[ffmpeg] Support splitting target based on extracted chapters (#24858)

This introduces a new FFmpegPostProcessor, which looks for "chapters"
extracted by IEs. For each identified chapter, it extracts a single
track from the source file, using ffmpeg's -ss / -t arguments.

youtube_dl/__init__.py

@@ -293,6 +293,8 @@ def _real_main(argv=None):
         })
         if not already_have_thumbnail:
             opts.writethumbnail = True
+    if opts.split_tracks:
+        postprocessors.append({'key': 'FFmpegSplitByTracks'})
     # XAttrMetadataPP should be run after post-processors that may change file
     # contents
     if opts.xattrs:

youtube_dl/options.py

@@ -525,6 +525,11 @@ def parseOpts(overrideArguments=None):
         '--external-downloader-args',
         dest='external_downloader_args', metavar='ARGS',
         help='Give these arguments to the external downloader')
+    downloader.add_option(
+        '--split-tracks',
+        dest='split_tracks', action='store_true', default=False,
+        help="Split tracks based on chapters information"
+    )
 
     workarounds = optparse.OptionGroup(parser, 'Workarounds')
     workarounds.add_option(

youtube_dl/postprocessor/__init__.py

@@ -12,6 +12,7 @@ from .ffmpeg import (
     FFmpegMetadataPP,
     FFmpegVideoConvertorPP,
     FFmpegSubtitlesConvertorPP,
+    FFmpegSplitByTracksPP,
 )
 from .xattrpp import XAttrMetadataPP
 from .execafterdownload import ExecAfterDownloadPP
@@ -27,6 +28,7 @@ __all__ = [
     'ExecAfterDownloadPP',
     'FFmpegEmbedSubtitlePP',
     'FFmpegExtractAudioPP',
+    'FFmpegSplitByTracksPP',
     'FFmpegFixupM3u8PP',
     'FFmpegFixupM4aPP',
     'FFmpegFixupStretchedPP',

youtube_dl/postprocessor/ffmpeg.py

@@ -1,6 +1,7 @@
 from __future__ import unicode_literals
 
 import io
+import logging
 import os
 import subprocess
 import time
@@ -655,3 +656,49 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                 }
 
         return sub_filenames, info
+
+
+class FFmpegSplitByTracksPP(FFmpegPostProcessor):
+    log = logging.getLogger(__name__)
+
+    def _ffmpeg_time_string(self, seconds):
+        t_minutes, t_seconds = divmod(seconds, 60)
+        t_hours, t_minutes = divmod(t_minutes, 60)
+        t_string = '{hrs:02}:{min:02}:{sec:02}'.format(hrs=t_hours, min=t_minutes, sec=t_seconds)
+        return t_string
+
+    def _build_track_name(self, chapter, information):
+        track_title = chapter.get("title", "")
+        track_title = encodeFilename(track_title)
+        track_title = track_title.replace("/", "_")
+
+        prefix, sep, ext = information['filepath'].rpartition('.')
+        track_name = "%s - %s%s%s" % (prefix, track_title, sep, ext)
+
+        return track_name
+
+    def _extract_track_from_chapter(self, chapter, information):
+        start = int(chapter['start_time'])
+        end = int(chapter['end_time'])
+        duration = end - start
+
+        start = self._ffmpeg_time_string(start)
+        duration = self._ffmpeg_time_string(duration)
+
+        destination = self._build_track_name(chapter, information)
+
+        self.run_ffmpeg(information['filepath'], destination, ['-c', 'copy', '-ss', start, '-t', duration])
+
+    def run(self, information):
+        chapters = information.get('chapters', [])
+        if not isinstance(chapters, list) or len(chapters) == 0:
+            self.log.warning('[ffmpeg] There are no tracks to extract')
+            return [], information
+
+        for idx, chapter in enumerate(chapters):
+            try:
+                self._extract_track_from_chapter(chapter, information)
+            except Exception as e:
+                self.log.error('Splitting track failed: ' + repr(e))
+
+        return [], information