Mopidy is a music server that is extended by extensions. This glossary covers Mopidy core and all extensions.

HTTP frontend: The bundled extension that serves Mopidy's HTTP API, the JSON-RPC WebSocket and all web apps on one port. Avoid: HTTP server, web server

Web app: Anything an extension mounts under /<name>/ in the HTTP frontend. A web client is one kind of web app. Avoid: HTTP app, web application

Web client: A web app that gives users a browser UI to control Mopidy. Avoid: Web UI, frontend

ASGI app: A web app that an extension registers with http:asgi. The public type is a plain ASGI callable, not a class from a framework.

Tornado app: A web app that an extension registers with http:app, as a list of Tornado request handlers. Avoid: Legacy app

Static app: A web app that an extension registers with http:static, as a directory of files.

Media framework: A third-party library that plays or reads media, such as GStreamer, FFmpeg or mpv. It is not a Mopidy component. Mopidy uses GStreamer today. Avoid: Audio backend, audio engine

Playback engine: The component that plays URIs to the audio output and reports stream metadata. There is one playback engine in each Mopidy process, and it lives as long as the process. It does not read media info of URIs that it does not play. Avoid: Audio engine, audio backend, player

Audio: Mopidy's current playback engine, mopidy.audio. GstAudio implements it with GStreamer.

Stream metadata: Metadata that the playback engine reports while it plays, and that can change during playback, such as the ICY title of a radio stream. Only the playback engine gives stream metadata. The reader does not. Avoid: Tags

Stream title: The field of the stream metadata that tells what plays now, for example the current song on a radio stream.

Reader: The object that reads media and playlist documents without playing them, mopidy.media.Reader. It has three methods: read media info, read playlist entries and find playback target. Each caller, for example a backend, makes its own reader with Reader.create() and closes it when it stops. A reader keeps an HTTP connection pool, but it does not cache results. Avoid: Scanner, prober, discoverer

Media info reader: The part of a reader that a media framework implements, MediaInfoReader. It gives the media info of one URI. Callers do not use it directly. It is not public until Mopidy has a way to plug in other implementations. Avoid: Scanner, metadata reader

Media info: What read media info gives for one URI, MediaInfo: the metadata as a track, if the media info reader decoded audio (playable), if the media allows seeking, and the embedded images. Playable does not mean that the playback engine can play the URI. Media info does not contain raw tags from the media framework. Avoid: Scan result

Read media info: Get the media info of one URI, Reader.read_media_info(). It reads only that URI. If the URI is a playlist document, the result is not playable. It does not follow playlist documents. Avoid: Scan, read metadata

Playlist format: A text format that lists URIs: M3U, PLS, XSPF, ASX, ASX reference or URI list. HLS and DASH are not playlist formats. They are stream formats that the playback engine plays.

Playlist document: Content in a playlist format at a URI, local or remote. It is read-only. It is not a playlist: a playlist belongs to a backend, which can save or delete it. Avoid: Playlist file, remote playlist, playlist

Playlist entry: One item in a playlist document, PlaylistEntry. It has a track with the metadata of the entry, and one or more alternatives. The URI of the track is the first alternative.

Alternative: One of the URIs of a playlist entry. All alternatives of an entry give the same content, in order of preference. Only XSPF and ASX can give more than one alternative. Avoid: Source, mirror, location

Parse playlist entries: Get the playlist entries from the bytes of a playlist document, parse_playlist_entries(). It does no I/O. The content decides the playlist format. The media type and the URI are only hints for which format to try first. It gives no entries for content that is not a playlist document, and for HLS and DASH. Avoid: Parse playlist, parse entries

Read playlist entries: Fetch a playlist document from a file, http or https URI and parse its playlist entries, Reader.read_playlist_entries(). It goes only one level deep. It does not read the media info of the entries, and it does not check that they play. Avoid: Expand, read entries

Find playback target: Get the playback target for a URI that the user wants to play now, Reader.find_playback_target(). It reads playlist entries and media info, nested to any depth, and tries the alternatives of each entry in order. It gives a PlaybackTarget: the URI, its media info and the playlist entry that it came from. It does not raise an error for a URI that it cannot read: it tries the next one. Avoid: Resolve, unwrap, translate, find stream

Playback target: The URI to give to the playback engine, such as a local audio file, an HTTP radio stream or an HLS stream. It is not a playlist document. It is not a promise that playback works: when find playback target cannot read a URI that is not a playlist document, it gives that URI as an unverified playback target, without media info. Avoid: Stream URI, stream