X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/3ae74f711947d73bf6627bf312edeec41cec85c3..d0938011ed52b620d91de214e425c8613a453f2f:/youtube_dl/extractor/dailymotion.py diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py index 1ea449c..3f012ae 100644 --- a/youtube_dl/extractor/dailymotion.py +++ b/youtube_dl/extractor/dailymotion.py @@ -3,18 +3,29 @@ import json import itertools from .common import InfoExtractor +from .subtitles import SubtitlesInfoExtractor + from ..utils import ( compat_urllib_request, + compat_str, get_element_by_attribute, get_element_by_id, ExtractorError, ) -class DailymotionIE(InfoExtractor): +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(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor): """Information Extractor for Dailymotion""" - _VALID_URL = r'(?i)(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/video/([^/]+)' + _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', @@ -33,10 +44,10 @@ class DailymotionIE(InfoExtractor): video_id = mobj.group(1).split('_')[0].split('?')[0] video_extension = 'mp4' + 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 @@ -55,8 +66,12 @@ class DailymotionIE(InfoExtractor): embed_url = 'http://www.dailymotion.com/embed/video/%s' % video_id embed_page = self._download_webpage(embed_url, video_id, u'Downloading embed page') - info = self._search_regex(r'var info = ({.*?}),', embed_page, 'video info') + 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 @@ -71,6 +86,12 @@ class DailymotionIE(InfoExtractor): raise ExtractorError(u'Unable to extract video URL') video_url = info[max_quality] + # subtitles + video_subtitles = self.extract_subtitles(video_id) + if self._downloader.params.get('listsubtitles', False): + self._list_available_subtitles(video_id) + return + return [{ 'id': video_id, 'url': video_url, @@ -78,33 +99,76 @@ class DailymotionIE(InfoExtractor): 'upload_date': video_upload_date, 'title': self._og_search_title(webpage), 'ext': video_extension, + 'subtitles': video_subtitles, 'thumbnail': info['thumbnail_url'] }] - -class DailymotionPlaylistIE(InfoExtractor): + def _get_available_subtitles(self, video_id): + 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)) + return {} + info = json.loads(sub_list) + if (info['total'] > 0): + sub_lang_list = dict((l['language'], l['url']) for l in info['list']) + return sub_lang_list + self._downloader.report_warning(u'video doesn\'t have subtitles') + return {} + + +class DailymotionPlaylistIE(DailymotionBaseInfoExtractor): + IE_NAME = u'dailymotion:playlist' _VALID_URL = r'(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/playlist/(?P.+?)/' _MORE_PAGES_INDICATOR = r'' + _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) + 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)) 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[^/]+)' + _MORE_PAGES_INDICATOR = r'' + _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'(.*?)