MarcoBuster/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-30 16:42:57 +01:00
Code Issues Projects Releases Wiki Activity

[extractor/foxnews] Fix extractors (#7222)

Browse Source 
Closes #6050
Authored by: bashonly
This commit is contained in:
bashonly 2023-06-04 08:37:59 -05:00 committed by GitHub
parent 5ee9a7d6e1
commit 97d60ad8cd
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 62 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
yt_dlp/extractor/amp.py
View File

@ -5,6 +5,7 @@
int_or_none, int_or_none,
mimetype2ext, mimetype2ext,
parse_iso8601, parse_iso8601,
strip_jsonp,
unified_timestamp, unified_timestamp,
url_or_none, url_or_none,
) )
@ -15,7 +16,7 @@ class AMPIE(InfoExtractor): # XXX: Conventionally, base classes should end with
def _extract_feed_info(self, url): def _extract_feed_info(self, url):
feed = self._download_json( feed = self._download_json(
url, None, 'Downloading Akamai AMP feed', url, None, 'Downloading Akamai AMP feed',
'Unable to download Akamai AMP feed') 'Unable to download Akamai AMP feed', transform_source=strip_jsonp)
item = feed.get('channel', {}).get('item') item = feed.get('channel', {}).get('item')
if not item: if not item:
raise ExtractorError('%s said: %s' % (self.IE_NAME, feed['error'])) raise ExtractorError('%s said: %s' % (self.IE_NAME, feed['error']))
@ -73,8 +74,10 @@ def get_media_node(name, default=None):
media_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', media_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124',
video_id, f4m_id='hds', fatal=False)) video_id, f4m_id='hds', fatal=False))
elif ext == 'm3u8': elif ext == 'm3u8':
formats.extend(self._extract_m3u8_formats( fmts, subs = self._extract_m3u8_formats_and_subtitles(
media_url, video_id, 'mp4', m3u8_id='hls', fatal=False)) media_url, video_id, 'mp4', m3u8_id='hls', fatal=False)
formats.extend(fmts)
self._merge_subtitles(subs, target=subtitles)
else: else:
formats.append({ formats.append({
'format_id': media_data.get('media-category', {}).get('@attributes', {}).get('label'), 'format_id': media_data.get('media-category', {}).get('@attributes', {}).get('label'),

77
yt_dlp/extractor/foxnews.py
View File

@ -7,8 +7,37 @@
class FoxNewsIE(AMPIE): class FoxNewsIE(AMPIE):
IE_NAME = 'foxnews' IE_NAME = 'foxnews'
IE_DESC = 'Fox News and Fox Business Video' IE_DESC = 'Fox News and Fox Business Video'
_VALID_URL = r'https?://(?P<host>video\.(?:insider\.)?fox(?:news|business)\.com)/v/(?:video-embed\.html\?video_id=)?(?P<id>\d+)' _VALID_URL = r'https?://video\.(?:insider\.)?fox(?:news|business)\.com/v/(?:video-embed\.html\?video_id=)?(?P<id>\d+)'
_TESTS = [ _TESTS = [
{
'url': 'https://video.foxnews.com/v/6320653836112',
'info_dict': {
'id': '6320653836112',
'ext': 'mp4',
'title': 'Tucker Carlson joins \'Gutfeld!\' to discuss his new documentary',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 404,
'upload_date': '20230217',
'description': 'md5:858a8a36f59e9ca897d758855bcdfa02',
'timestamp': 1676611344.0,
},
'params': {'skip_download': 'm3u8'},
},
{
# From http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words
'url': 'http://video.insider.foxnews.com/v/video-embed.html?video_id=5099377331001&autoplay=true&share_url=http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words&share_title=Student%20Group:%20Saying%20%27Politically%20Correct,%27%20%27Trash%27%20and%20%27Lame%27%20Is%20Offensive&share=true',
'info_dict': {
'id': '5099377331001',
'ext': 'mp4',
'title': '82416_censoring',
'description': '82416_censoring',
'upload_date': '20160826',
'timestamp': 1472169708.0,
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 521,
},
'params': {'skip_download': 'm3u8'},
},
{ {
'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips', 'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips',
'md5': '32aaded6ba3ef0d1c04e238d01031e5e', 'md5': '32aaded6ba3ef0d1c04e238d01031e5e',
@ -22,6 +51,7 @@ class FoxNewsIE(AMPIE):
'upload_date': '20110503', 'upload_date': '20110503',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*\.jpg$',
}, },
'skip': '404 page',
}, },
{ {
'url': 'http://video.foxnews.com/v/3922535568001/rep-luis-gutierrez-on-if-obamas-immigration-plan-is-legal/#sp=show-clips', 'url': 'http://video.foxnews.com/v/3922535568001/rep-luis-gutierrez-on-if-obamas-immigration-plan-is-legal/#sp=show-clips',
@ -36,10 +66,7 @@ class FoxNewsIE(AMPIE):
'upload_date': '20141204', 'upload_date': '20141204',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*\.jpg$',
}, },
'params': { 'skip': 'm3u8 HTTP error 400 in web browser',
# m3u8 download
'skip_download': True,
},
}, },
{ {
'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com', 'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com',
@ -49,11 +76,6 @@ class FoxNewsIE(AMPIE):
'url': 'http://video.foxbusiness.com/v/4442309889001', 'url': 'http://video.foxbusiness.com/v/4442309889001',
'only_matching': True, 'only_matching': True,
}, },
{
# From http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words
'url': 'http://video.insider.foxnews.com/v/video-embed.html?video_id=5099377331001&autoplay=true&share_url=http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words&share_title=Student%20Group:%20Saying%20%27Politically%20Correct,%27%20%27Trash%27%20and%20%27Lame%27%20Is%20Offensive&share=true',
'only_matching': True,
},
] ]
@classmethod @classmethod
@ -67,10 +89,10 @@ def _extract_embed_urls(cls, url, webpage):
yield f'https://video.foxnews.com/v/video-embed.html?video_id={mobj.group("video_id")}' yield f'https://video.foxnews.com/v/video-embed.html?video_id={mobj.group("video_id")}'
def _real_extract(self, url): def _real_extract(self, url):
host, video_id = self._match_valid_url(url).groups() video_id = self._match_id(url)
info = self._extract_feed_info( info = self._extract_feed_info(
'http://%s/v/feed/video/%s.js?template=fox' % (host, video_id)) f'https://api.foxnews.com/v3/video-player/{video_id}?callback=uid_{video_id}')
info['id'] = video_id info['id'] = video_id
return info return info
@ -78,6 +100,19 @@ def _real_extract(self, url):
class FoxNewsVideoIE(InfoExtractor): class FoxNewsVideoIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?foxnews\.com/video/(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.)?foxnews\.com/video/(?P<id>\d+)'
_TESTS = [{ _TESTS = [{
'url': 'https://www.foxnews.com/video/6328632286112',
'info_dict': {
'id': '6328632286112',
'ext': 'mp4',
'title': 'Review: 2023 Toyota Prius Prime',
'duration': 155,
'thumbnail': r're:^https://.+\.jpg$',
'timestamp': 1685720177.0,
'upload_date': '20230602',
'description': 'md5:b69aafb125b41c1402e9744f53d6edc4',
},
'params': {'skip_download': 'm3u8'},
}, {
'url': 'https://www.foxnews.com/video/6313058664112', 'url': 'https://www.foxnews.com/video/6313058664112',
'info_dict': { 'info_dict': {
'id': '6313058664112', 'id': '6313058664112',
@ -89,8 +124,7 @@ class FoxNewsVideoIE(InfoExtractor):
'title': 'Gutfeld! - Thursday, September 29', 'title': 'Gutfeld! - Thursday, September 29',
'timestamp': 1664527538, 'timestamp': 1664527538,
}, },
'expected_warnings': ['Ignoring subtitle tracks'], 'skip': '404 page',
'params': {'skip_download': 'm3u8'},
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -104,19 +138,22 @@ class FoxNewsArticleIE(InfoExtractor):
_TESTS = [{ _TESTS = [{
# data-video-id # data-video-id
'url': 'http://www.foxnews.com/politics/2016/09/08/buzz-about-bud-clinton-camp-denies-claims-wore-earpiece-at-forum.html', 'url': 'https://www.foxnews.com/politics/2016/09/08/buzz-about-bud-clinton-camp-denies-claims-wore-earpiece-at-forum.html',
'md5': '83d44e1aff1433e7a29a7b537d1700b5', 'md5': 'd2dd6ce809cedeefa96460e964821437',
'info_dict': { 'info_dict': {
'id': '5116295019001', 'id': '5116295019001',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Trump and Clinton asked to defend positions on Iraq War', 'title': 'Trump and Clinton asked to defend positions on Iraq War',
'description': 'Veterans react on \'The Kelly File\'', 'description': 'Veterans and Fox News host Dana Perino react on \'The Kelly File\' to NBC\'s presidential forum',
'timestamp': 1473301045, 'timestamp': 1473301045,
'upload_date': '20160908', 'upload_date': '20160908',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 426,
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
# iframe embed # iframe embed
'url': 'http://www.foxnews.com/us/2018/03/09/parkland-survivor-kyle-kashuv-on-meeting-trump-his-app-to-prevent-another-school-shooting.amp.html?__twitter_impression=true', 'url': 'https://www.foxnews.com/us/2018/03/09/parkland-survivor-kyle-kashuv-on-meeting-trump-his-app-to-prevent-another-school-shooting.amp.html?__twitter_impression=true',
'info_dict': { 'info_dict': {
'id': '5748266721001', 'id': '5748266721001',
'ext': 'flv', 'ext': 'flv',
@ -127,9 +164,7 @@ class FoxNewsArticleIE(InfoExtractor):
'timestamp': 1520594670, 'timestamp': 1520594670,
'upload_date': '20180309', 'upload_date': '20180309',
}, },
'params': { 'skip': '404 page',
'skip_download': True,
},
}, { }, {
'url': 'http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words', 'url': 'http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words',
'only_matching': True, 'only_matching': True,