fixing repo issues

Author: Saumya-R

scripts/workflow/parse_repos.py

@@ -63,7 +63,7 @@ def parse_known_good(json_file="./known_good.json"):
 
     for name, module in modules.items():
         repo_url = module.repo
-        ref = module.version or module.branch or module.commit_hash
+        ref = module.version or module.branch or module.hash
 
         repo_obj = {"name": name, "url": repo_url, "version": ref, "path": f"repos/{name}"}
         repos.append(repo_obj)
@@ -74,8 +74,8 @@ def parse_known_good(json_file="./known_good.json"):
             module_outputs[f"{name}_version"] = module.version
         if module.branch:
             module_outputs[f"{name}_branch"] = module.branch
-        if module.commit_hash:
-            module_outputs[f"{name}_hash"] = module.commit_hash
+        if module.hash:
+            module_outputs[f"{name}_hash"] = module.hash
 
     return repos, len(modules), module_outputs
 

scripts/workflow/recategorize_guidelines.py

@@ -32,20 +32,45 @@
 
 
 def validate_paths():
-    """Validate that required files and directories exist."""
-    required_files = [
+    """
+    Validate that required files exist.
+
+    Note: Only validates files needed for recategorization if SARIF exists.
+    Returns:
+        True if validation passes or SARIF doesn't exist, False on critical errors
+    """
+    # First check if SARIF file exists - if not, nothing to recategorize
+    if not Path(SARIF_FILE).exists():
+        print(f"Info: SARIF file not found at {SARIF_FILE}", file=sys.stderr)
+        return False  # Signal to skip recategorization
+
+    # SARIF exists, check for recategorization dependencies
+    optional_files = [
         RECATEGORIZE_SCRIPT,
-        CODING_STANDARDS_CONFIG,
         CODING_STANDARDS_SCHEMA,
         SARIF_SCHEMA,
-        SARIF_FILE,
     ]
 
+    required_files = [
+        CODING_STANDARDS_CONFIG,
+    ]
+
+    # Check required files (fail if missing)
     for file_path in required_files:
         if not Path(file_path).exists():
             print(f"Error: Required file not found: {file_path}", file=sys.stderr)
             return False
 
+    # Warn about optional files but don't fail
+    missing_optional = []
+    for file_path in optional_files:
+        if not Path(file_path).exists():
+            missing_optional.append(file_path)
+
+    if missing_optional:
+        print(f"Warning: Some recategorization files not found: {missing_optional}", file=sys.stderr)
+        print("Recategorization will be skipped, but filtering will still be applied.", file=sys.stderr)
+
     return True
 
 
@@ -54,8 +79,14 @@ def recategorize_sarif():
     Run the CodeQL recategorization script on SARIF results.
 
     Returns:
-        True if successful, False otherwise
+        True if successful or skipped, False on critical errors
     """
+    # Check if recategorization script exists
+    if not Path(RECATEGORIZE_SCRIPT).exists():
+        print(f"Info: Recategorization script not found at {RECATEGORIZE_SCRIPT}", file=sys.stderr)
+        print("Skipping recategorization step (will apply filtering only).", file=sys.stderr)
+        return True  # Not a failure, just skip this step
+
     # Create output directory
     output_path = Path(OUTPUT_DIR)
     output_path.mkdir(parents=True, exist_ok=True)
@@ -172,12 +203,17 @@ def filter_sarif_results():
 def main():
     """Main entry point."""
     # Validate required files exist
-    if not validate_paths():
-        sys.exit(1)
+    has_sarif = validate_paths()
 
-    # Run recategorization
+    if not has_sarif:
+        # No SARIF file to process - this is normal before CodeQL analysis runs
+        print("No SARIF file found - skipping recategorization.")
+        print("This is expected if CodeQL analysis hasn't completed yet.")
+        sys.exit(0)
+
+    # Run recategorization (will skip gracefully if script not available)
     if not recategorize_sarif():
-        sys.exit(1)
+        print("Warning: Recategorization failed, continuing with filtering...", file=sys.stderr)
 
     # Filter SARIF results to only include repos/*
     if not filter_sarif_results():