Smooth Streaming Support : TTML Captions

src/mss/parser/MssParser.js

@@ -470,6 +470,7 @@ function MssParser(config) {
         manifest.timeShiftBufferDepth = parseFloat(smoothStreamingMedia.getAttribute('DVRWindowLength')) / TIME_SCALE_100_NANOSECOND_UNIT;
         manifest.mediaPresentationDuration = (parseFloat(smoothStreamingMedia.getAttribute('Duration')) === 0) ? Infinity : parseFloat(smoothStreamingMedia.getAttribute('Duration')) / TIME_SCALE_100_NANOSECOND_UNIT;
         manifest.minBufferTime = mediaPlayerModel.getStableBufferTime();
+        manifest.ttmlTimeIsRelative = true;
 
         // In case of live streams, set availabilityStartTime property according to DVRWindowLength
         if (manifest.type === 'dynamic') {
@@ -597,4 +598,4 @@ function MssParser(config) {
 }
 
 MssParser.__dashjs_factory_name = 'MssParser';
-export default FactoryMaker.getClassFactory(MssParser);
+export default FactoryMaker.getClassFactory(MssParser);

src/streaming/MediaPlayer.js

@@ -1928,6 +1928,7 @@ function MediaPlayer() {
         textController = TextController(context).getInstance();
         textController.setConfig({
             errHandler: errHandler,
+            manifestModel: ManifestModel(context).getInstance(),
             dashManifestModel: dashManifestModel,
             mediaController: mediaController,
             streamController: streamController,

src/streaming/text/TextController.js

@@ -43,6 +43,7 @@ function TextController() {
     let allTracksAreDisabled,
         errHandler,
         dashManifestModel,
+        manifestModel,
         mediaController,
         videoModel,
         streamController,
@@ -71,6 +72,9 @@ function TextController() {
         if (config.dashManifestModel) {
             dashManifestModel = config.dashManifestModel;
         }
+        if (config.manifestModel) {
+            manifestModel = config.manifestModel;
+        }
         if (config.mediaController) {
             mediaController = config.mediaController;
         }
@@ -94,6 +98,7 @@ function TextController() {
         textSourceBuffer.setConfig({
             errHandler: errHandler,
             dashManifestModel: dashManifestModel,
+            manifestModel: manifestModel,
             mediaController: mediaController,
             videoModel: videoModel,
             streamController: streamController,

src/streaming/text/TextSourceBuffer.js

@@ -53,6 +53,7 @@ function TextSourceBuffer() {
         boxParser,
         errHandler,
         dashManifestModel,
+        manifestModel,
         mediaController,
         parser,
         vttParser,
@@ -203,6 +204,9 @@ function TextSourceBuffer() {
         if (config.dashManifestModel) {
             dashManifestModel = config.dashManifestModel;
         }
+        if (config.manifestModel) {
+            manifestModel = config.manifestModel;
+        }
         if (config.mediaController) {
             mediaController = config.mediaController;
         }
@@ -328,7 +332,10 @@ function TextSourceBuffer() {
                             subOffset += sample.subSizes[j];
                         }
                         try {
-                            result = parser.parse(ccContent, sampleStart / timescale, (sampleStart + sample.duration) / timescale, images);
+                            // Only used for Miscrosoft Smooth Streaming support - caption time is relative to sample time. In this case, we apply an offset.
+                            let manifest = manifestModel.getValue();
+                            let offsetTime = manifest.ttmlTimeIsRelative ? sampleStart / timescale : 0;
+                            result = parser.parse(ccContent, sampleStart / timescale, (sampleStart + sample.duration) / timescale, images, offsetTime);
                             textTracks.addCaptions(currFragmentedTrackIdx, firstSubtitleStart / timescale, result);
                         } catch (e) {
                             log('TTML parser error: ' + e.message);

src/streaming/utils/TTMLParser.js

@@ -171,9 +171,11 @@ function TTMLParser() {
      * If intervalStart and intervalEnd defined, use them to clip the interval.
      * Return null if no overlap with interval
      */
-    function getClippedInterval(entity, intervalStart, intervalEnd) {
+    function getClippedInterval(entity, intervalStart, intervalEnd, offsetTime) {
         let startTime = parseTimings(entity.begin);
         let endTime = parseTimings(entity.end);
+        startTime += offsetTime;
+        endTime += offsetTime;
         startTime = clipStartTime(startTime, intervalStart);
         endTime = clipEndTime(endTime, intervalEnd);
         if (typeof intervalStart !== 'undefined' && typeof intervalEnd !== 'undefined') {
@@ -207,9 +209,9 @@ function TTMLParser() {
      * @param {number} intervalStart
      * @param {number} intervalEnd
      * @param {array} imageArray - images represented as binary strings
+     * @param {number} offsetTime - offset time to apply to cue time
      */
-
-    function parse(data, intervalStart, intervalEnd, imageArray) {
+    function parse(data, intervalStart, intervalEnd, imageArray, offsetTime) {
         let tt, // Top element
             head, // head in tt
             body, // body in tt
@@ -364,7 +366,7 @@ function TTMLParser() {
                 // The timing may either be on paragraph or span level.
                 if (paragraph.hasOwnProperty('begin') && paragraph.hasOwnProperty('end')) {
                     // Timing on paragraph level
-                    let clippedInterval = getClippedInterval(paragraph, intervalStart, intervalEnd);
+                    let clippedInterval = getClippedInterval(paragraph, intervalStart, intervalEnd, offsetTime);
                     if (clippedInterval !== null) {
                         cueIntervals.push(clippedInterval);
                     }