gh-158986: Fix profiling.sampling module type for nested site-packages

extract_module_name() checked the stdlib directory first, so files in a
site-packages directory inside it were reported as stdlib modules. Use
the most specific matching directory instead.

Author: sricursion

Lib/profiling/sampling/module_utils.py

@@ -65,14 +65,17 @@ def extract_module_name(filename, path_info):
     except (ValueError, OSError):
         return (str(filename), 'other')
 
-    # Check if it's in stdlib
-    if path_info['stdlib'] and file_path.is_relative_to(path_info['stdlib']):
-        return (_path_to_module(file_path.relative_to(path_info['stdlib'])), 'stdlib')
-
-    # Check site-packages
-    for site_pkg in path_info['site_packages']:
-        if file_path.is_relative_to(site_pkg):
-            return (_path_to_module(file_path.relative_to(site_pkg)), 'site-packages')
+    # Check stdlib and site-packages, preferring the most specific directory
+    # because site-packages is usually located inside the stdlib directory
+    bases = [(path_info['stdlib'], 'stdlib')]
+    bases.extend((site_pkg, 'site-packages')
+                 for site_pkg in path_info['site_packages'])
+    matches = [(base, module_type) for base, module_type in bases
+               if base and file_path.is_relative_to(base)]
+    if matches:
+        # max() returns the first of equal candidates, so stdlib wins ties
+        base, module_type = max(matches, key=lambda m: len(m[0].parts))
+        return (_path_to_module(file_path.relative_to(base)), module_type)
 
     # Check other sys.path entries (project files)
     if not str(file_path).startswith(('<', '[')):  # Skip special files

Lib/test/test_profiling/test_heatmap.py

@@ -86,6 +86,25 @@ def test_extract_module_name_with_stdlib_file(self):
             self.assertEqual(module_type, 'stdlib')
             self.assertIn('os', module_name)
 
+    def test_extract_module_name_with_site_packages_inside_stdlib(self):
+        """Test extract_module_name when site-packages is inside stdlib."""
+        stdlib = Path('/usr/lib/python3.15')
+        site_packages = stdlib / 'site-packages'
+        # On Windows, site.getsitepackages() also includes sys.prefix
+        path_info = {
+            'stdlib': stdlib,
+            'site_packages': [Path('/usr'), site_packages],
+            'sys_path': [],
+        }
+
+        site_packages_file = str(site_packages / 'requests' / 'api.py')
+        self.assertEqual(extract_module_name(site_packages_file, path_info),
+                         ('requests.api', 'site-packages'))
+
+        stdlib_file = str(stdlib / 'json' / 'decoder.py')
+        self.assertEqual(extract_module_name(stdlib_file, path_info),
+                         ('json.decoder', 'stdlib'))
+
     def test_extract_module_name_with_project_file(self):
         """Test extract_module_name with a project file."""
         path_info = get_python_path_info()

Misc/NEWS.d/next/Library/2026-10-08-01-33-46.gh-issue-158986.iXoUCo.rst

@@ -0,0 +1,4 @@
+Fix :mod:`profiling.sampling` reporting modules in a ``site-packages``
+directory located inside the standard library directory as standard library
+modules. The heatmap now lists them under "Site Packages" and flamegraph
+labels no longer start with ``site-packages.``.