Merge pull request #2045 from Lynskylate/feature-symtable

Add is_imported and is_annotation for symtable

Lib/test/test_symtable.py

@@ -0,0 +1,251 @@
+"""
+Test the API of the symtable module.
+"""
+import symtable
+import unittest
+
+
+
+TEST_CODE = """
+import sys
+
+glob = 42
+some_var = 12
+
+class Mine:
+    instance_var = 24
+    def a_method(p1, p2):
+        pass
+
+def spam(a, b, *var, **kw):
+    global bar
+    bar = 47
+    some_var = 10
+    x = 23
+    glob
+    def internal():
+        return x
+    def other_internal():
+        nonlocal some_var
+        some_var = 3
+        return some_var
+    return internal
+
+def foo():
+    pass
+
+def namespace_test(): pass
+def namespace_test(): pass
+"""
+
+
+def find_block(block, name):
+    for ch in block.get_children():
+        if ch.get_name() == name:
+            return ch
+
+
+class SymtableTest(unittest.TestCase):
+
+    top = symtable.symtable(TEST_CODE, "?", "exec")
+    # These correspond to scopes in TEST_CODE
+    Mine = find_block(top, "Mine")
+    a_method = find_block(Mine, "a_method")
+    spam = find_block(top, "spam")
+    internal = find_block(spam, "internal")
+    other_internal = find_block(spam, "other_internal")
+    foo = find_block(top, "foo")
+
+    def test_type(self):
+        self.assertEqual(self.top.get_type(), "module")
+        self.assertEqual(self.Mine.get_type(), "class")
+        self.assertEqual(self.a_method.get_type(), "function")
+        self.assertEqual(self.spam.get_type(), "function")
+        self.assertEqual(self.internal.get_type(), "function")
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_id(self):
+        self.assertGreater(self.top.get_id(), 0)
+        self.assertGreater(self.Mine.get_id(), 0)
+        self.assertGreater(self.a_method.get_id(), 0)
+        self.assertGreater(self.spam.get_id(), 0)
+        self.assertGreater(self.internal.get_id(), 0)
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_optimized(self):
+        self.assertFalse(self.top.is_optimized())
+
+        self.assertTrue(self.spam.is_optimized())
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_nested(self):
+        self.assertFalse(self.top.is_nested())
+        self.assertFalse(self.Mine.is_nested())
+        self.assertFalse(self.spam.is_nested())
+        self.assertTrue(self.internal.is_nested())
+
+    def test_children(self):
+        self.assertTrue(self.top.has_children())
+        self.assertTrue(self.Mine.has_children())
+        self.assertFalse(self.foo.has_children())
+
+    def test_lineno(self):
+        self.assertEqual(self.top.get_lineno(), 0)
+        self.assertEqual(self.spam.get_lineno(), 12)
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_function_info(self):
+        func = self.spam
+        self.assertEqual(sorted(func.get_parameters()), ["a", "b", "kw", "var"])
+        expected = ['a', 'b', 'internal', 'kw', 'other_internal', 'some_var', 'var', 'x']
+        self.assertEqual(sorted(func.get_locals()), expected)
+        self.assertEqual(sorted(func.get_globals()), ["bar", "glob"])
+        self.assertEqual(self.internal.get_frees(), ("x",))
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_globals(self):
+        self.assertTrue(self.spam.lookup("glob").is_global())
+        self.assertFalse(self.spam.lookup("glob").is_declared_global())
+        self.assertTrue(self.spam.lookup("bar").is_global())
+        self.assertTrue(self.spam.lookup("bar").is_declared_global())
+        self.assertFalse(self.internal.lookup("x").is_global())
+        self.assertFalse(self.Mine.lookup("instance_var").is_global())
+        self.assertTrue(self.spam.lookup("bar").is_global())
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_nonlocal(self):
+        self.assertFalse(self.spam.lookup("some_var").is_nonlocal())
+        self.assertTrue(self.other_internal.lookup("some_var").is_nonlocal())
+        expected = ("some_var",)
+        self.assertEqual(self.other_internal.get_nonlocals(), expected)
+
+    def test_local(self):
+        self.assertTrue(self.spam.lookup("x").is_local())
+        self.assertFalse(self.spam.lookup("bar").is_local())
+
+    def test_free(self):
+        self.assertTrue(self.internal.lookup("x").is_free())
+
+    def test_referenced(self):
+        self.assertTrue(self.internal.lookup("x").is_referenced())
+        self.assertTrue(self.spam.lookup("internal").is_referenced())
+        self.assertFalse(self.spam.lookup("x").is_referenced())
+
+    def test_parameters(self):
+        for sym in ("a", "var", "kw"):
+            self.assertTrue(self.spam.lookup(sym).is_parameter())
+        self.assertFalse(self.spam.lookup("x").is_parameter())
+
+    def test_symbol_lookup(self):
+        self.assertEqual(len(self.top.get_identifiers()),
+                         len(self.top.get_symbols()))
+
+        self.assertRaises(KeyError, self.top.lookup, "not_here")
+
+    def test_namespaces(self):
+        self.assertTrue(self.top.lookup("Mine").is_namespace())
+        self.assertTrue(self.Mine.lookup("a_method").is_namespace())
+        self.assertTrue(self.top.lookup("spam").is_namespace())
+        self.assertTrue(self.spam.lookup("internal").is_namespace())
+        self.assertTrue(self.top.lookup("namespace_test").is_namespace())
+        self.assertFalse(self.spam.lookup("x").is_namespace())
+
+        # TODO(RUSTPYTHON): lookup should return same pythonref
+        # self.assertTrue(self.top.lookup("spam").get_namespace() is self.spam)
+        ns_test = self.top.lookup("namespace_test")
+        self.assertEqual(len(ns_test.get_namespaces()), 2)
+        self.assertRaises(ValueError, ns_test.get_namespace)
+
+    def test_assigned(self):
+        self.assertTrue(self.spam.lookup("x").is_assigned())
+        self.assertTrue(self.spam.lookup("bar").is_assigned())
+        self.assertTrue(self.top.lookup("spam").is_assigned())
+        self.assertTrue(self.Mine.lookup("a_method").is_assigned())
+        self.assertFalse(self.internal.lookup("x").is_assigned())
+
+    def test_annotated(self):
+        st1 = symtable.symtable('def f():\n    x: int\n', 'test', 'exec')
+        st2 = st1.get_children()[0]
+        self.assertTrue(st2.lookup('x').is_local())
+        self.assertTrue(st2.lookup('x').is_annotated())
+        self.assertFalse(st2.lookup('x').is_global())
+        st3 = symtable.symtable('def f():\n    x = 1\n', 'test', 'exec')
+        st4 = st3.get_children()[0]
+        self.assertTrue(st4.lookup('x').is_local())
+        self.assertFalse(st4.lookup('x').is_annotated())
+
+        # Test that annotations in the global scope are valid after the
+        # variable is declared as nonlocal.
+        st5 = symtable.symtable('global x\nx: int', 'test', 'exec')
+        self.assertTrue(st5.lookup("x").is_global())
+
+        # Test that annotations for nonlocals are valid after the
+        # variable is declared as nonlocal.
+        st6 = symtable.symtable('def g():\n'
+                                '    x = 2\n'
+                                '    def f():\n'
+                                '        nonlocal x\n'
+                                '    x: int',
+                                'test', 'exec')
+
+    def test_imported(self):
+        self.assertTrue(self.top.lookup("sys").is_imported())
+
+    def test_name(self):
+        self.assertEqual(self.top.get_name(), "top")
+        self.assertEqual(self.spam.get_name(), "spam")
+        self.assertEqual(self.spam.lookup("x").get_name(), "x")
+        self.assertEqual(self.Mine.get_name(), "Mine")
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_class_info(self):
+        self.assertEqual(self.Mine.get_methods(), ('a_method',))
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_filename_correct(self):
+        ### Bug tickler: SyntaxError file name correct whether error raised
+        ### while parsing or building symbol table.
+        def checkfilename(brokencode, offset):
+            try:
+                symtable.symtable(brokencode, "spam", "exec")
+            except SyntaxError as e:
+                self.assertEqual(e.filename, "spam")
+                self.assertEqual(e.lineno, 1)
+                self.assertEqual(e.offset, offset)
+            else:
+                self.fail("no SyntaxError for %r" % (brokencode,))
+        # TODO: RUSTPYTHON, now offset get 15
+        checkfilename("def f(x): foo)(", 14)  # parse-time
+        checkfilename("def f(x): global x", 11)  # symtable-build-time
+        symtable.symtable("pass", b"spam", "exec")
+        with self.assertWarns(DeprecationWarning), \
+             self.assertRaises(TypeError):
+            symtable.symtable("pass", bytearray(b"spam"), "exec")
+        with self.assertWarns(DeprecationWarning):
+            symtable.symtable("pass", memoryview(b"spam"), "exec")
+        with self.assertRaises(TypeError):
+            symtable.symtable("pass", list(b"spam"), "exec")
+
+    def test_eval(self):
+        symbols = symtable.symtable("42", "?", "eval")
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_single(self):
+        symbols = symtable.symtable("42", "?", "single")
+
+    def test_exec(self):
+        symbols = symtable.symtable("def f(x): return x", "?", "exec")
+
+    @unittest.skip("TODO: RUSTPYTHON")
+    def test_bytes(self):
+        top = symtable.symtable(TEST_CODE.encode('utf8'), "?", "exec")
+        self.assertIsNotNone(find_block(top, "Mine"))
+
+        code = b'# -*- coding: iso8859-15 -*-\nclass \xb4: pass\n'
+
+        top = symtable.symtable(code, "?", "exec")
+        self.assertIsNotNone(find_block(top, "\u017d"))
+
+
+if __name__ == '__main__':
+    unittest.main()

compiler/src/symboltable.rs

@@ -97,10 +97,13 @@ pub struct Symbol {
     pub name: String,
     // pub table: SymbolTableRef,
     pub scope: SymbolScope,
+    // TODO: Use bitflags replace
     pub is_referenced: bool,
     pub is_assigned: bool,
     pub is_parameter: bool,
     pub is_free: bool,
+    pub is_annotated: bool,
+    pub is_imported: bool,
 
     // indicates if the symbol gets a value assigned by a named expression in a comprehension
     // this is required to correct the scope in the analysis.
@@ -121,6 +124,8 @@ impl Symbol {
             is_assigned: false,
             is_parameter: false,
             is_free: false,
+            is_annotated: false,
+            is_imported: false,
             is_assign_namedexpr_in_comprehension: false,
             is_iter: false,
         }
@@ -394,7 +399,10 @@ enum SymbolUsage {
     Nonlocal,
     Used,
     Assigned,
+    Imported,
+    AnnotationAssigned,
     Parameter,
+    AnnotationParameter,
     AssignedNamedExprInCompr,
     Iter,
 }
@@ -460,7 +468,12 @@ impl SymbolTableBuilder {
     }
 
     fn scan_parameter(&mut self, parameter: &ast::Parameter) -> SymbolTableResult {
-        self.register_name(&parameter.arg, SymbolUsage::Parameter)
+        let usage = if parameter.annotation.is_some() {
+            SymbolUsage::AnnotationParameter
+        } else {
+            SymbolUsage::Parameter
+        };
+        self.register_name(&parameter.arg, usage)
     }
 
     fn scan_parameters_annotations(&mut self, parameters: &[ast::Parameter]) -> SymbolTableResult {
@@ -564,12 +577,12 @@ impl SymbolTableBuilder {
                 for name in names {
                     if let Some(alias) = &name.alias {
                         // `import mymodule as myalias`
-                        self.register_name(alias, SymbolUsage::Assigned)?;
+                        self.register_name(alias, SymbolUsage::Imported)?;
                     } else {
                         // `import module`
                         self.register_name(
                             name.symbol.split('.').next().unwrap(),
-                            SymbolUsage::Assigned,
+                            SymbolUsage::Imported,
                         )?;
                     }
                 }
@@ -601,7 +614,12 @@ impl SymbolTableBuilder {
                 annotation,
                 value,
             } => {
-                self.scan_expression(target, &ExpressionContext::Store)?;
+                // https://github.com/python/cpython/blob/master/Python/symtable.c#L1233
+                if let ast::ExpressionType::Identifier { ref name } = target.node {
+                    self.register_name(name, SymbolUsage::AnnotationAssigned)?;
+                } else {
+                    self.scan_expression(target, &ExpressionContext::Store)?;
+                }
                 self.scan_expression(annotation, &ExpressionContext::Load)?;
                 if let Some(value) = value {
                     self.scan_expression(value, &ExpressionContext::Load)?;
@@ -984,9 +1002,21 @@ impl SymbolTableBuilder {
                     });
                 }
             }
+            SymbolUsage::Imported => {
+                symbol.is_assigned = true;
+                symbol.is_imported = true;
+            }
             SymbolUsage::Parameter => {
                 symbol.is_parameter = true;
             }
+            SymbolUsage::AnnotationParameter => {
+                symbol.is_parameter = true;
+                symbol.is_annotated = true;
+            }
+            SymbolUsage::AnnotationAssigned => {
+                symbol.is_assigned = true;
+                symbol.is_annotated = true;
+            }
             SymbolUsage::Assigned => {
                 symbol.is_assigned = true;
             }

vm/src/stdlib/symtable.rs

@@ -109,10 +109,17 @@ impl PySymbolTable {
         if let Some(symbol) = self.symtable.symbols.get(name) {
             Ok(PySymbol {
                 symbol: symbol.clone(),
+                namespaces: self
+                    .symtable
+                    .sub_tables
+                    .iter()
+                    .filter(|table| table.name == name)
+                    .cloned()
+                    .collect(),
             }
             .into_ref(vm))
         } else {
-            Err(vm.new_lookup_error(name.to_owned()))
+            Err(vm.new_key_error(vm.new_str(format!("lookup {} failed", name))))
         }
     }
 
@@ -133,7 +140,20 @@ impl PySymbolTable {
             .symtable
             .symbols
             .values()
-            .map(|s| (PySymbol { symbol: s.clone() }).into_ref(vm).into_object())
+            .map(|s| {
+                (PySymbol {
+                    symbol: s.clone(),
+                    namespaces: self
+                        .symtable
+                        .sub_tables
+                        .iter()
+                        .filter(|&table| table.name == s.name)
+                        .cloned()
+                        .collect(),
+                })
+                .into_ref(vm)
+                .into_object()
+            })
             .collect();
         Ok(vm.ctx.new_list(symbols))
     }
@@ -158,6 +178,7 @@ impl PySymbolTable {
 #[pyclass(name = "Symbol")]
 struct PySymbol {
     symbol: symboltable::Symbol,
+    namespaces: Vec<symboltable::SymbolTable>,
 }
 
 impl fmt::Debug for PySymbol {
@@ -189,6 +210,25 @@ impl PySymbol {
         self.symbol.is_local()
     }
 
+    #[pymethod(name = "is_imported")]
+    fn is_imported(&self) -> bool {
+        self.symbol.is_imported
+    }
+
+    #[pymethod(name = "is_nested")]
+    fn is_nested(&self) -> bool {
+        // TODO
+        false
+    }
+
+    #[pymethod(name = "is_nonlocal")]
+    fn is_nonlocal(&self) -> bool {
+        match self.symbol.scope {
+            symboltable::SymbolScope::Nonlocal => true,
+            _ => false,
+        }
+    }
+
     #[pymethod(name = "is_referenced")]
     fn is_referenced(&self) -> bool {
         self.symbol.is_referenced
@@ -211,7 +251,32 @@ impl PySymbol {
 
     #[pymethod(name = "is_namespace")]
     fn is_namespace(&self) -> bool {
-        // TODO
-        false
+        !self.namespaces.is_empty()
+    }
+
+    #[pymethod(name = "is_annotated")]
+    fn is_annotated(&self) -> bool {
+        self.symbol.is_annotated
+    }
+
+    #[pymethod(name = "get_namespaces")]
+    fn get_namespaces(&self, vm: &VirtualMachine) -> PyResult {
+        let namespaces = self
+            .namespaces
+            .iter()
+            .map(|table| to_py_symbol_table(table.clone()).into_ref(vm).into_object())
+            .collect();
+        Ok(vm.ctx.new_list(namespaces))
+    }
+
+    #[pymethod(name = "get_namespace")]
+    fn get_namespace(&self, vm: &VirtualMachine) -> PyResult {
+        if self.namespaces.len() != 1 {
+            Err(vm.new_value_error("namespace is bound to multiple namespaces".to_owned()))
+        } else {
+            Ok(to_py_symbol_table(self.namespaces.first().unwrap().clone())
+                .into_ref(vm)
+                .into_object())
+        }
     }
 }