__init__.py 1.2 KB

1234567891011121314151617181920212223242526272829303132333435363738394041
  1. """Docling parser integration (raw client, cache, manifest, IR adapter).
  2. Public surface for the rest of the codebase. ``parse_docling`` imports
  3. only from this facade so the inner module layout stays free to evolve.
  4. """
  5. from lightrag.constants import DOCLING_RAW_DIR_SUFFIX
  6. from lightrag.parser.external._common import (
  7. clear_dir_contents,
  8. raw_dir_for_parsed_dir as _raw_dir_for_parsed_dir,
  9. )
  10. MANIFEST_ENGINE = "docling"
  11. def raw_dir_for_parsed_dir(parsed_dir):
  12. """``foo.parsed/`` → ``foo.docling_raw/`` (docling-specific binding)."""
  13. return _raw_dir_for_parsed_dir(parsed_dir, suffix=DOCLING_RAW_DIR_SUFFIX)
  14. # Imported after ``MANIFEST_ENGINE`` / ``DOCLING_RAW_DIR_SUFFIX`` because
  15. # the submodules read those constants at import time.
  16. from lightrag.parser.external.docling.ir_builder import ( # noqa: E402
  17. DoclingIRBuilder,
  18. )
  19. from lightrag.parser.external.docling.cache import ( # noqa: E402
  20. is_bundle_valid,
  21. )
  22. from lightrag.parser.external.docling.client import ( # noqa: E402
  23. DoclingRawClient,
  24. )
  25. __all__ = [
  26. "DOCLING_RAW_DIR_SUFFIX",
  27. "MANIFEST_ENGINE",
  28. "DoclingIRBuilder",
  29. "DoclingRawClient",
  30. "clear_dir_contents",
  31. "is_bundle_valid",
  32. "raw_dir_for_parsed_dir",
  33. ]