X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fceskatelevize.py;h=6f7b2a70de302aa23a80e24da4126d237e653798;hb=5c2266df4b9aeb7881ed8c026a038e2a25e43734;hp=345ac5e859d7e6b66af789f99beabdf9f1c36cc7;hpb=c067545c175bb421fcfc255ae7df7a2fff1c5638;p=youtube-dl diff --git a/youtube_dl/extractor/ceskatelevize.py b/youtube_dl/extractor/ceskatelevize.py index 345ac5e85..6f7b2a70d 100644 --- a/youtube_dl/extractor/ceskatelevize.py +++ b/youtube_dl/extractor/ceskatelevize.py @@ -3,69 +3,95 @@ from __future__ import unicode_literals import re -from .subtitles import SubtitlesInfoExtractor +from .common import InfoExtractor from ..compat import ( - compat_urllib_request, compat_urllib_parse, + compat_urllib_parse_unquote, compat_urllib_parse_urlparse, ) from ..utils import ( ExtractorError, float_or_none, - HEADRequest, + sanitized_Request, ) -class CeskaTelevizeIE(SubtitlesInfoExtractor): - _VALID_URL = r'https?://www\.ceskatelevize\.cz/(porady|ivysilani)/(.+/)?(?P[^?#]+)' - - _TESTS = [ - { - 'url': 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220', +class CeskaTelevizeIE(InfoExtractor): + _VALID_URL = r'https?://www\.ceskatelevize\.cz/(porady|ivysilani)/(?:[^/]+/)*(?P[^/#?]+)/*(?:[#?].*)?$' + _TESTS = [{ + 'url': 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220', + 'info_dict': { + 'id': '61924494876951776', + 'ext': 'mp4', + 'title': 'Hyde Park Civilizace', + 'description': 'md5:fe93f6eda372d150759d11644ebbfb4a', + 'thumbnail': 're:^https?://.*\.jpg', + 'duration': 3350, + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, { + 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/bonus/14716-zpevacka-z-duparny-bobina', + 'info_dict': { + 'id': '61924494876844374', + 'ext': 'mp4', + 'title': 'První republika: Zpěvačka z Dupárny Bobina', + 'description': 'Sága mapující atmosféru první republiky od r. 1918 do r. 1945.', + 'thumbnail': 're:^https?://.*\.jpg', + 'duration': 88.4, + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, { + # video with 18+ caution trailer + 'url': 'http://www.ceskatelevize.cz/porady/10520528904-queer/215562210900007-bogotart/', + 'info_dict': { + 'id': '215562210900007-bogotart', + 'title': 'Queer: Bogotart', + 'description': 'Alternativní průvodce současným queer světem', + }, + 'playlist': [{ 'info_dict': { - 'id': '214411058091220', + 'id': '61924494876844842', 'ext': 'mp4', - 'title': 'Hyde Park Civilizace', - 'description': 'Věda a současná civilizace. Interaktivní pořad - prostor pro vaše otázky a komentáře', - 'thumbnail': 're:^https?://.*\.jpg', - 'duration': 3350, - }, - 'params': { - # m3u8 download - 'skip_download': True, + 'title': 'Queer: Bogotart (Varování 18+)', + 'duration': 10.2, }, - }, - { - 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/bonus/14716-zpevacka-z-duparny-bobina', + }, { 'info_dict': { - 'id': '14716', + 'id': '61924494877068022', 'ext': 'mp4', - 'title': 'První republika: Zpěvačka z Dupárny Bobina', - 'description': 'Sága mapující atmosféru první republiky od r. 1918 do r. 1945.', + 'title': 'Queer: Bogotart (Queer)', 'thumbnail': 're:^https?://.*\.jpg', - 'duration': 88.4, - }, - 'params': { - # m3u8 download - 'skip_download': True, + 'duration': 1558.3, }, + }], + 'params': { + # m3u8 download + 'skip_download': True, }, - ] + }] def _real_extract(self, url): url = url.replace('/porady/', '/ivysilani/').replace('/video/', '') mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') + playlist_id = mobj.group('id') - webpage = self._download_webpage(url, video_id) + webpage = self._download_webpage(url, playlist_id) NOT_AVAILABLE_STRING = 'This content is not available at your territory due to limited copyright.' if '%s

' % NOT_AVAILABLE_STRING in webpage: raise ExtractorError(NOT_AVAILABLE_STRING, expected=True) - typ = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":"(.+?)","id":".+?"\}\],', webpage, 'type') - episode_id = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":".+?","id":"(.+?)"\}\],', webpage, 'episode_id') + typ = self._html_search_regex( + r'getPlaylistUrl\(\[\{"type":"(.+?)","id":".+?"\}\],', webpage, 'type') + episode_id = self._html_search_regex( + r'getPlaylistUrl\(\[\{"type":".+?","id":"(.+?)"\}\],', webpage, 'episode_id') data = { 'playlist[0][type]': typ, @@ -74,7 +100,7 @@ class CeskaTelevizeIE(SubtitlesInfoExtractor): 'requestSource': 'iVysilani', } - req = compat_urllib_request.Request( + req = sanitized_Request( 'http://www.ceskatelevize.cz/ivysilani/ajax/get-client-playlist', data=compat_urllib_parse.urlencode(data)) @@ -83,54 +109,70 @@ class CeskaTelevizeIE(SubtitlesInfoExtractor): req.add_header('X-Requested-With', 'XMLHttpRequest') req.add_header('Referer', url) - playlistpage = self._download_json(req, video_id) + playlistpage = self._download_json(req, playlist_id) playlist_url = playlistpage['url'] if playlist_url == 'error_region': raise ExtractorError(NOT_AVAILABLE_STRING, expected=True) - req = compat_urllib_request.Request(compat_urllib_parse.unquote(playlist_url)) + req = sanitized_Request(compat_urllib_parse_unquote(playlist_url)) req.add_header('Referer', url) - playlist = self._download_json(req, video_id) - - item = playlist['playlist'][0] - formats = [] - for format_id, stream_url in item['streamUrls'].items(): - formats.extend(self._extract_m3u8_formats(stream_url, video_id, 'mp4')) - self._sort_formats(formats) - - title = self._og_search_title(webpage) - description = self._og_search_description(webpage) - duration = float_or_none(item.get('duration')) - thumbnail = item.get('previewImageUrl') - - # subtitles - subtitles = self.extract_subtitles(video_id, webpage) - - if self._downloader.params.get('listsubtitles', False): - self._list_available_subtitles(video_id, webpage) - return - - subtitles = self._fix_subtitles(self.extract_subtitles(video_id, webpage)) - + playlist_title = self._og_search_title(webpage) + playlist_description = self._og_search_description(webpage) + + playlist = self._download_json(req, playlist_id)['playlist'] + playlist_len = len(playlist) + + entries = [] + for item in playlist: + formats = [] + for format_id, stream_url in item['streamUrls'].items(): + formats.extend(self._extract_m3u8_formats( + stream_url, playlist_id, 'mp4', entry_protocol='m3u8_native')) + self._sort_formats(formats) + + item_id = item.get('id') or item['assetId'] + title = item['title'] + + duration = float_or_none(item.get('duration')) + thumbnail = item.get('previewImageUrl') + + subtitles = {} + if item.get('type') == 'VOD': + subs = item.get('subtitles') + if subs: + subtitles = self.extract_subtitles(episode_id, subs) + + entries.append({ + 'id': item_id, + 'title': playlist_title if playlist_len == 1 else '%s (%s)' % (playlist_title, title), + 'description': playlist_description if playlist_len == 1 else None, + 'thumbnail': thumbnail, + 'duration': duration, + 'formats': formats, + 'subtitles': subtitles, + }) + + return self.playlist_result(entries, playlist_id, playlist_title, playlist_description) + + def _get_subtitles(self, episode_id, subs): + original_subtitles = self._download_webpage( + subs[0]['url'], episode_id, 'Downloading subtitles') + srt_subs = self._fix_subtitles(original_subtitles) return { - 'id': episode_id, - 'title': title, - 'description': description, - 'thumbnail': thumbnail, - 'duration': duration, - 'formats': formats, - 'subtitles': subtitles, + 'cs': [{ + 'ext': 'srt', + 'data': srt_subs, + }] } - def _fix_subtitles(self, subtitles): - """ Convert milisecond-based subtitles to SRT """ - if subtitles is None: - return subtitles # subtitles not requested + @staticmethod + def _fix_subtitles(subtitles): + """ Convert millisecond-based subtitles to SRT """ def _msectotimecode(msec): - """ Helper utility to convert miliseconds to timecode """ + """ Helper utility to convert milliseconds to timecode """ components = [] for divider in [1000, 60, 60, 100]: components.append(msec % divider) @@ -139,27 +181,12 @@ class CeskaTelevizeIE(SubtitlesInfoExtractor): def _fix_subtitle(subtitle): for line in subtitle.splitlines(): - m = re.match(r"^ *([0-9]+); *([0-9]+) +([0-9]+) *$", line) + m = re.match(r"^\s*([0-9]+);\s*([0-9]+)\s+([0-9]+)\s*$", line) if m: yield m.group(1) start, stop = (_msectotimecode(int(t)) for t in m.groups()[1:]) - yield "{} --> {}".format(start, stop) + yield "{0} --> {1}".format(start, stop) else: yield line - fixed_subtitles = {} - for k, v in subtitles.items(): - fixed_subtitles[k] = "\r\n".join(_fix_subtitle(v)) - return fixed_subtitles - - def _get_available_subtitles(self, video_id, webpage): - video_id = video_id.partition('-')[0] - url = 'http://imgct.ceskatelevize.cz/cache/data/ivysilani/' \ - 'subtitles/{}/{}/sub.txt'.format(video_id[:3], video_id) - req = HEADRequest(url) - sub = self._request_webpage( - req, video_id, - note="Checking subtitles", - errnote="No subtitles found", - fatal=False) - return {'cs': url} if sub else {} + return "\r\n".join(_fix_subtitle(subtitles))