]>
Commit | Line | Data |
---|---|---|
60d142aa | 1 | # encoding: utf-8 |
94a23d2a PH |
2 | from __future__ import unicode_literals |
3 | ||
60d142aa JMF |
4 | import re |
5 | import json | |
6 | ||
7 | from .common import InfoExtractor | |
1cc79574 PH |
8 | from ..compat import ( |
9 | compat_str, | |
10 | compat_urllib_parse, | |
11 | compat_urllib_request, | |
12 | ) | |
60d142aa | 13 | from ..utils import ( |
9032dc28 | 14 | ExtractorError, |
1cc79574 | 15 | orderedSet, |
8117df4c | 16 | str_to_int, |
60d142aa | 17 | unescapeHTML, |
e1e8b689 | 18 | unified_strdate, |
1cc79574 | 19 | ) |
60d142aa JMF |
20 | |
21 | ||
22 | class VKIE(InfoExtractor): | |
94a23d2a | 23 | IE_NAME = 'vk.com' |
cf9cf7dd S |
24 | _VALID_URL = r'''(?x) |
25 | https?:// | |
26 | (?: | |
27 | (?:m\.)?vk\.com/video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)| | |
28 | (?: | |
29 | (?:m\.)?vk\.com/(?:.+?\?.*?z=)?video| | |
30 | (?:www\.)?biqle\.ru/watch/ | |
31 | ) | |
32 | (?P<videoid>[^s].*?)(?:\?|%2F|$) | |
33 | ) | |
34 | ''' | |
5544e038 | 35 | _NETRC_MACHINE = 'vk' |
60d142aa | 36 | |
9032dc28 S |
37 | _TESTS = [ |
38 | { | |
39 | 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521', | |
40 | 'md5': '0deae91935c54e00003c2a00646315f0', | |
41 | 'info_dict': { | |
42 | 'id': '162222515', | |
43 | 'ext': 'flv', | |
44 | 'title': 'ProtivoGunz - Хуёвая песня', | |
36300346 | 45 | 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*', |
9032dc28 | 46 | 'duration': 195, |
42e1ff86 | 47 | 'upload_date': '20120212', |
8117df4c | 48 | 'view_count': int, |
9032dc28 | 49 | }, |
60d142aa | 50 | }, |
9032dc28 | 51 | { |
c52331f3 WS |
52 | 'url': 'http://vk.com/video205387401_165548505', |
53 | 'md5': '6c0aeb2e90396ba97035b9cbde548700', | |
9032dc28 | 54 | 'info_dict': { |
c52331f3 | 55 | 'id': '165548505', |
9032dc28 | 56 | 'ext': 'mp4', |
c52331f3 WS |
57 | 'uploader': 'Tom Cruise', |
58 | 'title': 'No name', | |
59 | 'duration': 9, | |
8117df4c S |
60 | 'upload_date': '20130721', |
61 | 'view_count': int, | |
9032dc28 S |
62 | } |
63 | }, | |
ca97a56e S |
64 | { |
65 | 'note': 'Embedded video', | |
66 | 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1', | |
67 | 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a', | |
68 | 'info_dict': { | |
69 | 'id': '162925554', | |
70 | 'ext': 'mp4', | |
71 | 'uploader': 'Vladimir Gavrin', | |
72 | 'title': 'Lin Dan', | |
73 | 'duration': 101, | |
42e1ff86 | 74 | 'upload_date': '20120730', |
8117df4c | 75 | 'view_count': int, |
ca97a56e S |
76 | } |
77 | }, | |
9032dc28 | 78 | { |
c52331f3 WS |
79 | # VIDEO NOW REMOVED |
80 | # please update if you find a video whose URL follows the same pattern | |
9032dc28 S |
81 | 'url': 'http://vk.com/video-8871596_164049491', |
82 | 'md5': 'a590bcaf3d543576c9bd162812387666', | |
83 | 'note': 'Only available for registered users', | |
84 | 'info_dict': { | |
85 | 'id': '164049491', | |
86 | 'ext': 'mp4', | |
87 | 'uploader': 'Триллеры', | |
57bdc730 | 88 | 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]', |
9032dc28 | 89 | 'duration': 8352, |
8117df4c S |
90 | 'upload_date': '20121218', |
91 | 'view_count': int, | |
9032dc28 S |
92 | }, |
93 | 'skip': 'Requires vk account credentials', | |
ca97a56e | 94 | }, |
57bdc730 S |
95 | { |
96 | 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d', | |
97 | 'md5': '4d7a5ef8cf114dfa09577e57b2993202', | |
98 | 'info_dict': { | |
99 | 'id': '168067957', | |
100 | 'ext': 'mp4', | |
101 | 'uploader': 'Киномания - лучшее из мира кино', | |
102 | 'title': ' ', | |
103 | 'duration': 7291, | |
42e1ff86 | 104 | 'upload_date': '20140328', |
57bdc730 S |
105 | }, |
106 | 'skip': 'Requires vk account credentials', | |
107 | }, | |
849086a1 S |
108 | { |
109 | 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540', | |
110 | 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6', | |
111 | 'note': 'ivi.ru embed', | |
112 | 'info_dict': { | |
113 | 'id': '60690', | |
114 | 'ext': 'mp4', | |
115 | 'title': 'Книга Илая', | |
116 | 'duration': 6771, | |
42e1ff86 | 117 | 'upload_date': '20140626', |
8117df4c | 118 | 'view_count': int, |
849086a1 | 119 | }, |
d518d06e | 120 | 'skip': 'Only works from Russia', |
849086a1 | 121 | }, |
a8363f3a PH |
122 | { |
123 | # removed video, just testing that we match the pattern | |
124 | 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a', | |
125 | 'only_matching': True, | |
126 | }, | |
cf9cf7dd S |
127 | { |
128 | # vk wrapper | |
129 | 'url': 'http://www.biqle.ru/watch/847655_160197695', | |
130 | 'only_matching': True, | |
131 | } | |
9032dc28 S |
132 | ] |
133 | ||
134 | def _login(self): | |
135 | (username, password) = self._get_login_info() | |
136 | if username is None: | |
137 | return | |
138 | ||
226b886c S |
139 | login_page = self._download_webpage( |
140 | 'https://vk.com', None, 'Downloading login page') | |
141 | ||
142 | login_form = dict(re.findall( | |
143 | r'<input\s+type="hidden"\s+name="([^"]+)"\s+(?:id="[^"]+"\s+)?value="([^"]*)"', | |
144 | login_page)) | |
145 | ||
146 | login_form.update({ | |
4f3bf679 S |
147 | 'email': username.encode('cp1251'), |
148 | 'pass': password.encode('cp1251'), | |
226b886c | 149 | }) |
9032dc28 | 150 | |
226b886c S |
151 | request = compat_urllib_request.Request( |
152 | 'https://login.vk.com/?act=login', | |
153 | compat_urllib_parse.urlencode(login_form).encode('utf-8')) | |
154 | login_page = self._download_webpage( | |
155 | request, None, note='Logging in as %s' % username) | |
9032dc28 S |
156 | |
157 | if re.search(r'onLoginFailed', login_page): | |
226b886c S |
158 | raise ExtractorError( |
159 | 'Unable to login, incorrect username and/or password', expected=True) | |
9032dc28 S |
160 | |
161 | def _real_initialize(self): | |
162 | self._login() | |
60d142aa JMF |
163 | |
164 | def _real_extract(self, url): | |
165 | mobj = re.match(self._VALID_URL, url) | |
ca97a56e S |
166 | video_id = mobj.group('videoid') |
167 | ||
168 | if not video_id: | |
169 | video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id')) | |
9032dc28 | 170 | |
275c0423 | 171 | info_url = 'https://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id |
60d142aa | 172 | info_page = self._download_webpage(info_url, video_id) |
9032dc28 | 173 | |
7f220b2f S |
174 | if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page): |
175 | raise ExtractorError( | |
176 | 'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.', | |
177 | expected=True) | |
178 | ||
e0c51cda S |
179 | ERRORS = { |
180 | r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<': | |
3d36cea4 PH |
181 | 'Video %s has been removed from public access due to rightholder complaint.', |
182 | ||
e0c51cda | 183 | r'<!>Please log in or <': |
3d36cea4 PH |
184 | 'Video %s is only available for registered users, ' |
185 | 'use --username and --password options to provide account credentials.', | |
186 | ||
187 | r'<!>Unknown error': | |
1aa5172f S |
188 | 'Video %s does not exist.', |
189 | ||
190 | r'<!>Видео временно недоступно': | |
191 | 'Video %s is temporarily unavailable.', | |
e0c51cda | 192 | } |
9032dc28 | 193 | |
e0c51cda S |
194 | for error_re, error_msg in ERRORS.items(): |
195 | if re.search(error_re, info_page): | |
196 | raise ExtractorError(error_msg % video_id, expected=True) | |
9334f8f1 | 197 | |
60d142aa JMF |
198 | m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page) |
199 | if m_yt is not None: | |
fbe80531 | 200 | self.to_screen('Youtube video detected') |
60d142aa | 201 | return self.url_result(m_yt.group(1), 'Youtube') |
849086a1 | 202 | |
7a1818c9 PH |
203 | m_rutube = re.search( |
204 | r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page) | |
7a1818c9 PH |
205 | if m_rutube is not None: |
206 | self.to_screen('rutube video detected') | |
207 | rutube_url = self._proto_relative_url( | |
208 | m_rutube.group(1).replace('\\', '')) | |
209 | return self.url_result(rutube_url) | |
210 | ||
054932f4 | 211 | m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page) |
849086a1 | 212 | if m_opts: |
054932f4 | 213 | m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1)) |
849086a1 S |
214 | if m_opts_url: |
215 | opts_url = m_opts_url.group(1) | |
216 | if opts_url.startswith('//'): | |
217 | opts_url = 'http:' + opts_url | |
218 | return self.url_result(opts_url) | |
219 | ||
054932f4 | 220 | data_json = self._search_regex(r'var\s+vars\s*=\s*({.+?});', info_page, 'vars') |
608bf698 | 221 | data = json.loads(data_json) |
60d142aa | 222 | |
02a12f9f WS |
223 | # Extract upload date |
224 | upload_date = None | |
7c7dd9dc | 225 | mobj = re.search(r'id="mv_date(?:_views)?_wrap"[^>]*>([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page) |
02a12f9f | 226 | if mobj is not None: |
9e1a5b84 | 227 | mobj.group(1) + ' ' + mobj.group(2) |
02a12f9f WS |
228 | upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2)) |
229 | ||
8117df4c S |
230 | view_count = str_to_int(self._search_regex( |
231 | r'"mv_views_count_number"[^>]*>([\d,.]+) views<', | |
232 | info_page, 'view count', fatal=False)) | |
233 | ||
913f3292 PH |
234 | formats = [{ |
235 | 'format_id': k, | |
236 | 'url': v, | |
237 | 'width': int(k[len('url'):]), | |
238 | } for k, v in data.items() | |
239 | if k.startswith('url')] | |
240 | self._sort_formats(formats) | |
241 | ||
60d142aa | 242 | return { |
608bf698 | 243 | 'id': compat_str(data['vid']), |
913f3292 | 244 | 'formats': formats, |
608bf698 | 245 | 'title': unescapeHTML(data['md_title']), |
913f3292 PH |
246 | 'thumbnail': data.get('jpg'), |
247 | 'uploader': data.get('md_author'), | |
02a12f9f WS |
248 | 'duration': data.get('duration'), |
249 | 'upload_date': upload_date, | |
8117df4c | 250 | 'view_count': view_count, |
60d142aa | 251 | } |
469d4c89 WS |
252 | |
253 | ||
254 | class VKUserVideosIE(InfoExtractor): | |
255 | IE_NAME = 'vk.com:user-videos' | |
cad985ab | 256 | IE_DESC = 'vk.com:All of a user\'s videos' |
53d1cd1f | 257 | _VALID_URL = r'https?://vk\.com/videos(?P<id>[0-9]+)(?:m\?.*)?' |
469d4c89 WS |
258 | _TEMPLATE_URL = 'https://vk.com/videos' |
259 | _TEST = { | |
260 | 'url': 'http://vk.com/videos205387401', | |
15ec6693 PH |
261 | 'info_dict': { |
262 | 'id': '205387401', | |
263 | }, | |
469d4c89 WS |
264 | 'playlist_mincount': 4, |
265 | } | |
266 | ||
469d4c89 | 267 | def _real_extract(self, url): |
021a0db8 | 268 | page_id = self._match_id(url) |
469d4c89 | 269 | page = self._download_webpage(url, page_id) |
d16abf43 PH |
270 | video_ids = orderedSet( |
271 | m.group(1) for m in re.finditer(r'href="/video([0-9_]+)"', page)) | |
272 | url_entries = [ | |
273 | self.url_result( | |
274 | 'http://vk.com/video' + video_id, 'VK', video_id=video_id) | |
275 | for video_id in video_ids] | |
9262867e | 276 | return self.playlist_result(url_entries, page_id) |