Public metadata catalog

The catalog is for discovery and linking. Bulk collection of book text stays forbidden by the terms. The catalog contains no book text, sentences, synopsis excerpts, or glosses.

JSON catalog

Fetch catalog.json without signing in. Each response has entries and a next URL, or null at the end. Follow that URL unchanged. Pages contain at most 100 books in ascending UUID order; this stable order avoids offset drift. It is a live catalog, not a snapshot: newly published books with earlier IDs appear on the next full traversal.

Each entry includes id, canonical url, original and English title, author and translator names and publication years where recorded, source_language, english_pairing, level (0–10), band, sentence_count, listening_length, narration, rights, cover_url, and updated_at. Unknown facts are null, never inferred publication dates or invented credits. Author years describe the original work when recorded, not the author's lifetime; a source printing's year is kept separately in the rights basis and is not substituted for an unknown original date. Listening seconds cover the source at normal speed, including paced reading, and are not a promise of recorded audio. Narration identifies our machine voice when speech is enabled; no human recordings or LibriVox reader credits are recorded in the current data.

Only ready, public house stories and imported classics appear. Reader-created books, including public books, uploads, deleted books and languages still behind a launch flag do not appear. Historical house books held by the library account are included.

The response is the same for every caller, cached at the CDN for 60 seconds with up to 60 seconds of stale revalidation. A cache miss uses one bounded metadata query, plus the existing public API rate-limit check. Both feed routes share an allowance of 240 requests per minute per network address at the origin, matching public search’s rate. A refusal returns 429 with Retry-After; invalid cursors return 400. Errors are not cached.

OPDS 2.0

The feed follows OPDS 2.0. Fetch the OPDS feed with the same pagination. The content type is application/opds+json. It uses the same source and eligibility as the JSON catalog. Acquisition links open the HTML reading page, never an EPUB, audio, or text file. Clients must support HTML acquisition to use those links.

Rights and citation

Use the per-entry rights record and linked terms. A public-domain original does not put Alloglot's translations, alignments, explanations, or audio in the public domain. House stories carry “Alloglot house story © Alloglot” with the existing non-commercial reuse terms; this is not a new open license. The terms explain that copyright may not reach machine-generated material. Source edition URLs, recorded licenses, and edition years supply provenance; public-domain statements are scoped to the United States where the edition says so.

To cite a book, name its author, title, original publication year when known, translator and translation year where credited, Alloglot as publisher, and its canonical reading URL. Include the access date for this live edition. Do not use updated_at as the original publication date.

Sitemaps

The catalog is a metadata feed, not a sitemap replacement. Submit the website and app's /sitemap.xml to search engines; the docs site has its own https://docs.alloglot.com/sitemap.xml for these documentation pages. Book-page indexing and language discovery lists have their own scope and eligibility checks.