#360

#360

#355

#360

feat: Caching and persistent scrobble queues

Fixes #356

Faster interval means less likely user will update queue and exit MS before cache is saved

Should be ~700kb, full up. This should be a generous upper limit for number of scrobbles queued/dead at the same time.

If cache files get large enough to cause noticeable delay it will be helpful to log progress

tmp dependency has been removed so previous buggy behavior(?) is gone

* Use typeson for stringified data marshalling
* Use clone for in-memory serializing to preserve dayjs
* Refactor cache backend init into indepedent, testable functions
* Add `false` as cache provider type to allow disabling secondary cache

* Update package to improve typing support/esm
* Remove dependency on sources/client/notifiers to prevent circular dependency

* Two-tier caching
* primary is always in-memory LRU
* secondary supports file and valkey
* Different caches for metadata and scrobbles
* no file support for metadata (too big)

feat: Add Access Token auth for Ntfy webhook

fix(lastfm): Full destructure recent response and log respon on error

Maybe recenttracks/track does not exist for totally new accounts, though it should.

Guard against this and log response if something unexpected happens so we can troubleshoot

#353

Youtube Music updates

Fixes LOG_LEVEL non-output issue

testing and improvements for artist string parsing

feat: Improve Player behavior for Sources only tracking Now Playing data

* show "Now Playing" instead of Unknown/Playing
* Don't show empty listened duration/percent

Not really needed since koito has its own docs

#338

feat: Add Maloja Source

Don't want to break 3p libraries (dayjs)

* Don't split artist with joiner when only one joiner is present
* Override naive parsing if metabrainz mapped artists fit with provided joiners

* Use more context-aware list parsing with ampersands
* Ignore artists with slashes when wrapped by word-boundary