GetAllProgressData (SSR handler) was missing LastUpdated, DeviceIcon,
DeviceName, DeviceType, and EpubCFI fields that the template expects.
All showed blank. Now matches the API handler's field population.
Three bugs fixed:
1. Schema seeded base_url with fake placeholder 'bookhoard.example.com'.
Removed seed; startup now seeds from BASE_URL env var only if DB row
is empty (admin changes persist across restarts). One-time UPDATE
clears the placeholder in existing installs.
2. config.GetBaseURL() had a broken type assertion (local SystemConfigRow
vs database.SystemConfig) that always failed, returning . Admin panel
showed env var fallback instead of actual DB value. Fixed with a
function-type getter that properly wraps the DB query.
3. OPDS handler read base_url only from DB with no fallback. When DB had
the placeholder, all feed links pointed to an unreachable domain,
breaking KOReader search/download. Added deriveBaseURL() helper that
falls back to the request Host/scheme when DB value is empty.
Setup gate improvements:
- isSetupComplete now requires both admin user AND non-empty base_url
- Setup middleware no longer exempts all /api/ routes; only allows
/api/auth/register, /api/auth/login, /api/system/config before setup
is complete. All other API routes get 503.
- Cache invalidated when base_url is saved via admin settings
Dev workflow:
- New bruno/NewDevDBSetup/SetBaseUrl.yml for dev DB setup
- NewDB.sh runs SetBaseUrl between RegisterUser and CreateEbookLibrary
Three root causes, all fixed:
1. Icon buttons were created with setAttribute('onclick', ...) in
populateIconGrid, but selectIcon is module-scoped (not on window),
so clicking threw ReferenceError. Switch to addEventListener with
a closure. Icon search/focus used plain oninput/onfocus attributes
with the same problem — convert to Alpine @input/@focus.
2. selectColor's highlight selector queried [onclick="selectColor(...)\]
System collections (Not Started, Continue Reading, etc.) compute
their contents dynamically from reading_progress, so manual
add/remove has no effect. Hide the search bar, Remove Selected
button, Add Books button, per-card checkboxes, and Remove buttons
when the collection is system, making the page visually read-only.
- Add IsSystem bool to CollectionData, populated from the database
is_system_collection flag.
- Add data-is-system to #collection-data so the JS renderer can
also conditionally omit controls on library switch.
- Wrap toolbar controls, book picker modal, card checkboxes, and
remove buttons in if !collection.IsSystem in the template.
Three fixes for the collection detail page:
1. Picker modal never showed because the outer overlay div had
style="display:none" with no x-show binding — add x-show bound
to $store.bookPicker.isOpen plus a backdrop and click-to-close.
2. Replace native confirm() with an in-page Alpine modal for UI
continuity. Add confirm dialog state (showConfirm, confirmMessage,
pendingAction) and methods (requestRemoveBook, requestBulkRemove,
executeConfirmed, closeConfirm) to the collections component.
The actual API calls (doRemoveBook/doBulkRemove) are triggered only
when the user confirms.
3. Rename removeBook → requestRemoveBook and bulkRemove →
requestBulkRemove in template + JS renderer so the dialog opens
instead of navigating or failing silently.
The /collections/:id page had several broken features because three
referenced functions (removeBook, toggleBookForRemoval,
filterCollectionBooks) were never defined, and every book card was
wrapped in <a href="/media/..."> so clicking the checkbox or remove
button navigated to the book detail page instead.
Card restructure:
- Remove the <a> wrapper; title and cover are now individual links.
- Checkbox sits in a <label> with expanded click area (p-2 -m-2).
- Checkbox uses Alpine :checked/@change bound to a reactive
selectedBooks array on the collections component.
Remove (single + bulk):
- Add removeBook(id) and bulkRemove() methods with confirm() dialogs.
- Wire the "Remove Selected" button with :disabled binding and @click.
- Selected-count badge is now Alpine-reactive (x-show/x-text).
Search within collection:
- Add filterCollectionBooks() that filters cards client-side by
title/author via data-* attributes and @input.
Book picker ("Add Books"):
- Point the HTMX search inputs at the existing /api/media-items/search
endpoint instead of the non-existent /api/media-items/filtered.
- Add hx-trigger="loadBooks" + hx-get to the grid so loadBooks()
actually fires an initial request when the picker opens.
- Merge the hidden limit/offset inputs into the #book-picker-filters
div so hx-include picks them up (was a separate <form id=filter-form>
that nobody referenced).
- Add show_checkbox mode to handleSearchHTML: when present, render a
new BookPickerGrid template with clickable, selectable cards instead
of the reader BookCard.
- Fix bookPicker submit() to location.reload() instead of a non-existent
reloadCollection HTMX event, and clearFilters() to target text inputs.
The reader's back button was hardcoded to the book detail page
(/media/{id}), so even when you launched the reader straight from the
dashboard the back button ignored that and sent you to the detail view.
Mirror the existing book-detail.ts referrer pattern: capture
document.referrer into sessionStorage on load (excluding other reader
pages and the reader's own URL), then override the back link's click to
navigate there. Falls back to the link's original href (/media/{id}) when
no valid referrer exists (direct URL access).
The dashboard re-renders its sections client-side (library switch,
refresh, saving settings) via renderBookCard in dashboard.ts, which was
still the old markup with no .book-card-action overlay. So the read
button appeared on the server-rendered cards but vanished as soon as the
dashboard re-rendered, while the bookshelf (always templ-rendered) kept
working.
- Rewrite renderBookCard to match the templ BookCard: detail link plus
the play/read action overlay, routing to the reader or the detail page
when the book has an active conflict.
- Add has_conflict to the BookInfo TS type and stamp it in the dashboard
sections API (GetSections) so client-rendered cards can route correctly.
- Add pointer-events-none / group-hover:pointer-events-auto to the
client-rendered carousel nav buttons so they no longer swallow hover
over edge cards, matching the templ fix.
The settings icon is a circle with radiating spokes, which reads as a
sun/light-mode toggle rather than a dashboard control. Swap it for the
grid icon, the standard dashboard-layout affordance.
The account and appearance menus sat in one container with no separation,
so expanding one made the other hard to find. Each collapsible menu now
lives in its own subtly lifted, bordered panel (.sidebar-panel) so the
expanded items stay visually contained and the menus never blend together.
Applied to the account, appearance, and sign-in menus.
The play button on book cards now opens the reader directly, instead of
always going to the detail page. Cards with an active progress sync
conflict route the play button to the detail page (which hosts the
conflict dialogue and resolves before writing progress), so the user is
never silently dropped into the reader with an unresolved conflict.
Backend:
- Add HasConflict to BookInfo and stamp it via ListSyncConflictsByUser
(MarkActiveConflicts / MarkActiveConflictsSections) on the dashboard,
bookshelf, series, tag, and search result card builders.
- Each page issues a single conflict query regardless of card count.
BookCard:
- Restructure into a detail link (cover + meta) with the play action as a
sibling overlay using a pointer-events split: the container passes
clicks through to detail while only the circular button routes to the
reader. No nested anchors.
- On touch devices (hover: none) the play button stays visible.
Fix: carousel nav buttons had opacity-0 without pointer-events-none, so
they swallowed hover/clicks over book cards on the dashboard. They are
now click-through until the carousel is hovered.
On phones the footer's progress cell rendered the full chapter title (e.g. 'Long Chapter Name · 5 / 12') in a div with no max-width or nowrap, so the text wrapped to multiple lines and ballooned the bottom bar. Combined with viewport offsets that were computed from assumed pixel heights with ~0px margin, the book text slipped underneath the bars.
Chapter label in footer: split #progress-display into two spans (progressLabel hidden on mobile via 'hidden sm:inline', progressMain always shown) and cap it with 'truncate whitespace-nowrap max-w-[5rem] sm:max-w-none' so it can never wrap or grow the bar. Phones now show just '5 / 12'; larger screens keep 'Chapter · 5 / 12'.
Viewport offset: replace the fragile hardcoded calc() with runtime measurement. Gave the chrome bars ids (reader-topbar/reader-bottombar) and added updateViewportInsets(), which sets #reader-viewport top/bottom from each bar's real offsetHeight (which already includes env(safe-area-inset-*) padding) plus a 6px margin. It runs on init and refreshes on resize, orientationchange, and via a ResizeObserver, so the content area tracks the actual chrome height on any DPI, notch, home-indicator, or zoom level instead of guessing.
Refactored formatProgress into formatProgressParts (returns {label, main}; only chapter mode sets a label) with a setProgress() helper wiring progressLabel/progressMain/progressText across the relocate, cycleProgressMode, and applyProgressMode call sites. Rebuilt reader_templ.go and style.css.
On phones the footer's progress cell rendered the full chapter title (e.g. 'Long Chapter Name · 5 / 12') in a div with no max-width or nowrap, so the text wrapped to multiple lines and ballooned the bottom bar. Combined with viewport offsets that were computed from assumed pixel heights with ~0px margin, the book text slipped underneath the bars.
Chapter label in footer: split #progress-display into two spans (progressLabel hidden on mobile via 'hidden sm:inline', progressMain always shown) and cap it with 'truncate whitespace-nowrap max-w-[5rem] sm:max-w-none' so it can never wrap or grow the bar. Phones now show just '5 / 12'; larger screens keep 'Chapter · 5 / 12'.
Viewport offset: replace the fragile hardcoded calc() with runtime measurement. Gave the chrome bars ids (reader-topbar/reader-bottombar) and added updateViewportInsets(), which sets #reader-viewport top/bottom from each bar's real offsetHeight (which already includes env(safe-area-inset-*) padding) plus a 6px margin. It runs on init and refreshes on resize, orientationchange, and via a ResizeObserver, so the content area tracks the actual chrome height on any DPI, notch, home-indicator, or zoom level instead of guessing.
Refactored formatProgress into formatProgressParts (returns {label, main}; only chapter mode sets a label) with a setProgress() helper wiring progressLabel/progressMain/progressText across the relocate, cycleProgressMode, and applyProgressMode call sites. Rebuilt reader_templ.go and style.css.
- Dashboard: page heading, bold section headers with accent icon tiles,
carousel chevrons as SVG icons with theme-aware gradients (wood-paneling
gradient classes preserved), and a cinematic BookCard (hover overlay with
a quick-action button). BookCard is now fluid so it fills both the
carousel slot and the bookshelf grid.
- Bookshelf: the filter wall becomes a search toolbar + a slide-in filter
drawer (filtersOpen state added to the bookshelf Alpine component). Every
filter input, the tristate cover toggle, tag autocomplete, save/load/clear
actions, HTMX search/sort, and pagination are preserved.
- Book detail: blurred cover backdrop hero, rounded-2xl cover, bold
typography, .btn action bar, progress/metadata cards, chip-style external
links. All interactive rating, modals, and data-attrs preserved.
- Auth/landing: brand-gradient hero for index/login/register, icon feature
cards, data-driven theme select (ThemeOptions). All ids (#theme-select,
#result, #auth-result, password-requirement ids) and Alpine init preserved.
Replace the top-nav with a fixed left sidebar + slim content topbar
(the Komga/Audiobookshelf layout), the signature change versus the
conservative tighten-ui branch.
- App shell: .app-sidebar (off-canvas on mobile via Alpine mobileMenuOpen,
pinned at 16rem on lg), .app-topbar (fixed, blurred, 4rem), and
.app-subbar for in-page sticky bars (parks under the topbar). Content is
auto-offset via body:has(.app-sidebar) so reader.templ/error.templ (which
have no sidebar) are untouched.
- Header rebuilt as the sidebar: logo + vertical nav (activeClass), an
inline Appearance picker driven by ThemeOptions/WoodOptions, and an
inline account menu / sign-in (preserving the inline htmx login). The
search lives in the topbar so its dropdown still anchors correctly.
- Bolder primitives: .card -> rounded-2xl, cinematic .book-card-cover
hover overlay with a quick-action affordance, .brand-gradient hero
surface, .hero-backdrop (blurred cover) and .stat-card utilities.
Add semantic design tokens and base primitives to replace the verbose
inline var() styling that made the UI feel dated.
- Rename colliding Tailwind color tokens (bg-primary/bg-secondary) to
semantic names (surface/surface-raised/content/content-muted/brand/line)
- Derive hover, overlay, border-strong, accent-muted and elevation tokens
once on <body> so they adapt to every theme automatically
- Make :root mirror Tokyo Night to kill the first-paint theme flash
- Add component primitives in @layer components: .card (surface + soft
shadow, no hard border), .btn variants, .input, .chip, .badge, .icon-btn
- Add theme-aware status/priority badges and global focus-visible styling
- Add an inline-SVG Icon() component (consistent stroke language) to
replace the mixed emoji/SVG iconography
- Data-drive theme/wood options and add an active-nav helper
- Add skeleton shimmer + x-cloak support
The reader chrome used the same dimensions at every screen size, and the book viewport offset was hardcoded to 52px. This made the header/footer oversized on phones and left the body text flush against (or overlapping) the bars, with no handling for notched-device safe areas.
- Add viewport-fit=cover so notched devices expose safe-area insets.
- Shrink the top bar on mobile (px-3 py-2 / text-base, scaling up at sm:) and hide the chapter title on phones (hidden sm:block sm:truncate).
- Shrink the bottom bar on mobile (tighter padding/gap, p-1.5 sm:p-2 on buttons) while keeping all controls visible.
- Replace the hardcoded top-[52px] bottom-[52px] viewport offsets with responsive calc() values (44/48px mobile, 60px at sm:) that fold in env(safe-area-inset-*), plus matching safe-area padding on the bars, so the book content always clears the chrome with a visible margin.
Regenerates reader_templ.go and rebuilds style.css.
The app service in docker-compose.yml had no restart policy (defaults to
'no'), so if the process exited -- e.g. the watcher-leak panic fixed in
the previous commit -- the container stayed down until a manual
restart. Adding restart: unless-stopped makes the container self-recover
from crashes or host reboots, while still honoring explicit 'docker
compose down'.
Defense-in-depth alongside the scanner leak/panic fix: even if a future
unforeseen panic occurs, the app comes back automatically.
The bookhoard container crashed with 'panic: Failed to create file
watcher: too many open files' (media_scanner.go) after running for a few
hours, preceded by floods of 'no space left on device' from watcher.Add.
Root cause: every scan job called NewMediaScanner(), which eagerly
created an fsnotify watcher. SetFolders() then walked the entire library
tree and registered one inotify watch per directory (~3,000+ across the
libraries), and ScanFolders() registered them again during its walk. The
worker never called scanner.Close() on these ephemeral per-job scanners,
and the worker loop had no recover(), so:
1. Leaked watchers accumulated until the kernel inotify watch cap was
hit (ENOSPC -> 'no space left on device'), then
2. the process fd limit (ulimit -n 1024) was exhausted, causing
fsnotify.NewWatcher() to fail with EMFILE, and
3. NewMediaScanner panicked on that error, taking down the whole
process (exit code 2). With no restart policy the container stayed
down.
The scan jobs run frequently (scan_poll_interval), so the leak built up
within hours. Note this was NOT a disk-space issue; df showed plenty free.
Fix:
- media_scanner.go: NewMediaScanner no longer creates a watcher eagerly
(s.watcher starts nil), which removes the panic site entirely -- there
is nothing to fail at construction. The watcher is created lazily only
when needed.
- media_scanner.go: SetFolders gains a [?1049h[22;0;0t[1;24r(B[m[4l[?7h[?25l[H[2JEvery 2.0s: bool[1;37Hgaruda-ser8: Fri 31 Jul 2026 10:45:41 AM EDT[2;66Hin 0.002s (127)[2;80H
[3dsh: line 1: bool: command not found
[4d[24;1H[?12l[?25h[?1049l[23;0;0t
[?1l> parameter. It creates
and populates a watcher (returning an error instead of panicking) only
when watch=true; otherwise it skips all watcher.Add calls. ScanFolders
guards its watcher.Add with a nil check, and the WatchChanges event
loop exits cleanly when there is no watcher (polling still runs).
- worker.go: the worker() loop now wraps each job in defer/recover() so a
panicking job is recorded as failed and can never kill the process.
- worker.go: the three ephemeral scan handlers (processScanJob,
processSetFoldersJob, processDirectoryScanJob) now defer scanner.Close()
and call SetFolders(..., false), so scan jobs allocate zero watchers and
zero inotify watches. Any pre-existing leak is also bounded by Close().
- handlers/scanner.go: the long-lived watch-mode scanners (StartScanner
and StartWatchModeForLibrary) pass watch=true since they actually read
watcher.Events for live change detection.
- calibre_integration_test.go: updated to the new SetFolders signature
(watch=false, matching one-off scan usage).
Auto-add is fully preserved: new files are still detected by the periodic
poller (startBackupScan), which is independent of fsnotify and unaffected
by these changes. The watch-mode event loop remains as bonus responsiveness
when inotify is available; through Docker bind mounts where inotify is
unreliable, polling is what catches new books.
Add a top-level run-name so the Gitea Actions runs list shows
'Release v0.3.0' rather than the tagged commit's subject. Uses the same
expression (inputs.tag || ref_name) as the TAG env, so it resolves for
both tag pushes and manual workflow_dispatch.
Replace the image-only tag pipeline with a full release workflow that also
publishes a Gitea Release whose body is the annotated tag's message,
generated from Conventional Commits by git-cliff. No hand-written release
notes are required.
- cliff.toml: group commits (Features, Bug Fixes, Refactor, Documentation,
Tests, Miscellaneous Tasks) with scopes and short-SHA links; emit only the
current tag's section rather than the full history.
- .gitea/workflows/release.yml: tag-driven. Reads the release body from the
annotated tag (git tag -l --format), so the tag message and release body are
a single source of truth. Idempotent create/PATCH; prints the Gitea API
error body on failure so a 403 names the missing token scope instead of
failing silently. Adds a workflow_dispatch tag input so manual re-runs
target the right tag instead of the default branch.
- Makefile: release VERSION=vX.Y.Z generates notes via git cliff --latest
against a throwaway tag, then creates an annotated tag with
--cleanup=verbatim so the markdown group headers are preserved (git's
default cleanup strips lines starting with "#").
- release: project-attached wrapper accepting a positional version arg
(./release 0.3.0 or ./release v0.3.0) and auto-prefixing v, for ergonomic
one-command releases.
Search results navigated to /bookshelf with no filters instead of the
selected book's page. Results now link to /media/:id and display cover
thumbnails, with cover URLs resolved server-side via ResolveMediaURL.
Removes the dead selectedBook localStorage plumbing.
The book detail page had no way to mark a book finished or reset its
read state from the UI. Reading state is modelled by reading_progress
alone, where 'read' is the canonical signal percentage >= 1.0 (used by
the dashboard Recently Read collection, analytics, and sync priority).
Add a single toggle button in the action row (after Read Now) whose
label is server-rendered from completion state:
- not read -> "Mark as Read" -> PUT /api/media-items/:id/progress
{ percentage: 1.0 }
- read -> "Mark as Unread" -> DELETE /api/media-items/:id/progress
Mark as Unread cannot use PUT { percentage: 0 }: the progress handler
silently ignores percentage < 0.005 when existing progress > 0.01
(internal/handlers/media.go anti-regression guard), so DELETE is the
only reliable reset.
If the book has an active sync mismatch (an unresolved sync_conflicts
row), the toggle resolves it first via POST /api/conflicts/:id/resolve
before writing progress. Order matters: resolving sets resolved_at,
arming the 10-minute HasRecentConflictResolution suppression window so
the subsequent progress write does not spawn a brand-new conflict. The
resolve winner is any valid source key from the conflict data (prefers
"web"); it does not affect the final state, which the progress write
sets. A 400 "already resolved" response is tolerated.
Notes, highlights, and ratings are independent of reading_progress (they
reference media_items, not progress) and are never affected by the
toggle. After toggling the page reloads so the progress card, Sync
Progress button, and conflict banner re-render server-side.
- templates/utils.go: add conflictWinnerSource and conflictID helpers.
- templates/book_detail.templ: data-conflict-id/winner on <body> and the
toggle button.
- web/src/book-detail.ts: toggleRead() + conflictId/conflictWinner/
readSaving state (read from <body> in init()).
- templates/book_detail_templ.go regenerated.
Running `templ generate` to pick up the book_detail changes also
resynced book_detail_modals_templ.go, whose committed output was stale
relative to its source. The regeneration (templ v0.3.1020) reformats
boolean attribute rendering (e.g. `selected`) via
templ.ResolveAttributeValue and reflects pre-existing source additions
such as id/for label associations.
No source (.templ) change in this file; generated output only.
The Bruno collection docs mislabeled the rating system and referenced
endpoints that do not exist.
- Update Media Rating.yml: the rating value is a 1-10 integer scale
(displayed as 1-5 stars with half-star precision), not "typically 1-5".
- opencollection.yml: the rating routes live under
/api/media-items/:id/rating (not /api/ratings/:media_id), GET returns
null (not 0) when unrated, and document the PUT upsert route. Correct
the scale to 1-10 here as well.
The book detail page only displayed user ratings as static, non-clickable
stars. The full rating CRUD stack already existed in the backend
(media_ratings table, POST/GET/PUT/DELETE /api/media-items/:id/rating)
but nothing in the web UI could create or update a rating.
Replace the display-only renderStars output for the user rating with an
Alpine.js widget that:
- Renders 5 stars, each split into two transparent hit zones so the
underlying 1-10 scale maps to half-star precision (left half = x.5,
right half = whole star).
- Shows a live hover preview via a ratingHover state field.
- Saves the rating in place through POST /api/media-items/:id/rating
(which upserts) and reflects the value immediately, with no full page
reload.
- Displays the numeric value (e.g. "3.5 / 5") and a Clear button that
issues DELETE to remove the rating.
- Reads the server-rendered value from a new data-rating attribute on
<body> during the bookDetail component init().
The community rating block is left as a display-only renderStars render
since it is imported metadata, not a user rating.
templates/book_detail_templ.go is regenerated (also picking up templ
v0.3.1020 reformatting of the generated output).
Update the OPDS section of the API reference to reflect the now-working
catalog:
- Document the page/per_page parameters and that paging is driven by the
rel=next/previous/first/last links plus OpenSearch paging metadata.
- Refresh the example feed XML to show the pagination links, opensearch
namespace/elements, and standard Atom <title>/<author> elements.
- Document the search endpoint's two modes: OpenSearch description
(application/opensearchdescription+xml, no q) and results feed (with q),
with an example description document.
The device catalog feed was unusable on paged OPDS clients such as
KOReader: it sliced results into pages but never advertised how to reach
the next page, so clients could only ever fetch the first page (~50 books)
and could not search the catalog.
GetDeviceCatalog:
- Emit the full set of OPDS pagination link relations (self, start, first,
previous, next, last) pointing at catalog?page=N&per_page=M, with the
device auth token appended for path-based auth.
- Emit OpenSearch totalResults/itemsPerPage/startIndex metadata.
- Point rel=search at the OpenSearch description (correct MIME type).
SearchDeviceCatalog now branches on the q parameter:
- No q: return an OpenSearch description document whose Url template
contains the {searchTerms} placeholder, so clients can formulate a query.
- With q: return the existing acquisition results feed, now including
totalResults.
A pure addCatalogPaginationLinks helper holds the page/URL logic so it can
be unit tested without a database. New handler tests cover middle/first/
last/single/empty pages (correct presence of next/previous) and token
appending.
Ordering is intentionally left unchanged (created_at DESC, grouped by
library).
Extend the OPDS feed model so clients can page through large catalogs and
discover how to search them.
Feed changes:
- Add the OpenSearch namespace (xmlns:opensearch) to all feeds.
- Add optional TotalResults/ItemsPerPage/StartIndex fields, serialized as
<opensearch:totalResults>, <opensearch:itemsPerPage> and
<opensearch:startIndex>, plus a SetPagination helper.
- Add OpenSearchDescription/OpenSearchUrl types and a NewSearchDescription
constructor with GenerateXML/GenerateXMLString. This produces the
OpenSearch description document (application/opensearchdescription+xml)
that OPDS clients like KOReader fetch to learn the {searchTerms} search
URL template.
These are building blocks; the handlers are wired up in a follow-up commit.
Tests cover SetPagination, omission when unset, XML emission of the
paging metadata, and OpenSearch description generation/serialization.
The OPDS feed test suite did not compile or pass:
- TestNewEntry asserted on entry.Creator, but the Entry struct stores the
creator under Author.Name (the Atom <author><name> element). Assert on
entry.Author.Name instead.
- TestFeedGenerateXML expected <dc:title>/<dc:creator> elements, but the
Entry struct emits standard Atom <title> and <author><name>. Update the
expected substrings to match the actual (correct) output.
These are pre-existing assertion errors unrelated to any field being
removed; the code under test was already correct.
The UI had no surface showing how many media items have been imported.
Surface the total in the library switcher shown on the Dashboard, Series,
and Collections pages (via the LibrarySwitcher component) and in the
Bookshelf's inline library filter.
- Add a MediaCount field to LibraryData and a TotalMediaCount helper to
sum counts for the "All Libraries" / "All Books" option.
- resolveLibrary() now fetches per-library counts (one query) and maps
them onto each LibraryData entry, so the switcher reflects the active
scope without changing the component's signature.
- Each library option renders "(N)" and the "All" option renders the
grand total across the user's visible libraries.
The "All" total is the sum of the user's visible libraries, correctly
respecting per-user library visibility rather than a raw global count.
Regenerated templ files for library_switcher and bookshelf.
Add GetVisibleLibraryMediaCounts, which returns the media item count for
each library visible to a given user in a single GROUP BY query over
media_items. It mirrors the visibility logic in GetUserVisibleLibraries
(libraries default to visible unless an explicit false row exists) so
counts can be resolved in one round-trip instead of N per-library
lookups.
Regenerated sqlc bindings (querier.go, queries.sql.go).
Gitea's auto GITHUB_TOKEN lacks the package scope needed to push to the container registry, causing the login step to fail. Switch the login password to a PAT stored as the REGISTRY_TOKEN repo Actions secret (scopes: write:package, read:package).
Adds .gitea/workflows/release.yml. On a v* git tag push (or manual dispatch), builds the Dockerfile and publishes to git.linuxhg.com/bookhoard/bookhoard under two tags: the version (${{ gitea.ref_name }}) and 'latest'. Auth uses the auto-provided GITHUB_TOKEN; no secret to manage. Pushes to main do nothing, so work-in-progress commits never ship.
Expose the recently-added env-driven compose settings as commented examples so self-hosters and deployers can discover them. All remain optional with defaults.
Replace hardcoded port literals with env-driven variables so a single change
in .env reconfigures the full stack consistently. Defaults are unchanged
(DB 5432, app 8765), so existing setups need no .env changes.
- DB_PORT (default 5432): drives the db host<->container port mapping,
Postgres PGPORT (so it listens on the chosen port), and the app's
DATABASE_PORT connection setting. Lets deployers avoid a host port conflict
(e.g. another local Postgres) by setting DB_PORT once.
- SERVER_PORT (default 8765): drives the app host<->container mapping, the
SERVER_PORT the app listens on, and the healthcheck target URL.
- Applied to both the base (docker-compose.yml) and the dev override
(docker-compose.dev.yml, tests service) so dev and prod stay in sync.
Address compose issues surfaced on first production deploy:
- Remove obsolete `version: "3.8"` (ignored by Compose v2; caused a warning).
- Fix BASE_URL: it used compose-time interpolation of ${SERVER_PORT}, which is
only defined as a runtime container env var (invisible to interpolation) and
absent from .env. This resolved to an empty string, producing a broken
`http://localhost:` (no port) and a startup warning. Now
${BASE_URL:-http://localhost:8765}, overridable per-deployment via .env.
- Move COOKIE_SECURE from the db service to the app service and make it
configurable (${COOKIE_SECURE:-false}). It controls the session cookie Secure
flag, an app concern; on the db service it was a no-op, so the app never
received it and cookies were always non-secure. Set COOKIE_SECURE=true behind
a TLS-terminating reverse proxy (Caddy/nginx/traefik), where the app speaks
plain HTTP internally.
- Image reference unchanged: ${IMAGE_TAG:-latest} (no hardcoded version).
Restructure the container setup to support registry-based deployment:
the default docker-compose.yml now pulls a prebuilt app image from the
Gitea container registry instead of building locally, while a new
docker-compose.dev.yml override preserves the local build + integration
test workflow for development.
Why:
- Production and self-hosting should consume a published image, not
rebuild from source on the host. The default `docker compose up` now
pulls the app image (git.linuxhg.com/bookhoard/bookhoard) alongside the
public postgres image, with no build step required.
- Development still needs to build from source and run integration
tests, so those concerns move to an override file the Makefile applies.
Shared config (env, volumes, ports, healthchecks) lives in one place to
avoid drift between environments.
Changes:
- docker-compose.yml (prod base): the app service now references
`image: git.linuxhg.com/bookhoard/bookhoard:${IMAGE_TAG:-latest}` instead
of a build context. The tests service is removed (moved to the
override). IMAGE_TAG lets deployers pin or roll back a specific version.
- docker-compose.dev.yml (new override): adds the local `build:` context
for the app and defines the integration `tests` service (profile-gated).
Everything else is inherited from the base file via compose merging.
- Makefile: introduce a COMPOSE variable that merges the base and
override (`-f docker-compose.yml -f docker-compose.dev.yml`); all dev
targets now use it. Plain `docker compose` against the base file only
remains the production path.
- README: quickstart updated to pull and start prebuilt images; clone URL
points at the Gitea instance.
The development workflow (`make rebuild-app`, `make test-integration`,
etc.) is functionally unchanged.
Complete the annotation sync pipeline across all ingest and serve paths.
Previously, annotations sent inline with KOReader progress pushes were
silently discarded, and no annotations were ever served back to devices.
INGEST (device → server):
KOReader (koreader.go):
- Add processBookAnnotations helper that processes inline highlights,
notes, and bookmarks from every progress push (immediate + checkpoint)
- Highlights get CRE→CFI position conversion before SaveHighlight
- KOReader 'notes' (text + notes) stored as highlights with NoteText
to ensure correct round-trip classification
- Bookmarks routed through SaveBookmark with device sync data
- Called from both updateProgressForBook and handleCheckpointSync
Kobo (kobo.go):
- Markup handler: annotations and bookmarks route through
AnnotationService (SaveHighlight/SaveBookmark)
- Bookmark handler: same routing with device sync data
- SyncFromServer handler: same routing
- All handlers fall back to direct DB calls when annotationSvc == nil
Web reader (media.go):
- CreateMediaHighlight → SaveHighlight (Source="web", ModifiedAt=now)
- CreateMediaNote → SaveNote (Source="web")
- DeleteMediaHighlight → TombstoneHighlightByID
- DeleteMediaNote → TombstoneNoteByID (was hard delete, now tombstone)
- All fall back to old behavior when annotationSvc == nil
SERVE (server → device):
KOReader GetMetadata (koreader.go):
- Query and serve bookmarks from media_bookmarks table (was missing)
- Serve deleted_highlights and deleted_bookmarks arrays containing
device_sync_data + dedup_key for client-side deletion
- Highlights/notes already served with reverse CFI conversion
Kobo Markup handler (kobo.go):
- Track processed books during sync
- Query tombstones per book, extract bookmark_id from device_sync_data
- Return DeletedAnnotations array in KoboSyncStatus response
Conflict resolution (conflicts.go):
- Enable annotation conflict types in ResolveConflict handler
- Add applyAnnotationResolution dispatching to:
applyHighlightResolution / applyBookmarkResolution / applyNoteResolution
- Each looks up by dedup_key and applies winner's fields
- Allow manual override of auto_resolved conflicts
(changed check from != "unresolved" to == "user_resolved")
Infrastructure:
- AnnotationService field + SetAnnotationService in router Config
- Inject AnnotationService into KOReader, Kobo, Media handlers
- Start tombstone purger goroutine in main.go (24h interval)
- Test helpers: construct AnnotationService in test setup
Wire AnnotationService into SyncQueueProcessor and implement the three
previously-stubbed execute methods:
- syncHighlight: unmarshals syncData JSON into SaveHighlightRequest,
applies CRE→CFI conversion via AnnotationService
- syncNote: unmarshals into SaveNoteRequest
- syncBookmark: unmarshals into SaveBookmarkRequest
- Add SyncTypeBookmark to executeSync switch (was hitting default error)
Add enqueue methods for future offline/batch use:
- EnqueueHighlight / EnqueueNote / EnqueueBookmark
- Shared enqueueAnnotation helper creates queue items with
PriorityCriticalNote and 3 max attempts
- Update types (HighlightUpdate, NoteUpdate, BookmarkUpdate) mirror the
existing ProgressUpdate pattern
Existing handler behavior is unchanged — annotations still sync
synchronously via AnnotationService. The queue path is available for
retry-on-failure and offline batch processing scenarios.
AnnotationService is the central service for cross-device annotation sync.
It provides SaveHighlight, SaveNote, and SaveBookmark methods that handle
the full sync lifecycle:
Identity (3-layer):
1. Server UUID (primary key)
2. Per-device native ID stored in device_sync_data JSONB
3. Content dedup_key: sha1(normalize(selection_text) + bucket_position)
- CFI character offsets are stripped for bucketing so the same
highlight at slightly different offsets still deduplicates
- Raw positions are preserved in the DB for precise restoration
Resolution policy (LWW):
- When the incoming annotation has an explicit ModifiedAt timestamp,
last_modified_at wins
- When the device sends zero ModifiedAt (creation time only), field-diff
mode compares content fields (text/color/note/percentage) — if all
match, the save is skipped; if any differ, the save is applied with
server-receive-time as the new last_modified_at
Conflict detection:
- When incoming and existing annotations have different sources (e.g.
koreader vs kobo) and content differs, an auto_resolved sync_conflict
is recorded with both sides' data for audit trail
- Broadcasts a WebSocket conflict notification for real-time UI updates
Tombstone management:
- Delete-wins: tombstoned annotations block recreation from stale pushes
- 30-day TTL before physical purge
- PurgeExpiredTombstones method + StartTombstonePurger goroutine (24h ticker)
Add locators.go with unified bidirectional CFI conversion:
ConvertToCanonical / ConvertFromCanonical
- CRE XPointer <-> standard EPUB CFI (for KOReader)
- KEPUB CFI passthrough (for Kobo)
- Skips non-reflowable formats (PDF, CBZ, fixed-layout EPUBs)
Add 25 unit tests covering:
- Dedup key determinism, text normalization, position sensitivity
- Offset insensitivity (CFI char-offset bucketing)
- Device sync data merge (preserves existing, overwrites same source)
- Cross-source detection
- LWW comparison (newer wins, older skipped, fallback to updated_at)
- Field-diff mode (identical content skipped, changes applied)
- Tombstone TTL constant
- CRE XPointer parsing and classification
- Standard EPUB CFI classification
Add migration columns to media_highlights, media_notes, and media_bookmarks
for cross-device annotation sync:
- dedup_key: SHA-1 of normalized selection text + bucketed position, used
as the stable cross-device identity for annotations
- last_modified_at / last_modified_source: edit clock for LWW resolution
and cross-source conflict detection
- deleted / deleted_at: sticky tombstone columns for delete-wins semantics
with a 30-day TTL before rows are physically purged
- note_text on highlights: stores attached notes from KOReader entries that
have both selected text and a user note
- Location columns on bookmarks (cfi_position, percentage_location,
epubcfi_location, chapter_reference, paragraph_reference)
- device_sync_data JSONB on all three tables: stores per-device native
identifiers (e.g. KOReader pos0/datetime, Kobo bookmark_id) so each
device can locate and manipulate its own copy of an annotation
Add partial unique indexes on (user_id, media_item_id, dedup_key) where
deleted = FALSE to enforce one active annotation per dedup key.
Add tombstone purge indexes on (deleted, deleted_at) for efficient GC.
New queries:
- GetByDedupKey for all three tables (returns active or most-recent tombstone)
- CreateFull / UpdateForSync for all three tables (populate sync columns)
- TombstoneByDedupKey / TombstoneByID for all three tables
- PurgeExpired* for all three tables (GC past TTL)
- GetActiveAnnotationsForBook (filtered union of highlights + notes)
- GetTombstonedAnnotationsForBook (union of all 3 deleted within TTL)
- GetMediaBookmarks with deleted filter
- GetMediaBookmark (singular) with deleted filter
- Added deleted=FALSE filter to GetMediaHighlights, GetAnnotationsForBook
- CreateAutoResolvedSyncConflict (INSERT with resolution_status='auto_resolved')
Quote DATABASE_PORT and SERVER_PORT ("5432", "8765") in docker-compose.yml so they are treated as strings rather than YAML integers, avoiding type-coercion warnings from compose runtimes.
Setup completion was previously tracked by a manually-flipped setup_complete row in system_settings, written via a JWT-protected PUT /api/setup/complete endpoint. This meant any admin user created outside the setup wizard (future CLI, seed scripts, direct DB inserts) would not flip the switch, leaving the app stuck redirecting to /setup.
The trigger is now derived from real data: setup is complete iff at least one admin user exists. This is self-correcting regardless of how users are created, and re-engages setup automatically if all admins are ever removed.
Changes:
- Add internal/setupstatus package with IsSetupComplete() (queries CountAdmins, 10s in-memory cache, fails open on DB error) and Invalidate() to clear the cache. Uses an AdminCounter interface to avoid importing the database package.
- Add CountAdmins sqlc query (SELECT COUNT(*) FROM users WHERE role = 'admin') and regenerate.
- Rewire router/setup.go isSetupComplete() to delegate to setupstatus; drop the old setup_complete setting read, cache vars, and the PUT /api/setup/complete route.
- Call setupstatus.Invalidate() in the auth handler after CreateUser, UpdateUserRole, and DeleteUser so the cache reflects admin-count changes immediately.
- Align first-user promotion in Register to key off !adminExists instead of len(users) == 0, so the two checks cannot diverge.
- Remove the now-dead SetSetupComplete/GetSetupStatus handlers.
- Drop the setup_complete seed row from schema.sql.
- Remove the apiPut('/setup/complete') call from the setup wizard finishSetup(); the admin account created in submitAdmin already marks setup complete server-side.
john-okeefe