[ast::Stmt],
+    metadata: Option<&Vec<Option<ast::Stmt>>>,
+    owner: &'static str,
+) -> PyResult<()> {
     validate_nonempty_seq(vm, body.len(), "body", owner)?;
+    validate_runtime_stmt_list_slots(vm, metadata)?;
     validate_stmts(vm, body)
 }
 
@@ -111,34 +150,83 @@ fn validate_interpolated_elements<'a>(
     for element in elements {
         if let ast::InterpolatedStringElementRef::Interpolation(interpolation) = element {
             validate_expr(vm, &interpolation.expression, ast::ExprContext::Load)?;
-            if let Some(format_spec) = &interpolation.format_spec {
-                for spec_element in &format_spec.elements {
-                    if let ast::InterpolatedStringElement::Interpolation(spec_interp) = spec_element
-                    {
-                        validate_expr(vm, &spec_interp.expression, ast::ExprContext::Load)?;
-                    }
-                }
+            if let Some(format_spec) = interpolation.runtime_formatted_value_format_spec.as_deref()
+            {
+                validate_expr(vm, format_spec, ast::ExprContext::Load)?;
+            } else if let Some(format_spec) =
+                interpolation.runtime_interpolation_format_spec.as_deref()
+            {
+                validate_expr(vm, format_spec, ast::ExprContext::Load)?;
+            } else if let Some(format_spec) = &interpolation.format_spec {
+                validate_interpolated_elements(
+                    vm,
+                    format_spec
+                        .elements
+                        .iter()
+                        .map(ast::InterpolatedStringElementRef::from),
+                )?;
             }
         }
     }
     Ok(())
 }
 
+fn ensure_literal_number(expr: &ast::Expr, allow_real: bool, allow_imaginary: bool) -> bool {
+    let ast::Expr::NumberLiteral(number) = expr else {
+        return false;
+    };
+    match number.value {
+        ast::Number::Int(_) | ast::Number::Float(_) => allow_real,
+        ast::Number::Complex { .. } => allow_imaginary,
+    }
+}
+
+fn ensure_literal_negative(expr: &ast::Expr, allow_real: bool, allow_imaginary: bool) -> bool {
+    let ast::Expr::UnaryOp(unary) = expr else {
+        return false;
+    };
+    if unary.op != ast::UnaryOp::USub {
+        return false;
+    }
+    ensure_literal_number(&unary.operand, allow_real, allow_imaginary)
+}
+
+fn ensure_literal_complex(expr: &ast::Expr) -> bool {
+    let ast::Expr::BinOp(bin) = expr else {
+        return false;
+    };
+    if !matches!(bin.op, ast::Operator::Add | ast::Operator::Sub) {
+        return false;
+    }
+    let real_left = ensure_literal_number(&bin.left, true, false)
+        || ensure_literal_negative(&bin.left, true, false);
+    real_left && ensure_literal_number(&bin.right, false, true)
+}
+
+fn ast_constant_value(expr: &ast::Expr) -> Option<ConstantData> {
+    expr.as_constant_expr()
+        .map(|expr| super::constant::ast_constant_value_to_constant_data(expr.value.clone()))
+}
+
 fn validate_pattern_match_value(vm: &VirtualMachine, expr: &ast::Expr) -> PyResult<()> {
     validate_expr(vm, expr, ast::ExprContext::Load)?;
+    if let Some(constant) = ast_constant_value(expr) {
+        return match &constant {
+            ConstantData::Integer { .. }
+            | ConstantData::Float { .. }
+            | ConstantData::Bytes { .. }
+            | ConstantData::Complex { .. }
+            | ConstantData::Str { .. } => Ok(()),
+            _ => Err(vm.new_value_error("unexpected constant inside of a literal pattern")),
+        };
+    }
     match expr {
         ast::Expr::NumberLiteral(_) | ast::Expr::StringLiteral(_) | ast::Expr::BytesLiteral(_) => {
             Ok(())
         }
         ast::Expr::Attribute(_) => Ok(()),
-        ast::Expr::UnaryOp(op) => match &*op.operand {
-            ast::Expr::NumberLiteral(_) => Ok(()),
-            _ => Err(vm.new_value_error("patterns may only match literals and attribute lookups")),
-        },
-        ast::Expr::BinOp(bin) => match (&*bin.left, &*bin.right) {
-            (ast::Expr::NumberLiteral(_), ast::Expr::NumberLiteral(_)) => Ok(()),
-            _ => Err(vm.new_value_error("patterns may only match literals and attribute lookups")),
-        },
+        ast::Expr::UnaryOp(_) if ensure_literal_negative(expr, true, true) => Ok(()),
+        ast::Expr::BinOp(_) if ensure_literal_complex(expr) => Ok(()),
         ast::Expr::FString(_) | ast::Expr::TString(_) => Ok(()),
         ast::Expr::BooleanLiteral(_)
         | ast::Expr::NoneLiteral(_)
@@ -162,7 +250,10 @@ fn validate_pattern(vm: &VirtualMachine, pattern: &ast::Pattern, star_ok: bool)
         ast::Pattern::MatchSingleton(singleton) => match singleton.value {
             ast::Singleton::None | ast::Singleton::True | ast::Singleton::False => Ok(()),
         },
-        ast::Pattern::MatchSequence(seq) => validate_patterns(vm, &seq.patterns, true),
+        ast::Pattern::MatchSequence(seq) => {
+            validate_runtime_pattern_list_slots(vm, seq.runtime_patterns.as_ref())?;
+            validate_patterns(vm, &seq.patterns, true)
+        }
         ast::Pattern::MatchMapping(mapping) => {
             if mapping.keys.len() != mapping.patterns.len() {
                 return Err(vm.new_value_error(
@@ -172,15 +263,34 @@ fn validate_pattern(vm: &VirtualMachine, pattern: &ast::Pattern, star_ok: bool)
             if let Some(rest) = &mapping.rest {
                 validate_capture(vm, rest)?;
             }
+            validate_runtime_expr_option_list_slots(vm, mapping.runtime_keys.as_ref())?;
             for key in &mapping.keys {
-                if let ast::Expr::BooleanLiteral(_) | ast::Expr::NoneLiteral(_) = key {
+                if matches!(
+                    key,
+                    ast::Expr::BooleanLiteral(_)
+                        | ast::Expr::NoneLiteral(_)
+                        | ast::Expr::Constant(ast::ExprConstant {
+                            value: ast::ConstantValue::Boolean(_) | ast::ConstantValue::None,
+                            ..
+                        })
+                ) {
                     continue;
                 }
                 validate_pattern_match_value(vm, key)?;
             }
+            validate_runtime_pattern_list_slots(vm, mapping.runtime_patterns.as_ref())?;
             validate_patterns(vm, &mapping.patterns, false)
         }
         ast::Pattern::MatchClass(match_class) => {
+            if let (Some(kwd_attrs), Some(kwd_patterns)) = (
+                match_class.runtime_kwd_attrs.as_ref(),
+                match_class.runtime_kwd_patterns.as_ref(),
+            ) && kwd_attrs.len() != kwd_patterns.len()
+            {
+                return Err(vm.new_value_error(
+                    "MatchClass doesn't have the same number of keyword attributes as patterns",
+                ));
+            }
             validate_expr(vm, &match_class.cls, ast::ExprContext::Load)?;
             let mut cls = match_class.cls.as_ref();
             loop {
@@ -199,7 +309,13 @@ fn validate_pattern(vm: &VirtualMachine, pattern: &ast::Pattern, star_ok: bool)
             for keyword in &match_class.arguments.keywords {
                 validate_name(vm, keyword.attr.id())?;
             }
+            if let Some(patterns) = &match_class.runtime_patterns {
+                validate_runtime_nullable_patterns(vm, patterns)?;
+            }
             validate_patterns(vm, &match_class.arguments.patterns, false)?;
+            if let Some(kwd_patterns) = &match_class.runtime_kwd_patterns {
+                validate_runtime_nullable_patterns(vm, kwd_patterns)?;
+            }
             for keyword in &match_class.arguments.keywords {
                 validate_pattern(vm, &keyword.pattern, false)?;
             }
@@ -234,11 +350,80 @@ fn validate_pattern(vm: &VirtualMachine, pattern: &ast::Pattern, star_ok: bool)
             if match_or.patterns.len() < 2 {
                 return Err(vm.new_value_error("MatchOr requires at least 2 patterns"));
             }
+            validate_runtime_pattern_list_slots(vm, match_or.runtime_patterns.as_ref())?;
             validate_patterns(vm, &match_or.patterns, false)
         }
     }
 }
 
+fn validate_runtime_pattern_list_slots(
+    vm: &VirtualMachine,
+    values: Option<&Vec<Option<ast::Pattern>>>,
+) -> PyResult<()> {
+    if values.is_some_and(|values| values.iter().any(Option::is_none)) {
+        return Err(vm.new_value_error("unexpected pattern"));
+    }
+    Ok(())
+}
+
+fn validate_runtime_expr_option_list_slots(
+    vm: &VirtualMachine,
+    values: Option<&Vec<Option<ast::Expr>>>,
+) -> PyResult<()> {
+    if values.is_some_and(|values| values.iter().any(Option::is_none)) {
+        return Err(vm.new_value_error("None disallowed in expression list"));
+    }
+    Ok(())
+}
+
+fn validate_runtime_expr_list_slots(
+    vm: &VirtualMachine,
+    values: Option<&Vec<Option<ast::Expr>>>,
+    ctx: ast::ExprContext,
+) -> PyResult<()> {
+    if let Some(values) = values {
+        for value in values {
+            let Some(value) = value else {
+                return Err(vm.new_value_error("None disallowed in expression list"));
+            };
+            validate_expr(vm, value, ctx)?;
+        }
+    }
+    Ok(())
+}
+
+fn validate_runtime_stmt_list_slots(
+    vm: &VirtualMachine,
+    values: Option<&Vec<Option<ast::Stmt>>>,
+) -> PyResult<()> {
+    if let Some(values) = values
+        && values.iter().any(Option::is_none)
+    {
+        return Err(vm.new_value_error("None disallowed in statement list"));
+    }
+    Ok(())
+}
+
+fn validate_runtime_except_handler_list_slots(
+    vm: &VirtualMachine,
+    values: Option<&Vec<Option<ast::ExceptHandler>>>,
+) -> PyResult<()> {
+    if values.is_some_and(|values| values.iter().any(Option::is_none)) {
+        return Err(vm.new_value_error("unexpected excepthandler"));
+    }
+    Ok(())
+}
+
+fn validate_runtime_nullable_patterns(
+    vm: &VirtualMachine,
+    patterns: &[Option<ast::Pattern>],
+) -> PyResult<()> {
+    if patterns.iter().any(Option::is_none) {
+        return Err(vm.new_value_error("unexpected pattern"));
+    }
+    Ok(())
+}
+
 fn validate_patterns(
     vm: &VirtualMachine,
     patterns: &[ast::Pattern],
@@ -282,6 +467,12 @@ fn validate_type_params(
     type_params: Option<&ast::TypeParams>,
 ) -> PyResult<()> {
     if let Some(type_params) = type_params {
+        if let Some(values) = type_params.runtime_type_params.as_ref() {
+            for tp in values.iter().flatten() {
+                validate_typeparam(vm, tp)?;
+            }
+            return Ok(());
+        }
         for tp in &type_params.type_params {
             validate_typeparam(vm, tp)?;
         }
@@ -337,6 +528,11 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
             if op.values.len() < 2 {
                 return Err(vm.new_value_error("BoolOp with less than 2 values"));
             }
+            validate_runtime_expr_list_slots(
+                vm,
+                op.runtime_values.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             validate_exprs(vm, &op.values, ast::ExprContext::Load, false)
         }
         ast::Expr::Named(named) => {
@@ -362,6 +558,11 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
             validate_expr(vm, &ifexp.orelse, ast::ExprContext::Load)
         }
         ast::Expr::Dict(dict) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                dict.runtime_values.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             for item in &dict.items {
                 if let Some(key) = &item.key {
                     validate_expr(vm, key, ast::ExprContext::Load)?;
@@ -370,7 +571,14 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
             }
             Ok(())
         }
-        ast::Expr::Set(set) => validate_exprs(vm, &set.elts, ast::ExprContext::Load, false),
+        ast::Expr::Set(set) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                set.runtime_elts.as_ref(),
+                ast::ExprContext::Load,
+            )?;
+            validate_exprs(vm, &set.elts, ast::ExprContext::Load, false)
+        }
         ast::Expr::ListComp(list) => {
             validate_comprehension(vm, &list.generators)?;
             validate_expr(vm, &list.elt, ast::ExprContext::Load)
@@ -409,34 +617,73 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
                     "Compare has a different number of comparators and operands",
                 ));
             }
+            validate_runtime_expr_list_slots(
+                vm,
+                compare.runtime_comparators.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             validate_exprs(vm, &compare.comparators, ast::ExprContext::Load, false)?;
             validate_expr(vm, &compare.left, ast::ExprContext::Load)
         }
         ast::Expr::Call(call) => {
             validate_expr(vm, &call.func, ast::ExprContext::Load)?;
+            validate_runtime_expr_list_slots(
+                vm,
+                call.arguments.runtime_args.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             validate_exprs(vm, &call.arguments.args, ast::ExprContext::Load, false)?;
             validate_keywords(vm, &call.arguments.keywords)
         }
-        ast::Expr::FString(fstring) => validate_interpolated_elements(
-            vm,
-            fstring
-                .value
-                .elements()
-                .map(ast::InterpolatedStringElementRef::from),
-        ),
-        ast::Expr::TString(tstring) => validate_interpolated_elements(
-            vm,
-            tstring
-                .value
-                .elements()
-                .map(ast::InterpolatedStringElementRef::from),
-        ),
+        ast::Expr::FString(fstring) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                fstring.runtime_values.as_ref(),
+                ast::ExprContext::Load,
+            )?;
+            if let Some(joined_str) = fstring.runtime_joined_str.as_ref() {
+                validate_exprs(vm, joined_str, ast::ExprContext::Load, false)
+            } else {
+                validate_interpolated_elements(
+                    vm,
+                    fstring
+                        .value
+                        .elements()
+                        .map(ast::InterpolatedStringElementRef::from),
+                )
+            }
+        }
+        ast::Expr::TString(tstring) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                tstring.runtime_values.as_ref(),
+                ast::ExprContext::Load,
+            )?;
+            if let Some(template_str) = tstring.runtime_template_str.as_ref() {
+                validate_exprs(vm, template_str, ast::ExprContext::Load, false)
+            } else {
+                validate_interpolated_elements(
+                    vm,
+                    tstring
+                        .value
+                        .elements()
+                        .map(ast::InterpolatedStringElementRef::from),
+                )
+            }
+        }
         ast::Expr::StringLiteral(_)
         | ast::Expr::BytesLiteral(_)
         | ast::Expr::NumberLiteral(_)
+        | ast::Expr::Constant(_)
         | ast::Expr::BooleanLiteral(_)
         | ast::Expr::NoneLiteral(_)
-        | ast::Expr::EllipsisLiteral(_) => Ok(()),
+        | ast::Expr::EllipsisLiteral(_) => {
+            if let Some(invalid_type) = super::constant::invalid_constant_type(expr) {
+                Err(vm.new_type_error(format!("got an invalid type in Constant: {invalid_type}")))
+            } else {
+                Ok(())
+            }
+        }
         ast::Expr::Attribute(attr) => validate_expr(vm, &attr.value, ast::ExprContext::Load),
         ast::Expr::Subscript(sub) => {
             validate_expr(vm, &sub.slice, ast::ExprContext::Load)?;
@@ -444,8 +691,14 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
         }
         ast::Expr::Starred(star) => validate_expr(vm, &star.value, ctx),
         ast::Expr::Name(_) => Ok(()),
-        ast::Expr::List(list) => validate_exprs(vm, &list.elts, ctx, false),
-        ast::Expr::Tuple(tuple) => validate_exprs(vm, &tuple.elts, ctx, false),
+        ast::Expr::List(list) => {
+            validate_runtime_expr_list_slots(vm, list.runtime_elts.as_ref(), ctx)?;
+            validate_exprs(vm, &list.elts, ctx, false)
+        }
+        ast::Expr::Tuple(tuple) => {
+            validate_runtime_expr_list_slots(vm, tuple.runtime_elts.as_ref(), ctx)?;
+            validate_exprs(vm, &tuple.elts, ctx, false)
+        }
         ast::Expr::Slice(slice) => {
             if let Some(lower) = &slice.lower {
                 validate_expr(vm, lower, ast::ExprContext::Load)?;
@@ -458,7 +711,7 @@ fn validate_expr(vm: &VirtualMachine, expr: &ast::Expr, ctx: ast::ExprContext) -
             }
             Ok(())
         }
-        ast::Expr::IpyEscapeCommand(_) => Ok(()),
+        ast::Expr::IpyEscapeCommand(_) => Err(invalid_syntax_error(vm)),
     }
 }
 
@@ -477,9 +730,14 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             } else {
                 "FunctionDef"
             };
-            validate_body(vm, &func.body, owner)?;
+            validate_body(vm, &func.body, func.runtime_body.as_ref(), owner)?;
             validate_type_params(vm, func.type_params.as_deref())?;
             validate_parameters(vm, &func.parameters)?;
+            validate_runtime_expr_list_slots(
+                vm,
+                func.runtime_decorator_list.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             validate_decorators(vm, &func.decorator_list)?;
             if let Some(returns) = &func.returns {
                 validate_expr(vm, returns, ast::ExprContext::Load)?;
@@ -487,12 +745,27 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             Ok(())
         }
         ast::Stmt::ClassDef(class_def) => {
-            validate_body(vm, &class_def.body, "ClassDef")?;
+            validate_body(
+                vm,
+                &class_def.body,
+                class_def.runtime_body.as_ref(),
+                "ClassDef",
+            )?;
             validate_type_params(vm, class_def.type_params.as_deref())?;
             if let Some(arguments) = &class_def.arguments {
+                validate_runtime_expr_list_slots(
+                    vm,
+                    arguments.runtime_bases.as_ref(),
+                    ast::ExprContext::Load,
+                )?;
                 validate_exprs(vm, &arguments.args, ast::ExprContext::Load, false)?;
                 validate_keywords(vm, &arguments.keywords)?;
             }
+            validate_runtime_expr_list_slots(
+                vm,
+                class_def.runtime_decorator_list.as_ref(),
+                ast::ExprContext::Load,
+            )?;
             validate_decorators(vm, &class_def.decorator_list)
         }
         ast::Stmt::Return(ret) => {
@@ -501,8 +774,20 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             }
             Ok(())
         }
-        ast::Stmt::Delete(del) => validate_assignlist(vm, &del.targets, ast::ExprContext::Del),
+        ast::Stmt::Delete(del) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                del.runtime_targets.as_ref(),
+                ast::ExprContext::Del,
+            )?;
+            validate_assignlist(vm, &del.targets, ast::ExprContext::Del)
+        }
         ast::Stmt::Assign(assign) => {
+            validate_runtime_expr_list_slots(
+                vm,
+                assign.runtime_targets.as_ref(),
+                ast::ExprContext::Store,
+            )?;
             validate_assignlist(vm, &assign.targets, ast::ExprContext::Store)?;
             validate_expr(vm, &assign.value, ast::ExprContext::Load)
         }
@@ -532,22 +817,30 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             let owner = if for_stmt.is_async { "AsyncFor" } else { "For" };
             validate_expr(vm, &for_stmt.target, ast::ExprContext::Store)?;
             validate_expr(vm, &for_stmt.iter, ast::ExprContext::Load)?;
-            validate_body(vm, &for_stmt.body, owner)?;
+            validate_body(vm, &for_stmt.body, for_stmt.runtime_body.as_ref(), owner)?;
+            validate_runtime_stmt_list_slots(vm, for_stmt.runtime_orelse.as_ref())?;
             validate_stmts(vm, &for_stmt.orelse)
         }
         ast::Stmt::While(while_stmt) => {
             validate_expr(vm, &while_stmt.test, ast::ExprContext::Load)?;
-            validate_body(vm, &while_stmt.body, "While")?;
+            validate_body(
+                vm,
+                &while_stmt.body,
+                while_stmt.runtime_body.as_ref(),
+                "While",
+            )?;
+            validate_runtime_stmt_list_slots(vm, while_stmt.runtime_orelse.as_ref())?;
             validate_stmts(vm, &while_stmt.orelse)
         }
         ast::Stmt::If(if_stmt) => {
             validate_expr(vm, &if_stmt.test, ast::ExprContext::Load)?;
-            validate_body(vm, &if_stmt.body, "If")?;
+            validate_body(vm, &if_stmt.body, if_stmt.runtime_body.as_ref(), "If")?;
             for clause in &if_stmt.elif_else_clauses {
                 if let Some(test) = &clause.test {
                     validate_expr(vm, test, ast::ExprContext::Load)?;
                 }
-                validate_body(vm, &clause.body, "If")?;
+                validate_body(vm, &clause.body, clause.runtime_body.as_ref(), "If")?;
+                validate_runtime_stmt_list_slots(vm, clause.runtime_orelse.as_ref())?;
             }
             Ok(())
         }
@@ -564,7 +857,7 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
                     validate_expr(vm, optional_vars, ast::ExprContext::Store)?;
                 }
             }
-            validate_body(vm, &with_stmt.body, owner)
+            validate_body(vm, &with_stmt.body, with_stmt.runtime_body.as_ref(), owner)
         }
         ast::Stmt::Match(match_stmt) => {
             validate_expr(vm, &match_stmt.subject, ast::ExprContext::Load)?;
@@ -574,7 +867,7 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
                 if let Some(guard) = &case.guard {
                     validate_expr(vm, guard, ast::ExprContext::Load)?;
                 }
-                validate_body(vm, &case.body, "match_case")?;
+                validate_body(vm, &case.body, case.runtime_body.as_ref(), "match_case")?;
             }
             Ok(())
         }
@@ -591,7 +884,7 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
         }
         ast::Stmt::Try(try_stmt) => {
             let owner = if try_stmt.is_star { "TryStar" } else { "Try" };
-            validate_body(vm, &try_stmt.body, owner)?;
+            validate_body(vm, &try_stmt.body, try_stmt.runtime_body.as_ref(), owner)?;
             if try_stmt.handlers.is_empty() && try_stmt.finalbody.is_empty() {
                 return Err(vm.new_value_error(format!(
                     "{owner} has neither except handlers nor finalbody"
@@ -602,14 +895,22 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
                     vm.new_value_error(format!("{owner} has orelse but no except handlers"))
                 );
             }
+            validate_runtime_except_handler_list_slots(vm, try_stmt.runtime_handlers.as_ref())?;
             for handler in &try_stmt.handlers {
                 let ast::ExceptHandler::ExceptHandler(handler) = handler;
                 if let Some(type_expr) = &handler.type_ {
                     validate_expr(vm, type_expr, ast::ExprContext::Load)?;
                 }
-                validate_body(vm, &handler.body, "ExceptHandler")?;
+                validate_body(
+                    vm,
+                    &handler.body,
+                    handler.runtime_body.as_ref(),
+                    "ExceptHandler",
+                )?;
             }
+            validate_runtime_stmt_list_slots(vm, try_stmt.runtime_finalbody.as_ref())?;
             validate_stmts(vm, &try_stmt.finalbody)?;
+            validate_runtime_stmt_list_slots(vm, try_stmt.runtime_orelse.as_ref())?;
             validate_stmts(vm, &try_stmt.orelse)
         }
         ast::Stmt::Assert(assert_stmt) => {
@@ -624,6 +925,11 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             Ok(())
         }
         ast::Stmt::ImportFrom(import) => {
+            if let Some(level) = import.runtime_level
+                && level < 0
+            {
+                return Err(vm.new_value_error("Negative ImportFrom level"));
+            }
             validate_nonempty_seq(vm, import.names.len(), "names", "ImportFrom")?;
             Ok(())
         }
@@ -636,10 +942,8 @@ fn validate_stmt(vm: &VirtualMachine, stmt: &ast::Stmt) -> PyResult<()> {
             Ok(())
         }
         ast::Stmt::Expr(expr) => validate_expr(vm, &expr.value, ast::ExprContext::Load),
-        ast::Stmt::Pass(_)
-        | ast::Stmt::Break(_)
-        | ast::Stmt::Continue(_)
-        | ast::Stmt::IpyEscapeCommand(_) => Ok(()),
+        ast::Stmt::Pass(_) | ast::Stmt::Break(_) | ast::Stmt::Continue(_) => Ok(()),
+        ast::Stmt::IpyEscapeCommand(_) => Err(invalid_syntax_error(vm)),
     }
 }
 
@@ -652,10 +956,17 @@ fn validate_stmts(vm: &VirtualMachine, stmts: &[ast::Stmt]) -> PyResult<()> {
 
 pub(super) fn validate_mod(vm: &VirtualMachine, module: &Mod) -> PyResult<()> {
     match module {
-        Mod::Module(module) => validate_stmts(vm, &module.body),
-        Mod::Interactive(module) => validate_stmts(vm, &module.body),
+        Mod::Module(module) => {
+            validate_runtime_stmt_list_slots(vm, module.module.runtime_body.as_ref())?;
+            validate_stmts(vm, &module.module.body)
+        }
+        Mod::Interactive(module) => {
+            validate_runtime_stmt_list_slots(vm, module.runtime_body.as_ref())?;
+            validate_stmts(vm, &module.body)
+        }
         Mod::Expression(expr) => validate_expr(vm, &expr.body, ast::ExprContext::Load),
         Mod::FunctionType(func_type) => {
+            validate_runtime_expr_option_list_slots(vm, func_type.runtime_argtypes.as_ref())?;
             validate_exprs(vm, &func_type.argtypes, ast::ExprContext::Load, false)?;
             validate_expr(vm, &func_type.returns, ast::ExprContext::Load)
         }

crates/vm/src/stdlib/_symtable.rs

@@ -153,7 +153,9 @@ mod _symtable {
                 CompilerScope::Class => TYPE_CLASS,
                 CompilerScope::Module => TYPE_MODULE,
                 CompilerScope::Annotation => TYPE_ANNOTATION,
+                CompilerScope::TypeAlias => TYPE_TYPE_ALIAS,
                 CompilerScope::TypeParams => TYPE_TYPE_PARAMETERS,
+                CompilerScope::TypeVariable => TYPE_TYPE_VARIABLE,
             }
         }
 

crates/vm/src/stdlib/builtins.rs

@@ -13,11 +13,12 @@ mod builtins {
             PyByteArray, PyBytes, PyDictRef, PyStr, PyStrRef, PyTuple, PyTupleRef, PyType,
             PyUtf8StrRef,
             enumerate::PyReverseSequenceIterator,
-            function::{PyCellRef, PyFunction},
+            function::{PyCell, PyCellRef, PyFunction},
             int::PyIntRef,
             iter::PyCallableIterator,
             list::{PyList, SortOptions},
         },
+        bytecode,
         common::hash::PyHash,
         function::{
             ArgBytesLike, ArgCallable, ArgIndex, ArgIntoBool, ArgIterable, ArgMapping,
@@ -29,9 +30,13 @@ mod builtins {
         readline::{Readline, ReadlineResult},
         stdlib::sys,
         types::PyComparisonOp,
+        vm::compile_mode::{
+            CompilerFlags, PY_EVAL_INPUT, PY_FILE_INPUT, PY_FUNC_TYPE_INPUT, PY_SINGLE_INPUT,
+            compile_future_feature_mask, compile_future_features_from_flags,
+        },
     };
     use itertools::Itertools;
-    use num_traits::{Signed, ToPrimitive, Zero};
+    use num_traits::{Signed, ToPrimitive};
     use rustpython_common::wtf8::CodePoint;
 
     #[cfg(not(feature = "rustpython-compiler"))]
@@ -103,8 +108,8 @@ mod builtins {
         filename: PyObjectRef,
         mode: PyUtf8StrRef,
         // CPython parity: flags / optimize accept any object with __index__,
-        // not just exact int. Matches the behavior of `int(x)` arg conversion
-        // used by Python/Python-ast.c::compile.
+        // not just exact int. Matches the argument conversion used by
+        // builtin_compile_impl.
         #[pyarg(any, optional)]
         flags: OptionalArg<ArgPrimitiveIndex<i32>>,
         // CPython parity: dont_inherit goes through PyObject_IsTrue, so
@@ -114,174 +119,67 @@ mod builtins {
         dont_inherit: OptionalArg<ArgIntoBool>,
         #[pyarg(any, optional)]
         optimize: OptionalArg<ArgPrimitiveIndex<i32>>,
-        #[pyarg(any, optional)]
+        #[pyarg(named, optional)]
         _feature_version: OptionalArg<i32>,
     }
 
-    /// Detect PEP 263 encoding cookie from source bytes.
-    /// Checks first two lines for `# coding[:=] <encoding>` pattern.
-    /// Returns the encoding name if found, or None for default (UTF-8).
-    #[cfg(feature = "parser")]
-    fn detect_source_encoding(source: &[u8]) -> Option<String> {
-        fn find_encoding_in_line(line: &[u8]) -> Option<String> {
-            // PEP 263: '#' must be preceded only by whitespace/formfeed
-            let hash_pos = line.iter().position(|&b| b == b'#')?;
-            if !line[..hash_pos]
-                .iter()
-                .all(|&b| matches!(b, b' ' | b'\t' | b'\x0c' | b'\r'))
-            {
-                return None;
-            }
-            let after_hash = &line[hash_pos..];
-
-            // Find "coding" after the #
-            let coding_pos = after_hash.windows(6).position(|w| w == b"coding")?;
-            let after_coding = &after_hash[coding_pos + 6..];
-
-            // Next char must be ':' or '='
-            let rest = if matches!(after_coding.first(), Some(b':' | b'=')) {
-                &after_coding[1..]
-            } else {
-                return None;
-            };
-
-            // Skip whitespace
-            let rest = rest
-                .iter()
-                .copied()
-                .skip_while(|&b| matches!(b, b' ' | b'\t'))
-                .collect::<Vec<_>>();
-
-            // Read encoding name: [-\w.]+
-            let name = rest
-                .iter()
-                .take_while(|&&b| b.is_ascii_alphanumeric() || matches!(b, b'-' | b'_' | b'.'))
-                .map(|&b| b as char)
-                .collect::<String>();
-
-            if name.is_empty() {
-                None
-            } else {
-                Some(normalize_source_encoding(&name))
-            }
+    fn merge_compile_future_features(
+        flags: i32,
+        dont_inherit: bool,
+        vm: &VirtualMachine,
+    ) -> bytecode::CodeFlags {
+        let mut future_features = compile_future_features_from_flags(flags);
+        if !dont_inherit && let Some(frame) = vm.current_frame() {
+            future_features |= bytecode::CodeFlags::from_bits_truncate(
+                frame.code.flags.bits() & compile_future_feature_mask().bits(),
+            );
         }
-
-        // Split into lines (first two only)
-        let mut lines = source.splitn(3, |&b| b == b'\n');
-
-        if let Some(first) = lines.next() {
-            // Strip BOM if present
-            let first = first.strip_prefix(b"\xef\xbb\xbf").unwrap_or(first);
-            if let Some(enc) = find_encoding_in_line(first) {
-                return Some(enc);
-            }
-            // Only check second line if first line is blank or a comment
-            let trimmed = first
-                .iter()
-                .find(|&&b| !matches!(b, b' ' | b'\t' | b'\x0c' | b'\r'))
-                .copied();
-
-            if trimmed.is_some_and(|b| b != b'#') {
-                return None;
-            }
-        }
-
-        lines.next().and_then(find_encoding_in_line)
+        future_features
     }
 
-    /// Match CPython's Parser/tokenizer/helpers.c:get_normal_name().
-    #[cfg(feature = "parser")]
-    fn normalize_source_encoding(name: &str) -> String {
-        let mut normalized = String::with_capacity(name.len().min(12));
-        for ch in name.chars().take(12) {
-            if ch == '_' {
-                normalized.push('-');
-            } else {
-                normalized.push(ch.to_ascii_lowercase());
-            }
-        }
+    fn audit_compile_source(vm: &VirtualMachine, source: &[u8], filename: &str) -> PyResult<()> {
+        vm.sys_module.get_attr("audit", vm)?.call(
+            (
+                vm.ctx.new_str("compile"),
+                vm.ctx.new_bytes(source.to_vec()),
+                vm.ctx.new_str(filename),
+            ),
+            vm,
+        )?;
+        Ok(())
+    }
 
-        if normalized == "utf-8" || normalized.starts_with("utf-8-") {
-            "utf-8".to_owned()
-        } else if normalized == "latin-1"
-            || normalized == "iso-8859-1"
-            || normalized == "iso-latin-1"
-            || normalized.starts_with("latin-1-")
-            || normalized.starts_with("iso-8859-1-")
-            || normalized.starts_with("iso-latin-1-")
+    fn trim_eval_source_bytes(mut source: &[u8]) -> &[u8] {
+        while let Some((&first, rest)) = source.split_first()
+            && matches!(first, b' ' | b'\t')
         {
-            "iso-8859-1".to_owned()
-        } else {
-            name.to_owned()
+            source = rest;
         }
+        source
     }
 
-    /// Decode source bytes to a string, handling PEP 263 encoding declarations
-    /// and BOM. Raises SyntaxError for invalid UTF-8 without an encoding
-    /// declaration.
-    #[cfg(feature = "parser")]
-    fn is_utf8_encoding(name: &str) -> bool {
-        name == "utf-8"
-    }
-
-    #[cfg(feature = "parser")]
-    fn decode_source_bytes(source: &[u8], filename: &str, vm: &VirtualMachine) -> PyResult<String> {
-        let has_bom = source.starts_with(b"\xef\xbb\xbf");
-        let encoding = detect_source_encoding(source);
-
-        let is_utf8 = encoding.as_deref().is_none_or(is_utf8_encoding);
-
-        // Validate BOM + encoding combination
-        if has_bom && !is_utf8 {
-            let enc = encoding.as_deref().unwrap_or("utf-8");
-            return Err(vm.new_exception_msg(
-                vm.ctx.exceptions.syntax_error.to_owned(),
-                format!("encoding problem: {enc} with BOM").into(),
-            ));
+    fn decode_eval_exec_source_bytes(
+        vm: &VirtualMachine,
+        source: &[u8],
+        filename: &str,
+    ) -> PyResult<String> {
+        #[cfg(feature = "parser")]
+        {
+            vm.decode_source_bytes(source, filename, false)
         }
-
-        if is_utf8 {
-            let src = if has_bom { &source[3..] } else { source };
-            match core::str::from_utf8(src) {
-                Ok(s) => Ok(s.to_owned()),
-                Err(e) => {
-                    let bad_byte = src[e.valid_up_to()];
-                    let line = src[..e.valid_up_to()]
-                        .iter()
-                        .filter(|&&b| b == b'\n')
-                        .count()
-                        + 1;
-                    Err(vm.new_exception_msg(
-                        vm.ctx.exceptions.syntax_error.to_owned(),
-                        format!(
-                            "Non-UTF-8 code starting with '\\x{bad_byte:02x}' \
-                             on line {line}, but no encoding declared; \
-                             see https://peps.python.org/pep-0263/ for details \
-                             ({filename}, line {line})"
-                        )
-                        .into(),
-                    ))
-                }
-            }
-        } else {
-            // Use codec registry for non-UTF-8 encodings
-            let enc = encoding.as_deref().unwrap();
-            let bytes_obj = vm.ctx.new_bytes(source.to_vec());
-            let decoded = vm
-                .state
-                .codec_registry
-                .decode_text(bytes_obj.into(), enc, None, vm)
-                .map_err(|exc| {
-                    if exc.fast_isinstance(vm.ctx.exceptions.lookup_error) {
-                        vm.new_exception_msg(
-                            vm.ctx.exceptions.syntax_error.to_owned(),
-                            format!("unknown encoding for '{filename}': {enc}").into(),
-                        )
-                    } else {
-                        exc
-                    }
-                })?;
-            Ok(decoded.to_string_lossy().into_owned())
+        #[cfg(not(feature = "parser"))]
+        {
+            _ = filename;
+            core::str::from_utf8(source)
+                .map(str::to_owned)
+                .map_err(|err| {
+                    let msg = format!(
+                        "(unicode error) 'utf-8' codec can't decode byte 0x{:x?} in position {}: invalid start byte",
+                        source[err.valid_up_to()],
+                        err.valid_up_to()
+                    );
+                    vm.new_exception_msg(vm.ctx.exceptions.syntax_error.to_owned(), msg.into())
+                })
         }
     }
 
@@ -303,30 +201,60 @@ mod builtins {
 
             use crate::{class::PyClassImpl, stdlib::_ast};
 
-            let feature_version = feature_version_from_arg(args._feature_version, vm)?;
+            let feature_version = args._feature_version.into_option().unwrap_or(-1);
 
             let mode_str = args.mode.as_str();
+            let flags: i32 = args.flags.map_or(0, |v| v.value);
+            let cf = CompilerFlags::from_bits_retain(flags);
+
+            if (flags & !CompilerFlags::ALLOWED_FLAGS.bits()) != 0 {
+                return Err(vm.new_value_error("compile(): unrecognised flags"));
+            }
 
             let optimize: i32 = args.optimize.map_or(-1, |v| v.value);
             let optimize: u8 = match optimize {
-                -1 => vm.state.config.settings.optimize,
+                -1 => vm.state.config.settings.optimize.min(2),
                 0..=2 => optimize as u8,
                 _ => return Err(vm.new_value_error("compile(): invalid optimize value")),
             };
+            let dont_inherit = args.dont_inherit.map_or(false, ArgIntoBool::into_bool);
+            let is_ast_only = cf.contains(CompilerFlags::ONLY_AST);
+            let future_features = merge_compile_future_features(flags, dont_inherit, vm);
 
-            if args
-                .source
-                .fast_isinstance(&_ast::NodeAst::make_static_type())
-            {
-                let flags: i32 = args.flags.map_or(0, |v| v.value);
-                let is_ast_only = !(flags & _ast::PY_CF_ONLY_AST).is_zero();
-
-                // func_type mode requires PyCF_ONLY_AST
-                if mode_str == "func_type" && !is_ast_only {
+            let start = if mode_str == "exec" {
+                PY_FILE_INPUT
+            } else if mode_str == "eval" {
+                PY_EVAL_INPUT
+            } else if mode_str == "single" {
+                PY_SINGLE_INPUT
+            } else if mode_str == "func_type" {
+                if !is_ast_only {
                     return Err(vm.new_value_error(
                         "compile() mode 'func_type' requires flag PyCF_ONLY_AST",
                     ));
                 }
+                PY_FUNC_TYPE_INPUT
+            } else {
+                let msg = if is_ast_only {
+                    "compile() mode must be 'exec', 'eval', 'single' or 'func_type'"
+                } else {
+                    "compile() mode must be 'exec', 'eval' or 'single'"
+                };
+                return Err(vm.new_value_error(msg));
+            };
+
+            let ast_type = _ast::NodeAst::make_static_type().as_object().to_owned();
+            if args.source.is_instance(&ast_type, vm)? {
+                let explicit_future_annotations =
+                    future_features.contains(bytecode::CodeFlags::FUTURE_ANNOTATIONS);
+                vm.sys_module.get_attr("audit", vm)?.call(
+                    (
+                        vm.ctx.new_str("compile"),
+                        args.source.clone(),
+                        vm.ctx.none(),
+                    ),
+                    vm,
+                )?;
 
                 // compile(ast_node, ..., PyCF_ONLY_AST) returns the AST after validation
                 if is_ast_only {
@@ -336,15 +264,29 @@ mod builtins {
                                 "compile() mode must be 'exec', 'eval', 'single' or 'func_type'",
                             )
                         })?;
-                    if !args.source.fast_isinstance(&expected_type) {
+                    if !args.source.is_instance(expected_type.as_object(), vm)? {
                         return Err(vm.new_type_error(format!(
                             "expected {} node, got {}",
                             expected_name,
                             args.source.class().name()
                         )));
                     }
-                    _ast::validate_ast_object(vm, args.source.clone())?;
-                    return Ok(args.source);
+                    #[cfg(not(feature = "rustpython-codegen"))]
+                    {
+                        _ast::validate_ast_object(vm, args.source.clone())?;
+                        return Ok(args.source);
+                    }
+                    #[cfg(feature = "rustpython-codegen")]
+                    {
+                        return _ast::preprocess_ast_object(
+                            vm,
+                            args.source,
+                            &filename.to_string_lossy(),
+                            optimize,
+                            cf.contains(CompilerFlags::OPTIMIZED_AST),
+                            explicit_future_annotations,
+                        );
+                    }
                 }
 
                 #[cfg(not(feature = "rustpython-codegen"))]
@@ -353,133 +295,104 @@ mod builtins {
                 }
                 #[cfg(feature = "rustpython-codegen")]
                 {
+                    let (expected_type, expected_name) = _ast::mode_type_and_name(mode_str)
+                        .ok_or_else(|| {
+                            vm.new_value_error("compile() mode must be 'exec', 'eval' or 'single'")
+                        })?;
+                    if !args.source.is_instance(expected_type.as_object(), vm)? {
+                        return Err(vm.new_type_error(format!(
+                            "expected {} node, got {}",
+                            expected_name,
+                            args.source.class().name()
+                        )));
+                    }
                     let mode = mode_str
                         .parse::<crate::compiler::Mode>()
                         .map_err(|err| vm.new_value_error(err.to_string()))?;
-                    return _ast::compile(
-                        vm,
-                        args.source,
-                        &filename.to_string_lossy(),
-                        mode,
-                        Some(optimize),
-                    );
+                    let mut opts = vm.compile_opts();
+                    opts.optimize = optimize;
+                    opts.allow_top_level_await = cf.contains(CompilerFlags::ALLOW_TOP_LEVEL_AWAIT);
+                    opts.future_features = future_features;
+                    return _ast::compile(vm, args.source, &filename.to_string_lossy(), mode, opts);
                 }
             }
 
             #[cfg(not(feature = "parser"))]
-            return Err(vm.new_type_error(
-                "can't compile() source code when the `parser` feature of rustpython is disabled",
-            ));
-
+            {
+                Err(vm.new_type_error(
+                    "can't compile() source code when the `parser` feature of rustpython is disabled",
+                ))
+            }
             #[cfg(feature = "parser")]
             {
-                use crate::convert::ToPyException;
-
-                use ruff_python_parser as parser;
-
                 let source = ArgStrOrBytesLike::try_from_object(vm, args.source)?;
-                let source = source.borrow_bytes();
 
-                let source = decode_source_bytes(&source, &filename.to_string_lossy(), vm)?;
-                let source = source.as_str();
-
-                let flags: i32 = args.flags.map_or(0, |v| v.value);
-
-                if !(flags & !_ast::PY_COMPILE_FLAGS_MASK).is_zero() {
-                    return Err(vm.new_value_error("compile(): unrecognised flags"));
-                }
-
-                let allow_incomplete = !(flags & _ast::PY_CF_ALLOW_INCOMPLETE_INPUT).is_zero();
-                let type_comments = !(flags & _ast::PY_CF_TYPE_COMMENTS).is_zero();
-
-                let optimize_level = optimize;
-
-                if (flags & _ast::PY_CF_ONLY_AST).is_zero() {
-                    #[cfg(not(feature = "compiler"))]
-                    {
-                        Err(vm.new_value_error(CODEGEN_NOT_SUPPORTED))
-                    }
-                    #[cfg(feature = "compiler")]
-                    {
-                        if let Some(feature_version) = feature_version {
-                            let mode = mode_str
-                                .parse::<parser::Mode>()
-                                .map_err(|err| vm.new_value_error(err.to_string()))?;
-                            let _ = _ast::parse(
-                                vm,
-                                source,
-                                mode,
-                                optimize_level,
-                                Some(feature_version),
-                                type_comments,
-                            )
-                            .map_err(|e| (e, Some(source), allow_incomplete).to_pyexception(vm))?;
-                        }
-
-                        let mode = mode_str
-                            .parse::<crate::compiler::Mode>()
-                            .map_err(|err| vm.new_value_error(err.to_string()))?;
-
-                        let mut opts = vm.compile_opts();
-                        opts.optimize = optimize;
-
-                        let code = vm
-                            .compile_with_opts(source, mode, &filename.to_string_lossy(), opts)
-                            .map_err(|err| {
-                                (err, Some(source), allow_incomplete).to_pyexception(vm)
-                            })?;
-                        Ok(code.into())
-                    }
-                } else {
-                    if mode_str == "func_type" {
-                        return _ast::parse_func_type(vm, source, optimize_level, feature_version)
-                            .map_err(|e| (e, Some(source), allow_incomplete).to_pyexception(vm));
-                    }
-
-                    let mode = mode_str
-                        .parse::<parser::Mode>()
-                        .map_err(|err| vm.new_value_error(err.to_string()))?;
-                    let parsed = _ast::parse(
-                        vm,
+                let mut compile_flags = flags | future_features.bits() as i32;
+                #[cfg(feature = "rustpython-compiler")]
+                let compile_source = |source: &[u8], compile_flags: i32| {
+                    vm.compile_string_object_with_flags(
                         source,
-                        mode,
-                        optimize_level,
+                        &filename.to_string_lossy(),
+                        start,
+                        compile_flags,
                         feature_version,
-                        type_comments,
+                        optimize as i32,
                     )
-                    .map_err(|e| (e, Some(source), allow_incomplete).to_pyexception(vm))?;
-
-                    if mode_str == "single" {
-                        return _ast::wrap_interactive(vm, parsed);
+                };
+                match &source {
+                    ArgStrOrBytesLike::Str(source) => {
+                        if source.as_bytes().contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        audit_compile_source(
+                            vm,
+                            source.as_bytes(),
+                            filename.to_string_lossy().as_ref(),
+                        )?;
+                        compile_flags |= CompilerFlags::IGNORE_COOKIE.bits();
+                        #[cfg(feature = "rustpython-compiler")]
+                        {
+                            compile_source(source.as_bytes(), compile_flags)
+                        }
+                        #[cfg(not(feature = "rustpython-compiler"))]
+                        {
+                            Err(vm.new_value_error(CODEGEN_NOT_SUPPORTED))
+                        }
+                    }
+                    ArgStrOrBytesLike::Buf(source) => {
+                        let source_bytes = source.borrow_buf();
+                        let source_bytes: &[u8] = &source_bytes;
+                        if source_bytes.contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        audit_compile_source(
+                            vm,
+                            source_bytes,
+                            filename.to_string_lossy().as_ref(),
+                        )?;
+                        #[cfg(feature = "rustpython-compiler")]
+                        {
+                            compile_source(source_bytes, compile_flags)
+                        }
+                        #[cfg(not(feature = "rustpython-compiler"))]
+                        {
+                            Err(vm.new_value_error(CODEGEN_NOT_SUPPORTED))
+                        }
                     }
-
-                    Ok(parsed)
                 }
             }
         }
     }
 
-    #[cfg(feature = "ast")]
-    fn feature_version_from_arg(
-        feature_version: OptionalArg<i32>,
-        vm: &VirtualMachine,
-    ) -> PyResult<Option<ruff_python_ast::PythonVersion>> {
-        let Some(minor) = feature_version.into_option() else {
-            return Ok(None);
-        };
-
-        if minor < 0 {
-            return Ok(None);
-        }
-
-        u8::try_from(minor)
-            .map(|v| Some(ruff_python_ast::PythonVersion { major: 3, minor: v }))
-            .map_err(|_| vm.new_value_error("compile() _feature_version out of range"))
-    }
-
     #[pyfunction]
     fn delattr(obj: PyObjectRef, attr: PyObjectRef, vm: &VirtualMachine) -> PyResult<()> {
-        let attr = attr.try_to_ref::<PyStr>(vm).map_err(|_| {
+        let attr = attr.try_to_ref::<PyStr>(vm).map_err(|_e| {
             vm.new_type_error(format!(
                 "attribute name must be string, not '{}'",
                 attr.class().name()
@@ -507,42 +420,39 @@ mod builtins {
     }
 
     impl ScopeArgs {
-        fn validate_globals_dict(
-            globals: &PyObject,
-            vm: &VirtualMachine,
-            func_name: &'static str,
-        ) -> PyResult<()> {
-            if globals.fast_isinstance(vm.ctx.types.dict_type) {
-                return Ok(());
-            }
-
-            let msg = match func_name {
-                "eval" => {
-                    let is_mapping = globals.mapping_unchecked().check();
-                    if is_mapping {
-                        "globals must be a real dict; try eval(expr, {}, mapping)".into()
-                    } else {
-                        "globals must be a dict".into()
-                    }
-                }
-                "exec" => format!(
-                    "exec() globals must be a dict, not {}",
-                    globals.class().name()
-                ),
-                _ => "globals must be a dict".into(),
-            };
-
-            Err(vm.new_type_error(msg))
-        }
-
         fn make_scope(
             self,
             vm: &VirtualMachine,
             func_name: &'static str,
         ) -> PyResult<crate::scope::Scope> {
+            fn validate_globals_dict(
+                globals: &PyObject,
+                vm: &VirtualMachine,
+                func_name: &'static str,
+            ) -> PyResult<()> {
+                if !globals.fast_isinstance(vm.ctx.types.dict_type) {
+                    return Err(match func_name {
+                        "eval" => {
+                            let is_mapping = globals.mapping_unchecked().check();
+                            vm.new_type_error(if is_mapping {
+                                "globals must be a real dict; try eval(expr, {}, mapping)"
+                            } else {
+                                "globals must be a dict"
+                            })
+                        }
+                        "exec" => vm.new_type_error(format!(
+                            "exec() globals must be a dict, not {}",
+                            globals.class().name()
+                        )),
+                        _ => vm.new_type_error("globals must be a dict"),
+                    });
+                }
+                Ok(())
+            }
+
             let (globals, locals) = match self.globals {
                 Some(globals) => {
-                    Self::validate_globals_dict(&globals, vm, func_name)?;
+                    validate_globals_dict(&globals, vm, func_name)?;
 
                     let globals = PyDictRef::try_from_object(vm, globals)?;
                     if !globals.contains_key(identifier!(vm, __builtins__), vm) {
@@ -570,6 +480,61 @@ mod builtins {
         }
     }
 
+    #[derive(FromArgs)]
+    struct ExecArgs {
+        #[pyarg(positional)]
+        source: Either<ArgStrOrBytesLike, PyRef<crate::builtins::PyCode>>,
+        #[pyarg(any, default)]
+        globals: Option<PyObjectRef>,
+        #[pyarg(any, default)]
+        locals: Option<ArgMapping>,
+        #[pyarg(named, optional)]
+        closure: OptionalOption<PyObjectRef>,
+    }
+
+    fn exec_closure(
+        code_obj: &PyRef<crate::builtins::PyCode>,
+        closure: Option<PyObjectRef>,
+        vm: &VirtualMachine,
+    ) -> PyResult<Option<PyRef<PyTuple<PyCellRef>>>> {
+        let num_free = code_obj.freevars.len();
+        let Some(closure) = closure else {
+            if num_free == 0 {
+                return Ok(None);
+            }
+            return Err(vm.new_type_error(format!(
+                "code object requires a closure of exactly length {num_free}"
+            )));
+        };
+
+        if num_free == 0 {
+            return Err(vm.new_type_error("cannot use a closure with this code object"));
+        }
+
+        let closure_tuple = closure
+            .downcast_exact::<PyTuple>(vm)
+            .map_err(|_| {
+                vm.new_type_error(format!(
+                    "code object requires a closure of exactly length {num_free}"
+                ))
+            })?
+            .into_pyref();
+        if closure_tuple.len() != num_free {
+            return Err(vm.new_type_error(format!(
+                "code object requires a closure of exactly length {num_free}"
+            )));
+        }
+
+        closure_tuple
+            .try_into_typed::<PyCell>(vm)
+            .map(Some)
+            .map_err(|_| {
+                vm.new_type_error(format!(
+                    "code object requires a closure of exactly length {num_free}"
+                ))
+            })
+    }
+
     #[pyfunction]
     fn eval(
         source: Either<ArgStrOrBytesLike, PyRef<crate::builtins::PyCode>>,
@@ -581,38 +546,93 @@ mod builtins {
         // source as string
         let code = match source {
             Either::A(either) => {
-                let source: &[u8] = &either.borrow_bytes();
-                if source.contains(&0) {
-                    return Err(vm.new_exception_msg(
-                        vm.ctx.exceptions.syntax_error.to_owned(),
-                        "source code string cannot contain null bytes".into(),
-                    ));
-                }
-
-                let source = core::str::from_utf8(source).map_err(|err| {
-                    let msg = format!(
-                        "(unicode error) 'utf-8' codec can't decode byte 0x{:x?} in position {}: invalid start byte",
-                        source[err.valid_up_to()],
-                        err.valid_up_to()
-                    );
-
-                    vm.new_exception_msg(vm.ctx.exceptions.syntax_error.to_owned(), msg.into())
-                })?;
-                Ok(Either::A(vm.ctx.new_utf8_str(source.trim_start())))
+                let source = match &either {
+                    ArgStrOrBytesLike::Str(source) => {
+                        if source.as_bytes().contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        let source = source.expect_str().trim_start_matches([' ', '\t']);
+                        audit_compile_source(vm, source.as_bytes(), "<string>")?;
+                        source.to_owned()
+                    }
+                    ArgStrOrBytesLike::Buf(source) => {
+                        let source: &[u8] = &source.borrow_buf();
+                        if source.contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        let source = trim_eval_source_bytes(source);
+                        audit_compile_source(vm, source, "<string>")?;
+                        decode_eval_exec_source_bytes(vm, source, "eval")?
+                    }
+                };
+                Ok(Either::A(vm.ctx.new_utf8_str(source)))
             }
             Either::B(code) => Ok(Either::B(code)),
         }?;
-        run_code(vm, code, scope, crate::compiler::Mode::Eval, "eval")
+        run_code(vm, code, scope, crate::compiler::Mode::Eval, "eval", None)
     }
 
     #[pyfunction]
-    fn exec(
-        source: Either<PyUtf8StrRef, PyRef<crate::builtins::PyCode>>,
-        scope: ScopeArgs,
-        vm: &VirtualMachine,
-    ) -> PyResult {
-        let scope = scope.make_scope(vm, "exec")?;
-        run_code(vm, source, scope, crate::compiler::Mode::Exec, "exec")
+    fn exec(args: ExecArgs, vm: &VirtualMachine) -> PyResult {
+        let ExecArgs {
+            source,
+            globals,
+            locals,
+            closure,
+        } = args;
+        let scope = ScopeArgs { globals, locals }.make_scope(vm, "exec")?;
+        let closure = closure.flatten();
+        let (source, closure) = match source {
+            Either::A(either) => {
+                if closure.is_some() {
+                    return Err(
+                        vm.new_type_error("closure can only be used when source is a code object")
+                    );
+                }
+                let source = match &either {
+                    ArgStrOrBytesLike::Str(source) => {
+                        if source.as_bytes().contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        audit_compile_source(vm, source.as_bytes(), "<string>")?;
+                        source.expect_str().to_owned()
+                    }
+                    ArgStrOrBytesLike::Buf(source) => {
+                        let source: &[u8] = &source.borrow_buf();
+                        if source.contains(&0) {
+                            return Err(vm.new_exception_msg(
+                                vm.ctx.exceptions.syntax_error.to_owned(),
+                                "source code string cannot contain null bytes".into(),
+                            ));
+                        }
+                        audit_compile_source(vm, source, "<string>")?;
+                        decode_eval_exec_source_bytes(vm, source, "exec")?
+                    }
+                };
+                (Either::A(vm.ctx.new_utf8_str(source)), None)
+            }
+            Either::B(code) => {
+                let closure = exec_closure(&code, closure, vm)?;
+                (Either::B(code), closure)
+            }
+        };
+        run_code(
+            vm,
+            source,
+            scope,
+            crate::compiler::Mode::Exec,
+            "exec",
+            closure,
+        )
     }
 
     fn run_code(
@@ -621,28 +641,39 @@ mod builtins {
         scope: crate::scope::Scope,
         #[allow(unused_variables)] mode: crate::compiler::Mode,
         func: &str,
+        closure: Option<PyRef<PyTuple<PyCellRef>>>,
     ) -> PyResult {
         // Determine code object:
         let code_obj = match source {
             #[cfg(feature = "rustpython-compiler")]
             Either::A(string) => {
                 let source = string.as_str();
-                vm.compile(source, mode, "<string>")
-                    .map_err(|err| vm.new_syntax_error(&err, Some(source)))?
+                let mut opts = vm.compile_opts();
+                if let Some(frame) = vm.current_frame() {
+                    opts.future_features = bytecode::CodeFlags::from_bits_truncate(
+                        frame.code.flags.bits() & compile_future_feature_mask().bits(),
+                    );
+                }
+                vm.compile_with_opts(source, mode, "<string>", opts)
+                    .map_err(|err| err.into_pyexception(vm, Some(source)))?
             }
             #[cfg(not(feature = "rustpython-compiler"))]
             Either::A(_) => return Err(vm.new_type_error(CODEGEN_NOT_SUPPORTED)),
             Either::B(code_obj) => code_obj,
         };
 
-        if !code_obj.freevars.is_empty() {
+        vm.sys_module
+            .get_attr("audit", vm)?
+            .call((vm.ctx.new_str("exec"), code_obj.clone()), vm)?;
+
+        if closure.is_none() && !code_obj.freevars.is_empty() {
             return Err(vm.new_type_error(format!(
                 "code object passed to {func}() may not contain free variables"
             )));
         }
 
         // Run the code:
-        vm.run_code_obj(code_obj, scope)
+        vm.run_code_obj_with_closure(code_obj, scope, closure)
     }
 
     #[pyfunction]
@@ -1002,8 +1033,8 @@ mod builtins {
             modulus,
         } = args;
         let modulus = modulus
-            .as_ref()
-            .map_or_else(|| vm.ctx.none.as_object(), |m| m);
+            .as_deref()
+            .unwrap_or_else(|| vm.ctx.none.as_object());
         vm._pow(&x, &y, modulus)
     }
 

crates/vm/src/stdlib/sys.rs

@@ -751,7 +751,7 @@ pub mod sys {
             .read_to_string(&mut source)
             .map_err(|e| vm.new_os_error(format!("Error reading from stdin: {e}")))?;
         vm.compile(&source, crate::compiler::Mode::Single, "<stdin>")
-            .map_err(|e| vm.new_os_error(format!("Error running stdin: {e}")))?;
+            .map_err(|e| e.into_pyexception(vm, Some(&source)))?;
         Ok(())
     }
 
@@ -1228,7 +1228,7 @@ pub mod sys {
             vm.state.int_max_str_digits.store(maxdigits);
             Ok(())
         } else {
-            let error = format!("maxdigits must be 0 or larger than {threshold:?}");
+            let error = format!("maxdigits must be 0 or larger than {threshold}");
             Err(vm.new_value_error(error))
         }
     }
@@ -1744,12 +1744,54 @@ pub mod sys {
         }
 
         for hook in hooks {
-            hook.call((event.clone(), args.clone()), vm)?;
+            call_audit_hook(&hook, event.clone().into(), args, vm)?;
         }
 
         Ok(())
     }
 
+    fn audit_hook_can_trace(hook: &PyObjectRef, vm: &VirtualMachine) -> PyResult<bool> {
+        match hook.get_attr("__cantrace__", vm) {
+            Ok(can_trace) => can_trace.try_to_bool(vm),
+            Err(exc)
+                if exc
+                    .class()
+                    .fast_issubclass(vm.ctx.exceptions.attribute_error) =>
+            {
+                Ok(false)
+            }
+            Err(exc) => Err(exc),
+        }
+    }
+
+    fn call_audit_hook(
+        hook: &PyObjectRef,
+        event: PyObjectRef,
+        args: &PyObjectRef,
+        vm: &VirtualMachine,
+    ) -> PyResult<()> {
+        // Tracing is suppressed while dispatching Python audit hooks,
+        // except for hooks that explicitly opt in with __cantrace__.
+        vm.enter_tracing();
+        let can_trace = audit_hook_can_trace(hook, vm);
+        let result = match can_trace {
+            Ok(can_trace) => {
+                if can_trace {
+                    vm.leave_tracing();
+                }
+                let result = hook.call((event, args.clone()), vm).map(|_| ());
+                if can_trace {
+                    vm.enter_tracing();
+                }
+                result
+            }
+            Err(exc) => Err(exc),
+        };
+
+        vm.leave_tracing();
+        result
+    }
+
     #[pyfunction]
     fn audit(event: PyStrRef, args: PosArgs, vm: &VirtualMachine) -> PyResult<()> {
         if vm.audit_hooks.borrow().is_empty() {
@@ -1773,10 +1815,13 @@ pub mod sys {
         let event: PyObjectRef = vm.ctx.new_str("sys.addaudithook").into();
 
         for existing_hook in hooks {
-            let Err(exc) = existing_hook.call((event.clone(), args.clone()), vm) else {
+            let Err(exc) = call_audit_hook(&existing_hook, event.clone(), &args, vm) else {
                 continue;
             };
-            if exc.class().fast_issubclass(vm.ctx.exceptions.runtime_error) {
+            if exc
+                .class()
+                .fast_issubclass(vm.ctx.exceptions.exception_type)
+            {
                 return Ok(());
             }
             return Err(exc);

crates/vm/src/stdlib/sys/monitoring.rs

@@ -747,7 +747,7 @@ fn fire(
     cb_extra: &[PyObjectRef],
 ) -> PyResult<()> {
     // Prevent recursive event firing
-    if FIRING.with(|f| f.get()) {
+    if vm.tracing_is_suppressed() || FIRING.with(|f| f.get()) {
         return Ok(());
     }
 
@@ -795,6 +795,7 @@ fn fire(
     let args = FuncArgs::from(args_vec);
 
     FIRING.with(|f| f.set(true));
+    vm.enter_tracing();
     let result = (|| {
         for (tool, cb) in callbacks {
             let result = cb.call(args.clone(), vm)?;
@@ -817,6 +818,7 @@ fn fire(
         }
         Ok(())
     })();
+    vm.leave_tracing();
     FIRING.with(|f| f.set(false));
     result
 }

crates/vm/src/vm/compile_mode.rs

@@ -0,0 +1,83 @@
+use crate::bytecode;
+
+pub(crate) const PY_SINGLE_INPUT: i32 = 256;
+pub(crate) const PY_FILE_INPUT: i32 = 257;
+pub(crate) const PY_EVAL_INPUT: i32 = 258;
+pub(crate) const PY_FUNC_TYPE_INPUT: i32 = 345;
+
+bitflags::bitflags! {
+    /// `PyCF_*` compiler flags together with the `__future__` `CO_FUTURE_*`
+    /// bits, mirroring `PyCompilerFlags.cf_flags`.
+    ///
+    /// Caveat emptor: these flags are undocumented on purpose and depending on
+    /// their effect outside the standard library is **unsupported**.
+    #[derive(Copy, Clone, Debug, PartialEq, Eq)]
+    pub(crate) struct CompilerFlags: i32 {
+        const SOURCE_IS_UTF8 = 0x0100;
+        const DONT_IMPLY_DEDENT = 0x0200;
+        const ONLY_AST = 0x0400;
+        const IGNORE_COOKIE = 0x0800;
+        const TYPE_COMMENTS = 0x1000;
+        const ALLOW_TOP_LEVEL_AWAIT = 0x2000;
+        const ALLOW_INCOMPLETE_INPUT = 0x4000;
+        const OPTIMIZED_AST = 0x8000 | Self::ONLY_AST.bits();
+
+        // __future__ flags - sync with Lib/__future__.py and Include/cpython/compile.h.
+        const NESTED = 0x0010;
+        const FUTURE_DIVISION = 0x20000;
+        const FUTURE_ABSOLUTE_IMPORT = 0x40000;
+        const FUTURE_WITH_STATEMENT = 0x80000;
+        const FUTURE_PRINT_FUNCTION = 0x100000;
+        const FUTURE_UNICODE_LITERALS = 0x200000;
+        const FUTURE_BARRY_AS_BDFL = 0x400000;
+        const FUTURE_GENERATOR_STOP = 0x800000;
+        const FUTURE_ANNOTATIONS = 0x1000000;
+    }
+}
+
+impl CompilerFlags {
+    const FUTURE_MASK: Self = Self::FUTURE_DIVISION
+        .union(Self::FUTURE_ABSOLUTE_IMPORT)
+        .union(Self::FUTURE_WITH_STATEMENT)
+        .union(Self::FUTURE_PRINT_FUNCTION)
+        .union(Self::FUTURE_UNICODE_LITERALS)
+        .union(Self::FUTURE_BARRY_AS_BDFL)
+        .union(Self::FUTURE_GENERATOR_STOP)
+        .union(Self::FUTURE_ANNOTATIONS);
+    const MASK_OBSOLETE: Self = Self::NESTED;
+    const COMPILE_MASK: Self = Self::ONLY_AST
+        .union(Self::ALLOW_TOP_LEVEL_AWAIT)
+        .union(Self::TYPE_COMMENTS)
+        .union(Self::DONT_IMPLY_DEDENT)
+        .union(Self::ALLOW_INCOMPLETE_INPUT)
+        .union(Self::OPTIMIZED_AST);
+    pub(crate) const ALLOWED_FLAGS: Self = Self::FUTURE_MASK
+        .union(Self::MASK_OBSOLETE)
+        .union(Self::COMPILE_MASK);
+}
+
+// Python-visible `ast.PyCF_*` attribute values. The flags cross the
+// `compile()` boundary as a plain `int`, so the exposed surface stays `i32`.
+pub(crate) const PY_CF_SOURCE_IS_UTF8: i32 = CompilerFlags::SOURCE_IS_UTF8.bits();
+pub(crate) const PY_CF_DONT_IMPLY_DEDENT: i32 = CompilerFlags::DONT_IMPLY_DEDENT.bits();
+pub(crate) const PY_CF_ONLY_AST: i32 = CompilerFlags::ONLY_AST.bits();
+pub(crate) const PY_CF_IGNORE_COOKIE: i32 = CompilerFlags::IGNORE_COOKIE.bits();
+pub(crate) const PY_CF_TYPE_COMMENTS: i32 = CompilerFlags::TYPE_COMMENTS.bits();
+pub(crate) const PY_CF_ALLOW_TOP_LEVEL_AWAIT: i32 = CompilerFlags::ALLOW_TOP_LEVEL_AWAIT.bits();
+pub(crate) const PY_CF_ALLOW_INCOMPLETE_INPUT: i32 = CompilerFlags::ALLOW_INCOMPLETE_INPUT.bits();
+pub(crate) const PY_CF_OPTIMIZED_AST: i32 = CompilerFlags::OPTIMIZED_AST.bits();
+
+pub(crate) fn compile_future_feature_mask() -> bytecode::CodeFlags {
+    // RustPython accepts barry_as_FLUFL but leaves its parser mode disabled.
+    bytecode::CodeFlags::FUTURE_DIVISION
+        | bytecode::CodeFlags::FUTURE_ABSOLUTE_IMPORT
+        | bytecode::CodeFlags::FUTURE_WITH_STATEMENT
+        | bytecode::CodeFlags::FUTURE_PRINT_FUNCTION
+        | bytecode::CodeFlags::FUTURE_UNICODE_LITERALS
+        | bytecode::CodeFlags::FUTURE_GENERATOR_STOP
+        | bytecode::CodeFlags::FUTURE_ANNOTATIONS
+}
+
+pub(crate) fn compile_future_features_from_flags(flags: i32) -> bytecode::CodeFlags {
+    bytecode::CodeFlags::from_bits_truncate(flags as u32 & compile_future_feature_mask().bits())
+}

crates/vm/src/vm/interpreter.rs

@@ -300,8 +300,10 @@ impl Default for InterpreterBuilder {
 ///     let scope = vm.new_scope_with_builtins();
 ///     let source = r#"print("Hello World!")"#;
 ///     let code_obj = vm.compile(
-///         source, Mode::Exec, "<embedded>"
-///     ).map_err(|err| vm.new_syntax_error(&err, Some(source))).unwrap();
+///             source,
+///             Mode::Exec,
+///             "<embedded>",
+///     ).map_err(|err| err.into_pyexception(vm, Some(source))).unwrap();
 ///     vm.run_code_obj(code_obj, scope).unwrap();
 /// });
 /// ```

crates/vm/src/vm/mod.rs

@@ -5,6 +5,9 @@
 
 #[cfg(feature = "rustpython-compiler")]
 mod compile;
+pub(crate) mod compile_mode;
+#[cfg(feature = "rustpython-compiler")]
+pub use compile::VmCompileError;
 mod context;
 mod interpreter;
 mod method;
@@ -82,6 +85,7 @@ pub struct VirtualMachine {
     pub profile_func: RefCell<PyObjectRef>,
     pub trace_func: RefCell<PyObjectRef>,
     pub use_tracing: Cell<bool>,
+    tracing_depth: Cell<usize>,
     pub recursion_limit: Cell<usize>,
     pub(crate) signal_handlers: OnceCell<SignalHandlers>,
     pub(crate) signal_rx: Option<signal::UserSignalReceiver>,
@@ -739,6 +743,7 @@ impl VirtualMachine {
             profile_func,
             trace_func,
             use_tracing: Cell::new(false),
+            tracing_depth: Cell::new(0),
             recursion_limit: Cell::new(if cfg!(debug_assertions) { 256 } else { 1000 }),
             signal_handlers,
             signal_rx: None,
@@ -1095,30 +1100,22 @@ impl VirtualMachine {
     }
 
     pub fn run_code_obj(&self, code: PyRef<PyCode>, scope: Scope) -> PyResult {
-        use crate::builtins::{PyFunction, PyModule};
+        self.run_code_obj_with_closure(code, scope, None)
+    }
 
-        // Create a function object for module code, similar to CPython's PyEval_EvalCode
-        let func = PyFunction::new(code.clone(), scope.globals.clone(), self)?;
-        let func_obj = func.into_ref(&self.ctx).into();
+    pub(crate) fn run_code_obj_with_closure(
+        &self,
+        code: PyRef<PyCode>,
+        scope: Scope,
+        closure: Option<PyRef<crate::builtins::PyTuple<crate::builtins::function::PyCellRef>>>,
+    ) -> PyResult {
+        use crate::builtins::PyFunction;
 
-        // Extract builtins from globals["__builtins__"], like PyEval_EvalCode
-        let builtins = match scope
-            .globals
-            .get_item_opt(identifier!(self, __builtins__), self)?
-        {
-            Some(b) => {
-                if let Some(module) = b.downcast_ref::<PyModule>() {
-                    module.dict().into()
-                } else {
-                    b
-                }
-            }
-            None => self.builtins.dict().into(),
-        };
-
-        let frame =
-            Frame::new(code, scope, builtins, &[], Some(func_obj), false, self).into_ref(&self.ctx);
-        self.run_frame(frame)
+        // Create a function object for module code, similar to PyEval_EvalCode
+        let mut func = PyFunction::new(code, scope.globals.clone(), self)?;
+        func.closure = closure;
+        let func = func.into_ref(&self.ctx);
+        func.invoke_with_locals(FuncArgs::default(), scope.locals, self)
     }
 
     #[cold]
@@ -1434,9 +1431,11 @@ impl VirtualMachine {
     }
 
     /// Stack margin bytes (like _PyOS_STACK_MARGIN_BYTES).
-    /// 2048 * sizeof(void*) = 16KB for 64-bit.
+    /// The margin is doubled for debug/sanitized builds because frame
+    /// evaluation consumes more native stack in those configurations.
     #[cfg_attr(any(miri, target_env = "musl"), allow(dead_code))]
-    const STACK_MARGIN_BYTES: usize = 2048 * core::mem::size_of::<usize>();
+    const STACK_MARGIN_BYTES: usize =
+        (if cfg!(debug_assertions) { 4096 } else { 2048 }) * core::mem::size_of::<usize>();
 
     /// Get the stack boundaries using platform-specific APIs.
     /// Returns (base, top) where base is the lowest address and top is the highest.
@@ -1709,11 +1708,33 @@ impl VirtualMachine {
     #[cfg(feature = "rustpython-codegen")]
     pub fn compile_opts(&self) -> crate::compiler::CompileOpts {
         crate::compiler::CompileOpts {
-            optimize: self.state.config.settings.optimize,
+            optimize: self.state.config.settings.optimize.min(2),
             debug_ranges: self.state.config.settings.code_debug_ranges,
+            int_max_str_digits: self.state.int_max_str_digits.load(),
+            allow_top_level_await: false,
+            future_features: crate::bytecode::CodeFlags::empty(),
+            dont_imply_dedent: false,
+            recursion_limit: self.recursion_limit.get(),
         }
     }
 
+    #[inline]
+    pub(crate) fn enter_tracing(&self) {
+        self.tracing_depth.set(self.tracing_depth.get() + 1);
+    }
+
+    #[inline]
+    pub(crate) fn leave_tracing(&self) {
+        let depth = self.tracing_depth.get();
+        debug_assert!(depth > 0);
+        self.tracing_depth.set(depth.saturating_sub(1));
+    }
+
+    #[inline]
+    pub(crate) fn tracing_is_suppressed(&self) -> bool {
+        self.tracing_depth.get() != 0
+    }
+
     // To be called right before raising the recursion depth.
     fn check_recursive_call(&self, _where: &str) -> PyResult<()> {
         if self.recursion_depth.get() >= self.recursion_limit.get() {
@@ -2286,7 +2307,7 @@ mod tests {
                 let source = "from dir_module.dir_module_inner import value2";
                 let code_obj = vm
                     .compile(source, vm::compiler::Mode::Exec, "<embedded>")
-                    .map_err(|err| vm.new_syntax_error(&err, Some(source)))
+                    .map_err(|err| err.into_pyexception(vm, Some(source)))
                     .unwrap();
 
                 if let Err(e) = vm.run_code_obj(code_obj, scope) {

crates/vm/src/vm/python_run.rs

@@ -22,7 +22,7 @@ impl VirtualMachine {
     pub fn run_string(&self, scope: Scope, source: &str, source_path: &str) -> PyResult {
         let code_obj = self
             .compile(source, compiler::Mode::Exec, source_path)
-            .map_err(|err| self.new_syntax_error(&err, Some(source)))?;
+            .map_err(|err| err.into_pyexception(self, Some(source)))?;
         // linecache._register_code(code, source, filename)
         let _ = self.register_code_in_linecache(&code_obj, source);
         self.run_code_obj(code_obj, scope)
@@ -47,7 +47,7 @@ impl VirtualMachine {
     pub fn run_block_expr(&self, scope: Scope, source: &str) -> PyResult {
         let code_obj = self
             .compile(source, compiler::Mode::BlockExpr, "<embedded>")
-            .map_err(|err| self.new_syntax_error(&err, Some(source)))?;
+            .map_err(|err| err.into_pyexception(self, Some(source)))?;
         self.run_code_obj(code_obj, scope)
     }
 }
@@ -105,11 +105,19 @@ mod file_run {
                 if path != "<stdin>" {
                     set_main_loader(module_dict, path, "SourceFileLoader", self)?;
                 }
-                match crate::host_env::fs::read_to_string(path) {
-                    Ok(source) => {
+                match crate::host_env::fs::read(path) {
+                    Ok(source_bytes) => {
+                        if source_bytes.contains(&0) {
+                            return Err(self.new_exception_msg(
+                                self.ctx.exceptions.syntax_error.to_owned(),
+                                "source code cannot contain null bytes".into(),
+                            ));
+                        }
+                        let source = String::from_utf8(source_bytes)
+                            .map_err(|err| self.new_os_error(err.to_string()))?;
                         let code_obj = self
                             .compile(&source, compiler::Mode::Exec, path)
-                            .map_err(|err| self.new_syntax_error(&err, Some(&source)))?;
+                            .map_err(|err| err.into_pyexception(self, Some(&source)))?;
                         self.run_code_obj(code_obj, scope)?;
                     }
                     Err(err) => {

crates/vm/src/vm/thread.rs

@@ -719,6 +719,7 @@ impl VirtualMachine {
             profile_func: RefCell::new(global_profile.unwrap_or_else(|| self.ctx.none())),
             trace_func: RefCell::new(global_trace.unwrap_or_else(|| self.ctx.none())),
             use_tracing: Cell::new(use_tracing),
+            tracing_depth: Cell::new(0),
             recursion_limit: self.recursion_limit.clone(),
             signal_handlers: core::cell::OnceCell::new(),
             signal_rx: None,

crates/vm/src/vm/vm_new.rs

@@ -54,15 +54,6 @@ impl SyntaxErrorInfo {
         Self { msg, narrow_caret }
     }
 
-    fn with_msg(&mut self, msg: &str) {
-        self.msg = msg.into();
-    }
-
-    #[cfg(feature = "parser")]
-    const fn with_narrow_caret(&mut self, narrow_caret: bool) {
-        self.narrow_caret = narrow_caret;
-    }
-
     #[cfg(feature = "parser")]
     #[must_use]
     const fn handle_expected_token(expected: TokenKind, found: TokenKind) -> &'static str {
@@ -114,7 +105,7 @@ impl SyntaxErrorInfo {
             }
 
             ParseErrorType::InvalidStarredExpressionUsage => {
-                self.with_narrow_caret(true);
+                self.narrow_caret = true;
                 "invalid syntax".into()
             }
 
@@ -134,7 +125,7 @@ impl SyntaxErrorInfo {
             ParseErrorType::EmptyTypeParams => "Type parameter list cannot be empty".into(),
 
             ParseErrorType::InvalidStarPatternUsage => {
-                self.with_narrow_caret(true);
+                self.narrow_caret = true;
                 "cannot use starred expression here".into()
             }
 
@@ -179,12 +170,22 @@ impl SyntaxErrorInfo {
             | ParseErrorType::SimpleAndCompoundStatementOnSameLine
             | ParseErrorType::ExpectedExpression => "invalid syntax".into(),
 
+            ParseErrorType::OtherError(s) if s.starts_with("Expected an identifier") => {
+                "invalid syntax".into()
+            }
+
             ParseErrorType::OtherError(s)
-                if s.starts_with("Expected an identifier, but found a keyword") =>
+                if s.eq_ignore_ascii_case(
+                    "Expected a type parameter or the end of the type parameter list",
+                ) =>
             {
                 "invalid syntax".into()
             }
 
+            ParseErrorType::OtherError(s) if s.eq_ignore_ascii_case("Expected a statement") => {
+                "invalid syntax".into()
+            }
+
             ParseErrorType::OtherError(s)
                 if s.eq_ignore_ascii_case(
                     "bytes literal cannot be mixed with non-bytes literals",
@@ -262,7 +263,7 @@ impl SyntaxErrorInfo {
             _ => return,
         };
 
-        self.with_msg(&msg);
+        self.msg = msg;
     }
 }
 
@@ -576,6 +577,16 @@ impl VirtualMachine {
         source: Option<&str>,
         allow_incomplete: bool,
     ) -> PyBaseExceptionRef {
+        if matches!(
+            error,
+            crate::compiler::CompileError::Codegen(crate::compiler::codegen::error::CodegenError {
+                error: crate::compiler::codegen::error::CodegenErrorType::RecursionError,
+                ..
+            })
+        ) {
+            return self.new_recursion_error(error.to_string());
+        }
+
         let incomplete_or_syntax = |allow| -> &'static Py<crate::builtins::PyType> {
             if allow {
                 self.ctx.exceptions.incomplete_input_error
@@ -687,7 +698,9 @@ impl VirtualMachine {
                 raw_location,
                 ..
             }) => {
-                if s.starts_with("Expected an indented block after") {
+                if s.starts_with("Expected an indented block after")
+                    || s.starts_with("expected an indented block after")
+                {
                     if allow_incomplete {
                         // Check that all chars in the error are whitespace, if so, the source is
                         // incomplete. Otherwise, we've found code that might violates
@@ -717,6 +730,12 @@ impl VirtualMachine {
                     } else {
                         self.ctx.exceptions.indentation_error
                     }
+                } else if allow_incomplete
+                    && source.is_some_and(|source| {
+                        raw_location.end().to_usize() >= source.len() && !source.ends_with('\n')
+                    })
+                {
+                    self.ctx.exceptions.incomplete_input_error
                 } else {
                     self.ctx.exceptions.syntax_error
                 }
@@ -738,7 +757,29 @@ impl VirtualMachine {
         let statement = source.and_then(|src| get_statement(src, error.location()));
 
         let mut msg = error.to_string();
-        if let Some(msg) = msg.get_mut(..1) {
+        if !msg.starts_with("Exceeds the limit ")
+            && !msg.starts_with("Did you mean ")
+            && !msg.starts_with("Invalid star expression")
+            && !msg.starts_with("Function parameters cannot be parenthesized")
+            && !msg.starts_with("Lambda expression parameters cannot be parenthesized")
+            && !msg.starts_with("Cannot have two type comments on def")
+            && !msg.starts_with("Variable annotation syntax is")
+            && !msg.starts_with("The '@' operator is")
+            && !msg.starts_with("Async functions are")
+            && !msg.starts_with("Async comprehensions are")
+            && !msg.starts_with("Async for loops are")
+            && !msg.starts_with("Async with statements are")
+            && !msg.starts_with("Exception groups are")
+            && !msg.starts_with("Positional-only parameters are")
+            && !msg.starts_with("Pattern matching is")
+            && !msg.starts_with("Type statement is")
+            && !msg.starts_with("Type parameter lists are")
+            && !msg.starts_with("Type parameter defaults are")
+            && !msg.starts_with("Assignment expressions are")
+            && !msg.starts_with("Await expressions are")
+            && !msg.starts_with("Underscores in numeric literals are")
+            && let Some(msg) = msg.get_mut(..1)
+        {
             msg.make_ascii_lowercase();
         }
 
@@ -753,17 +794,26 @@ impl VirtualMachine {
         };
 
         if syntax_error_type.is(self.ctx.exceptions.tab_error) {
-            syntax_error_info.with_msg("inconsistent use of tabs and spaces in indentation");
+            syntax_error_info.msg = "inconsistent use of tabs and spaces in indentation".to_owned();
+        }
+        if syntax_error_type.is(self.ctx.exceptions.incomplete_input_error) {
+            syntax_error_info.msg = "incomplete input".to_owned();
         }
 
         let SyntaxErrorInfo { msg, narrow_caret } = syntax_error_info;
+        let check_version_suite_error = msg.starts_with("Async functions are")
+            || msg.starts_with("Async for loops are")
+            || msg.starts_with("Async with statements are")
+            || msg.starts_with("Exception groups are")
+            || msg.starts_with("except expressions without parentheses are")
+            || msg.starts_with("Pattern matching is");
+        let line_end_binary_operator_error = msg.starts_with("The '@' operator is");
 
         let syntax_error = self.new_exception_msg(syntax_error_type, msg.into());
 
-        let (lineno, offset) = error.python_location();
-        let lineno = self.ctx.new_int(lineno);
-        let offset = self.ctx.new_int(offset);
-
+        let (lineno_raw, offset_raw) = error.python_location();
+        let lineno = self.ctx.new_int(lineno_raw);
+        let offset = self.ctx.new_int(offset_raw);
         set_attrs!(
             syntax_error.as_object(), self, unwrap,
             "lineno" => lineno,
@@ -772,15 +822,23 @@ impl VirtualMachine {
 
         // Set end_lineno and end_offset if available
         if let Some((end_lineno, end_offset)) = error.python_end_location() {
-            let (end_lineno, end_offset) = if narrow_caret {
+            let (end_lineno, end_offset) = if check_version_suite_error
+                && statement
+                    .as_deref()
+                    .and_then(|line| line.chars().next())
+                    .is_some_and(|ch| ch.is_ascii_whitespace())
+            {
+                (end_lineno, -1)
+            } else if line_end_binary_operator_error && end_offset == offset_raw {
+                (end_lineno, (end_offset + 1) as isize)
+            } else if narrow_caret {
                 let (l, o) = error.python_location();
-                (l, o + 1)
+                (l, (o + 1) as isize)
             } else {
-                (end_lineno, end_offset)
+                (end_lineno, end_offset as isize)
             };
             let end_lineno = self.ctx.new_int(end_lineno);
             let end_offset = self.ctx.new_int(end_offset);
-
             set_attrs!(
                 syntax_error.as_object(), self, unwrap,
                 "end_lineno" => end_lineno,

crates/wasm/Cargo.toml

@@ -20,6 +20,7 @@ no-start-func = []
 rustpython-common = { workspace = true }
 rustpython-pylib = { workspace = true, optional = true }
 rustpython-stdlib = { workspace = true, default-features = false, optional = true }
+ruff_text_size = { workspace = true }
 # make sure no threading! otherwise wasm build will fail
 rustpython-vm = { workspace = true, features = ["compiler", "encodings", "serde", "wasmbind"] }
 

crates/wasm/src/lib.rs

@@ -70,7 +70,11 @@ pub mod eval {
         if let Some(js_vars) = js_vars {
             vm.add_to_scope("js_vars".into(), js_vars.into())?;
         }
-        vm.run(source, mode, None)
+        if matches!(mode, Mode::Single) {
+            vm.exec_single(source, None)
+        } else {
+            vm.run(source, mode, None)
+        }
     }
 
     /// Evaluate Python code

crates/wasm/src/vm_class.rs

@@ -6,9 +6,14 @@ use crate::{
 use alloc::rc::{Rc, Weak};
 use core::cell::RefCell;
 use js_sys::{Object, TypeError};
+use ruff_text_size::Ranged;
 use rustpython_vm::{
-    Interpreter, PyObjectRef, PyRef, PyResult, Settings, VirtualMachine, builtins::PyWeak,
-    compiler::Mode, function::ArgMapping, scope::Scope,
+    Interpreter, PyObjectRef, PyRef, PyResult, Settings, VirtualMachine,
+    builtins::PyWeak,
+    compiler::{self, Mode},
+    function::ArgMapping,
+    scope::Scope,
+    vm::VmCompileError,
 };
 use std::collections::HashMap;
 use wasm_bindgen::prelude::*;
@@ -21,6 +26,25 @@ pub(crate) struct StoredVirtualMachine {
     held_objects: RefCell<Vec<PyObjectRef>>,
 }
 
+fn compile_err_to_js(vm: &VirtualMachine, err: VmCompileError) -> JsValue {
+    match err {
+        VmCompileError::Compile(err) => convert::syntax_err(err).into(),
+        err => convert::py_err_to_js_err(vm, &err.into_pyexception(vm, None)),
+    }
+}
+
+fn statement_chunks(source: &str) -> Option<Vec<&str>> {
+    let module = compiler::parser::parse_module(source).ok()?.into_syntax();
+    module
+        .body
+        .iter()
+        .map(|stmt| {
+            let range = stmt.range();
+            source.get(range.start().to_usize()..range.end().to_usize())
+        })
+        .collect()
+}
+
 #[pymodule]
 mod _window {
     use super::{js_module, wasm_builtins};
@@ -263,8 +287,8 @@ impl WASMVirtualMachine {
     ) -> Result<(), JsValue> {
         self.with_vm(|vm, _| {
             let code = vm
-                .compile(source, Mode::Exec, &name)
-                .map_err(convert::syntax_err)?;
+                .compile(source, Mode::Exec, name.as_str())
+                .map_err(|err| compile_err_to_js(vm, err))?;
             let attrs = vm.ctx.new_dict();
             attrs
                 .set_item("__name__", vm.new_pyobj(name.as_str()), vm)
@@ -327,13 +351,46 @@ impl WASMVirtualMachine {
     ) -> Result<JsValue, JsValue> {
         self.with_vm(|vm, StoredVirtualMachine { scope, .. }| {
             let source_path = source_path.unwrap_or_else(|| "<wasm>".to_owned());
-            let code = vm.compile(source, mode, &source_path);
-            let code = code.map_err(convert::syntax_err)?;
+            let code = vm.compile(source, mode, source_path.as_str());
+            let code = code.map_err(|err| compile_err_to_js(vm, err))?;
             let result = vm.run_code_obj(code, scope.clone());
             convert::pyresult_to_js_result(vm, result)
         })?
     }
 
+    pub(crate) fn run_single(
+        &self,
+        source: &str,
+        source_path: Option<String>,
+    ) -> Result<JsValue, JsValue> {
+        self.with_vm(|vm, StoredVirtualMachine { scope, .. }| {
+            let source_path = source_path.unwrap_or_else(|| "<wasm>".to_owned());
+            let Some(chunks) = statement_chunks(source) else {
+                let code = vm.compile(source, Mode::Single, source_path.as_str());
+                let code = code.map_err(|err| compile_err_to_js(vm, err))?;
+                let result = vm.run_code_obj(code, scope.clone());
+                return convert::pyresult_to_js_result(vm, result);
+            };
+
+            if chunks.is_empty() {
+                return Ok(convert::py_to_js(vm, vm.ctx.none()));
+            }
+
+            let displayhook = vm
+                .sys_module
+                .get_attr("displayhook", vm)
+                .map_err(|_| TypeError::new("lost sys.displayhook"))?;
+            let mut result = vm.ctx.none();
+            for chunk in chunks {
+                let code = vm.compile(chunk, Mode::BlockExpr, source_path.as_str());
+                let code = code.map_err(|err| compile_err_to_js(vm, err))?;
+                result = vm.run_code_obj(code, scope.clone()).into_js(vm)?;
+                displayhook.call((result.clone(),), vm).into_js(vm)?;
+            }
+            Ok(convert::py_to_js(vm, result))
+        })?
+    }
+
     pub fn exec(&self, source: &str, source_path: Option<String>) -> Result<JsValue, JsValue> {
         self.run(source, Mode::Exec, source_path)
     }
@@ -348,6 +405,6 @@ impl WASMVirtualMachine {
         source: &str,
         source_path: Option<String>,
     ) -> Result<JsValue, JsValue> {
-        self.run(source, Mode::Single, source_path)
+        self.run_single(source, source_path)
     }
 }

examples/hello_embed.rs

@@ -6,7 +6,7 @@ fn main() -> vm::PyResult<()> {
         let source = r#"print("Hello World!")"#;
         let code_obj = vm
             .compile(source, vm::compiler::Mode::Exec, "<embedded>")
-            .map_err(|err| vm.new_syntax_error(&err, Some(source)))?;
+            .map_err(|err| err.into_pyexception(vm, Some(source)))?;
 
         vm.run_code_obj(code_obj, scope)?;
 

examples/mini_repl.rs

@@ -66,7 +66,7 @@ def fib(n):
         // (note that this is only the case when compiler::Mode::Single is passed to vm.compile)
         match vm
             .compile(&input, vm::compiler::Mode::Single, "<embedded>")
-            .map_err(|err| vm.new_syntax_error(&err, Some(&input)))
+            .map_err(|err| err.into_pyexception(vm, Some(&input)))
             .and_then(|code_obj| vm.run_code_obj(code_obj, scope.clone()))
         {
             Ok(output) => {