X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fdailymotion.py;h=42b20a46ddefc1e4a7e66aacd0d959a1e062618f;hb=bfc993cc9183d5f001e30267551bcdf9f0a98be9;hp=26d077eff0e2d6ea431ace761c2515b732dd48c0;hpb=ce6815aad37ecb58b91181ffe3f6e75215fc89d2;p=youtube-dl diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py index 26d077eff..42b20a46d 100644 --- a/youtube_dl/extractor/dailymotion.py +++ b/youtube_dl/extractor/dailymotion.py @@ -1,20 +1,25 @@ +# 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): @@ -24,7 +29,8 @@ class DailymotionBaseInfoExtractor(InfoExtractor): request.add_header('Cookie', '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[^/?_]+)' @@ -79,11 +85,7 @@ class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor): ] 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 @@ -95,12 +97,12 @@ class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor): # It may just embed a vevo video: m_vevo = re.search( - r'[\w]*)', + r'[\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) @@ -111,9 +113,9 @@ class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor): embed_url = 'http://www.dailymotion.com/embed/video/%s' % video_id embed_page = self._download_webpage(embed_url, video_id, - u'Downloading embed page') + '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'] @@ -136,66 +138,76 @@ class DailymotionIE(DailymotionBaseInfoExtractor, SubtitlesInfoExtractor): '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)]*>(.*?)', 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.+?)/' _MORE_PAGES_INDICATOR = r'(?s)
.*?[^/]+)' _PAGE_TEMPLATE = 'http://www.dailymotion.com/user/%s/%s' + _TESTS = [{ + 'url': 'https://www.dailymotion.com/user/nqtv', + 'info_dict': { + 'id': 'nqtv', + 'title': 'Rémi Gaillard', + }, + 'playlist_mincount': 100, + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) @@ -221,7 +241,7 @@ class DailymotionUserIE(DailymotionPlaylistIE): webpage = self._download_webpage(url, user) full_user = unescapeHTML(self._html_search_regex( r'' % re.escape(user), - webpage, u'user', flags=re.DOTALL)) + webpage, 'user')) return { '_type': 'playlist',