| name | pdf-acquisition |
| description | Provenance-first public PDF acquisition for Idea2Repo. Use when downloading, validating, hashing, licensing, or reviewing paper PDF manifests and docs/reference/pdfs artifacts. |
PDF Acquisition
Acquire PDFs only from public/legal sources and keep failures explicit.
Workflow
- Start from verified
PaperCandidate.pdf_urls.
- Skip unknown license sources with
skipped_license; do not download questionable files.
- Download only when the caller explicitly enabled PDF download and network permission.
- Validate the file starts as a PDF, parse enough text for title matching, then write it under
docs/reference/pdfs/.
- Record
paper_id, source_url, downloaded_at, sha256, bytes, license_hint, title_match_score, and status in docs/reference/pdf_manifest.json.
Guardrails
- A missing or failed PDF must not block the whole pipeline.
- Never fabricate PDF files or hashes.
- Keep tests on tiny fixture PDFs or mocked fetch/parser output.