Fix HizoManga showing no chapters - #269
Merged
Merged
Conversation
Cloudflare answers every GalaxyNovels request with a hard WAF 403
("Sorry, you have been blocked"), not a solvable JS challenge, so
cf_clearance/WebView solving cannot help. Two independent request
signatures trigger it, both measured against the live site:
1. OkHttp's default `okhttp/x.y.z` User-Agent is rejected on every path.
The source sent no User-Agent of its own, so the app-wide default was
used and even list pages that the site serves fine to a real browser
came back 403.
2. `Accept-Language` sent together with BOTH `Sec-Fetch-Dest: document`
AND `Sec-Fetch-Mode: navigate`. Either Sec-Fetch header alone passes,
and Accept-Language alone passes, but the combination does not.
The interceptor now pins a desktop Chrome User-Agent and stops sending
Sec-Fetch-Dest/Sec-Fetch-Mode. Referer and Origin already come from
KeiSource's headersBuilder(), so the hand-rolled copies in the interceptor
and in fetchPageText are gone - fetchPageText had the same broken set and
so 403'd on every chapter.
This also unblocks the chapter list: because the detail page 403'd, the
data-novel-id was never parsed, so the WorReader manifest/pack fetch never
ran and every novel fell back to the ~30 chapters in the detail page HTML.
With the detail page reachable, the manifest+pack path is used again and
returns the full list (verified: 790 chapters for one novel, 939 for
another). The legacy per-novel JSON endpoint is removed - it is now dead
(403 even with a browser UA), so it only cost a wasted request.
Finally, fetchPageText looked for WordPress-style containers, none of
which exist on the current WorReader theme, so it fell through to <article>
and returned the whole page. It now selects `div.wor-reader-text-surface`,
the reader's actual text container (verified: 52 paragraphs, ~6.4k chars),
keeping the old selectors as a fallback.
Co-Authored-By: Claude Code <noreply@anthropic.com>
Reading a chapter showed sentences that are not part of the novel. The site interleaves anti-piracy notice lines *inside* the chapter body -- "اقرأ من galaxynovels.com", "لا تدعم من يسرق الروايات وينسبها لنفسه" and rotations of the same idea -- one every few paragraphs. They are plain inline elements (<small>/<b>/<i>) that a real browser hides, so the site itself never shows them, but a raw HTML scrape keeps them. Each carries data-wor-decoy="1" plus hidden/aria-hidden="true" and an inline style that collapses it to zero size, so on the measured chapter 17 of the 69 children of div.wor-reader-text-surface were decoys: 1323 of 7667 characters, roughly 17% of the extracted text. Dropped by attribute rather than by wording, since the site rotates the sentence. data-wor-decoy is the theme's own marker and is matched first; the accessibility attributes are kept as a fallback in case a future theme drops it. No real prose carries these attributes, so all 52 real paragraphs survive. Co-Authored-By: Claude Code <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
hizomanga.net rejects every Madara chapter endpoint at the CDN: POST /content/<id>/ajax/chapters/ answers 403 and the legacy admin-ajax.php?action=manga_get_chapters answers 400, so the shared implementation returned an empty list. The novel page itself answers 200 and already renders the complete list into .listing-chapters_wrap .wp-manga-chapter, so parse it from there instead. Verified on a 239-chapter series (contiguous 1..239 in the initial HTML, no pagination), a 130-chapter one, a 3-chapter one, and a series with no chapters yet. Co-Authored-By: Claude <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The HizoManga source (
hizomanga.net) returned an empty chapter list, so no chapters showed at all.Cause
Madara normally fetches chapters from a separate ajax endpoint rather than from the novel page. On this host every chapter endpoint is rejected at the CDN:
POST /content/<id>/ajax/chapters/error_docs)POST /ajax/chapters/?t=1POST /wp-admin/admin-ajax.php+action=manga_get_chapters0)Tried with a desktop browser UA,
Referer,Origin,X-Requested-With,Sec-Fetch-*, and a cookie jar primed from the novel page — the 403 persists on every combination. The novel page itself returns 200 throughout, so this is a targeted block of the ajax paths, not of the site.Fix
The novel page already renders the complete chapter list server-side, so the list is read straight from it:
Chapter parsing mirrors the theme's own logic — number-prefix stripping, relative URL normalisation, release-date parsing, and
🔒marking ofpremium-blockentries.Verification
Simulated the parser against live pages:
4289842370183916183858All parsed URLs are relative and well-formed. The 239-chapter list confirms the site does not paginate the chapter list, so a single page request is sufficient.
Note
mangaDetailsParseandparseChaptersFromHtmlareprivatein the shared theme, so the detail parsing is mirrored locally (URLs taken from the existingSMangaas fallback). This is scoped tosrc/ar/hizomanga— the sharedmadaranoveltemplate is untouched.🤖 Generated with Claude Code