torchcell.literature.capture_by_doi

capture_by_doi(lib: ZoteroLibrary, doi: str, *, dryad_doi: str | None = None, extra_si_urls: list[str] | None = None, si_expected: list[str] | None = None, do_ocr: bool = True, data_root: str | Path | None = None) → Path[source]

Capture a paper’s full artifact from Zotero + external SI data.

Pipeline: resolve the Zotero item by DOI -> download main + SI PDFs -> fetch SI data files (DRYAD / extra URLs) -> OCR every PDF to markdown -> write manifest.json recording provenance + per-file sha256. The manifest is written last so it captures every produced file.

Parameters:

  • lib – Connected Zotero library.

  • doi – Paper DOI (the join key); must exist in the library.

  • dryad_doi – Optional DRYAD dataset DOI for the SI data files.

  • extra_si_urls – Additional direct URLs to pull into si/si_data/.

  • si_expected – SI items the paper says should exist (completeness check).

  • do_ocr – Run MinerU OCR (slow on first run – downloads models).

  • data_root – Mirror root; defaults to $DATA_ROOT.

Returns:

The artifact directory path.