adding filter for coding guidelines files

Author: Saumya-R

scripts/workflow/recategorize_guidelines.py

@@ -15,6 +15,8 @@
 Recategorize CodeQL SARIF results according to coding standards.
 """
 
+import json
+import re
 import subprocess
 import sys
 from pathlib import Path
@@ -108,6 +110,65 @@ def recategorize_sarif():
         return False
 
 
+def filter_sarif_results():
+    """
+    Filter SARIF results to only include entries with paths matching repos/*.
+
+    Returns:
+        True if successful, False otherwise
+    """
+    sarif_path = Path(SARIF_FILE)
+
+    if not sarif_path.exists():
+        print(f"Warning: SARIF file not found: {SARIF_FILE}", file=sys.stderr)
+        return False
+
+    try:
+        # Load SARIF file
+        with open(sarif_path, "r") as f:
+            sarif_data = json.load(f)
+
+        print("Filtering SARIF results to only include entries with paths matching repos/* ...")
+
+        # Filter runs and results
+        if "runs" in sarif_data:
+            for run in sarif_data["runs"]:
+                if "results" in run:
+                    filtered_results = []
+
+                    for result in run["results"]:
+                        # Check if result has locations
+                        locations = result.get("locations", [])
+                        if not locations:
+                            continue
+
+                        # Check if first location URI matches repos/ pattern
+                        first_location = locations[0].get("physicalLocation", {})
+                        artifact_uri = first_location.get("artifactLocation", {}).get("uri", "")
+
+                        # Pattern: (^|/)repos/ - matches repos/ at start or after a /
+                        if artifact_uri and re.search(r"(^|/)repos/", artifact_uri):
+                            filtered_results.append(result)
+
+                    # Update results with filtered list
+                    run["results"] = filtered_results
+                    print(
+                        f"Run '{run.get('tool', {}).get('driver', {}).get('name', 'unknown')}' "
+                        f"now has {len(filtered_results)} results"
+                    )
+
+        # Write filtered SARIF back to file
+        with open(sarif_path, "w") as f:
+            json.dump(sarif_data, f, indent=2)
+
+        print(f"Filtered SARIF written to {SARIF_FILE}")
+        return True
+
+    except (json.JSONDecodeError, IOError) as e:
+        print(f"Error: Failed to filter SARIF file: {e}", file=sys.stderr)
+        return False
+
+
 def main():
     """Main entry point."""
     # Validate required files exist
@@ -118,6 +179,10 @@ def main():
     if not recategorize_sarif():
         sys.exit(1)
 
+    # Filter SARIF results to only include repos/*
+    if not filter_sarif_results():
+        sys.exit(1)
+
     print("Recategorization workflow completed successfully")
     sys.exit(0)