* fix: add missing TXT worker files for large import path
* fix(ios): decode import paths and stabilize large TXT conversion
* style(txt): run prettier on TXT conversion files
* fix(txt): restore chunk iterator API and stream cancel behavior
* fix(txt): avoid unused private segment iterator
* chore: clean up log for unit tests
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
* feat: add RSVP speed reading feature
Implements Rapid Serial Visual Presentation (RSVP) speed reading mode:
- Display words one at a time with ORP (Optimal Recognition Point) highlighting
- Adjustable WPM speed (100-1000) with keyboard/button controls
- Punctuation pause settings (25-200ms)
- Progress tracking with chapter navigation
- Context panel showing surrounding text when paused
- Keyboard shortcuts (Space, Escape, arrows) and touch gestures
- Chapter selector for quick navigation
- Respects current theme colors
- Persists settings (WPM, pause duration, position) per book
New files:
- services/rsvp/RSVPController.ts - Main controller with playback logic
- services/rsvp/types.ts - TypeScript interfaces
- components/rsvp/RSVPOverlay.tsx - Full-screen RSVP reader UI
- components/rsvp/RSVPControl.tsx - Integration component
- components/rsvp/RSVPStartDialog.tsx - Start position selection
Closes#3111
* test(rsvp): add Playwright e2e tests for RSVP feature
- Set up Playwright test infrastructure with config
- Add comprehensive e2e tests for RSVP speed reading:
- Opening RSVP from View menu
- Start dialog options
- Play/pause toggle
- Speed adjustment
- Skip navigation
- Keyboard shortcuts
- Progress bar
- Chapter navigation
- Accessibility tests
- Add test data attributes and ARIA labels to RSVPOverlay
- Add test scripts to package.json
* refactor: remove test files, unsure of use case for now
* chore: remove Playwright e2e test scripts from package.json
- Deleted e2e test scripts related to Playwright from package.json as they are no longer needed.
- Removed Playwright as a dev dependency to streamline the project.
* chore: sync pnpm-lock.yaml after removing @playwright/test
* fix: lint errors and timezone-aware date formatting
* fix(rsvp): restore correct position when exiting RSVP mode
Fix bug where exiting RSVP after changing speed would navigate to wrong
position (several pages backwards). The issue was that Range objects
become stale if the document re-renders, and the position recovery used
global word index instead of per-document index.
- Add docWordIndex to track word position within each document
- Add docTotalWords to RsvpStopPosition for accurate recovery
- Validate Range before using it, recreate from current DOM if stale
- Use same word extraction logic as RSVPController for consistency
* fix(rsvp): prevent arrow keys from changing chapter dropdown
Arrow keys during RSVP mode were inadvertently navigating the chapter
dropdown instead of controlling RSVP speed. Fixed by using capture phase
for keyboard events and stopping propagation to prevent native elements
from receiving the events.
* chore: remove playwright-mcp screenshots from repo
* chore: update gitignore
* fix(i18n): wrap all RSVP overlay strings with translation function
Restore missing translation wrappers for all user-facing strings:
- Close RSVP button labels and tooltips
- Select Chapter dropdown
- WPM display
- Context panel header
- Ready state text
- Chapter Progress label
- Words count and time remaining
- Reading progress slider
- Pause/punctuation label
- Skip back/forward buttons
- Play/Pause button
- Speed control buttons
* fix(rsvp): use CFI-based position tracking for accurate page positioning
- Add CFI generation for each extracted word during RSVP processing
- Implement CFI path and character offset comparison for precise position matching
- Update startFromCurrentPosition to use CFI matching instead of word index
- Add extractFullDocPathWithOffset to handle CFI character offsets
- Simplify RSVPControl by removing unused helper functions
- Remove word index from RsvpPosition in favor of CFI-only tracking
* fix(ViewMenu): re-enable Speed Reading Mode in production environment
* refactor(ViewMenu, RSVPController): clean up code formatting and improve readability
* refactor(ViewMenu): remove unused IoSpeedometer import
* refactor(RSVPController): streamline CFI handling and improve section comparison
- Removed redundant logic and re-using CFI util functions for CFI operations
* refactor(RsvpPosition): remove legacy wordIndex field for cleaner type definition
* feat(library): implement grouping of books by series or author
- Added functionality to group books in the library by series or author.
- Introduced new utility functions for creating book groups and parsing author strings.
- Updated the LibraryPageContent to track the current group for navigation.
- Created a new GroupHeader component to display the current group context.
- Enhanced the Bookshelf component to support displaying grouped books.
- Updated settings to include grouping options in the view menu.
New files:
- src/__tests__/utils/libraryUtils.test.ts - Unit tests for new utility functions.
- src/app/library/components/GroupHeader.tsx - Component for displaying group context.
- src/app/library/utils/libraryUtils.ts - Utility functions for grouping and parsing authors.
* Use group by and sort by constants, and split sort/group by menus into their own elements
* Format code with autoformatter
* Remove any casting from tests
* Translate missing strings
* refactor: cleaner layout with collapsible view menu options
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
* feat: add RSVP speed reading feature
Implements Rapid Serial Visual Presentation (RSVP) speed reading mode:
- Display words one at a time with ORP (Optimal Recognition Point) highlighting
- Adjustable WPM speed (100-1000) with keyboard/button controls
- Punctuation pause settings (25-200ms)
- Progress tracking with chapter navigation
- Context panel showing surrounding text when paused
- Keyboard shortcuts (Space, Escape, arrows) and touch gestures
- Chapter selector for quick navigation
- Respects current theme colors
- Persists settings (WPM, pause duration, position) per book
New files:
- services/rsvp/RSVPController.ts - Main controller with playback logic
- services/rsvp/types.ts - TypeScript interfaces
- components/rsvp/RSVPOverlay.tsx - Full-screen RSVP reader UI
- components/rsvp/RSVPControl.tsx - Integration component
- components/rsvp/RSVPStartDialog.tsx - Start position selection
Closes#3111
* fix(rsvp): use portal to fix overlay stacking context issue
- Render RSVP overlay and start dialog via React Portal to document.body
- This ensures the overlay appears above all other content regardless of
parent component CSS transforms or stacking contexts
- Add fallback colors for theme values to ensure solid background
* fix(rsvp): improve UX with progress sync, sentence highlight, and better dialogs
- Fix start dialog transparency by using solid opaque background with proper
fallback colors for both light and dark modes
- Increase context words from 50 to 100 words before/after current word
- Add progress sync on RSVP exit - navigates reader to the last word position
- Add temporary bright green sentence underline on exit (5 second duration)
to easily locate where reading left off
- Helper function to expand word range to full sentence boundaries
* fix(rsvp): store full BookNote for proper annotation removal
The addAnnotation API requires the full BookNote object (including CFI)
when removing annotations, not just the ID. Changed tempHighlightIdRef
to tempHighlightRef to store the complete BookNote object.
* test(rsvp): add Playwright e2e tests for RSVP feature
- Set up Playwright test infrastructure with config
- Add comprehensive e2e tests for RSVP speed reading:
- Opening RSVP from View menu
- Start dialog options
- Play/pause toggle
- Speed adjustment
- Skip navigation
- Keyboard shortcuts
- Progress bar
- Chapter navigation
- Accessibility tests
- Add test data attributes and ARIA labels to RSVPOverlay
- Add test scripts to package.json
* fix(rsvp): clarify start dialog option labels
Update the RSVP start dialog to use clearer language:
- "From Beginning" → "From Chapter Start" (since it starts from chapter beginning)
- "From Current Position" → "From Current Page" (starts from visible page)
* fix(rsvp): use correct theme colors from themeCode
The RSVP components were using incorrect palette key names (camelCase
like `base100` instead of hyphenated like `base-100`), causing the
fallback colors to always be used instead of the reader's actual theme.
Fix by using themeCode.bg, themeCode.fg, and themeCode.primary directly,
which are already resolved from the palette with correct keys.
* fix(rsvp): use theme accent color for sentence underline and persist until page change
- Change underline color from hardcoded green (#22c55e) to theme accent
color (themeCode.primary), matching the ORP focal point highlight
- Remove 5-second timeout that auto-removed the underline
- Add cleanup on page navigation, new RSVP session start, and unmount
- Add removeRsvpHighlight helper function for consistent cleanup
* fix(rsvp): transition to next chapter when reaching end of section
When RSVP reached the end of a chapter, it would restart the current
chapter instead of moving to the next one. This happened because RSVP
extracts ALL words from the current section via renderer.getContents(),
so when words run out, the entire chapter is done.
- Always use view.renderer.nextSection() when RSVP exhausts words
- This moves to the next chapter instead of staying in the current one
* refactor: remove test files, unsure of use case for now
* chore: remove Playwright e2e test scripts from package.json
- Deleted e2e test scripts related to Playwright from package.json as they are no longer needed.
- Removed Playwright as a dev dependency to streamline the project.
* fix(rsvp): ensure CFI retrieval occurs before navigation
- Updated comments to clarify the necessity of obtaining CFI for both the navigation and sentence highlight before invoking the goTo() method, as it may re-render the document and invalidate Range objects.
- Introduced a new variable for sentence text to enhance readability and maintainability of the code.
* chore: sync pnpm-lock.yaml after removing @playwright/test
* style: format RSVP components
* fix: lint errors and timezone-aware date formatting
* i18n: support CJK text and add translations
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
* feat: enable linking to settings items
* feat: integrate command palette with global state and styles
* feat: add command pallete ui and fuzzy search
* feat: add command palette with global state and styles
* feat: add command registry and search dependencies
* fix: open command palette with shortcuts in reader page
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
* feat(highlight): extend types and constants for custom hex colours
* feat(highlight): update colour to support hex strings
* refactor(annotator): update component to accept hex colours
* feat(highlight): add colour picker with max 4 custom colors
* feat(settings): add custom colour editor with limit
* refactor: custom highlight colors can only be modified in settings
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
* feat(ai): add dependencies
* chore: bump zod version to default version
* feat(ai): define types and model constants
* feat(ai): ollama provider for local LLM
* feat(ai): implement openrouter provider for cloud models
* feat(settings): register ai settings panel in global dialog
* refactor(ai): expose provider factory and service layer entry point
* test(ai): add unit tests for the providers
* test(ai): add unit tests for the providers
* feat(ai): settings panel for ai configurations
* refactor(ai): rewrite aipanel with autosave and greyed out disabled state
* fix: remove unused onClose prop from aipanel
* test(ai): update mock data
* refactor(ai): remove models
* refactor: use centralised defaults in system defaults
* chore(ai): remove comments
* fix(ai): merge default ai settings on load to prevent undefined values
* refactor(ai): rewrite settings panel with autosave and model input
* feat(ai): add ai tab with simplified highlighting
* feat(sidebar): render AIAssistant for ai tab
* feat(ai): add chat UI
* feat(ai); add chat service with RAG context
* feat(ai): temp debug logger
* feat(ai): add RAG service
* feat(ai): add text chunking utility
* feat(ai): add structured method
* feat(ai): add chatstructured method
* feat(ai): add rag types nd structured output schema
* feat(ai): add aistore, indexdb, bm25
* fix: update lock file
* feat(ai): update types for AI SDK v5
* feat(ai): add placeholder gateway model constants
* refactor(ai): update OllamaProvider for AI SDK
* feat(ai): add native gateway provider
* refactor(ai): update provider exports
* refactor(ai): use streamText from AI SDK
* refactor(ai): use embed from AI sdk
* refactor(ai): update provider factory exports
* feat(ai): add AI Elements and shadcn components
* config: add shadcn component config
* deps: add AI SDK and AI Elements dependencies
* config: add ai packages to transpilePackages
* refactor(ai): remove OpenRouterProvider and old tests
* feat(ai):add assistant-ui components
* feat(ai): add TauriChatAdapter for assistant-ui runtime
* refactor(ai): remove ai-elements components
* dep(ai): install assistant-ui and update next config
* chore(ai): export adapters from service index
* feat(ui): enhance ui components for assistant integration
* feat(settings): migrate ai settings to gateway
* feat(sidebar): integrate assistant-ui
* feat: add ai settings toggle to sidebar content
* feat: conditionally show ai tab in sidebar navigation
* feat: update ai model constants for cheaper options
* feat: add gateway provider with proxied embedding
* feat: add timeouts to ollama provider health checks
* feat: add retry logic to rag service embeddings
* feat: add error recovery to ai store
* feat: add ai feature tests
* feat: add ai api endpoints
* feat: add proxied gateway embedding provider
* feat: add ai runtime utilities
* feat: add ai retry utilities
* feat: add tauri env example template
* feat: add web env example template
* chore: add env
* feat(ai): update models and pricing, remove GLM-4.7-FlashX
* feat(ai): improve system prompt with official headings and no numeric citations
* feat(ai): optimize system prompt for tauri chat
* feat(ui): refine ai chat UI and relocate sources
* feat(ui): update ai settings panel with model pricing and custom model support
* feat(ai): add custom model support to ai settings
* test(ai): update constants tests for removed model
* feat(api): implement ai chat proxy route
* feat(api): implement ai embedding proxy route
* feat(ai): implement ai gateway health check and proxy logic
* feat(ai): simplify proxied embedding provider
* feat(ui): improve markdown text rendering
* feat(ui): add input group component
* test(ai): update ai provider tests
* feat(ai): add pageNumber to text chunk schema
* feat(ai): implement page-based chunking with 1500 char formula
* feat(ai): bump db to v2 and add store reset migration
* feat(ai): transition rag pipeline to page level spoiler filtering
* feat(ai): overhaul readest persona and antijailbreak prompt
* feat(ai): update tauri adapter for page tracking and persona
* chore(ai): export aiStore and logger from core index
* feat(reader): integrate page tracking and manual index reset
* feat(ui): add re-index action and reset logic to chat
* chore: sync pnpm lockfile with ai dependencies
* feat(utils): add browser-safe file utilities for web builds
* refactor(utils): use dynamic tauri fs import to prevent web crashes
* refactor(services): defer osType call to init() for web compatibility
* refactor(services): import RemoteFile from file.web
* refactor(services): import ClosableFile from file.web
* fix(libs): cast Entry to any for getData access
* fix(annotator): cast Overlayer to any for bubble access
* refactor(ai): replace SparklesIcon with BookOpenIcon for index prompt
* test(ai): add pageNumber to TextChunk mocks
* test(ai): fix chunkSection signature in tests
* chore: update files
* fix(ai): prevent useLocalRuntime crash when adapter is null
* refactor: optimize annotator overlay drawing
* feat: stabilize AI assistant runtime and adapter
* refactor: improve document zip loader type safety
* feat: update tauri chat adapter for dynamic options
* fix: restore architecture comments and refine platform properties
* build: update lockfile with assistant-ui patch
* fix(library): patch @assistant-ui/react for runtime initialization
* build: update dependencies in readest-app
* build: update root dependencies and patch configuration
* fix(ai): patch @assistant-ui/react for thread deletion and runtime init
* fix(ai): update assistant-ui patch with dist guards and deletion fallback
* build: sync lockfile with assistant-ui patch updates
* chore(env): update .gitignore by removing .env files from it
* chore(env): update .gitignore by adding .env.local
* chore(env): update .gitignore by adding .env*.local
* fix: restore static osType import
* chore: sync submodules with upstream/main
* refactor: remove redundant file.web module and revert import
* chore: update pnpm-lock.yaml
* refactor: revert guards
* refactor; remove deprecated codes and extract prompts.ts
* refactor(ai): remove unused ragservice exports
* refactor: remove unused ollama and embedding models
* refactor: remove unused type
* test: remove test for the now deleted constants
* refactor: remove unused export
* style: fix ui component formatting
* style: fix core and style file formatting
* test: fix broken ai provider import
* fix: typescript error
* fix: add eslint disable command
* fix(deps): remove unused ai sdk provider util after v6 ai sdk migration
* fix(patch): add lookbehind regex patch
* feat(dep): upgrade vercel ai sdk to v6 and ai-sdk-ollama to v3
* chore: update lockfile for vercel ai sdk v6
* refactor(ai): remove EmbeddingModel generic for ai sdk v6
* refactor(ai): remove EmbeddingModel generic for ai sdk v6
* test(ai): update mock to use embeddingModel
* fix(patch): add lookbehind regex patch for email autolinks in markdown
* refactor(ai): use ai sdk v6 syntax
* fix: prettier formatting
* chore: revert cargo.lock
* fix(ai): update proxied embedding model to v3 spec
* feat(ai): add aiconversation types for chat persistence
* feat(ai): add conversation/message indexeddb and crud operations
* feat(ai): create aiChatStore zustand store for chat state management
* feat(notebook): add notebookactivetab state for Notes/AI
* refactor(ai): refine conversation and message types for persistence
* feat(types): add notebookActiveTab to ReadSettings type
* chore: update deps
* feat: add notebookactive tab default value
* feat: add hook for ai chat
* feat: update left side panel with history/chat icon
* feat: integrate ChatHistoryView into sidebar content
* feat: create UI for managing AI chat history
* feat: implement persistent history with assistant-ui adapter
* feat: create tab navigation component for notes and AI
* feat: add tab navigation and AI assistant view
* feat: update header to display active tab title
* fix: formatting
* feat: remove title and update new chat button
* fix: formatting
* fix: revert tooltip and styling
* feat: implement cross-platform ask dialog bridge
* feat(ai): preserve history during ui clear & use native dialogs
* fix: align notebook navigation height with sidebar tabs
* fix(ai): add missing dependency to handleDeleteConversation hook
* docs: update PROJECT.md with session highlights
* chore: delete projectmd
* chore: update package.json and lock file
* chore: update package.json
* chore: remove patch
* chore: upgrade react types to 19 and show ai features only in development mode for now
---------
Co-authored-by: Huang Xin <chrox.huang@gmail.com>
Convert to   before sanitization and restore after XMLSerializer
serialization. This handles the platform difference where XMLSerializer
produces different representations of non-breaking spaces ( ,  ,
or \u00A0) across different WebView implementations.
Huang Xin