Fix bytes.isspace (#5655)

* Fix implementation of bytes.isspace to match the implementation in CPython.

* Fix implementation of bytes.istitle so that test_bigmem now passes.
Also added some extra_test to capture this.

* Fix some clippy issues

Signed-off-by: Hanif Ariffin <hanif.ariffin.4326@gmail.com>

* bytes.istitle returns false when the bytes is empty.

* Fix some edges of istitle with non-alphanumeric characters

* Disable a line because python 3.12 errors on bad escape sequence

* Match CPython's istitle a little closer

* Fix clippy warning

* Rewrite bytes.istitle along _Py_bytes_istitle

The lookahead-based check accepted b"Not--a Titlecase String", which
test_bytes covers. Track whether the previous byte was cased instead.

isspace compares bytes directly rather than converting to char.

Restore the b"omkmok\Xaa" assert: 3.14 reports an invalid escape
sequence as a SyntaxWarning, not an error.

Assisted-by: Claude

---------

Signed-off-by: Hanif Ariffin <hanif.ariffin.4326@gmail.com>
Co-authored-by: Jeong YunWon <jeong@youknowone.org>

Lib/test/test_bigmem.py

@@ -789,14 +789,6 @@ class BytesTest(unittest.TestCase, BaseStrTest):
     def test_swapcase(self, size):
         self._test_swapcase(size)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
-    def test_isspace(self):
-        return super().test_isspace()
-
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
-    def test_istitle(self):
-        return super().test_istitle()
-
 
 class BytearrayTest(unittest.TestCase, BaseStrTest):
 
@@ -823,14 +815,6 @@ class BytearrayTest(unittest.TestCase, BaseStrTest):
     test_hash = None
     test_split_large = None
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
-    def test_isspace(self):
-        return super().test_isspace()
-
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
-    def test_istitle(self):
-        return super().test_istitle()
-
 class TupleTest(unittest.TestCase):
 
     # Tuples have a small, fixed-sized head and an array of pointers to

crates/vm/src/bytes_inner.rs

@@ -397,44 +397,40 @@ impl PyBytesInner {
         self.elements.py_isupper()
     }
 
+    // _Py_bytes_isspace
     pub fn isspace(&self) -> bool {
+        // is_ascii_whitespace excludes vertical tab, while Py_ISSPACE accepts it
         !self.elements.is_empty()
             && self
                 .elements
                 .iter()
-                .all(|x| char::from(*x).is_ascii_whitespace())
+                .all(|x| x.is_ascii_whitespace() || *x == b'\x0b')
     }
 
+    // _Py_bytes_istitle
     pub fn istitle(&self) -> bool {
-        if self.elements.is_empty() {
-            return false;
-        }
+        let mut cased = false;
+        let mut previous_is_cased = false;
 
-        let mut iter = self.elements.iter().peekable();
-        let mut prev_cased = false;
-
-        while let Some(c) = iter.next() {
-            let current = char::from(*c);
-            let next = if let Some(k) = iter.peek() {
-                char::from(**k)
-            } else if current.is_uppercase() {
-                return !prev_cased;
+        for byte in &self.elements {
+            if byte.is_ascii_uppercase() {
+                if previous_is_cased {
+                    return false;
+                }
+                previous_is_cased = true;
+                cased = true;
+            } else if byte.is_ascii_lowercase() {
+                if !previous_is_cased {
+                    return false;
+                }
+                previous_is_cased = true;
+                cased = true;
             } else {
-                return prev_cased;
-            };
-
-            let is_cased = current.to_uppercase().next().unwrap() != current
-                || current.to_lowercase().next().unwrap() != current;
-            if (is_cased && next.is_uppercase() && !prev_cased)
-                || (!is_cased && next.is_lowercase())
-            {
-                return false;
+                previous_is_cased = false;
             }
-
-            prev_cased = is_cased;
         }
 
-        true
+        cased
     }
 
     pub fn lower(&self) -> Vec<u8> {

extra_tests/snippets/builtin_bytes.py

@@ -708,3 +708,42 @@ b = B1.fromhex("a0a1a2")
 assert b.foo == "bar"
 
 skip_if_unsupported(3, 11, test__bytes__)
+
+assert " \f\n\r\t\v".encode("utf-8").isspace()
+assert " \f\n\r\t\v".encode("latin-1").isspace()
+
+# bytes.istitle tests
+s = b"Aa6A"
+assert s.istitle(), f"{s}"
+s = b"Aa6aA"
+assert not s.istitle(), f"{s}"
+s = b"Python Is Fun"
+assert s.istitle(), f"{s}"
+s = b"Python is fun"
+assert not s.istitle(), f"{s}"
+s = b"PYTHON IS FUN"
+assert not s.istitle(), f"{s}"
+s = b"Python 3.9 Is Awesome!"
+assert s.istitle(), f"{s}"
+s = b""
+assert not s.istitle(), f"{s}"
+s = b"Hello Is Amazing"
+assert s.istitle(), f"{s}"
+s = b"Not--a Titlecase String"
+assert not s.istitle(), f"{s}"
+s = b"123A"
+assert s.istitle(), f"{s}"
+s = b"123a"
+assert not s.istitle(), f"{s}"
+s = b"123A\ta"
+assert not s.istitle(), f"{s}"
+SUBSTR = b"123456"
+s = b"".join([b"A", b"a" * 64, SUBSTR])
+assert s.istitle(), f"{s}"
+s += b"A"
+assert s.istitle(), f"{s}"
+s += b"aA"
+assert not s.istitle(), f"{s}"
+assert "123A".istitle(), f"{s}"
+assert not "123a".istitle(), f"{s}"
+assert not "123A\ta".istitle(), f"{s}"