Skip to content

source-adapters.md

Latest commit

 

History

History
48 lines (39 loc) · 3.49 KB

source-adapters.md

File metadata and controls

48 lines (39 loc) · 3.49 KB
Raw

Source Adapters

Verified against internal/source/ and the live Comix source on 2026-08-28.

All adapters implement the small domain.Source contract:

  • Discover returns manga metadata and its chapter map.
  • Pages returns an ordered page transport description for one chapter.
  • adapters return values. They do not mutate caller-owned manga or chapter values.
  • source.Select is the single registry used by download and monitor commands.

Matrix

Identifier Input (-m) Extra args Validation highlights Notes
tcbscans manga title none non-empty title HTML scraping
mangadex manga UUID -g optional, -l optional valid manga UUID API-based
mangaplus numeric title ID none strict numeric regex mobile protobuf API; lazily registers a deterministic device secret
flamecomics full series URL none https://flamecomics.xyz prefix HTML + embedded JSON
asurascans full series URL none https://asurascans.com/comics/... server-rendered HTML; filters locked early-access chapters from discovery
cubari gist URL -g required valid URL + non-empty group images listed in the payload, or fetched from the /proxy/... path the gist points at
weebcentral full series URL none https://weebcentral.com prefix scraper + chapter image fragment fetch
comix full manga URL -g optional https://comix.to/title/... prefix; numeric group when set private API codec + referer-protected images + tile reconstruction
atsumaru full manga URL -g required https://atsu.moe/manga/... prefix + non-empty scan ID API-based

Rules

  • validate early
  • keep selectors and parsing local to the adapter
  • prefer stable attributes or embedded data over brittle DOM traversal
  • reuse internal/sharedhttp/ before adding new transport helpers
  • when an adapter drifts, add a regression test around the parsing seam if practical

Shared Behavior

  • unknown source values fail fast in source selection
  • mangaplus uses the mobile API because the web protobuf endpoint rejects current unauthenticated access; chapter discovery reads the current chapter_list_v2 field with legacy list fields kept as fallback
  • asurascans removes chapters marked is_locked=true or is_premium=true before returning the chapter map
  • weebcentral fetches chapter images from the /chapters/<id>/images HTML fragment
  • comix implements frontend build 35595e3de3c99889c1aa70; it generates request tokens, decodes encrypted API envelopes, sends image request headers, and reconstructs scrambled tile images
  • Comix scramble hashes are opaque routing keys. The two known legacy hashes select explicit seed prefixes; unknown hashes use prefix zero, matching the frontend fallback.
  • atsumaru fetches chapter metadata from /api/manga/info, filters chapters by scan ID, and resolves relative page paths from /api/read/chapter
  • source-specific image transforms use domain.ImageProcessor; the acquisition path owns transport and output while the source adapter owns the transform
  • shared request retries use internal/sharedhttp/; see the retry policy and scraper limitations
  • Manga Plus request errors omit query strings so registration and device secrets do not enter logs
  • HTTP and Colly-backed requests inherit caller cancellation
  • fixture-backed parser flows cover every supported source
  • manhwa/long-strip handling flows through selectedManga.IsManhwa