gh-135801: Improve filtering by module in warn_explicit() without module argument

* Try to match the module name pattern with module names constructed
  starting from different parent directories of the filename.
  E.g., for "/path/to/package/module" try to match with
  "path.to.package.module", "to.package.module", "package.module" and
  "module".
* Ignore trailing "/__init__".
* Keep matching with the full filename (without optional ".py" extension)
  for compatibility.
* Only ignore the case of the ".py" extension on Windows.

Author: serhiy-storchaka

Lib/_py_warnings.py

@@ -520,20 +520,44 @@ def warn(message, category=None, stacklevel=1, source=None,
     )
 
 
+def _match_filename(pattern, filename):
+    if not filename:
+        return pattern.match('<unknown>') is not None
+    if filename[0] == '<' and filename[-1] == '>':
+        return pattern.match(filename) is not None
+
+    if sys.platform == 'win32':
+        if filename[-3:].lower() == '.py':
+            filename = filename[:-3]
+        if pattern.match(filename):
+            return True
+        filename = filename.replace('\\', '/')
+    else:
+        filename = filename.removesuffix('.py')
+        if pattern.match(filename):
+            return True
+    filename = filename.removesuffix('/__init__')
+    filename = filename.replace('/', '.')
+    i = 0
+    while True:
+        if pattern.match(filename, i):
+            return True
+        i = filename.find('.', i) + 1
+        if not i:
+            return False
+
+
 def warn_explicit(message, category, filename, lineno,
                   module=None, registry=None, module_globals=None,
                   source=None):
     lineno = int(lineno)
-    if module is None:
-        module = filename or "<unknown>"
-        if module[-3:].lower() == ".py":
-            module = module[:-3] # XXX What about leading pathname?
     if isinstance(message, Warning):
         text = str(message)
         category = message.__class__
     else:
         text = message
         message = category(message)
+    modules = None
     key = (text, category, lineno)
     with _wm._lock:
         if registry is None:
@@ -549,9 +573,11 @@ def warn_explicit(message, category, filename, lineno,
             action, msg, cat, mod, ln = item
             if ((msg is None or msg.match(text)) and
                 issubclass(category, cat) and
-                (mod is None or mod.match(module)) and
-                (ln == 0 or lineno == ln)):
-                break
+                (ln == 0 or lineno == ln) and
+                (mod is None or (_match_filename(mod, filename)
+                                 if module is None else
+                                 mod.match(module)))):
+                    break
         else:
             action = _wm.defaultaction
         # Early exit actions

Lib/test/test_ast/test_ast.py

@@ -13,6 +13,7 @@
 import textwrap
 import types
 import unittest
+import warnings
 import weakref
 from io import StringIO
 from pathlib import Path
@@ -1124,6 +1125,19 @@ def test_tstring(self):
         self.assertIsInstance(tree.body[0].value.values[0], ast.Constant)
         self.assertIsInstance(tree.body[0].value.values[1], ast.Interpolation)
 
+    def test_filter_syntax_warnings_by_module(self):
+        filename = support.findfile('test_warnings/data/syntax_warnings.py')
+        with open(filename, 'rb') as f:
+            source = f.read()
+        with warnings.catch_warnings(record=True) as wlog:
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=r'<unknown>\z')
+            ast.parse(source)
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 21])
+        for wm in wlog:
+            self.assertEqual(wm.filename, '<unknown>')
+            self.assertIs(wm.category, SyntaxWarning)
+
 
 class CopyTests(unittest.TestCase):
     """Test copying and pickling AST nodes."""

Lib/test/test_builtin.py

@@ -1088,6 +1088,28 @@ def four_freevars():
             three_freevars.__globals__,
             closure=my_closure)
 
+    def test_exec_filter_syntax_warnings_by_module(self):
+        filename = support.findfile('test_warnings/data/syntax_warnings.py')
+        with open(filename, 'rb') as f:
+            source = f.read()
+        with warnings.catch_warnings(record=True) as wlog:
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=r'<string>\z')
+            exec(source, {})
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21])
+        for wm in wlog:
+            self.assertEqual(wm.filename, '<string>')
+            self.assertIs(wm.category, SyntaxWarning)
+
+        with warnings.catch_warnings(record=True) as wlog:
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=r'<string>\z')
+            exec(source, {'__name__': 'package.module', '__file__': filename})
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21])
+        for wm in wlog:
+            self.assertEqual(wm.filename, '<string>')
+            self.assertIs(wm.category, SyntaxWarning)
+
 
     def test_filter(self):
         self.assertEqual(list(filter(lambda c: 'a' <= c <= 'z', 'Hello World')), list('elloorld'))

Lib/test/test_cmd_line_script.py

@@ -810,6 +810,12 @@ def test_script_as_dev_fd(self):
                 out, err = p.communicate()
                 self.assertEqual(out, b"12345678912345678912345\n")
 
+    def test_filter_syntax_warnings_by_module(self):
+        filename = support.findfile('test_warnings/data/syntax_warnings.py')
+        rc, out, err = assert_python_ok('-Werror', '-Walways:::test.test_warnings.data.syntax_warnings', filename)
+        self.assertEqual(err.count(b': SyntaxWarning: '), 6)
+        rc, out, err = assert_python_ok('-Werror', '-Walways:::syntax_warnings', filename)
+        self.assertEqual(err.count(b': SyntaxWarning: '), 6)
 
 
 def tearDownModule():

Lib/test/test_compile.py

@@ -6,6 +6,7 @@
 import opcode
 import os
 import unittest
+import re
 import sys
 import ast
 import _ast
@@ -1745,6 +1746,20 @@ def test_compile_warning_in_finally(self):
             self.assertEqual(wm.category, SyntaxWarning)
             self.assertIn("\"is\" with 'int' literal", str(wm.message))
 
+    def test_filter_syntax_warnings_by_module(self):
+        filename = support.findfile('test_warnings/data/syntax_warnings.py')
+        with open(filename, 'rb') as f:
+            source = f.read()
+        module_re = r'test\.test_warnings\.data\.syntax_warnings\z'
+        with warnings.catch_warnings(record=True) as wlog:
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=module_re)
+            compile(source, filename, 'exec')
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21])
+        for wm in wlog:
+            self.assertEqual(wm.filename, filename)
+            self.assertIs(wm.category, SyntaxWarning)
+
 
 class TestBooleanExpression(unittest.TestCase):
     class Value:

Lib/test/test_import/__init__.py

@@ -15,6 +15,7 @@
 import os
 import py_compile
 import random
+import re
 import shutil
 import stat
 import subprocess
@@ -23,6 +24,7 @@
 import threading
 import time
 import types
+import warnings
 import unittest
 from unittest import mock
 import _imp
@@ -43,6 +45,7 @@
     Py_GIL_DISABLED,
     no_rerun,
     force_not_colorized_test_class,
+    findfile
 )
 from test.support.import_helper import (
     forget, make_legacy_pyc, unlink, unload, ready_to_import,
@@ -51,7 +54,7 @@
     TESTFN, rmtree, temp_umask, TESTFN_UNENCODABLE)
 from test.support import script_helper
 from test.support import threading_helper
-from test.test_importlib.util import uncache
+from test.test_importlib.util import uncache, temporary_pycache_prefix
 from types import ModuleType
 try:
     import _testsinglephase
@@ -1250,6 +1253,36 @@ class Spec2:
                 origin = "a\x00b"
             _imp.create_dynamic(Spec2())
 
+    def test_filter_syntax_warnings_by_module(self):
+        filename = findfile('test_warnings/data/syntax_warnings.py')
+        module_re = r'test\.test_warnings\.data\.syntax_warnings\z'
+        unload('test.test_warnings.data.syntax_warnings')
+        with (os_helper.temp_dir() as tmpdir,
+              temporary_pycache_prefix(tmpdir),
+              warnings.catch_warnings(record=True) as wlog):
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=module_re)
+            import test.test_warnings.data.syntax_warnings
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21])
+        filename = test.test_warnings.data.syntax_warnings.__file__
+        for wm in wlog:
+            self.assertEqual(wm.filename, filename)
+            self.assertIs(wm.category, SyntaxWarning)
+
+        module_re = r'syntax_warnings\z'
+        unload('test.test_warnings.data.syntax_warnings')
+        with (os_helper.temp_dir() as tmpdir,
+              temporary_pycache_prefix(tmpdir),
+              warnings.catch_warnings(record=True) as wlog):
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=module_re)
+            import test.test_warnings.data.syntax_warnings
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21])
+        filename = test.test_warnings.data.syntax_warnings.__file__
+        for wm in wlog:
+            self.assertEqual(wm.filename, filename)
+            self.assertIs(wm.category, SyntaxWarning)
+
 
 @skip_if_dont_write_bytecode
 class FilePermissionTests(unittest.TestCase):

Lib/test/test_symtable.py

@@ -5,6 +5,7 @@
 import re
 import textwrap
 import symtable
+import warnings
 import unittest
 
 from test import support
@@ -586,6 +587,20 @@ def test__symtable_refleak(self):
         # check error path when 'compile_type' AC conversion failed
         self.assertRaises(TypeError, symtable.symtable, '', mortal_str, 1)
 
+    def test_filter_syntax_warnings_by_module(self):
+        filename = support.findfile('test_warnings/data/syntax_warnings.py')
+        with open(filename, 'rb') as f:
+            source = f.read()
+        module_re = r'test\.test_warnings\.data\.syntax_warnings\z'
+        with warnings.catch_warnings(record=True) as wlog:
+            warnings.simplefilter('error')
+            warnings.filterwarnings('always', module=module_re)
+            symtable.symtable(source, filename, 'exec')
+        self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10])
+        for wm in wlog:
+            self.assertEqual(wm.filename, filename)
+            self.assertIs(wm.category, SyntaxWarning)
+
 
 class ComprehensionTests(unittest.TestCase):
     def get_identifiers_recursive(self, st, res):

Lib/test/test_warnings/__init__.py

@@ -241,6 +241,77 @@ def test_once(self):
                                     42)
             self.assertEqual(len(w), 0)
 
+    def test_filter_module(self):
+        with self.module.catch_warnings(record=True) as w:
+            self.module.simplefilter('error')
+            self.module.filterwarnings('always', module=r'package\.module\z')
+            self.module.warn_explicit('msg', UserWarning, 'filename', 42,
+                                      module='package.module')
+            self.assertEqual(len(w), 1)
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module', 42)
+            self.assertEqual(len(w), 2)
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module.py', 42)
+            self.assertEqual(len(w), 3)
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module/__init__.py', 42)
+            self.assertEqual(len(w), 4)
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module/__init__', 42)
+            self.assertEqual(len(w), 5)
+            if sys.platform == 'win32':
+                self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module.PY', 42)
+                self.assertEqual(len(w), 6)
+                self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module\__init__.PY', 42)
+                self.assertEqual(len(w), 6)
+
+        with self.module.catch_warnings(record=True) as w:
+            self.module.simplefilter('error')
+            self.module.filterwarnings('always', module='package')
+            self.module.warn_explicit('msg', UserWarning, 'filename', 42,
+                                      module='package.module')
+            self.assertEqual(len(w), 1)
+            with self.assertRaises(UserWarning):
+                self.module.warn_explicit('msg', UserWarning, 'filename', 42,
+                                          module='other.package.module')
+            with self.assertRaises(UserWarning):
+                self.module.warn_explicit('msg', UserWarning, '/path/to/otherpackage/module.py', 42)
+
+        with self.module.catch_warnings(record=True) as w:
+            self.module.simplefilter('error')
+            self.module.filterwarnings('always', module=r'/path/to/package/module\z')
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module', 42)
+            self.assertEqual(len(w), 1)
+            self.module.warn_explicit('msg', UserWarning, '/path/to/package/module.py', 42)
+            self.assertEqual(len(w), 2)
+            with self.assertRaises(UserWarning):
+                self.module.warn_explicit('msg', UserWarning, '/PATH/TO/PACKAGE/MODULE', 42)
+            if sys.platform == 'win32':
+                self.module.warn_explicit('msg', UserWarning, r'/path/to/package/module.PY', 42)
+                self.assertEqual(len(w), 3)
+                with self.assertRaises(UserWarning):
+                    self.module.warn_explicit('msg', UserWarning, r'\path\to\package\module', 42)
+
+        if sys.platform == 'win32':
+            with self.module.catch_warnings(record=True) as w:
+                self.module.simplefilter('error')
+                self.module.filterwarnings('always', module=r'C:\\path\\to\\package\\module\z')
+                self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module', 42)
+                self.assertEqual(len(w), 1)
+                self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module.py', 42)
+                self.assertEqual(len(w), 2)
+                self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module.PY', 42)
+                self.assertEqual(len(w), 3)
+                with self.assertRaises(UserWarning):
+                    self.module.warn_explicit('msg', UserWarning, r'C:\PATH\TO\PACKAGE\MODULE', 42)
+                with self.assertRaises(UserWarning):
+                    self.module.warn_explicit('msg', UserWarning, r'C:/path/to/package/module', 42)
+                with self.assertRaises(UserWarning):
+                    self.module.warn_explicit('msg', UserWarning, r'C:\path\to\package\module\__init__', 42)
+
+        with self.module.catch_warnings(record=True) as w:
+            self.module.simplefilter('error')
+            self.module.filterwarnings('always', module=r'<unknown>\z')
+            self.module.warn_explicit('msg', UserWarning, '', 42)
+            self.assertEqual(len(w), 1)
+
     def test_module_globals(self):
         with self.module.catch_warnings(record=True) as w:
             self.module.simplefilter("always", UserWarning)
@@ -320,7 +391,7 @@ def test_message_matching(self):
 
     def test_mutate_filter_list(self):
         class X:
-            def match(self, a):
+            def match(self, a, start=0):
                 L[:] = []
 
         L = [("default",X(),UserWarning,X(),0) for i in range(2)]

Lib/test/test_warnings/data/syntax_warnings.py

@@ -0,0 +1,21 @@
+# Syntax warnings emitted in different parts of the Python compiler.
+
+# Parser/lexer/lexer.c
+x = 1or 0  # line 4
+
+# Parser/tokenizer/helpers.c
+'\z'  # line 7
+
+# Parser/string_parser.c
+'\400'  # line 10
+
+# _PyCompile_Warn() in Python/codegen.c
+assert(x, 'message')  # line 13
+x is 1  # line 14
+
+# _PyErr_EmitSyntaxWarning() in Python/ast_preprocess.c
+def f():
+    try:
+        pass
+    finally:
+        return 42  # line 21