X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fdailymotion.py;h=62b0747a5c5a73f7628a3e500cbffada661f3212;hb=1e19ff2984f4887bbe79d9306841c3e2a96b2f37;hp=428556213239651a6994cedf22c78d62d81e0434;hpb=1faa66f005aaefe1493d150f145523a0ea34b5c2;p=youtube-dl diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py index 428556213..62b0747a5 100644 --- a/youtube_dl/extractor/dailymotion.py +++ b/youtube_dl/extractor/dailymotion.py @@ -7,15 +7,16 @@ import itertools from .common import InfoExtractor -from ..compat import compat_str from ..utils import ( - ExtractorError, determine_ext, + error_to_compat_str, + ExtractorError, int_or_none, parse_iso8601, sanitized_Request, str_to_int, unescapeHTML, + mimetype2ext, ) @@ -37,7 +38,7 @@ class DailymotionBaseInfoExtractor(InfoExtractor): class DailymotionIE(DailymotionBaseInfoExtractor): - _VALID_URL = r'(?i)(?:https?://)?(?:(www|touch)\.)?dailymotion\.[a-z]{2,3}/(?:(embed|#)/)?video/(?P[^/?_]+)' + _VALID_URL = r'(?i)(?:https?://)?(?:(www|touch)\.)?dailymotion\.[a-z]{2,3}/(?:(?:embed|swf|#)/)?video/(?P[^/?_]+)' IE_NAME = 'dailymotion' _FORMATS = [ @@ -104,9 +105,20 @@ class DailymotionIE(DailymotionBaseInfoExtractor): { 'url': 'http://www.dailymotion.com/video/x20su5f_the-power-of-nightmares-1-the-rise-of-the-politics-of-fear-bbc-2004_news', 'only_matching': True, + }, + { + 'url': 'http://www.dailymotion.com/swf/video/x3n92nf', + 'only_matching': True, } ] + @staticmethod + def _extract_urls(webpage): + # Look for embedded Dailymotion player + matches = re.findall( + r'<(?:(?:embed|iframe)[^>]+?src=|input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=)(["\'])(?P(?:https?:)?//(?:www\.)?dailymotion\.com/(?:embed|swf)/video/.+?)\1', webpage) + return list(map(lambda m: unescapeHTML(m[1]), matches)) + def _real_extract(self, url): video_id = self._match_id(url) @@ -118,10 +130,13 @@ class DailymotionIE(DailymotionBaseInfoExtractor): description = self._og_search_description(webpage) or self._html_search_meta( 'description', webpage, 'description') - view_count = str_to_int(self._search_regex( - [r']+itemprop="interactionCount"[^>]+content="UserPlays:(\d+)"', - r'video_views_count[^>]+>\s+([\d\.,]+)'], - webpage, 'view count', fatal=False)) + view_count_str = self._search_regex( + (r']+itemprop="interactionCount"[^>]+content="UserPlays:([\s\d,.]+)"', + r'video_views_count[^>]+>\s+([\s\d\,.]+)'), + webpage, 'view count', fatal=False) + if view_count_str: + view_count_str = re.sub(r'\s', '', view_count_str) + view_count = str_to_int(view_count_str) comment_count = int_or_none(self._search_regex( r']+itemprop="interactionCount"[^>]+content="UserComments:(\d+)"', webpage, 'comment count', fatal=False)) @@ -146,21 +161,19 @@ class DailymotionIE(DailymotionBaseInfoExtractor): type_ = media.get('type') if type_ == 'application/vnd.lumberjack.manifest': continue - ext = determine_ext(media_url) - if type_ == 'application/x-mpegURL' or ext == 'm3u8': - m3u8_formats = self._extract_m3u8_formats( - media_url, video_id, 'mp4', m3u8_id='hls', fatal=False) - if m3u8_formats: - formats.extend(m3u8_formats) - elif type_ == 'application/f4m' or ext == 'f4m': - f4m_formats = self._extract_f4m_formats( - media_url, video_id, preference=-1, f4m_id='hds', fatal=False) - if f4m_formats: - formats.extend(f4m_formats) + ext = mimetype2ext(type_) or determine_ext(media_url) + if ext == 'm3u8': + formats.extend(self._extract_m3u8_formats( + media_url, video_id, 'mp4', preference=-1, + m3u8_id='hls', fatal=False)) + elif ext == 'f4m': + formats.extend(self._extract_f4m_formats( + media_url, video_id, preference=-1, f4m_id='hds', fatal=False)) else: f = { 'url': media_url, - 'format_id': quality, + 'format_id': 'http-%s' % quality, + 'ext': ext, } m = re.search(r'H264-(?P\d+)x(?P\d+)', media_url) if m: @@ -278,7 +291,7 @@ class DailymotionIE(DailymotionBaseInfoExtractor): '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('unable to download video subtitles: %s' % compat_str(err)) + self._downloader.report_warning('unable to download video subtitles: %s' % error_to_compat_str(err)) return {} info = json.loads(sub_list) if (info['total'] > 0): @@ -318,7 +331,9 @@ class DailymotionPlaylistIE(DailymotionBaseInfoExtractor): for video_id in re.findall(r'data-xid="(.+?)"', webpage): if video_id not in video_ids: - yield self.url_result('http://www.dailymotion.com/video/%s' % video_id, 'Dailymotion') + yield self.url_result( + 'http://www.dailymotion.com/video/%s' % video_id, + DailymotionIE.ie_key(), video_id) video_ids.add(video_id) if re.search(self._MORE_PAGES_INDICATOR, webpage) is None: @@ -339,7 +354,7 @@ class DailymotionPlaylistIE(DailymotionBaseInfoExtractor): class DailymotionUserIE(DailymotionPlaylistIE): IE_NAME = 'dailymotion:user' - _VALID_URL = r'https?://(?:www\.)?dailymotion\.[a-z]{2,3}/(?!(?:embed|#|video|playlist)/)(?:(?:old/)?user/)?(?P[^/]+)' + _VALID_URL = r'https?://(?:www\.)?dailymotion\.[a-z]{2,3}/(?!(?:embed|swf|#|video|playlist)/)(?:(?:old/)?user/)?(?P[^/]+)' _PAGE_TEMPLATE = 'http://www.dailymotion.com/user/%s/%s' _TESTS = [{ 'url': 'https://www.dailymotion.com/user/nqtv', @@ -379,7 +394,7 @@ class DailymotionUserIE(DailymotionPlaylistIE): class DailymotionCloudIE(DailymotionBaseInfoExtractor): - _VALID_URL_PREFIX = r'http://api\.dmcloud\.net/(?:player/)?embed/' + _VALID_URL_PREFIX = r'https?://api\.dmcloud\.net/(?:player/)?embed/' _VALID_URL = r'%s[^/]+/(?P[^/?]+)' % _VALID_URL_PREFIX _VALID_EMBED_URL = r'%s[^/]+/[^\'"]+' % _VALID_URL_PREFIX @@ -395,13 +410,13 @@ class DailymotionCloudIE(DailymotionBaseInfoExtractor): }] @classmethod - def _extract_dmcloud_url(self, webpage): - mobj = re.search(r']+src=[\'"](%s)[\'"]' % self._VALID_EMBED_URL, webpage) + def _extract_dmcloud_url(cls, webpage): + mobj = re.search(r']+src=[\'"](%s)[\'"]' % cls._VALID_EMBED_URL, webpage) if mobj: return mobj.group(1) mobj = re.search( - r']+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=[\'"](%s)[\'"]' % self._VALID_EMBED_URL, + r']+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=[\'"](%s)[\'"]' % cls._VALID_EMBED_URL, webpage) if mobj: return mobj.group(1)