]>
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, |
60d142aa | 16 | unescapeHTML, |
e1e8b689 | 17 | unified_strdate, |
1cc79574 | 18 | ) |
60d142aa JMF |
19 | |
20 | ||
21 | class VKIE(InfoExtractor): | |
94a23d2a | 22 | IE_NAME = 'vk.com' |
469d4c89 | 23 | _VALID_URL = r'https?://(?:m\.)?vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:.+?\?.*?z=)?video(?P<videoid>[^s].*?)(?:\?|%2F|$))' |
5544e038 | 24 | _NETRC_MACHINE = 'vk' |
60d142aa | 25 | |
9032dc28 S |
26 | _TESTS = [ |
27 | { | |
28 | 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521', | |
29 | 'md5': '0deae91935c54e00003c2a00646315f0', | |
30 | 'info_dict': { | |
31 | 'id': '162222515', | |
32 | 'ext': 'flv', | |
33 | 'title': 'ProtivoGunz - Хуёвая песня', | |
01ba1780 | 34 | 'uploader': 're:Noize MC.*', |
9032dc28 | 35 | 'duration': 195, |
42e1ff86 | 36 | 'upload_date': '20120212', |
9032dc28 | 37 | }, |
60d142aa | 38 | }, |
9032dc28 | 39 | { |
c52331f3 WS |
40 | 'url': 'http://vk.com/video205387401_165548505', |
41 | 'md5': '6c0aeb2e90396ba97035b9cbde548700', | |
9032dc28 | 42 | 'info_dict': { |
c52331f3 | 43 | 'id': '165548505', |
9032dc28 | 44 | 'ext': 'mp4', |
c52331f3 WS |
45 | 'uploader': 'Tom Cruise', |
46 | 'title': 'No name', | |
47 | 'duration': 9, | |
48 | 'upload_date': '20130721' | |
9032dc28 S |
49 | } |
50 | }, | |
ca97a56e S |
51 | { |
52 | 'note': 'Embedded video', | |
53 | 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1', | |
54 | 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a', | |
55 | 'info_dict': { | |
56 | 'id': '162925554', | |
57 | 'ext': 'mp4', | |
58 | 'uploader': 'Vladimir Gavrin', | |
59 | 'title': 'Lin Dan', | |
60 | 'duration': 101, | |
42e1ff86 | 61 | 'upload_date': '20120730', |
ca97a56e S |
62 | } |
63 | }, | |
9032dc28 | 64 | { |
c52331f3 WS |
65 | # VIDEO NOW REMOVED |
66 | # please update if you find a video whose URL follows the same pattern | |
9032dc28 S |
67 | 'url': 'http://vk.com/video-8871596_164049491', |
68 | 'md5': 'a590bcaf3d543576c9bd162812387666', | |
69 | 'note': 'Only available for registered users', | |
70 | 'info_dict': { | |
71 | 'id': '164049491', | |
72 | 'ext': 'mp4', | |
73 | 'uploader': 'Триллеры', | |
57bdc730 | 74 | 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]', |
9032dc28 | 75 | 'duration': 8352, |
c52331f3 | 76 | 'upload_date': '20121218' |
9032dc28 S |
77 | }, |
78 | 'skip': 'Requires vk account credentials', | |
ca97a56e | 79 | }, |
57bdc730 S |
80 | { |
81 | 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d', | |
82 | 'md5': '4d7a5ef8cf114dfa09577e57b2993202', | |
83 | 'info_dict': { | |
84 | 'id': '168067957', | |
85 | 'ext': 'mp4', | |
86 | 'uploader': 'Киномания - лучшее из мира кино', | |
87 | 'title': ' ', | |
88 | 'duration': 7291, | |
42e1ff86 | 89 | 'upload_date': '20140328', |
57bdc730 S |
90 | }, |
91 | 'skip': 'Requires vk account credentials', | |
92 | }, | |
849086a1 S |
93 | { |
94 | 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540', | |
95 | 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6', | |
96 | 'note': 'ivi.ru embed', | |
97 | 'info_dict': { | |
98 | 'id': '60690', | |
99 | 'ext': 'mp4', | |
100 | 'title': 'Книга Илая', | |
101 | 'duration': 6771, | |
42e1ff86 | 102 | 'upload_date': '20140626', |
849086a1 | 103 | }, |
d518d06e | 104 | 'skip': 'Only works from Russia', |
849086a1 | 105 | }, |
a8363f3a PH |
106 | { |
107 | # removed video, just testing that we match the pattern | |
108 | 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a', | |
109 | 'only_matching': True, | |
110 | }, | |
9032dc28 S |
111 | ] |
112 | ||
113 | def _login(self): | |
114 | (username, password) = self._get_login_info() | |
115 | if username is None: | |
116 | return | |
117 | ||
118 | login_form = { | |
119 | 'act': 'login', | |
120 | 'role': 'al_frame', | |
121 | 'expire': '1', | |
122 | 'email': username, | |
123 | 'pass': password, | |
124 | } | |
125 | ||
126 | request = compat_urllib_request.Request('https://login.vk.com/?act=login', | |
9e1a5b84 | 127 | compat_urllib_parse.urlencode(login_form).encode('utf-8')) |
9032dc28 S |
128 | login_page = self._download_webpage(request, None, note='Logging in as %s' % username) |
129 | ||
130 | if re.search(r'onLoginFailed', login_page): | |
131 | raise ExtractorError('Unable to login, incorrect username and/or password', expected=True) | |
132 | ||
133 | def _real_initialize(self): | |
134 | self._login() | |
60d142aa JMF |
135 | |
136 | def _real_extract(self, url): | |
137 | mobj = re.match(self._VALID_URL, url) | |
ca97a56e S |
138 | video_id = mobj.group('videoid') |
139 | ||
140 | if not video_id: | |
141 | video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id')) | |
9032dc28 | 142 | |
60d142aa JMF |
143 | info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id |
144 | info_page = self._download_webpage(info_url, video_id) | |
9032dc28 | 145 | |
e0c51cda S |
146 | ERRORS = { |
147 | r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<': | |
3d36cea4 PH |
148 | 'Video %s has been removed from public access due to rightholder complaint.', |
149 | ||
e0c51cda | 150 | r'<!>Please log in or <': |
3d36cea4 PH |
151 | 'Video %s is only available for registered users, ' |
152 | 'use --username and --password options to provide account credentials.', | |
153 | ||
154 | r'<!>Unknown error': | |
155 | 'Video %s does not exist.' | |
e0c51cda | 156 | } |
9032dc28 | 157 | |
e0c51cda S |
158 | for error_re, error_msg in ERRORS.items(): |
159 | if re.search(error_re, info_page): | |
160 | raise ExtractorError(error_msg % video_id, expected=True) | |
9334f8f1 | 161 | |
60d142aa JMF |
162 | m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page) |
163 | if m_yt is not None: | |
fbe80531 | 164 | self.to_screen('Youtube video detected') |
60d142aa | 165 | return self.url_result(m_yt.group(1), 'Youtube') |
849086a1 | 166 | |
7a1818c9 PH |
167 | m_rutube = re.search( |
168 | r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page) | |
7a1818c9 PH |
169 | if m_rutube is not None: |
170 | self.to_screen('rutube video detected') | |
171 | rutube_url = self._proto_relative_url( | |
172 | m_rutube.group(1).replace('\\', '')) | |
173 | return self.url_result(rutube_url) | |
174 | ||
849086a1 S |
175 | m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.*?});', info_page) |
176 | if m_opts: | |
177 | m_opts_url = re.search(r"url\s*:\s*'([^']+)", m_opts.group(1)) | |
178 | if m_opts_url: | |
179 | opts_url = m_opts_url.group(1) | |
180 | if opts_url.startswith('//'): | |
181 | opts_url = 'http:' + opts_url | |
182 | return self.url_result(opts_url) | |
183 | ||
94a23d2a | 184 | data_json = self._search_regex(r'var vars = ({.*?});', info_page, 'vars') |
608bf698 | 185 | data = json.loads(data_json) |
60d142aa | 186 | |
02a12f9f WS |
187 | # Extract upload date |
188 | upload_date = None | |
189 | mobj = re.search(r'id="mv_date_wrap".*?Added ([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page) | |
190 | if mobj is not None: | |
9e1a5b84 | 191 | mobj.group(1) + ' ' + mobj.group(2) |
02a12f9f WS |
192 | upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2)) |
193 | ||
913f3292 PH |
194 | formats = [{ |
195 | 'format_id': k, | |
196 | 'url': v, | |
197 | 'width': int(k[len('url'):]), | |
198 | } for k, v in data.items() | |
199 | if k.startswith('url')] | |
200 | self._sort_formats(formats) | |
201 | ||
60d142aa | 202 | return { |
608bf698 | 203 | 'id': compat_str(data['vid']), |
913f3292 | 204 | 'formats': formats, |
608bf698 | 205 | 'title': unescapeHTML(data['md_title']), |
913f3292 PH |
206 | 'thumbnail': data.get('jpg'), |
207 | 'uploader': data.get('md_author'), | |
02a12f9f WS |
208 | 'duration': data.get('duration'), |
209 | 'upload_date': upload_date, | |
60d142aa | 210 | } |
469d4c89 WS |
211 | |
212 | ||
213 | class VKUserVideosIE(InfoExtractor): | |
214 | IE_NAME = 'vk.com:user-videos' | |
cad985ab | 215 | IE_DESC = 'vk.com:All of a user\'s videos' |
53d1cd1f | 216 | _VALID_URL = r'https?://vk\.com/videos(?P<id>[0-9]+)(?:m\?.*)?' |
469d4c89 WS |
217 | _TEMPLATE_URL = 'https://vk.com/videos' |
218 | _TEST = { | |
219 | 'url': 'http://vk.com/videos205387401', | |
220 | 'playlist_mincount': 4, | |
221 | } | |
222 | ||
469d4c89 | 223 | def _real_extract(self, url): |
021a0db8 | 224 | page_id = self._match_id(url) |
469d4c89 | 225 | page = self._download_webpage(url, page_id) |
d16abf43 PH |
226 | video_ids = orderedSet( |
227 | m.group(1) for m in re.finditer(r'href="/video([0-9_]+)"', page)) | |
228 | url_entries = [ | |
229 | self.url_result( | |
230 | 'http://vk.com/video' + video_id, 'VK', video_id=video_id) | |
231 | for video_id in video_ids] | |
9262867e | 232 | return self.playlist_result(url_entries, page_id) |