v0.1.9

Bug Fixes

• fix(docker): set HF_HOME for faster-whisper model caching

Other Changes

See commit history since v0.1.8

v0.1.8

Bug fixes

• Storyteller: Handle DNS failures cleanly — connection backoff prevents log spam and repeated failed requests when Storyteller is unreachable
• KoSync: Use datetime.min instead of 0 for last_updated fallback
• Sync manager: Treat check_connection() returning False as a failure (previously only caught exceptions)

v0.1.7

Security, KoSync refactor, suggestions & Grimmory rename

Security

• Non-root Docker container — app now runs as appuser (uid 1000) instead of root
• Rate limiter for KoSync auth endpoints — token bucket throttling prevents brute-force attempts
• XML security — added defusedxml for safe XML parsing
• Secret redaction expanded from 3 to 12 environment variables in error logs
• Session hardening — SameSite=Lax, HttpOnly cookies, X-Frame-Options and X-Content-Type-Options headers
• Path traversal guards on file-serving endpoints
• XSS fixes on KoSync API endpoints (Snyk SAST findings)

KoSync Refactor

• Monolith decomposition — kosync_server.py (1000+ lines) split into 5 focused modules: protocol routes, auth, admin dashboard, business logic service, and debounce manager
• KoSync Document Management page — new dedicated UI for triaging documents (healthy, needs attention, stale 30+ days) with search, linking, and deletion
• Debounce manager — batches rapid KoReader PUT events before triggering sync cycles
• Auto-discovery resilience — better error handling, misleading log fixes, correct UTC timestamps

Suggestions

• Decoupled from Audiobookshelf — suggestions now work for KoReader-only, Storyteller, and Grimmory users without ABS configured
• Cross-ebook pairings — match KoReader documents to Grimmory/Storyteller ebooks for position sync
• Dashboard suggestion banner — high-confidence suggestions surface on the home page with one-click mapping
• Suggestion count badge in navbar
• Match scoring fix — no longer penalizes ebook sources missing author metadata

Hardcover

• Increased API timeouts (10s → 20s)
• distinct_on dedup prevents duplicate user book entries
• Format label mapping for edition types
• Read ID caching to reduce API calls during sync

Booklore → Grimmory Rename (closes #49)

• Full rename across codebase, database, and UI
• Alembic migration handles table, column, constraint, settings, and data renames
• Legacy booklore_cache.json files are auto-migrated on upgrade

Code Quality

• Ebook utils decomposition — ebook_utils.py split into koreader_xpath.py and locator_search.py with facade pattern
• Shared constants module — magic strings extracted to src/utils/constants.py
• Test suite expanded to 805 tests — 7 new test files covering debounce, rate limiter, KoSync service, KoReader XPath, locator search, and Grimmory client
• Dead code cleanup, unused import removal, and null-safety fixes across 196 files

Bug Fixes

• Fixed NOT IN with nullable column silently skipping stale suggestion cleanup
• Fixed cached Hardcover read_id overwriting actual start/finish dates
• Fixed JSON null becoming literal string "None" in BookFusion uploads
• Fixed BookFusion highlight unlink silently succeeding when book not found
• Fixed XPath offset calculation when falling back to parent text nodes
• Fixed BS4/LXML text ordering divergence in hybrid anchor mapping
• Fixed datetime.utcnow().timestamp() using local timezone instead of UTC

v0.1.6

Frontend overhaul, testing expansion & UX improvements

This release focuses on frontend code quality, test coverage, and UX polish across the app.

Frontend

• Extracted frontend from 5 templates into 7 external files, making the codebase easier to maintain and debug
• Shared utility library (utils.js) put in place to replace 6+ duplicated implementations

Testing

• Test suite expanded from 461 to 719 tests (+258 new tests across 17 files)
• New coverage for: BookFusion routes, logs routes, dashboard errors, matching errors, reading blueprint errors, API errors, BackgroundJobService, ReadingDateService, BookMetadataService, ReadingService, ClientPoller, settings hot-reload, sync concurrency, and more
• Canonical MockContainer, shared pytest fixtures, and make_test_book/make_test_state helpers for consistent test authoring

UX Improvements

• Smart mode defaults — auto-detect which services are configured and default to available options
• Service logo placeholders — when book covers are unavailable, show the service logo instead of a broken image
• Portrait covers — ebook cards now display portrait-oriented covers instead of square
• Book metadata caching — author and subtitle are cached locally for faster page loads
• Suggestions page — inline refresh, modal errors, hidden when ABS is not configured
• Batch match — ebook-only support, de-centered from ABS
• Storyteller UI hidden when unconfigured
• Cover proxy — fixed streaming and N+1 query, added fallback for missing covers

Bugs Fixed

• Dashboard auto-refresh pausing permanently after closing a modal via button click
• BookFusion covers not loading due to broken onerror fallback
• Offset/pagination bug with missing null guards
• Silent error handling in multiple catch blocks (now surfaces actual error messages)
• Dead code cleanup (lastLogTimestamp, unused getattr calls)
• Live log fetch failures now display inline errors instead of failing silently
• Log filter state detection replaced fragile DOM text matching with proper boolean flag

Upgrading from v0.1.5

No database migrations in this release. Standard container update:

docker compose pull && docker compose up -d

v0.1.5

Decouple data model from Audiobookshelf ID (#20)

PageKeeper no longer requires Audiobookshelf. Technical tldr: Books now use an internal integer primary key (book.id) instead of the ABS library item ID (abs_id). This enables better standalone ebook tracking, setups that only use something like BookFusion, KoSync/KoReader, and Hardcover, and other various configurations without ABS. As apps like Grimmory progress this will also allow using something like Grimmory as your main audiobook driver instead of ABS.

What changed (the details)

• New integer primary key on the books table - all 11 child tables (states, jobs, kosync_documents, hardcover_details, etc.) now reference book.id via foreign key
• abs_id is now no longer needed - books can exist without an Audiobookshelf counterpart
• abs_title renamed to title - the column name no longer implies ABS ownership
• Backward compatible - existing ABS-linked books keep their abs_id and all external integrations (KoSync, Hardcover, Storyteller, Booklore) continue to work unchanged
• Write-suppression and debounce keyed by book.id - fixes a bug where ebook-only books (with abs_id=None) would collide into a single cache/debounce entry
• KoSync document linking now correctly sets both linked_book_id and linked_abs_id, fixing an issue where freshly linked documents were invisible in the admin UI
• get_book_by_abs_id(None) guarded - returns None immediately instead of matching NULL rows
• Schema creation gated on migration success - Base.metadata.create_all() no longer runs after a failed Alembic migration, preventing half-built schemas

Upgrading from v0.1.4

⚠️ Back up your database before upgrading.

This release runs 7 schema migrations that restructure the primary key and foreign keys across the entire database. The migrations are designed for production SQLite and include re-run safety, but a backup is strongly recommended.

cp /path/to/pagekeeper/data/database.db /path/to/pagekeeper/data/database.db.v0.1.4.bak

Migrations run automatically on container startup. No manual steps required after the backup. Please report any bugs you run into when updating.

Other Bugs Fixed

• Reading detail page: all action buttons broken due to undefined _rdAbsId JS variable
• KoSync debounce: ebook-only books collapsed into single debounce entry
• Client poller: poll cache and write-suppression collisions for ebook-only books
• KoSync admin: freshly linked documents not appearing in linked list
• Database startup: create_all() could mask migration failures

v0.1.4

What's Changed

• feat: add dev environment with separate Docker compose and tooling
• fix: replace startup sleep with KoSync readiness probe
• fix: add scripts to dev-restart-reminder hook pattern
• ci: add multi-arch build (amd64 + arm64)

v0.1.3

What's New in 0.1.3

• Bug fix - Allow deletion of resumed/paused/dnf journal entries
• Housekeeping - Remove stale CHANGELOG.md, replace CLAUDE.md and run-tests-fast.sh with CONTRIBUTING.md

v0.1.2

What's New in 0.1.2

Continued forking and updating PageKeeper including:

• Hardcover sync overhaul - Bidirectional status/progress sync, journal push, reading date pull/push, sync log, edition selection
• Reading tracker - Reading detail pages with journal, manual progress, ratings, and dates
• TBR list - To Be Read list with Open Library enrichment, Hardcover import, detail pages, and mark-as-read flow
• BookFusion integration - Highlight sync and shelf sync
• Dual Booklore servers - Multi-instance support
• Storyteller improvements - Direct submission from match page, force mode, status badges, and native word-level alignment
• UI/UX rework - Redesigned match page, settings, cover picker, and suggestions page
• Architecture fixes - Extracted services (Suggestion, BackgroundJob, ProgressReset, Reading, HardcoverService, StatusMachine) from monolithic files
• Security hardening - Sanitized exception logging, path traversal protection, XSS escaping, CVE-2025-43859 fix
• Infrastructure - Docker distribution prep with ARM64 docs, CI linting, fast test runner, 429 tests

v0.1.1

v0.1.1 — Security, Integrations, Reading Tracker, and TBR

• Hardcover bidirectional sync — Reading status and progress now syncs both directions with Hardcover
• Hardcover journal push — Journal notes are pushed to Hardcover as user journal entries
• Hardcover reading dates — Historical started_at/finished_at dates tracked and synced
• Hardcover sync log — New sync log tab showing all operations with status, direction, and timestamps
• BookFusion integration — Upload books, sync highlights via Obsidian plugin API, save highlights to reading journals
• Reading tracker — Reading detail page with journal entries, manual progress, ratings, read counts, and yearly reading goals
• TBR (To Be Read) list — Want to Read tab with search, manual add, Hardcover import, enrichment, and detail pages
• Storyteller submission service — Submit books directly to Storyteller for narrated EPUB3 creation
• Suggestions system — Dedicated page with card grid, socket-driven discovery, and Dismiss/Link/Never Ask actions
• Cover management — Cover picker modal, custom cover URL support, Hardcover cover search
• Architecture decomposition — Extracted SuggestionService, BackgroundJobService, ProgressResetService, ReadingService, BookMetadataService, and database repositories
• Dashboard UX rework — Processing section with progress bars, status filters, and live polling

Full Changelog: v0.1.0...v0.1.1