Two fixes: the ChannelList local-AI warning kept showing after the
runtime became available because status changes were not re-evaluated
on refresh; and selecting a language in the player did not switch the
stream variant immediately. Refresh warning visibility from the
latest laya status and apply the chosen variant/language on
selection, with tests for both behaviors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Offer the right actions in Settings when Laya is ready/available
instead of leaving irrelevant ones, and stop showing the ChannelList
local-AI warning once the runtime is actually available. Derive both
from layaStatusPresentation and cover the ready/available action set
with tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The model was reported as ready while the AI runtime was actually
unreachable, and the Download action stayed visible afterwards.
Distinguish model-downloaded from runtime-available in
layaStatusPresentation, derive the Download action from the real
combined state, and align Settings and ChannelList rendering on that
single source of truth. Cover the combined status cases with tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Track download progress in layaService (bytes/total, phase, error
state) and expose it through the laya status route. Render a progress
bar during model download in Settings via a layaStatusPresentation
service, and display download/status errors inline instead of failing
silently. Cover progress tracking and status presentation with tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a languageDetectionMode setting (algo | laya | laya+algo) exposed
in Settings with three engine choices. When Laya is enabled, a laya
service (Dockerfile.laya, optional compose profile, models stored
outside the repo under ./data/models/laya) can be started, with model
download and deletion from the UI via /laya routes to manage disk
space. Add a channelLanguageFusion service combining Laya results
with the existing algorithm, and wire detection mode through
grouping, channel list and player. Cover engine selection, fusion
and laya service behavior with tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a liveChannelListMode setting (default 'grouped') with a Settings
toggle to display the Live TV channel list flat, without category
group sections. When enabled, logical channels are listed directly
with their stream variants still grouped as in #9/#10. Cover the
setting default and flat-list behavior with tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a channelLanguageDetection service that scores language evidence
per IPTV variant (country/language prefixes in name and group, source
metadata, original title words) and picks the highest-scored language.
Integrate detected languages into channel variant grouping and add a
language selector to the live player to switch to a variant of the
logical channel in the desired language, keeping raw streams
non-destructive.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
docker-compose.yml referenced only the Gitea registry image, so
`docker compose up --build` ignored local source and never shipped
the channel variant grouping feature. Add a build section (root
context, local Dockerfile) alongside the registry image tag so
compose builds local code while keeping the published image name.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a channelVariantGrouping service that normalizes channel names
(strips region prefixes and quality tokens) to fingerprint duplicate
stream entries and cluster them under a single logical channel with
quality/source variants. Wire it into the Live TV channel list and
player variant selection, reducing duplicates without destructively
merging streams.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Manual quality profiles (source/1080p/720p/480p) in the player overflow
menu plus an honest/preparatory Auto contract: adaptive when the HLS
manifest exposes multiple levels, single-rendition automatic selection
otherwise.
- server: transcodeQualityProfiles service and qualityMode session option
- player: quality menu, persisted preference, live switch with rollback
- tests: npm test script and unit tests for quality profiles
Was running into an issue where epg wouldn't process passed the first batch. Refactored the batching to use an array and added backpressure to keep memory usage low.
