gh-132933: zipapp - apply the filter when creating the list of files to add (gh-132934)

Author: Jolmberg

Lib/test/test_zipapp.py

@@ -113,6 +113,19 @@ def test_target_overwrites_source_file(self):
         with self.assertRaises(zipapp.ZipAppError):
             zipapp.create_archive(source, target)
 
+    def test_target_overwrites_filtered_source_file(self):
+        # If there's a filter that excludes the target,
+        # the overwrite check shouldn't trigger.
+        source = self.tmpdir
+        (source / '__main__.py').touch()
+        target = source / 'target.pyz'
+        target.touch()
+        pyz_filter = lambda p: not p.match('*.pyz')
+        zipapp.create_archive(source, target, filter=pyz_filter)
+        with zipfile.ZipFile(target, 'r') as z:
+            self.assertEqual(len(z.namelist()), 1)
+            self.assertIn('__main__.py', z.namelist())
+
     def test_create_archive_filter_exclude_dir(self):
         # Test packing a directory and using a filter to exclude a
         # subdirectory (ensures that the path supplied to include

Lib/zipapp.py

@@ -134,7 +134,11 @@ def create_archive(source, target=None, interpreter=None, main=None,
     # Create the list of files to add to the archive now, in case
     # the target is being created in the source directory - we
     # don't want the target being added to itself
-    files_to_add = sorted(source.rglob('*'))
+    files_to_add = {}
+    for path in sorted(source.rglob('*')):
+        relative_path = path.relative_to(source)
+        if filter is None or filter(relative_path):
+            files_to_add[path] = relative_path
 
     # The target cannot be in the list of files to add. If it were, we'd
     # end up overwriting the source file and writing the archive into
@@ -159,10 +163,8 @@ def create_archive(source, target=None, interpreter=None, main=None,
         compression = (zipfile.ZIP_DEFLATED if compressed else
                        zipfile.ZIP_STORED)
         with zipfile.ZipFile(fd, 'w', compression=compression) as z:
-            for child in files_to_add:
-                arcname = child.relative_to(source)
-                if filter is None or filter(arcname):
-                    z.write(child, arcname.as_posix())
+            for path, relative_path in files_to_add.items():
+                z.write(path, relative_path.as_posix())
             if main_py:
                 z.writestr('__main__.py', main_py.encode('utf-8'))
 

Misc/NEWS.d/next/Library/2025-04-25-21-41-45.gh-issue-132933.yO3ySJ.rst

@@ -0,0 +1 @@
+The zipapp module now applies the filter when creating the list of files to add, rather than waiting until the file is being added to the archive.