fix: normalize complete inventory paths

Author: renecannao

scripts/pg_compat/extract_statements.py

@@ -258,6 +258,13 @@ def _jsonl_text(rows):
     return text
 
 
+def _portable_path_key(path):
+    return tuple(
+        unicodedata.normalize("NFC", component).casefold()
+        for component in path.parts
+    )
+
+
 def _validate_distinct_paths(input_path, inventory_path, diagnostics_path):
     named_paths = (
         ("--input", Path(input_path)),
@@ -283,27 +290,9 @@ def _validate_distinct_paths(input_path, inventory_path, diagnostics_path):
                     aliases = False
 
             if not aliases:
-                left_parent = left_resolved.parent
-                right_parent = right_resolved.parent
-                same_parent = left_parent == right_parent
-                if not same_parent:
-                    try:
-                        same_parent = os.path.samefile(
-                            left_parent,
-                            right_parent,
-                        )
-                    except OSError:
-                        same_parent = False
-
-                left_name = unicodedata.normalize(
-                    "NFC",
-                    left_resolved.name,
-                ).casefold()
-                right_name = unicodedata.normalize(
-                    "NFC",
-                    right_resolved.name,
-                ).casefold()
-                aliases = same_parent and left_name == right_name
+                aliases = _portable_path_key(
+                    left_resolved
+                ) == _portable_path_key(right_resolved)
 
             if aliases:
                 raise ValueError(

tests/pg_compat/test_extract_statements.py

@@ -524,6 +524,50 @@ def test_rejects_nonexistent_output_names_with_equivalent_unicode(self):
             self.assertFalse(inventory_path.exists())
             self.assertFalse(diagnostics_path.exists())
 
+    def test_rejects_nonexistent_parent_names_differing_only_by_case(self):
+        with tempfile.TemporaryDirectory() as directory:
+            directory = Path(directory)
+            input_path = directory / "raw.jsonl"
+            inventory_path = directory / "Reports" / "output.jsonl"
+            diagnostics_path = directory / "reports" / "output.jsonl"
+            input_path.write_text("", encoding="utf-8")
+
+            result = self.run_cli(
+                "--input",
+                input_path,
+                "--inventory",
+                inventory_path,
+                "--diagnostics",
+                diagnostics_path,
+            )
+
+            self.assertNotEqual(result.returncode, 0)
+            self.assertIn("distinct files", result.stderr)
+            self.assertFalse(inventory_path.parent.exists())
+            self.assertFalse(diagnostics_path.parent.exists())
+
+    def test_rejects_nonexistent_parent_names_with_equivalent_unicode(self):
+        with tempfile.TemporaryDirectory() as directory:
+            directory = Path(directory)
+            input_path = directory / "raw.jsonl"
+            inventory_path = directory / "caf\u00e9" / "output.jsonl"
+            diagnostics_path = directory / "cafe\u0301" / "output.jsonl"
+            input_path.write_text("", encoding="utf-8")
+
+            result = self.run_cli(
+                "--input",
+                input_path,
+                "--inventory",
+                inventory_path,
+                "--diagnostics",
+                diagnostics_path,
+            )
+
+            self.assertNotEqual(result.returncode, 0)
+            self.assertIn("distinct files", result.stderr)
+            self.assertFalse(inventory_path.parent.exists())
+            self.assertFalse(diagnostics_path.parent.exists())
+
     @unittest.skipUnless(hasattr(os, "symlink"), "symlinks are unavailable")
     def test_rejects_existing_symlink_path_aliases(self):
         with tempfile.TemporaryDirectory() as directory: