* fix(vi/Hako): fix chapter path parsing and decrypt protected chapter content
- Fix `parseNovel` extracting full URL instead of relative path for chapters.
- Implement decryption logic for protected chapter content (`#chapter-c-protected`) using XOR shuffle, base64 reverse, and plain base64 decoding.
* fix(vi/hako): format note tags, suggestion banners, and bump version to 1.1.1
- Format `[note(\d+)]` tags into HTML span/anchor tags for proper rendering (following the original website's logic)
- Remove recommendation/suggestion banners at the end of chapters
- Bump plugin version to 1.1.1
* Refactor NovelMania plugin for API updates
Well, the Novel Mania website updated and revamped everything, so I tried to fix it manually, until I gave up and used Replit's AI to discover the new APIs. I first tested it with a fork of the repository using my own app, and so far it's working well here; everything is functioning perfectly, from the filters to the search system and content recognition.
Note: Translated from Portuguese to English using Google Translate because my English version is unusable lol
* Fixes after using "npm run lint"
* fix(novelbin): fix missing summary due to multiple classes
* Add case for 'desc-text desc-text-collapsed' in parser
* Add version increments to Novel Bin source options
* feat(novelbin): remove NB Reader app advertisements
* fix(faqwikius): update site URL to include '/novel' for correct navigation
* chore: bumped version of faqwiki
* fix(faqwikius): remove script tags from parsed HTML to clean up novel data
* fix(faqwikius): move the remove script tags from parseNovels method to parseNovel method
* chore(faqwikius): rename parseNovels method to parseNovelList and parseNovel to parseNovelDetails for clarity
* fix(faqwikius): reversion of previous commit
* fix: update BoxNovel source configuration
* feat: add filter configuration for boxnovel plugin
* chore: remove redundant boxnovel filter file from readnovelfull multiscr
* fix: boxnovel uses newchapterendpoint
Chapter list was not working
Co-authored-by: K1ngfish3r <26593485+K1ngfish3r@users.noreply.github.com>
* chore(madara): bump template base version to 2.2.0, set boxnovel to increment 2
* chore(madara): reset other plugin version increments in sources.json
---------
Co-authored-by: K1ngfish3r <26593485+K1ngfish3r@users.noreply.github.com>
* fix(novelbuddy): fix watermark regex and bump version to 2.1.2
* fix(novelbuddy): fix watermark regex, remove chapter api fetch and bump version to 2.1.2
* fix(proxy): propagate origin status code and handle 304 Not Modified correctly
* Add API call for fetching manga chapters
Fetch chapters from API if available, otherwise use initial data.
* revert(novelbuddy): revert novelbuddy to version 2.1.1 and simple watermark removal
* fix(proxy): improve redirect handling, status propagation, and 304 handling
Restored verbose logging and implemented robust internal redirect following with infinite loop protection. Verified via web debugger.
* fix: type errors for buffer and redact sensitive logs
* fix: replace any with strictly typed separate variables
* chore(proxy): revert log redaction and run prettier
* fix(novelbuddy): fix watermark regex and bump version to 2.1.2
* fix(novelbuddy): fix watermark regex, remove chapter api fetch and bump version to 2.1.2
* Update novelbuddy.ts
* feat(novelbuddy): optimize performance and improve watermark removal
* feat(novelbuddy): optimize performance and improve watermark removal (v2.1.2)
* chore: restore original index.ts to ensure clean PR
* refactor: address reviewer feedback and update watermark regex
- Replace regex-based URL path manipulation with startsWith and slice.
- Update fwnRegex with a more comprehensive pattern.
- Run npm run format.
* refactor: update watermark regex to latest comprehensive pattern
- Integrate the newest provided fwnRegex pattern.
- Ensure non-capturing groups and special characters are correctly handled.
- Maintain addressing of reviewer feedback regarding URL manipulation.
- Run npm run format.
* Add TruyenSS Vietnamese source for truyenss.com
Implements novel pages, hash-style chapter list, POST chapter loader
(/layout/xem-chuong.php), genre filters for popular listings, and
multi-pattern search. Includes saved HTML fixtures used for selectors.
* fix(truyenss): fetch chapter HTML via GET like the site
TruyenSS loads chapters with jQuery $.ajax, which defaults to GET.
POST to xem-chuong.php returns an empty body, causing empty chapter
errors in the app while WebView (GET + hash) still worked.
Bump plugin version to 1.0.1.
* fix(truyenss): address PR review — version 1.0.0, defaultCover, cleanup
Remove README note and truyenss HTML fixtures. Drop unused resolveUrl and
full-URL handling in parseNovel. Use defaultCover for listing items and when
novel page has no cover image.
* fix(truyenss): browse list covers — scrape thumbs, stable placeholder URL
LNReader applies plugin Referer headers to browse tiles; external blob
placeholders can fail. List layout omits defaultCover when cover is unset.
Prefer real listing images; use raw GitHub placeholder URL aligned with the app.
* fix(truyenss): resolve listing covers (.card sibling img, ../ URLs)
Genre/search cards put the thumbnail in col-2 beside the stretched link,
not inside the anchor. Image src uses ../user-upload/... which must be
resolved against the list page URL. Use the site no_avatar.jpg placeholder
so browse tiles load with the plugin Referer.
* fix(playground): readable novel summary + TruyenSS paragraph breaks
- Parse playground: normalize CR/LF and stray spaces; use whitespace-pre-line,
max-w-prose, leading-7, break-words, explicit font-sans for robust wrapping
when webfonts misbehave.
- TruyenSS: preserve <br>/<p> structure like novelfire (join with newlines)
instead of collapsing the whole intro to one line.
* fix(proxy): avoid static zstd import so Vite starts on older Node
zstdDecompressSync is not exported from zlib on Node versions before it was
added. Importing it at load time broke vite.config (and the dev server)
with SyntaxError. Only call zstd at runtime when content-encoding is zstd
and throw a clear error if the runtime lacks support.
* fix(proxy): correct decompress error message for zstd
* Revert "fix(proxy): correct decompress error message for zstd"
This reverts commit ab5c4f0f6b0059589956693da534294a49500be0.
* Revert "fix(proxy): avoid static zstd import so Vite starts on older Node"
This reverts commit 58ddf6c6003fa3828a5a6627f77b4ea3aa3cfca4.
* fix(dev): proxy URL for https targets + disable auto-open by default
- Reconstruct target URL when the dev server path is /https://... (from
window.fetch prefixing localhost); the old + path form produced
an invalid URL and broke all proxied plugin requests.
- Only open a browser when VITE_OPEN=true so missing xdg-open does not error.
* Revert "fix(dev): proxy URL for https targets + disable auto-open by default"
This reverts commit b0de564fca75e756ef11899e77038fc594f18f97.
* revert: drop out-of-scope parse-novel playground summary changes
Restore src/components/parse-novel.tsx to pre-093668b state; TruyenSS
summary parsing in the plugin is unchanged.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
`$()` only treats the input as HTML when it starts with `<`; otherwise it
parses as a CSS selector. The novelbuddy API returns `initialManga.summary`
as plain text for many novels (e.g. "When I opened my eyes, I was inside
a novel."), and cheerio's selector parser then chokes on the first `.`
in the punctuation with `Expected name, found .` — `parseNovel` throws
and the novel page fails to load entirely.
Wrap the summary in a `<div>` before passing it to `$()` so cheerio always
takes the HTML branch. Plain text becomes `<div>plain text</div>` (parses
fine, `.text()` still returns the original string), HTML stays HTML, and
the existing `.find('br').replaceWith('\n')` / `.find('p').before/after`
behavior is preserved either way.
Bumps version 2.1.0 → 2.1.1.
Repro:
- Open the source, search/browse to "Trash of the Count's Family" (or any
novel where the API returns a plain-text summary)
- Tap the entry; details fail with `QuickJsException: Expected name, found .`
Noah Verburg