from worst to best quality.
Potential fields:
- * url Mandatory. The URL of the video file or URL of
- the manifest file in case of fragmented media
- (DASH, hls, hds).
+ * url Mandatory. The URL of the video file
+ * manifest_url
+ The URL of the manifest file in case of
+ fragmented media (DASH, hls, hds)
* ext Will be calculated from URL if missing
* format A human-readable description of the format
("mp4 container with h264/opus").
formats.append({
'format_id': format_id,
'url': manifest_url,
+ 'manifest_url': manifest_url,
'ext': 'flv' if bootstrap_info is not None else None,
'tbr': tbr,
'width': width,
# format_id intact.
if not live:
format_id.append(stream_name if stream_name else '%d' % (tbr if tbr else len(formats)))
+ manifest_url = format_url(line.strip())
f = {
'format_id': '-'.join(format_id),
- 'url': format_url(line.strip()),
+ 'url': manifest_url,
+ 'manifest_url': manifest_url,
'tbr': tbr,
'ext': ext,
'fps': float_or_none(last_info.get('FRAME-RATE')),
mpd_base_url = re.match(r'https?://.+/', urlh.geturl()).group()
return self._parse_mpd_formats(
- compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url, formats_dict=formats_dict)
+ compat_etree_fromstring(mpd.encode('utf-8')), mpd_id, mpd_base_url,
+ formats_dict=formats_dict, mpd_url=mpd_url)
- def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}):
+ def _parse_mpd_formats(self, mpd_doc, mpd_id=None, mpd_base_url='', formats_dict={}, mpd_url=None):
"""
Parse formats from MPD manifest.
References:
def extract_multisegment_info(element, ms_parent_info):
ms_info = ms_parent_info.copy()
+
+ # As per [1, 5.3.9.2.2] SegmentList and SegmentTemplate share some
+ # common attributes and elements. We will only extract relevant
+ # for us.
+ def extract_common(source):
+ segment_timeline = source.find(_add_ns('SegmentTimeline'))
+ if segment_timeline is not None:
+ s_e = segment_timeline.findall(_add_ns('S'))
+ if s_e:
+ ms_info['total_number'] = 0
+ ms_info['s'] = []
+ for s in s_e:
+ r = int(s.get('r', 0))
+ ms_info['total_number'] += 1 + r
+ ms_info['s'].append({
+ 't': int(s.get('t', 0)),
+ # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60])
+ 'd': int(s.attrib['d']),
+ 'r': r,
+ })
+ start_number = source.get('startNumber')
+ if start_number:
+ ms_info['start_number'] = int(start_number)
+ timescale = source.get('timescale')
+ if timescale:
+ ms_info['timescale'] = int(timescale)
+ segment_duration = source.get('duration')
+ if segment_duration:
+ ms_info['segment_duration'] = int(segment_duration)
+
+ def extract_Initialization(source):
+ initialization = source.find(_add_ns('Initialization'))
+ if initialization is not None:
+ ms_info['initialization_url'] = initialization.attrib['sourceURL']
+
segment_list = element.find(_add_ns('SegmentList'))
if segment_list is not None:
+ extract_common(segment_list)
+ extract_Initialization(segment_list)
segment_urls_e = segment_list.findall(_add_ns('SegmentURL'))
if segment_urls_e:
ms_info['segment_urls'] = [segment.attrib['media'] for segment in segment_urls_e]
- initialization = segment_list.find(_add_ns('Initialization'))
- if initialization is not None:
- ms_info['initialization_url'] = initialization.attrib['sourceURL']
else:
segment_template = element.find(_add_ns('SegmentTemplate'))
if segment_template is not None:
- start_number = segment_template.get('startNumber')
- if start_number:
- ms_info['start_number'] = int(start_number)
- segment_timeline = segment_template.find(_add_ns('SegmentTimeline'))
- if segment_timeline is not None:
- s_e = segment_timeline.findall(_add_ns('S'))
- if s_e:
- ms_info['total_number'] = 0
- ms_info['s'] = []
- for s in s_e:
- r = int(s.get('r', 0))
- ms_info['total_number'] += 1 + r
- ms_info['s'].append({
- 't': int(s.get('t', 0)),
- # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60])
- 'd': int(s.attrib['d']),
- 'r': r,
- })
- else:
- timescale = segment_template.get('timescale')
- if timescale:
- ms_info['timescale'] = int(timescale)
- segment_duration = segment_template.get('duration')
- if segment_duration:
- ms_info['segment_duration'] = int(segment_duration)
+ extract_common(segment_template)
media_template = segment_template.get('media')
if media_template:
ms_info['media_template'] = media_template
if initialization:
ms_info['initialization_url'] = initialization
else:
- initialization = segment_template.find(_add_ns('Initialization'))
- if initialization is not None:
- ms_info['initialization_url'] = initialization.attrib['sourceURL']
+ extract_Initialization(segment_template)
return ms_info
+ def combine_url(base_url, target_url):
+ if re.match(r'^https?://', target_url):
+ return target_url
+ return '%s%s%s' % (base_url, '' if base_url.endswith('/') else '/', target_url)
+
mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration'))
formats = []
for period in mpd_doc.findall(_add_ns('Period')):
f = {
'format_id': '%s-%s' % (mpd_id, representation_id) if mpd_id else representation_id,
'url': base_url,
+ 'manifest_url': mpd_url,
'ext': mimetype2ext(mime_type),
'width': int_or_none(representation_attrib.get('width')),
'height': int_or_none(representation_attrib.get('height')),
}
representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info)
if 'segment_urls' not in representation_ms_info and 'media_template' in representation_ms_info:
- if 'total_number' not in representation_ms_info and 'segment_duration':
- segment_duration = float(representation_ms_info['segment_duration']) / float(representation_ms_info['timescale'])
- representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration))
+
media_template = representation_ms_info['media_template']
media_template = media_template.replace('$RepresentationID$', representation_id)
media_template = re.sub(r'\$(Number|Bandwidth|Time)\$', r'%(\1)d', media_template)
# As per [1, 5.3.9.4.4, Table 16, page 55] $Number$ and $Time$
# can't be used at the same time
- if '%(Number' in media_template:
- representation_ms_info['segment_urls'] = [
- media_template % {
+ if '%(Number' in media_template and 's' not in representation_ms_info:
+ segment_duration = None
+ if 'total_number' not in representation_ms_info and 'segment_duration':
+ segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale'])
+ representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration))
+ representation_ms_info['fragments'] = [{
+ 'url': media_template % {
'Number': segment_number,
'Bandwidth': representation_attrib.get('bandwidth'),
- }
- for segment_number in range(
- representation_ms_info['start_number'],
- representation_ms_info['total_number'] + representation_ms_info['start_number'])]
+ },
+ 'duration': segment_duration,
+ } for segment_number in range(
+ representation_ms_info['start_number'],
+ representation_ms_info['total_number'] + representation_ms_info['start_number'])]
else:
- representation_ms_info['segment_urls'] = []
+ # $Number*$ or $Time$ in media template with S list available
+ # Example $Number*$: http://www.svtplay.se/klipp/9023742/stopptid-om-bjorn-borg
+ # Example $Time$: https://play.arkena.com/embed/avp/v2/player/media/b41dda37-d8e7-4d3f-b1b5-9a9db578bdfe/1/129411
+ representation_ms_info['fragments'] = []
segment_time = 0
+ segment_d = None
+ segment_number = representation_ms_info['start_number']
def add_segment_url():
- representation_ms_info['segment_urls'].append(
- media_template % {
- 'Time': segment_time,
- 'Bandwidth': representation_attrib.get('bandwidth'),
- }
- )
+ segment_url = media_template % {
+ 'Time': segment_time,
+ 'Bandwidth': representation_attrib.get('bandwidth'),
+ 'Number': segment_number,
+ }
+ representation_ms_info['fragments'].append({
+ 'url': segment_url,
+ 'duration': float_or_none(segment_d, representation_ms_info['timescale']),
+ })
for num, s in enumerate(representation_ms_info['s']):
segment_time = s.get('t') or segment_time
+ segment_d = s['d']
add_segment_url()
+ segment_number += 1
for r in range(s.get('r', 0)):
- segment_time += s['d']
+ segment_time += segment_d
add_segment_url()
- segment_time += s['d']
- if 'segment_urls' in representation_ms_info:
+ segment_number += 1
+ segment_time += segment_d
+ elif 'segment_urls' in representation_ms_info and 's' in representation_ms_info:
+ # No media template
+ # Example: https://www.youtube.com/watch?v=iXZV5uAYMJI
+ # or any YouTube dashsegments video
+ fragments = []
+ s_num = 0
+ for segment_url in representation_ms_info['segment_urls']:
+ s = representation_ms_info['s'][s_num]
+ for r in range(s.get('r', 0) + 1):
+ fragments.append({
+ 'url': segment_url,
+ 'duration': float_or_none(s['d'], representation_ms_info['timescale']),
+ })
+ representation_ms_info['fragments'] = fragments
+ # NB: MPD manifest may contain direct URLs to unfragmented media.
+ # No fragments key is present in this case.
+ if 'fragments' in representation_ms_info:
f.update({
- 'segment_urls': representation_ms_info['segment_urls'],
+ 'fragments': [],
'protocol': 'http_dash_segments',
})
if 'initialization_url' in representation_ms_info:
initialization_url = representation_ms_info['initialization_url'].replace('$RepresentationID$', representation_id)
- f.update({
- 'initialization_url': initialization_url,
- })
if not f.get('url'):
f['url'] = initialization_url
+ f['fragments'].append({'url': initialization_url})
+ f['fragments'].extend(representation_ms_info['fragments'])
+ for fragment in f['fragments']:
+ fragment['url'] = combine_url(base_url, fragment['url'])
try:
existing_format = next(
fo for fo in formats
for track_tag in re.findall(r'<track[^>]+>', media_content):
track_attributes = extract_attributes(track_tag)
kind = track_attributes.get('kind')
- if not kind or kind == 'subtitles':
+ if not kind or kind in ('subtitles', 'captions'):
src = track_attributes.get('src')
if not src:
continue
media_info['subtitles'].setdefault(lang, []).append({
'url': absolute_url(src),
})
- if media_info['formats']:
+ if media_info['formats'] or media_info['subtitles']:
entries.append(media_info)
return entries
projects / youtube-dl / blobdiff