import re
import json
import itertools
-import socket
from .common import InfoExtractor
-from .subtitles import NoAutoSubtitlesInfoExtractor
+from .subtitles import SubtitlesInfoExtractor
from ..utils import (
- compat_http_client,
- compat_urllib_error,
compat_urllib_request,
compat_str,
get_element_by_attribute,
)
-class DailymotionIE(NoAutoSubtitlesInfoExtractor):
+class DailymotionIE(SubtitlesInfoExtractor):
"""Information Extractor for Dailymotion"""
_VALID_URL = r'(?i)(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/(?:embed/)?video/([^/]+)'
info = self._search_regex(r'var info = ({.*?}),$', embed_page,
'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']
+ raise ExtractorError(msg, expected=True)
# TODO: support choosing qualities
video_url = info[max_quality]
# subtitles
- video_subtitles = None
- video_webpage = None
-
- if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False):
- video_subtitles = self._extract_subtitles(video_id)
- elif self._downloader.params.get('writeautomaticsub', False):
- video_subtitles = self._request_automatic_caption(video_id, video_webpage)
-
+ video_subtitles = self.extract_subtitles(video_id)
if self._downloader.params.get('listsubtitles', False):
self._list_available_subtitles(video_id)
return
}]
def _get_available_subtitles(self, video_id):
- request = compat_urllib_request.Request('https://api.dailymotion.com/video/%s/subtitles?fields=id,language,url' % video_id)
try:
- sub_list = compat_urllib_request.urlopen(request).read().decode('utf-8')
- except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
+ 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))
return {}
info = json.loads(sub_list)
class DailymotionPlaylistIE(InfoExtractor):
+ 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>'
+ _PAGE_TEMPLATE = 'https://www.dailymotion.com/playlist/%s/%s'
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- playlist_id = mobj.group('id')
+ def _extract_entries(self, id):
video_ids = []
-
for pagenum in itertools.count(1):
- webpage = self._download_webpage('https://www.dailymotion.com/playlist/%s/%s' % (playlist_id, pagenum),
- playlist_id, u'Downloading page %s' % pagenum)
+ webpage = self._download_webpage(self._PAGE_TEMPLATE % (id, pagenum),
+ 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))
if re.search(self._MORE_PAGES_INDICATOR, webpage, re.DOTALL) is None:
break
-
- entries = [self.url_result('http://www.dailymotion.com/video/%s' % video_id, 'Dailymotion')
+ return [self.url_result('http://www.dailymotion.com/video/%s' % video_id, 'Dailymotion')
for video_id in video_ids]
+
+ def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ playlist_id = mobj.group('id')
+ webpage = self._download_webpage(url, playlist_id)
+
return {'_type': 'playlist',
'id': playlist_id,
'title': get_element_by_id(u'playlist_name', webpage),
- 'entries': entries,
+ 'entries': self._extract_entries(playlist_id),
}
+
+
+class DailymotionUserIE(DailymotionPlaylistIE):
+ IE_NAME = u'dailymotion:user'
+ _VALID_URL = r'(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/user/(?P<user>[^/]+)'
+ _MORE_PAGES_INDICATOR = r'<div class="next">.*?<a.*?href="/user/.+?".*?>.*?</a>.*?</div>'
+ _PAGE_TEMPLATE = 'http://www.dailymotion.com/user/%s/%s'
+
+ def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ user = mobj.group('user')
+ webpage = self._download_webpage(url, user)
+ full_user = self._html_search_regex(
+ r'<a class="label" href="/%s".*?>(.*?)</' % re.escape(user),
+ webpage, u'user', flags=re.DOTALL)
+
+ return {
+ '_type': 'playlist',
+ 'id': user,
+ 'title': full_user,
+ 'entries': self._extract_entries(user),
+ }