X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/a070911bf98c751e4425008f1335cc57c86d30fc..5a9d00a5190582368d533f12db7638e735d7f7c8:/youtube_dl/extractor/foxnews.py diff --git a/youtube_dl/extractor/foxnews.py b/youtube_dl/extractor/foxnews.py index 917f76b..dc0662f 100644 --- a/youtube_dl/extractor/foxnews.py +++ b/youtube_dl/extractor/foxnews.py @@ -1,14 +1,15 @@ from __future__ import unicode_literals +import re + +from .amp import AMPIE from .common import InfoExtractor -from ..utils import ( - parse_iso8601, - int_or_none, -) -class FoxNewsIE(InfoExtractor): - _VALID_URL = r'https?://video\.foxnews\.com/v/(?:video-embed\.html\?video_id=)?(?P\d+)' +class FoxNewsIE(AMPIE): + IE_NAME = 'foxnews' + IE_DESC = 'Fox News and Fox Business Video' + _VALID_URL = r'https?://(?Pvideo\.(?:insider\.)?fox(?:news|business)\.com)/v/(?:video-embed\.html\?video_id=)?(?P\d+)' _TESTS = [ { 'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips', @@ -17,11 +18,11 @@ class FoxNewsIE(InfoExtractor): 'id': '3937480', 'ext': 'flv', 'title': 'Frozen in Time', - 'description': 'Doctors baffled by 16-year-old girl that is the size of a toddler', + 'description': '16-year-old girl is size of toddler', 'duration': 265, 'timestamp': 1304411491, 'upload_date': '20110503', - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', }, }, { @@ -31,64 +32,109 @@ class FoxNewsIE(InfoExtractor): 'id': '3922535568001', 'ext': 'mp4', 'title': "Rep. Luis Gutierrez on if Obama's immigration plan is legal", - 'description': "Congressman discusses the president's executive action", + 'description': "Congressman discusses president's plan", 'duration': 292, 'timestamp': 1417662047, 'upload_date': '20141204', - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', + }, + 'params': { + # m3u8 download + 'skip_download': True, }, }, { 'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com', 'only_matching': True, }, + { + 'url': 'http://video.foxbusiness.com/v/4442309889001', + '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, + }, ] def _real_extract(self, url): - video_id = self._match_id(url) - - video = self._download_json( - 'http://video.foxnews.com/v/feed/video/%s.js?template=fox' % video_id, video_id) - - item = video['channel']['item'] - title = item['title'] - description = item['description'] - timestamp = parse_iso8601(item['dc-date']) - - media_group = item['media-group'] - duration = None - formats = [] - for media in media_group['media-content']: - attributes = media['@attributes'] - video_url = attributes['url'] - if video_url.endswith('.f4m'): - formats.extend(self._extract_f4m_formats(video_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', video_id)) - elif video_url.endswith('.m3u8'): - formats.extend(self._extract_m3u8_formats(video_url, video_id, 'flv')) - elif not video_url.endswith('.smil'): - duration = int_or_none(attributes.get('duration')) - formats.append({ - 'url': video_url, - 'format_id': media['media-category']['@attributes']['label'], - 'preference': 1, - 'vbr': int_or_none(attributes.get('bitrate')), - 'filesize': int_or_none(attributes.get('fileSize')) - }) - self._sort_formats(formats) - - media_thumbnail = media_group['media-thumbnail']['@attributes'] - thumbnails = [{ - 'url': media_thumbnail['url'], - 'width': int_or_none(media_thumbnail.get('width')), - 'height': int_or_none(media_thumbnail.get('height')), - }] if media_thumbnail else [] + host, video_id = re.match(self._VALID_URL, url).groups() + + info = self._extract_feed_info( + 'http://%s/v/feed/video/%s.js?template=fox' % (host, video_id)) + info['id'] = video_id + return info + + +class FoxNewsArticleIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?foxnews\.com/(?!v)([^/]+/)+(?P[a-z-]+)' + IE_NAME = 'foxnews:article' + + _TEST = { + 'url': 'http://www.foxnews.com/politics/2016/09/08/buzz-about-bud-clinton-camp-denies-claims-wore-earpiece-at-forum.html', + 'md5': '62aa5a781b308fdee212ebb6f33ae7ef', + 'info_dict': { + 'id': '5116295019001', + 'ext': 'mp4', + 'title': 'Trump and Clinton asked to defend positions on Iraq War', + 'description': 'Veterans react on \'The Kelly File\'', + 'timestamp': 1473299755, + 'upload_date': '20160908', + }, + } + + def _real_extract(self, url): + display_id = self._match_id(url) + webpage = self._download_webpage(url, display_id) + + video_id = self._html_search_regex( + r'data-video-id=([\'"])(?P[^\'"]+)\1', + webpage, 'video ID', group='id') + return self.url_result( + 'http://video.foxnews.com/v/' + video_id, + FoxNewsIE.ie_key()) + + +class FoxNewsInsiderIE(InfoExtractor): + _VALID_URL = r'https?://insider\.foxnews\.com/([^/]+/)+(?P[a-z-]+)' + IE_NAME = 'foxnews:insider' + + _TEST = { + 'url': 'http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words', + 'md5': 'a10c755e582d28120c62749b4feb4c0c', + 'info_dict': { + 'id': '5099377331001', + 'display_id': 'univ-wisconsin-student-group-pushing-silence-certain-words', + 'ext': 'mp4', + 'title': 'Student Group: Saying \'Politically Correct,\' \'Trash\' and \'Lame\' Is Offensive', + 'description': 'Is campus censorship getting out of control?', + 'timestamp': 1472168725, + 'upload_date': '20160825', + 'thumbnail': r're:^https?://.*\.jpg$', + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + 'add_ie': [FoxNewsIE.ie_key()], + } + + def _real_extract(self, url): + display_id = self._match_id(url) + + webpage = self._download_webpage(url, display_id) + + embed_url = self._html_search_meta('embedUrl', webpage, 'embed URL') + + title = self._og_search_title(webpage) + description = self._og_search_description(webpage) return { - 'id': video_id, + '_type': 'url_transparent', + 'ie_key': FoxNewsIE.ie_key(), + 'url': embed_url, + 'display_id': display_id, 'title': title, 'description': description, - 'duration': duration, - 'timestamp': timestamp, - 'formats': formats, - 'thumbnails': thumbnails, }