* Added support for ODT and FODT file formats.
* Improved `OdtParser` to support enhanced document formatting and external assets
* Added support for viewing CSV, TSV, JSON, XML, logs, and source code files by dynamically converting them to HTML.
* Add visual options and seamless transitions to the EPUB reader
* Add support for DOCX file format
* Refine padding and status bar inset handling in Epub pagination mode
* Add file size parameter for recent files and sorting
* Refactor book data cleanup logic into a centralized method
* Bump version to 1.0.40(41)
* Added support for FB2 book format
* Add support for CBZ files by introducing a `ReaderDocument` abstraction.
Key changes:
- Defined `ReaderDocument`, `ReaderPage`, and `ReaderTextPage` interfaces to provide a unified API for different document types.
- Implemented `PdfDocumentWrapper` and `CbzDocumentWrapper` to handle PDF and CBZ files respectively.
- Updated `PdfViewerScreen`, `PdfPageComposable`, and `MainViewModel` to use the new unified document interfaces.
- Added CBZ file type detection and cover generation logic.
* Add support for CBR and CB7 comic book formats
- Add `me.zhanghai.android.libarchive` dependency to support RAR and 7z archives.
- Implement `ArchiveDocumentWrapper` using `libarchive` to handle CBZ, CBR, and CB7 files uniformly, replacing the previous CBZ-only implementation.
- Update `MainViewModel` and `DocumentFactory` to recognize and process `.cbr` and `.cb7` extensions and their associated MIME types.
- Add support for extracting and caching cover images from CBR and CB7 archives.
- Update UI components (`HomeScreen`, `LibraryScreen`, `AppNavigation`) to handle the new comic book file types.
* perf(pdf): native Pdfium-based reflow engine with auto-open
Migrate PDF-to-Markdown reflow generation from PDFBox to a custom
native implementation using Pdfium. This transition improves
processing speed by ~10x and significantly reduces memory overhead.
- Native JNI Bridge: Implemented `pdfium_bridge.cpp` using `dlopen`
to hook into the existing `libpdfium.so` memory space.
- Optimized Extraction: Replaced character-by-character JNI calls
with bulk array retrieval (`getPageFontSizes`, `getPageFontWeights`),
drastically reducing JNI boundary overhead.
- Enhanced Accuracy: Improved Markdown formatting logic by using
native font weight (bold) and relative font size variance (headers).
- Thread Safety: Refactored generator to process pages sequentially
while synchronized with the global `PdfiumCore.lock` to ensure
stability across concurrent UI operations.
- Seamless UX: Implemented a reactive auto-open system in the
PDF viewer that tracks user intent and navigates to the reflow
view immediately upon background task completion.
* Improved PDF to Markdown conversion and added cache cleanup.
- Implemented automated cleanup of imported file caches when deleting books.
- Enhanced `PdfToMarkdownGenerator` with support for font flags (italics), improved kerning, and smarter paragraph wrapping.
- Updated `NativePdfiumBridge` and C++ JNI code to extract font information flags from PDFium.
* Implemented seamless file switching and enhanced reflow transition logic.
Key changes include:
- Added `switchToFileSeamlessly` to `MainViewModel` to handle state transitions and navigation when switching between PDF and reflowed text views.
- Updated `generateAndImportReflowFile` to support automatic opening of the generated file at a specific page/chapter.
- Integrated `NavigationEvent` and `CompletableDeferred` to manage asynchronous navigation and state updates during file switches.
- Modified `EpubReaderScreen` and `PdfViewerScreen` to pass the current position when toggling between PDF and text modes.
- Updated `AppNavigation` to move navigation logic out of the `NavHost` and added loading overlays to viewers to improve UI feedback during transitions.
* Implemented PDF reflow mode by introducing a mechanism to convert PDF content to Markdown/HTML for viewing in the EPUB reader.
Specific changes include:
- Added `PdfReflowGenerator` and `PdfToMarkdownGenerator` to handle PDF text extraction and conversion to reflowable formats.
- Updated `MainViewModel` with `toggleReflowMode` logic to switch between original PDF and reflowed views.
- Modified `RecentFileEntity` and `RecentFileDao` to persist user reflow preferences, including a Room database migration (v12 to v13).
- Updated `PdfViewerScreen` and `EpubReaderControls` to include UI options for toggling reflow mode.
- Integrated reflow preference check into the book opening workflow to automatically load the preferred view.
- Updated `AppNavigation` and `EpubReaderScreen` to support the new view switching state.
* Implemented background processing and incremental loading for PDF reflow mode.
- Added `reflowProgress` to `MainViewModel` to track and display PDF-to-Markdown conversion progress in the UI.
- Refactored `PdfToMarkdownGenerator` to generate a skeleton EPUB structure immediately while processing page content (text and images) asynchronously.
- Switched PDF text extraction to use `PDFBox` with optimized memory settings and JPEG compression for images.
- Implemented priority page processing in reflow mode, starting with the user's current page.
- Added "Clear Reflow Cache" debug option to the Home Screen.
- Enhanced `BookPaginator` to support lazy loading of chapter content from disk and improved cache hit detection.
* Refactored PDF Reflow Mode to generate standalone Markdown files instead of temporary EPUB books.
* perf(reflow): optimize PDF-to-Markdown conversion and fix viewing lag
- Re-architected PdfToMarkdownGenerator to use a single-pass stream (O(N) complexity), fixing performance bottlenecks and timeouts on large PDFs.
- Implemented "Virtual Chaptering" in SingleFileImporter for Markdown files to split content into page-level HTML files, eliminating UI lag during reading.
- Simplified ReflowWorker to delegate progress tracking and looping to the generator.
- Enhanced PdfViewerScreen with a prominent top-bar progress indicator and a completion snackbar with an "OPEN" action.
* Optimized EPUB parsing performance and fixed PDF viewer UI layout.
- Optimized `EpubParser` by implementing parallel chapter parsing using coroutines and a semaphore to limit concurrency.
- Reduced memory usage in `EpubParser` and `SingleFileImporter` by no longer storing full HTML content in memory for chapters.
- Updated `EpubXMLFileParser` to support an existing `Document` object to avoid redundant Jsoup parsing.
- Fixed an issue in `PdfViewerScreen` where the snackbar was appearing under the bottom app bar.
* feat: rework Folder Sync architecture to separate Managed vs Linked books
- Introduced "Managed" (imported to app storage) vs "Linked" (SAF URI) book logic.
- Disabled Google Drive/Metadata synchronization for Folder-Linked books to respect privacy and storage.
- Updated deletion logic: Folder books are now marked as deleted in the DB to blacklist them from future auto-scans, while Managed books are permanently purged from local and cloud storage.
- Enhanced folder sync discoverability by adding a direct "Sync Folder" navigation button in Home screen.
- Implemented reactive pager navigation in MainScreen and LibraryScreen via LaunchedEffect to allow programmatic tab switching from the ViewModel.
* Implemented a local folder synchronization system to track and update book metadata across devices.
Key changes:
- Added `FolderBookMetadata` and `LocalSyncUtils` for serializing and managing metadata files in a `.episteme` directory.
- Implemented bidirectional sync in `FolderSyncWorker` to reconcile local database state with folder-based metadata using timestamps and Syncthing conflict resolution.
- Updated `MainViewModel` to trigger folder metadata sync when closing a book and verify for remote updates upon opening.
- Enhanced folder management with manual scan support via `WorkManager` and automatic cleanup of database entries when a folder is disconnected.
- Added `RecentFileDao` methods to support bulk deletion and prefix-based file lookups.
* Removed API 34 restriction for chapter processing and updated local folder sync directory name
* Increased TTS timeouts, added edge-to-edge support in MainScreen, and improved TtsChunk mapping safety
* Improved folder synchronization and metadata management.
- Added metadata-only sync option and triggered it on app start.
- Implemented hidden metadata filenames (prefixed with `.`) to reduce file clutter.
- Added automatic creation of `.nomedia` files in the sync directory.
- Enhanced sync conflict resolution by picking the latest metadata and cleaning up obsolete or orphaned files.
- Added "Sync Metadata" button to the Library screen.
- Updated UI labels for local folder sync to clarify Google Drive integration.
- Refactored `FolderSyncWorker` to support targeted metadata-only synchronization.
* Implement folder sync migration and refactoring
- Added a migration dialog to inform users about the folder sync refactor, where books are now read directly from external storage.
- Updated `MainViewModel` to handle migration state and trigger a full scan upon completion or dismissal of the dialog.
- Modified `FolderSyncWorker` to support legacy book matching during migration, updating file URIs and cleaning up internal app storage for migrated books.
- Improved synchronization logic between local and remote metadata, including cleanup of orphaned metadata files.
- Refined `RecentFileItem` updates during sync to preserve progress and bookmarks based on last modified timestamps.
* Implemented pull-to-refresh for library sync and enhanced folder file deletion.
- Added `isRefreshing` state to `MainViewModel` and integrated `PullToRefreshBox` in `HomeScreen`.
- Implemented `refreshLibrary` to trigger cloud and folder metadata synchronization.
- Updated deletion logic to physically remove files and metadata from synced local folders.
- Added a warning to the delete confirmation dialog when removing folder-synced items.
- Improved folder sync reliability by performing lazy cleanup of missing files during interaction and sync.
- Fixed a bug where sync loading indicators would not retract on failure or cancellation.
* Improved bookmark navigation and scroll synchronization in EPUB reader
- Refined bookmark navigation logic for vertical scroll mode to handle chunk injection more reliably.
- Added `isNavigatingToBookmark` state to show a loading overlay during long jumps.
- Implemented `onScrollFinished` callback in `CfiJsBridge` to synchronize UI state with WebView scroll completion.
- Updated `ChapterWebView` to use `rememberUpdatedState` for JavaScript bridge callbacks to ensure data consistency.
- Improved `getCurrentCfi` and `scrollToCfi` in `epub_reader.js` with better visibility probing and retry logic for detached nodes.
* Implemented background metadata extraction for folder sync.
Key changes:
- Added `MetadataExtractionWorker` to handle heavy metadata and cover extraction for files in the background.
- Refactored `FolderSyncWorker` to perform fast file discovery using placeholders, enqueuing the metadata worker upon completion.
- Added `getFolderBooksWithoutCovers` query to `RecentFileDao`.
- Updated UI components (`EmptyState`, `MainViewModel`) to improve user feedback during sync and provide clear setup actions.
- Enhanced `EmptyState` composable to support secondary actions and custom button text.
* Updated Library Screen UI
docs: Revamp README and add APK naming convention
This commit introduces a complete overhaul of the `README.md` to improve project presentation. Key changes include:
* Adds a new header with the app icon and a Google Play badge.
* Updates the feature graphic and includes a note distinguishing the OSS version from the Play Store version.
* Refines the "Open Source Libraries" section to be more structured.
* Moves the main preview image into the `docs` directory.
Additionally, the `app/build.gradle.kts` file is updated to configure a standardized naming convention for output APK files, including flavor, version, and build type.
* refactor copyright headers and fix README icon alignment
This commit introduces two main changes across the codebase:
The `spotless/copyright.kt` template have been updated for consistency:
* The author name is simplified from "Episteme Authors" to "Episteme".
* The contact email prefix is shortened from "Electronic mail:" to "mail:".
Second, the `README.md` file is adjusted to correct the vertical alignment of the app icon with the project title.
