+# coding: utf-8
+from __future__ import unicode_literals
+
import re
import json
import itertools
from .common import InfoExtractor
-from .subtitles import SubtitlesInfoExtractor
-from ..utils import (
- compat_urllib_request,
+from ..compat import (
compat_str,
+ compat_urllib_request,
+)
+from ..utils import (
+ ExtractorError,
+ int_or_none,
orderedSet,
str_to_int,
- int_or_none,
- ExtractorError,
unescapeHTML,
)
+
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')
- request.add_header('Cookie', 'ff=off')
+ request.add_header('Cookie', 'family_filter=off; ff=off')
return request
-class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor):
+
+class DailymotionIE(DailymotionBaseInfoExtractor):
"""Information Extractor for Dailymotion"""
_VALID_URL = r'(?i)(?:https?://)?(?:(www|touch)\.)?dailymotion\.[a-z]{2,3}/(?:(embed|#)/)?video/(?P<id>[^/?_]+)'
_TESTS = [
{
- 'url': 'http://www.dailymotion.com/video/x33vw9_tutoriel-de-youtubeur-dl-des-video_tech',
- 'md5': '392c4b85a60a90dc4792da41ce3144eb',
+ 'url': 'https://www.dailymotion.com/video/x2iuewm_steam-machine-models-pricing-listed-on-steam-store-ign-news_videogames',
+ 'md5': '2137c41a8e78554bb09225b8eb322406',
'info_dict': {
- 'id': 'x33vw9',
+ 'id': 'x2iuewm',
'ext': 'mp4',
- 'uploader': 'Amphora Alex and Van .',
- 'title': 'Tutoriel de Youtubeur"DL DES VIDEO DE YOUTUBE"',
+ 'uploader': 'IGN',
+ 'title': 'Steam Machine Models, Pricing Listed on Steam Store - IGN News',
}
},
# Vevo video
]
def _real_extract(self, url):
- # Extract id and simplified title from URL
- mobj = re.match(self._VALID_URL, url)
-
- video_id = mobj.group('id')
-
+ video_id = self._match_id(url)
url = 'http://www.dailymotion.com/video/%s' % video_id
# Retrieve video webpage to extract further information
# It may just embed a vevo video:
m_vevo = re.search(
- r'<link rel="video_src" href="[^"]*?vevo.com[^"]*?videoId=(?P<id>[\w]*)',
+ r'<link rel="video_src" href="[^"]*?vevo.com[^"]*?video=(?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')
+ self.to_screen('Vevo video detected: %s' % vevo_id)
+ return self.url_result('vevo:%s' % vevo_id, ie='Vevo')
age_limit = self._rta_search(webpage)
video_upload_date = mobj.group(3) + mobj.group(2) + mobj.group(1)
embed_url = 'http://www.dailymotion.com/embed/video/%s' % video_id
- embed_page = self._download_webpage(embed_url, video_id,
- u'Downloading embed page')
+ embed_request = self._build_request(embed_url)
+ embed_page = self._download_webpage(
+ embed_request, video_id, 'Downloading embed page')
info = self._search_regex(r'var info = ({.*?}),$', embed_page,
- 'video info', flags=re.MULTILINE)
+ 'video info', flags=re.MULTILINE)
info = json.loads(info)
if info.get('error') is not None:
msg = 'Couldn\'t get video, Dailymotion says: %s' % info['error']['title']
'height': height,
})
if not formats:
- raise ExtractorError(u'Unable to extract video URL')
+ raise ExtractorError('Unable to extract video URL')
# subtitles
video_subtitles = self.extract_subtitles(video_id, webpage)
- if self._downloader.params.get('listsubtitles', False):
- self._list_available_subtitles(video_id, webpage)
- return
- view_count = self._search_regex(
- r'video_views_count[^>]+>\s+([\d\.,]+)', webpage, u'view count', fatal=False)
- if view_count is not None:
- view_count = str_to_int(view_count)
+ view_count = str_to_int(self._search_regex(
+ r'video_views_count[^>]+>\s+([\d\.,]+)',
+ webpage, 'view count', fatal=False))
+
+ title = self._og_search_title(webpage, default=None)
+ if title is None:
+ title = self._html_search_regex(
+ r'(?s)<span\s+id="video_title"[^>]*>(.*?)</span>', webpage,
+ 'title')
return {
- 'id': video_id,
+ 'id': video_id,
'formats': formats,
'uploader': info['owner.screenname'],
- 'upload_date': video_upload_date,
- 'title': self._og_search_title(webpage),
- 'subtitles': video_subtitles,
+ 'upload_date': video_upload_date,
+ 'title': title,
+ 'subtitles': video_subtitles,
'thumbnail': info['thumbnail_url'],
'age_limit': age_limit,
'view_count': view_count,
}
- def _get_available_subtitles(self, video_id, webpage):
+ def _get_subtitles(self, video_id, webpage):
try:
sub_list = self._download_webpage(
'https://api.dailymotion.com/video/%s/subtitles?fields=id,language,url' % video_id,
video_id, note=False)
except ExtractorError as err:
- self._downloader.report_warning(u'unable to download video subtitles: %s' % compat_str(err))
+ self._downloader.report_warning('unable to download video subtitles: %s' % compat_str(err))
return {}
info = json.loads(sub_list)
if (info['total'] > 0):
- sub_lang_list = dict((l['language'], l['url']) for l in info['list'])
+ sub_lang_list = dict((l['language'], [{'url': l['url'], 'ext': 'srt'}]) for l in info['list'])
return sub_lang_list
- self._downloader.report_warning(u'video doesn\'t have subtitles')
+ self._downloader.report_warning('video doesn\'t have subtitles')
return {}
class DailymotionPlaylistIE(DailymotionBaseInfoExtractor):
- IE_NAME = u'dailymotion:playlist'
+ IE_NAME = 'dailymotion:playlist'
_VALID_URL = r'(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/playlist/(?P<id>.+?)/'
_MORE_PAGES_INDICATOR = r'(?s)<div class="pages[^"]*">.*?<a\s+class="[^"]*?icon-arrow_right[^"]*?"'
_PAGE_TEMPLATE = 'https://www.dailymotion.com/playlist/%s/%s'
+ _TESTS = [{
+ 'url': 'http://www.dailymotion.com/playlist/xv4bw_nqtv_sport/1#video=xl8v3q',
+ 'info_dict': {
+ 'title': 'SPORT',
+ 'id': 'xv4bw_nqtv_sport',
+ },
+ 'playlist_mincount': 20,
+ }]
def _extract_entries(self, id):
video_ids = []
for pagenum in itertools.count(1):
request = self._build_request(self._PAGE_TEMPLATE % (id, pagenum))
webpage = self._download_webpage(request,
- id, u'Downloading page %s' % pagenum)
+ id, 'Downloading page %s' % pagenum)
video_ids.extend(re.findall(r'data-xid="(.+?)"', webpage))
if re.search(self._MORE_PAGES_INDICATOR, webpage) is None:
break
return [self.url_result('http://www.dailymotion.com/video/%s' % video_id, 'Dailymotion')
- for video_id in orderedSet(video_ids)]
+ for video_id in orderedSet(video_ids)]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
class DailymotionUserIE(DailymotionPlaylistIE):
- IE_NAME = u'dailymotion:user'
+ IE_NAME = 'dailymotion:user'
_VALID_URL = r'https?://(?:www\.)?dailymotion\.[a-z]{2,3}/user/(?P<user>[^/]+)'
_PAGE_TEMPLATE = 'http://www.dailymotion.com/user/%s/%s'
+ _TESTS = [{
+ 'url': 'https://www.dailymotion.com/user/nqtv',
+ 'info_dict': {
+ 'id': 'nqtv',
+ 'title': 'RĂ©mi Gaillard',
+ },
+ 'playlist_mincount': 100,
+ }]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
webpage = self._download_webpage(url, user)
full_user = unescapeHTML(self._html_search_regex(
r'<a class="nav-image" title="([^"]+)" href="/%s">' % re.escape(user),
- webpage, u'user', flags=re.DOTALL))
+ webpage, 'user'))
return {
'_type': 'playlist',