X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Frutube.py;h=5b1c3577a02bb541d912faa6958533086534af01;hb=ecd1936695e73ba850d0618828b4a40d7d16c091;hp=83459c52251110c9adfa3c5cb1cfcc577f3c7354;hpb=00ff8f92a54de32eb2bde54f3c52fe4acf854ee1;p=youtube-dl diff --git a/youtube_dl/extractor/rutube.py b/youtube_dl/extractor/rutube.py index 83459c522..5b1c3577a 100644 --- a/youtube_dl/extractor/rutube.py +++ b/youtube_dl/extractor/rutube.py @@ -2,26 +2,28 @@ from __future__ import unicode_literals import re -import json import itertools from .common import InfoExtractor -from ..utils import ( +from ..compat import ( compat_str, - unified_strdate, +) +from ..utils import ( ExtractorError, + unified_strdate, ) class RutubeIE(InfoExtractor): IE_NAME = 'rutube' - IE_DESC = 'Rutube videos' + IE_DESC = 'Rutube videos' _VALID_URL = r'https?://rutube\.ru/video/(?P[\da-z]{32})' _TEST = { 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/', - 'file': '3eac3b4561676c17df9132a9a1e62e3e.mp4', 'info_dict': { + 'id': '3eac3b4561676c17df9132a9a1e62e3e', + 'ext': 'mp4', 'title': 'Раненный кенгуру забежал в аптеку', 'description': 'http://www.ntdtv.ru ', 'duration': 80, @@ -36,22 +38,22 @@ class RutubeIE(InfoExtractor): } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') - - api_response = self._download_webpage('http://rutube.ru/api/video/%s/?format=json' % video_id, - video_id, 'Downloading video JSON') - video = json.loads(api_response) - - api_response = self._download_webpage('http://rutube.ru/api/play/trackinfo/%s/?format=json' % video_id, - video_id, 'Downloading trackinfo JSON') - trackinfo = json.loads(api_response) - + video_id = self._match_id(url) + video = self._download_json( + 'http://rutube.ru/api/video/%s/?format=json' % video_id, + video_id, 'Downloading video JSON') + # Some videos don't have the author field - author = trackinfo.get('author') or {} - m3u8_url = trackinfo['video_balancer'].get('m3u8') + author = video.get('author') or {} + + options = self._download_json( + 'http://rutube.ru/api/play/options/%s/?format=json' % video_id, + video_id, 'Downloading options JSON') + + m3u8_url = options['video_balancer'].get('m3u8') if m3u8_url is None: raise ExtractorError('Couldn\'t find m3u8 manifest url') + formats = self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4') return { 'id': video['id'], @@ -59,8 +61,7 @@ class RutubeIE(InfoExtractor): 'description': video['description'], 'duration': video['duration'], 'view_count': video['hits'], - 'url': m3u8_url, - 'ext': 'mp4', + 'formats': formats, 'thumbnail': video['thumbnail_url'], 'uploader': author.get('name'), 'uploader_id': compat_str(author['id']) if author else None, @@ -69,25 +70,63 @@ class RutubeIE(InfoExtractor): } +class RutubeEmbedIE(InfoExtractor): + IE_NAME = 'rutube:embed' + IE_DESC = 'Rutube embedded videos' + _VALID_URL = 'https?://rutube\.ru/video/embed/(?P[0-9]+)' + + _TEST = { + 'url': 'http://rutube.ru/video/embed/6722881?vk_puid37=&vk_puid38=', + 'info_dict': { + 'id': 'a10e53b86e8f349080f718582ce4c661', + 'ext': 'mp4', + 'upload_date': '20131223', + 'uploader_id': '297833', + 'description': 'Видео группы ★http://vk.com/foxkidsreset★ музей Fox Kids и Jetix

восстановлено и сделано в шикоформате subziro89 http://vk.com/subziro89', + 'uploader': 'subziro89 ILya', + 'title': 'Мистический городок Эйри в Индиан 5 серия озвучка subziro89', + }, + 'params': { + 'skip_download': 'Requires ffmpeg', + }, + } + + def _real_extract(self, url): + embed_id = self._match_id(url) + webpage = self._download_webpage(url, embed_id) + + canonical_url = self._html_search_regex( + r'\d+)' + _TESTS = [{ + 'url': 'http://rutube.ru/tags/video/1800/', + 'info_dict': { + 'id': '1800', + }, + 'playlist_mincount': 68, + }] _PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json' def _extract_videos(self, channel_id, channel_title=None): entries = [] for pagenum in itertools.count(1): - api_response = self._download_webpage(self._PAGE_TEMPLATE % (channel_id, pagenum), - channel_id, 'Downloading page %s' % pagenum) - page = json.loads(api_response) + page = self._download_json( + self._PAGE_TEMPLATE % (channel_id, pagenum), + channel_id, 'Downloading page %s' % pagenum) results = page['results'] - if len(results) == 0: - break; + if not results: + break entries.extend(self.url_result(result['video_url'], 'Rutube') for result in results) - if page['has_next'] is False: - break; + if not page['has_next']: + break return self.playlist_result(entries, channel_id, channel_title) def _real_extract(self, url): @@ -98,18 +137,18 @@ class RutubeChannelIE(InfoExtractor): class RutubeMovieIE(RutubeChannelIE): IE_NAME = 'rutube:movie' - IE_DESC = 'Rutube movies' + IE_DESC = 'Rutube movies' _VALID_URL = r'http://rutube\.ru/metainfo/tv/(?P\d+)' + _TESTS = [] _MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json' _PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json' def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - movie_id = mobj.group('id') - api_response = self._download_webpage(self._MOVIE_TEMPLATE % movie_id, movie_id, - 'Downloading movie JSON') - movie = json.loads(api_response) + movie_id = self._match_id(url) + movie = self._download_json( + self._MOVIE_TEMPLATE % movie_id, movie_id, + 'Downloading movie JSON') movie_name = movie['name'] return self._extract_videos(movie_id, movie_name) @@ -118,5 +157,12 @@ class RutubePersonIE(RutubeChannelIE): IE_NAME = 'rutube:person' IE_DESC = 'Rutube person videos' _VALID_URL = r'http://rutube\.ru/video/person/(?P\d+)' + _TESTS = [{ + 'url': 'http://rutube.ru/video/person/313878/', + 'info_dict': { + 'id': '313878', + }, + 'playlist_mincount': 37, + }] - _PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json' \ No newline at end of file + _PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json'