57501cc5208a9c79b59cf5cd7b44caae1387e91d
561 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Huang Xin
|
57501cc520 | feat(updater): nightly update channel (Android/Windows/macOS/Linux) (#4577) | ||
|
Huang Xin
|
bfb85c2f68 |
feat(reader): sync paragraph mode & speed reader with TTS read-along (#3235) (#4576)
* docs(reader): TTS-sync design spec for paragraph mode + RSVP (#3235) Hardened via brainstorming + /autoplan (CEO/Design/Eng dual-voice review). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(tts): emit canonical tts-position event from TTSController (#3235) Controller emits { cfi, kind, sectionIndex, sequence } alongside the existing tts-highlight-mark/-word events. Monotonic sequence lets downstream consumers (paragraph mode, RSVP — later slices) drop out-of-order positions. Additive; existing events untouched. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): containment+cursor CFI->index mappers for TTS sync (#3235) RSVPController.syncToCfi + setExternallyDriven: containment match (fixes mid-token skip), monotonic cursor + binary search (avoids O(N)-per-word jank, no per-word getCFI), -1/no-op on no match (no silent jump to word 0), timer suspension while externally driven. ParagraphIterator.findIndexByRange: hinted + binary-search containment mapper returning -1 on no match (never first()). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(tts): forward tts-position + tts-playback-state onto the app bus (#3235) useTTSControl republishes the controller's canonical tts-position (tagged with bookKey) via a dedicated listener — NOT inside the suppression-gated highlight handlers, so page-follow suppression can't silently desync the modes. Adds tts-playback-state (playing/paused/stopped) so RSVP can track playback without the hook-local isPlaying. Verified by extending the real-foliate-view browser harness. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): paragraph mode follows TTS playback (#3235) When paragraph mode + TTS are both active, the focused paragraph follows the spoken position (sentence granularity, all engines). Section-generation contract (stash cross-section position, apply after the iterator re-inits); sync-focus path that does NOT arm isFocusingRef (avoids the relocate-eaten wrong-section paragraph-0 bug); stale-sequence drop; decouple on manual nav, re-engage on next playing. Start-alignment + visible indicator deferred to later slices. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(rsvp): speed reader follows TTS playback (#3235) Edge word-boundary voices: RSVP shows the spoken word via syncToCfi. Non-Edge (sentence-only) voices: sentence-paced estimator (clamp 60..600 wpm from voice rate, hold at +60 words cap, snap to first word on each new sentence mark). RSVP auto-advance suspended while TTS-driven. Decouple on manual nav via a rsvp-manual-nav signal; re-engage on next playing. Cross-section positions re-extract then apply. Pure decideRsvpTtsPosition helper unit-tested. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): fixed-layout gate + ttsSyncStatus for TTS sync (#3235) Gate sync to reflowable books (D7): fixed-layout reports 'unsupported' and never engages. Both modes expose ttsSyncStatus (idle/following/syncing/ decoupled/unsupported) as the data source for the upcoming indicator. RSVPControl now forwardRef-exposes the status via an imperative handle. Cross-bookKey events ignored (regression-tested). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): 'following audio' indicator for TTS sync (#3235) 5-state pill (following/syncing/decoupled, idle+unsupported render null) shown top-center in the paragraph overlay and as a status row in the RSVP overlay. Decoupled state is the tap-to-resume control; first decouple fires a one-time toast. eink-bordered, glyph+text (no color-only), RTL logical props, touch targets, safe-area top inset. RSVP 'plain' variant matches its themed surface; non-Edge shows '· estimated'. New i18n keys need extraction before merge. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(rsvp): in-overlay TTS toggle + audio-paced speed control (#3235) Voice-glyph audio toggle in the RSVP control row (trailing, by the gear) starts/ stops read-along from inside the full-screen overlay, start-aligned to the current word (range validated against the live doc). While TTS-driven, the WPM control shows a locked 'Audio pace' affordance that opens a compact rate picker; rate changes go through a new tts-set-rate bus event reusing the existing throttled setRate path. Pure buildRsvpTtsSpeakDetail helper unit-tested. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(reader): e2e paragraph mode follows TTS across a section boundary (#3235) Real <foliate-view> browser e2e: with paragraph mode active, the focused paragraph follows the TTS walk and re-targets to the new section after a Ch4->Ch5 boundary (proves no stuck wrong-section paragraph-0 / isFocusingRef trap). Asserts on the owning section of the current range. Test-only. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * i18n(reader): translate TTS-sync strings across 33 locales (#3235) Following audio / · estimated / Resume audio / Stopped following audio / Play audio / Pause audio / Audio pace / Speed follows audio. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(reader): resolve TTS CFI anchors across iframe realms (#3235) RSVP and paragraph follow silently failed to track the spoken word: the CFI anchor from view.resolveCFI(...).anchor(doc) is a Range created in the book iframe's realm, so 'anchor instanceof Range' (top realm) was always false (cross-realm instanceof) -> resolveCfiToRange/applySyncCfi returned null -> syncToCfi never advanced. Add isRangeLike() duck-type (cloneRange is unique to Range) and use it at all 4 CFI-resolution sites. Confirmed live via CDP: before = syncToCfi false (frozen); after = exact word map + RSVP follows Edge TTS at ~171 wpm (audio pace). Unit tests reproduce the cross-realm anchor (jsdom is single-realm so the old code passed there but died in the app). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(rsvp): stop estimator/word fight + map transport to TTS play/pause (#3235) Two read-along refinements (verified live via CDP with Edge TTS): 1. No more jump-ahead-then-snap-back flashing. Word-boundary engines (Edge) emit BOTH sentence marks and word boundaries; RSVP was routing sentence -> the estimator (self-paces ~190xrate, up to +60 words ahead) while word positions snapped it back. Now once a word position is seen, sentence positions are ignored and any running estimator is stopped, so words alone drive RSVP. 2. The RSVP transport (center play/pause, Space, center-tap) maps to TTS play/pause while read-along is engaged (tts-toggle-play), instead of RSVP's own suspended timer. Pausing TTS keeps RSVP suspended (no runaway); a full stop releases it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(rsvp): keep indicator on pause + reach dict management from RSVP (#3235) - Pausing read-along no longer dismisses the 'following audio' indicator / 'Audio pace' lock (layout shift). New 'paused' sync status keeps the indicator row and WPM lock present while TTS is engaged-but-paused; only a full stop clears them. Verified live via CDP: pause keeps the layout, no shift. - Dict management is reachable from RSVP: the settings dialog is z-50, far below the full-screen RSVP overlay (z-[10000]), so it opened invisibly behind it. handleManageDictionary now exits RSVP first (position saved/resumable) so management shows over the reader. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(rsvp): show dict management over RSVP instead of exiting it (#3235) Per feedback: opening dictionary management from the RSVP lookup popup no longer closes the speed reader. The settings dialog is raised above the full-screen RSVP overlay (z-[10000] -> SettingsDialog !z-[10050]) so it shows on top, and RSVP's capture-phase keyboard handler bails while the settings dialog is open so its inputs accept Space and Escape closes settings (not RSVP). Verified live via CDP: management opens over RSVP, RSVP stays active behind it, Escape returns to it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(dict): only apply drag-handle margin compensation when the handle shows (#3235) The dictionary sheet header used -mt-4 to compensate for Dialog's drag handle, but that handle is sm:hidden (shown only below sm). On sm+ the handle is display:none, so -mt-4 pulled the header up into the top edge (broken layout when the lookup renders as a sheet on a short/wide window). Mirror the handle's breakpoint: -mt-4 sm:mt-0. Verified live via CDP. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): in-mode TTS audio toggle for paragraph mode (#3235) Paragraph mode already follows TTS, but there was no way to start read-along from inside it. Add an audio toggle to the ParagraphBar (mirroring RSVP's): it starts TTS start-aligned to the focused paragraph (range validated live, + section index) and stops it. Track session-active vs playing so a pause keeps the indicator ('paused' status) instead of collapsing to idle. Pure buildParagraphTtsSpeakDetail helper unit-tested. Verified live via CDP: tapping the icon starts audio from the focused paragraph and the focus follows speech. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): highlight current TTS word/sentence in paragraph mode (#3235) Paragraph mode follows TTS by advancing the focused paragraph, but the spoken word wasn't highlighted within it like normal mode. The overlay renders a CLONE of the paragraph, so the iframe's TTS highlight isn't visible there — reproduce it on the clone with the CSS Custom Highlight API (no DOM mutation, spans inline boundaries natively, leaves the fade-in animation untouched). - TTSController already tags tts-position with kind word|sentence. The hook decides granularity: word boundaries (Edge) drive a per-word highlight; once seen, the coarse sentence event is skipped so the whole sentence doesn't flicker over the current word. Engines without word boundaries (WebSpeech/Native) fall back to the sentence highlight. - Offsets are computed relative to the paragraph start (so they map 1:1 onto the clone's text) and tagged with the paragraph index so a stale highlight never paints the wrong paragraph. Cleared on stop / section change / disabled. - The ::highlight() style mirrors the user's ttsHighlightOptions color+style. Pure helpers (offset math, word/sentence decision, css builder) unit-tested. Verified live via CDP: word highlight tracks Edge word-by-word and follows across paragraph boundaries (news -> ... -> ladies), matching the TTS color. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
67c22c770b |
feat(reader): Share intent + customizable annotation toolbar (#4014) (#4570)
* docs(spec): annotation Share tool + customizable toolbar (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(plan): implementation plan for Share tool + customizable toolbar (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(annotator): add 'share' annotation tool type and button (#4014) * feat(annotator): add pure toolbar order/visibility helpers (#4014) * feat(annotator): add annotationToolbarItems view setting (#4014) * feat(annotator): add shareSelectedText ladder helper (#4014) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * feat(annotator): render Share tool and honor toolbar order in selection popup (#4014) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * feat(settings): add drag-and-drop annotation toolbar customizer (#4014) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * feat(settings): open the toolbar customizer from the Behavior panel (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(i18n): extract and translate annotation share/toolbar strings (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * refactor(annotator): extract canShareText helper, preserve hidden Share on cross-platform edit (#4014) Addresses final-review findings: de-duplicate the triplicated canShare definition into share.ts::canShareText, trim ShareCapableService to the fields actually read, and stop the toolbar customizer from dropping a synced 'share' tool when edited on a non-share-capable device. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(settings): WYSIWYG drag-and-drop toolbar customizer (#4014) Rework the customizer per live testing: - Render 'In toolbar' as a faithful, content-width, start-aligned preview of the real selection popup (gray bar, icon-only buttons); 'Available' tools show as labeled chips. - Multi-container dnd-kit pattern: in-place dragging (no DragOverlay, which a transformed modal offsets), pointerWithin collision so empty zones accept drops, live onDragOver reparent, itemsRef to dodge dnd-kit's drag-start handler-capture stale closure. - Add 'Add all' (canonical predefined order) and 'Clear all' shortcuts. - Align zone labels with the SubPageHeader breadcrumb. - Empty toolbar now suppresses the selection popup entirely (no empty bar), while still allowing highlight-edit/notes popups. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore(i18n): translate Add all / Clear all toolbar shortcuts (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(annotator): size selection popup to visible tool count (#4014) With the customizable toolbar a fixed-width popup looked sparse for a 2-3 tool toolbar (buttons spread to the corners). Size the popup to the number of visible tools (responsive) capped at the previous max; annotated selections keep the max width since they show highlight options / notes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(settings): reword empty-toolbar hint to 'No tools, drag one here' (#4014) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(annotator): render default tools (not Share) in popup layout screenshot (#4014) The visual regression test rendered every annotationToolButtons entry, so adding the Share tool shifted the toolbar to 9 buttons and broke the baselines. Share is hidden by default (added via Customize Toolbar), so the popup screenshot should mirror the default-enabled set — filter to DEFAULT_ANNOTATION_TOOLBAR_ITEMS, keeping the existing baselines valid. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
a56cc6c61a |
feat(tts): word-by-word highlighting for Edge TTS, closes #4017 (#4566)
Highlight each word as it is spoken (Edge TTS only) instead of keeping the whole sentence highlighted, and keep the view tracking the spoken word across page boundaries. Word boundaries - Capture Edge's audio.metadata WordBoundary frames (offset/duration in 100ns ticks plus the verbatim input-text span) in the Tauri, browser, and Cloudflare-Workers WebSocket transports. - Carry boundaries through the authenticated HTTPS proxy route via an X-TTS-Word-Boundaries response header (percent-encoded JSON, ASCII-safe), so word highlighting works on the web where the browser cannot open the wss connection directly. Cache them alongside the audio blob URL. Highlighting - Sync a requestAnimationFrame loop to audio.currentTime against the boundary table and highlight the word sub-range within the spoken sentence. Synthesis stays sentence-level (natural prosody); only the visual highlight is word-level. - Suppress the sentence highlight when the active client reports word boundaries and draw the first word immediately, so the whole sentence never flashes before the first word. Fall back to the sentence highlight when a chunk has no boundaries (other engines, empty metadata). - Re-apply the current word (not the sentence) when the view relocates. Page following - Turn the page as soon as the spoken word crosses a page boundary (a tts-highlight-word event scrolls only when the word is outside the visible range), instead of waiting for the next sentence. - Check the word's position for the "back to TTS location" badge so it no longer appears while the view follows the word onto the next page. Also fixes a pre-existing bug where the browser WebSocket was constructed with an options object (valid only for the Node ws package), which threw in browsers and made the wss path unusable on the web. Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Huang Xin
|
763b579c8f |
fix(android): launch installed dictionary for system lookup, closes #4559 (#4568)
On targetSdk 36, ACTION_PROCESS_TEXT handlers were hidden by Android 11+ package-visibility filtering — only auto-visible web browsers resolved the intent, so system-dictionary lookups landed in the OEM browser (VIVO/iQOO) even with a dictionary like Eudic installed. Add a <queries> declaration so dictionary apps are visible, and filter web browsers out of the handler set so an OEM browser that registers PROCESS_TEXT can't swallow the lookup: - no browser among handlers → unchanged implicit dispatch (keeps native Always) - browser + one dictionary → launch it directly (explicit component) - browser + several dictionaries → chooser excluding browsers, remembering the pick via EXTRA_CHOSEN_COMPONENT so later lookups go straight through - only a browser installed → report unavailable instead of opening it Routing is a pure, JUnit-tested decideLookupDispatch(). Adds get/clear lookup-dictionary commands + an Android-only reset row in the dictionary settings to switch the remembered app. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
c72afe269a |
fix(tts): keep voice list stable across region variants of a language, closes #4033 (#4565)
The voice panel filtered voices by the full locale of the currently speaking text (v.lang.startsWith(locale)), so a book mixing region variants of one language flip-flopped its voice list: Standard Ebooks tag their boilerplate front matter en-US (17 Edge voices) while the body text is en-GB (5 Edge voices). Filter by primary language instead (isSameLang) in all three TTS clients so every English variant yields the same voice set, and sort voices matching the requested locale first (TTSUtils.sortVoicesPreferLocaleFunc) so default-voice resolution via getVoiceIdFromLang still picks an exact-locale voice. This also fixes languages whose tags never matched a voice locale prefix at all (e.g. zh-Hans books previously got an empty Edge voice list). Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
loveheaven
|
1c392de0fa |
perf(reader): throttle library.json writes and cache known dirs to cut IPC (#4556)
useProgressAutoSave fires saveConfig ~once per second of reading. Two write-time sources were doubling its IPC cost: 1. saveConfig wrote the WHOLE library.json (+ backup) on every call, so a user with N books paid 2*JSON.stringify(N) per save. Chrome DevTools' Bottom-Up profile on a release Android build showed processIpcMessage chewing ~25% of main-thread time during a reading session. 2. nativeFileSystem.writeFile / copyFile defensively called plugin:fs|exists before every write to ensure the parent dir existed. Same dir gets probed once per save -- ~50% of IPC time per save was just exists() round-trips against directories that have been there since the book was opened. Fix: - LIBRARY_SAVE_THROTTLE_MS=30s coalesces a swipe burst into a single library.json write. Per-book config.json is still written eagerly -- it's the sync source-of-truth and is small. flushPendingLibrarySave() is called on hook unmount + window blur so closing the book always flushes. - In-process knownExistingDirs Set caches verified directories per app session. createDir adds, removeDir (incl. recursive) clears. Cold start still does the original exists+createDir dance once per dir. |
||
|
Huang Xin
|
61d804a54f |
fix(dict): resolve Android content-URI filenames via native basename (#4553)
On some Android devices the SAF picker returns an opaque, extension-less content:// document URI (e.g. .../downloads.documents/document/msf%3A20). Dictionary bundle grouping derived each filename from getFilename() — a pure string-parse of the URI — so no .ifo/.idx/.dict marker was found, every file was orphaned, and the user saw "Skipped incomplete bundles" even though the bundle was complete. Devices whose URI happens to embed the name (e.g. primary%3ADictionaries%3A21cen.dict.dz) worked, which is why it reproduced only on some Android devices. The same string-parse also wrote the bundle files (and synced metadata / contentId) under the mangled URI-segment names, so a re-import elsewhere did not dedupe. tauri's Android path.file_name (basename) special-cases content:// / file:// URIs and queries the content resolver for the real DISPLAY_NAME — the same call AppService.openFile already relies on. Resolve the display name once at selection time, store it on SelectedFile.name, and have bundle grouping classify by that name instead of re-parsing the URI. The old extension filter already used basename but discarded the resolved name; threading it through removes that divergence. Also fix the Settings -> Dictionaries "+" badges (Import Dictionary / Add Web Search) collapsing to a black spot in e-ink mode by adding eink-inverted, mirroring the font import button (#4454). Fixes #4489 Fixes #4472 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
9dc41e7adf |
feat(reader): reference page numbers from EPUB page-list with manual page count fallback (#4549)
Add a 'Reference Pages' reading progress style that shows physical book page numbers in the footer progress info: - When the book carries a page list (EPUB3 nav page-list or EPUB2 NCX pageList — foliate-js already parses both and resolves the current pageItem on relocate; it was just never consumed), display the current page label and use the highest numeric label as the total, so a trailing roman-numeral index page can't corrupt the total (#672). - When the book has none, a per-book 'Reference Page Count' input appears; the reading fraction is mapped linearly onto the entered count (#4542). The count is saved per book only and never propagates to global view settings. - Falls back to percentage display when neither source is available. Verified with the sample books from #672: Caleb's Crossing (EPUB3 page-list, 419 pages) and Count Zero (EPUB2 NCX pageList/page-map, 346 pages — chapter 2 lands exactly on page 22 per its page-map), plus a stripped no-pagelist copy for the manual-count path (175/350 at 50%). Closes #672 Closes #4542 Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Huang Xin
|
ceddee3793 |
feat(library): search a book on Goodreads from the library and reader (#4543) (#4548)
Adds a quick "Search on Goodreads" action so readers can jump straight to Goodreads to track a book instead of retyping the title there. - Library: a Goodreads button in the Book Details view (works on web, desktop and mobile) searching the book's title + author, plus a "Search on Goodreads" item in the desktop right-click context menu. - Reader: Goodreads is added as a built-in web-search provider so highlighted text (e.g. a short-story title inside a magazine) can be looked up on Goodreads. Disabled by default like the other built-ins; enable it in Settings -> Dictionaries. Both surfaces are used because the native context menu is desktop-only; the Book Details button covers web and mobile. Adds a shared openExternalUrl() helper and translates "Search on Goodreads" across all locales (the Goodreads brand name is kept verbatim). Closes #4543 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
390c711070 |
feat(rsvp): configurable start delay, word stepping, context dictionary lookup, and keyboard shortcut (#4541)
- #4478: add a configurable pre-start countdown (Off / 1s / 2s / 3s, default 3s) that now ticks at honest one-second intervals and applies to start, resume, and page loads; Off starts instantly. - #4476: add manual next/previous word controls (buttons flanking Play plus the "," / "." keys) that pause playback and step exactly one word. - #4475: allow selecting text in the context panel to look it up in the dictionary (anchored popup on desktop, bottom sheet on small screens), reusing the reader's dictionary view; auto-scroll/seek are suppressed during selection and an outside click dismisses the popup. - #4473: add a Speed Reading keyboard shortcut (Shift+V), shown on the View menu item; ignore repeat triggers while a session is active. Adds i18n strings for the new UI across all locales. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
loveheaven
|
1e26c5d765 |
fix(nav): bound section-scan concurrency to keep zip.js writers from ERRORED-ing (#4528)
On Tauri, section.loadText() drives a plugin-fs open/read/close round trip per call. The unbounded Promise.all in computeBookNav and enrichTocFromNavElements can fire 200+ concurrent IPC chains against a long-spine EPUB, saturate the JS↔Rust bridge or the fd pool, and cause individual reads to reject. The zip.js TextWriter then transitions to ERRORED, surfacing as 'Cannot close a ERRORED writable stream' and silently dropping TOC fragments for the affected sections. In the worst case the rejection propagates through Promise.all and prevents the reader from opening the book.
Hoist the OPDS module's runWithConcurrency to utils/concurrency.ts (zero behaviour change for OPDS) and reuse it in computeBookNav and enrichTocFromNavElements, capped at 128. The cap was binary-searched against the worst-case repro (Android emulator + dev mode + 250-section EPUB): 30/64/128 pass, 200 fails. Section-internal loadText/createDocument dedupe is unchanged.
The worker pool also isolates per-section failures: the outcome shape ({item,result}|{item,error}) lets us log and skip the offending section instead of aborting the entire build as Promise.all did. Even if a future workload pushes past the cap, the reader still opens.
|
||
|
Huang Xin
|
82bd90afc5 |
feat(reader): random-access file reads on Android via rangefile scheme (#4534)
* feat(reader): random-access file reads on Android via rangefile scheme NativeFile's per-chunk Tauri IPC (open+seek+read+close) is slow on Android, and RemoteFile can't replace it because the WebView mishandles Range requests on intercepted custom-protocol responses — it re-applies the offset to the already-sliced body, so any non-zero-start range returns corrupt data or net::ERR_FAILED (Chromium 40739128, tauri-apps/tauri#12019/#3725). Add a `rangefile` custom URI scheme that carries the byte range in the URL query (?path=&start=&end=) instead of a Range header. With no Range header the WebView delivers the 200 body verbatim, while bytes still stream through the network stack rather than the IPC bridge. The handler is scope-gated by asset_protocol_scope (same boundary as the asset protocol) plus an explicit traversal/NUL/relative guard. RemoteFile.fromNativePath() drives the scheme on Android (query-carried range, X-Total-Size for size); nativeAppService.openFile routes Android reads through it with a NativeFile fallback. Verified on-device (Android 16 / WebView 147) via CDP: byte-equal reads at every offset, ~1.8x faster small scattered reads, real book opens/renders; all out-of-scope/traversal/NUL paths rejected 403. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * ci(rust): run cargo unit tests in rust_lint The rust_lint job ran only fmt + clippy, so the crate's ~40 Rust unit tests (parsers, parser_common, and the new range_file tests) never executed in CI. Add `cargo test -p Readest --lib` to rust_lint — the frontend dist is absent there, but generate_context! already compiles without it (clippy proves this) and the unit tests run headless. Also add a `test:rust` pnpm script and document it as verification done-condition #6. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
loveheaven
|
11d796361e |
perf(import+open): native Rust EPUB/MOBI parser, OPF prefetch, parallel TOC enrichment (#4369)
* perf(epub): add native EPUB parser in Rust
Introduce a Rust-side EPUB pre-parser exposing three Tauri commands:
* parse_epub_metadata - title/author/cover + partialMD5 in one
shot, for the import hot path
* parse_epub_full - OPF + nav.xhtml + toc.ncx bytes plus a
manifest size table, for the reader open
hot path
* extract_epub_cover_full - full-resolution cover bytes, for the
lock-screen wallpaper writer
All three avoid ferrying multi-MB blobs across the JS<->Rust IPC
boundary. Cover bytes returned by parse_epub_metadata are downscaled
to a webview-friendly JPEG when the long edge exceeds the library
thumbnail size.
No JS callers yet -- wired up in the following commits.
* perf(import): use native EPUB parser and downscale covers on Tauri targets
On Tauri (desktop/iOS/Android), importBook now forwards EPUB
metadata + cover extraction to the Rust parse_epub_metadata
command and reuses the partialMD5 it returns, skipping the
foliate-js full archive parse and the second pass over the file
for hashing.
As a side effect, the cover written to cover.png is downscaled
to a webview-friendly JPEG (long edge <= 512px), shrinking the
on-disk thumbnail from multi-MB to ~30-60KB per book. To keep
the lock-screen wallpaper feature unchanged, useAutoSaveBookCover
now pulls the original full-resolution cover via the Rust
extract_epub_cover_full command instead of copying the (now
downscaled) cover.png; falls back to the thumbnail when the
native path is unavailable.
Web targets and non-EPUB formats keep the existing path.
* perf(reader): prefetch EPUB OPF/nav from Rust on book open
When opening an EPUB on Tauri targets, DocumentLoader now calls the
Rust parse_epub_full command up-front to pull the OPF, EPUB3 nav,
NCX and the central-directory size map in a single IPC. The
foliate-js zip loader is wrapped so that loadText() of these
entries (and a synthetic META-INF/container.xml) is served from
that in-memory cache without inflating through zip.js, while
all other assets keep flowing through the original loader.
A small in-flight dedupe is added to the spine-text loader so the
nav pipeline (loadText + createDocument back-to-back on the same
href) doesn't pay for two zip.js inflate calls per chapter on
first open.
Reader store / app service plumbing: readerStore.openBook now
resolves an absolute on-disk path via the new
appService.resolveNativeBookFilePath / bookService.resolveNativeBookFilePath
helper and threads it into DocumentLoader as nativeFilePath so
the prefetch can fire. Web targets, non-EPUB formats and books
without a managed/external on-disk path skip the prefetch and
take the original code path.
* perf(nav): parallelize section scans and memoize fragment lookups
computeBookNav now processes sections via Promise.all instead of
a sequential for-loop, and within each section issues loadText()
and createDocument() concurrently. Combined with the in-flight
loadText dedupe added to the zip loader, each chapter pays for a
single zip inflate per nav build, and the inflates of different
chapters overlap.
enrichTocFromNavElements is restructured into two concurrent
phases: a cheap '<nav' substring filter on the inflated text, and
a parsed-document walk for the survivors. Most chapters fall out
in phase 1 without ever being parsed.
In fragments.ts, calculateFragmentSize now consults a
per-section position cache (makeFragmentPositionCache) so the
N-fragment loop is O(N) over the chapter HTML instead of O(N²).
A small isCfiAddressable guard is added to skip elements that
foliate-js's CFI generator can't address (documentElement, body
itself, detached nodes, nodes outside <body>) — these previously
threw and spammed console.warn for every fragment, now they
silently fall back to the section CFI.
* perf(import): use native MOBI/AZW/AZW3 parser on Tauri targets
On Tauri (desktop/iOS/Android), importBook now forwards
MOBI/AZW/AZW3/PRC metadata + cover extraction to the Rust
parse_mobi_metadata command and reuses the partialMD5 it returns,
skipping the foliate-js full-buffer parse and the second pass over
the file for hashing. Mirrors the existing EPUB native fast-path
added in e3fc4767 — bookService tries EPUB first, then MOBI; both
bridges fall back to the foliate-js DocumentLoader when the native
path is unavailable (web target, parse error, format mismatch).
The new mobi_parser is built on the mobi crate (KF7+KF8 reader,
zero JS-side touch). It reads title, author, publisher, ISBN, ASIN,
publish date, language, subjects and description from the MobiHeader
+ EXTH records, resolves the EXTH 201 cover offset against the PDB
image-record table (with ThumbOffset / first-image fallbacks), and
strips KindleGen's HTML wrapping in EXTH 103 so the description goes
into the library DB as plain text. The parsed cover is funneled
through the same maybe_resize_cover path as EPUB, so MOBI library
thumbnails are also clamped to a 512px-long-edge JPEG.
Cover-resize / partialMD5 / RawCoverImage are extracted into a new
parser_common module shared between epub_parser and mobi_parser, so
a single tweak (e.g. raising the thumbnail target) applies to every
native importer and the partialMD5 implementation can't drift between
the two paths (a divergent algorithm would silently re-import every
existing book under a new hash on the first run).
Web targets and non-Kindle formats keep the existing path.
* test(tauri): verify native Rust EPUB parser parity with foliate-js
Add a Tauri WebView parity suite (epub-parser-parity.tauri.test.ts) that
cross-checks the native Rust parser against foliate-js on the same fixtures:
parse_epub_metadata / parse_epub_full (title, author, language, identifier,
publisher, published, subjects, partialMD5, OPF + per-entry size table), and
that opening with the native prefetch produces the same BookDoc and
computeBookNav (TOC) output as the pure-JS path.
Fix a parity divergence the suite caught: the Rust OPF parser mapped
dcterms:modified onto `published`, but foliate-js keeps them separate and
leaves `published` empty -- so EPUB3 books carrying only the mandatory
dcterms:modified got a bogus publication date on the native import path. Map
only dc:date now; add regression tests.
Test infra:
- vitest.tauri.config.mts: add optimizeDeps (mirroring vitest.browser.config)
so foliate-js-importing tauri tests load -- otherwise esbuild's dep scan
can't resolve '@pdfjs/pdf.min.mjs', pre-bundling is skipped, and the CJS
deps fail to import ("Importing a module script failed").
- capabilities-extra/webdriver.json: fix __test__ -> __tests__ fs scope typo
so import tests can open fixtures under src/__tests__/.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(import): foliate-js owns EPUB/MOBI metadata via standalone extractors
Rust contributes only the mechanical work that's expensive on a
WebView — partialMD5, the downscaled cover, and (for EPUB) the raw
OPF bytes Rust already had to read for cover resolution. Metadata
extraction is delegated to foliate-js's two new standalone entry
points (`parseEpubMetadataFromXML`, `readMobiMetadata`) so the
import-path BookDoc and the reader-path BookDoc share a single
parser implementation.
EPUB
- `parse_epub_metadata` returns
`{ partialMd5, cover, coverMime, opfPath, opfBytes }`. OPF bytes
are a free byproduct of the cover-resolution scan.
- `tryNativeParseEpub` runs `parseEpubMetadataFromXML` on the OPF
bytes and assembles a lightweight BookDoc stub (metadata +
getCover). The importer doesn't drive `DocumentLoader.open()`, so
no zip central-directory scan, no nav/ncx inflate, no spine walk.
- `coverMime` is preserved so `bookService.importBook`'s
`cover.type === 'image/svg+xml'` branch still routes SVG covers
through svg2png.
MOBI / AZW / AZW3 / PRC
- `parse_mobi_metadata` returns `{ partialMd5, cover, coverMime }`.
`tryNativeParseMobi` runs foliate's `readMobiMetadata` on the
same File, which uses `MOBI.open(file, { metadataOnly: true })`
to parse PalmDB + MobiHeader + EXTH and short-circuit before the
MOBI6 / KF8 init() that walks every text record.
- `Book.metadata.identifier` is foliate's `mobi.uid.toString()`
(PalmDB UID), the canonical MOBI identifier the reader path uses.
bookService.importBook
- EPUB and MOBI native branches consume the bridge's BookDoc stub
directly. The stub's `getCover()` returns the Rust-downscaled
blob, falling back to foliate's own `getCover` thunk when Rust
didn't extract a cover.
Other
- Drop the unused `base64` Rust dependency: cover bytes go over IPC
as `Vec<u8>` (Tauri 2 transports them natively, like opfBytes /
navBytes / ncxBytes).
- Drop the `nativePrefetch` option on `DocumentLoaderOptions`; no
caller passes it. `nativeFilePath` keeps driving `parse_epub_full`
on the open hot path.
Tests
- vitest.tauri parity test asserts byte-equal partialMD5, cover
presence parity, OPF bytes that decode to a real `<package>`
document, and that `parseEpubMetadataFromXML` on those bytes
produces the same user-visible metadata fields (title / author /
language / identifier / published) as `DocumentLoader.open()`.
* test(tauri): add War and Peace MOBI fixture for native parser parity
The .tauri parser-parity suite previously had no .mobi/.azw3 asset, so the native MOBI parser (metadata + EXTH cover resolution) was uncovered. Adds a real KF8 MOBI ("War and Peace") to enable MOBI parity coverage against foliate-js.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* chore(foliate-js): bump submodule to readest/foliate-js main (91191ca)
Replaces the ad-hoc 02f435a with the merged main commit 91191ca, which lands the standalone OPF/MOBI metadata extractors (parseEpubMetadataFromXML, readMobiMetadata) the import fast-path depends on (foliate#19), plus the RTL multi-view rect-mapper fix (foliate#20). The extractor code is byte-identical to 02f435a, so the bridges are unaffected.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
Huang Xin
|
8425d0b91f |
fix(opds): render HTML in publication descriptions (#4510)
* fix(opds): render HTML in publication descriptions, closes #4503 OPDS publication descriptions showed raw HTML tags (literal `<p>`, `"`, `'`) instead of rendering them. Some aggregator feeds serve the description as an Atom `type="text"` summary whose HTML has been escaped twice; foliate's getContent only un-escapes `type="html"`/ `"xhtml"`, so the markup survives parsing as entity text and the detail view dumped it straight into an unsanitized `dangerouslySetInnerHTML` (also an XSS sink for untrusted feed content). Add `getOPDSDescriptionHtml`: decode one extra entity level only when the value is entirely escaped markup (mixed content like `<p>see <code>` is left literal), then sanitize with the shared DOMPurify sanitizer. Wire it into PublicationView and render the sanitized HTML. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * refactor: consolidate HTML sanitizers into @/utils/sanitize sanitizeHtml/sanitizeForParsing are generic DOMPurify wrappers, not specific to Send-to-Readest. Now that OPDS description rendering also needs sanitizeHtml, move them out of services/send/conversion into the shared @/utils/sanitize module (alongside sanitizeString) so neither consumer reaches across the other's feature boundary. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
1eaf16ffc2 |
fix(opds): tolerate junk after document element in feeds (#4479) (#4506)
The Hungarian MEK catalog (a PHP backend) returns a valid Atom feed
followed by trailing junk after </feed> — a stray PHP warning, an extra
tag, or text. Chrome's DOMParser ignores it, but Firefox's strict parser
fails with "junk after document element" and replaces the whole document
with a <parsererror>. The reader then sees a non-feed root, treats the
response as HTML, finds no OPDS link, and silently navigates back, so
browsing the catalog on Firefox web is broken on nearly every subpage.
Add parseOPDSXML(): on a parser error, re-parse the slice from the root
element's start tag to its last matching end tag, dropping any leading
prolog and trailing junk. If recovery still fails the original error
document is returned, so callers fall through to their existing
HTML/non-OPDS handling. Wire it into the three OPDS XML parse sites:
the reader (page.tsx), validateOPDSURL (adding a catalog), and the
subscription/auto-download feed checker. feedChecker also switches its
text.startsWith('<') detection to looksLikeXMLContent so the MEK feed's
leading newlines (no <?xml?> declaration, #4181) are recognized.
jsdom mirrors Firefox's strict behavior (same parsererror namespace), so
the regression tests run in the normal unit suite.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
loveheaven
|
4eeed74cdd |
fix(webdav): always sync book covers, not just when syncBooks is on (#4445)
Cover uploads were nested inside the `syncBooks` toggle in both the
batch path (`syncLibrary`) and the per-book reader path
(`pushBookFileNow`). With the default `syncBooks: false`, covers
silently never reached the WebDAV server even though `config.json`
did, so receiving devices ended up with progress + notes synced but
no shelf art.
Covers are conceptually metadata, not bytes:
- they're tiny (~30–60 KB after the import-time downscale);
- they cannot be regenerated on a fresh device that doesn't hold
the book bytes (custom covers from metadata services in
particular are completely unrecoverable without sync);
- cloudService already treats them as metadata-grade in its
download path (`downloadBookCovers`, `downloadBook(onlyCover)`).
Two changes:
1. `WebDAVSync.ts::syncLibrary` — moved `pushBookCover` out of the
`if (options.syncBooks)` block; it now runs alongside
`pushBookConfig`, before `pushBookFile`. Step ordering in the
header doc-comment was updated to match.
2. `useWebDAVSync.ts` — extracted a standalone `pushBookCoverNow`
callback (gated only on `allowPush`, with its own
`coverSyncedRef` for per-instance dedupe), and dropped the cover
ride-along that lived at the tail of `pushBookFileNow`. The
open-book effect now fires `pushBookCoverNow` and
`pushBookFileNow` in parallel via `Promise.all` (different remote
paths, no reason to serialize), and the manual-push event handler
triggers both independently.
The WebDAV pull path was already independent of `syncBooks`, so no
changes are needed there — receiving devices will pick up the newly
mirrored covers automatically.
|
||
|
Huang Xin
|
e4bb9fc4b7 |
refactor(share): make saveFile content nullable for path-based shares (#4424)
The book "Send" flow had to pass `new ArrayBuffer(0)` to saveFile purely to satisfy the type-checker: the content arg is ignored on the native share path when `options.filePath` points at an already-on-disk file. Widen saveFile's content parameter to `string | ArrayBuffer | null` across the AppService contract so callers can hand off a file by path without buffering it into memory, and pass `null` from the Send flow instead of a throwaway empty buffer. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
loveheaven
|
fe853554a9 |
feat(library): send book file from bookshelf selection popup (#4402)
* feat(library): send book file from bookshelf selection popup
Adds a Send button to the bottom popup that appears when one or more
books are selected on the bookshelf. Hands the actual book file
(epub/pdf/...) to the OS share sheet via tauri-plugin-sharekit
(UIActivityViewController on iOS, Intent.ACTION_SEND on Android,
NSSharingServicePicker on macOS), so users can fire the file off to
Mail / Messages / WeChat / AirDrop / etc.
This is intentionally distinct from the per-item context-menu "Share
Book", which uploads the book to the readest backend and generates a
public link. "Send" is offline file egress; "Share Book" is remote
collaboration. They share zero infra.
Resolution rules mirror bookContent.resolveBookContentSource: managed
copy under Books/<hash>/ first, then the device-local in-place import
path. Cloud-only books warn rather than silently no-op.
Path is handed to shareFile via options.filePath. Without that,
saveFile() falls back to writing a temp copy under BaseDirectory.Temp,
which on Android resolves to /data/local/tmp/ — the app sandbox has
no write permission there and the call fails with EACCES ("failed to
open file at path: /data/local/tmp/...epub Permission denied (os error
13)"). Passing the absolute path also avoids re-buffering the entire
epub/pdf into memory.
On macOS the NSSharingServicePicker is anchored to the selected
book's cover rather than to the Send button — the user's visual
focus is on the cover they just tapped, not on the bottom toolbar.
BookshelfItem stamps a data-book-hash attribute on its root div so
the Send handler can locate the cell via querySelector and pass its
rect through saveFile's sharePosition option. preferredEdge='bottom'
maps to NSMinYEdge, so the popover renders above the cover (and only
auto-flips below when there's no room above). iOS / Android share
sheets are modal and ignore sharePosition, so the same code is a
no-op there.
The button is hidden on Linux (no system share sheet), Windows
(WebView2 share UI deadlocks the main thread, see #4343), and web
browsers (no "send file to <app>" affordance for arbitrary downloads).
* fix(share): don't fall back to saveDialog when shareFile is cancelled
The native saveFile({ share: true }) path used to swallow any error
from sharekit's shareFile() and fall through to saveDialog. The plugin
treats user cancellation the same as a failure (it rejects with
'Share cancelled' on Android when the user dismisses the share sheet),
so cancelling a share popped up an unwanted 'Save As...' dialog right
after the user explicitly chose not to share.
Mirror what webAppService already does for the navigator.share()
AbortError path: once we entered the share branch, return true
regardless of whether the share completed or was cancelled. The
saveDialog path is now reserved for Linux/Windows desktop, which never
hit the share branch in the first place (wantShare gates them out).
If a future caller wants 'try share, fall back to save on hard
failure', that decision belongs to the caller — saveFile shouldn't
silently override an explicit share intent.
|
||
|
loveheaven
|
7283a8ac21 |
fix(android): open book directly when launched via 'Open with' (#4407)
On Android, tapping an EPUB/MOBI/AZW3 in the system file browser and choosing Readest only opens the library — the book itself never opens in the reader. Now Readest can open the book now. |
||
|
wfjack
|
5ff18b8f32 |
fix(readwise): use Tauri HTTP transport in desktop app (#4413)
Switch ReadwiseClient to @tauri-apps/plugin-http when running in Tauri desktop mode, matching the pattern already used by WebDAV, Hardcover, AI providers, translators, OPDS, and KOReader sync. In a Tauri webview context the standard window.fetch is still subject to CORS preflight rules and—on Android—the platform's cleartext-traffic policy. @tauri-apps/plugin-http sends the request from the Rust side via reqwest, bypassing the renderer entirely (no Origin header, no preflight, no cleartext block). ReadwiseClient was one of the few remaining services that had not yet adopted this transport, so users on the desktop build could hit CORS or network errors when validating tokens or pushing highlights. Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
3a81e09911 |
fix(reader): scroll oversized blocks in-place instead of turning the page (#4400) (#4415)
Wide or tall tables, code blocks and display equations overflowed the reading column and a scroll gesture over them turned the page instead of scrolling the content (#4400). - Wrap tables and display equations in a horizontally/vertically scrollable container; route touch + wheel along the box's scrollable axis so it scrolls the box and never turns the page, even at the edge (both axes). - A box that fits its column is marked fit (overflow:visible) so it never clips or captures gestures; the fit decision is measured once after layout via a self-disconnecting ResizeObserver, so it never relayerizes during a page turn. - The scroll wrapper carries a new cfi-skip attribute that makes it transparent to CFI: epubcfi.js hoists a cfi-skip node's children into its parent (unlike cfi-inert which drops the subtree), and xcfi.ts mirrors this for CFI<->XPointer so existing highlights, bookmarks and KOSync positions inside a wrapped table or equation still resolve. The sanitizer whitelists cfi-skip. - Bump foliate-js submodule (cfi-skip support + raf fallback for large sections). Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
ef603852b7 |
feat(tts): hotkey to highlight the currently-spoken sentence (#4085) (#4368)
Add a "Highlight Current Sentence" keyboard action (default Shift+M, in the Text to Speech shortcut section) that persists the sentence TTS is reading aloud as a normal highlight using the user's default style/color — no text selection, eyes-off, silent, and idempotent (a repeat press on the same sentence is a no-op rather than a duplicate). Flow: the shortcut handler in useBookShortcuts dispatches tts-highlight-sentence → useTTSControl (which owns the TTSController) resolves the current sentence via the new TTSController.getSpokenSentence() and relays create-tts-highlight → Annotator builds the BookNote with the pure, unit-tested buildTTSSentenceHighlight helper and persists/renders it like any other highlight. Closes #4085 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
6605ae8242 |
feat(dictionary): import companion MDD files that share the MDX prefix (#4363)
A single MDX commonly ships its resources across several MDD files sharing the MDX's filename prefix (e.g. `Name.mdd` for images, `Name-02.mdd` for scripts, `Name-03.mdd` for audio). The importer only grouped the exact-stem `Name.mdd`, so the other MDDs were dropped as orphans and their resources (notably audio) could never be loaded. `groupBundlesByStem` now attaches every `.mdd` whose stem starts with an `.mdx` stem at a separator boundary to that MDX bundle (longest prefix wins on overlap); the boundary check prevents false merges like `dict.mdx` claiming `dictionary-words.mdd`. The runtime provider, contentId, and replica sync (binary upload + manifest apply) already treat `files.mdd` as a list, so multi-MDD bundles sync across devices with no further changes. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
78794499a2 |
fix(dictionary): correct System Dictionary platform gating on web and iPad (#4362)
* fix(dictionary): keep other dictionaries usable when System Dictionary syncs to an unsupported platform `dictionarySettings.providerEnabled` is whole-field synced across devices, so enabling System Dictionary on macOS/iOS sets the flag on web/Linux/Windows too. There the row is hidden and the feature is a no-op, but the settings UI read the raw flag and locked every other dictionary's toggle read-only. Gate the lock on `isSystemDictionaryEnabled(settings)` — the same platform-aware check the annotator uses — so it matches real lookup behavior and never triggers where the system dictionary can't run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(dictionary): dispatch system dictionary handoff by native OS (fixes iPad) iPadOS sends a desktop "Macintosh" user agent, so the UA-based `getOSPlatform()` reported iPad as 'macos' and the handoff invoked the macOS-only `show_lookup_popover` Rust command that iOS never registers ("Command show_lookup_popover not found"). Derive the OS from the app service's `is*App` capability flags (sourced from the Tauri OS plugin, correct on iPad) via a new synchronous `getInitializedAppService()` accessor, so iPad routes to the iOS plugin command path. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(ui): constrain reader View Options dropdown to h-8 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(agent): note System Dictionary platform-detection and synced-flag patterns Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
36e11de332 |
feat(reader): swipe-to-adjust brightness gesture on mobile (#3021) (#4356)
* docs: design spec for gesture-based brightness control (#3021) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs: revise brightness-gesture spec per /autoplan review (#3021) CEO+Design+Eng dual-voice review. Key fixes: capture-phase listener (bubble-phase could not suppress foliate paginator), opt-out toggle, 18px threshold, selection guard, brightness seed race, rAF teardown, e-ink stepped overlay, contrast capsule, perceptual curve reuse, listener-level test harness. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(reader): swipe-to-adjust brightness gesture on mobile (#3021) Left-edge vertical swipe adjusts screen brightness on iOS/Android, with a Sun-icon progress overlay. Capture-phase non-passive listener suppresses the foliate paginator / page-flip / UI-toggle handlers; selection guard, strip reservation in scrolled mode, eager brightness seed, rAF throttle + teardown. Opt-out toggle in Settings > Behavior > Device (default on). Perceptual curve shared with the menu slider. Pure-helper + listener-level tests. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(reader): detect brightness-swipe edge by screenX; i18n + shorter label (#3021) On-device fix: paginated mode lays the iframe doc out as wide side-by-side columns, so clientX/documentElement.clientWidth are document coordinates and a left-edge touch on a later page never fell inside the strip (armed stayed false). Detect with screenX against the parent window width, matching usePagination. Also: translate the two new setting strings across all locales and shorten the toggle description to 'Slide along the left edge'. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
2f5e583653 |
feat(annotations): configurable export link type + dedicated Import Annotations modal (#4350)
* feat(export): make annotation export link type configurable Add an Annotation Link selector (App / Web) to the Export Annotations dialog. Defaults to the app deeplink in the native app and the universal web link on the web, so web exports no longer emit readest:// links that only the desktop/mobile app can open. The default markdown template now uses the configurable annotation.link variable. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(annotations): move Moon+ Reader import into a dedicated Import Annotations modal Replace the single 'Import from Moon+ Reader' menu item with an 'Import Annotations' entry (below 'Export Annotations') that opens a dedicated modal listing import sources. Currently lists Moon+ Reader; the boxed-list layout makes adding future providers a one-row change. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
ce0ab5cc61 |
feat(library): add secondary "Then by" sort with smart defaults (#4347)
Adds a primary/secondary sort pair so users can group by author and have each author's books drilled-in list sort by series, without touching the sort menu each time. Closes #4307. - New "Then by..." picker in the library view menu (None + same keys as primary). Secondary acts as tiebreaker for the global sort, and as the in-group ordering when the user drills into a non-series group. - Smart defaults derived from groupBy, surfaced as "(Auto)" in the menu and resolved at sort time so user picks are never overwritten: - groupBy=Author + secondary=none -> Series - groupBy=Series + librarySortByAuto -> primary becomes Series - librarySortByAuto flips off as soon as the user makes any explicit primary pick; subsequent groupBy changes then respect that choice. Settings: librarySortBy2, librarySortByAuto. URL: ?sort2 syncs the secondary; auto is settings-only (no URL representation). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
10a223b0e2 |
fix(export): export uses save dialog on Windows to avoid share UI freeze (#4343) (#4346)
Windows WebView2's native share UI, invoked via tauri-plugin-sharekit, synchronously blocks the Tauri command thread waiting on ShareCompleted/ShareCanceled callbacks. When the user dismisses the picker without those callbacks firing, the app freezes and has to be killed from Task Manager. Skip the native share path on Windows and fall through to the save dialog, mirroring Linux. |
||
|
loveheaven
|
3c134380b7 |
feat: add empty state hints and loading indicators for annotations, bookmarks, notes, font import, and Moon+ Reader import (#4338)
* feat: add empty state hints and loading indicators for annotations, bookmarks, notes, font import, and Moon+ Reader import - BooknoteView: show 'No annotation yet' / 'No bookmark yet' when empty - Notebook: show 'No note yet' when no notes/excerpts exist - Annotator: add loading overlay with spinner during Moon+ Reader import - mrexpt: yield to event loop every 5 entries to keep spinner animating - CustomFonts: show in-place loading card during font import, spinner transitions to font name without layout jump * fix(ui): respect e-ink overlay styling and drop dead className branch - Annotator: use modal-box on the mrexpt import overlay so eink picks up the no-shadow + 1px border override automatically. - CustomFonts: collapse importing-card clsx ternary whose branches were identical into a flat className. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Huang Xin <chrox.huang@gmail.com> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
744d5b3a03 |
fix(dict): restore MDD eager init; CSP-safe audio handler; gate binary uploads on sync category (#4337)
* fix(dict): restore MDD eager init; CSP-safe Vocabulary.com audio handler
Two MDict fixes:
1. **Regression**: #4334 turned on `MDD.create(..., { lazy: true })` so
the audio / resource side of every MDict would skip the upfront
key-block decode + sort. That's the right trade-off on the MDX side
(millions of headwords, ~80 s init saving), but it doesn't transfer
to MDDs: js-mdict's lazy lookup binary-searches `keyInfoList`
envelopes with `comp` (localeCompare), and MDDs store keys in
byte-sorted order with `\` prefixes and case folding. The two orders
disagree just often enough that `mdd.locateBytes('sound.png')`
misses on resource paths that do exist — the icon's CSS
`background-image: url(sound.png)` then stays unresolved and the
speaker glyph disappears. Bisected to
|
||
|
Huang Xin
|
648c35b334 |
feat(reader): add disableSwipe option to disable swipe-to-paginate (#4335)
Issue #4288: users with hardware page-turn buttons (e.g. e-ink readers) want to disable swipe-to-paginate so accidental finger drags during highlight selection don't flip the page mid-annotation. The existing "Tap to Paginate" toggle only covers taps; swipe was always on. - New `disableSwipe: boolean` on `BookLayout` (default `false`, declared right after `disableClick`). - foliate-js submodule bump: paginator gates `#onTouchMove` and `#onTouchEnd`'s snap-to-page on a new `no-swipe` attribute, so native touch behaviour (text selection) stays intact. - `FoliateViewer` sets/removes the `no-swipe` attribute alongside `animated`, and the `ControlPanel` toggle pushes the change to the live renderer so it takes effect without a viewer reset. - The fixed-layout swipe interceptor in `usePagination` also bails when `disableSwipe` is on, covering both reflowable and fixed- layout books. - New "Swipe to Paginate" UI row directly below "Tap to Paginate"; both can be off simultaneously. - i18n: 33 locales translated. Also polish: rephrase the two helper texts under "Read books in place" in `ImportFromFolderDialog`. The previous copy ("Copy no book into the library to save space.") used an awkward double-negative; the new wording is clearer and the locked variant drops "registered as" / em-dash for a plain two-sentence form. i18n updated. Closes #4288 Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
93abca8960 |
feat(dict): faster MDict/StarDict import + lazy lookup; raw .dict; UX (#4334)
Make the dictionary import path usable on large bundles and bring the multi-device flow up to par. Import perf - Skip `MDX.create()` at import time. The factory triggers full init — decompresses every key block and sorts millions of keys with localeCompare just to expose the header. Replace with a tiny `readMdxHeader()` that only reads the small XML header for Title / Encoding / Encrypted (saves ~17 s on a 250 MB MDX on web). - `partialMD5`: read the 9 sample slices in parallel rather than sequentially. Each freshly-picked-File slice round-trip on Chrome costs ~100 ms cold; parallelisation collapses 9 of them into one. - Native fast-path in `nativeAppService.writeFile`: when the source is a `NativeFile`, delegate to Tauri's `copyFile` rather than streaming the file through `NativeFile.stream()`. Streaming a 250 MB body through 1 MB IPC chunks on Android took ~100 s; native copy is bound by disk throughput instead. Exposes `NativeFile.getNativeLocation()` for the FS layer to use the underlying path + baseDir directly. Lazy lookup - Pass `lazy: true` to `MDX.create` / `MDD.create`. The js-mdict change in this PR skips the upfront decompress-every-block + sort during init (~80 s on the same 250 MB bundle) and decodes only the relevant key block on demand per lookup. First-lookup main-thread block drops from ~81 s to ~230 ms. (Closes #4228.) Raw .dict - Drop the import-time gate that flagged non-gzip dict bodies as `unsupported`. The runtime body loader (`loadDictBody`) already probes the gzip header and falls through to a passthrough buffer for raw files, so the gate was the only thing preventing raw `.dict` bundles from importing on devices that received them via cloud sync. (Closes #4179, partially addresses #4248.) Import-flow UX - `handleImport` now always surfaces a toast for every non-cancelled attempt: picker errors, missing app service, no-op imports, and unsupported-but-imported bundles each get their own message instead of failing silently. - Call `markAvailableByContentId(newDict.contentId)` after add/replace so the "Bundle is missing on this device" warning clears immediately — no need to close-and-reopen the panel. System Dictionary - Drop the cascading toggle behavior in `setEnabled`. Each provider's enabled flag persists independently; exclusivity is enforced at lookup time. Toggling System on/off no longer wipes the user's preferred set of in-app providers. - Render non-system rows as read-only when System is on (toggle still shows what's queued to restore; tooltip explains the lock). - `isSystemDictionaryEnabled` short-circuits to `false` on platforms where the handoff isn't implemented. `providerEnabled` is whole- field synced across devices, so a flag set on macOS would otherwise leak to a Windows device with no way to look up a word. js-mdict - Submodule bump to e6dbc99 which adds the opt-in `lazy: true` `MDictOptions` flag (skip `_readKeyBlocks` + post-init sort; new `lookupKeyBlockByWordLazy` path on `MDX` and `MDD`). Eager mode is unchanged and every existing js-mdict test still passes. i18n - 208 new translations across 33 locales for the new UX strings. Closes #4228 Closes #4248 Closes #4179 Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
cf44e85180 |
fix(reader): fit duokan-page-fullscreen cover image without cropping (#4328)
Closes #3914. Switch the page-fullscreen image to object-fit: contain (and SVG preserveAspectRatio meet) so cover images fit the page and center without cropping. Also adds a duokan-image-gallery-cell layout helper and drops the mobile marginBottomPx override. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
loveheaven
|
64651a65ef |
fix(sync): skip replica upload when not authenticated (#4327)
Prevents 'Please log in to continue' error when importing fonts or images locally without logging in. The replica upload is now gated on getAccessToken() so unauthenticated users are never enqueued. |
||
|
loveheaven
|
ff605e000d |
feat(library): in-place import from registered external folders (#4315)
* feat(library): in-place import with cloud sync and symmetric local delete
Adds an `inPlace` option to importBook so a source file inside a
registered external library folder is referenced directly via
`book.filePath` instead of being copied into Books/<hash>/. Sidecars
(cover, config, nav) still live under Books/<hash>/.
ingestService routes through shouldImportInPlace, which marks an
import in-place when the absolute source path lives under any of
`settings.externalLibraryFolders` and is NOT inside a per-root
`Books/` subtree. The Readest data dir (`customRootDir`) is
intentionally excluded — that directory is Readest's home and
should freely hold hash copies; in-place is for user-registered
roots (Duokan, Calibre, Moon+ Reader, an iCloud mirror, …).
Cloud sync treats in-place books as first-class:
- uploadBook reads bytes from (book.filePath, 'None') when set.
The cloud key is unchanged, so a peer downloading the book
lands it under Books/<hash>/ as a normal hash copy.
- useBooksSync strips `book.filePath` before pushing — it is a
device-local path that is meaningless on any other device.
- ingestService no longer skips upload for in-place books;
autoUpload / forceUpload behave like any other book. Only
transient imports opt out.
- deleteBook 'local'/'both' now physically removes the source
file at book.filePath (base 'None'). Local-delete semantics
are symmetric with hash-copy books: the local copy is gone,
the cloud backup remains, a future pull restores under
Books/<hash>/. removeFile errors are swallowed.
New `SystemSettings.externalLibraryFolders?: string[]` (no UI yet;
registration entrypoint lands in a follow-up). Added to
BACKUP_SETTINGS_BLACKLIST alongside `localBooksDir` /
`customRootDir` so device-local paths don't ride cloud backups.
Tests: cloud-service, ingest-service, and backup-settings suites
cover in-place delete, multi-root matching, per-root `Books/`
guard, and the backup-strip.
* feat(library): one-tap "read in place" toggle in folder import
Surface the in-place / copy choice as a single "Read books in place (don't copy)" checkbox in the Import-from-Folder dialog. When the user opts in, the chosen directory is registered in `settings.externalLibraryFolders` and ingestService's `shouldImportInPlace` will route the books straight to importBook with `inPlace: true` — no copy into Books/<hash>/, sync still works, local delete still removes the source file (the symmetry was set up in the previous in-place commit).
User experience:
- First-time users hit the toggle once per library folder. The choice is also persisted to localStorage so subsequent dialog opens default to whatever they picked last.
- Repeat imports from a folder that's already registered as an external library folder force the toggle ON and disable it, with a help line explaining that imports from this folder are always in-place. The check is exact-string (after path normalization) so registering /Users/me/Duokan only locks the toggle for that exact path — picking /Users/me/Downloads after Duokan still shows the toggle in its normal state.
- URL-ingress / drag-drop replays go through `runFolderImport` without the dialog and default `readInPlace: false`. They still benefit from in-place automatically when the dropped path lives under an already-registered root, because that decision is made by `shouldImportInPlace` based on settings, not by the dialog flag.
Mechanics:
- ImportFromFolderResult gains `readInPlace: boolean`. ImportFromFolderDialog gains an `initialReadInPlace` prop (seeded from the new `readest:lastImportFolderReadInPlace` localStorage key) and an `isRegisteredExternalRoot` predicate it uses to render the locked / unlocked toggle.
- runFolderImport calls a new `registerExternalLibraryFolder` helper that appends the chosen directory to `settings.externalLibraryFolders` and persists settings, but only when `result.readInPlace` is true. `isRegisteredExternalRoot` does the inverse lookup the dialog needs. Both helpers normalize paths the same way `shouldImportInPlace` does so the predicate matches the ingest layer.
- The new feature has no effect for users who never flip the toggle: `externalLibraryFolders` stays empty, the path-prefix check in `shouldImportInPlace` returns false for every import, and books continue to be copied into Books/<hash>/ exactly as before.
Self-healing for externally-removed in-place books:
Once the dialog lets users opt their library into in-place mode, the source file becomes a piece of state Readest doesn't control — another app may rewrite it (e.g. Duokan persisting reading progress into the epub), the user may move it in Finder, or an external drive may unmount between sessions. Previously, clicking such a book would navigate into the reader, fail inside loadBookContent's `fs.openFile(book.filePath, 'None')` with a low-level IO error, flash an "Unable to open book" toast, and auto-bounce back to the library — leaving the stale library record in place so the next tap reproduces the same dance.
BookshelfItem.handleBookClick now probes availability before navigating, but only for purely-local in-place books (`book.filePath && !book.uploadedAt && !book.deletedAt`). If `appService.isBookAvailable` returns false — which for in-place books means the recorded `book.filePath` no longer exists at the OS level — we dispatch `delete-books` for that hash and show an info toast explaining the removal, instead of opening the reader.
Scope is intentionally narrow:
- Cloud-synced books still flow through `makeBookAvailable`'s on-demand download path; missing local copies trigger a re-download, not a deletion.
- Hash-copy books (no `filePath` set) are not probed: a missing Books/<hash>/ file under normal use signals a bug or filesystem corruption, not user intent, and silently dropping the record would hide the real problem.
- The dispatched delete-books event reuses the existing Bookshelf deletion path, so sidecar metadata and selection state are cleaned up the same way as a user-initiated delete. For in-place books that path doesn't touch any file outside Books/<hash>/, so the now-missing source location (or whatever the user did with it externally) is left alone — symmetric with 165f15a6.
* fix(library): centralize book content resolution
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
|
||
|
loveheaven
|
5a092f16f7 |
feat(ios): folder import with security-scoped bookmark persistence (#4314)
* feat(import): support folder picker on iOS via native-bridge
Tauri's dialog plugin rejects folder picks on both mobile platforms with FolderPickerNotImplemented, so previously only Android could pick an import directory (it already routed through the native-bridge plugin's ACTION_OPEN_DOCUMENT_TREE). iOS users had no working folder-import entry point at all.
Add an iOS implementation of the native-bridge select_directory command using UIDocumentPickerViewController(forOpeningContentTypes: [.folder], asCopy: false), with a dedicated FolderPickerDelegate that:
- holds a strong reference until the picker dismisses (UIKit keeps the delegate weak), and
- calls startAccessingSecurityScopedResource on the picked URL and retains it for the app's lifetime so plain Foundation/POSIX reads against url.path work for the rest of the session.
Route NativeAppService.selectDirectory through the bridge for both iOS and Android, then call allowPathsInScopes so the picked directory is reachable via fs_scope and the asset protocol. The library page's pickImportDirectory entry point now also takes the mobile branch on iOS, while keeping the Android-only MANAGE_EXTERNAL_STORAGE prompt gated behind isAndroidApp.
* feat(ios): persist security-scoped bookmarks for picked folders
iOS hands the folder picker back a security-scoped URL whose access
right is granted only to the running process. The previous
implementation kept the URL alive for the lifetime of the process via a
static `urlsToKeepAlive` array, which worked for the current session
but forced the user to re-pick the same folder after every relaunch.
Add a `FolderBookmarkStore` that:
- Right after the picker returns, calls
`URL.bookmarkData(.minimalBookmark)` and stashes the bytes in
`UserDefaults` keyed by the POSIX path.
- On every `NativeBridgePlugin.load(webview:)`, walks every persisted
bookmark, resolves it back into a URL, and calls
`startAccessingSecurityScopedResource`. Holds the URL alive in a
process-scoped dictionary so subsequent Foundation / POSIX reads
against `url.path` succeed.
- Handles `isStale` by re-encoding the bookmark against the resolved
URL, and drops permanently unresolvable bookmarks (folder gone,
provider uninstalled) from `UserDefaults` so the next launch
doesn't re-attempt them.
Pair this with a Tauri-side change so the same paths are reachable
through both `dir_scanner::read_dir` and the fs plugin's `readDir`:
- `allow_paths_in_scopes` now has an iOS branch that widens
`fs_scope` / `asset_protocol_scope` for any path the frontend hands
it, intentionally without the desktop-side "must already be in
fs_scope" gate. The OS sandbox + bookmark store is the real
access-control boundary on iOS; widening Tauri's in-memory scope
set cannot escalate access beyond what the OS already grants. The
security comment on the command was rewritten to spell this
contract out.
- `allow_file_in_scopes` is now compiled for iOS too (previously
desktop-only) so the file-grant path is available when needed.
|
||
|
loveheaven
|
4c539e6be1 |
fix(opds): show 'Open & Read' for publications already in the library (#4313)
* fix(opds): show 'Open & Read' for publications already in the library
PublicationView only flipped the acquisition button to 'Open & Read'
when the user downloaded the book within the current component
lifecycle — the downloadedBook state started as null on every mount
and was never reconciled against the actual library. So reopening the
detail page (after navigating away in the OPDS browser, or coming back
from the reader) always showed 'Download' / 'Open Access' again and
asked the user to re-download a book they already had.
Two real-world feeds (m.gutenberg.org and ManyBooks) exposed two
distinct failure modes that made naive title/author equality unusable:
(1) Gutenberg's OPDS entries never carry <dc:identifier>. The only work
identifier lives in <atom:id>urn:gutenberg:1342:2</atom:id>, which
foliate-js's opds.js parses into metadata.id — a field
getMetadataHashInfo never reads. So the identifier candidate list
was empty and identifier-overlap matching silently skipped every
book.
(2) Author strings disagree across the boundary: the feed emits
'<author><name>Austen, Jane</name>' (Lastname-first) while the EPUB
inside the same feed ships <dc:creator>Jane Austen</dc:creator>.
Plain string equality (even after normalization) never matched.
Add findExistingBookForPublication in app/opds/utils/findExistingBook.ts
with a layered matcher:
- Pass 1: full metaHash equality (the strongest signal — same fingerprint
the bookService uses internally).
- Pass 2: identifier overlap. collectPublicationIdentifiers() splices
metadata.id into the candidate list alongside metadata.identifier, so
Gutenberg's 'urn:gutenberg:1342:2' feeds into identifierKeys(), which
extracts the '1342' digit-tail (>=3 digits, so the trailing ':2'
version suffix is ignored) and matches the EPUB's
'http://www.gutenberg.org/1342' → '1342'.
- Pass 3: tolerant title + author match. hasAuthorOverlap() does a
token-set fallback: each name is split on whitespace/comma/semicolon,
single-letter tokens (initials) and year-range tokens ('1775-1817')
are dropped, and we require >=2 shared tokens by default. So
'Austen, Jane' ↔ 'Jane Austen' match via {austen, jane} but
'Author A' ↔ 'Author B' don't (both collapse to {author} after
dropping single-letter tokens — we track raw token count to refuse
the single-token shortcut when discriminative parts were filtered).
Genuine mononyms still match on a single shared token because raw
count is 1 on both sides.
OPDSPerson[] is fed through the library's own formatAuthors so the
produced author string matches Book.author byte-for-byte (Intl.ListFormat
output, locale-aware). Soft-deleted books are skipped so removing a copy
locally reverts the button.
Wire it in opds/page.tsx by subscribing to useLibraryStore.library
(rather than the snapshot reads inside handleDownload) and memoizing
existingBookForPublication. Pass it to PublicationView, which now
seeds downloadedBook from the prop and resyncs when the prop changes
(switching publications inside the browser) — but does not clobber a
download success it observed locally before the parent recomputed.
Covered by unit tests for the helper, including the real Gutenberg
URN-in-atom-id case, ':version' suffix on the URN, 'Lastname, Firstname'
vs 'Firstname Lastname', year-range stripping, and the 'Author A' vs
'Author B' non-match.
* fix(opds): dedupe downloads by source url
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
|
||
|
Huang Xin
|
64492d6551 | feat(reedy): wire MemoryConsolidator + live memory providers into ReedyAssistant (#4310) | ||
|
Huang Xin
|
6be7606da4 | feat(reedy): wire Phase 5 skills + Phase 3 memory tools into the agent runtime (#4309) | ||
|
Huang Xin
|
e0ce6c8c22 | feat(reedy): Appendix A · Phase 4 — custom thread UI on AgentRuntime (#4308) | ||
|
Huang Xin
|
4aaf416c43 | feat(reedy): Appendix A · Phase 5.1 — SkillRegistry + 3 seed skills (#4306) | ||
|
Huang Xin
|
49664ecb75 | feat(reedy): Appendix A · Phase 3.2 — MemoryConsolidator (#4304) | ||
|
Huang Xin
|
568b8c0a88 | feat(reedy): Appendix A · Phase 3.1 — Memory services + memory tools (#4302) | ||
|
Huang Xin
|
df2698fa0e | feat(reedy): Appendix A · Phase 2.6 — AgentRuntime + abort helper (#4301) | ||
|
Huang Xin
|
a86b09dba5 | feat(reedy): Appendix A · Phase 2.5 — PromptContextBuilder + layers + tokenBudget (#4300) | ||
|
Huang Xin
|
5c71ccb908 | feat(reedy): Appendix A · Phase 2.4 — built-in tools (non-memory families) (#4299) | ||
|
Huang Xin
|
4d96c0d54a |
feat(reedy): Appendix A · Phase 2.1–2.3 — agent runtime foundation (#4298)
* feat(reedy): Appendix A · Phase 2.1 — ChatModel + model registry First piece of the deferred agent runtime, building alongside the shipped Phase 1 MVP per the locked decision (no rip-and-replace; legacy + Reedy MVP + agent paths coexist). ChatModel is a thin interface over Vercel SDK's LanguageModel that surfaces contextWindow / reservedOutput / supportsTools — the metadata the M2.5 PromptContextBuilder and M2.6 AgentRuntime need for prompt budgeting and tool-calling decisions. createReedyModels(settings) returns the active (chat, embedding) pair from the user's currently configured AIProvider, rewrapped in the Reedy interfaces without touching the legacy AIProvider. CONTEXT_WINDOW_TABLE hardcodes per-model metadata (Gemini 2.5 → 2M, GPT-5/Llama-4 → 128K, local Ollama llama → 4K with tools off, etc.) with a conservative 8K fallback for unknown ids. No runtime wired yet — Phase 2.6 consumes this. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(reedy): Appendix A · Phases 2.2–2.3 — runtime event/error taxonomy + ToolRegistry Phase 2.2 — pure types and factories for everything AgentRuntime.runTurn() will yield: turn_start | text_delta | tool_call | tool_result(ok/err) | citation | memory_write | step_finish | usage | error | abort | done ReedyError covers turn-level failures (context_overflow, turn_timeout, model_error, provider_unavailable, stream_parse, abort, ...) with default retryability per kind so the M2.6 streamLoop's retry-with-backoff can decide without string-matching. ReedyToolError covers tool-execution failures with a narrower kind union so the streamLoop can re-emit them as { tool_result, ok: false, error } events without losing structure. Phase 2.3 — ToolRegistry that holds the runtime's tool catalog and adapts each tool to the Vercel ai-SDK ToolSet streamText consumes. Every registered ReedyTool gets wrapped with: - Zod input validation → tool_invalid_args on mismatch - Permission gate → tool_permission_denied (read auto-approves) - Per-call wall-clock → tool_timeout (default 10s) - AbortSignal propagation → tool_aborted on turn cancel - Per-tool serialization → parallelSafe=false tools queue The local AbortController per call composes the turn-level signal with the per-tool timeout so we classify timeout-vs-abort correctly based on which controller fired first, and listener cleanup avoids leaks. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Huang Xin
|
6bc4a96b99 |
feat(reedy): Phase 1B — wire Reedy into the chat, settings, and Sources UI (#4296)
* feat(reedy): wire RetrievalBackend interface + metrics into the chat adapter
Phase 1B backend integration. Adds a RetrievalBackend interface so
TauriChatAdapter holds a uniform reference instead of branching across the
file, with two impls — LegacyIdbBackend (wraps the existing IDB ragService
unchanged) and ReedyBackend (lazy-opens reedy.db, adapts the active
provider's embedding model to Reedy's narrower shape, exposes a Vercel
`lookupPassage` tool). selectBackend() gates Reedy behind both
aiSettings.reedy.enabled AND isTauriAppPlatform() per plan D15 so the MVP
cohort is desktop-only.
The Reedy path streams via streamText({ tools: { lookupPassage }, stopWhen:
stepCountIs(3) }) with a status-aware system prompt that tells the model
how to phrase responses for each RetrieverStatus value.
Replaces the module-global `lastSources` + 500ms poll with a per-instance
ReedySourceStore keyed by a synthetic per-turn id the adapter generates,
so the Sources dropdown stops racing on global state. Both legacy and
Reedy backends now feed citations through the same store; the UI is
backend-agnostic via a shared SourceItem shape both ScoredChunk and
RetrievedChunk satisfy.
Adds the reedy_metrics table to the reedy migration (versioned with
app_version + session_id + turn_id per row) plus a ReedyMetrics writer
that ReedyBackend uses to record indexing-lifecycle and tool-use events.
Always-on local; no network egress. NoopReedyMetrics keeps construction
cheap before the DB opens.
Tests: retrievalBackend selectBackend gates, ReedySourceStore semantics
(append/replace/subscribe/clear), ReedyMetrics debounced batching +
exportBundle, and a TauriChatAdapter contract test that asserts the
Reedy/legacy code-paths pass the right args to streamText.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(reedy): UI wiring — settings toggle, clickable sources, feedback bundle
Phase 1B UI integration. AIAssistant now constructs the active backend
(legacy or Reedy) via selectBackend with the platform gate from M1.7,
wires a ReedySourceStore for the chat adapter, and routes Sources-dropdown
clicks to `getView(bookKey)?.goTo(source.cfi)` when the source has a CFI.
Legacy-path sources still render as static rows because they have no CFI.
AIPanel grows a 'Reedy Retrieval (Beta)' BoxedList with the toggle and a
'Send Reedy feedback' button that calls exportReedyMetricsBundle and
triggers a JSON download of the last 90 days of events. The toggle is
disabled on web with an explanatory description per plan D15.
Thread accepts an onSourceClick callback and renders each source as a
button when its source carries a CFI, otherwise as a static div — so the
Sources dropdown is backend-agnostic via the shared SourceItem shape.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Huang Xin
|
a1046f5684 |
feat(reedy): Phase 1A — MVP retrieval primitives (#4293)
* feat(db): add reedy schema migration with Tantivy FTS + lazy embeddings Registers a new `reedy` migration set bound to reedy.db. Creates reedy_book_meta + reedy_book_chunks with a Tantivy FTS index on chunks.text (ngram tokenizer) and a per-book position index used by BookRetriever. The vector embeddings table is intentionally NOT created here — the indexer creates it lazily on first index so the vector32(<dim>) column matches the active embedding model. Tests cover the migration applies cleanly, is idempotent, and that the FTS index is queryable. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(reedy): retrieval primitives — DB, chunker, indexer, retriever, lookupPassage tool Wires the MVP retrieval pipeline behind reedy.db, all under src/services/reedy/ and with no integration into the existing AI module yet (Phase 1B will do that). - ReedyDb wrapper over DatabaseService: book-meta CRUD, lazy embeddings table at the active model's dim, bulk chunk + embedding writes via batch(), hybridSearch (brute-force cosine + Tantivy FTS + reciprocal-rank fusion with 3× per-path over-fetch), per-book and global wipe. Internal write queue serializes batch() calls so Turso's single-writer transaction guard doesn't trip when BookIndexer runs across books in parallel. - CfiChunker: TreeWalker over the section's DOM, ~maxChunkSize windows with paragraph > sentence > word break-points, full epubcfi(/6/N!/…) anchors, round-trip verified via CFI.toRange before each chunk lands. - BookIndexer: per-book mutex, lazy embeddings-table creation, model.batchSize embedding batches with dim assertion, terminal status transitions (indexed | empty_index | failed). Re-indexing clears prior chunks via a new ReedyDb.clearBookChunks helper. - BookRetriever: status-typed results (ok | not_indexed | empty_index | stale_index | degraded). Embedding has a 5s wall-clock budget; on timeout it falls through to FTS-only with status=degraded. - lookupPassage Vercel ai-SDK tool: Zod-validated query/topK, per-turn composite-key dedupe, parallel-call serialization, 10s per-turn budget, 6000-char result clamp, status-with-hint passthrough, and a separate serializeForModel that wraps each passage in <retrieved trust="untrusted"> with XML-escaped content so book text cannot escape the envelope. 59 unit tests; pnpm test + pnpm lint clean. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |