trtomasz / stremio-local-addon

Add-on for stremio (using v3 protocol) meant to be ran locally which indexes locally found torrent and video files

stremio-local-addon

An add-on for stremio meant to be ran locally which indexes locally found torrent and video files

It does a few things:

• Scans the filesystem for video files (currently mp4, mkv, avi and others - depends on the implementation in lib/findFiles) and torrent files containing videos
• Tries to recognize video files as entertainment content and associate them with an IMDB ID
• Presents a catalog to Stremio containing all the found items, where IMDB-recognized video files are grouped by IMDB ID and torrents are grouped by BitTorrent infohash; non-recognized video files are omitted
• Allows Stremio to open any BitTorrent infoHash using /meta/bt:<infoHash> request to this add-on

Testing

npm start

npm test

Data structure

The data is kept in an set (dictionary) of entries (filePath=>entry). Each entry represents one file on the filesystem.

Each entry is defined by { itemId, name, files }. When the entry represents a filePath of no interest (not an indexable video), we just save { files: [] }. Entries may contain other extra properties such as the Bittorrent-specific ih and sources.

The reason files is an array is that one file on the filesystem may contain multiple indexable videos (e.g. a .torrent file).

itemId

itemId is the stremio metadata ID this entry corresponds to.

It may be formed in two ways:

bt:<bittorrent info hash>

local:<IMDB ID>

In other words, to make a single Stremio MetaItem object, files are grouped either by the torrent they belong in, or by the IMDB ID they are indexed by. So this add-on will only display videos either indexed by IMDB ID or belonging in a torrent.

Storage

The persistence layer is defined in lib/storage, and it keeps each entry as one line in the storage file. Entries may only be added to the storage file, and no entries may be removed.

It allows referencing entries by file path (.byFilePath) or by item ID (.byItemId). There may be more than one entry per item ID.

The in-memory structure is as follows

byFilePath: filePath=>entry

byItemId: itemId=>(filePath=>entry)

Finally, we have the storage.getAggrEntry function, which gives us an aggregate entry for an itemId, by taking all entries for the given itemId and merging them by concatting files and taking the leftmost values of the other properties (name, ih, sources). Taking the leftmost values is OK for torrents since itemId implies grouping by torrent anyway (in other words all ih and sources values will be the same).