Commit a836926a authored by Jan Reimes's avatar Jan Reimes
Browse files

♻️ refactor(workspace): extract _process_member and show doc ID in progress bar 

- Extract per-member processing into _process_member() helper
  (reduces workspace_process statement count below PLR0915 limit)
- Progress bar now shows current document ID being processed
- Rename local variable manager→cache_manager for clarity
parent 2071de0f

src/tdoc_crawler/cli/_workspace_commands.py

+51 −25
Original line number Diff line number Diff line
@@ -243,6 +243,44 @@ def _coerce_profile(value: str) -> ExtractionProfile | None: 
        return None





def _process_member(

    member: Any,

    wiki_source_dir_base: Path,

    extraction_profile: ExtractionProfile,

    force: bool,

    md_yaml_frontmatter: bool,

) -> tuple[str, bool, bool]:

    """Process a single workspace member.



    Returns:

        Tuple of (source_id, succeeded, failed).

    """

    source_id = member.source_item_id

    wiki_source_dir = wiki_source_dir_base / source_id

    wiki_source_dir.mkdir(parents=True, exist_ok=True)



    try:

        result_path = convert_for_wiki(

            document_id=source_id,

            wiki_source_dir=wiki_source_dir,

            source_kind=member.source_kind,

            profile=extraction_profile,

            force=force,

            release=member.release,

            md_yaml_frontmatter=md_yaml_frontmatter,

        )

        if result_path:

            suffix = result_path.suffix.lstrip(".")

            logger.debug("%s [%s] → %s", source_id, extraction_profile.value, suffix)

            return source_id, True, False

        logger.debug("No output for %s", source_id)

        return source_id, False, False

    except Exception as e:

        console.print(f"[red]  Failed {source_id}: {e}[/red]")

        logger.error("Failed to process %s: %s", source_id, e)

        return source_id, False, True





@app.command("process", help="Process workspace members.")

def workspace_process(

    workspace_name: str = typer.Argument(None, help="Workspace name (default: active workspace)"),
@@ -282,12 +320,11 @@ def workspace_process( 
    if limit is not None:

        members = members[:limit]



    # Pre-start hybrid server if extraction profile requires it

    if not ensure_hybrid_server_for_profile(extraction_profile):

        raise typer.Exit(1)



    manager = resolve_cache_manager()

    wiki_source_dir_base = manager.workspaces_dir / normalized / "sources"

    cache_manager = resolve_cache_manager()

    wiki_source_dir_base = cache_manager.workspaces_dir / normalized / "sources"



    processed = 0

    failed = 0
@@ -307,28 +344,17 @@ def workspace_process( 
                progress.advance(task)

                continue



            wiki_source_dir = wiki_source_dir_base / source_id

            wiki_source_dir.mkdir(parents=True, exist_ok=True)



            try:

                result_path = convert_for_wiki(

                    document_id=source_id,

                    wiki_source_dir=wiki_source_dir,

                    source_kind=member.source_kind,

                    profile=extraction_profile,

                    force=force,

                    release=member.release,

                    md_yaml_frontmatter=md_yaml_frontmatter,

            progress.update(task, description=f"Processing [{extraction_profile.value}] {source_id}")

            _, succeeded, errored = _process_member(

                member,

                wiki_source_dir_base,

                extraction_profile,

                force,

                md_yaml_frontmatter,

            )

                if result_path:

                    suffix = result_path.suffix.lstrip(".")

                    logger.debug("%s [%s] → %s", source_id, extraction_profile.value, suffix)

            if succeeded:

                processed += 1

                else:

                    logger.debug("No output for %s", source_id)

            except Exception as e:

                console.print(f"[red]  Failed {source_id}: {e}[/red]")

                logger.error("Failed to process %s: %s", source_id, e)

            if errored:

                failed += 1



            progress.advance(task)