Auto-added airing episodes came in metadata-empty (no synopsis, no still) — the
handler only passed season/episode/title/air_date, dropping the overview the
calendar already returns and never fetching the still URL (calendar_upcoming
only returned a has_still flag, not the URL). Now calendar_upcoming also returns
e.still_url, and the handler carries overview + still_url through. The wishlist
renders the (Plex-relative) still via the same pimg() proxy as the show poster,
so it resolves. Idempotent upsert backfills the already-added empty rows on the
next run.
_fix_video_scan_default set its 'done' flag even on runs where it deleted
nothing, so once the flag latched True the standalone 'Scan Video Library'
system automation survived forever (the row the post-download chain replaced).
Drop the flag entirely — get_system_automation_by_action already matches only
the is_system-seeded row, so the cleanup is safe to run every startup and
no-ops once the row is gone.
'Holy Marvels with Dennis Quaid' is stored with a leading space in the shows
table, so the watchlist sort key fell to ' holy marvels…' — and a leading space
sorts before 'a', jumping it to the top. .strip() the sort key so dirty titles
sort by their real first letter.
The default sort put manual follows first (newest date_added), then airing
shows A–Z — so recently-followed shows like 'Welcome to Widows Bay' and 'From'
jumped to the top. A manual follow is no more special than an auto-added airing
show; default now sorts everything by name A–Z. 'added' (newest first) stays as
an opt-in sort.
Two Sonarr-parity features.
1) Per-episode live tracking. "Grab season" was headless (only a button label
changed); episode rows had a status span that was never populated. Now every
episode ROW shows its own live state — Searching → Downloading % → Downloaded
/ Failed — via epTrack() polling /downloads/status?id, matching the inline
movie tracker. Grab season lights all target rows at once; manual + per-source
auto grabs also light their row; reopening the modal resumes tracking in-flight
episodes (resumeEpisodeTracking via /downloads/active + search_ctx match).
Season batch grabs through the same payload as a manual grab (_pickAndGrab →
sendGrab(buildGrabPayload)).
2) Auto-wishlist airing episodes. New daily automation (video_add_airing_episodes):
reads the calendar for episodes airing TODAY for followed shows, skips owned
ones, adds the rest to the wishlist (idempotent). Handler uses injected seams
(calendar read + wishlist write) so it's unit-tested without a DB/server.
Registered + action block + seeded as a daily system automation (01:00),
owned_by=video.
Two over-rejections that filtered out legit releases as "unknown quality":
- the source parser only matched WEB-DL/WEBDL, not plain "WEB" (very common)
- a release with a known resolution but no recognized source had no tier
Now plain WEB parses as web-dl, and a resolution-only release assumes web so
it lands on a tier instead of being rejected (ffprobe verifies the real quality
after download). Truly quality-less packs still reject. Tests added.
End-to-end import for video grabs, mirroring the music side's rigor and the
Radarr/Sonarr standard, fully isolated in core/video + api/video.
- Importer: parse release -> ffprobe-verify (true resolution, reject corrupt/
samples) -> templated rename into Movie (Year)/ + Show/Season NN/ -> copy or
move, carry subtitles, upgrade-replace a worse copy.
- Library Organization settings: editable $token path templates + toggles
(transfer mode, verify, replace, carry subs, save artwork, write NFO,
download subtitles + langs). Stored in video.db; matches the music File
Organization section's look.
- Sidecar writer: movie.nfo / tvshow.nfo + full artwork set (poster, fanart,
clearlogo, season posters) from on-demand TMDB detail, and external .srt from
OpenSubtitles. Owned re-grabs resolve their library tmdb_id; tmdb_full_detail
bypasses the owned->library redirect so they enrich too.
- Import page: surfaces import_failed downloads, resolve by hand (library-first
-> TMDB picker -> force-place) or dismiss; fires a library refresh on place.
- "Grab whole season": episode-level batch (reuses searchInto + _autoPick).
- Brutalist redesign of the download modal sources + result cards.
All new logic has seam-level tests (pure parsers/planners + injected I/O);
sidecars/subtitles are best-effort and never break an import.
Reworked the whole sources/results area for a more premium, Netflix-y feel.
Results — flat divider list → real CARDS: a bold cinematic quality badge (resolution
over source), the release name as the hero, the meta as a row of crisp PILLS (codec /
audio / HDR / repack / group) plus a clean availability stat, then size · verdict · Get
grouped on the right. Cards have depth, a hover-lift, an accent edge on accepted
releases, and the auto/grabbed pick lights up with an accent ring + glow.
Sources — toned the full per-source colour wash down to a sophisticated dark-glass row
with the brand colour as an ACCENT only (left-edge light that extends on hover + the
glassy icon tile), so it reads calm and designed instead of busy.
All functional hooks preserved (data-vdl-card/grab, .vdl-res-main tracker dock, status
states, auto-pick). node --check + CSS balance clean; tracking/auto wiring tests green.
The wishlist page had no way to empty a tab — only per-item remove. Added a red-tinted
'Clear all' button in the toolbar that empties the ACTIVE tab in one click (after a
confirm), shown only when that tab has items.
- db.clear_wishlist(kind) maps the tab to its rows (movie→'movie', show→'episode',
youtube→'video') and deletes them; returns the count.
- POST /api/video/wishlist/clear {kind: movie|show|youtube}.
- Toolbar button + clearAll() (confirm → clear → reload) + updateClearBtn() visibility.
Tests: per-tab clear leaves the others intact, unknown-kind/empty no-ops, the endpoint,
and the frontend wiring. node --check clean.
A scan wrote server-provided fields straight over the row, so an incremental/deep
re-read wiped the TMDB-backfilled 'status' (Plex returns it blank) — clearing the
airing watchlist. Now matches the intended model: incremental (add recent) and deep
(coverage + prune) PRESERVE enrichment-owned fields the server left blank; only a FULL
scan clobbers them (an explicit reset / fresh start).
- _resilient_upsert gains preserve_enrichment (default True): on a conflict UPDATE,
enrichment-owned columns (per _ENRICH_META_COLS: status/network/ratings/air dates/…)
take the server value only when non-blank, else keep what's stored. A real server
value still wins.
- upsert_movie/upsert_show_tree thread the flag; scanner passes preserve=(mode!='full').
Tests: preserve-on-blank, server-value-wins, full-resets, and the scanner picking the
right mode. 88 DB + 18 scanner tests + isolation green.
Superseded by the post-download chain (Auto-Scan Video After Downloads → Auto-Update
Video Database After Scan), which keeps the library fresh without a separate scheduled
scan. Two small changes: drop it from SYSTEM_AUTOMATIONS (no longer seeded), and a
one-time flag-guarded cleanup (_fix_video_scan_default, v3) that DELETES the existing
system row — so it actually stops running, not just hidden. The video_scan_library
action/handler/block stay, so a custom scan automation can still be built later.
Engine seed test updated. 66 automation tests pass.
Mirrors the music flow: a finished download batch → refresh the media server → (after
it indexes) pull the new media into the DB — so a downloaded movie/episode shows as
owned without waiting for the 6h scheduled scan.
Two event-based system automations (owned_by='video'):
- 'Auto-Scan Video After Downloads' (video_batch_complete → video_scan_server)
- 'Auto-Update Video Database After Scan' (video_library_scan_completed → video_update_database)
Pieces:
- core/video/download_events.py: a callback registry (core/video can't import the
engine — isolation). The monitor publishes batch-complete; web_server bridges it to
automation_engine.emit('video_batch_complete', …), like music's web_scan_manager.
- download_monitor: fires the batch-complete event once, when the last in-flight
download finishes (none queued/downloading/searching left).
- video_scan_server handler (stage 1): refresh server, wait a debounce for indexing,
then emit 'video_library_scan_completed' (mirrors music's time-based completion).
- video_update_database handler (stage 2): incremental read (newest-first, stop after
25 consecutive known — same as music).
- blocks: 2 video triggers + 2 video actions (scope='video'); registration; seeds.
kettui: seam tests for both handlers (refresh/wait/emit, incremental read, error paths),
the event registry (idempotent register, isolated failures), and the monitor (fires once
on last completion, never while work remains). 298 automation tests + isolation green.
The scheduled 'Scan Video Library' defaulted to a FULL scan (re-reads everything +
prunes) that fired ~5 min after every app start. A recurring scan should be light:
switched the seed to INCREMENTAL (newest-only, no prune) and pushed the first run a
full interval out so it runs on its 6h cadence rather than right after startup.
One-time migration (_fix_video_scan_default, flag-guarded) corrects existing rows that
still carry the old full+startup default, without overriding a deliberate user choice.
Alongside the per-worker 'Retry all failed', the worker modal now has a topbar
'Retry all failed' that re-queues every failed/not_found item across ALL workers and
kinds in one click — one-shot recovery after an API outage left lots errored.
- db.retry_all_failed() derives the full service+kind set from the same _ENRICH /
_BACKFILL maps the workers use (tmdb/tvdb + omdb + fanart/opensubtitles/trakt/
tvmaze/anilist/wikidata + ryd/sponsorblock/dearrow), loops enrichment_retry, returns
the total re-queued. POST /api/video/enrichment/retry-all-failed.
- Topbar button (amber, text) → calls it, toasts the count, refreshes the modal.
DB test (resets across matcher + backfill + youtube service, deterministic count) +
frontend wiring test. ruff + node --check clean.
The calendar pulled every airing show in the library regardless of whether you
follow it. Now it's scoped to the EFFECTIVE watchlist by default — explicit show
follows ∪ airing library shows (not muted), same logic as the Shows watchlist tab —
so it tracks what you actually care about, and you can mute a show off it. A
'Watchlist / All library' toggle on the calendar lets you flip to everything you
own (remembered in localStorage).
- calendar_upcoming(watchlist_only=) adds the watchlist filter; /calendar takes
?scope=watchlist|all (default watchlist).
- Calendar page gets a scope toggle (defaults watchlist, persists, refetches on
change).
Tests: DB scope (followed-only / airing-default / mute drops out / all-library sees
all) + frontend wiring. node --check clean.
Person follows are already supported (video_watchlist kind='person', add/remove/check
API, and the button shows on person CARDS) — but the person DETAIL page had no way to
follow or see if a person is followed. Added the standard watchlist button to the
person hero: renderWatchlist() builds it + lazily checks the followed state,
toggleWatch() adds/removes via the person-kind watchlist API, wired through the page's
delegated click handler. Same chrome as the movie/show pages.
4 wiring tests; node --check clean.
On a deep scan the progress bar hits 100% when the last item is read, but the prune
(delete orphaned rows + cascades) runs AFTER that and — on a big cleanup — takes a
few seconds, during which the scan still reads as running (can't start a new one,
workers still paused). Looked stuck at 100%. Now the scanner sets a 'cleaning up
removed movies/shows' phase around the prune so the UI shows it's finalizing, not
frozen. Test spies the phase at prune time. 18 scanner tests pass.
1) Scan only the MAPPED libraries — never fall back to 'all'. The scan path used
_sections/_views with the selected name, but an empty name returned ALL sections
of that type — so a missing/unreadable selection silently scanned every library
(how the 4K movie + 'YouTube' TV libraries leaked in as movies/shows). New
_scan_sections / _scan_views return [] when a kind isn't mapped; available_libraries
still lists all (for the Settings dropdown). Now an unmapped kind scans NOTHING.
2) A library scan (full/incremental/deep) now pauses EVERY enricher, including the
YouTube date enricher — a separate singleton outside engine.workers that kept
running through scans. pause_for_scan/resume_after_scan pause+resume it too, only
if it wasn't already manually paused (never override the user).
kettui: source-scope tests (mapped-only / unmapped-scans-nothing / listing still
shows all) + engine pause tests (pauses+resumes YT / preserves manual pause).
123 scanner/source/enrichment tests + isolation guard green.
enrichment_retry handled ryd/sponsorblock + the keyed backfills, but DeArrow (also a
youtube_video_stats backfill, keyed on dearrow_status) fell through to nothing — so the
modal's Retry button did nothing for DeArrow. Fold dearrow into the youtube_video_stats
retry branch. Regression test re-queues failed dearrow rows, leaves matched ones.
Two UX fixes from feedback:
- Clicking a coverage card on a worker was also setting the GLOBAL 'process first'
priority (and silently re-queuing that kind's failed items) — so picking a worker's
coverage reached across and re-prioritised every worker. Now a coverage card just
switches the view; priority changes only via the top Movies/Shows/Auto tabs, and
re-queuing is the explicit 'Retry all failed' button. (Removed the now-dead
requeueFailed helper.)
- 'Retry all failed' now re-queues EVERY coverage kind the worker handles (movie+show,
etc.), not just the tab you're viewing — matching the 'all' in the label.
3 wiring tests; node --check clean.
The details backfill queue is keyed on tmdb_id, but the gate (hasattr match) let the
TVDB worker run it too — feeding TMDB ids to TVDB's /series/{id}/extended (→ 404 on
every show) and double-processing each show (TMDB backfilled it, TVDB then 404'd but
still logged 'Backfilled'). Gate on self.service=='tmdb' so only the TMDB worker runs
it. Regression test: the TVDB worker no-ops (never calls its client, leaves the item
pending). 95 enrichment tests pass.
The media server pre-matches shows/movies (tmdb_id set), so the enrichment matcher
skips them and never fetches TMDB details — leaving details-only fields like `status`
(airing vs ended) blank on almost the whole library. That's why the watchlist's
airing-shows default only ever saw the handful of shows whose detail page had been
opened (the one path that force-fetches details). Library here: 3,371 matched shows,
only 18 with status.
Fix: a one-time-per-item details backfill that runs in the enrichment worker's idle
loop (after the episode-sync pass). New `details_synced` marker column on shows+movies;
detail_backfill_next/mark_details_synced/pending_count; worker._detail_backfill_one()
re-fetches an already-matched item's TMDB details and gap-fills (never clobbers server
data), then marks it done so it's attempted once. No re-scan needed — it heals the
existing library in place, and once status is populated the airing-watchlist reflects
real TV.
It's a background gap-fill on already-matched items (like episode coverage), so it
doesn't block the worker's 'Complete' status. kettui: DB seam tests + worker tests
(fills status / enrich-by-id / marks-done-when-absent). ruff + isolation guards green;
94 enrichment tests pass.
Auto: replaced 'Manual all' + 'Auto all' (which fired auto on every source = up to
one download PER source = duplicate copies) with a single header 'Auto' button. It
searches every source, waits for them all to settle, compares the accepted+grabbable
hits across ALL sources by quality-profile score (tie-break on availability), and
grabs exactly ONE winner — the chosen row gets the auto ring + live tracker. Per-source
Manual/Auto buttons are unchanged.
Bug fix: viewing a YouTube channel populated the playlist section in the SHARED
show-detail DOM; opening a real movie/show afterward still showed those playlists,
because ytResetPlaylists() used the kind-scoped q() (pointing at the wrong root on a
movie load) and wasn't called on normal loads. Now it targets the show subpage
directly and runs from resetExtras() (every detail load), so stale playlists are
always cleared.
node --check clean; 20 wiring/regression tests green; all video-only.
Two fixes from feedback:
1) Reopening the download modal now KNOWS a download is already running. The view
gets a persistent active-download banner at the top that looks the title up by
media identity (/downloads/status?media_id=) on every open and polls while active
— progress bar + release name + 'Track on Downloads ↗'. Suppressed while a result
card is already tracking inline (fresh-grab case) so there's never a double
indicator.
2) Result cards completely redesigned (third time's the charm): dropped the rounded
cards + big resolution tile for a flat, release-list layout (Radarr/Prowlarr
style) — hairline dividers, a small colour quality tag + source word on the left,
the RELEASE NAME as the hero line, dense inline meta (codec · audio · HDR ·
uploader · group) under it, then size · a compact ✓/✕ verdict flag · a compact
accent 'Get' pill. The selected/auto/grabbed row tints + rings in place, and the
live tracker still docks under the chosen row.
All video-only. node --check clean; 16 tracking/auto wiring tests + the status
endpoint test green.
After a grab (manual or Auto) the user now SEES what happened and can follow it:
- The chosen result card is spotlighted (Auto scrolls it into view) and grows a live
tracker: a progress bar that follows the real download + a 'Track on Downloads ↗'
button that closes the modal and jumps to the Downloads page. Polls the new
GET /api/video/downloads/status?id= until the download reaches a terminal state.
- A movie's detail page shows a live download chip (progress bar + %) for any in-flight
download of that title; clicking it jumps to Downloads. Looks up by media identity
via /downloads/status?media_id=&media_source=, polls while active, clears on
navigate-away. (video-detail.js)
- Result CARDS redesigned (the part you didn't like): a column card with a colour
resolution badge, a green 'accepted' edge, cleaner hierarchy, and the grab button is
now an accent 'Get' pill matching the source Auto button language.
Plumbing: new lightweight /downloads/status endpoint (by id, or by media for detail
pages); soulsync:video-navigate event in video-side.js to reach a top-level page from
anywhere; VideoGet.close exported so the tracker can dismiss the modal. All video-only.
Tests: status endpoint (by id + by media + null cases) in test_video_api.py;
tracking/detail/nav wiring in test_video_download_tracking.py. node --check clean;
isolation guards green.
The sources half felt weak next to the animated top half (quality chips + glowing
verdict), and the dark-text lightning Auto button looked off. Redesigned the whole
sources block:
- Buttons are now a cohesive pair: Manual = quiet ghost (outline), Auto = hero —
brand-filled gradient, white text w/ shadow for legibility on bright brands, a
soft continuous brand glow (vdlAutoGlow, --glow set per element) + a sheen sweep
and a sparkle twinkle on hover. Swapped the harsh ⚡ for a clean monochrome ✦
that inherits the button colour. 'Manual all'/'Auto all' header buttons speak the
same language (ghost vs accent hero).
- Source rows: richer brand card — bigger glassy icon tile w/ inner highlight +
halo, a stronger brand gradient, an inset top highlight, a glowing left brand
edge (::before), and the status is now a brand-tinted pill (was bare dot+text)
with state colours (scanning/done/none). Scan bar moved to ::after.
- Section labels get a small accent tick so both halves read 'designed'.
- Reduced-motion + mobile (full-width stacked buttons) handled.
Icon+label split into spans for finer control. node --check clean; 7 tests
updated/green.
Each source in the movie/YouTube download view had one 'Search' button (manual —
you pick a release). Adds a second 'Auto' button beside it that runs the SAME
search and then auto-grabs the best release for your quality profile; renamed the
pair to 'Manual' / 'Auto' for clarity (+ a matching 'Auto all' beside 'Manual all').
How 'best' is chosen: the backend already returns hits ranked best-first
(accepted → score → availability — see test_downloads_search_endpoint_ranks_and_filters),
so Auto just waits for the search to settle, then takes the first accepted hit
that has an uploader and grabs it. The chosen release card gets a ring + the row
shows Auto-grabbing → Sent, so the pick is transparent.
- searchInto/_pollSearch gain an onDone callback (fires when results settle); the
immediate (mock) path fires it too.
- doGrab refactored into shared buildGrabPayload + sendGrab so the manual button
and _autoPick send an identical /grab request (incl. the auto-retry candidate pool).
- searchInto now drops stale _rows on start so an empty Auto search can't grab a
prior search's hit.
Soulseek-grab-only for now (same as the manual button); non-soulseek sources say
'no release met your profile' until that grab path lands. TV show view (separate
onShowClick, still stub searches) untouched. 7 wiring tests; node --check clean.
The video automations page had no way to CREATE automations and its card cog did
nothing — it called the music global showAutomationBuilder(), which swaps views
inside the (hidden-on-video) music page, so the builder 'opened' on the music tab
instead. Now the video side has its own builder.
- index.html: the video automations subpage gets its own list-view + builder-view
(vauto- prefixed ids) and a '+ New Automation' button, swapping exactly like the
music page. Save/Cancel/Back reuse the shared builder functions.
- stats-automations.js: the builder is now context-aware. A builder context holds
the element ids + blocks endpoint + owned_by + reload callback. Music context is
the default and byte-identical (all 17 id lookups go through _bEl() resolving the
music ids). showVideoAutomationBuilder() sets a video context (vauto- ids,
/api/video/automations/blocks, owned_by='video'); editAutomation() routes the
card cog to the right builder by active side. Opening clears BOTH builders'
canvases so cfg-* ids can't collide. Save tags owned_by from context and calls
the context's reload. A generic config_fields renderer/reader (video-gated so
music keeps its bespoke renderers) drives video block config like the mode select.
- video-automations.js: exposes window._reloadVideoAutomations so a save refreshes
the video list.
11 wiring tests (test_video_automations_builder.py); node --check clean; music
builder path unchanged.
conftest already redirected the MUSIC DB to a tmp path after test writes once
corrupted a user's music library (WSL/NTFS + WAL). The VIDEO side had the SAME
hazard but no guard: VideoDatabase()/get_video_db() with no path resolve from
VIDEO_DATABASE_PATH, defaulting to the real database/video_library.db, and its
enrichment threads WRITE — a default-path open during tests corrupted the real
video library (one table's btree pages; recovered via row-by-row salvage).
Fix: set VIDEO_DATABASE_PATH to the same throwaway tmp dir at conftest import,
before anything loads. Adds guard tests proving VideoDatabase()/get_video_db()
never resolve to database/video_library.db (mirrors the music guards).
The video side gets its OWN automations at music-side parity, kept separate so
nothing on the music side breaks. First twin: Scan Video Library — tells the media
server to rescan the user's SELECTED video sections (movies/TV, never music), then
reads the result into video.db so freshly-downloaded media shows as owned.
Architecture (scope tags + video twins on the shared engine):
- Handler core/automation/handlers/video_scan_library.py — pure function with
injected I/O (server_refresh / run_video_scan); production lazily binds
refresh_video_server_sections() + the video scanner. Owns its own progress.
Lives on the SHARED automation side so it may import core.video (isolation only
forbids core/video & api/video from importing music, not the reverse).
- blocks.py gains a 'scope' tag ('both' generic / 'video' video-only / absent=music)
+ blocks_for_scope(). The music /api/automations/blocks now filters out video
blocks; new isolated /api/video/automations/blocks serves the video palette.
- automation_engine seeds 'Scan Video Library' (owned_by='video', schedule 6h) so
it appears ONLY on the video Automations page; ensure_system_automations now
honours owned_by + action_config. Music page excludes owned_by='video' rows.
kettui: seam-level tests for every handler path (happy/no-server/scan-error/never-
raises/mode), scope filtering (music excludes video, video gets generics, music
parity preserved), seeding (owned_by + mode), registration drift guard. 39 new
tests; full automation suite (288) + isolation guards green.
The foundation for the 'Scan Video Library' automation (the video twin of music's
Scan Library). PlexVideoSource.refresh_sections() triggers a Plex scan on the selected
movie/TV sections (section.update()); JellyfinVideoSource.refresh_sections() POSTs
/Items/{id}/Refresh per selected video view (the GET-only _make_request can't POST).
Module helper refresh_video_server_sections() gets the active source and refreshes —
scoped to the user's chosen video libraries (Settings), so it scans the CORRECT media,
not music. Video-only, additive; isolation guard + (my code) ruff clean.
Per Boulder — the music system automations target music resources and don't belong on
the video side. The page now filters to owned_by='video' (a tag the upcoming video
automations will carry); none exist yet, so the System section is hidden and the empty
state reads 'Video automations coming soon — separate from the music ones'. The hub +
layout stay (video hub content + video automations come next). Balance clean.
The Automation Hub's Pipelines / Singles / Quick Start / Tips panes are music content
(playlist pipelines, music recipes/guides). The video side will get its own content
there; for now those four panes are emptied to a 'Video … coming soon' placeholder
(scoped to the detached hub element, no id clash). Reference stays — it's generic
automation reference. Hub tabs + structure otherwise unchanged. Balance clean.
Stopped hand-rolling the cards/section and now call the music page's GLOBAL builders
(_buildAutomationSection / renderAutomationCard / _buildAutomationHub from
stats-automations.js) — so the System section, every automation card, AND the
Automation Hub (Pipelines/Singles/Quick Start/Tips/Reference tabs) render byte-for-byte
identical to the music page. The System section uses a unique id (no clash) and is the
only thing re-rendered on refresh; the hub is built once (static). Run/toggle go through
the reused music card handlers; we re-sync the section after. Balance clean.
Was a bespoke .vauto- header; now mirrors the music automations page structure 1:1 —
.page-shell.automations-container > .dashboard-header (sweep + automation.png icon +
header-title/subtitle) > .automations-stats ('N Active · N System') > .automations-list
holding the protected '.automations-section.section-protected' System group (chevron +
label + count + collapse, persisted) with the same .automation-card rows. Reuses every
music class; driven by video JS via data-vauto-* hooks (no #id clash). Removed the dead
.vauto-* CSS. Balance clean.
Adds an Automations nav entry + page to the video side. The automation engine is
app-wide, so this surfaces the SAME system automations the music side runs — filtered
to is_system, EXCLUDING Refresh Beatport Cache and any user/playlist-pipeline ones.
Reads the shared /api/automations (no music imports), reuses the music .automation-*
card look, and supports run-now + enable toggle (system automations aren't editable).
Polls every 5s while the page is open. Frontend-only; balance clean.
Next: the library-refresh/scan automation wiring (Boulder has questions on how server
scanning works first).
The active-downloads poll kept running in the background — the page-change event that
stops it only fires for video-side navigations, not when switching to the music side,
so /api/video/downloads/active was being hit forever. poll() now bails (and clears the
timer) whenever the Downloads page isn't on screen (data-side!=video or the subpage is
hidden). Also matched the music page's 2s active cadence (6s idle). Still HTTP polling,
same as the music downloads page (which polls /api/downloads/all every 2s — the app's
SocketIO is for other realtime, not the downloads list).
When a grabbed release fails (transfer error / peer-cancel / never lands), the engine
now retries instead of giving up — the music-style depth:
- Grab stores the OTHER accepted results as a retry pool + the search context (schema
v16: candidates / search_ctx / tried_queries / tried_files / attempts).
- core/video/retry.py (pure, tested): plan_retry() → try the next-best candidate; when
the pool is dry, next_query() generates an ALTERNATE query (movie: drop the year; TV:
numbering variants) to re-search; budget MAX_ATTEMPTS=6. merge_candidates dedupes
against already-tried releases.
- Monitor: on failure, _fail_or_retry hops to the next candidate inline; if none, flips
the row to a new 'searching' state and a background requery thread re-searches the
alternate query, evaluates against the profile, and starts the best fresh hit — or
marks failed once truly exhausted. 'searching' rows are owned by their thread.
- Page: 'Searching' status (Trying another release…) + a 'Nx' attempt badge.
16 tests (retry engine + candidate-retry transition); ruff clean on touched files.
The poll capped at 32s but slskd results trickle in over ~50s (the music side waits
the whole search_timeout), so slow searches like 'Project Hail Mary' returned 'none'
before results landed. Now:
- /search/start returns poll_ms (slskd search_timeout + 8s); the UI polls that long
(capped 80s), streaming results as they arrive, stopping early only once results
clearly plateau (≥20s + stable) or hit 25.
- /search/poll returns total_files; when 0 video releases but slskd DID return files,
the panel says 'returned N files, but none are video — likely audio/other for this
title' instead of a blank 'none' (Soulseek is audio-heavy; many movie titles are
audiobooks there). slskd_search.poll_search() returns {hits, total_files}.
Tests green, ruff + balance clean.
Grabs now carry the movie's identity so the Downloads cards are rich, not anonymous:
- video_downloads gains media_id / media_source / year / poster_url (schema v15 +
migration); grab stores them (passed from the get-modal → download view → grab).
- Cards show the POSTER in the art tile (emoji fallback), 'Title (Year)', a quality
chip (1080p · BluRay · X265), and an ↗ Open button that jumps to the movie/show
detail page (dispatches soulsync:video-open-detail). Cancel/retry unchanged.
16 tests green, ruff + balance clean.
Scrapped the bespoke centered .vdpg- design and reused the music downloads page's
.adl-* classes for real parity: full-width .adl-layout, the segmented .adl-filter-pills,
the title with the accent download glyph, and the compact .adl-row (44px art tile +
.adl-row-info title/meta/error + .adl-row-status dot+label) — driven by the video JS
via data-vdpg-* hooks (no #id clashes with the music page). Per-row cancel reuses the
music hover-reveal .adl-row-cancel; retry mirrors it in accent. Kept the smooth
in-place patching (slim progress line for active rows). Removed the old .vdpg-* block;
balance clean.
Brings the Downloads page up toward the music page's depth:
- Filter tabs (music-style pills): All / Active / Completed / Failed, each with a live
count; clicking filters the list. Cancelled rolls under Failed.
- Header actions: Cancel all (active) + Clear finished, shown only when relevant; a
live 'N active · N done · N failed' subline.
- Per-row actions: ✕ Cancel on active rows (→ /downloads/cancel), ↻ Retry on
failed/cancelled rows (→ /downloads/retry, re-grabs the release).
- Cancelled status styling (pill + dimmed row). Still the smooth in-place patching
(no blink), adaptive polling, empty + filtered-empty states.
Balance clean. Phase C (auto-retry + alternate-query retry) next.
Fixes the 'cancelled but still shows running' stuck bug and adds real depth:
- classify_state now distinguishes 'cancelled' from 'failed'.
- Monitor is robust to slskd forgetting a transfer: if it's gone, it first tries to
complete from the FILE on disk (survives the music 'Clean Completed Downloads'
auto-clear), else counts misses and fails the row after ~8 polls instead of hanging
on 'downloading' forever. Cancelled transfers → cancelled status.
- POST /downloads/cancel (slskd DELETE transfer + mark cancelled) and /downloads/retry
(re-grab the same release). get_video_download(id) added; clear includes cancelled.
process_download stays pure (fs/slskd injected); 15 tests, ruff + guard clean.
Phase B (page: tabs/queue/history/cancel+retry buttons) next.
The old search did a 4.5s slskd search + 8s wait; slskd responses trickle in over
10-30s, so the window closed before results arrived (you'd see them in slskd but the
panel said none). Now it works like the music side:
- slskd_search.start_search() (uses the shared soulseek.search_timeout) + poll_responses().
- POST /downloads/search/start (mock = immediate; soulseek = returns a search id) +
GET /downloads/search/poll. Shared _evaluate_hits ranks each poll's hits.
- UI streams: starts the search, polls every 1.3s, renders results live with a
'searching…' badge, stops when results plateau (4 stable polls) or ~32s. Live
re-renders suppress per-card entrance so it doesn't blink.
Tests green, ruff clean.
backfill_next returns rows keyed by title/kind/id/imdb_id (no 'name'), so the
'Enriched movie None via Trakt' spam was just the log line reading the wrong key.
Now logs the real title in the 'enriched' line, the fetch-failed line, and the
current-item status. Cosmetic only — the Trakt sweep itself was working correctly
(one-time backfill of the movie catalogue; each row marked once, never loops).
The blink was every poll re-running each card's entrance animation via innerHTML
churn. Now cards are created ONCE and PATCHED in place (data-st attribute swaps,
progress width glides over a 1.4s transition, meta only rewrites when its text
changes), so nothing re-animates on a tick — the music-downloads smoothness.
Also: adaptive polling (1.5s while active, 6s idle), border/pill colours TRANSITION
on status change, downloading cards get an accent tint + glow, a relative 'started Xs
ago', and the % moved inline. Balance clean.
Phase 3 — the UX:
- Grab button now actually grabs (Soulseek results only — they carry the slskd
username/filename): POSTs /downloads/grab with the result + search context, shows
✓ on success, toasts 'Sent to Downloads', fires a refresh event. Endpoint returns
size_bytes for the payload.
- New Downloads PAGE (video-downloads-page.js, .vdpg-*): every grab lands here.
Polls /downloads/active while open and shows live status — type icon (🎬/📺/▶️),
title + release, a pulsing 'Downloading' pill with a shimmering progress bar, then
'Completed' with the file's library destination (→ /media/movies/…), or 'Failed'
with the reason. Clear-finished button, empty state, staggered entrance, vibes.
Reduced-motion honored. JS/CSS/HTML balance clean (verified w/ a real tokenizer).
Phase 2 — the engine:
- core/video/download_monitor.py: a daemon thread polls slskd for active video
downloads, updates progress, and on completion MOVES the file from the shared
download folder into the per-type library folder + marks it completed. The
per-download decision (process_download) is pure (fs + slskd injected) — 6 tests.
- POST /downloads/grab: validates (Soulseek-only v1), resolves the target library by
kind, starts the slskd download, records the row, lazily starts the monitor.
- GET /downloads/active (list + ensure monitor running) · POST /downloads/clear
(drop finished).
14 tests, isolation guard + ruff clean. Grab button + Downloads page next.
- Input/download folder is now the SAME shared config key the music side uses
(soulseek.download_path via config_manager) — change it on either side and both
follow, one physical download dir, simpler Docker mounts. Output libraries stay
video-specific in video.db. ZERO music code touched (read/write the shared key only).
- docker-compose: documented the shared ./downloads input mount, and added the three
video library OUTPUT mounts (./Movies→/media/movies, ./TV→/media/tv,
./YouTube→/media/youtube) matching the in-app placeholders.
- Test monkeypatches config_manager: asserts the input folder writes the shared
soulseek.download_path (music sees it) and is NOT in video.db; libraries persist to
video.db; legacy migration intact. ruff + guard + balance clean.
