]>
Commit | Line | Data |
---|---|---|
45d7bc2f JMF |
1 | from __future__ import unicode_literals |
2 | ||
70d1924f | 3 | import re |
70d1924f JMF |
4 | |
5 | from .common import InfoExtractor | |
ad30dc1e YCH |
6 | from ..compat import ( |
7 | compat_etree_fromstring, | |
8 | compat_urlparse, | |
9 | ) | |
1cc79574 | 10 | from ..utils import ( |
70d1924f | 11 | ExtractorError, |
7d3d06a1 | 12 | int_or_none, |
5c2266df | 13 | sanitized_Request, |
70d1924f JMF |
14 | ) |
15 | ||
88bd97e3 | 16 | |
70d1924f | 17 | class VevoIE(InfoExtractor): |
1c251cd9 | 18 | """ |
0577177e | 19 | Accepts urls from vevo.com or in the format 'vevo:{id}' |
3266f0c6 | 20 | (currently used by MTVIE and MySpaceIE) |
1c251cd9 | 21 | """ |
f25571ff | 22 | _VALID_URL = r'''(?x) |
91683085 | 23 | (?:https?://www\.vevo\.com/watch/(?:[^/]+/(?:[^/]+/)?)?| |
f25571ff | 24 | https?://cache\.vevo\.com/m/html/embed\.html\?video=| |
ebce53b3 | 25 | https?://videoplayer\.vevo\.com/embed/embedded\?videoId=| |
f25571ff PH |
26 | vevo:) |
27 | (?P<id>[^&?#]+)''' | |
fd5e6f7e | 28 | |
72321ead | 29 | _TESTS = [{ |
45d7bc2f | 30 | 'url': 'http://www.vevo.com/watch/hurts/somebody-to-die-for/GB1101300280', |
b686fc18 | 31 | "md5": "95ee28ee45e70130e3ab02b0f579ae23", |
45d7bc2f | 32 | 'info_dict': { |
27579b9e PH |
33 | 'id': 'GB1101300280', |
34 | 'ext': 'mp4', | |
45d7bc2f JMF |
35 | "upload_date": "20130624", |
36 | "uploader": "Hurts", | |
37 | "title": "Somebody to Die For", | |
38 | "duration": 230.12, | |
39 | "width": 1920, | |
40 | "height": 1080, | |
fd5e6f7e PH |
41 | # timestamp and upload_date are often incorrect; seem to change randomly |
42 | 'timestamp': int, | |
6f5ac90c | 43 | } |
27579b9e PH |
44 | }, { |
45 | 'note': 'v3 SMIL format', | |
46 | 'url': 'http://www.vevo.com/watch/cassadee-pope/i-wish-i-could-break-your-heart/USUV71302923', | |
b686fc18 | 47 | 'md5': 'f6ab09b034f8c22969020b042e5ac7fc', |
27579b9e PH |
48 | 'info_dict': { |
49 | 'id': 'USUV71302923', | |
50 | 'ext': 'mp4', | |
51 | 'upload_date': '20140219', | |
52 | 'uploader': 'Cassadee Pope', | |
53 | 'title': 'I Wish I Could Break Your Heart', | |
54 | 'duration': 226.101, | |
6cadf8c8 | 55 | 'age_limit': 0, |
fd5e6f7e | 56 | 'timestamp': int, |
6cadf8c8 PH |
57 | } |
58 | }, { | |
59 | 'note': 'Age-limited video', | |
60 | 'url': 'https://www.vevo.com/watch/justin-timberlake/tunnel-vision-explicit/USRV81300282', | |
61 | 'info_dict': { | |
62 | 'id': 'USRV81300282', | |
63 | 'ext': 'mp4', | |
64 | 'age_limit': 18, | |
65 | 'title': 'Tunnel Vision (Explicit)', | |
66 | 'uploader': 'Justin Timberlake', | |
aba77bbf PH |
67 | 'upload_date': 're:2013070[34]', |
68 | 'timestamp': int, | |
6cadf8c8 PH |
69 | }, |
70 | 'params': { | |
71 | 'skip_download': 'true', | |
27579b9e | 72 | } |
ff51983e YCH |
73 | }, { |
74 | 'note': 'No video_info', | |
75 | 'url': 'http://www.vevo.com/watch/k-camp-1/Till-I-Die/USUV71503000', | |
76 | 'md5': '8b83cc492d72fc9cf74a02acee7dc1b0', | |
77 | 'info_dict': { | |
78 | 'id': 'USUV71503000', | |
79 | 'ext': 'mp4', | |
80 | 'title': 'Till I Die - K Camp ft. T.I.', | |
81 | 'duration': 193, | |
82 | }, | |
b6afc225 | 83 | 'expected_warnings': ['Unable to download SMIL file'], |
72321ead PH |
84 | }] |
85 | _SMIL_BASE_URL = 'http://smil.lvl3.vevo.com/' | |
70d1924f | 86 | |
f0b5d6af | 87 | def _real_initialize(self): |
5c2266df | 88 | req = sanitized_Request( |
f0b5d6af PH |
89 | 'http://www.vevo.com/auth', data=b'') |
90 | webpage = self._download_webpage( | |
91 | req, None, | |
92 | note='Retrieving oauth token', | |
93 | errnote='Unable to retrieve oauth token', | |
94 | fatal=False) | |
95 | if webpage is False: | |
96 | self._oauth_token = None | |
97 | else: | |
4bf99ade YCH |
98 | if 'THIS PAGE IS CURRENTLY UNAVAILABLE IN YOUR REGION' in webpage: |
99 | raise ExtractorError('%s said: This page is currently unavailable in your region.' % self.IE_NAME, expected=True) | |
100 | ||
f0b5d6af PH |
101 | self._oauth_token = self._search_regex( |
102 | r'access_token":\s*"([^"]+)"', | |
103 | webpage, 'access token', fatal=False) | |
104 | ||
72321ead | 105 | def _formats_from_json(self, video_info): |
ff51983e YCH |
106 | if not video_info: |
107 | return [] | |
108 | ||
88bd97e3 JMF |
109 | last_version = {'version': -1} |
110 | for version in video_info['videoVersions']: | |
111 | # These are the HTTP downloads, other types are for different manifests | |
112 | if version['sourceType'] == 2: | |
113 | if version['version'] > last_version['version']: | |
114 | last_version = version | |
115 | if last_version['version'] == -1: | |
45d7bc2f | 116 | raise ExtractorError('Unable to extract last version of the video') |
88bd97e3 | 117 | |
36e6f62c | 118 | renditions = compat_etree_fromstring(last_version['data']) |
88bd97e3 JMF |
119 | formats = [] |
120 | # Already sorted from worst to best quality | |
121 | for rend in renditions.findall('rendition'): | |
122 | attr = rend.attrib | |
72321ead | 123 | format_note = '%(videoCodec)s@%(videoBitrate)4sk, %(audioCodec)s@%(audioBitrate)3sk' % attr |
88bd97e3 | 124 | formats.append({ |
e54fd4b2 PH |
125 | 'url': attr['url'], |
126 | 'format_id': attr['name'], | |
127 | 'format_note': format_note, | |
88bd97e3 JMF |
128 | 'height': int(attr['frameheight']), |
129 | 'width': int(attr['frameWidth']), | |
130 | }) | |
72321ead PH |
131 | return formats |
132 | ||
b6afc225 | 133 | def _formats_from_smil(self, smil_doc): |
72321ead | 134 | formats = [] |
72321ead PH |
135 | els = smil_doc.findall('.//{http://www.w3.org/2001/SMIL20/Language}video') |
136 | for el in els: | |
137 | src = el.attrib['src'] | |
138 | m = re.match(r'''(?xi) | |
139 | (?P<ext>[a-z0-9]+): | |
140 | (?P<path> | |
141 | [/a-z0-9]+ # The directory and main part of the URL | |
142 | _(?P<cbr>[0-9]+)k | |
143 | _(?P<width>[0-9]+)x(?P<height>[0-9]+) | |
144 | _(?P<vcodec>[a-z0-9]+) | |
145 | _(?P<vbr>[0-9]+) | |
146 | _(?P<acodec>[a-z0-9]+) | |
147 | _(?P<abr>[0-9]+) | |
148 | \.[a-z0-9]+ # File extension | |
149 | )''', src) | |
150 | if not m: | |
151 | continue | |
152 | ||
153 | format_url = self._SMIL_BASE_URL + m.group('path') | |
72321ead PH |
154 | formats.append({ |
155 | 'url': format_url, | |
45d7bc2f | 156 | 'format_id': 'SMIL_' + m.group('cbr'), |
91c7271a PH |
157 | 'vcodec': m.group('vcodec'), |
158 | 'acodec': m.group('acodec'), | |
159 | 'vbr': int(m.group('vbr')), | |
160 | 'abr': int(m.group('abr')), | |
72321ead PH |
161 | 'ext': m.group('ext'), |
162 | 'width': int(m.group('width')), | |
163 | 'height': int(m.group('height')), | |
164 | }) | |
165 | return formats | |
166 | ||
ad30dc1e | 167 | def _download_api_formats(self, video_id, video_url): |
f0b5d6af PH |
168 | if not self._oauth_token: |
169 | self._downloader.report_warning( | |
170 | 'No oauth token available, skipping API HLS download') | |
171 | return [] | |
172 | ||
ad30dc1e YCH |
173 | api_url = compat_urlparse.urljoin(video_url, '//apiv2.vevo.com/video/%s/streams/hls?token=%s' % ( |
174 | video_id, self._oauth_token)) | |
f0b5d6af PH |
175 | api_data = self._download_json( |
176 | api_url, video_id, | |
177 | note='Downloading HLS formats', | |
178 | errnote='Failed to download HLS format list', fatal=False) | |
179 | if api_data is None: | |
180 | return [] | |
181 | ||
182 | m3u8_url = api_data[0]['url'] | |
183 | return self._extract_m3u8_formats( | |
184 | m3u8_url, video_id, entry_protocol='m3u8_native', ext='mp4', | |
185 | preference=0) | |
186 | ||
72321ead | 187 | def _real_extract(self, url): |
4b942883 | 188 | video_id = self._match_id(url) |
72321ead | 189 | |
ff51983e YCH |
190 | webpage = None |
191 | ||
72321ead | 192 | json_url = 'http://videoplayer.vevo.com/VideoService/AuthenticateVideo?isrc=%s' % video_id |
e97e53ee | 193 | response = self._download_json(json_url, video_id) |
ff51983e | 194 | video_info = response['video'] or {} |
e97e53ee | 195 | |
ff51983e | 196 | if not video_info and response.get('statusCode') != 909: |
e97e53ee S |
197 | if 'statusMessage' in response: |
198 | raise ExtractorError('%s said: %s' % (self.IE_NAME, response['statusMessage']), expected=True) | |
199 | raise ExtractorError('Unable to extract videos') | |
72321ead | 200 | |
ff51983e YCH |
201 | if not video_info: |
202 | if url.startswith('vevo:'): | |
203 | raise ExtractorError('Please specify full Vevo URL for downloading', expected=True) | |
204 | webpage = self._download_webpage(url, video_id) | |
205 | ||
206 | title = video_info.get('title') or self._og_search_title(webpage) | |
207 | ||
72321ead | 208 | formats = self._formats_from_json(video_info) |
27579b9e | 209 | |
6cadf8c8 PH |
210 | is_explicit = video_info.get('isExplicit') |
211 | if is_explicit is True: | |
212 | age_limit = 18 | |
213 | elif is_explicit is False: | |
214 | age_limit = 0 | |
215 | else: | |
216 | age_limit = None | |
217 | ||
f0b5d6af | 218 | # Download via HLS API |
ad30dc1e | 219 | formats.extend(self._download_api_formats(video_id, url)) |
f0b5d6af | 220 | |
7d3d06a1 PH |
221 | # Download SMIL |
222 | smil_blocks = sorted(( | |
ff51983e | 223 | f for f in video_info.get('videoVersions', []) |
7d3d06a1 PH |
224 | if f['sourceType'] == 13), |
225 | key=lambda f: f['version']) | |
226 | smil_url = '%s/Video/V2/VFILE/%s/%sr.smil' % ( | |
227 | self._SMIL_BASE_URL, video_id, video_id.lower()) | |
228 | if smil_blocks: | |
229 | smil_url_m = self._search_regex( | |
230 | r'url="([^"]+)"', smil_blocks[-1]['data'], 'SMIL URL', | |
231 | default=None) | |
232 | if smil_url_m is not None: | |
233 | smil_url = smil_url_m | |
234 | if smil_url: | |
b6afc225 YCH |
235 | smil_doc = self._download_smil(smil_url, video_id, fatal=False) |
236 | if smil_doc: | |
237 | formats.extend(self._formats_from_smil(smil_doc)) | |
7d3d06a1 | 238 | |
bff74bdd | 239 | self._sort_formats(formats) |
ff51983e | 240 | timestamp = int_or_none(self._search_regex( |
7d3d06a1 | 241 | r'/Date\((\d+)\)/', |
ff51983e YCH |
242 | video_info['launchDate'], 'launch date', fatal=False), |
243 | scale=1000) if video_info else None | |
244 | ||
245 | duration = video_info.get('duration') or int_or_none( | |
246 | self._html_search_meta('video:duration', webpage)) | |
9d2ecdbc | 247 | |
45d7bc2f | 248 | return { |
88bd97e3 | 249 | 'id': video_id, |
ff51983e | 250 | 'title': title, |
88bd97e3 | 251 | 'formats': formats, |
ff51983e YCH |
252 | 'thumbnail': video_info.get('imageUrl'), |
253 | 'timestamp': timestamp, | |
254 | 'uploader': video_info['mainArtists'][0]['artistName'] if video_info else None, | |
255 | 'duration': duration, | |
6cadf8c8 | 256 | 'age_limit': age_limit, |
88bd97e3 | 257 | } |