[internetvideoarchive] extract all formats

This commit is contained in:
Remita Amine 2016-09-04 11:42:15 +01:00
parent 919cf1a62f
commit 0def758782
2 changed files with 19 additions and 10 deletions

View file

@ -1163,13 +1163,6 @@ def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
m3u8_id=None, note=None, errnote=None,
fatal=True, live=False):
formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)]
format_url = lambda u: (
u
if re.match(r'^https?://', u)
else compat_urlparse.urljoin(m3u8_url, u))
res = self._download_webpage_handle(
m3u8_url, video_id,
note=note or 'Downloading m3u8 information',
@ -1180,6 +1173,13 @@ def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
m3u8_doc, urlh = res
m3u8_url = urlh.geturl()
formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)]
format_url = lambda u: (
u
if re.match(r'^https?://', u)
else compat_urlparse.urljoin(m3u8_url, u))
# We should try extracting formats only from master playlists [1], i.e.
# playlists that describe available qualities. On the other hand media
# playlists [2] should be returned as is since they contain just the media

View file

@ -48,13 +48,23 @@ def _real_extract(self, url):
# There are multiple videos in the playlist whlie only the first one
# matches the video played in browsers
video_info = configuration['playlist'][0]
title = video_info['title']
formats = []
for source in video_info['sources']:
file_url = source['file']
if determine_ext(file_url) == 'm3u8':
formats.extend(self._extract_m3u8_formats(
file_url, video_id, ext='mp4', m3u8_id='hls'))
m3u8_formats = self._extract_m3u8_formats(
file_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
if m3u8_formats:
formats.extend(m3u8_formats)
file_url = m3u8_formats[0]['url']
formats.extend(self._extract_f4m_formats(
file_url.replace('.m3u8', '.f4m'),
video_id, f4m_id='hds', fatal=False))
formats.extend(self._extract_mpd_formats(
file_url.replace('.m3u8', '.mpd'),
video_id, mpd_id='dash', fatal=False))
else:
a_format = {
'url': file_url,
@ -70,7 +80,6 @@ def _real_extract(self, url):
self._sort_formats(formats)
title = video_info['title']
description = video_info.get('description')
thumbnail = video_info.get('image')
else: