Skip to content

PDFs

Open-access only. Sources run in this order and stop at the first valid PDF:

  1. Existing attachment (skipped; stale fulltext:pdf-missing and fulltext:check-needed tags cleared)
  2. Unpaywall
  3. PubMed Central
  4. Europe PMC
  5. Semantic Scholar
  6. Crossref (open-licensed links only; follows preprint links)
  7. CORE
  8. OpenAIRE (only products it rates open)
  9. Preprint servers: arXiv, bioRxiv, medRxiv, OSF family, PeerJ Preprints, Zenodo, Figshare
  10. HAL, DOAJ, OSTI, Internet Archive Scholar
  11. OpenAlex links, then hosted full text if you set a key
  12. Internet Archive retry for dead links
  13. Free-to-read publisher pages through Zotero's own downloader (on by default)

Safety checks

  • Bytes must start with %PDF- and be 10 KB to 100 MB.
  • If the PDF's embedded DOI differs from the item's, it is not attached and the item gets fulltext:check-needed. Preprint DOIs inside a published item's PDF are tolerated.
  • Title-only items match on exact normalized title and year within 1. Near matches are never accepted.

Politeness

3 items in parallel, 400 ms pause per host, one retry on network errors and 5xx, a 429 makes that source rest for a minute. Items with no open copy are skipped for a day. Runs over 60 items ask first. Cancel stops in-flight requests.

Tip

Preview PDF Sources shows which source would supply each PDF and changes nothing.

Tags written: fulltext:pdf-found, fulltext:pdf-missing, fulltext:check-needed. Tools > RefWright > Export Missing PDFs writes a CSV for interlibrary loan.