bug_reports_message,
clean_html,
compiled_regex_type,
+ determine_ext,
ExtractorError,
fix_xml_ampersands,
float_or_none,
self.to_screen(msg)
time.sleep(timeout)
- def _extract_f4m_formats(self, manifest_url, video_id, preference=None, f4m_id=None):
+ def _extract_f4m_formats(self, manifest_url, video_id, preference=None, f4m_id=None,
+ transform_source=lambda s: fix_xml_ampersands(s).strip()):
manifest = self._download_xml(
manifest_url, video_id, 'Downloading f4m manifest',
'Unable to download f4m manifest',
# Some manifests may be malformed, e.g. prosiebensat1 generated manifests
# (see https://github.com/rg3/youtube-dl/issues/6215#issuecomment-121704244)
- transform_source=lambda s: fix_xml_ampersands(s).strip())
+ transform_source=transform_source)
formats = []
manifest_version = '1.0'
media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media')
for i, media_el in enumerate(media_nodes):
if manifest_version == '2.0':
- media_url = media_el.attrib.get('href') or media_el.attrib['url']
+ media_url = media_el.attrib.get('href') or media_el.attrib.get('url')
+ if not media_url:
+ continue
manifest_url = (
media_url if media_url.startswith('http://') or media_url.startswith('https://')
else ('/'.join(manifest_url.split('/')[:-1]) + '/' + media_url))
+ # If media_url is itself a f4m manifest do the recursive extraction
+ # since bitrates in parent manifest (this one) and media_url manifest
+ # may differ leading to inability to resolve the format by requested
+ # bitrate in f4m downloader
+ if determine_ext(manifest_url) == 'f4m':
+ formats.extend(self._extract_f4m_formats(manifest_url, video_id, preference, f4m_id))
+ continue
tbr = int_or_none(media_el.attrib.get('bitrate'))
formats.append({
'format_id': '-'.join(filter(None, [f4m_id, compat_str(i if tbr is None else tbr)])),
def _parse_smil_video(self, video, video_id, base, rtmp_count):
src = video.get('src')
if not src:
- return ([], rtmp_count)
+ return [], rtmp_count
bitrate = int_or_none(video.get('system-bitrate') or video.get('systemBitrate'), 1000)
width = int_or_none(video.get('width'))
height = int_or_none(video.get('height'))
proto = 'http'
ext = video.get('ext')
if proto == 'm3u8':
- return (self._extract_m3u8_formats(src, video_id, ext), rtmp_count)
+ return self._extract_m3u8_formats(src, video_id, ext), rtmp_count
elif proto == 'rtmp':
rtmp_count += 1
streamer = video.get('streamer') or base