All versions since 0.18.1
0.18.1
🐛 Bug Fixes
- (plex) Re-add missing component artwork api route #718
- (ui) Add duped to common play state so its available for source play filtering
0.19.0
This is huge release that brings a ton of new Play/Scrobble modification options to MS!
What’s New?
Simpler Config For Teal/Rocksky Sources
App Password has been removed from required config for Teal.fm and Rocksky Sources. It still required for Clients (scrobblers).
Simpler Config for Musicbrainz Transformer
Due to the lack of required config since the change to the Musicbrainz API in 0.18.0, explicit setup of the Musicbrainz transformer is no longer required. It is now built-in to MS with a default configuration that can be used without needing to specify apis or anything in your AIO/ENV config.
You can still configure rate limiting for the default server, or add additional servers, using the same config. It’s just not required to explicitly set anything anymore.
Example
If you currently have this in your config.json
{ // ... "transformers": [ // your empty musicbrainz transform { "type": "musicbrainz", "name": "MyMB", "data": { "apis": [ { "enable": true } ] }, }, // ...maybe more transforms here ], // ...}You can remove the musicbrainz object entirely. If it’s the only entry in your transforms array then you can remove that too.
Then, in your Source/Client transform hooks change this
"playTransform": { "preCompare": [ { "type": "musicbrainz", "name": "MyMB" } ] }to this
"playTransform": { "preCompare": [ { "type": "musicbrainz", "name": "MSDefault" } ] }or entirely remove "name": "MSDefault" and MS will use whatever is configured first.
Rocksky PDS Client
The Rocksky Client has been rewritten to use the sdk agent for scrobbling rather than scrobbling through the Rocksky server.
This change, along with using the new Rocksky matchSong metadata API, means Multi-Scrobbler can now:
- give you immediate feedback on when a scrobble does not meet Rocksky’s standard of quality and
- directly write scrobbles to your PDS repository
This is a huge improvement in user experience over the old “submit scrobble to rocksky server and find out later if it was accepted” approach. Use this in conjunction with the new Rocksky Transformer to get (essentially) guaranteed scrobbles to Rocksky with verifiable writes.
An additional benefit of the PDS client is that MS now supports Now Playing for Rocksky.
Rocksky Transformer
A new scrobble enhancement stage is available, the Rocksky Transformer.
This stage uses Rocksky’s metadata API to match multiple providers (deezer, spotify, etc…) and returns an enriched metadata result that Multi-Scrobbler can use to add Musicbrainz Recording MBID, ISRC, artists, etc… similar to how the Musicbrainz Stage works.
Rocksky’s metadata API is still a work-in-progress so the resulting transformation is limited for now, but it is extremely fast and provides good results!
Use this in conjunction with the Rocksky PDS Client to guarantee scrobbles for Rocksky or as a fast-and-cheap alternative to Musicbrainz.
Spotify Transformer
A new scrobble enhancement stage is available, the Spotify Transformer. All of the research, config options, and initial implementation was contributed by @DigitalStarSys :tada:
This stage works similarly to Rocksky and Musicbrainz transformers in that it is useful for fixing/enriching your scrobble data.
It is advantageous to use, over Musicbrainz, if you prefer a simpler setup with fast API responses and formatted data that is familiar to a mainstream audience (vs. having to configure Musicbrainz and filter through all the different release types).
Note: This transform requires a premium Spotify account (same as the Spotify Source) to use the Spotify API.
Cover Art Archive Transformer and Art Transforms
A new scrobble enhancement stage is available, the Cover Art Archive Transformer. This is standalone version of the Cover Art Archive functionality found in Discord.
When this Stage is used, if your scrobble data has Musicbrainz’ MBIDs for album or release-group and is missing album art, MS will use the CAA API to fetch an album art URL.
Art In Other Transformers
As part of this improvement, both the Rocksky and Spotify Transformer can also populate album art URLs when they are missing. This is enabled automatically as part of Stage Rules and can be explicitly disabled with a rule "art": false.
Device Info for Listenbrainz-Compatible Scrobblers
Thanks to @rpschube reported device info from Sources can now be safely submitted as scrobble data to listenbrainz-compatible clients (Listenbrainz and Koito) without leaking any PII by using an allowlist of scrubbed identifiers to match with real device ids.
See the Scrobbling Device Data expandable section on Listenbrainz and Koito clients to see how this works.
Reduced Docker Build Size
More cleanup done after 0.18.0 revealed a large number of files in docker file build layers that could be easily pruned. A minor rewrite of the Dockerfile has eliminated these and improved build times as well. The cleaned up docker image is 20% smaller than 0.18.1.
Typescript Strict Mode
The entire project has been refactored to compile with typescript strict mode on. This “enables a wide range of type checking behavior that results in stronger guarantees of program correctness” which in laymen’s terms means that ongoing development of Multi-Scrobbler is less likely to introduce new bugs. Just moving to strict mode fixed at least 30+ undiscovered bugs!
Dynamic Transformer Loading and Usage
The logic that loads code for tranformers has been refactored so that all transformers and their dependencies loaded just before a Source/Client actually uses them instead of on startup. This should result in less memory usage for all users but especially those that don’t use any transformers at all.
Additionally, the names of built-in transformers and the order they are used in has been changed to be deterministic and give you more options for which you would like to use, between built-in, ENV configured, or file/aio configured. See the Breaking Changes section below for these changes.
Breaking Changes
Rocksky Auth API Key Removal and Token Deprecation
- The previously-deprecated API Key authentication for Rocksky has been removed
- Access Token authentication for Rocksky is now deprecated in favor of handle/app password auth that uses the new functionality from above.
- Review the Rocksky Client docs for how to set this up
Discord Cover Art Archive Deprecation
The automatic Cover Art Archive usage currently found in the Discord Scrobbler is now deprecated and should be replaced with a Cover Art Archive Transformer.
Example
For Discord Scrobbler with ENV add this to your environmental variables:
DISCORD_TRANSFORMS=coverartarchiveFor File/AIO add the example playTransform block to your config options (or just add the coverartarchive stage config if you already have hooks configured):
{ "name": "MyDiscord", "data": { /* ... */}, // add this "options": { "playTransform": { "preCompare": [ { "type": "coverartarchive", } ] } } }A new Kitchensink example shows a more complex example, here.
Transform Names and Fallbacks
Transform stages are specified by type and name when used in a Source/Client, however the name is optional.
Example
[ { "name": "MySubsonic", "data": { /* ... */}, "options": { "playTransform": { "preCompare": [ { "type": "musicbrainz", "name": "MyMB" // name is optional } ] } } }]Previously when name was omitted MS used the first non-default transformer, then tried the default.
Now, names and order of selection (when name is omitted in configuration) looks like this:
- MS looks for a non-default, non-ENV config first
- MS falls back to ENV config (name is
MSEnv) if it exists - MS falls back to default config (name is
MSDefault) if it exists
Steps to migrate:
- If you only use
named configs that aren’tMSDefaultthere are no changes necessary - If you only use ENV configuration there are no changes necessary
- If you use the
nameMSDefaultbut do not mix ENV transforms with AIO/File config, no changes are necessary - If you use the
nameMSDefaultand also have ENV transforms configured (musicbrainz)- Rename
nametoMSEnv
- Rename
See more about Stage Names and Defaults in this new section in the docs.
Logs Directory for Non-Docker Users
More attention has gone into the Data Directory Standardization work that began in 0.15.0. Now, the directory logs are written to has been standardized, alongside the existing data and config directories.
This change only affects non-docker users.
The logs directory is now determined by, in order of priority:
LOGS_DIRENVLOGS_DIRECTORYENV (used by systemd)- OS-specific standardized path
You can see OS-specific paths here in the documentation.
If you are running multi-scrobbler locally with nodejs you’ll likely want to set LOGS_DIR=./config/logs
Other Changelog Highlights
- fix(endpointlz): Missing ‘import’ enum for listen_type on submit payload #732
- fix: Update dbcomponent at runtime when name changes #721
0.19.1
Changelog
⛰️ Features
- (musicbrainz) Disallow audiobook and music videos with config options #734
🐛 Bug Fixes
- (transform) Add missing musicbrainz default transform #737
0.19.2
Changelog
🐛 Bug Fixes
- (transform) Don’t try to initialize transformers that are already initializing, reduces noisy logging on startup
- (transform) Fix async transform initialization race condition and type/name key case sensitivity #739
v0.20.0
What’s New?
Play Editing and Search
Plays can now (finally) be edited! If your scrobbles fail you can now edit the data to correct it manually before resending it.
A plain text mode is initially used where you can edit everything related to fields and metadata ids
Additionally a search mode is available to search all of the metadata providers you have setup. Musicbrainz, Spotify, and Rocksky can all be searched concurrently. Selecting a result fills in that field, any associated fields (track can set artists + album), and meta ids.
Musicbrainz and Rocksky providers are automatically setup. All providers automatically inherit from your existing transformers. You can also set up preferred providers to only use some of them, or use with alternative configuration.
See the docs for more information.
Tuneshine Client
Tuneshine, a wifi-connected hardware device with an aesthetic, low-count LED display for showcasing the album art of the music you are currently listening to, is now integrated as a Scrobble Client.
Like Discord, it can be used to display Now Playing activity from any of your Sources.
Mixcloud Source
Mixcloud is now integrated as a Scrobble Source thanks to @anandghegde
This source records the mixes you listen to as scrobbles. It supports both free and Pro Mixcloud accounts.
Metadata Overhaul
Previously, metadata like track identifiers from Spotify/Musicbrainz/etc… and album art, were stored alongside a Play but they were not strongly associated with specific parts of the Play. This lead to situations where some functionality would require assertions like “Artist 1 has metadata artist ID 1” but it was not possible to guarantee this.
Now, Play data has been restructured to tightly associate all of this metadata to the field it belongs to. This change does not affect any end-user functionality but it dramatically simplifies developer experience. The Play Editing functionality above would have been much more difficult without this change, for example.
Changelog Highlights
⛰️ Features
- (endpointlz) Allow disabling artist extraction from ingress payload #750 #751
- (deezer) Add ISRC and optional title variant to parsed data #752 #754
- (koito) Add includePlayPosition to koito scrobbler implementation #741 #742
- (spotify) Periodically backfill songs from spotify history api for missed live playback state #735 thanks to @sudbrack
- (No Category) Consolidate SSE for logs into events to reduce connection pool usage for HTTP/1.1
- (No Category) Consolidate global SSE connections with coordinator
🐛 Bug Fixes
- (docs) Resource static asset link broken by trailingSlash #573
- (listenbrainz) Only use original artist_name in naive play if no artists credits were extracted #750
- (stats) Persist lifetime discovered/scrobbled counts across restarts #745 @dylangovender
- (No Category) Improve detection of repeat plays when listen range is close to end of a position player
📚 Documentation
- (discord) Update kasmweb/discord service example
- (metadata) Add editing and metadata search
- (No Category) Add reverse proxy recommendation and tips
🚜 Refactor
- (No Category) Decouple transform manager init
- (No Category) Circular dependency elimination
⚙️ Miscellaneous Tasks
- (No Category) Add docs and devcontainer modifications to ease discord IPC development with kasmcord
v0.20.1
This release bumps the Nodejs version used in the docker image to 24.22.1 in order to include the october security release fixes in the distributed docker image.
You should upgrade as soon as possible in any case, but users who run MS in a way that exposes it to the public internet (API or UI accessible from a non-authenticated URL) are particularly vulnerable as this fixes an exploit for remote code execution. Do not delay upgrading if you are in this scenario.
v0.20.2
This is another security fix that removes compromised eMudhra certificates from the base docker image. #759
v0.20.3
What’s New?
Docs Moved to Starlight
Documentation has been migrated from Docusaurus to Astro ✨Starlight✨
Docusaurus has been great but Starlight, and Astro being a general static site build system, is more flexible and lighter. Along with the already-strong community plugin support it has enabled more functionality for the MS docs that was not possible with Docusaurus.
The docs now have a fresh look with clearer/more readable styling. Particularly awesome new features:
- Code blocks get walkthroughs, footnotes, inline linking, stylized focus, dynamic placeholders, and much more thanks to starlight-codeblocks
- Changelogs are now built into the public site
Additionally, the developer experience for docs is greatly simplified. Same build commands but many things, like og image generation and open api doc generation, are all built-in now.
Improved Shutdown Behavior
MS now handles all exit signals (SIGNIT and SIGTERM) instead of just SIGINT. On exit signal it now tries to gracefully stop all Source/Clients by letting all currently processing Plays/scrobbles finish before closing the database and shutting down.
v0.21.0
What’s New?
This is a huge release that hardens Multi-Scrobbler deployments so that you can use it in a public environment.
Included in this release:
- A new, hardened Docker Image
- Single-User Authentication
- Public Dashboard :eyes:
- Improved Service ID UI
Hardened Docker Image
Alongside the Authentication introduced below, MS now uses a rebuilt-from-scratch, hardened, distroless docker image.
This image entirely removes root and shell functionality, includes only necessary dependencies to run MS (basically no OS), and defaults to running the container as a non-privileged user. These changes make it much harder for a malicious actor to compromise the container, use any kind of exploits, and gain a privileged foothold in your host system.
- The drawback of this change is that there are breaking changes for your compose stack related to container and file/folder ownership.
- The good news is that these changes are relatively straightforward and for the majority of users should be only a few changed lines in their compose file.
Single-User Authentication
I have planned features that will allow more modification of MS and downstream scrobble data. However MS (until now) still did not have any kind of authentication which puts users in a hard position of exposing their instance to malicious actors or potentially losing connectivity to some Sources. This would have been made worse by the planned features. So the time has come…
Multi-Scrobbler now supports single-user authentication using Better-Auth :tada:
Upgrading and Initial Setup
I have gone to great lengths to make sure upgrading is painless.
When you upgrade, before even setting up auth, your MS instance will continue to work exactly as it does today for backend, Sources, Clients, polling, etc…the only thing that changes is how the web dashboard is initially accessed.
Additionally, nothing changes for your Ingress Sources, like Listenbrainz Endpoint, even after setting up auth. Any external clients that use these Sources will continue to work as-is.
Use the 0.21.0 Upgrade Guide in the docs for upgrading/setting up Auth.
Api Keys
Since your instance’s API will now be authenticated, you can now use API Keys found in Settings -> Account to generate keys for use with external services (Prometheus, other apps that want to consume MS apis).
Users with Existing Auth Solutions
If you have already secured MS with your own auth like a reverse proxy + authelia/authentik/keycloak etc.. and are happy with that there is an escape hatch for you:
To disable all auth and have MS behave (almost) exactly as it does pre 0.21.0, use Bring Your Own Auth (BYOA) mode by setting the ENV AUTH_MODE=disabled. This is also covered in the upgrade guide.
Data Directory Usage
Standardized application directories for storing config, data, and log files separately were introduced in 0.15.0. However, many of the cache/auth flow files used by MS are still stored in the config directory.
In 0.21.0, MS now looks for these files in the data directory.
Public Dashboard
As a carrot :carrot: to entice you to (please!) upgrade, this release also include a Public Dashboard that unauthenticated users can view when visiting your secured instance.
This is a strictly opt-in feature and you control what (privacy-restricted) data is shown and how often it is shown. Find this feature’s setting in the logged-in dashboard in the menu Settings -> Publish Dashboard.
Some highlights:
- Animated stats for total number of Plays, Scrobbles, and minutes listened
- Animated stats/icons for number of Sources/Clients and what types you are running
- A configurable Recent Scrobbles lists, optionally with a Now Playing card
- An embeddable/iframe page so you can show off your activity and stats on your blog!
Service ID Indicator Improvements
Service ID/Icons got a glowup with a new, animated stacking style, merged icons when multiple of the same id exist (like Musicbrainz release and release-group), and a popup to make ids/links acessible to screen readers, keyboards, and mobile touch navigation.
Other Changelog Highlights
🐛 Bug Fixes
- (spotify) fixed regression that dropped
localeparam in search requests #768 thanks @DigitalStarSys - (listenbrainz) fixed incorrectly parsing forward slashes in artist names #760
📚 Documentation
- Added new Security section for authentication guidance
- Reworked usage of Base URL everywhere so it uses placeholders
⚙️ CI/CD
- Rewrote Github Actions to remove reliance on
pull_request_target- Publishing PRs from external contributors now requires an explicit comment with commit SHA each time it should be built
- Moved testing workflow into a docker image build step to improve security and leverage already built layers
v0.21.1 Latest
What’s New?
Blackbox Client
A new Client, Blackbox, is available that can be used as a glorified “webhook” for scrobble/now playing activity from MS.
This client can be configured to POST a json payload (scrobble and/or now playing) to any URL you give it. It can optionally be configured with headers for authentication.
The payload can be either a Full MS Play object which contains a ton of information, or a normal Listenbrainz submit-listens payload.
This client is generally useful to feed your own app, script, or home automation with listening activity from any of your Sources.
Date Time Formatting on Dashboard
MS now defaults to using the date/time format associated with your browser’s locale instead of hardcoding it.
Additionally, you can now also explicitly set preferences for these in the MS dashboard under Setting -> Appearance
More Security Hardening and Dependency Updates
Select dependencies have been updated to fix long-standing vulnerabilities due to these dependencies/packages being mostly unmaintained (but still used by MS like for chromecast).
Additionally, dependencies made it into the 0.21.0 production image that were not used. These are now manually removed when building the image.
The result is a ~25MB drop in image size and 90% reduction in vulnerabilities reported by dockerhub image scanning.
Changelog Highlights
🐛 Bug Fixes
- (lastfm) Parse artists with an mbid present as one artist (don’t parse by delimiter)
- (listenbrainz) Prevent
duration_played: 0#781 - (tests) Prevent test artists generator from duplicating artists
- (tests) Fix client config tests accidentally clearing non-test related ENVs during the run
- (No Category) fix LOGS_DIR being ignored
⛰️ Features
- (ui) Date and time formatting preferences
- (No Category) Implement blackbox scrobble client
⚙️ Miscellaneous Tasks
- (deps) Upgrade drizzle-orm and drizzle-kit to 1.0.0-rc.5, remove unnecessary drizzle-kit prod dependency
- (No Category) Update dockerfile base image pinned versions
- (No Category) Regenerate lockfile with newer npm to get correct libc deps
- (No Category) Force bump protobufjs to fix vulnerabilities with castv2
- (No Category) Migrate eslint to oxlint
📚 Documentation
- (No Category) Spruce up source/client list on landing page
- (No Category) Blackbox docs
🧑🍳 Continuous Integration
- (No Category) Update tests to run in parallel
🧪 Testing
- (No Category) Ensure a client with upstreamRefresh disabled still detects dupes from in-house plays
- (No Category) Allow longer timeout for drizzle tests on less powerful machines
- (No Category) Setup chai-as-promised globally
- (No Category) More robust tmp dir usage for tests