# Develop Changelog

## 2026-06-12 — Chapter editor: optional visual (WYSIWYG) mode alongside Monaco

### Added
- The chapter editor now has an optional **Visual (WYSIWYG) mode** next to the existing Monaco HTML-source editor, ported from the Tiptap editor built in novela-ng and adapted to Novela's own markup conventions. A **Visual / HTML** toggle button in the toolbar switches between the two; Monaco remains the default and the backing source of truth.
  - **Tiptap is bundled, not CDN-loaded.** A new `containers/novela/editor-src/` npm project bundles Tiptap + the custom extensions into a single committed file, `static/editor-bundle.js` (`window.NovelaVisual`), loaded by the editor page with a plain `<script>` tag — the runtime frontend stays build-less. Rebuild with `cd containers/novela/editor-src && npm install && npm run build`. `node_modules` is git-ignored; the bundle is committed.
  - **Novela conventions modelled as a real ProseMirror schema** so they apply and round-trip as formatting instead of raw tags: subheading (`span.subheading`), chat (`span.chat`), author comment (`div.novela-comment`), indented paragraph (`<p style="padding-left: 40px;">`), and scene break (`<center><img src="…break.png" style="height:15px;"/></center>` — the original img `src` is preserved, so both DB `/static/break.png` and on-disk EPUB `../Images/break.png` breaks round-trip exactly). Standard bold/italic/underline/superscript/subscript, H2/H3 headings, and bullet/numbered lists are also available as visual-only toolbar buttons.
  - **Detect & block (no silent data loss).** ProseMirror drops any markup its schema doesn't model, which would silently strip formatting from arbitrary EPUB/DB HTML. Switching into Visual mode is therefore gated by a round-trip safety check (`NovelaVisual.roundtripSafe`): the chapter is fed through the editor and the result is compared, canonicalised, against the original (ignoring insignificant formatting differences such as attribute order, whitespace, self-closing style, and `b`/`strong` & `i`/`em` equivalence). If anything — a tag, attribute, or text — would be altered or dropped, Visual mode is refused for that chapter and editing stays in HTML mode. In practice this means Visual mode is offered for clean, convention-conforming content (typically DB books) and declined for heterogeneous EPUB markup or whole-xhtml documents, so a book's existing formatting can never be quietly destroyed.
  - The semantic toolbar buttons (Break, S, C, →|, [ ]) route to the corresponding Tiptap commands in Visual mode and to the existing Monaco text-wrapping in HTML mode. Visual edits participate in the existing per-chapter dirty-tracking, Save / Save-all, chapter switching, add/delete, and Find & Replace flows; the visual content is synced back into Monaco (the backing store) at every save/switch/replace boundary, and the safety gate is re-evaluated whenever a chapter is loaded. Ctrl/Cmd+S also saves while editing visually.
  - Files: new `containers/novela/editor-src/{package.json,build.mjs,index.js,extensions.js,test.mjs,test-commands.mjs,test.html,.gitignore}`; new committed `static/editor-bundle.js`; `templates/editor.html` (bundle script, toggle button, visual-only toolbar group, `#visual-pane`); `static/editor.css` (visual pane, toolbar and convention styling); `static/editor.js` (pane abstraction, `toggleVisual`/`mountVisual`/`syncVisualToMonaco`/`refreshVisualAfterLoad`, boundary syncing, toolbar routing). No backend changes — chapter save is unchanged and does not sanitise, so the editor's clean output is stored as-is.

## 2026-06-03 — Reader: Page Up/Down scroll within the page

### Changed
- In the reader, **Page Up / Page Down no longer switch chapters** — they now fall through to the browser's default in-page scrolling. Chapter navigation is still bound to the **arrow keys** (Left/Right). Previously both the arrow keys and the page keys triggered chapter navigation, so accidentally pressing Page Down while reading jumped to the next chapter and lost the reading position.
  - `templates/reader.html`: the `keydown` handler now matches only `ArrowRight`/`ArrowLeft` for `navigate(±1)`; `PageDown`/`PageUp` removed from the bindings.

*Released as v0.2.14 on 2026-06-03*

## 2026-06-01 — Full Database Restore: live progress + download-to-local first

### Added
- Full database restores (Dropbox, local and upload) now run as a **background task with live byte-level progress**, polled by the Backup page, instead of a silent blocking request. This matters for large (e.g. ~1 GB) production databases where the restore previously showed nothing for minutes.
  - Phases reported and shown with a percentage / size: **downloading** (Dropbox only) → **safety snapshot** → **resetting schema** → **loading** → done / rolled back / failed.
  - Load progress is derived from feeding the dump to `psql` over stdin in chunks; pipe back-pressure makes the byte counter track psql's real progress closely.
- **Dropbox restore now downloads the dump to the local store first**, then restores from that file — so the restore no longer depends on the network connection mid-load, and a reusable local copy is kept for token-free restores. (Previously the Dropbox dump was streamed straight into memory and not persisted.)
  - `routers/backup.py`: `RESTORE_PROGRESS` state + `_restore_prog`; `GET /api/backup/restore/progress`; `_apply_pg_dump_file` (stdin-streamed load with progress, stderr captured to a temp file to avoid pipe deadlock), `_pg_dump_safety_to_local` (safety snapshot with live size), `_download_dropbox_to_file`, `_restore_worker_sync`, `_start_restore` (single restore at a time). The three restore endpoints now start the background job and return `{started: true}`; `_run_pg_restore`/`_apply_pg_dump` (bytes-based) were replaced.
  - Local dump retention now sorts by mtime (robust across `pre-restore-…` and downloaded dump names).
  - `templates/backup.html`: shared `pollRestoreProgress` with phase labels and percentage/size; the three restore buttons start the job then poll; progress resumes on page reload if a restore is still running.

## 2026-06-01 — Sidebar version is display-only

### Changed
- The build version at the bottom of the sidebar is no longer a clickable link to the changelog — it is now a plain text display. It exists only to show that the running build number has incremented (so it's clear the new code is active).
  - `templates/_sidebar.html`: `<a href="/changelog" class="sidebar-version">` → `<span class="sidebar-version">`.
  - `static/sidebar.css`: dropped the link/hover styling; added `cursor: default` and `user-select: none`.

## 2026-06-01 — Local pre-restore snapshots + token-free / upload restore

### Added
- Every restore now writes its **pre-restore safety dump to a local store** on the config volume (`CONFIG_DIR/postgres_dumps/`, named `pre-restore-…`, with snapshot-equal retention), so the database can be restored or rolled back **without a Dropbox token**. Regular backups are intentionally left **Dropbox-only** — production backs up hourly, so writing every backup to disk would fill the volume.
  - `routers/backup.py`: `LOCAL_DUMP_DIR`, helpers `_save_local_dump`, `_list_local_dumps`, `_enforce_local_dump_retention`, `_resolve_local_dump`; `_run_pg_restore` persists the safety dump locally (and prunes by retention) before the destructive load. `_run_backup_internal` is unchanged (no local copy).
- New **Local Database Restore** card on the Backup page with two token-free paths:
  - Restore from a locally stored dump — `GET /api/backup/local/dumps`, `POST /api/backup/local/restore`.
  - Upload a `.sql` dump (e.g. downloaded manually from dropbox.com) and restore it — `POST /api/backup/upload-restore`. The upload itself is not persisted; the automatic pre-restore safety snapshot covers the local copy.
  - Both reuse the safe restore path (pre-restore safety dump + automatic rollback) and require `psql`.
  - `templates/backup.html`: new card with a local-dump selector, an upload field, double-guarded restore buttons, and `fmtBytes` sizes.

## 2026-06-01 — Full Database Restore: safety dump + automatic rollback

### Fixed
- Full Database Restore could leave the database **empty** if the dump failed to load: the restore dropped and recreated the public schema first and only then applied the dump, so any failure during the load (e.g. a header `transaction_timeout` error with `ON_ERROR_STOP`) wiped all data with nothing to fall back to.
  - `routers/backup.py`: split the load into `_apply_pg_dump`; `_run_pg_restore` now takes a safety `pg_dump` of the current database before the destructive load and, if applying the requested dump fails, automatically rolls back to that safety snapshot. A failed restore therefore no longer leaves an empty or broken database.

## 2026-06-01 — Full Database Restore: tolerate PostgreSQL version mismatch

### Fixed
- Full Database Restore failed with `ERROR: unrecognized configuration parameter "transaction_timeout"` when the dump was produced by a newer `pg_dump` (PostgreSQL 17+, which emits `SET transaction_timeout = 0;` in the header) but restored into an older server (<17) that doesn't know that session parameter. The restore ran with `ON_ERROR_STOP=1` and aborted on that harmless header line.
  - `routers/backup.py` (`_run_pg_restore`): the dump is now applied without `ON_ERROR_STOP`; stderr is inspected afterwards via `_real_restore_errors`, which ignores benign "unrecognized configuration parameter" errors but still fails the restore on any other `ERROR:` line. The schema-reset step keeps `ON_ERROR_STOP=1` (it is our own controlled SQL).

## 2026-06-01 — Backup/Restore: database-stored books are now restorable

### Fixed
- Database-stored books (`storage_type='db'`, synthetic `db/...` filenames) could not be restored through the Backup → Restore option. The restore UI only listed and restored files from the on-disk library object store, but db-books have no file on disk — their content lives entirely in PostgreSQL (`book_chapters` + `library` row + `book_tags` + `library_cover_cache`). They were captured only in the full `pg_dump`, which the UI offered no way to restore. As a result db-books never appeared in the restore list and were effectively unrecoverable per-book.

### Added
- **Per-book db restore.** The backup writer now serializes each database book to JSON (library row, all chapters, tags, and the cached cover) and stores it in the same content-addressed Dropbox object store used for file books, referenced from the snapshot with a `"storage": "db"` marker. Such books now appear in the Restore table (format **DB**) and can be restored individually; restore re-inserts the library row, chapters, tags and cover into PostgreSQL. Inline chapter images are unaffected — they already live on disk under `library/images/` and are backed up as ordinary files.
  - `routers/backup.py`: new `_db_book_filenames`, `_serialize_db_book`, `_restore_db_book`, `_entry_storage` helpers; db-books serialized during `_run_backup_internal`; `_download_and_restore` branches on storage type; `/api/backup/snapshots/{name}/files` now reports `storage` and computes `exists_locally` for db-books from the `library` table.
  - Note: db-books only appear in snapshots created *after* this change. Older backups still contain them only in the `pg_dump` — recover those via Full Database Restore below.
- **Full Database Restore.** New Backup-page card and endpoints to restore the entire PostgreSQL database from any Dropbox `pg_dump`. This recovers everything (all db-books, reading progress, tags, settings) from existing backups too. It is destructive: the public schema is dropped and recreated before the dump is applied, so any plain dump (with or without `--clean`) restores cleanly. Guarded behind a double confirmation in the UI.
  - `routers/backup.py`: `_psql_base_args`, `_run_pg_restore`, `_list_pg_dump_paths`; endpoints `GET /api/backup/postgres/dumps` and `POST /api/backup/postgres/restore`; health endpoint now also reports `psql_available`/`psql_path`.
  - `templates/backup.html`: db-books shown with **DB** format and "in library / not in library" status in the restore table; new "Full Database Restore" card with dump selector, double-confirm, and `psql` health row.

---
*Released as v0.2.13 on 2026-06-01*

## 2026-06-01 — Reader: Sepia reading theme

### Added
- The reader now offers a **Sepia** theme alongside the existing Dark theme, for easier long-form reading (warm paper background with dark brown text instead of light-on-black). New **Theme** toggle (Dark / Sepia) at the top of the reading settings drawer.
  - `static/theme.css`: added a `:root[data-theme="sepia"]` palette overriding `--bg`, `--surface`, `--surface2`, `--border`, `--accent`, `--accent2`, `--text`, `--text-dim`, `--text-faint`, `--success`, `--warning`, `--error`. Also added a `--shadow-ring` variable (light hairline in dark, dark hairline in sepia) so swatch rings read correctly in both themes.
  - `templates/reader.html`: theme toggle buttons + `.theme-btn` styling; an inline head script applies the saved theme before paint to avoid a flash; the Text colour row now has two theme-specific swatch sets (light tints for dark, dark brown tints for sepia), toggled by visibility.
  - JS: `applyTheme()`/`loadTheme()` set the `data-theme` attribute and persist to `localStorage` (`reader-theme`). Text colour is now stored per theme (`reader-text-colour-dark` / `reader-text-colour-sepia`); the old single `reader-text-colour` key is migrated into the dark slot. `init()` calls `loadTheme()` (which loads the active theme's text colour).
  - Replaced two hard-coded `rgba(255,255,255,...)` values in the reader CSS (swatch ring, series-button border) with theme variables so they don't show as white halos in sepia.

## 2026-05-31 — Build version in the sidebar

### Added
- The sidebar now shows the running build version at the bottom (e.g. `v0.2.11` for releases, `v0.2.11.3` for dev builds), linking to the changelog page.
  - New `containers/novela/version.py` exposes `display_version()`. The semantic release version stays the single source in `changelog.py` (`CHANGELOG[0]["version"]`); a `BUILD` segment is appended for dev builds.
  - `shared_templates.py` registers `app_version` as a Jinja global; `_sidebar.html` renders it as a `.sidebar-version` link styled in `sidebar.css`.
  - `main.py` adds a `/api/version` endpoint returning `{"version": display_version()}`.

## 2026-05-31 — Find & Replace: scope option

### Added
- Editor Find & Replace: new **Current chapter only** checkbox in the modal options. When checked, the search/replace runs against the currently open chapter instead of every chapter in the book.
  - `templates/editor.html`: added the `rp-current` checkbox next to Regex/Case sensitive; modal title changed from "Find & Replace — all chapters" to "Find & Replace" since scope is now selectable.
  - `static/editor.js` (`replaceInAllChapters()`): reads `rp-current`; builds a `targets` list of either just `currentCh()` or all `chapters`, and iterates that. Shows an error ("No chapter open.") if Current-chapter-only is selected with no open chapter. Result message reads "… in current chapter" for the single-chapter case.
  - Default is unchecked, so the existing all-chapters behaviour is preserved.

---
*Released as v0.2.12 on 2026-06-01*

## 2026-05-10
- Reader: subheading/chat styling now also wins when the wrapper contains block elements with their own color rule (e.g. `<div class="subheading"><p>…</p></div>`)
  - Previously the `<p>`/`<h*>` rules in `reader.html` (`#chapter-content p { color: var(--text); }`, etc.) had the same specificity as `.subheading` and applied more directly to the inner element, so the parent's color was effectively overridden — the wrapped paragraph stayed in the default text color
  - CSS now applies the rule to the wrapper *and* all descendants: `#chapter-content .subheading, #chapter-content .subheading * { … }` (and the same for `.chat`)
  - Reason: bug reported after v0.2.10 — `.chat` worked because it wraps plain inline text inside a `<p>`, but `.subheading` wrapping a `<p>` lost its color

---
*Released as v0.2.11 on 2026-05-10*

---
*Released as v0.2.10 on 2026-05-10*

## 2026-05-09 (2)
- Editor / Reader: subheading (**S**) and chat (**C**) styling now also applies when the wrapper is a `<div>` instead of a `<span>`
  - `wrapSpan()` in `editor.js` falls back to `<div class="…">` whenever the selection contains a block element (`<h1..6>`, `<p>`, `<div>`, etc.) so the resulting HTML stays valid; the previous CSS in `reader.html` (`#chapter-content span.subheading` / `span.chat`) only matched the `<span>` form, so anything wrapped around or inside a heading silently lost its color/weight
  - CSS selectors are now class-only (`#chapter-content .subheading` / `.chat`), matching both span and div wrappers

---

## 2026-05-09
- Reader: progress is now monotonic across devices — saved position only advances, never rewinds
  - `POST /library/progress/{filename}` in `routers/reader.py` parses the incoming `cfi` as `(chapterIndex, scrollFrac)` and the currently stored row the same way, then skips the write when `new_pos <= cur_pos`
  - Same `cfi` format is used for EPUB, PDF and CBR/CBZ, so one tuple comparison covers all readers
  - Explicit Read/Unread actions still clear the row (`mark-read` / `mark-unread` delete from `reading_progress`), so users can deliberately reset and start over
  - Reason: when reading the same book on device A (chapter 12) and then continuing on device B (chapter 15), opening device A again previously sent its stale chapter-12 cfi back to the server and overwrote the further progress; now the older position is ignored

---
*Released as v0.2.9 on 2026-05-09*

## 2026-04-22 (5)
- Grabber: newly converted books now appear in the **New** view again
  - Both the DB-storage and file-EPUB branches in `routers/grabber.py` set `needs_review: True` on `upsert_book` (was `False`); the New view filters on `needs_review == True`, so previously grabbed books never showed up there
  - Matches the behavior of disk-scanned imports (`library.py` already sets `needs_review = True` for freshly discovered files)

---
*Released as v0.2.8 on 2026-04-22*

## 2026-04-22 (4)
- Break detection: runs of consecutive break images are now collapsed to a single break
  - New helper `collapse_consecutive_breaks()` in `xhtml.py` matches 2+ consecutive `<center><img src=".../break.png" .../></center>` occurrences (with optional whitespace in between) and replaces them with one
  - Applied in `normalize_wysiwyg_html()` (editor save path) and in `routers/grabber.py` on both the preview `converted_xhtml` and the per-chapter `content_html` produced during scraping
- Docs: `docs/TECHNICAL.md` updated to cover previously missing changes — `POST /api/edit/intro/{filename}` and the `title` field on file-EPUB chapter save; FlareSolverr sidecar and `BaseScraper.close()`; `AwesomeDudeScraper` uses FlareSolverr; `make_epub(include_intro=…)` and `epub_utils.build_book_info_body_html`; grabber DB flow stores Book Info as chapter 0; `"Book Info"` h1-strip skip in reader; new env vars (`FLARESOLVERR_URL`, `FLARESOLVERR_TIMEOUT_MS`, `NOVELA_PORT`, `ADMINER_PORT`); `collapse_consecutive_breaks()` helper

---
*Released as v0.2.7 on 2026-04-22*

## 2026-04-22 (3)
- Scrapers: Cloudflare "Just a moment…" challenges are now solved via a new FlareSolverr sidecar service so books on protected sites (awesomedude.org) can be scraped again
  - New service `flaresolverr` in `stack/stack.yml` (image `ghcr.io/flaresolverr/flaresolverr:latest`, internal-only, on `novela-net`); `novela` service gains `FLARESOLVERR_URL=http://flaresolverr:8191/v1` and a `depends_on: flaresolverr`
  - New helpers in `scrapers/base.py`: `flaresolverr_get(url, timeout_ms=None, session=None)` posts `cmd: request.get` and returns a `SimpleNamespace(text, url)` (drop-in for `httpx.Response` attributes); `flaresolverr_session_create()` and `flaresolverr_session_destroy(sid)` manage browser sessions so Cloudflare cookies stay warm across per-chapter requests (first page solves the challenge, all further chapters reuse the same browser — much faster)
  - Configurable via `FLARESOLVERR_URL` and `FLARESOLVERR_TIMEOUT_MS` env vars (defaults: service DNS name and 60000 ms)
  - `BaseScraper` gained an async `close()` method (default no-op) for releasing scraper-scoped resources; `scrapers/awesomedude.py` creates a FlareSolverr session in `fetch_book_info`, reuses it in all `fetch_chapter` calls, and destroys it in `close()`
  - `routers/grabber.py` now wraps all scraper usage in `try/finally: await scraper.close()` so sessions are released even on errors
  - Stack uses `${NOVELA_PORT}` and `${ADMINER_PORT}` (defined in `stack/novela.env` as 8099 / 8098) so host-port values don't diverge between environments

---
*Released as v0.2.6 on 2026-04-22*

## 2026-04-22 (2)
- Editor: chapter titles are now editable for file-EPUB books as well (DB books already supported this)
  - Frontend (`editor.js`): the `chapter-title-input` is always shown (the read-only `header-chapter` label is hidden for both storage types), title changes mark the chapter dirty, and the title is sent in both `saveChapter` and `saveAllChapters`
  - Backend (`routers/editor.py`): `POST /api/edit/chapter/{index}/{filename}` for file EPUBs now accepts `title` alongside `content`; if the title changed it calls new helper `_update_epub_navpoint_title(path, href, new_title)` which locates the matching NCX `navPoint` by content src and rewrites its `<text>` via `rewrite_epub_entries`

---

## 2026-04-22
- Book Info page: new "Info page" button in the editor toolbar generates a gayauthors-style info page and inserts it as the first chapter
  - New endpoint `POST /api/edit/intro/{filename}` builds the page from stored metadata (author, genres, sub-genres, tags, description, source, updated) and prepends it
  - DB books: shifts existing `chapter_index` values up by one and inserts `"Book Info"` at index 0 (two-step negation to avoid unique-constraint violations)
  - File EPUBs: writes a new `intro_<hex>.xhtml` via `make_intro_xhtml`, adds a manifest item, places the `itemref` at the start of the spine, and inserts a navPoint at the top of the NCX with renumbered `playOrder`
  - Empty metadata fields are skipped
  - Option A: no duplicate detection — clicking the button on a book that already has one will add a second page
- Grabber convert: DB-storage conversions now persist the Book Info page as a real stored chapter at index 0 (so it shows up in the editor and reader); EPUB-storage conversions continue to produce `intro.xhtml` via `make_epub` as before
- DB → EPUB export (`reader.py`): no longer synthesises `intro.xhtml` (`make_epub(..., include_intro=False)`) — the stored chapter 0 is the single source of truth
  - Legacy DB books converted before this change have no stored info page; their exports will lack an intro page until the button is used
- `make_epub` gained an `include_intro: bool = True` parameter; when false, the `intro.xhtml` file, its manifest item, its spine `itemref`, and its NCX navPoint are all omitted (remaining `playOrder` values start at 1)
- Shared helper: `epub_utils.build_book_info_body_html(title, author, info)` — returns the inner-body HTML fragment used for DB storage, starting with `<h1>{title}</h1>`; skips empty fields and separates description and source/updated blocks with `<hr/>`
- DB-storage info page: chapter title is `"Book Info"`; to preserve the leading `<h1>{book title}</h1>` in its body, the leading-h-tag stripping in `reader.py` (`get_chapter_html` and the DB→EPUB export) is now skipped when `title == "Book Info"`

---
*Released as v0.2.5 on 2026-04-22*

## 2026-04-21 (2)
- Backup: separate "scanned" vs "uploaded" counters in backup_log and UI
  - New migration `backup_log_scanned_files` adds `scanned_files INTEGER` column to `backup_log`
  - `_run_backup_internal` now returns `(scanned_files, uploaded_count, uploaded_size)`
  - `_finish_backup_log(...)` accepts `scanned_files=`; `_run_backup_job` passes the value through
  - `/api/backup/status` and `/api/backup/history` return `uploaded_files` (old `files_count`) and `scanned_files`
  - `backup.html`: Latest Status now shows separate "Scanned" and "Uploaded" rows; History table renamed "Files" column to two columns "Scanned" / "Uploaded" (colspan of loading/empty states updated to 8)
- Backup page: live phase indicator under the Run buttons while a backup is running
  - New `#run-progress` status line is filled by `pollRunProgress()` which polls `/api/backup/progress` every 3 s
  - Phase labels spelled out: `scanning library`, `uploading library objects`, `uploading snapshot`, `uploading pg_dump (may take minutes)` — so users understand why the counter sits at `N / N` during the final phases
  - Polling starts on page load (via `refreshAll()`) and is re-kicked when Run Live/Dry is clicked; it auto-stops when progress reports `running=false`

---
*Released as v0.2.4 on 2026-04-21*

## 2026-04-21
- Backup: fix Dropbox `Read timed out. (read timeout=120)` error on large uploads
  - `dropbox.Dropbox(...)` `timeout` raised from `120` to `300` in `routers/backup.py` (`_dbx()`, both refresh-token and legacy-token branches)
  - `_DROPBOX_UPLOAD_CHUNK` reduced from `100 * 1024 * 1024` (100 MB) to `16 * 1024 * 1024` (16 MB)
  - `_DROPBOX_UPLOAD_THRESHOLD` lowered to match (`16 * 1024 * 1024`) so the session upload path is used earlier
  - Net effect: each chunk PUT finishes well within the socket timeout, and a stalled connection gets 5 minutes instead of 2 before erroring

---
*Released as v0.2.3 on 2026-04-21*

## 2026-04-16 (2)
- Editor: four inline formatting buttons added to the chapter editor toolbar
  - **S** (subheading) — wraps selection in `<span class="subheading">` (red, bold in reader)
  - **C** (chat) — wraps selection in `<span class="chat">` (orange in reader)
  - **→|** (indent) — wraps selection in `<p style="padding-left: 40px;">` (or `<div>` when the selection contains block elements)
  - **[ ]** (comment) — wraps selection in `<div class="novela-comment">` (blue left border + subtle background in reader)
  - Without a selection: the tag is inserted at the cursor with the cursor positioned inside
  - All four buttons disabled when no chapter is loaded; enabled state mirrors the existing Break button
  - Wrap logic auto-detects block content in the selection: `wrapSpan` and `insertIndent` use a `<div>` wrapper instead of `<span>`/`<p>` when the selection contains `<p>`, `<div>`, `<h*>` etc. to keep the HTML valid
  - Reader CSS extended: `span.subheading`, `span.chat`, `p[style*="padding-left"]`, and `.novela-comment` are styled in `reader.html`

## 2026-04-16 (1)
- Startup: migration logging zichtbaar in Docker logs
  - `logging.basicConfig(level=logging.INFO)` toegevoegd aan `main.py`
  - `migrations.py` logt per migratie of deze overgeslagen of uitgevoerd wordt (met duur in ms)
  - Samenvattingsregel bij afsluiting: "all already applied" of "N executed"

---
*Released as v0.2.1 on 2026-04-16*

## 2026-04-15 (4)
- Scraper: fix encoding for Codey's World
  - Pages are decoded with `cp1252` (Windows-1252) instead of relying on `r.text` or `html.parser` auto-detection via `iso-8859-1`
  - `cp1252` correctly maps the 0x80–0x9F byte range: `…`, `'`, `'`, `"`, `"`, `–`, `—` etc. now render correctly instead of producing replacement characters
- XHTML: normalize non-breaking spaces globally
  - `\xa0` (HTML `&nbsp;`) is now replaced with a regular space before HTML-escaping in both `element_to_xhtml` and `normalize_wysiwyg_html`
  - Consecutive spaces resulting from this substitution are collapsed to a single space
  - Applies to all scrapers

## 2026-04-15 (3)
- Editor: chapter add/delete deferred until Save is clicked
  - Adding or deleting a chapter no longer triggers an immediate API call
  - All structural changes are collected in memory and applied (in the correct order) when the Save button is pressed
  - Deletes are applied in reverse server-index order to avoid index shifting errors; new chapters are appended afterwards
  - Fix: `saving` flag is set at the start of a delete operation to prevent a concurrent `saveChapter` from flushing pending changes during the async gap in `loadChapter`
- Refactor: fix unclosed file handles in `epub.py`
  - `make_epub` and `write_epub_file` now use `Path.read_text()` / `Path.read_bytes()` instead of bare `open()` calls
- Refactor: eliminate temp file in `pdf_cover_thumb`
  - Cover thumbnail is now generated fully in-memory via `Image.frombytes()` from the PyMuPDF pixmap — removes the race condition when multiple requests hit the same PDF simultaneously
- Refactor: harden `security.py`
  - Hardcoded fallback encryption key removed; raises `RuntimeError` with a clear message when neither `NOVELA_MASTER_KEY` nor `POSTGRES_PASSWORD` is set
  - Fernet instance cached with `@functools.lru_cache(maxsize=1)` — key derivation runs only once per process
- Add `GET /health` endpoint
  - Runs `SELECT 1` against the database and returns `{"ok": true/false}`
- Performance: cap in-memory job dicts at 50 entries
  - `JOBS` in `grabber.py` and `BACKUP_TASKS`/`BACKUP_PROGRESS` in `backup.py` are trimmed (oldest-first) whenever a new entry is added
- Performance: improve ETag accuracy for the library API
  - ETag now includes `MAX(updated_at)` from `reading_progress` and `MAX(id)` from `book_tags`, so tag changes and progress updates correctly invalidate the client cache
- Performance: cache CBR/CBZ page list
  - `cbr_page_list` is cached per `(path, mtime)` via `lru_cache(maxsize=64)` — avoids opening the archive twice per page request
- Refactor: normalize transaction handling in `builder.py`
  - All `conn.commit()` calls replaced with `with conn:` context manager, consistent with the rest of the codebase

---
*Released as v0.2.0 on 2026-04-15*

## 2026-04-15 (2)
- Editor: fix chapter add failing with UniqueViolation on DB-stored books
  - PostgreSQL checks the unique constraint on `(filename, chapter_index)` per row during `UPDATE`, so incrementing consecutive indices in a single statement (e.g. 1→2 while 2 exists) raised a `UniqueViolation`
  - Fixed by using a two-step approach: first shift affected rows to temporary negative values, then to their final positive values
- Refactor: eliminate duplicated EPUB helper functions across `reader.py`, `editor.py`, `common.py`
  - New `epub_utils.py` with shared `find_opf_path`, `norm_href`, `epub_spine`, `make_new_chapter_xhtml`, `rewrite_epub_entries`
  - Fixes the double-escaped `\\\\s*` regex in the old `_epub_spine` OPF path lookup (was silently falling back to directory scan)
  - `rewrite_epub_entries` combines crash-safe `.tmp.epub` write with `ZIP_STORED` for the `mimetype` entry (EPUB spec requirement)
  - All private `_epub_spine`, `_norm_href`, `_find_opf_path`, `_make_new_chapter_xhtml`, `_rewrite_epub_entries` copies removed from `reader.py`, `editor.py`, `common.py`
- Migrations: run each migration only once via `schema_migrations` tracking table
  - Eliminates heavy `rebuild_chapter_tsv_with_title` UPDATE running on every container restart
  - Reduces startup from 37 separate DB connections to 1

## 2026-04-15 (1)
- Reader: font size control in reading settings
  - New "Font size" slider (80–150%, default 105%) in the settings drawer, between "Content width" and "Text colour"
  - Applies via CSS custom property `--reader-font-size` on `#chapter-content`
  - Persisted per-device in `localStorage` as `reader-font-size` — iPad and desktop each remember their own preference

## 2026-04-13 (1)
- Edit metadata: comma-separated tag input fix
  - `PillInput._add` in `book.js` now splits the incoming value on commas before adding — each trimmed, non-empty, non-duplicate part is pushed individually
  - Applies to genres, subgenres and plain tags; pasting e.g. `Fiction, Thriller, Adventure` adds 3 separate pills instead of 1

---
*Released as v0.1.11 on 2026-04-13*

## 2026-04-12 (2)
- Series navigation in the reader
  - New `GET /api/series-nav/{filename}` endpoint: returns `{prev, next}` with `{filename, title, index, suffix}` for adjacent books in the same series, ordered by `series_index ASC, series_suffix ASC`; returns `{prev: null, next: null}` for books without a series
  - Reader header now shows prev/next series buttons (skip-to-start/skip-to-end icons); hidden for books without a series, visible once `loadSeriesNav()` resolves
  - Hovering a series button shows a tooltip: `#<index><suffix> <title>` (e.g. `#4 Batman: Year One`)
  - `markRead()` in the reader redirects to `/library/read/{next.filename}` when a next volume exists, so reading continues without leaving the reader; falls back to the book detail page when the series is complete or the book has no series

## 2026-04-12 (1)
- Comics: series_volume support for annual series (issue numbers restart each year)
  - New `series_volume VARCHAR(20)` column on `library` (migration `migrate_series_volume`); default `''`
  - Stored in EPUB OPF as `<meta name="novela:series_volume" content="…"/>`; read back by `scan_epub`
  - `upsert_book` inserts/preserves `series_volume` with the same COALESCE strategy as `series_suffix`
  - `list_library_json` ORDER BY now: `publisher → author → series → series_volume → series_index → title`
  - `PATCH /library/book/{filename}` reads `series_volume` from request body; persists to DB and OPF for both file-based and DB-stored books
  - Book detail page: displays `(year)` after series name when `series_volume` is set (e.g. `Donald Duck (1982) [15]`); edit panel has a new "Year/Volume" input field
  - `book.js`: "Auto" button next to the Title field generates `Series (Year/Vol) #Number` from the current series fields
- Bulk importer: `series_volume` support
  - New `%series_volume%` placeholder token (orange) for filename pattern parsing
  - New "Year/Vol." shared metadata field (applies to all files; overridden by per-row value or pattern)
  - Preview table has a new "Yr/Vol" column
  - New "Auto-generate titles from series info" checkbox: when enabled, rows without a parsed title get title `Series (Year/Vol) #Number`
  - Skip checkbox now always visible for every row (previously only shown when duplicates were detected); any file can be manually excluded before import; skipped rows are dimmed; stats bar shows "X skipped"
  - `POST /api/bulk-check-duplicates` extended: now also checks `(series, series_index, author)` as fallback — detects duplicates even when the title format has changed; items must include `series` field
- Library front-end: sorting within a series now respects `series_volume` before `series_index`
  - `groupBySeries`, `renderAuthorDetail`, `renderPublisherDetail` all sort by `series_volume → series_index → series_suffix`
  - `getSeriesSlots` refactored: when any book in the series has `series_volume` set, gap-detection runs per volume (year) independently — prevents `#5 (1982)` and `#5 (1983)` from colliding in the same slot
  - Slot index label shows `(year) #index` for annual series (e.g. `(1982) #5`); unchanged for regular series

---
*Released as v0.1.10 on 2026-04-12*

## 2026-04-08 (13)
- Library: archive series in one action
  - "Archive series" / "Unarchive series" button in the series detail view (`#series/{name}`)
  - New endpoint `POST /library/archive-series` — sets `archived` for all books in the series via a single SQL UPDATE; body: `{"series": "…", "archive": true|false}`; returns `{ok, archived, count}`
  - Button label reflects current state: "Archive series" when any book is active, "Unarchive series" when all are archived
  - After the call, `allBooks` is updated in place and sidebar counters are recalculated without a full page reload

## 2026-04-08 (12)
- TedLouisScraper: title extraction fix
  - `<h2 class="story-page-title">` also contains a "Back" button (`<a class="btn">`) and the author byline (`<span class="story-author-by-line">`)
  - Fix: only direct `NavigableString` children of the h2 are used as the title, so link and span text is skipped

## 2026-04-08 (11)
- New scraper: `TedLouisScraper` (`scrapers/tedlouis.py`) for `tedlouis.com`
  - Matches all `tedlouis.com` URLs; no login required
  - Entry point is the story index page (e.g. `?t=CWYSqpOryu2rQmT1`); raises an error when a chapter URL is used as entry point
  - Title from direct text nodes of `<h2 class="story-page-title">`
  - Author from `<span class="story-author-by-line"> <a>`
  - Status from `<span class="story-status-text">` (strips "Status: " prefix)
  - Updated date from `<span class="story-last-updated">` → `YYYY-MM-DD`
  - Chapters from all `<ul class="story-index-list">` elements (three columns); relative `?t=TOKEN` links resolved to absolute URLs
  - `fetch_chapter()`: content from `<div id="chapter">`; removes story title, chapter title and copyright blocks

## 2026-04-08 (10)
- Settings: break image upload added
  - New "Break image" card on the settings page: upload a PNG/JPG/WebP as the scene break image
  - Stored in the imagestore (sha256-addressed) and overwrites `static/break.png` so EPUB export uses the same image
  - `app_settings` extended with `break_image_sha256` and `break_image_ext` columns (migration `migrate_app_settings_break_image`)
  - New endpoints: `POST /api/app-settings/break-image`; `GET /api/app-settings` now also returns `break_image_url`
  - Preview of the current break image visible on the settings page

## 2026-04-08 (9)
- Grabber: break image fix for DB-stored books
  - Break images (`<center><img src="../Images/break.png">`) contain a relative EPUB path that does not exist in the DB context
  - Fix: `storage_mode` determined earlier in `_run_scrape`; for DB mode `/static/break.png` is passed as `break_img_path` to `element_to_xhtml`, for EPUB mode `../Images/break.png` is kept
  - Additional fix: when a break image is detected in the image loop, the parent (`<center>`) is replaced with `<hr>` instead of decomposing the `<img>`; this allows `element_to_xhtml` to correctly detect the break after the image loop

## 2026-04-08 (8)
- New scraper: `IomfatsScraper` (`scrapers/iomfats.py`) for `iomfats.org`
  - Matches all `iomfats.org` URLs; no login required
  - Entry point is a chapter URL (e.g. `.../grasshopper/justhitsend-part1/justhitsend01.html`); automatically navigates to the author page to fetch all metadata and chapters
  - Author page as entry point: raises a user-visible error message
  - Detects two structures on the author page:
    - Single story: outer `<h3>` = book title; chapters directly in `<ul>`
    - Multi-part series: outer `<h3>` = series name; nested `<li><h3>` = book title per part; chapters in sub-`<ul>`
  - Series index from folder name: `*-part{N}` or `*-{N}` → `series_index_hint`
  - Publication status from `<p><small>[...]</small></p>` after book title
  - `fetch_chapter()`: content via `<div id="content">`; removes `<h2>`/`<h3>` headings, chapternav divs, footer elements

## 2026-04-08 (7)
- New scraper: `CodeysWorldScraper` (`scrapers/codeysworld.py`) for `codeysworld.org`
  - Matches all `codeysworld.org` URLs; no login required
  - Title from `<h1>`; author from `<h2>` ("by …"); fallback to URL slug (`/{author}/{category}/filename`)
  - Category from URL path → tag (e.g. "Remembrances")
  - Single-file stories (no chapter links): the page itself is the only chapter
  - Multi-chapter: links to `.htm`/`.html` files in the same directory (audio/image links skipped)
  - `fetch_chapter()`: removes all `<h1>`/`<h2>` headings, navigation links ("Back to …"), audio links (`.mp3`), mailto links
- Nifty scraper: category/subcategory moved from `genres`/`subgenres` to `tags`

## 2026-04-08 (6)
- NiftyNewScraper: chapter extraction made more robust
  - `fetch_chapter()` now first tries the standard chapter HTML (`{url}`) and reads `<article><p>` directly
  - Fallback added for Next payloads with escaped paragraphs (`\u003cp...\u003c/p`) via `_extract_escaped_html_paragraphs()`
  - Last fallback remains `?_rsc=1`: first `_parse_rsc_paragraphs()`, then escaped-paragraph fallback again
- Nifty (classic + new): standard boilerplate no longer visible in reader, but preserved in chapter
  - Lead/tail detection added for common blocks (e.g. `NOTICE This is a work of fiction…`, `If you enjoy this story…`, donate text)
  - Detected intro/closing boilerplate removed from visible paragraphs
  - Removed text stored as invisible HTML comment:
    - `<!-- NIFTY_HIDDEN_LEAD: ... -->`
    - `<!-- NIFTY_HIDDEN_TAIL: ... -->`
  - Detection also works when text contains inline HTML (e.g. `<a>` or `<svg>` in donate links)

## 2026-04-08 (5)
- NiftyNewScraper: chapter content fix
  - `new.nifty.org` uses Next.js RSC — chapter content is not present in the static HTML but in the RSC payload (React component tree)
  - `fetch_chapter()` now fetches `{url}?_rsc=1` instead of the regular HTML page
  - RSC parser added: `_parse_rsc_paragraphs()`, `_rsc_find_paragraphs()`, `_rsc_text()` — parse the RSC stream line by line (format: `{hex_id}:{json}`), recursively search for `["$","p",null,{...}]` nodes and extract text from `children`

## 2026-04-08 (4)
- NiftyNewScraper added (`scrapers/nifty_new.py`) for `new.nifty.org`
  - Matches all `new.nifty.org` URLs; no login required
  - `_to_index_url()`: strips trailing `/N` (chapter number) so both index and chapter URLs can be used as entry point
  - `fetch_book_info()`: title from `<h1>`, author from `<strong itemprop="name">` in author link, dates from `<time itemprop="datePublished/dateModified">`, tags from all `<ul aria-label="Tags">` containers (category links and generated tags, deduplicated), description from `<meta name="description">`, chapter list via `<a>` links matching `/stories/{slug}/N` (RSC stream regex as fallback)
  - `fetch_chapter()`: title from JSON-LD `@type: "Chapter"`, content from `<article>`; no email header stripping, no line joining, no boilerplate detection
  - `NiftyScraper.matches()` updated: excludes `new.nifty.org` (`"nifty.org" in url and "new.nifty.org" not in url`)
  - `NiftyNewScraper` registered before `NiftyScraper` in `scrapers/__init__.py`

## 2026-04-08 (3)
- Settings: develop mode added
  - New `app_settings` table (single row, `id = 1`) with `develop_mode` boolean; created via `migrate_create_app_settings()`
  - `shared_templates.py`: shared `Jinja2Templates` instance for all routers; `develop_mode()` registered as Jinja2 global so all templates can access it without explicit context injection
  - All 11 routers now import `templates` from `shared_templates` instead of each creating their own instance
  - New endpoints in `routers/settings.py`: `GET /api/app-settings` and `PATCH /api/app-settings`
  - Diagonal **DEVELOP** banner in the top-left of every page (CSS in `static/sidebar.css`, HTML in `templates/_sidebar.html`); only visible when develop mode is enabled
  - All 17 HTML templates: `<title>` shows **Novela Develop — …** when develop mode is active
  - Settings page: new card with checkbox toggle; page reloads after saving so banner and title apply immediately

## 2026-04-08 (2)
- Nifty scraper: fix `_strip_email_headers` now tolerates blank lines between header fields
  - Some Nifty pages place `Subject:` after a blank line (`Date:\nFrom:\n\nSubject:\n`) — the previous implementation stopped at the first blank line causing `Subject:` to appear as a paragraph in chapter text
  - Fix: on a blank line, look ahead to check whether a header field still follows; if so, skip the blank line(s) and continue stripping

## 2026-04-08 (1)
- Nifty scraper added (`scrapers/nifty.py`)
  - Matches all `nifty.org` URLs; no login required
  - `fetch_book_info`: accepts index or chapter URL; normalises to index; title from URL slug, author and publication date from email headers of chapter 1, `updated_date` from email headers of the last chapter; genres/subgenres from URL path (`/nifty/{category}/{subcategory}/…`)
  - Boilerplate detection: compares first paragraphs of chapters 1 and 2 after header stripping; matching paragraphs are skipped in all chapters (`preamble_count` per chapter dict)
  - `fetch_chapter`: retrieves `<pre>` content (fallback: body); subject header stored as invisible HTML comment `<!-- Subject: … -->` at the top of chapter content; email headers stripped; hard line breaks within paragraphs joined into a single line; scene breaks (`***`, `---`, etc.) → `<hr/>`
  - Date parsing via `email.utils.parsedate` → `YYYY-MM-DD`
- `xhtml.py`: `element_to_xhtml` now handles `bs4.Comment` objects as XML comments (`<!-- … -->`); `--` in the body is sanitised to `- -` (illegal in XML comments)

## 2026-04-06 (3)
- Book detail: rating moved from clickable stars to dropdown in the Edit metadata panel
  - Stars on the detail page are now purely visual (no longer clickable)
  - New `<select id="ed-rating">` field in the edit panel, directly below Status
  - `openEdit()` populates the select with the current rating; `saveEdit()` calls `POST /library/rating/…` if the value has changed
  - `rateBook()` function removed from `book.js`
  - Overview pages unchanged

## 2026-04-06 (2)
- Library: cover upload now also supported for DB-stored books
  - `POST /library/cover/{filename}` previously returned an error for DB books (`"File not found"`) because no physical file exists
  - Fix: DB books are now detected via `is_db_filename`; cover is stored directly in `library_cover_cache` and `has_cover = TRUE` set in the `library` table — `add_cover_to_epub` is not called

## 2026-04-06 (1)
- Search: filter for unread novels / unread shorts
  - New `filter` parameter on `GET /api/search?q=…&filter=all|unread_novels|unread_shorts` (default: `all`)
  - `unread_novels`: restricts to books with no reading sessions/progress and no `Shorts` tag
  - `unread_shorts`: restricts to books with no reading sessions/progress and with a `Shorts` tag
  - UI: second toggle row (All / Unread novels / Unread shorts) below the Phrase/All words toggle
  - Filter persisted in URL (`?filter=…`) and restored on page load

## 2026-04-05 (4)
- Backup: large files (> 148 MB) now uploaded via Dropbox upload session in 100 MB chunks
  - `_dropbox_upload_bytes`: files ≤ 148 MB go via `files_upload` (unchanged); larger files via `upload_session_start` → `upload_session_append_v2` → `upload_session_finish`
  - Fixes `ApiError: UploadError('payload_too_large')`

## 2026-04-05 (3)
- Filenames: spaces now replaced with underscores when saving new files
  - `clean_segment` (common.py) and `_clean_segment` (reader.py): `\s+` → `_` instead of space
  - Series separator changed from ` - ` to `_-_` (applies to EPUB, CBR, CBZ and DB filenames)
  - Existing files are not renamed

## 2026-04-05 (2)
- Search: removed result limit; added Phrase / All words mode toggle
  - `LIMIT 30` removed — all matching chapters are returned
  - New `mode` parameter on `GET /api/search?q=…&mode=phrase|words`: `phrase` (default) requires words in order (`phraseto_tsquery`); `words` requires all words present in any order (`plainto_tsquery`)
  - Toggle in the UI (Phrase / All words) above the results; mode is included in the URL

## 2026-04-05 (1)
- Export EPUB: double chapter titles fixed — same heading-stripping logic as the reader now applied before passing content to `make_chapter_xhtml`
- Library: authors and publishers with only archived books now remain visible in the Authors and Publishers list views
  - `renderAuthorsView` and `renderPublishersView` switched from `activeBooks()` to `allBooks` — consistent with `renderAuthorDetail` and `renderPublisherDetail`

## 2026-04-04 (2)
- Reader: fixed double chapter titles for books where the heading is wrapped in a `<section>` element (pandoc-style)
  - Previous regex only stripped `<h1>`–`<h4>` at the very start of content; pandoc-converted EPUBs wrap headings as `<section …><h1>…</h1>` — heading was not at position 0 so the regex didn't fire
  - Added a second `re.sub` pass that removes the first `<h1>`–`<h4>` found directly after an opening `<section>` or `<div>` tag, preserving the wrapper element
  - Same stripping applied in the DB→EPUB export (`export_epub`) before passing content to `make_chapter_xhtml`
- Search: switched from `plainto_tsquery` to `phraseto_tsquery` in the FTS WHERE clause
  - `plainto_tsquery` ANDs all words but treats them as independent terms (any order, any distance) — multi-word queries like "4 years later" matched chapters where the words appeared far apart
  - `phraseto_tsquery` requires all words to appear in sequence; `ts_rank` and `ts_headline` still use `plainto_tsquery` for correct scoring and highlighting

## 2026-04-04 (1)
- Reader: fixed double chapter titles in DB-stored books
  - Chapter endpoint (`GET /library/chapter/{index}/{filename}`) now strips all leading `<h1>`–`<h4>` tags from stored content before prepending its own `<h2 class="chapter-title">` — books scraped before front-matter stripping was added to the scraper showed the title (and chapter heading) twice
- Library: archived books now visible in author and publisher detail views, with an indicator badge on the cover
  - `renderAuthorDetail` and `renderPublisherDetail` reverted to use `allBooks` (including archived) so archived books remain accessible from the author/publisher pages
  - New `.badge-archived` overlay (bottom-left of cover): dark circle with archive icon, consistent with existing `badge-status` and `read-pill` overlays; added to all three card-rendering code paths

## 2026-04-03 (3)
- DB chapter editor: Monaco-based editor now supports DB-stored books
  - `GET /library/editor/{filename}` handles `db/…` filenames; `is_db` flag passed to template
  - `GET /api/edit/chapter/{index}/{filename}` and `POST …`: DB branches query/update `book_chapters` directly; save calls `upsert_chapter` (updates `content_tsv` too)
  - `POST /api/edit/chapter/add/{filename}` and `DELETE …`: DB branches insert/delete with `chapter_index` shift via `UPDATE … SET chapter_index = chapter_index ± 1`
  - Title editing: header chapter-name replaced with a text input for DB books; `pendingTitles` map preserves unsaved titles across chapter switches (parallel to `pendingContent`); title-only dirty chapters correctly saved in Save All
  - `insertBreak`: scene-break image path is `/static/break.png` for DB books (vs `../Images/break.png` for EPUB)
  - Fix: `editor.focus()` called after content load so Monaco receives keyboard focus immediately
  - Fix: `header-chapter` "Loading…" text suppressed for DB books where that element is hidden
  - `book.html`: "Edit chapters" button shown for `storage_type = 'db'` books
- Search: chapter titles now included in FTS
  - `upsert_chapter` prepends title to the plain-text input for `to_tsvector`: `title + " " + stripped_html`
  - `GET /api/search`: added `OR LOWER(bc.title) LIKE LOWER('%…%')` fallback for chapters whose title matches but content doesn't
  - Startup migration `migrate_rebuild_chapter_tsv_with_title()` rebuilds existing `content_tsv` values to include titles
- Grabber: added DB/EPUB storage toggle on the Convert page
  - UI toggle above Convert button ("Save as: DB | EPUB file"); `storageMode` JS variable sent in POST body
  - `POST /convert`: reads `storage_mode` from body; stored in job as `'db'` or `'epub'`
  - `_run_scrape`: EPUB path builds chapters via `make_chapter_xhtml`, calls `make_epub`, writes file, calls `upsert_book(storage_type='file')`; DB path unchanged
  - `done` SSE event includes `storage_type`; `conversion.js` updates the download button label/action accordingly
- EPUB → DB conversion: fixed double chapter title
  - `_epub_body_inner` strips the first `<h1>`/`<h2>`/`<h3>` heading from each chapter body before storing; the editor prepends its own heading, so storing the EPUB heading too caused it to appear twice
  - Fix for `NavigableString` crash: `getattr(child, "name", None) is None` used instead of `hasattr(child, "name")` — `NavigableString` has `name = None` but no `decompose()` method
- Sidebar: Search link styling fixed
  - Stray `<li>Search</li>` moved inside the Library `<ul class="sidebar-nav">` (was outside, causing incorrect HTML structure)
  - `sidebar.css`: added `a:visited { color: var(--text-dim) }` and `a.active:visited { color: var(--accent) }` to prevent the browser's default purple visited color

## 2026-04-03 (2)
- DB-stored books (Fase 4–6): EPUB→DB conversion, DB→EPUB export, full-text search
  - **Fase 4** — EPUB-to-DB conversion: `POST /api/library/convert-to-db/{filename}` converts an existing on-disk EPUB to a DB-stored book; extracts chapters via `_epub_body_inner` (rewrites img src to imagestore URLs), migrates all child rows (book_tags, reading_progress, reading_sessions, bookmarks, library_cover_cache) to the new `db/…` filename using INSERT→UPDATE→DELETE to respect FK constraints, then deletes the EPUB file
  - **Fase 5** — DB→EPUB export: `GET /api/library/export-epub/{filename}` builds and streams an EPUB from DB content; `_rewrite_db_images_for_epub` rewrites `/library/db-images/…` URLs back to `OEBPS/Images/…` paths, deduplicating by sha256; `Content-Disposition: attachment` response
  - **Fase 6** — Full-text search: new `routers/search.py` with `GET /search` (page) and `GET /api/search?q=…` (FTS over `book_chapters.content_tsv` via `plainto_tsquery('simple', q)`, `ts_headline` for snippets, `ts_rank` for ordering, LIMIT 30, excludes archived); new `templates/search.html` with highlight (`<mark>`), "Read here" link (`?bm_ch=N&bm_scroll=0`), and "Book detail" link; Search entry added to sidebar
  - `book.html`: DB books show "Export EPUB" instead of "Download"; "Edit EPUB" and "Convert to DB" buttons only shown for `.epub` files; delete modal text differs for DB vs file books
  - `PATCH /library/book/{filename}`: DB book branch added — skips file move, recomputes synthetic `db/…` filename via `make_rel_path`, applies same FK-safe rename pattern, updates `book_chapters` and `bookmarks` in addition to standard child tables

## 2026-04-03 (1)
- DB-stored books (Fase 1–3): grabber now stores scraped books in PostgreSQL instead of EPUB files on disk
  - New `book_chapters` table: `filename FK, chapter_index, title, content TEXT, content_tsv TSVECTOR`; GIN index on `content_tsv` for future FTS
  - New `book_images` table: `sha256 PK, ext, media_type, size_bytes`; content-addressed imagestore at `library/images/{sha2}/{sha256}{ext}`
  - New `storage_type VARCHAR(10) DEFAULT 'file'` column on `library`; DB-stored books use `'db'`
  - New utilities in `common.py`: `is_db_filename`, `write_image_file`, `store_db_image`, `html_to_plain`, `upsert_chapter`, `ensure_unique_db_filename`; `make_rel_path` now handles `media_type="db"` → synthetic `db/{pub}/{auth}/...` filename
  - `upsert_book` and `list_library_json` updated to include `storage_type`
  - Grabber: `_run_scrape` stores chapters in `book_chapters`, chapter images in imagestore (absolute `/library/db-images/` URLs embedded in HTML), cover in `library_cover_cache`; no EPUB file written
  - New `GET /library/db-images/{path:path}` endpoint serves imagestore files
  - Reader: `GET /library/chapters/` and `GET /library/chapter/` have DB branches for `storage_type='db'` books (query `book_chapters` directly)
  - Reader page (`/library/read/`), book detail page, mark-read, and rating endpoints all handle DB filenames (no file existence required)
  - Cover endpoints (`/library/cover/`, `/library/cover-cached/`) serve DB books from `library_cover_cache`

## 2026-04-02 (1)
- Added Restore functionality to the Backup page
  - New `GET /api/backup/snapshots` endpoint: lists available Dropbox snapshots (name + date parsed from filename, no downloads needed)
  - New `GET /api/backup/snapshots/{snapshot_name}/files` endpoint: loads a snapshot from Dropbox and returns all files with path, size, sha256, and whether the file currently exists locally
  - New `POST /api/backup/restore` endpoint: downloads file objects from Dropbox, writes to disk, and re-indexes via `scan_media` + `upsert_book`; returns per-file result with errors
  - New "Restore" card on the backup page: snapshot dropdown (auto-loaded on page open), file list with filter/search, per-file "Restore" button, multi-select + "Restore selected", on-disk indicator, inline status feedback
  - After restore, the file list refreshes to reflect updated on-disk state

## 2026-03-29 (10)
- Duplicates: fixed `updateCounts` crashing with a TypeError (`g.books.length` → `g.length`); the crash prevented `renderGrid` from running, so the duplicates view never rendered and the counter was stale

## 2026-03-29 (9)
- Duplicates: volume-aware duplicate detection — a book is only a duplicate when title + author + volume all match
  - `_duplicateGroups` in `library.js` (Duplicates view): key now includes `series_index` when > 0, so different volumes of the same series are no longer grouped as duplicates
  - `preload` in `grabber.py` (Grabber): when the scraper returns a `series_index_hint`, the DB lookup only flags a match when title + author + volume all match; falls back to title + author when no volume is known — same logic as the bulk importer

## 2026-03-29 (8)
- Shared code: eliminated all remaining duplication across templates and JS files
  - CSS custom properties: extracted single `:root { }` block into `static/theme.css`; removed duplicate inline `:root` from all 15 templates and from `library.css`, `book.css` — `editor.css` keeps editor-specific vars (`--danger`, `--header-h`, `--panel-w`), `reader.html` keeps page-specific vars (`--header-h`, `--footer-h`, `--content-w`), `backup.html` keeps (`--ok`, `--warn`, `--err`)
  - Cover helpers: moved `strHash`, `COVER_PALETTES`, `makePlaceholderCover`, `wrapText`, `truncate` from `library.js` and `book.js` into `books.js`; removed from `home.html` and `following.html`
  - HTML escape: `esc()` added to `books.js`; removed from `library.js`, `editor.js`, and all 8 templates that defined it inline
  - SSE/EventSource: extracted shared `connectConversionStream(job_id)` and `addLog()` into new `static/conversion.js`; both `index.html` and `grabber.html` now call the shared function (removed ~70 duplicate lines)

## 2026-03-29 (7)
- Search: extracted shared book helpers and search logic into `static/books.js`
  - `_filenameBase`, `bookTitle`, `bookAuthor`, `tagValuesByType`, `bookGenres`, `bookSubgenres`, `bookPlainTags`, `filterBooks`, `setupSearchInput` moved from `library.js` and `home.html` to a single shared file
  - Both pages now use identical search behaviour: Enter to search, × to clear
  - Both pages now search across title, author, publisher, genre, sub-genre, and tag

## 2026-03-29 (6)
- Search: changed from search-as-you-type (250 ms debounce) to Enter-to-search on both Library and Home — prevents iPad keyboard from locking up on large collections

## 2026-03-29 (5)
- Accent colour updated to match logo orange (`#ffa20e`); secondary accent updated to `#ffb840`
  - Applied across all CSS files, templates, and inline styles

## 2026-03-29 (4)
- Branding: added logo, favicon, and Apple touch icon to all pages
  - Static assets: `logo.png` (sidebar), `favicon.ico` (16×16), `favicon-32.png` (32×32), `favicon-256.png` (256×256), `apple-touch-icon.png` (180×180 with `#0f0e0c` background)
  - Favicon `<link>` tags added to all 15 templates
  - Sidebar: image logo (`logo.png`) placed next to the existing "No**vela**" wordmark using flexbox
  - `apple-touch-icon.png` uses dark `#0f0e0c` background — renders as a native-looking iOS home screen icon

## 2026-03-29 (3)
- Dockerfile: replaced `unrar-free` with proprietary `unrar` (RARLAB v6.2.6) from Debian non-free — fixes "Failed to read enough data" errors on RAR archives using newer compression methods

## 2026-03-29 (2)
- CBR reader: detect archive format via magic bytes instead of file extension — `.cbr` files that are actually ZIP or 7-zip archives now open correctly; added `py7zr` dependency for 7-zip support

## 2026-03-29 (1)
- Bulk Import: duplicate check now volume-aware — books with the same title+author but a different volume number (e.g. recoloured reprints) are no longer flagged as duplicates; `volume` is included in the API call and matched against `series_index` in the DB

## 2026-03-28 (11)
- Bulk Import: duplicate detection against existing library
  - New `POST /api/bulk-check-duplicates` endpoint: case-insensitive title+author+volume match, single SQL query with OR conditions for all pairs
  - Duplicate rows highlighted in red in the preview table; a skip checkbox appears per duplicate row
  - Stats bar shows "X duplicates · Skip all · Import all" action buttons
  - Duplicate rows are skipped by default; user can toggle individual rows or use bulk actions
  - `startImport()` filters out skipped rows before batching; skipped files appear in the result summary as "Duplicate – skipped"
  - If all rows are skipped (nothing to import), result shown immediately without sending any request

## 2026-03-28 (10)
- After "Remove from New" succeeds, the library is now reloaded from the server (`loadLibrary()`) so the New list updates immediately without a manual refresh

## 2026-03-28 (9)
- Bulk delete is now batched: new `POST /library/bulk-delete` endpoint accepts a JSON list of filenames, deletes files and removes DB rows in one query per batch; JS sends 20 files per batch with a progress bar in the confirmation dialog

## 2026-03-28 (8)
- Disk usage warning in sidebar: `GET /api/disk` returns partition usage for the library directory; sidebar polls every 60 s and shows a warning bar (amber ≥ 85% or < 2 GB free, red ≥ 95% or < 500 MB free) above the backup status bar

## 2026-03-28 (7)
- Fixed `mark-reviewed` JS: UI (allBooks update + renderGrid) now only runs after confirmed server success; `catch` no longer fires a false error after a successful response where renderGrid throws

## 2026-03-28 (6)
- Bulk Import: rewrote pattern editor from token-pills to free-text `%placeholder%` syntax
  - Pattern input: free-text field; placeholders: `%series%` `%volume%` `%title%` `%year%` `%month%` `%day%` `%author%` `%publisher%` `%ignore%`
  - Colored chip row: click to insert at cursor position; drag onto input also supported
  - Colored live preview below the pattern input; regex-based parser replaces delimiter+token logic
  - Shared metadata now wins over filename-parsed values (previously filename won)
  - All UI text translated from Dutch to English

## 2026-03-28 (5)
- Status badge (top-right cover) and want-to-read star (top-left cover) now use dark fill `rgba(15,14,12,0.82)` + `box-shadow: 0 0 0 2px #0f0e0c` ring — always readable regardless of cover colour

## 2026-03-28 (4)
- Added `Temporary Hold` status; renamed `Hiatus` → `Long-Term Hold`
  - Startup migration `migrate_rename_hiatus()` converts existing DB records automatically
  - Status colours: Complete=green, Ongoing=blue, Temporary Hold=amber, Long-Term Hold=orange
  - `statusBadgeHtml()` helper in `library.js` replaces three identical inline badge blocks
  - Grabber: `Temporary-Hold` (gayauthors.org) now maps to `Temporary Hold`; `Long-Term Hold` passes through unchanged
  - Status dropdowns updated in Book Detail and Bulk Import

## 2026-03-28 (3)
- Added Bulk Import page (`/bulk-import`) for batch importing CBR/CBZ/EPUB/PDF files with filename-based metadata parsing
  - Configurable token pattern: Serie, Volume, Titel, Jaar, Auteur, Uitgever, Negeer (in any order)
  - Configurable delimiter (default ` - `); year token right-anchored when last so title absorbs overflow segments
  - Live test-parse box: type any filename and see how it parses before selecting files
  - Shared metadata card: author, publisher, status, genres, tags applied to all files (overridden by pattern tokens or manual edits)
  - Preview table: all parsed fields editable via contenteditable cells; warning indicator for rows with fewer segments than tokens; sorted by filename
  - Batch upload in groups of 5 with progress bar (N / total files processed)
  - Result summary with list of skipped files and reasons
  - New `routers/bulk_import.py`: `GET /bulk-import`, `POST /library/bulk-import`
  - Sidebar link under Tools between Book Builder and Credentials

This file tracks changes on the `develop` line.
`changelog.md` can later be used for release summaries.

## 2026-03-28 (2)
- Performance: library page now loads instantly for large collections (1000+ books)
  - `IntersectionObserver` defers both cover image loading and placeholder canvas drawing until cards enter the viewport — eliminates hundreds of upfront canvas ops that blocked the initial render
  - `ETag` caching on `/library/list`: server returns `304 Not Modified` when nothing changed, client skips JSON parse and re-download
  - Single DOM pass in `renderBooksGrid`, `renderDuplicatesView`, `renderSeriesDetail`: canvas and img set up via `card.querySelector` immediately after `innerHTML`, removing a second iteration with `document.getElementById` per card
  - `book_tags` joined via `json_agg` in the main `list_library_json()` query, eliminating a separate `SELECT * FROM book_tags` query and Python merge loop
  - `loadLibrary` now shows an error message instead of staying stuck on "Loading…" when the fetch or render fails

## 2026-03-28 (1)
- Added Following page (`/following`): track external author URLs outside Library and Tools
  - New `authors` table: `name` (unique), `url`, `created_at`, `updated_at`
  - New `routers/following.py`: `GET /following` page, `GET /api/following` (all authors + URL + book count + last added), `POST /api/following/{name}` (set/clear URL)
  - Sidebar: new Following section between Library and Tools; counter shows number of followed authors
  - Following page: two tabs — Following (authors with URL) and All Authors; inline URL editing with Enter/Escape keyboard support; Visit button opens external URL in a new tab; author name links to library author view
- Added Incomplete view to Library (`#incomplete`): shows all non-archived books where `publication_status ≠ Complete`; sidebar counter included; entry placed after New in the Library section

## 2026-03-27 (1)
- Convert page: duplicate warning shown after loading metadata when a book with the same title+author already exists in the library; warning includes a link to the existing book; user can still proceed with conversion
- Library: added Duplicates section to sidebar (between Rated and Statistics); counter shows total number of books that are part of a duplicate group (same title+author, case-insensitive); Duplicates view groups books by title+author with a subheading per group
- Fixed Duplicates view not loading covers: card renderer now uses the same canvas + two-pass img/`makePlaceholderCover` pattern as `renderBooksGrid`

## 2026-03-26 (2)
- Fixed Book Builder page showing white background: `library.css` added to `builder.html` to load `:root` CSS variables and dark `body` background; all CSS variable references in `builder.css` aligned with library theme names (`--text`, `--surface`, `--surface2`, `--text-dim`, `--border`, `--accent`, `--sidebar`)

## 2026-03-26 (1)
- Added Book Builder: create EPUB books manually from scratch via a WYSIWYG editor
  - New `builder_drafts` table (`id UUID`, `title`, `author`, `publisher`, `source_url`, `chapters JSONB`)
  - `build_epub()` in `epub.py`: builds a standards-compliant EPUB 2.0 ZIP from title/author/publisher/chapters; embeds inline CSS and `break.png` if present
  - `normalize_wysiwyg_html()` in `xhtml.py`: converts contenteditable HTML to EPUB-safe XHTML; handles scene-breaks, `<blockquote class="author-note">`, inline formatting, and `<hr>` → break image
  - `routers/builder.py`: draft CRUD, chapter CRUD (`GET/POST/PUT/DELETE`), normalize preview endpoint, publish endpoint (normalizes all chapters → builds EPUB → writes to `library/epub/` → upserts to DB → deletes draft → redirects to book detail)
  - `templates/builder.html` + `static/builder.js` + `static/builder.css`: index page (new draft form + draft list) and editor (chapter panel, contenteditable pane, toolbar with bold/italic/underline/blockquote/author-note/scene-break/normalize, autosave every 30 s, Ctrl+S, publish button)
  - Book Builder link added to sidebar Tools section (between Convert and Credentials)
  - `.author-note` blockquote style added to `static/epub-style.css`

## 2026-03-25 (20)
- Added Rated section to library sidebar: shows all non-archived books with `rating > 0`, sorted by rating descending then title alphabetically; badge displays total count; navigable via `#rated` URL hash

## 2026-03-25 (19)
- Fixed series index 0 not displaying in series slot view, grid cards, list volume column, and book detail:
  - Series slot labels now show `#0` for all slots when the series has at least one positively-indexed sibling (consistent label height across all cards)
  - Grid card and list volume column use the same cross-book heuristic (`indexedSeriesSet`) to show `[0]` where appropriate
  - Book detail page queries whether any sibling in the same series has `series_index > 0` (`series_is_indexed`) and shows `[0]` accordingly

## 2026-03-25 (18)
- Added autocomplete for Author, Publisher, and Series in the book edit panel: typing shows a filtered dropdown of existing values from the library; selecting a value fills the field (same dropdown styling as genres/tags)
- Status field now defaults to "Complete" when opening the edit panel for a book that has no status set yet

## 2026-03-25 (17)
- Fixed CBR reader showing only first page: `cbr_page_count` was missing from the `cbr` import in `reader.py`; `/api/cbr/info/` was returning an error, causing `page_count` to fall back to 1 and the Next button to remain disabled

## 2026-03-25 (16)
- Fixed CBR/CBZ reader stuck on loading: added `/api/cbr/info/{filename}` endpoint returning `{page_count}`, and added a CBR/CBZ branch in `reader.html` that mirrors the PDF paged reader (page images served via `/library/cbr/{filename}/{page}`)

## 2026-03-25 (15)
- Added series volume 0 and letter suffix support (e.g. "21a", "21b"):
  - New `series_suffix VARCHAR(10)` column added to `library` table via `migrate_series_suffix`
  - `series_index` lower bound changed from 1 to 0 throughout (index 0 = special/prequel edition)
  - Volume field in the book editor changed from `type="number"` to `type="text"` — accepts "0", "1", "21a", etc.
  - Server parses the combined volume string into `series_index` (INTEGER) + `series_suffix` (VARCHAR) via `parse_volume_str`
  - File naming includes suffix: `021a - Title.epub`
  - `novela:series_suffix` meta tag written to/read from EPUB OPF for persistence across rescans
  - Series detail view: books sorted by `(series_index, series_suffix)`; slot labels show "21a"; index 0 shown as slot when any sibling has index > 0
  - Grid cards, list view Volume column, author/publisher sort all include suffix

## 2026-03-25 (14)
- Added multi-select and bulk delete to `All books` List view:
  - Checkbox column added to the list table (List mode only; Grid mode unchanged)
  - Select all / Clear all / Clear selection controls in the controls bar
  - `Shift+click` range selection on checkboxes
  - `Delete selected` button (red, disabled when nothing is selected) triggers a confirmation dialog; confirmed deletions remove files from disk and database, then reload the library

## 2026-03-25 (13)
- Fixed EPUB cover replacement reverting to original after upload:
  - `GET /library/cover/{filename}` for EPUBs now checks the DB cover cache first (populated on upload) before falling back to extracting from the EPUB file; the cache is also warmed on cache-miss so subsequent requests are fast.
  - `add_cover_to_epub` fully rewritten: locates the OPF via `META-INF/container.xml`, finds the existing cover image in the OPF manifest, removes it from the ZIP, and writes the new cover in the same directory — works for any EPUB directory structure (`OEBPS/`, `EPUB/`, etc.).

## 2026-03-25 (12)
- Fixed chapter images not loading in EPUBs that use a non-standard root directory (e.g. `EPUB/` instead of `OEBPS/`): image paths are now passed as full ZIP paths instead of stripping the root segment, and the image endpoint no longer hardcodes an `OEBPS/` prefix. Case-insensitive fallback retained for EPUBs with mismatched image folder casing.

## 2026-03-25 (11)
- Fixed reader progress bar jumping to 100% immediately when navigating to the second chapter in short EPUBs (2 chapters): changed progress formula denominator from `total - 1` to `total` so 100% is only reached after fully scrolling through the last chapter. Same fix applied to PDF page progress.

## 2026-03-25 (10)
- Added bookmark feature:
  - `bookmarks` table in DB (`migrate_create_bookmarks`): `filename`, `chapter_index`, `scroll_frac`, `chapter_title`, `note`, `created_at`
  - New API endpoints: `GET/POST /library/bookmarks/{filename}`, `PATCH/DELETE /library/bookmarks/{id}`, `GET /api/bookmarks`
  - Bookmark button in reader header (orange, bookmark icon): opens modal with optional note field
  - Bookmark modal closes on Escape, backdrop click, or Cancel
  - Reader supports `?bm_ch=N&bm_scroll=F` URL params to jump directly to bookmarked position (overrides saved progress)
  - Library sidebar: Bookmarks section in Library nav with live count badge
  - Library `#bookmarks` view: card list showing cover, book title, author, chapter, note, date, Go-to and Delete buttons

## 2026-03-22
- Added blueprint/technical documentation structure in `docs/`.
- Completed router split and bootstrap structure (`main.py`, routers, migrations, DB pool).
- Expanded media support to EPUB/PDF/CBR/CBZ in import and scan flows.
- Expanded Home UI with:
  - import dropzone for EPUB/PDF/CBR/CBZ
  - search functionality
  - alignment matching Library (search top-right, dropzone below)
- Updated Library import texts and drag/drop filtering for multi-format support.
- Expanded Library `New` view with:
  - `Grid`/`List` toggle
  - column visibility filter in `List`
  - multi-select + bulk `Remove from New`
  - selection only in `List` mode
  - `Shift+click` range selection on checkboxes
- Added route: `POST /library/new/mark-reviewed` (bulk set `needs_review=false`).
- Improved library performance:
  - `/api/library` fast-path (no full rescan on every page load)
  - optional `rescan=true` / `include_file_info=true`
  - SQL optimizations in `list_library_json()`
  - additional DB indexes for scale
- Restored `/api/home` full dataset output:
  - `continue_reading`
  - `shorts_unread`
  - `novels_unread`
  - `shorts_read`
  - `novels_read`
- Explicitly filtered series books out of Home sections.
- Corrected Home read ordering: `shorts_read` and `novels_read` now show oldest first (`ORDER BY MAX(read_at) ASC`).
- Restored Statistics page by returning the full `/api/stats` payload required for charts, favorites, top books, and reading history.
- Improved backup implementation:
  - Dropbox token stored encrypted in DB
  - Dropbox backup root configurable via web UI and stored encrypted in DB
  - versioned snapshots + object-store deduplication in Dropbox (`library_snapshots` / `library_objects`)
  - configurable snapshot retention (`snapshots to keep`) via backup settings
  - object pruning based on retained snapshots
  - scheduled backup (enable + interval in hours)
  - backup runs as a background process, so page navigation does not block execution
  - stale running state recovery after restart/crash (old running logs marked interrupted/error)
  - dry-run support in the new flow
- Updated Docker image with `postgresql-client` for `pg_dump`.
- Multiple test builds pushed to `gitea.oskamp.info/ivooskamp/novela:dev`.

## 2026-03-25 (9)
- Fixed backup progress not visible immediately after clicking Run Live Backup: sidebar `loadBackupProgress()` is now called directly from the backup page after a successful start
- Fixed backup status timestamp showing wrong relative time (e.g. "1h ago" for a recent backup): server UTC timestamps without timezone suffix are now correctly interpreted as UTC in the browser
- Fixed backup status bar using browser-default purple link color: now uses `var(--accent)` orange consistent with the rest of the UI
- Added password manager ignore attributes to App Key and App Secret fields: `data-1p-ignore` (1Password), `data-lpignore="true"` (LastPass), `data-form-type="other"` (1Password v8)

## 2026-03-25 (8)
- Added Dropbox OAuth2 refresh token support (no redirect URI needed):
  - New endpoints: `POST /api/backup/oauth/prepare` (generates auth URL) and `POST /api/backup/oauth/exchange` (exchanges code for refresh token)
  - `_dbx()` now prefers refresh token mode (app key + secret + refresh token) with automatic renewal; falls back to legacy access token for backwards compatibility
  - App key and secret stored encrypted in `credentials` table (`dropbox_app_key`, `dropbox_app_secret`)
  - `DELETE /api/backup/credentials` now also cleans up `dropbox_app_key` and `dropbox_app_secret`
  - Backup page updated with two-step OAuth flow UI (app key/secret → auth URL → paste code)
  - Settings status shows `• refresh token` or `• legacy token` indicator

## 2026-03-25 (7)
- Added backup progress tracking: `GET /api/backup/progress` returns `{running, done, total, phase}`; sidebar shows live file count (e.g. `312 / 652 · uploading`) polling every 3 s while running

## 2026-03-25 (6)
- Fixed `GET /api/backup/health` missing `schedule_enabled` and `schedule_interval_hours` in response (values were read but not returned; Health card showed "disabled" even when schedule was on)
- Added backup status indicator in sidebar above Rescan library button: coloured dot + status text (OK/running/failed/never), links to `/backup`, auto-refreshes every 8 s while running, shows time-ago for last successful backup

## 2026-03-25 (5)
- One-time path migration: all 571 existing library files moved to correct format-prefixed structure (`epub/`, `pdf/`, `comics/`) and all DB references updated (`migrate_paths.py --execute`)
- Empty old publisher root directories pruned after migration
- Recovery script (`recover_decock049.py`) written; confirmed file was added after last Dropbox backup so not recoverable — to be re-imported manually

## 2026-03-25 (4)
- Added {publisher} directory to PDF and CBR/CBZ paths: `pdf/{publisher}/{author}/{title}.pdf`, `comics/{publisher}/{author}/{title}.cbr/cbz` — consistent with EPUB structure

## 2026-03-25 (3)
- Fixed CBZ extension in import: `common.make_rel_path` always generated `.cbr` for CBZ files; now accepts `ext` parameter; `library.py` passes actual suffix so CBZ files land at `comics/{author}/{title}.cbz`
- Added missing `GET /download/{filename}` endpoint (referenced in book.html but was 404)
- TECHNICAL.md fully rewritten: added File Storage Paths section, complete endpoint lists for all routers including settings.py, corrected path documentation

## 2026-03-25 (2)
- Fixed PDF metadata editing (PATCH /library/book):
  - `_sync_epub_metadata` is now only called for `.epub` files; PDFs update DB only
  - `_make_rel_path` now includes the format prefix matching import: EPUB → `epub/{publisher}/{author}/…`, PDF → `pdf/{author}/{title}.pdf`, CBR/CBZ → `comics/{author}/{title}{ext}`; previously files were moved outside their format directory on metadata save
- Fixed PDF reader (infinite loading screen):
  - `reader_page` now passes `format` (epub/pdf/cbr/cbz) to `reader.html`
  - Added `GET /api/pdf/info/{filename}` endpoint returning `{"page_count": N}`
  - `reader.html` branches on `FORMAT`: PDFs render page images from `/library/pdf/{filename}?page=N`, EPUB flow unchanged
  - PDF progress tracked per page; keyboard and button navigation work identically to EPUB
- `Edit EPUB` button in Book Detail hidden for non-EPUB files

## 2026-03-23
- Added `All books` Grid/List toggle in Library:
  - same columns as `New` view (Publisher, Author, Series, Volume, Title, Has cover, Updated, Genres, Sub-genres, Tags, Status)
  - column visibility filter in `List` mode
  - no selection checkboxes or bulk actions
  - view mode and column visibility persisted separately in `localStorage` (`novela.all.viewMode`, `novela.all.visibleColumns`)
- Added 1–5 star rating for books:
  - stored in the database (`rating SMALLINT DEFAULT 0`)
  - written to EPUB OPF as `<meta name="novela:rating" content="N"/>` on rating change
  - written to CBZ `ComicInfo.xml` as `<NovelaRating>N</NovelaRating>` on rating change
  - CBR and PDF are DB-only (file format constraints)
  - rating is recovered from file metadata during rescan/DB rebuild (`upsert_book` preserves file rating over default 0)
  - stars displayed under the cover (outside `.book-info`) in all grid views (Library, Home)
  - stars in grid cards are display-only (no click) to prevent accidental taps while scrolling
  - stars in Book Detail are interactive (larger, 1.1rem), clicking same star removes rating
  - star color: amber (`#c8a03a`) for filled, `rgba(200, 160, 58, 0.25)` for unfilled — consistent across Library, Home, and Book Detail
- Added text colour setting to reader hamburger menu:
  - 5 warm-tone presets from bright (`#e8e2d9`) to dim (`#938d86`)
  - active preset shown with accent-coloured ring
  - choice persisted in `localStorage` (`reader-text-colour`) and restored on next open
- Increased spacing between hamburger button and back link in reader header (`margin-left: 1rem`) to prevent accidental taps
- Removed `Cover Missing` auto-tag:
  - tag is no longer added on import, rescan, or grabber download
  - `ensure_cover_missing_tag()` removed from `common.py`, `library.py`, and `grabber.py`
  - startup migration removes all existing `Cover Missing` tags from the database
- Fixed Tags/Genres/Sub-Genres not saving in book edit panel on desktop:
  - `,` (comma) now acts as a confirmation key alongside Enter in `PillInput`
  - `flush()` added to `PillInput`: any text still in the input field is auto-confirmed when Save is clicked
- Fixed tag/genre search and tag-pill navigation being broken:
  - `renderGenreView` was filtering on `b.genres` (non-existent field); now uses `bookGenres()`, `bookSubgenres()`, `bookPlainTags()`
  - `renderSearchResults` had the same bug; search now covers title, author, genres, sub-genres, and tags
- Made reading progress monotonic across devices:
  - `POST /library/progress/{filename}` now rejects any save whose `(chapter_index, scrollFrac)` is not strictly ahead of the stored position (returns `{"ok": true, "skipped": true}`)
  - prevents device A from overwriting further progress saved by device B when switching between devices without closing the book
  - also prevents bookmark-based backward navigation (e.g. jumping back to correct an earlier chapter) from clobbering the furthest-read position
  - progress reset remains via the explicit Read/Unread actions, which clear the `reading_progress` row