v0.8.1

What's New?

This release is entirely for documentation and developer experience improvement.

100% Docusaurus and Self-Hosted Docs

While MS was already using Docusaurus for the docs website the prior approach to writing docs was to keep them as plain markdown as possible so they were still accessible without needing access to an external website. With the "plain" markdown docs you could potentially browse from your own markdown reader in a cloned repository.

The downside to this interoperability is that docs, specifically configuration, had grown to be very long and full of nested/too many headers for each iteration of env/file + regular sections for clients/services. Adding additional examples made the problem even worse.

The solution, then, was to make the Docusaurus documentation part of multi-scrobbler! Now the same docs found at https://foxxmd.github.io/multi-scrobbler can also be found at the /docs URL on your multi-scrobbler instance (http://yourHost:9078/docs)

This enabled me to lean fully into Docusaurus and all of the layout, styling, and organization it offers. Alongside general improvements new docs have been added to help onboarding. Some highlights:

• A Quickstart guide has been added for aiding onboarding
• Improved readability and usefulness of all content in Configuration docs
  • ENV, File, and all-in-one File configuration has moved to Tabs under each Source/Client
    • Switching config type syncs tabs selection across all sections and in the URL
  • Full configuration examples are displayed for file and AIO-file parsed from the examples in the config directory
• Improved styling across all docs
  • Copious usage of collapsed sections for examples to make scanning docs easier
  • Improved Table of Contents with only important headings
  • Admonition elements for increasing readability for tips/notes/warnings...
• Installation docs use Tabs for docker and docker-compose examples
• Actual development docs and tutorial (!)
  • Overview of common development elements
  • Full tutorial for creating a new Source
  • Moved flatpak instructions into docs

And many more...check out the docs site or click on Docs in your MS dashboard to see the updated site.

Flatpak Convenience Setup

A bash script for automating building flatpak from source has been added at flatpak/setup.sh:

• Removes node_modules and generates sources for flatpak-builder for both MS and the docsite
• Builds the flatpak app in a given directory

Effectively making flatpak builds a one-line process rather than the manual 7-8 commands it was before.

See flatpak build instructions in the new docs.

Full Changelog

Documentation

• (feat) Add architecture overview, common dev structures, and dev guidance for Sources
• (fix) Update package scripts to use npx to invoke docusarus
• (fix) Add missing docsite build steps to alpine docker variant
• (flatpak) Implement bash setup script and move flatpak instructions to docsite
• (No Category) Update docusaurus to v3
• (No Category) Update docusaurs playground schema for MS to latest
• (No Category) Improve TOC for docusaurus and github
• (No Category) Implement self-hosted docs
• (No Category) Replace manual notes with admonitions
• (No Category) Lean 100% into docusaurus documentation
• (No Category) Refactor installation.md for docusaurus
• (No Category) Update kitchensink.md code block titles
• (No Category) Parse and render config examples from repository
• (No Category) Add quick start guide
• (No Category) Use relative paths for volume mounting

v0.8.0

What's New?

Musikcube Source

Musikcube, a cross-platform terminal-based music player, is now supported as a Source.

Configuration Normalization

In previous versions many of the common, optional options for source/clients, such as maxRequestRetries or maxPollRetries, were mixed into the data property for configs alongside properties that were specific to that source/client. Additionally, there was an existing options property that already had some common properties, but not all...

jellyfin.json

[
  {
    "name": "default",
    "enable": false,
    "clients": [],
    "data": {
      "maxRequestRetries": 3,
      "scrobbleThresholds": {
        "percent": 30,
        "duration": 30
      },
      "users": ["MyUser"]
    },
    "options": {
      "logPayload": false
    }
  }
]

These common properties have been refactored into the options object so that data is now solely for data that is specific to the source/client. This should make configs easier to reason about as well as making programmatic config validity checks easier.

[
  {
    "name": "default",
    "enable": false,
    "clients": [],
    "data": {
      "users": ["MyUser"]
    },
    "options": {
      "logPayload": false,
      "maxRequestRetries": 3,
      "scrobbleThresholds": {
        "percent": 30,
        "duration": 30
      }
    }
  }
]

This is potentially a breaking change, behaviorally, but MS will continue to operate without any changes to your config. The options that have been moved are all optional, minor settings that help fine-tune your experience with MS but do not majorly affect how it works or functionally affect how your sources/clients behave.

If you use any of the properties that have been migrated but have not moved them then MS will simply ignore them until they are moved.

For a complete list of moved properties and how to migrate them see this comment.

Docker Image Change

Due to consistent issues with DNS in alpine-based images ( #160 #126 #134) the image base is being changed to Debian. The image size will increase relatively significantly ( 60MB -> 120MB ) but the benefit is a more stable experience for docker users. Thanks to @Ardakilic for helping diagnose docker image issues and extensive testing/usage of the debian image.

This is not a breaking change but is still a significant refactor. If you have issues with the new default image please open an issue. Alternatively, there will still be an alpine-based image published with a -alpine suffix so you can continue to use the original image if you would prefer, though this may go away in the future.

TLDR

Previously...

• latest or develop or 0.7.1 -> alpine based
• latest-debian or develop-debian ... -> debian-based

Now...

• latest or develop or 0.8.0 -> debian based
• latest-alpine or develop-alpine ... -> alpine-based and may go away in the future

Youtube Music Improvements

The monitoring behavior for YTM has been improved to prevent duplicate scrobbling as well as provide insight into YTM history issues. #156

Additionally, the authentication method for YTM has been improved to save updated cookies from YTM if/when they are changed which may help with authentication issues in long-running MS instances.

Full Changelog

Bug Fixes

• (chromecast) Add missing avahi-utils package for debian image
• (spotify) Convert URL to string for redirectUri
• (vite) Use default base string if no other path detected
• (webscrobbler) Add preflight route #157
• (webscrobbler) Fix formatting of slug url in logs
• (youtube) Catch auth issues that occur during polling #158
• (No Category) Refactor Base URL normalization to cover more cases #155

Documentation

• (No Category) BREAKING: Update schema
• (No Category) Clarify BASE_URL usage and add more examples #155
• (No Category) Add scripting on deploy for plausible analytics
• (No Category) Update screenshot

Features

• (client) Implement initial refresh count and config control #153
• (docker) BREAKING: Use debian image as default
• (youtube) Surface library history playlist and raw recently played for debugging #156
• (youtube) Sync cookies with those returned from YTM responses #158
• (youtube) Use full-fat list diffing to watch for new plays #156
• (No Category) Update normalize-url to prevent port erasure #155

Miscellaneous Tasks

• (ci) Update setup-node to v4 to squash node deprecation warnings
• (No Category) Npm audit fixes
• (No Category) Update dependencies

Refactor

• (client) Use same init logic as source
• (config) BREAKING: Move scrobble thresholds to options
• (config) BREAKING: Move source retry config to options
• (config) BREAKING: Move remaining common options out of source data and into options
• (config) BREAKING: Move common options out of client config and into new option property
• (No Category) Move all init logic into common abstract base class
• (No Category) Use @foxxmd/get-version to render app version

Ci

• (No Category) Use checked out repo to get git info rather than GH variables
• (No Category) Implement gated automated PR building
• (No Category) Add package cleanup and manual triggers

v0.7.1

What's New?

Apprise Notification

Support for notifying of scrobbles and errors via Apprise using an apprise-api server. See the docs for configuration.

Disable Web

The API and Dashboard can now be disabled. This can be useful for users who do not want a point of ingress in to MS on insecure networks.

Full Changelog

Documentation

• (No Category) Add config section and config example for disabling web server #150

Features

• (mpris) Less noisy error handling
• (No Category) Implement option for disabling web server #150
• (No Category) Implement notifications via Apprise

Miscellaneous Tasks

• (No Category) Remove some unused packages
• (No Category) Update schema
• (No Category) Bump version for release

Refactor

• (logging) Reduce noise for polling interval logging
• (logging) Balance debug/verbose level logging
• (No Category) Reduce usage of ErrorWithCause
• (No Category) Optimize imports
• (No Category) Use updated logger to reduce log file path logic complexity
• (No Category) Remove pony-cause dependency

v0.7.0

See the majority of changes in the RC1 release

New Since RC1

• Updated flatpak appstream metadeta to conform with new guidelines
• Fixed flatpak build manifest to work with new ESM module configuration

v0.7.0-RC1

This is a pre-release that makes some major changes to dependency versions and internal(ish) tooling but no changes to configuration or primary functionality.

What's New?

Improved Logging

Replaced logging solution, which was based on an outdated (100's of commits behind) fork of winston, with @foxxmd/logging:

• Built on top of Pino with modern code and fewer dependencies
• Adds coloring to all parts of the logs! Coloring is supported by console, docker logs, and UI
  • UI log coloring is rendered using existing log coloring (ansi)
  • Labels and error stacktraces are dimmer to make main messages stand out more
• (BREAKING) Log timestamp format has been changed to be more human-friendly IE 2024-03-21 16:18:35.520 -0400
• (BREAKING) The filenames for log files has been changed from scrobble-2024-03-22.log.X to scrobble-2024-03-22.X.log

Bumped Minimum Node Version

To support simpler testing with Mocha the minimum node version required for development and testing has been bumped from v18.16.0 to v18.19.1 in order to enable --import usage.

Bug Fixes and Improvements

Several bug fixes and general improvements are also included in this release. Specifically, improvements for subsonic sources and duplicate scrobble matching. See the full changelog below for all changes.

Full Changelog

Bug Fixes

• (flatpak) Add missing developer name
• (flatpak) Reduce summary length
• (webscrobbler) Fix missing wildcard to match slug route #137
• (No Category) Handle when no existing scrobble match with score greater than 0 #130
• (No Category) Fix ts schema generation import issue
• (No Category) Remove alt single quote when normalizing strings
• (No Category) Refactor static arrow function class fields to normal functions #143
• (No Category) Add comment for future typed linting

Documentation

• (No Category) Fix which host debian image is available for

Features

• (ui) Use ansi-to-react for log coloring
• (No Category) Add legacy fallback options for subsonic communication #136
• (No Category) Log more URL when route is unknown
• (No Category) Add initial linting config
• (No Category) Enable colorizing docker log output

Miscellaneous Tasks

• (!) Upgrade Node to LTS for --import support
• (docker) Remove unused dependencies
• (No Category) Remove unused es6-error package
• (No Category) Upgrade string-sameness
• (No Category) Bump mocha version

Refactor

• (logging) (!) Replace winston with @foxxmd/logging
• (No Category) Replace dbus-next with dbus-ts #142

Testing

• (No Category) Add test that handles no existing scrobble match with score greater than 0 #130
• (No Category) Use tsx through mocha config

Ci

• (No Category) Implement test checks for PRs and image publishing

0.6.5

What's New?

Upstream Recent

Prior to 0.5.0 MS let you see "recently played" directly returned from a Source's upstream API (for those that have it IE spotify, lastfm, listenbrainz...) This was useful for determining if the Source's API was "behind" what had actually been played -- for instance sometimes Spotify's "recently played" can be delayed by minutes/hours which could result in backlogging not returning what you "know" was played more recently.

This is entirely a convenience for the user, rather than new functionality for MS, but had been missing as a feature. This release returns this feature to the frontend.

UI Improvements for Source-of-Truth

For some sources (Lastfm, Listenbrainz) MS always uses the recently played data as the source-of-truth for scrobbling to your clients, rather than MS's "player". This is due to information returned for "now playing" for these sources being less accurate than recently played data. This has functionally not changed but this discrepancy is now more clearly presented to the user in the UI via a ? tooltip icon.

Google Cast Stability Improvements

I have rewritten parts of the third-party library providing google cast (chromecast) functionality to be more stable and less prone to crashing the entire MS application. Reconnecting events should now be more reliable as well.

Last.fm Timeout Fixes

An issue causing requests to Last.fm to hang forever has been fixed and will now cause the request to timeout in a reasonable amount of time instead of freeze up MS #134

Existing Scrobble checks for non-english characters

A bug causing scrobbles with non-english character-symbols to potentially not match existing scrobbles from client has been fixed. #121

Docker Image Alternatives

An alternative debian-based docker image is now provided for x86/x64 host architectures through a *-debian flavored tag. This image may be used as a workaround if you experience sporadic networking issues with the regular docker images. #126

Full Changelog

Features

• (chrome) Used improved library
• (lastfm) Implement nowPlaying lfm api call
• (source) Implement recent from API
• (sources) Improve Source of Truth usage and presentation to user #134
• (No Category) Add debug logging for user api requests
• (No Category) Improved logging for last.fm api network errors #115
• (No Category) Implement debian-based image #126

Bug Fixes

• (frontend) Fix scss division deprecation warnings
• (lastfm) Patch lfm library to actually implement a request timeout #134
• (scrobbler) Fix erasing non-english characters #121
• (No Category) Add patch-package as postinstall script so packages actually get patched

Documentation

• (No Category) Add workarounds for rollup issues and flatpak disclaimer
• (No Category) Add debian docker image network workaround to docs #126 #134

Miscellaneous Tasks

• (No Category) Remove some unused packages
• (No Category) Ignore apple cert extensions
• (No Category) Make container image build tags agnostic to default branch name
• (No Category) Add matrix for building debian flavor tag

v0.6.4

What's New?

Google Cast Source

Support for Google Cast devices as Sources. This is particularly exciting because music platforms that do not have an API for exposing user Now Playing info (Pandora, Tidal, MixCloud, Soundcloud) can now be monitored if you play them through a Google Cast device. Additionally, since the Cast platform exposes generic Now Playing information virtually any app that plays music through Casting can now be monitored by MS.

Application tooling restructure and frontend platform migration

This is no functional change for end-users. This is an internal/development refactor but brings with it many benefits:

• Refactored project from cjs to esm (new module standard)
• Moved away from dead frontend platform (CRA) to Vite
  • Removed dependency on webpack
  • Removed api proxy required by CRA
  • Simplified app entry -- Vite is served through express middleware instead of having nodemon spawn separate processes for front/backend
• Refactored project to use tsx runtime wrapper -- no longer need to compile to js to run dev/prod
  • Side-effect of this is stack traces reported in prod now point to actual source files with correct lines numbers

The only visible changes are:

• a slight increase in docker image size due to bundling typescript/source files
• developers do not need to use API_PORT anymore -- use PORT for both dev/prod now.

Source init improvements

All sources have been refactored to have more granular init stages:

• build data -- any data that needs to be parsed (URLs for mopidy), built, or verified (credentials) happens here
• test connection -- if the source can be checked for reach-ability it is done here
• test auth -- if the source requires auth it is tested here

If any stage fails more descriptive errors are logged for the failure and the stage it failed at is also described, giving users more insight into why a source failed to start as well as improving debugging.

App Crash Logging

App-crashing errors should now be logged before the app crashes. The easiest way to capture these logs is to set file logging to warn log level so they are stored to file. In your file-based config.json`:

{
  "logging": {
    "level": "debug",
    "file": "warn"
  }
  // ...
}

Deprecations and Future Changes

Maloja

• Support for Maloja versions below 3.0.0 is now deprecated
  • Likely no one is using these versions as they are almost 2 years old but MS still supports them with additional logic that could (should) be simplified
• A warning is emitting if Maloja version is below 3.2.0 due to lack of support for scrobbling album information

Spotify Credentials

A future release will include a migration from the current spotify api library to the official spotify web sdk. The information stored for Spotify credentials in releases prior to 0.6.4 is insufficient and will require reauthentication by the user if, for example, the user upgrades directly from 0.6.2 -> 0.6.5+.

0.6.4 will store all required credential information needed for future releases. Simply upgrade to this version and wait a few hours for Spotify credentials to be refreshed.

Full Changelog

Features

• (chrome) Improve mdns discovery for docker
• (chrome) Implement user filtering media by type
• (chrome) Allow manual device connections
• (chrome) Refactor mdns discovery to run on source heartbeat as well as startup
• (chrome) Catch buffering media
• (deezer) Fix credentials check during build data stage
• (maloja) Improve error handling
• (spotify) Prepare for future spotify lib migration by writing completing credential data
• (subsonic) Improve error/response parsing and report errors correctly on startup #129
• (No Category) Improve discovery for network exceptions in error causes
• (No Category) Improve data init stage log wording
• (No Category) Log when heartbeat tasks start

Bug Fixes

• (No Category) Actually throw auth failure so initialization fails correctly
• (No Category) Fix isReady logic
• (No Category) Fix status indicator color based on status text
• (No Category) Fix status text based on initialization OK values
• (No Category) Fix return values for source init startup
• (No Category) Fix when to check connection
• (No Category) Fix where retries are checked
• (No Category) Better handling of uncaught exceptions and rejections

Documentation

• (No Category) Chromecast docs and schema
• (No Category) Add local source reference to flatpak manifest

Miscellaneous Tasks

• (chrome) Switch to scoped chromecast-client to fix flatpak source generation limitation
• (maloja) Deprecate support for versions below 3.0.0

v0.6.3

Changelog

Features

• (player) Handle state transfer for single-platform sources #124
• (player) Improved player deletion behavior
• (scrobbler) Change fuzzy diff threshold to be inclusive
• (scrobbler) Use play data to extract non-joined track name when comparing for duplicate check
• (scrobbler) Improve log readability for duplicate checking
• (scrobblers) Implement user-initiated (re)start mechanism #114
• (source) Improve (re)start behavior and ui
• (webscrobbler) Log the best-guess URL for requests using configured slug
• (No Category) Implement play completion context #121
• (No Category) Improve track credits parsing to recognize post-feat suffixes #121

Bug Fixes

• (jellyfin) Split musicbrainz mbid artist identifier semicolon #114
• (lastfm) Fix handling of non-string mbid string values #114
• (listenbrainz) Treat 400 responses as non-showstopping #114
• (mopidy) Fix album artist parsing #110
• (player) Fix listen range default value
• (spotify) Undefined player state data when in private session - thanks @brunohpaiva
• (webscrobbler) Correct ENV names to match those documented #111

Refactor

• (scrobblers) Consolidate scrobbler client startup surface area
• (source) Start consolidating isReady behavior
• (No Category) Improve string comparison robustness
• (No Category) Move Play time comparison code into own file
• (No Category) Temporal comparison result data improvements #121

Documentation

• (webscrobbler) Add where to find connector list #112
• (webscrobbler) Move slug into data #111
• (No Category) Update schema for logging enums
• (No Category) Add contributing guidelines and PR template
• (No Category) Correct json-schema.app links - thanks @Bujiraso

Miscellaneous Tasks

• (tests) Fix TS compile warning on test for bad data

v0.6.2

What's New?

Album Artists

MS now tracks Album Artists and correctly scrobbles this info if the client supports it.

Targeted (Smarter) Polling for Play Durations #108

For polling sources that report track duration (Spotify), MS will now decrease polling interval near the beginning and end of the track in order to get more accurate "listened duration" and track start timestamp. NOTE: Spotify's Automix feature may interfere with this.

Changelog

Features

• (player) Pad scrobble listen duration when confidence is high #108
• (scrobbler) Use listenedFor when checking for existing scrobbles
• (source) Smart interval calculation based on track position and end time #108
• (No Category) Allow disabling console and file logging through ENV
• (No Category) Refactor auth testing to differentiate between network issues and actual auth errors #102
• (No Category) Implement album artist and scrobble client payload function
• (No Category) Implement album artist for all sources #104
• (No Category) Add app version output

Refactor

• (No Category) Simplify network error detection during auth test
• (No Category) Don't stringify error when using upstream options

Bug Fixes

• (lastfm) Fix api error when listening history is private by sending session key #105
• (maloja) Fix missing return value for auth test
• (player) Correctly output value of play timestamps for player based on undefined or not #107
• (plex) Use album artist if present and remove "Various" artist #104
• (source) Fix polling interval drift
• (spotify) Improve api error handling #103
• (No Category) Test auth correctly on auth flow completion
• (No Category) Catch errors during source backlog processing #103

Documentation

• (spotify) Add note about duration accuracy when using Spotify's Automix feature

Miscellaneous Tasks

• (flatpak) Update metainfo and sources
• (No Category) Update string-sameness version

v0.6.1

What's New?

Webscrobbler Source

The Webscrobbler browser extension has been added as a source using its native webhooks behavior.

Display and Controls for Queued/Dead Letter Scrobbles

Quietly implemented in 0.6.0, MS has two new control structures for making scrobbling more resilient to offline clients and network issues:

• Queued Scrobbles -- When a Source discovers a new Play it is added to a queue for each Client that should scrobble it. If the Client fails to scrobble due to networking or "showstopping" errors from the downstream API then it stopped processing the queue until the next heartbeat. This ensures all scrobbles are preserved until the downstream service is ready to accept scrobbles again.
• Dead Letter Scrobbles -- If the downstream service fails to scrobble for reasons that are expected by MS (EX Last.fm's "ignored" response) then the Scrobble is added to the Dead Letter Queue, a queue of failed scrobbles.
  • The queue is retried a couple times on heartbeat to account for ephemeral issues downstream.
  • If scrobbles still fail the MS user can at least see what failed, when, and why. The queue display in the dashboard also allows users to retry scrobbling manually.

Changelog

Features

• (No Category) Implement WebScrobbler source
• (No Category) Implement dead letter endpoints
• (No Category) Add dead letter error to data
• (No Category) Log storage improvements and limit controls
• (No Category) Implement retry/remove all dead letter scrobbles
• (No Category) Implement live updates for dead letter and scrobble counts
• (No Category) Implement live client/source status updates
• (No Category) Implement queued scrobbles display

Bug Fixes

• (No Category) Deadletter endpoint responses

Documentation

• (webscrobbler) Add disclaimer about hostname for firefox
• (No Category) Add queued scrobble feature highlight

Refactor

• (No Category) Reduce noisy logging at INFO level