mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-28 15:46:54 +01:00
[ooyala] split long lines, fix test duration and add hdcode param to hds url
This commit is contained in:
parent
bf96b45238
commit
6101f45ef9
@ -13,6 +13,9 @@
|
|||||||
|
|
||||||
|
|
||||||
class OoyalaBaseIE(InfoExtractor):
|
class OoyalaBaseIE(InfoExtractor):
|
||||||
|
_PLAYER_BASE = 'http://player.ooyala.com/'
|
||||||
|
_CONTENT_TREE_BASE = _PLAYER_BASE + 'player_api/v1/content_tree/'
|
||||||
|
_AUTHORIZATION_URL_TEMPLATE = _PLAYER_BASE + 'sas/player_api/v1/authorization/embed_code/%s/%s?'
|
||||||
|
|
||||||
def _extract(self, content_tree_url, video_id, domain='example.org'):
|
def _extract(self, content_tree_url, video_id, domain='example.org'):
|
||||||
content_tree = self._download_json(content_tree_url, video_id)['content_tree']
|
content_tree = self._download_json(content_tree_url, video_id)['content_tree']
|
||||||
@ -31,24 +34,33 @@ def _extract(self, content_tree_url, video_id, domain='example.org'):
|
|||||||
formats = []
|
formats = []
|
||||||
for supported_format in ('mp4', 'm3u8', 'hds', 'rtmp'):
|
for supported_format in ('mp4', 'm3u8', 'hds', 'rtmp'):
|
||||||
auth_data = self._download_json(
|
auth_data = self._download_json(
|
||||||
'http://player.ooyala.com/sas/player_api/v1/authorization/embed_code/%s/%s?' % (pcode, embed_code) + compat_urllib_parse.urlencode({'domain': domain, 'supportedFormats': supported_format}),
|
self._AUTHORIZATION_URL_TEMPLATE % (pcode, embed_code) +
|
||||||
|
compat_urllib_parse.urlencode({
|
||||||
|
'domain': domain,
|
||||||
|
'supportedFormats': supported_format
|
||||||
|
}),
|
||||||
video_id, 'Downloading %s JSON' % supported_format)
|
video_id, 'Downloading %s JSON' % supported_format)
|
||||||
|
|
||||||
cur_auth_data = auth_data['authorization_data'][embed_code]
|
cur_auth_data = auth_data['authorization_data'][embed_code]
|
||||||
|
|
||||||
if cur_auth_data['authorized']:
|
if cur_auth_data['authorized']:
|
||||||
for stream in cur_auth_data['streams']:
|
for stream in cur_auth_data['streams']:
|
||||||
url = base64.b64decode(stream['url']['data'].encode('ascii')).decode('utf-8')
|
url = base64.b64decode(
|
||||||
|
stream['url']['data'].encode('ascii')).decode('utf-8')
|
||||||
if url in urls:
|
if url in urls:
|
||||||
continue
|
continue
|
||||||
urls.append(url)
|
urls.append(url)
|
||||||
delivery_type = stream['delivery_type']
|
delivery_type = stream['delivery_type']
|
||||||
if delivery_type == 'hls' or '.m3u8' in url:
|
if delivery_type == 'hls' or '.m3u8' in url:
|
||||||
formats.extend(self._extract_m3u8_formats(url, embed_code, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
|
formats.extend(self._extract_m3u8_formats(
|
||||||
|
url, embed_code, 'mp4', 'm3u8_native',
|
||||||
|
m3u8_id='hls', fatal=False))
|
||||||
elif delivery_type == 'hds' or '.f4m' in url:
|
elif delivery_type == 'hds' or '.f4m' in url:
|
||||||
formats.extend(self._extract_f4m_formats(url, embed_code, f4m_id='hds', fatal=False))
|
formats.extend(self._extract_f4m_formats(
|
||||||
|
url + '?hdcore=3.7.0', embed_code, f4m_id='hds', fatal=False))
|
||||||
elif '.smil' in url:
|
elif '.smil' in url:
|
||||||
formats.extend(self._extract_smil_formats(url, embed_code, fatal=False))
|
formats.extend(self._extract_smil_formats(
|
||||||
|
url, embed_code, fatal=False))
|
||||||
else:
|
else:
|
||||||
formats.append({
|
formats.append({
|
||||||
'url': url,
|
'url': url,
|
||||||
@ -62,7 +74,8 @@ def _extract(self, content_tree_url, video_id, domain='example.org'):
|
|||||||
'fps': float_or_none(stream.get('framerate')),
|
'fps': float_or_none(stream.get('framerate')),
|
||||||
})
|
})
|
||||||
else:
|
else:
|
||||||
raise ExtractorError('%s said: %s' % (self.IE_NAME, cur_auth_data['message']), expected=True)
|
raise ExtractorError('%s said: %s' % (
|
||||||
|
self.IE_NAME, cur_auth_data['message']), expected=True)
|
||||||
self._sort_formats(formats)
|
self._sort_formats(formats)
|
||||||
|
|
||||||
video_info['formats'] = formats
|
video_info['formats'] = formats
|
||||||
@ -120,7 +133,7 @@ def _real_extract(self, url):
|
|||||||
url, smuggled_data = unsmuggle_url(url, {})
|
url, smuggled_data = unsmuggle_url(url, {})
|
||||||
embed_code = self._match_id(url)
|
embed_code = self._match_id(url)
|
||||||
domain = smuggled_data.get('domain')
|
domain = smuggled_data.get('domain')
|
||||||
content_tree_url = 'http://player.ooyala.com/player_api/v1/content_tree/embed_code/%s/%s' % (embed_code, embed_code)
|
content_tree_url = self._CONTENT_TREE_BASE + 'embed_code/%s/%s' % (embed_code, embed_code)
|
||||||
return self._extract(content_tree_url, embed_code, domain)
|
return self._extract(content_tree_url, embed_code, domain)
|
||||||
|
|
||||||
|
|
||||||
@ -147,7 +160,7 @@ class OoyalaExternalIE(OoyalaBaseIE):
|
|||||||
'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG',
|
'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG',
|
||||||
'ext': 'mp4',
|
'ext': 'mp4',
|
||||||
'title': 'dm_140128_30for30Shorts___JudgingJewellv2',
|
'title': 'dm_140128_30for30Shorts___JudgingJewellv2',
|
||||||
'duration': 1302000,
|
'duration': 1302.0,
|
||||||
},
|
},
|
||||||
'params': {
|
'params': {
|
||||||
# m3u8 download
|
# m3u8 download
|
||||||
@ -157,5 +170,5 @@ class OoyalaExternalIE(OoyalaBaseIE):
|
|||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
partner_id, video_id, pcode = re.match(self._VALID_URL, url).groups()
|
partner_id, video_id, pcode = re.match(self._VALID_URL, url).groups()
|
||||||
content_tree_url = 'http://player.ooyala.com/player_api/v1/content_tree/external_id/%s/%s:%s' % (pcode, partner_id, video_id)
|
content_tree_url = self._CONTENT_TREE_BASE + 'external_id/%s/%s:%s' % (pcode, partner_id, video_id)
|
||||||
return self._extract(content_tree_url, video_id)
|
return self._extract(content_tree_url, video_id)
|
||||||
|
Loading…
Reference in New Issue
Block a user