Function decode_parse_engine

lightrag/parser/routing.py:108–134

Decode a stored ``parse_engine`` field into ``(engine, params, errors)``. ``value`` may be a bare engine (``mineru``) or an encoded directive (``mineru(page_range=1-3,language=en)``). ``engine`` is the bare, normalised engine name; ``params`` is the canonical coerced dict; ``errors``

(
    value: Any,
)

def decode_parse_engine(
    value: Any,
) -> tuple[str, dict[str, Any], list[str]]:
    """Decode a stored ``parse_engine`` field into ``(engine, params, errors)``.

    ``value`` may be a bare engine (``mineru``) or an encoded directive
    (``mineru(page_range=1-3,language=en)``).  ``engine`` is the bare,
    normalised engine name; ``params`` is the canonical coerced dict; ``errors``
    is non-empty for a malformed/unbalanced block or invalid params (callers on
    the parse path raise so the doc fails visibly instead of dropping params).
    """
    raw = str(value or "").strip()
    if not raw:
        return "", {}, []
    idx = raw.find("(")
    if idx == -1:
        return normalize_parser_engine(raw), {}, []
    engine = normalize_parser_engine(raw[:idx])
    inner, after = take_paren_block(raw, idx)
    if inner is None:
        return engine, {}, [f"unbalanced '(' in parse_engine {raw!r}"]
    if raw[after:].strip():
        return engine, {}, [f"unexpected text after ')' in parse_engine {raw!r}"]
    params, errors = parse_engine_params(
        inner, engine=engine, label=f"parse_engine {raw!r}"
    )
    return engine, params, errors


# ---------------------------------------------------------------------------