Groups subtitles (srt files) by sync. Useful when retrieving results from OpenSubtitles.
Install a tagged release from Stremio/node-subtitles-grouping on GitHub.
For example: npm install github:Stremio/node-subtitles-grouping#<tag>.
OpenSubtitles has two ways of matching subtitles: metadata (e.g. IMDB ID) and movieHash. While the first one looks subtitles for a particular movie or TV episode, the second method retrieves them for a specific video.
Obviously, matching by MovieHash is the much better method since it ensures the subtitles will be synced to that particular video file.
However, there are users in OpenSubtitles who upload subtitles for a particular video file but with wrong sync. This is where grouping the subtitles by sync helps.
The basic philosophy is that once we have the subtitles grouped, we select the group that has the most MovieHash matches in it. As long as most MovieHash matches are correct (always), this is the right sync group for that video. That way we weed out the "odd" subtitles and we also pick the correctly synced subtitles from the metadata-based matches (usually 80% of the matches).
This is how the example looks like. Each line represents a heatmap of an srt file (by time) and each colour represents a sync group. The brighter lines are picked by moviehash.
You can clearly see how the two groups are different in sync. You can also see how the red group has more bright lines, meaning most moviehash picks are there. This is the correct sync group. The green (wrong) group has one bright line - meaning we just found a wrong moviehash pick and filtered it. If we had also grouped all non-moviehash picks, the correctly synced would be added to the red group.
var groupSubtitles = require("subtitles-grouping");
groupSubtitles([
{ id: "3562019", uri: "http://dl.opensubtitles.org/en/download/filead/src-api/vrf-52c7037c6b/sid-vo81ml26hrarcsciua7gd44ta6/1952189414.gz" },
{ id: "3963807", uri: "./examples/dexter-4x1/3963807.srt" },
{ id: "3567323": uri: "./examples/dexter-4x1/3567323.srt" },
{ id: "3562666": uri: "./examples/dexter-4x1/3562666.srt" }
], function(err, groups) { console.log(groups) }, { /* OPTIONS agent, sensitivity */ });
// subtitles URI is a URL/local path to an srt file, gzip-compressed srt or a zip containing an srt
// groups will be an array of the groups, each group being an array of subtitles as given to groupSubtitles() but also with a ``.heatmap`` property
// see example/example.js// Retrieves an srt string from path/URL to srt, gz or zip
// Also converts encoding to UTF8
require("./lib/retriever").retrieveSrt(/* path/URL to an srt, gz or zip file */, function(err,buf) { })``
// Builds a heatmap of an srt file
require("./lib/heatmap")(/* string in a srt format */) //returns an array heatmap of that srtHTTP downloads preserve the original bytes. Decoding uses an explicit options.charset
(unless it is "auto"), then a UTF-8/UTF-16 byte-order mark, then valid UTF-8 text,
then a supported non-UTF-8 charset from the HTTP Content-Type, and finally automatic
detection. NUL-containing data is left to the byte-order mark, declared charset or
detector so UTF-16 is not mistaken for UTF-8.
The declared charset also applies to gzip subtitles, but never to entries inside a
ZIP archive. Generic iso-8859-1, latin1, us-ascii, ascii and binary labels are
ignored because servers and download tools may apply them regardless of the file's
encoding. A wrong specific legacy charset can still
override a correct detector guess; unlabelled legacy subtitles still depend on the
detector's supported encodings.
subtitles-grouping [path to srt] [path to srt] ...- Big thanks to OpenSubtitles for providing subtitles dumps to test with