torchcell.literature.ocr_artifact

ocr_artifact(artifact_dir: str | Path, **kwargs: Any) → list[Path]

OCR every PDF in an artifact directory: paper.pdf and each si/si*.pdf.

The SI PDFs run in natural order (si2 before si10). A directory with no paper.pdf raises MissingPaperPdfError before any OCR: every artifact is captured with its article as paper.pdf (ZoteroLibrary.download_artifact), so its absence means a broken capture, not an SI-only paper.

Returns the list of markdown paths produced (paper.md, si/si1.md, si/si2.md…).