Improve wildcard matching to use glob-like semantics

Co-authored-by: joocer <[email protected]>

Author: Copilot

opteryx/utils/paths.py

@@ -106,13 +106,39 @@ def split_wildcard_path(path: str):
 
 def match_wildcard(pattern: str, path: str) -> bool:
     """
-    Match a path against a wildcard pattern.
+    Match a path against a wildcard pattern using glob-like semantics.
+    
+    Unlike fnmatch, this function treats path separators specially:
+    - '*' matches any characters EXCEPT path separators
+    - '?' matches any single character EXCEPT path separators
+    - Use '**' to match across directory boundaries (not yet supported)
+    
+    This ensures consistent behavior with glob.glob() used for local files.
     
     Args:
         pattern: Pattern with wildcards (e.g., "bucket/path/*.parquet")
         path: Path to match (e.g., "bucket/path/file1.parquet")
         
     Returns:
         True if path matches pattern
+        
+    Examples:
+        >>> match_wildcard("bucket/path/*.parquet", "bucket/path/file.parquet")
+        True
+        >>> match_wildcard("bucket/path/*.parquet", "bucket/path/sub/file.parquet")
+        False
     """
-    return fnmatch.fnmatch(path, pattern)
+    # Split pattern and path into parts
+    pattern_parts = pattern.split(OS_SEP)
+    path_parts = path.split(OS_SEP)
+    
+    # Must have same number of parts for a match (unless using ** which we don't support yet)
+    if len(pattern_parts) != len(path_parts):
+        return False
+    
+    # Match each part using fnmatch
+    for pattern_part, path_part in zip(pattern_parts, path_parts):
+        if not fnmatch.fnmatch(path_part, pattern_part):
+            return False
+    
+    return True

tests/unit/utils/test_paths_wildcards.py

@@ -55,13 +55,16 @@ def test_split_wildcard_path():
 
 
 def test_match_wildcard():
-    """Test wildcard pattern matching"""
-    # Asterisk matches multiple characters
+    """Test wildcard pattern matching with glob-like semantics"""
+    # Asterisk matches multiple characters (but not path separators)
     assert paths.match_wildcard("bucket/path/*.parquet", "bucket/path/file1.parquet") is True
     assert paths.match_wildcard("bucket/path/*.parquet", "bucket/path/file2.parquet") is True
     assert paths.match_wildcard("bucket/path/*.parquet", "bucket/path/data.csv") is False
 
-    # Question mark matches single character
+    # Asterisk does NOT match across directory boundaries (glob-like behavior)
+    assert paths.match_wildcard("bucket/path/*.parquet", "bucket/path/subdir/file.parquet") is False
+    
+    # Question mark matches single character (but not path separators)
     assert paths.match_wildcard("bucket/path/file?.parquet", "bucket/path/file1.parquet") is True
     assert paths.match_wildcard("bucket/path/file?.parquet", "bucket/path/file2.parquet") is True
     assert paths.match_wildcard("bucket/path/file?.parquet", "bucket/path/file10.parquet") is False
@@ -71,6 +74,10 @@ def test_match_wildcard():
     assert paths.match_wildcard("bucket/path/file[0-9].parquet", "bucket/path/file5.parquet") is True
     assert paths.match_wildcard("bucket/path/file[0-9].parquet", "bucket/path/fileA.parquet") is False
 
+    # Wildcard in middle of path
+    assert paths.match_wildcard("bucket/*/data.parquet", "bucket/subdir/data.parquet") is True
+    assert paths.match_wildcard("bucket/*/data.parquet", "bucket/a/b/data.parquet") is False
+    
     # No wildcards - exact match
     assert paths.match_wildcard("bucket/path/data.parquet", "bucket/path/data.parquet") is True
     assert paths.match_wildcard("bucket/path/data.parquet", "bucket/path/other.parquet") is False