Internet Archive Library Search & Metadata Export

Search the Internet Archive library and export structured metadata for books, movies, audio, software and other indexed media. Researchers and data workflows can retrieve capped, paginated query results without writing an API client. Up to five queries share one identifier-deduplicated results and an uncharged coverage summary.

This Actor uses Archive.org public Advanced Search. It returns catalog metadata, not full texts, downloadable media, lending access, private content or Wayback website history. The source index may omit metadata and change during pagination.

Independent, unofficial tool; not affiliated with or endorsed by Internet Archive.

Input

queries accepts one to five Archive.org Advanced Search expressions, each 1–500 characters. The default is ["collection:gutenberg AND mediatype:texts"]. Use e.g. collection:nasa AND mediatype:movies for a different media profile. maxItems is the global maximum of unique delivered item rows, 1–1,000, default 100. Queries are processed in input order, in pages of 50 sorted by identifier ascending. Earlier queries may consume the shared limit, leaving later queries unvisited. Query coverage makes this explicit.

Example:

{"queries":["collection:gutenberg AND mediatype:texts","collection:nasa AND mediatype:movies"],"maxItems":200}

Output

archive_item rows contain identifier, short title, mediaType, creators, subjects, collections, languages, source date, download count, indexed licenseUrl and a short rights statement, itemUrl, matchedQueries, source and observedAt. Scalar/list metadata is normalized. Missing metadata remains null or an empty list; a missing rights/license field does not imply public domain or permission to reuse. Download counts are source observations, not buyers or payments.

One archive_search_summary row reports exportedItems, duplicatesRemoved, per-query queryCoverage, complete, stopReasons and contactsRedacted. matchedQueries covers only retrieved pages. Query totals are individual index counts and must not be summed into unique library items. If a live query total changes between pages, queryCoverage marks sourceWindowChanged and complete stays false; the result may omit items from the changing index. complete means all supplied queries exhausted in the observed live index; pagination is not a frozen snapshot. Standard Apify results exports provide JSON/CSV/Excel.

Pricing

PPE price is $0.001 per unique archive_item row delivered ($1 per 1,000). Duplicates and the summary are uncharged; no explicit start event. Non-PPE development runs deliver the same results without event charges or event-budget limits.

Limits

Sequential lightweight HTTP, 256 MB target memory, at most 30 total HTTP requests including robots and retries, 180 seconds for source collection. Pages contain 50 items, identifier ascending. Each request is bounded; a global item, event-budget or request cap returns explicit partial coverage. Delays and Retry-After are honored. Source authentication blocks, rate limits after bounded retries, network timeouts and invalid/error/repeated-page responses fail the run; they are not empty success. If all queries genuinely match zero items, fail with a no-results error and no item charge. No per-item metadata or media request is made.

Privacy

Only selected public catalog fields are requested. Uploader identities, review authors, contact details, account/server metadata, descriptions and media files are excluded. Contacts found inside allowed metadata are removed before saving. Public bibliographic creator names may remain. Use public non-sensitive search expressions. Account/favorite selectors and source records, and personal favorite collection identifiers, are excluded. Source license/rights statements are informational and do not authorize reuse. Archive.org terms, robots and automated-access guidance apply.

Research and automation use

Example user question: Which Gutenberg library items match this research query, and what source metadata and rights information are available? An AI agent supplies the query and reads item rows plus queryCoverage to understand missing metadata and export limits. The Actor does not promise every historical item or solve website snapshot extraction.

Validation status

Local source feasibility uses Gutenberg texts and NASA movies with multiple pages. Local regression tests cover output, failure handling, privacy, limits, PPE delivery and the installed SDK's non-PPE charging mode. Earlier real local runs delivered 100 Gutenberg items, 100 NASA movie items and 1,000 Gutenberg items; every item count matched mock billing and each run added one uncharged summary. See docs/reviews/2026-10-09-internet-archive-local-validation.json for those source runs and docs/reviews/2026-10-10-demand-system-review.md for the subsequent repair verification in the source repository. Private cloud source reachability, full dataset delivery and live PPE parity are checked before Store publication.