Commit e70e1df2 authored by Jan Reimes's avatar Jan Reimes
Browse files

fix(02-checkout-graph-deprecation-config-01): improve empty folder detection 

- Check recursively for files, excluding .ai/ subfolder
- Log folder contents for debugging when empty
- Triggers re-download when only .ai/ processing outputs remain
parent 28ddf849

packages/3gpp-ai/threegpp_ai/operations/workspaces.py

+11 −2
Original line number Diff line number Diff line
@@ -459,11 +459,20 @@ def _checkout_tdoc_if_needed(tdoc_id: str, metadata: TDocMetadata, checkout_base 
    existing_path = resolve_tdoc_checkout_path(tdoc_id, checkout_base)

    if existing_path:

        # Check if folder has actual document files (not just .ai subfolder)

        has_files = any(f.is_file() for f in existing_path.iterdir())

        # Recursively check for files, excluding the .ai/ subfolder which contains processing outputs

        has_files = any(

            f.is_file() and f.parent.name != ".ai" and not any(part == ".ai" for part in f.relative_to(existing_path).parts)

            for f in existing_path.rglob("*")

        )

        if has_files:

            _logger.info(f"TDoc {tdoc_id} already checked out at {existing_path}")

            return existing_path

        _logger.info(f"TDoc {tdoc_id} folder exists but is empty at {existing_path}, re-downloading...")

        

        # Log what was found for debugging

        all_items = list(existing_path.iterdir())

        _logger.info(

            f"TDoc {tdoc_id} folder exists but has no document files at {existing_path} (found: {len(all_items)} items: {[i.name for i in all_items]}), re-downloading..."

        )



    # Checkout the TDoc - returns Path directly or raises exception

    try: