Wikipedia recovery — lost-files subset

WIKIPEDIA RECOVERY · DARE · 10 MAY 2026

Programmatic recovery of Wikipedia/Wikimedia images from the lost-files list. Re-resolves canonical URLs via the MediaWiki API, downloads to staging, and (optionally) uploads to R2.

TL;DR

4 Wikimedia-pattern basenames attempted.
3 resolved via the MediaWiki API (1 not found in Commons or en.wikipedia).
3 downloaded to /Users/dansellars/Downloads/dare-recovered.
R2 upload skipped (no --r2-bucket given).

Per-file outcome

Basename	Source	Bytes	R2 status
`240px-Le_Dernier_Combat.jpg`	en.wikipedia	17,567	(skipped)
`330px-ZM1210-operating_edit2.jpg`	commons	1,184,308	(skipped)
`CRW_8331lg.jpeg`	not found	—	(skipped)
`Polarlicht_2.jpg`	commons	4,632,133	(skipped)

Next steps

Files staged at /Users/dansellars/Downloads/dare-recovered. To upload to R2 manually: bash for f in ~/Downloads/dare-recovered/*; do npx wrangler r2 object put YOUR-BUCKET/posts/$(basename "$f") --file="$f" done
Re-run dare_lost_files_audit.py to confirm the recovered files now resolve as in-both rather than lost.
For the lost subset that ISN’T Wikipedia (Flickr-style, generic), different recovery sources apply (Internet Archive, replacement images, or removing the reference).

Generated 2026-05-10 17:20:14 via dare_recover_wikipedia.py. Staging: /Users/dansellars/Downloads/dare-recovered. Source filter: \d+px- thumbnail prefix or Commons-style underscore casing.

Source: dare_recover_wikipedia_2026-05-10.md · Rendered 2026-05-10 17:20