euterpe – Self-hosted music streaming server with RESTful API and Web interface. (Demo, Source Code) GPL-3.0
Go
Euterpe
Euterpe is self-hosted streaming service for music. Formerly known as “HTTPMS (HTTP Media Server)”.
A way to listen to your music library from everywhere. Once set up you won’t need anything but a browser. Think of it as your own Spotify service over which you have full control. Euterpe will let you browse through and listen to your music over HTTP(s). Up until now I’ve had a really bad time listening to my music which is stored back home. I would create a mount over ftp, sshfs or something similar and point the local player to the mounted library. Every time it resulted in some upleasantries. Just imagine searching in a network mounted directory!
It comes with a custom jPlayer which can handle playlists with thousands of songs. Which is an imrovement over the original which never included this performance patch.
I feel obliged to say that the music on the screenshot is written and performed by my close friend Velislav Ivanov.
Features
- Simple. It is just one binary, that’s it! You don’t need to faff about with interpreters or web servers
- Fast. A typical response time on my more than a decade old mediocre computer is 26ms for a fairly large collection
- Supports the most common audio formats such as mp3, oga, ogg, wav, flac, opus, web and m4a audio formats
- Built-in fast and simple Web UI so that you can play your music on every device
- Media and UI could be served over HTTP(S) natively without the need for other software
- User authentication (HTTP Basic, query token, Bearer token)
- Media artwork from local files or automatically downloaded from the Cover Art Archive
- Artist images could be downloaded automatically from Discogs
- Search by track name, artist or album
- Download whole album in a zip file with one click
- Controllable via media keys in OSX with the help of BeardedSpice
- Extensible via stable API
- Multiple clients and player plugins
- Uses jplayer to play your music on really old browsers
Demo
Just show, don’t talk, will ya? I will! You may take the server a spin with the live demo if you would like to. Feel free to thank all the artists who made their music available for this!
Requirements
If you want to install it from source you will need:
-
Go 1.16 or later installed and properly configured.
-
taglib – Read the install instructions or better yet the one inside your downloaded version. Most operating systems will have it in their package manager, though. Better use this one.
-
International Components for Unicode – The Euterpe binary dynamically links to
libicu
. Your friendly Linux distribution probably already has a package. For other OSs one should go here.
Install
The safest route is installing one of the releases.
Linux & macOS
If you have one of the releases (for example euterpe_1.1.0_linux.tar.gz
) it includes an install
script which would install Euterpe in /usr/bin/euterpe
. You will have to uninstall any previously installed versions first. An uninstall
script is provided as well.
Windows
Automatically creating a release version for Windows is in progress at the moment. For the time being check out the next section, “From Source”. Pay attention to the requirements section above. As of writing this the author hasn’t been yet initiated in the secret art of building and installing libraries on Windows so you are on your own.
From Source (any OS)
If installing from source running go install
in the project root directory will compile euterpe
and move its binary in your $GOPATH
. Releases from v1.0.1
onward have their go dependencies vendored in.
So, to install the master
branch, you can just run
go install github.com/ironsmile/euterpe
Or alternatively, if you want to produce a release version you will have to get the repository. Then in the root of the project run
make release
This will produce a binary euterpe
which is ready for distribution. Check its version with
./euterpe -v
First Run
Once installed, you are ready to use your media server. After its initial run it will create a configuration file which you will have to edit to suit your needs.
-
Start it with
euterpe
-
Edit the config.json and add your library paths to the “library” field. This is an important step. Without it,
euterpe
will not know where your media files are.
Docker
Alternatively to installing everything in your environment you can use the Docker image.
Start the server by running:
docker run -v "${HOME}/Music/:/root/Music" -p 8080:9996 -d ironsmile/euterpe:latest euterpe
Then point your browser to https://localhost:8080 and you will see the Euterpe web UI. The -v
flag in the Docker command will mount your $HOME/Music
directory to be discoverable by Euterpe.
Building the Image Yourself
You can use the Dockerfile in this repository to build the image yourself.
docker build -t ironsmile/euterpe github.com/ironsmile/euterpe
The euterpe
binary there is placed in /usr/local/bin/euterpe
.
Configuration
HTTPS configuration is saved in a JSON file, different for every user in the system. Its location is as follows:
- Linux or BSD:
$HOME/.euterpe/config.json
- Windows:
%APPDATA%\euterpe\config.json
When started for the first time Euterpe will create one for you. Here is an example:
{
// Address and port on which Euterpe will listen. It is in the form hostname[:port]
// For exact explanation see the Addr field in the Go's net.http.Server
// Make sure the user running Euterpe have permission to bind on the specified
// port number
"listen": ":443",
// true if you want to access Euterpe over HTTPS or false for plain HTTP.
// If set to true the "ssl_certificate" field must be configured as well.
"ssl": true,
// Provides the paths to the certificate and key files. Must be full paths, not
// relatives. If "ssl" is false this can be left out.
"ssl_certificate": {
"crt": "/full/path/to/certificate/file.crt",
"key": "/full/path/to/key/file.key"
},
// true if you want the server to require HTTP basic authentication. Credentials
// are set by the 'authentication' field below.
"basic_authenticate": true,
// User and password for the HTTP basic authentication.
"authentication": {
"user": "example",
"password": "example"
},
// An array with all the directories which will be scanned for media. They must be
// full paths and formatted according to your OS. So for example a Windows path
// have to be something like "D:\Media\Music".
// As expected Euterpe will need permission to read in the library folders.
"libraries": [
"/path/to/my/files",
"/some/more/files/can/be/found/here"
],
// Optional configuration on how to scan libraries. Note that this configuration
// is applied to each library separately.
"library_scan": {
// Will wait this much time before actually starting to scan a library.
// This might be useful when scanning is resource hungry operation and you
// want to postpone it on start up.
"initial_wait_duration": "1s",
// With this option a "operation" is defined by this number of scanned files.
"files_per_operation": 1500,
// After each "operation", sleep this amount of time.
"sleep_after_operation": "15ms"
},
// When true, Euterpe will search for images on the internet. This means album artwork
// and artists images. Cover Art Archive is used for album artworks when none is
// found locally. And Discogs for artist images. Anything found will be saved in
// the Euterpe database and later used to prevent further calls to the archive.
"download_artwork": true,
// If download_artwork is true the server will try to find artist artwork in the
// Discogs database. In order for this to work an authentication is required
// with their API. This here must be a personal access token. In effect the server
// will make requests on your behalf.
//
// See the API docs for more information:
// https://www.discogs.com/developers/#page:authentication,header:authentication-discogs-auth-flow
"discogs_auth_token": "some-personal-token"
}
List with all directives can be found in the configration wiki.
As an API
You can use Euterpe as a REST API and write your own player. Or maybe a plugin for your favourite player which would use your Euterpe installation as a back-end.
v1 Compatibility Promise
The API presented in this README is stable and will continue to be supported as long as version one of the service is around. And this should be very long time. I don’t plan to make backward incompatible changes. Ever. It has survived in this form since 2013. So it should be good for at least double than this amount of time in the future.
This means that clients written for Euterpe will continue to work. I will never break them on purpose and if this happened it will be considered a bug to be fixed as soon as possible.
Authentication
When your server is open you don’t have to authenticate requests to the API. Installations protected by user name and password require you to authenticate requests when using the API. For this the following methods are supported:
- Bearer token in the
Authorization
HTTP header (as described in RFC 6750):
Authorization: Bearer token
- Basic authentication (RFC 2617) with your username and password:
Authorization: Basic base64(username:password)
Authentication tokens can be acquired using the /v1/login/token/
endpoint described below. Using tokens is the preferred method since it does not expose your username and password in every request. Once acquired users must register the tokens using the /v1/register/token/
endpoint in order to “activate” them. Tokens which are not registered may or may not work. Tokens may have expiration date or they may not. Integration applications must provide a mechanism for token renewal.
Endpoints
Search
One can do a search query at the following endpoint
GET /v1/search/?q={query}
wich would return an JSON array with tracks. Every object in the JSON represents a single track which matches the query
. Example:
[
{
"album" : "Battlefield Vietnam",
"title" : "Somebody to Love",
"track" : 10,
"artist" : "Jefferson Airplane",
"artist_id": 33,
"id" : 18,
"album_id" : 2,
"format": "mp3",
"duration": 180000
},
{
"album" : "Battlefield Vietnam",
"artist" : "Jefferson Airplane",
"track" : 14,
"format": "flac",
"title" : "White Rabbit",
"album_id" : 2,
"id" : 22,
"artist_id": 33,
"duration": 308000
}
]
The most important thing here is the track ID at the id
key. It can be used for playing this track. The other interesting thing is album_id
. Tracks can be grouped in albums using this value. And the last field of particular interest is track
. It is the position of this track in the album.
Note that the track duration is in milliseconds.
Browse
A way to browse through the whole collection is via the browse API call. It allows you to get its albums or artists in an ordered and paginated manner.
GET /v1/browse/[?by=artist|album][&per-page={number}][&page={number}][&order-by=id|name][&order=desc|asc]
The returned JSON contains the data for the current page, the number of all pages for the current browse method and URLs of the next or previous pages.
{
"pages_count": 12,
"next": "/v1/browse/?page=4&per-page=10",
"previous": "/v1/browse/?page=2&per-page=10",
"data": [ /* different data types are returned, determined by the `by` parameter */ ]
}
For the moment there are two possible values for the by
parameter. Consequently there are two types of data
that can be returned: “artist” and “album” (which is the default).
by=artist
would result in value such as
{
"artist": "Jefferson Airplane",
"artist_id": 73
}
by=album
would result in value such as
{
"album": "Battlefield Vietnam"
"artist": "Jefferson Airplane",
"album_id": 2
}
Additional parameters
per-page: controls how many items would be present in the data
field for every particular page. The default is 10.
page: the generated data would be for this page. The default is 1.
order-by: controls how the results would be ordered. The value id
means the ordering would be done by the album or artist ID, depending on the by
argument. The same goes for the name
value. Defaults to name
.
order: controls if the order would ascending (with value asc
) or descending (with value desc
). Defaults to asc
.
Play a Song
GET /v1/file/{trackID}
This endpoint would return you the media file as is. A song’s trackID
can be found with the search API call.
Download an Album
GET /v1/album/{albumID}
This endpoint would return you an archive which contains the songs of the whole album.
Album Artwork
Euterpe supports album artwork. Here are all the methods for managing it through the API.
Get Artwork
GET /v1/album/{albumID}/artwork
Returns a bitmap image with artwork for this album if one is available. Searching for artwork works like this: the album’s directory would be scanned for any images (png/jpeg/gif/tiff files) and if anyone of them looks like an artwork, it would be shown. If this fails, you can configure Euterpe to search in the MusicBrainz Cover Art Archive. By default no external calls are made, see the ‘download_artwork’ configuration property.
By default the full size image will be served. One could request a thumbnail by appending the ?size=small
query.
Upload Artwork
PUT /v1/album/{albumID}/artwork
Can be used to upload artwork directly on the Euterpe server. This artwork will be stored in the server database and will not create any files in the library paths. The image should be sent in the body of the request in binary format without any transformations. Only images up to 5MB are accepted. Example:
curl -i -X PUT \
--data-binary @/path/to/file.jpg \
http://127.0.0.1:9996/v1/album/18/artwork
Remove Artwork
DELETE /v1/album/{albumID}/artwork
Will remove the artwork from the server database. Note, this will not touch any files on the file system. Thus it is futile to call it for artwork which was found on disk.
Artist Image
Euterpe could build a database with artists’ images. Which it could then be used throughout the interfaces. Here are all the methods for managing it through the API.
Get Artist Image
GET /v1/artist/{artistID}/image
Returns a bitmap image representing an artist if one is available. Searching for artwork works like this: if artist image is found in the database then it will be used. In case there is not and Euterpe is configured to download images from internet and has a Discogs access token then it will use the MusicBrainz and Discogs APIs in order to retrieve an image. By default no internet requests are made.
By default the full size image will be served. One could request a thumbnail by appending the ?size=small
query.
Upload Artist Image
PUT /v1/artist/{artistID}/image
Can be used to upload artist image directly on the Euterpe server. It will be stored in the server database and will not create any files in the library paths. The image should be sent in the body of the request in binary format without any transformations. Only images up to 5MB are accepted. Example:
curl -i -X PUT \
--data-binary @/path/to/file.jpg \
http://127.0.0.1:9996/v1/artist/23/image
Remove Artist Image
DELETE /v1/artist/{artistID}/image
Will remove the artist image the server database. Note, this will not touch any files on the file system.
Token Request
POST /v1/login/token/
{
"username": "your-username",
"password": "your-password"
}
You have to send your username and password as a JSON in the body of the request as described above. Provided they are correct you will receive the following response:
{
"token": "new-authentication-token"
}
Before you can use this token for accessing the API you will have to register it with on “Register Token” endpoint.
Register Token
POST /v1/register/token/
This endpoint registers the newly generated tokens with Euterpe. Only registered tokens will work. Requests at this endpoint must authenticate themselves using a previously generated token.