]>
Commit | Line | Data |
---|---|---|
1074982e S |
1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
4 | from ..utils import ( | |
5 | ExtractorError, | |
50e93e03 | 6 | int_or_none, |
7 | str_to_int | |
1074982e S |
8 | ) |
9 | ||
10 | ||
a7515ec2 S |
11 | class RUTVIE(InfoExtractor): |
12 | IE_DESC = 'RUTV.RU' | |
342f630d | 13 | _VALID_URL = r'''(?x) |
624bd010 S |
14 | https?:// |
15 | (?:test)?player\.(?:rutv\.ru|vgtrk\.com)/ | |
16 | (?P<path> | |
17 | flash\d+v/container\.swf\?id=| | |
18 | iframe/(?P<type>swf|video|live)/id/| | |
19 | index/iframe/cast_id/ | |
20 | ) | |
21 | (?P<id>\d+) | |
22 | ''' | |
1074982e S |
23 | |
24 | _TESTS = [ | |
25 | { | |
a7515ec2 | 26 | 'url': 'http://player.rutv.ru/flash2v/container.swf?id=774471&sid=kultura&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972347/video_id/978186/brand_id/31724', |
1074982e | 27 | 'info_dict': { |
a7515ec2 | 28 | 'id': '774471', |
d37c07c5 | 29 | 'ext': 'mp4', |
a7515ec2 S |
30 | 'title': 'Монологи на все времена', |
31 | 'description': 'md5:18d8b5e6a41fb1faa53819471852d5d5', | |
32 | 'duration': 2906, | |
d37c07c5 S |
33 | }, |
34 | 'params': { | |
35 | # m3u8 download | |
36 | 'skip_download': True, | |
37 | }, | |
38 | }, | |
1074982e | 39 | { |
a7515ec2 | 40 | 'url': 'https://player.vgtrk.com/flash2v/container.swf?id=774016&sid=russiatv&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972098/video_id/977760/brand_id/57638', |
1074982e | 41 | 'info_dict': { |
a7515ec2 | 42 | 'id': '774016', |
1074982e | 43 | 'ext': 'mp4', |
a7515ec2 S |
44 | 'title': 'Чужой в семье Сталина', |
45 | 'description': '', | |
46 | 'duration': 2539, | |
1074982e S |
47 | }, |
48 | 'params': { | |
49 | # m3u8 download | |
50 | 'skip_download': True, | |
51 | }, | |
52 | }, | |
d37c07c5 | 53 | { |
a7515ec2 | 54 | 'url': 'http://player.rutv.ru/iframe/swf/id/766888/sid/hitech/?acc_video_id=4000', |
d37c07c5 S |
55 | 'info_dict': { |
56 | 'id': '766888', | |
57 | 'ext': 'mp4', | |
58 | 'title': 'Вести.net: интернет-гиганты начали перетягивание программных "одеял"', | |
59 | 'description': 'md5:65ddd47f9830c4f42ed6475f8730c995', | |
60 | 'duration': 279, | |
61 | }, | |
62 | 'params': { | |
63 | # m3u8 download | |
64 | 'skip_download': True, | |
65 | }, | |
66 | }, | |
b773ead7 | 67 | { |
a7515ec2 | 68 | 'url': 'http://player.rutv.ru/iframe/video/id/771852/start_zoom/true/showZoomBtn/false/sid/russiatv/?acc_video_id=episode_id/970443/video_id/975648/brand_id/5169', |
b773ead7 | 69 | 'info_dict': { |
a7515ec2 | 70 | 'id': '771852', |
b773ead7 | 71 | 'ext': 'mp4', |
a7515ec2 S |
72 | 'title': 'Прямой эфир. Жертвы загадочной болезни: смерть от старости в 17 лет', |
73 | 'description': 'md5:b81c8c55247a4bd996b43ce17395b2d8', | |
74 | 'duration': 3096, | |
b773ead7 S |
75 | }, |
76 | 'params': { | |
77 | # m3u8 download | |
78 | 'skip_download': True, | |
79 | }, | |
b773ead7 S |
80 | }, |
81 | { | |
a7515ec2 | 82 | 'url': 'http://player.rutv.ru/iframe/live/id/51499/showZoomBtn/false/isPlay/true/sid/sochi2014', |
b773ead7 S |
83 | 'info_dict': { |
84 | 'id': '51499', | |
85 | 'ext': 'flv', | |
86 | 'title': 'Сочи-2014. Биатлон. Индивидуальная гонка. Мужчины ', | |
87 | 'description': 'md5:9e0ed5c9d2fa1efbfdfed90c9a6d179c', | |
88 | }, | |
d0fd3050 S |
89 | 'skip': 'Translation has finished', |
90 | }, | |
91 | { | |
dfad3aac | 92 | 'url': 'http://player.rutv.ru/iframe/live/id/21/showZoomBtn/false/isPlay/true/', |
d0fd3050 S |
93 | 'info_dict': { |
94 | 'id': '21', | |
95 | 'ext': 'mp4', | |
96 | 'title': 're:^Россия 24. Прямой эфир [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', | |
97 | 'is_live': True, | |
98 | }, | |
b773ead7 | 99 | 'params': { |
d0fd3050 | 100 | # m3u8 download |
b773ead7 S |
101 | 'skip_download': True, |
102 | }, | |
b773ead7 | 103 | }, |
624bd010 S |
104 | { |
105 | 'url': 'https://testplayer.vgtrk.com/iframe/live/id/19201/showZoomBtn/false/isPlay/true/', | |
106 | 'only_matching': True, | |
107 | }, | |
1074982e S |
108 | ] |
109 | ||
a7515ec2 S |
110 | @classmethod |
111 | def _extract_url(cls, webpage): | |
d37c07c5 | 112 | mobj = re.search( |
624bd010 | 113 | r'<iframe[^>]+?src=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/(?:iframe/(?:swf|video|live)/id|index/iframe/cast_id)/.+?)\1', webpage) |
d37c07c5 | 114 | if mobj: |
a7515ec2 | 115 | return mobj.group('url') |
d37c07c5 S |
116 | |
117 | mobj = re.search( | |
624bd010 | 118 | r'<meta[^>]+?property=(["\'])og:video\1[^>]+?content=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/flash\d+v/container\.swf\?id=.+?\2)', |
a7515ec2 | 119 | webpage) |
1074982e | 120 | if mobj: |
a7515ec2 | 121 | return mobj.group('url') |
1074982e | 122 | |
a7515ec2 | 123 | def _real_extract(self, url): |
5ad28e7f | 124 | mobj = self._match_valid_url(url) |
a7515ec2 | 125 | video_id = mobj.group('id') |
342f630d | 126 | video_path = mobj.group('path') |
1074982e | 127 | |
1600ed1f | 128 | if re.match(r'flash\d+v', video_path): |
a7515ec2 | 129 | video_type = 'video' |
342f630d S |
130 | elif video_path.startswith('iframe'): |
131 | video_type = mobj.group('type') | |
132 | if video_type == 'swf': | |
133 | video_type = 'video' | |
134 | elif video_path.startswith('index/iframe/cast_id'): | |
135 | video_type = 'live' | |
1074982e | 136 | |
e038d5c4 S |
137 | is_live = video_type == 'live' |
138 | ||
1074982e | 139 | json_data = self._download_json( |
5c23e3af | 140 | 'http://player.vgtrk.com/iframe/data%s/id/%s' % ('live' if is_live else 'video', video_id), |
1074982e S |
141 | video_id, 'Downloading JSON') |
142 | ||
143 | if json_data['errors']: | |
a7515ec2 | 144 | raise ExtractorError('%s said: %s' % (self.IE_NAME, json_data['errors']), expected=True) |
1074982e S |
145 | |
146 | playlist = json_data['data']['playlist'] | |
147 | medialist = playlist['medialist'] | |
148 | media = medialist[0] | |
149 | ||
150 | if media['errors']: | |
a7515ec2 | 151 | raise ExtractorError('%s said: %s' % (self.IE_NAME, media['errors']), expected=True) |
1074982e S |
152 | |
153 | view_count = playlist.get('count_views') | |
154 | priority_transport = playlist['priority_transport'] | |
155 | ||
156 | thumbnail = media['picture'] | |
e477dcf6 S |
157 | width = int_or_none(media['width']) |
158 | height = int_or_none(media['height']) | |
1074982e S |
159 | description = media['anons'] |
160 | title = media['title'] | |
161 | duration = int_or_none(media.get('duration')) | |
162 | ||
163 | formats = [] | |
164 | ||
165 | for transport, links in media['sources'].items(): | |
166 | for quality, url in links.items(): | |
e038d5c4 | 167 | preference = -1 if priority_transport == transport else -2 |
1074982e S |
168 | if transport == 'rtmp': |
169 | mobj = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>.+))/(?P<playpath>.+)$', url) | |
170 | if not mobj: | |
171 | continue | |
172 | fmt = { | |
173 | 'url': mobj.group('url'), | |
174 | 'play_path': mobj.group('playpath'), | |
175 | 'app': mobj.group('app'), | |
176 | 'page_url': 'http://player.rutv.ru', | |
1600ed1f | 177 | 'player_url': 'http://player.rutv.ru/flash3v/osmf.swf?i=22', |
1074982e S |
178 | 'rtmp_live': True, |
179 | 'ext': 'flv', | |
50e93e03 | 180 | 'vbr': str_to_int(quality), |
1074982e S |
181 | } |
182 | elif transport == 'm3u8': | |
e038d5c4 | 183 | formats.extend(self._extract_m3u8_formats( |
f983b875 | 184 | url, video_id, 'mp4', quality=preference, m3u8_id='hls')) |
c6df6924 | 185 | continue |
1074982e S |
186 | else: |
187 | fmt = { | |
188 | 'url': url | |
189 | } | |
190 | fmt.update({ | |
3f168f0e LNO |
191 | 'width': int_or_none(quality, default=height, invscale=width, scale=height), |
192 | 'height': int_or_none(quality, default=height), | |
1074982e | 193 | 'format_id': '%s-%s' % (transport, quality), |
3f168f0e | 194 | 'source_preference': preference, |
1074982e S |
195 | }) |
196 | formats.append(fmt) | |
197 | ||
1074982e S |
198 | self._sort_formats(formats) |
199 | ||
200 | return { | |
201 | 'id': video_id, | |
39ca3b5c | 202 | 'title': title, |
1074982e S |
203 | 'description': description, |
204 | 'thumbnail': thumbnail, | |
205 | 'view_count': view_count, | |
206 | 'duration': duration, | |
207 | 'formats': formats, | |
8dab1e90 | 208 | 'is_live': is_live, |
5f6a1245 | 209 | } |