compat_str,
get_element_by_attribute,
get_element_by_id,
+ orderedSet,
ExtractorError,
)
+class DailymotionBaseInfoExtractor(InfoExtractor):
+ @staticmethod
+ def _build_request(url):
+ """Build a request with the family filter disabled"""
+ request = compat_urllib_request.Request(url)
+ request.add_header('Cookie', 'family_filter=off')
+ return request
-class DailymotionIE(SubtitlesInfoExtractor):
+class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor):
"""Information Extractor for Dailymotion"""
_VALID_URL = r'(?i)(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/(?:embed/)?video/([^/]+)'
IE_NAME = u'dailymotion'
- _TEST = {
- u'url': u'http://www.dailymotion.com/video/x33vw9_tutoriel-de-youtubeur-dl-des-video_tech',
- u'file': u'x33vw9.mp4',
- u'md5': u'392c4b85a60a90dc4792da41ce3144eb',
- u'info_dict': {
- u"uploader": u"Amphora Alex and Van .",
- u"title": u"Tutoriel de Youtubeur\"DL DES VIDEO DE YOUTUBE\""
- }
- }
+ _TESTS = [
+ {
+ u'url': u'http://www.dailymotion.com/video/x33vw9_tutoriel-de-youtubeur-dl-des-video_tech',
+ u'file': u'x33vw9.mp4',
+ u'md5': u'392c4b85a60a90dc4792da41ce3144eb',
+ u'info_dict': {
+ u"uploader": u"Amphora Alex and Van .",
+ u"title": u"Tutoriel de Youtubeur\"DL DES VIDEO DE YOUTUBE\""
+ }
+ },
+ # Vevo video
+ {
+ u'url': u'http://www.dailymotion.com/video/x149uew_katy-perry-roar-official_musi',
+ u'file': u'USUV71301934.mp4',
+ u'info_dict': {
+ u'title': u'Roar (Official)',
+ u'uploader': u'Katy Perry',
+ u'upload_date': u'20130905',
+ },
+ u'params': {
+ u'skip_download': True,
+ },
+ u'skip': u'VEVO is only available in some countries',
+ },
+ ]
def _real_extract(self, url):
# Extract id and simplified title from URL
url = 'http://www.dailymotion.com/video/%s' % video_id
# Retrieve video webpage to extract further information
- request = compat_urllib_request.Request(url)
- request.add_header('Cookie', 'family_filter=off')
+ request = self._build_request(url)
webpage = self._download_webpage(request, video_id)
# Extract URL, uploader and title from webpage
self.report_extraction(video_id)
+ # It may just embed a vevo video:
+ m_vevo = re.search(
+ r'<link rel="video_src" href="[^"]*?vevo.com[^"]*?videoId=(?P<id>[\w]*)',
+ webpage)
+ if m_vevo is not None:
+ vevo_id = m_vevo.group('id')
+ self.to_screen(u'Vevo video detected: %s' % vevo_id)
+ return self.url_result(u'vevo:%s' % vevo_id, ie='Vevo')
+
video_uploader = self._search_regex([r'(?im)<span class="owner[^\"]+?">[^<]+?<a [^>]+?>([^<]+?)</a>',
# Looking for official user
r'<(?:span|a) .*?rel="author".*?>([^<]+?)</'],
video_url = info[max_quality]
# subtitles
- video_subtitles = self.extract_subtitles(video_id)
+ video_subtitles = self.extract_subtitles(video_id, webpage)
if self._downloader.params.get('listsubtitles', False):
- self._list_available_subtitles(video_id)
+ self._list_available_subtitles(video_id, webpage)
return
return [{
'thumbnail': info['thumbnail_url']
}]
- def _get_available_subtitles(self, video_id):
+ def _get_available_subtitles(self, video_id, webpage):
try:
sub_list = self._download_webpage(
'https://api.dailymotion.com/video/%s/subtitles?fields=id,language,url' % video_id,
return {}
-class DailymotionPlaylistIE(InfoExtractor):
+class DailymotionPlaylistIE(DailymotionBaseInfoExtractor):
IE_NAME = u'dailymotion:playlist'
_VALID_URL = r'(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/playlist/(?P<id>.+?)/'
_MORE_PAGES_INDICATOR = r'<div class="next">.*?<a.*?href="/playlist/.+?".*?>.*?</a>.*?</div>'
def _extract_entries(self, id):
video_ids = []
for pagenum in itertools.count(1):
- webpage = self._download_webpage(self._PAGE_TEMPLATE % (id, pagenum),
+ request = self._build_request(self._PAGE_TEMPLATE % (id, pagenum))
+ webpage = self._download_webpage(request,
id, u'Downloading page %s' % pagenum)
playlist_el = get_element_by_attribute(u'class', u'video_list', webpage)
- video_ids.extend(re.findall(r'data-id="(.+?)" data-ext-id', playlist_el))
+ video_ids.extend(re.findall(r'data-id="(.+?)"', playlist_el))
if re.search(self._MORE_PAGES_INDICATOR, webpage, re.DOTALL) is None:
break
return [self.url_result('http://www.dailymotion.com/video/%s' % video_id, 'Dailymotion')
- for video_id in video_ids]
+ for video_id in orderedSet(video_ids)]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)