Online Sources

The Sources view (🌐 in the icon rail) — "Find books online" — searches free, legal digital libraries and imports works into your library in one click. Book search is per language: the active language decides which libraries are searched and where imports land.

The view

A Language switcher sits next to the heading. It is the same active language used everywhere else in the app; changing it here re-scopes every tab and clears the previous language's results. Below it, a row of tabs shows the sources available for that language:

A single search box serves the search tabs; its behavior differs per tab. While an import is running, a spinner with an "importing N…" counter appears in the header, and the Library nav icon shows the same count.

Imported works go through the normal import pipeline and land in the Library exactly like a file import: they are tokenized, difficulty-graded, and openable in the Reader immediately. Imports land in the active language's library and run through its analyzer, so the language switcher doubles as "import this book as this language."

Aozora Bunko

Aozora Bunko (青空文庫) is a volunteer archive of public-domain Japanese literature. This tab appears only when Japanese is active.

The catalog

• Shiori fetches Aozora's full catalog CSV once, in the background, when the app starts. Per Aozora's politeness guidance, the download comes from the project's GitHub mirror, not aozora.gr.jp itself.
• The catalog is cached in the data directory (aozora_catalog.zip). If you are offline, the last cached copy is used; if there is no cached copy yet, the view shows "Catalog not available" until you reconnect.
• The ⟳ Reload catalog button next to the search box deletes the cache and fetches the current catalog.

Only importable entries are kept: public-domain works only (copyright column なし), hosted on aozora.gr.jp, one row per work (the 著者 author row).

Searching and importing

Search is local against the cached catalog — instant and offline — matching the query as a substring of the title, title reading (kana), or author. Clicking ⬇ Import downloads the work's XHTML (GitHub mirror first, aozora.gr.jp as fallback) and feeds it through the HTML pipeline (furigana ruby stripped). Aozora sends no charset header, so the file is decoded per the catalog's encoding column (UTF-8 or Shift_JIS). The publisher field is set to 青空文庫.

Wikisource

The Wikisource tab queries the active language's Wikisource wiki — classic literature, historical documents, speeches, and law texts. The subdomain is resolved from the language code (fr → fr.wikisource.org, la → Vicifons, and so on). A language with no Wikisource of its own simply doesn't get the tab — Koine Greek, for example, has no Koine-specific Wikisource (the only Greek wiki is Modern Greek's), so it relies on Project Gutenberg and its Libraries directory instead. Koine and Modern Greek are treated as entirely separate languages, sharing no book sources.

Type a query and press Enter or click Search: Shiori runs a MediaWiki full-text search over mainspace (readable prose, not the Page: proofreading namespace) and shows up to 20 hits with title, word count, and a snippet. Clicking ⬇ Import fetches the page's rendered HTML from the MediaWiki REST API, strips it to text, and imports it — title from the page, publisher set to the wiki's host, author left empty (editable afterward in the library).

Project Gutenberg

The Project Gutenberg tab searches Gutendex, a JSON API over Gutenberg's catalog, filtered to the active language (Gutenberg indexes records by romanized metadata, so search "Akutagawa", not "芥川"). Results list the title, author, and download count.

Clicking ⬇ Import downloads the book, preferring a UTF-8 plain-text edition (falling back to HTML, then EPUB). Project Gutenberg's license header and footer are stripped so only the work itself is imported; the publisher is set to Project Gutenberg.

OPDS distributors

OPDS (Open Publication Distribution System) is the standard catalog format e-book libraries expose. You add distributors per language and search them here.

• Add a distributor with the form at the bottom of the tab: a name and the feed's root URL (https://…). Two anonymous, multilingual feeds — Project Gutenberg and Open Library — are offered as one-click suggestions. Added distributors are saved in your settings, per language.
• Search the selected distributor with the search box. Shiori follows the feed's advertised search (an OpenSearch description for OPDS 1.x, or a templated search link for OPDS 2.0); feeds without search are listed and filtered locally, and navigation-only feeds (like Gutenberg's) are followed one hop to reach the books.
• Import with ⬇ Import: EPUB and PDF go through the file pipeline (extracted and copied into your library), HTML and plain text are imported directly. Both OPDS 1.x (Atom) and OPDS 2.0 (JSON) feeds are supported.

Some catalogs (e.g. Standard Ebooks' patron feed) require authentication; those are not supported yet — use their public pages or another distributor.

Libraries

The Libraries tab is a read-only directory, compiled from public catalog metadata, of free and legal digital libraries: the collections dedicated to the active language, followed by multilingual aggregators (Internet Archive, Open Library, HathiTrust, …). Each entry links out to the site. Use it to discover where a language's books live — then download a file and import it from the Library page, or add the site's OPDS feed under the OPDS tab to search it in-app.

Politeness

Shiori is deliberately gentle with every service:

• Every request carries a descriptive User-Agent identifying the app, as the MediaWiki and Gutendex policies expect.
• Requests run serially — catalog fetches, searches, and imports happen one at a time, never in parallel bursts.
• The bulk Aozora catalog comes from the GitHub mirror rather than the Aozora site, and individual work downloads also try the mirror first.
• Wikisource search requests pass maxlag=5, so they back off automatically when the servers are under load.

Offline behavior

After the first successful catalog fetch, Aozora search works fully offline. Everything else — Wikisource, Project Gutenberg, OPDS, and all imports — requires a connection. These fetches, optional LLM calls, and the language-pack downloads under Settings → Languages are the app's only network features — see Getting-Started for the first-run reference-data download.