feat(version-scanner): support globbing and subpath patterns in ignore file

Author: chalmerlowe

scripts/version_scanner/tests/unit/test_version_scanner.py

@@ -364,6 +364,53 @@ def test_scan_repository_ignores_version_scanner(tmp_path):
     assert len(results) == 0
 
 
+def test_scan_repository_wildcard_ignores(tmp_path):
+    # Create files
+    (tmp_path / "test.jpg").write_text("dummy version 3.7\n")
+    (tmp_path / "test.py").write_text("python_requires = '>=3.7'\n")
+    
+    rules = [
+        {"name": "python_requires_check", "pattern": "python_requires\\s*=\\s*['\"]>=3\\.7['\"]"},
+        {"name": "explicit_version_string", "pattern": "3\\.7"}
+    ]
+    
+    from version_scanner import scan_repository
+    # Without ignore
+    results = scan_repository(str(tmp_path), rules)
+    assert len(results) >= 2
+    
+    # With wildcard ignore for *.jpg
+    results_ignored = scan_repository(str(tmp_path), rules, ignore_dirs=["*.jpg"])
+    # test.jpg should be ignored completely
+    for match in results_ignored:
+        assert not match["file_path"].endswith("test.jpg")
+
+
+def test__should_ignore():
+    from version_scanner import _should_ignore
+    
+    ignore_patterns = [
+        ".git",
+        "*.jpg",
+        "packages/pkg_a/.nox",
+        "*.egg-info"
+    ]
+    
+    # Exact match
+    assert _should_ignore(".git", ".git", ignore_patterns) is True
+    # Case insensitivity
+    assert _should_ignore(".GIT", ".GIT", ignore_patterns) is True
+    # Wildcard match
+    assert _should_ignore("some/path/image.jpg", "image.jpg", ignore_patterns) is True
+    assert _should_ignore("image.JPG", "image.JPG", ignore_patterns) is True
+    # Subpath match
+    assert _should_ignore("packages/pkg_a/.nox", ".nox", ignore_patterns) is True
+    # Wildcard directory match
+    assert _should_ignore("google_cloud_pubsub.egg-info", "google_cloud_pubsub.egg-info", ignore_patterns) is True
+    # Negative match
+    assert _should_ignore("setup.py", "setup.py", ignore_patterns) is False
+
+
 def test_load_ignore_file(tmp_path):
     from version_scanner import load_ignore_file
 

scripts/version_scanner/version_scanner.py

@@ -20,6 +20,7 @@
 import argparse
 import csv
 import datetime
+import fnmatch
 import os
 import re
 import sys
@@ -498,6 +499,28 @@ def read_package_file(file_path: str) -> List[str]:
     return packages
 
 
+def _should_ignore(rel_path: str, name: str, ignore_patterns: List[str]) -> bool:
+    """Check if a file or directory matches any of the ignore patterns."""
+    if not ignore_patterns:
+        return False
+    name_lower = name.lower()
+    rel_path_norm = rel_path.replace(os.sep, '/').lower()
+    
+    for pattern in ignore_patterns:
+        pattern_lower = pattern.lower()
+        if '/' in pattern:
+            if pattern_lower.startswith('/'):
+                p = pattern_lower[1:]
+            else:
+                p = pattern_lower
+            if fnmatch.fnmatchcase(rel_path_norm, p) or fnmatch.fnmatchcase(rel_path_norm, f"*/{p}"):
+                return True
+        else:
+            if fnmatch.fnmatchcase(name_lower, pattern_lower):
+                return True
+    return False
+
+
 def scan_repository(
     root_path: str,
     rules: List[Dict[str, Any]],
@@ -528,7 +551,6 @@ def scan_repository(
     Returns:
         A list of dictionaries detailing each match.
     """
-    ignore_lower = {i.lower() for i in ignore_dirs} if ignore_dirs else set()
     results = []
 
     filename_targets = []
@@ -557,13 +579,23 @@ def scan_repository(
         print(f"Filtering for packages: {target_packages}")
 
     for root, dirs, files in os.walk(root_path):
+        rel_root = os.path.relpath(root, root_path)
+        
+        # Helper to construct relative path for ignore matching
+        def get_rel_path(name):
+            return name if rel_root == "." else os.path.join(rel_root, name)
+            
         # Prune ignore directories (case-insensitive)
-        dirs[:] = [d for d in dirs if d.lower() not in ignore_lower]
+        dirs[:] = [
+            d for d in dirs 
+            if not _should_ignore(get_rel_path(d), d, ignore_dirs)
+        ]
 
         # Filter ignore files (case-insensitive)
-        files = [f for f in files if f.lower() not in ignore_lower]
-        
-        rel_root = os.path.relpath(root, root_path)
+        files = [
+            f for f in files 
+            if not _should_ignore(get_rel_path(f), f, ignore_dirs)
+        ]
 
         # Layout-agnostic generic subdirectory filtering
         if target_packages: