Improve solacc script

tests/integration/source_code/solacc-malformed.txt

@@ -0,0 +1,11 @@
+redkite-pricing/01_Data_Prep_Analysis.py
+user-behavior-analytics-for-cloud-services/Anomaly Detection Pipeline/0 - Ingest and Explore Raw User Data.py
+user-behavior-analytics-for-cloud-services/Anomaly Detection Pipeline/Full Data Engineering Pipeline/Train Anomaly Model on Historical Data.py
+user-behavior-analytics-for-cloud-services/Anomaly Detection Pipeline/Full Data Engineering Pipeline/Run Inference on Active Batch.py
+product-er-with-images/01_Prep_Data.py
+fsi-mrm-generation/templates/Credit Adjudication - Example.py
+lr-recommender/03_Train_Model.py
+dist/multi-touch-attribution/04_markov_chains.py
+security-analysis-tool/notebooks/Utils/workspace_bootstrap.py
+# the following fails to parse due to module dependency azure/devops/v5_1/client_factory.py
+edge-ml-for-manufacturing/03_Trigger_AzureDevOps_Job.py

tests/integration/source_code/solacc.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import sys
 from pathlib import Path
 
@@ -44,16 +45,27 @@ def clone_all():
         run_command(f'git clone {url} {dst}')
 
 
-def lint_all():
+def lint_all(file_to_lint: str | None):
     # pylint: disable=too-many-nested-blocks
     ws = WorkspaceClient(host='...', token='...')
     ctx = LocalCheckoutContext(ws).replace(
         linter_context_factory=lambda session_state: LinterContext(MigrationIndex([]), session_state)
     )
     parseable = 0
     missing_imports = 0
-    all_files = list(dist.glob('**/*.py'))
+    all_files = list(dist.glob('**/*.py')) if file_to_lint is None else [Path(dist, file_to_lint)]
+    if file_to_lint is None:
+        unparsed = Path(Path(__file__).parent, "solacc-unparsed.txt")
+        if unparsed.exists():
+            os.remove(unparsed)
+    skipped: set[str] | None = None
+    malformed = Path(Path(__file__).parent, "solacc-malformed.txt")
+    if file_to_lint is None and malformed.exists():
+        text = malformed.read_text()
+        skipped = set(text.split("\n"))
     for file in all_files:
+        if skipped and file.relative_to(dist).as_posix() in skipped:
+            continue
         try:
             for located_advice in ctx.local_code_linter.lint_path(file):
                 if located_advice.advice.code == 'import-not-found':
@@ -63,17 +75,34 @@ def lint_all():
             parseable += 1
         except Exception as e:  # pylint: disable=broad-except
             # here we're most likely catching astroid & sqlglot errors
-            logger.error(f"Error during parsing of {file}: {e}".replace("\n", " "), exc_info=e)
+            if file_to_lint:
+                logger.error(f"Error during parsing of {file}: {e}".replace("\n", " "), exc_info=e)
+            else:
+                logger.error(f"Error during parsing of {file}: {e}".replace("\n", " "))
+            if file_to_lint is None:
+                # populate solacc-unparsed.txt
+                with unparsed.open(mode="a") as f:
+                    f.write(file.relative_to(dist).as_posix())
+                    f.write("\n")
     parseable_pct = int(parseable / len(all_files) * 100)
-    logger.info(f"Parseable: {parseable_pct}% ({parseable}/{len(all_files)}), missing imports: {missing_imports}")
+    logger.info(
+        f"Skipped: {len(skipped or [])}, parseable: {parseable_pct}% ({parseable}/{len(all_files)}), missing imports: {missing_imports}"
+    )
     if parseable_pct < 100:
         sys.exit(1)
 
 
-if __name__ == "__main__":
+def main(args: list[str]):
     install_logger()
     logging.root.setLevel(logging.INFO)
-    logger.info("Cloning...")
-    clone_all()
+    file_to_lint = args[1] if len(args) > 1 else None
+    if not file_to_lint:
+        # don't clone if linting just one file, assumption is we're troubleshooting
+        logger.info("Cloning...")
+        clone_all()
     logger.info("Linting...")
-    lint_all()
+    lint_all(file_to_lint)
+
+
+if __name__ == "__main__":
+    main(sys.argv)