- Replace client-side IndexedDB with server-first document storage
- Add document caching layer for offline capability and performance
- Implement document transfer during anonymous-to-authenticated account linking
- Add database indexes for improved query performance
- Update document contexts to use new caching system
- Refactor document upload and deletion APIs
- Add audiobook pruning for missing files
- Improve test isolation with namespace support
- Add unit tests for document cache and transfer functions
Ensure all metadata operations consistently use the database layer,
removing filesystem-only fallback paths and conditional DB checks.
- Simplified migration scripts to always run on startup
- Updated document/audiobook APIs to always query DB
- Added ensureDbIndexed() calls across all routes
- Extracted test namespace utilities to dedicated module
- Removed migration-manager.ts (functionality consolidated)
- Updated rate limiter to assume DB is always available
BREAKING CHANGE: Database is now required in all configurations.
When auth is disabled, SQLite is used by default at /app/docstore/sqlite3.db.
- Change audiobooks and audiobookChapters tables to use composite PK (id, userId)
- Migrate audiobooks from flat storage to user-specific directories under audiobooks_users
- Add support for claiming unclaimed audiobooks on account creation
- Improve auth rate limiting with retry logic and DISABLE_AUTH_RATE_LIMIT for tests
- Fix iOS/Safari audio playback with unlock mechanism and playback rate watchdog
- Update API routes to handle user-scoped audiobook access with fallback to unclaimed
- Transfer audiobooks when linking anonymous accounts to real accounts
- Remove foreign key constraint from audiobookChapters to support composite PK
- Add cascade delete to account and session foreign keys
Replace privacy popup with PrivacyModal and reorganize auth UI components
Remove signed-out persistence and centralize session handling in useAuthSession
Move document viewers and Kokoro voice helpers into dedicated modules
Relocate migration scripts to `drizzle/scripts`, update `start` and
`migrate`
to use the new path, and add a `generate` script for contributors.
Update README migration guidance for SQLite vs Postgres workflows.
BREAKING CHANGE: remove `migrate:force` and change migration script
paths; consumers relying on `scripts/migrate*.mjs` or `pnpm
migrate:force` must update to the new commands.
Add dedicated Drizzle config files for sqlite and postgres and generate
new migration outputs for each dialect.
Update migrate scripts to load env files and auto-pick the correct
config unless explicitly provided, and tighten API auth handling by
using shared auth context helpers and enforcing userId checks when auth
is enabled.
- Smooth react-pdf page turns with stable canvas staging and fade-in
- Retry transient empty page text extraction during fast page turns
- Avoid noisy TTS preload toasts, guard requests when quota is exhausted,
and improve Howler retry/unload behavior to prevent pool exhaustion
- Enforce auth FK cascade deletes and tighten rate-limiter bucket updates
using affected-row checks
- replace custom DB adapter with Drizzle setup (SQLite/Postgres schemas) and add `drizzle.config.ts` plus migrations in `drizzle/` and `drizzle_pg/`
- switch better-auth to drizzleAdapter, add auth helpers (`getAuthContext`, `requireAuthContext`, `requireAudiobookOwned`), and make `useAuth`/`useAuthSession` safe no-ops when auth is disabled
- persist documents/audiobooks in DB with ownership checks, unclaimed fallback, and ref-counted deletes; add FS scan helper for no-auth mode
- gate docx-to-pdf, library, voices, whisper, and migration endpoints behind auth when enabled
- add unclaimed data scan/claim flow with `/api/user/claim`, `ClaimDataModal`, and server-side scan/claim helpers
- refactor rate limiting and account deletion to use Drizzle tables (`user_tts_chars`, `user`)
- run migrations via `scripts/migrate-if-auth.mjs` (auto on `pnpm start` + `pnpm migrate`), remove Docker entrypoint and old better-auth migration file
- update README and lockfile for the new migration workflow and dependencies
Update modal components to start from top on mobile devices and center on larger screens, improving usability across screen sizes. Also refine transition handling in privacy popup for cleaner unmounting.
- Remove unused SettingsIcon and UserMenu imports from epub and pdf pages
- Remove unused UserMenu import from home page
- Remove unused isDev const from home page
- Update SettingsModal to only import clearSignedOut
- Fix apostrophe encoding in privacy popup description
- Add device ID and IP-based rate limiting to prevent abuse
- Refactor UI components for better header menus and settings
- Update privacy popup with detailed data usage info
- Improve PDF handling and caching to prevent react-pdf warnings
- Update README with new Docker instructions and environment variables
- Implement user sign-in, sign-up, and account management using better-auth
- Add rate limiting for TTS API with daily character limits for authenticated and anonymous users
- Integrate SQLite and PostgreSQL database support for user sessions and data persistence
- Update UI components to include authentication flows, user menu, and privacy popup
- Modify Dockerfile and package.json for new dependencies and entrypoint script
- Bump version to v1.3.0
* **Bug Fixes**
* More reliable PDF sentence and word highlighting during layout changes and rapid updates, preventing stale or overlapping highlights.
* **Enhancements**
* Revamped text normalization and TTS block splitting for better handling of long/complex content, EPUB texts, dialogue/quotes, and PDF artifacts.
* **Tests**
* Expanded unit tests covering block-splitting, EPUB behavior, oversized inputs, and boundary cases.
* **Chores**
* Package version bumped; test runner config updated to ignore unit tests for select browser projects.
- Add logic to split oversized sentences in `splitTextToTtsBlocksEPUB` using `splitOversizedText` to ensure blocks stay within `MAX_BLOCK_LENGTH`
- Remove `extractRawSentences` and `processTextWithMapping` functions as they are no longer needed
- Update tests to reflect the changes, including a new test for oversized sentence splitting and removal of related test suites
Introduce `splitTextToTtsBlocksEPUB` function to handle EPUB text splitting,
treating single newlines as paragraph boundaries for precise highlighting.
Update TTS context to conditionally use EPUB splitting based on document type.
Enhance PDFViewer to clear highlights when current sentence is null.
Add comprehensive tests for the new functionality and refactor existing ones.
Handle case where currentSentence is null by canceling retry loops and clearing stale highlights to prevent lingering highlights. Remove redundant check inside the highlight function.
- Rename and enhance text processing functions in nlp.ts for better handling of oversized texts, sentence boundaries, and PDF artifacts
- Update PDFViewer to add layout-aware highlighting with retry logic for sentence and word highlights
- Adjust PDFContext and TTSContext to use new normalized text functions
- Expand unit tests for new splitting behaviors, including long texts and punctuation preferences
Add support for building and publishing multi-architecture Docker images (amd64 and arm64) using a matrix strategy. Introduce a new 'prepare' job for metadata computation and a 'merge' job for creating manifest lists. Enable manual workflow dispatch with optional latest tagging. Improve caching and provenance handling for better efficiency and reliability.
Add 'force-dynamic' export to audiobook and documents API routes to ensure
dynamic rendering. Remove 'nodejs' runtime export from TTS route for consistency.
- Introduces `DocumentPreview` component to display rich previews for documents.
- PDFs generate a thumbnail of their first page.
- EPUBs extract and display their cover image.
- HTML, TXT, and Markdown files show a text snippet.
- Refactors document list and folder views to a responsive grid layout.
- Expands main content areas to accommodate the new grid.
- Updates Dockerfile to use Node.js LTS image.
- Updates various development and runtime dependencies.
- Add scroll functionality to the chapter list to prevent content
overflow.
- Conditionally adjust the positioning of chapter action menus
to ensure they remain visible within the modal, preventing
them from being cut off for chapters at the top of the list.
- Remove unused `isDev` variable from `Footer.tsx`.
- Add `inline` and `inline-flex` classes to footer elements for improved layout.
- Move custom `xs` breakpoint definition to `theme.extend` in `tailwind.config.ts`.
- Introduce a toggleable grid/list view for the document list, enhancing organization and display flexibility. This involved updates across `DocumentList`, `DocumentFolder`, `DocumentListItem`, and `SortControls`.
- Add a new `CodeBlock` component and integrate detailed Docker-based self-hosting instructions into the footer.
- Enhance privacy policy popover with clearer details on Deepinfra usage and strongly recommend self-hosting for secure experience.
- Implement conditional rendering and feature gating based on the `isDev` environment variable, differentiating features between the production demo and self-hosted instances. Affected components include `page.tsx`, `DocumentSettings.tsx`, `SettingsModal.tsx`, and `config.ts`.
- Clarify that advanced features like audiobook export and word-by-word highlighting via `whisper.cpp` are exclusive to self-hosted setups and are disabled in the demo.
- Expand the list of supported document types on the homepage to include MD and TXT.
- Integrate new `ListIcon`, `GridIcon`, and `CopyIcon` to support UI enhancements.
- Add a custom `xs` breakpoint in `tailwind.config.ts` for improved responsive design.
Removed `isPopoverOpen` state and associated `useEffect` hook.
Auto-focus and select logic for the input are now directly handled
within `handlePopoverOpen`, reducing state management complexity.
- New buildCacheKey function creates unique identifiers for TTS cache entries.
- Cache keys now include sentence, voice, speed, provider, and model parameters.
- Prevents serving cached audio or alignment data that mismatches current TTS parameters.
- Removes redundant audioCache.clear() calls when voice or speed change, as entries are now distinct.
- Implement custom previous/next page buttons for EPUB viewer.
- Display current page number out of total pages.
- Introduce an in-viewer, toggleable table of contents (TOC) for quick chapter navigation.
- Hide default `react-reader` navigation arrows and title bar to prevent redundancy.
- Adjust `react-reader` styles to optimize content area and ensure custom controls are visible.
- Update description for EPUB theme setting in document settings.
- Add `ChevronLeftIcon` and `ChevronRightIcon` to icon library.
- Update Playwright tests to reflect new navigation button labels.
The Dockerfile has been refactored to a multi-stage build, allowing the `whisper.cpp` CLI binary to be compiled and embedded within the application's runtime image. This enables word-by-word highlighting functionality when deployed via Docker. The `README.md` has been updated to include installation and configuration instructions for `whisper.cpp` when running locally. Additionally, the `WHISPER_CPP_BIN` environment variable has been added to `template.env` and the package version has been bumped to v1.1.0.
Abstracted direct fetch calls across components and contexts into new functions within `src/lib/client.ts`. This provides a consistent and centralized interface for interacting with backend APIs.
- Introduced `src/lib/client.ts` to encapsulate API request logic.
- Standardized audio buffer types (`TTSAudioBuffer`, `TTSAudioBytes`) in `src/types/tts.ts`.
- Moved client-specific request types (`TTSRequestPayload`, `TTSRequestHeaders`, `TTSRetryOptions`) to `src/types/client.ts`.
- Updated API routes and consumer components/contexts to leverage the new client library functions and type definitions.
- Removed `src/utils/audio.ts` as its utility functions are now part of `src/lib/client.ts`.
