Function extract

explainshell/manager.py:585–898

Extract options from manpages and store in DB.

(
    ctx: click.Context,
    mode: str,
    files: tuple[str, ...],
    dry_run: bool,
    overwrite: bool,
    filter_db: tuple[str, ...],
    drop: bool,
    jobs: int,
    batch: int | None,
    debug: bool,
    small_only: bool,
    large_only: bool,
    reason: str | None,
)

@click.argument("files", nargs=-1, required=True)
@click.pass_context
def extract(
    ctx: click.Context,
    mode: str,
    files: tuple[str, ...],
    dry_run: bool,
    overwrite: bool,
    filter_db: tuple[str, ...],
    drop: bool,
    jobs: int,
    batch: int | None,
    debug: bool,
    small_only: bool,
    large_only: bool,
    reason: str | None,
) -> None:
    """Extract options from manpages and store in DB."""
    try:
        parsed_mode, model = _parse_mode(mode)
    except ValueError as e:
        raise click.UsageError(str(e))

    if jobs < 1:
        raise click.UsageError("--jobs must be >= 1")
    if small_only and large_only:
        raise click.UsageError("--small-only and --large-only are mutually exclusive")
    if drop and dry_run:
        raise click.UsageError("--drop and --dry-run are mutually exclusive")

    filter_specs: list[tuple[str, str | None]] = []
    if filter_db:
        if not overwrite:
            raise click.UsageError("--filter-db requires --overwrite")
        for spec in filter_db:
            try:
                filter_specs.append(_parse_mode(spec))
            except ValueError as e:
                raise click.UsageError(f"--filter-db: {e}")

    if batch is not None:
        if batch < 1:
            raise click.UsageError("--batch must be >= 1")
        if not model:
            raise click.UsageError(
                "--batch requires a model (e.g. llm:gemini/<model>, llm:openai/<model>, or llm:azure/<deployment>)"
            )
        if not model.startswith(_BATCH_MODEL_PREFIXES):
            raise click.UsageError(
                "--batch only supports gemini/, openai/, and azure/ models"
            )

    try:
        gz_files = util.collect_gz_files(list(files))
    except ValueError as e:
        raise click.UsageError(str(e))
    if not gz_files:
        raise click.UsageError("No .gz files found.")
    if not dry_run and len(gz_files) > 100 and not reason:
        raise click.UsageError(