test: improve path normalization tests with parametrize and edge cases

Author: AlePouroullis

src/humanloop/path_utils.py

@@ -3,13 +3,17 @@
 
 def normalize_path(path: str, strip_extension: bool = False) -> str:
     """Normalize a path to the standard API format: path/to/resource, where resource can be a file or a directory."""
-    # Remove leading/trailing slashes
-    path_obj = Path(path.strip("/"))
+    # Handle backslashes for Windows paths before passing to PurePosixPath
+    # This is needed because some backslash sequences are treated as escape chars
+    path = path.replace("\\", "/")
 
-    # Handle extension
+    # Use PurePosixPath to normalize the path (handles consecutive slashes)
+    path_obj = Path(path)
+
+    # Strip extension if requested
     if strip_extension:
         path_obj = path_obj.with_suffix("")
 
-    # Convert path to string with forward slashes, regardless of OS
-    # This ensures consistent path format (e.g., "path/to/resource") across Windows/Unix
-    return path_obj.as_posix()
+    # Convert to string and remove any leading/trailing slashes
+    # We use the path as a string and not as_posix() since we've already normalized separators
+    return str(path_obj).strip("/")

tests/custom/unit/test_path_utils.py

@@ -1,28 +1,38 @@
+import pytest
+
 from humanloop import path_utils
 
 
-def test_normalize_path():
-    """Test path normalization functionality."""
-    # GIVEN various file paths with different formats
-    test_cases = [
-        # Input path, expected with strip_extension=False, expected with strip_extension=True
+@pytest.mark.parametrize(
+    "input_path, expected_with_extension, expected_without_extension",
+    [
+        # Basic cases
         ("path/to/file.prompt", "path/to/file.prompt", "path/to/file"),
         ("path\\to\\file.agent", "path/to/file.agent", "path/to/file"),
         ("/leading/slashes/file.prompt", "leading/slashes/file.prompt", "leading/slashes/file"),
         ("trailing/slashes/file.agent/", "trailing/slashes/file.agent", "trailing/slashes/file"),
         ("multiple//slashes//file.prompt", "multiple/slashes/file.prompt", "multiple/slashes/file"),
-    ]
-
-    # Test with strip_extension=False (default)
-    for input_path, expected_with_ext, _ in test_cases:
-        # WHEN they are normalized without stripping extension
-        normalized = path_utils.normalize_path(input_path, strip_extension=False)
-        # THEN they should be converted to the expected format with extension
-        assert normalized == expected_with_ext
+        # Edge cases
+        ("path/to/file with spaces.prompt", "path/to/file with spaces.prompt", "path/to/file with spaces"),
+        (
+            "path/to/file\\with\\backslashes.prompt",
+            "path/to/file/with/backslashes.prompt",
+            "path/to/file/with/backslashes",
+        ),
+        ("path/to/unicode/文件.prompt", "path/to/unicode/文件.prompt", "path/to/unicode/文件"),
+        (
+            "path/to/special/chars/!@#$%^&*().prompt",
+            "path/to/special/chars/!@#$%^&*().prompt",
+            "path/to/special/chars/!@#$%^&*()",
+        ),
+    ],
+)
+def test_normalize_path(input_path, expected_with_extension, expected_without_extension):
+    """Test path normalization with various path formats."""
+    # Test without stripping extension
+    normalized = path_utils.normalize_path(input_path, strip_extension=False)
+    assert normalized == expected_with_extension, f"Failed with strip_extension=False for '{input_path}'"
 
-    # Test with strip_extension=True
-    for input_path, _, expected_without_ext in test_cases:
-        # WHEN they are normalized with extension stripping
-        normalized = path_utils.normalize_path(input_path, strip_extension=True)
-        # THEN they should be converted to the expected format without extension
-        assert normalized == expected_without_ext
+    # Test with extension stripping
+    normalized = path_utils.normalize_path(input_path, strip_extension=True)
+    assert normalized == expected_without_extension, f"Failed with strip_extension=True for '{input_path}'"