docling-project / docling-project/docling
Add HWP/HWPX input backend with layout provenance
- Dominant language
- Python
- Stars
- 66.4k
- Forks
- 4.8k
- Avg merge
- 2d 21h
- Merged PRs (30d)
- 84
Description
### Requested feature
Add input support for Korean Hangul Word Processor documents (`.hwp`, `.hwpx`).
These formats are widely used in Korean public-sector, legal, academic, and enterprise workflows. Docling currently has no `InputFormat` for them, so users must convert HWP/HWPX outside Docling before they can use the CLI/API, exporters, chunkers, and the provenance-aware document model.
Ideally HWP/HWPX would flow through the normal Docling pipeline like the other supported formats, preserving the document's semantic structure where present (paragraphs, headings, lists, tables, pictures, formulas, footnotes/endnotes, page headers/footers, page breaks). The implementation approach is left to the maintainers.
### Alternatives
Convert HWP/HWPX to an already-supported format outside Docling and ingest the result. This works today as a preprocessing step, but adds a manual conversion stage and gives Docling no direct control over warnings, errors, or provenance mapping.
Contributor guide
Assessment
This issue has not been assessed yet.