Add an option to discard style information from TTML subtitles

youtube_dl/__init__.py

@@ -280,6 +280,7 @@ def _real_main(argv=None):
         postprocessors.append({
             'key': 'FFmpegSubtitlesConvertor',
             'format': opts.convertsubtitles,
+            'ttml_convert_style': opts.ttmlconvertstyle,
         })
     if opts.embedsubtitles:
         postprocessors.append({

youtube_dl/options.py

@@ -862,6 +862,10 @@ def parseOpts(overrideArguments=None):
         '--convert-subs', '--convert-subtitles',
         metavar='FORMAT', dest='convertsubtitles', default=None,
         help='Convert the subtitles to other format (currently supported: srt|ass|vtt|lrc)')
+    postproc.add_option(
+        '--ttml-convert-no-style',
+        action='store_false', dest='ttmlconvertstyle', default=True,
+        help='Discard all style information when converting subtitles from ttml/dfxp')
 
     parser.add_option_group(general)
     parser.add_option_group(network)

youtube_dl/postprocessor/ffmpeg.py

@@ -599,9 +599,10 @@ class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 
 
 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
-    def __init__(self, downloader=None, format=None):
+    def __init__(self, downloader=None, format=None, ttml_convert_style=True):
         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
         self.format = format
+        self.ttml_convert_style = ttml_convert_style
 
     def run(self, info):
         subs = info.get('requested_subtitles')
@@ -634,7 +635,7 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                 srt_file = subtitles_filename(filename, lang, 'srt', info.get('ext'))
 
                 with open(dfxp_file, 'rb') as f:
-                    srt_data = dfxp2srt(f.read())
+                    srt_data = dfxp2srt(f.read(), self.ttml_convert_style)
 
                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
                     f.write(srt_data)

youtube_dl/utils.py

@@ -4494,7 +4494,7 @@ def srt_subtitles_timecode(seconds):
     return '%02d:%02d:%02d,%03d' % (seconds / 3600, (seconds % 3600) / 60, seconds % 60, (seconds % 1) * 1000)
 
 
-def dfxp2srt(dfxp_data):
+def dfxp2srt(dfxp_data, convert_style=True):
     '''
     @param dfxp_data A bytes-like object containing DFXP data
     @returns A unicode object containing converted SRT data
@@ -4536,7 +4536,7 @@ def dfxp2srt(dfxp_data):
         def start(self, tag, attrib):
             if tag in (_x('ttml:br'), 'br'):
                 self._out += '\n'
-            else:
+            elif convert_style:
                 unclosed_elements = []
                 style = {}
                 element_style_id = attrib.get('style')
@@ -4579,7 +4579,7 @@ def dfxp2srt(dfxp_data):
                 self._unclosed_elements.append(unclosed_elements)
 
         def end(self, tag):
-            if tag not in (_x('ttml:br'), 'br'):
+            if convert_style and tag not in (_x('ttml:br'), 'br'):
                 unclosed_elements = self._unclosed_elements.pop()
                 for element in reversed(unclosed_elements):
                     self._out += '</%s>' % element