Merge pull request #499 from calzoneman/gdrive-captions

Support captions/subtitles for Google Drive videos

.gitignore

@@ -10,3 +10,4 @@ node_modules
 *.key
 torlist
 www/cache
+google-drive-subtitles

NEWS.md

@@ -1,3 +1,11 @@
+2015-07-25
+==========
+
+  * CyTube now supports subtitles for Google Drive videos.  In order to take
+    advantage of this, you must upgrade mediaquery by running `npm install
+    cytube/mediaquery`.  Subtitles are cached in the google-drive-subtitles
+    folder.
+
 2015-07-07
 ==========
 

docs/google-drive-subtitles.md

@@ -0,0 +1,24 @@
+Adding subtitles to Google Drive
+================================
+
+  1. Upload your video to Google Drive
+  2. Right click the video in Google Drive and click Manage caption tracks
+  3. Click Add new captions or transcripts
+  4. Upload a supported subtitle file
+    * I have verified that Google Drive will accept .srt and .vtt subtitles.  It
+      might accept others as well, but I have not tested them.
+
+Once you have uploaded your subtitles, they should be available the next time
+the video is refreshed by CyTube (either restart it or delete the playlist item
+and add it again).  On the video you should see a speech bubble icon in the
+controls, which will pop up a menu of available subtitle tracks.
+
+## Limitations ##
+
+  * Google Drive converts the subtitles you upload into a custom format which
+    loses some information from the original captions.  For example, annotations
+    for who is speaking are not preserved.
+  * As far as I know, Google Drive is not able to automatically detect when
+    subtitle tracks are embedded within the video file.  You must upload the
+    subtitles separately (there are plenty of tools to extract
+    captions/subtitles from MKV and MP4 files).

lib/google2vtt.js

@@ -0,0 +1,173 @@
+var cheerio = require('cheerio');
+var https = require('https');
+var fs = require('fs');
+var path = require('path');
+var querystring = require('querystring');
+var crypto = require('crypto');
+
+var Logger = require('./logger');
+
+function md5(input) {
+    var hash = crypto.createHash('md5');
+    hash.update(input);
+    return hash.digest('base64').replace(/\//g, ' ')
+        .replace(/\+/g, '#')
+        .replace(/=/g, '-');
+}
+
+var slice = Array.prototype.slice;
+var subtitleDir = path.resolve(__dirname, '..', 'google-drive-subtitles');
+var ONE_HOUR = 60 * 60 * 1000;
+var ONE_DAY = 24 * ONE_HOUR;
+
+function padZeros(n) {
+    n = n.toString();
+    if (n.length < 2) n = '0' + n;
+    return n;
+}
+
+function formatTime(time) {
+    var hours = Math.floor(time / 3600);
+    time = time % 3600;
+    var minutes = Math.floor(time / 60);
+    time = time % 60;
+    var seconds = Math.floor(time);
+    var ms = time - seconds;
+
+    var list = [minutes, seconds];
+    if (hours) {
+        list.unshift(hours);
+    }
+
+    return list.map(padZeros).join(':') + ms.toFixed(3).substring(1);
+}
+
+function fixText(text) {
+    return text.replace(/&amp;/g, '&')
+        .replace(/&lt;/g, '<')
+        .replace(/&gt;/g, '>')
+        .replace(/&quot;/g, '"')
+        .replace(/&#39;/g, "'")
+        .replace(/-->/g, '--&gt;');
+}
+
+exports.convert = function convertSubtitles(subtitles) {
+    var $ = cheerio.load(subtitles, { xmlMode: true });
+    var lines = slice.call($('transcript text').map(function (index, elem) {
+        var start = parseFloat(elem.attribs.start);
+        var end = start + parseFloat(elem.attribs.dur);
+        var text;
+        if (elem.children.length) {
+            text = elem.children[0].data;
+        } else {
+            text = '';
+        }
+
+        var line = formatTime(start) + ' --> ' + formatTime(end);
+        line += '\n' + fixText(text) + '\n';
+        return line;
+    }));
+
+    return 'WEBVTT\n\n' + lines.join('\n');
+};
+
+exports.attach = function setupRoutes(app) {
+    app.get('/gdvtt/:id/:lang/(:name)?.vtt', handleGetSubtitles);
+};
+
+function handleGetSubtitles(req, res) {
+    var id = req.params.id;
+    var lang = req.params.lang;
+    var name = req.params.name || '';
+    var vid = req.query.vid;
+    if (typeof vid !== 'string' || typeof id !== 'string' || typeof lang !== 'string') {
+        return res.sendStatus(400);
+    }
+    var file = [id, lang, md5(name)].join('_') + '.vtt';
+    var fileAbsolute = path.join(subtitleDir, file);
+
+    fs.exists(fileAbsolute, function (exists) {
+        if (exists) {
+            res.sendFile(file, { root: subtitleDir });
+        } else {
+            fetchSubtitles(id, lang, name, vid, fileAbsolute, function (err) {
+                if (err) {
+                    Logger.errlog.log(err.stack);
+                    return res.sendStatus(500);
+                }
+
+                res.sendFile(file, { root: subtitleDir });
+            });
+        }
+    });
+}
+
+function fetchSubtitles(id, lang, name, vid, file, cb) {
+    var query = {
+        id: id,
+        v: id,
+        vid: vid,
+        lang: lang,
+        name: name,
+        type: 'track',
+        kind: undefined
+    };
+
+    var url = 'https://drive.google.com/timedtext?' + querystring.stringify(query);
+    https.get(url, function (res) {
+        if (res.statusCode !== 200) {
+            return cb(new Error(res.statusMessage));
+        }
+
+        var buf = '';
+        res.setEncoding('utf-8');
+        res.on('data', function (data) {
+            buf += data;
+        });
+
+        res.on('end', function () {
+            try {
+                buf = exports.convert(buf);
+            } catch (e) {
+                return cb(e);
+            }
+
+            fs.writeFile(file, buf, function (err) {
+                if (err) {
+                    cb(err);
+                } else {
+                    Logger.syslog.log('Saved subtitle file ' + file);
+                    cb();
+                }
+            });
+        });
+    }).on('error', function (err) {
+        cb(err);
+    });
+}
+
+function clearOldSubtitles() {
+    fs.readdir(subtitleDir, function (err, files) {
+        if (err) {
+            Logger.errlog.log(err.stack);
+            return;
+        }
+
+        files.forEach(function (file) {
+            fs.stat(path.join(subtitleDir, file), function (err, stats) {
+                if (err) {
+                    Logger.errlog.log(err.stack);
+                    return;
+                }
+
+                if (stats.mtime.getTime() < Date.now() - ONE_DAY) {
+                    Logger.syslog.log('Deleting old subtitle file: ' + file);
+                    fs.unlink(path.join(subtitleDir, file));
+                }
+            });
+        });
+    });
+}
+
+setInterval(clearOldSubtitles, ONE_HOUR);
+clearOldSubtitles();

lib/media.js

@@ -37,7 +37,8 @@ Media.prototype = {
                 codec: this.meta.codec,
                 bitrate: this.meta.bitrate,
                 scuri: this.meta.scuri,
-                embed: this.meta.embed
+                embed: this.meta.embed,
+                gdrive_subtitles: this.meta.gdrive_subtitles
             }
         };
     },

lib/server.js

@@ -14,6 +14,11 @@ module.exports = {
         fs.exists(chandumppath, function (exists) {
             exists || fs.mkdir(chandumppath);
         });
+
+        var gdvttpath = path.join(__dirname, "../google-drive-subtitles");
+        fs.exists(gdvttpath, function (exists) {
+            exists || fs.mkdir(gdvttpath);
+        });
         singleton = new Server();
         return singleton;
     },

lib/web/webserver.js

@@ -243,6 +243,7 @@ module.exports = {
         require("./auth").init(app);
         require("./account").init(app);
         require("./acp").init(app);
+        require("../google2vtt").attach(app);
         app.use(static(path.join(__dirname, "..", "..", "www"), {
             maxAge: Config.get("http.max-age") || Config.get("http.cache-ttl")
         }));

package.json

@@ -2,7 +2,7 @@
   "author": "Calvin Montgomery",
   "name": "CyTube",
   "description": "Online media synchronizer and chat",
-  "version": "3.8.2",
+  "version": "3.9.0",
   "repository": {
     "url": "http://github.com/calzoneman/sync"
   },

player/videojs.coffee

@@ -66,6 +66,20 @@ window.VideoJSPlayer = class VideoJSPlayer extends Player
                 ).appendTo(video)
             )
 
+            if data.meta.gdrive_subtitles
+                data.meta.gdrive_subtitles.available.forEach((subt) ->
+                    label = subt.lang_original
+                    if subt.name
+                        label += " (#{subt.name})"
+                    $('<track/>').attr(
+                        src: "/gdvtt/#{data.id}/#{subt.lang}/#{subt.name}.vtt?\
+                                vid=#{data.meta.gdrive_subtitles.vid}"
+                        kind: 'subtitles'
+                        srclang: subt.lang
+                        label: label
+                    ).appendTo(video)
+                )
+
             @player = videojs(video[0], autoplay: true, controls: true)
             @player.ready(=>
                 @setVolume(VOLUME)
@@ -91,6 +105,21 @@ window.VideoJSPlayer = class VideoJSPlayer extends Player
                 @player.on('seeked', =>
                     $('.vjs-waiting').removeClass('vjs-waiting')
                 )
+
+                # Workaround for Chrome-- it seems that the click bindings for
+                # the subtitle menu aren't quite set up until after the ready
+                # event finishes, so set a timeout for 1ms to force this code
+                # not to run until the ready() function returns.
+                setTimeout(->
+                    $('#ytapiplayer .vjs-subtitles-button .vjs-menu-item').each((i, elem) ->
+                        if elem.textContent == localStorage.lastSubtitle
+                            elem.click()
+
+                        elem.onclick = ->
+                            if elem.attributes['aria-selected'].value == 'true'
+                                localStorage.lastSubtitle = elem.textContent
+                    )
+                , 1)
             )
         )
 

www/css/cytube.css

@@ -631,3 +631,11 @@ table td {
 #videowrap .embed-responsive:-ms-full-screen { width: 100%; }
 #videowrap .embed-responsive:-o-full-screen { width: 100%; }
 #videowrap .embed-responsive:full-screen { width: 100%; }
+
+li.vjs-menu-item.vjs-selected {
+    background-color: #66a8cc !important;
+}
+
+.video-js video::-webkit-media-text-track-container {
+    bottom: 50px;
+}

www/js/player.js

@@ -445,7 +445,7 @@
   })(Player);
 
   sortSources = function(sources) {
-    var flv, flvOrder, i, idx, len, nonflv, pref, qualities, quality, qualityOrder, sourceOrder;
+    var flv, flvOrder, idx, j, len, nonflv, pref, qualities, quality, qualityOrder, sourceOrder;
     if (!sources) {
       console.error('sortSources() called with null source list');
       return [];
@@ -459,8 +459,8 @@
     qualityOrder = qualities.slice(idx).concat(qualities.slice(0, idx));
     sourceOrder = [];
     flvOrder = [];
-    for (i = 0, len = qualityOrder.length; i < len; i++) {
-      quality = qualityOrder[i];
+    for (j = 0, len = qualityOrder.length; j < len; j++) {
+      quality = qualityOrder[j];
       if (quality in sources) {
         flv = [];
         nonflv = [];
@@ -524,6 +524,21 @@
               'data-quality': source.quality
             }).appendTo(video);
           });
+          if (data.meta.gdrive_subtitles) {
+            data.meta.gdrive_subtitles.available.forEach(function(subt) {
+              var label;
+              label = subt.lang_original;
+              if (subt.name) {
+                label += " (" + subt.name + ")";
+              }
+              return $('<track/>').attr({
+                src: "/gdvtt/" + data.id + "/" + subt.lang + "/" + subt.name + ".vtt?vid=" + data.meta.gdrive_subtitles.vid,
+                kind: 'subtitles',
+                srclang: subt.lang,
+                label: label
+              }).appendTo(video);
+            });
+          }
           _this.player = videojs(video[0], {
             autoplay: true,
             controls: true
@@ -547,9 +562,21 @@
                 return sendVideoUpdate();
               }
             });
-            return _this.player.on('seeked', function() {
+            _this.player.on('seeked', function() {
               return $('.vjs-waiting').removeClass('vjs-waiting');
             });
+            return setTimeout(function() {
+              return $('#ytapiplayer .vjs-subtitles-button .vjs-menu-item').each(function(i, elem) {
+                if (elem.textContent === localStorage.lastSubtitle) {
+                  elem.click();
+                }
+                return elem.onclick = function() {
+                  if (elem.attributes['aria-selected'].value === 'true') {
+                    return localStorage.lastSubtitle = elem.textContent;
+                  }
+                };
+              });
+            }, 1);
           });
         };
       })(this));