]>
Commit | Line | Data |
---|---|---|
7f09a662 YCH |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
7a8fef31 | 4 | import base64 |
265bfa2c | 5 | import datetime |
7a8fef31 | 6 | import hashlib |
57031161 | 7 | import re |
7f09a662 | 8 | import time |
7f09a662 YCH |
9 | |
10 | from .common import InfoExtractor | |
265bfa2c | 11 | from ..compat import ( |
1e399778 | 12 | compat_ord, |
7b7507d6 | 13 | compat_str, |
7a8fef31 | 14 | compat_urllib_parse, |
265bfa2c PH |
15 | ) |
16 | from ..utils import ( | |
17 | determine_ext, | |
7a8fef31 | 18 | encode_data_uri, |
265bfa2c | 19 | ExtractorError, |
7a8fef31 YCH |
20 | int_or_none, |
21 | orderedSet, | |
265bfa2c | 22 | parse_iso8601, |
5c2266df | 23 | sanitized_Request, |
5f432ac8 | 24 | str_or_none, |
bec30224 | 25 | url_basename, |
265bfa2c | 26 | ) |
7f09a662 YCH |
27 | |
28 | ||
8c5a9944 | 29 | class LeIE(InfoExtractor): |
963d0ce7 | 30 | IE_DESC = '乐视网' |
f3bc19a9 | 31 | _VALID_URL = r'http://www\.le\.com/ptv/vplay/(?P<id>\d+)\.html' |
5e73a67d | 32 | |
f3bc19a9 | 33 | _URL_TEMPLATE = 'http://www.le.com/ptv/vplay/%s.html' |
7f09a662 YCH |
34 | |
35 | _TESTS = [{ | |
5e73a67d | 36 | 'url': 'http://www.le.com/ptv/vplay/22005890.html', |
1e399778 | 37 | 'md5': 'edadcfe5406976f42f9f266057ee5e40', |
7f09a662 YCH |
38 | 'info_dict': { |
39 | 'id': '22005890', | |
40 | 'ext': 'mp4', | |
41 | 'title': '第87届奥斯卡颁奖礼完美落幕 《鸟人》成最大赢家', | |
67706359 | 42 | 'description': 'md5:a9cb175fd753e2962176b7beca21a47c', |
1e399778 YCH |
43 | }, |
44 | 'params': { | |
45 | 'hls_prefer_native': True, | |
46 | }, | |
7f09a662 | 47 | }, { |
5e73a67d | 48 | 'url': 'http://www.le.com/ptv/vplay/1415246.html', |
7f09a662 | 49 | 'info_dict': { |
67706359 | 50 | 'id': '1415246', |
7f09a662 | 51 | 'ext': 'mp4', |
67706359 YCH |
52 | 'title': '美人天下01', |
53 | 'description': 'md5:f88573d9d7225ada1359eaf0dbf8bcda', | |
54 | }, | |
1e399778 YCH |
55 | 'params': { |
56 | 'hls_prefer_native': True, | |
57 | }, | |
91410c9b PH |
58 | }, { |
59 | 'note': 'This video is available only in Mainland China, thus a proxy is needed', | |
5e73a67d | 60 | 'url': 'http://www.le.com/ptv/vplay/1118082.html', |
1e399778 | 61 | 'md5': '2424c74948a62e5f31988438979c5ad1', |
91410c9b PH |
62 | 'info_dict': { |
63 | 'id': '1118082', | |
64 | 'ext': 'mp4', | |
65 | 'title': '与龙共舞 完整版', | |
66 | 'description': 'md5:7506a5eeb1722bb9d4068f85024e3986', | |
67 | }, | |
1e399778 YCH |
68 | 'params': { |
69 | 'hls_prefer_native': True, | |
70 | }, | |
051df9ad | 71 | 'skip': 'Only available in China', |
7f09a662 YCH |
72 | }] |
73 | ||
74 | @staticmethod | |
75 | def urshift(val, n): | |
76 | return val >> n if val >= 0 else (val + 0x100000000) >> n | |
77 | ||
265bfa2c | 78 | # ror() and calc_time_key() are reversed from a embedded swf file in KLetvPlayer.swf |
7f09a662 YCH |
79 | def ror(self, param1, param2): |
80 | _loc3_ = 0 | |
81 | while _loc3_ < param2: | |
82 | param1 = self.urshift(param1, 1) + ((param1 & 1) << 31) | |
83 | _loc3_ += 1 | |
84 | return param1 | |
85 | ||
265bfa2c | 86 | def calc_time_key(self, param1): |
7f09a662 YCH |
87 | _loc2_ = 773625421 |
88 | _loc3_ = self.ror(param1, _loc2_ % 13) | |
89 | _loc3_ = _loc3_ ^ _loc2_ | |
90 | _loc3_ = self.ror(_loc3_, _loc2_ % 17) | |
91 | return _loc3_ | |
92 | ||
1e399778 YCH |
93 | # see M3U8Encryption class in KLetvPlayer.swf |
94 | @staticmethod | |
95 | def decrypt_m3u8(encrypted_data): | |
96 | if encrypted_data[:5].decode('utf-8').lower() != 'vc_01': | |
97 | return encrypted_data | |
98 | encrypted_data = encrypted_data[5:] | |
99 | ||
2ebd2eac YCH |
100 | _loc4_ = bytearray(2 * len(encrypted_data)) |
101 | for idx, val in enumerate(encrypted_data): | |
102 | b = compat_ord(val) | |
103 | _loc4_[2 * idx] = b // 16 | |
104 | _loc4_[2 * idx + 1] = b % 16 | |
1e399778 YCH |
105 | idx = len(_loc4_) - 11 |
106 | _loc4_ = _loc4_[idx:] + _loc4_[:idx] | |
2ebd2eac YCH |
107 | _loc7_ = bytearray(len(encrypted_data)) |
108 | for i in range(len(encrypted_data)): | |
109 | _loc7_[i] = _loc4_[2 * i] * 16 + _loc4_[2 * i + 1] | |
1e399778 YCH |
110 | |
111 | return bytes(_loc7_) | |
112 | ||
7f09a662 YCH |
113 | def _real_extract(self, url): |
114 | media_id = self._match_id(url) | |
115 | page = self._download_webpage(url, media_id) | |
116 | params = { | |
117 | 'id': media_id, | |
118 | 'platid': 1, | |
119 | 'splatid': 101, | |
120 | 'format': 1, | |
265bfa2c | 121 | 'tkey': self.calc_time_key(int(time.time())), |
5e73a67d | 122 | 'domain': 'www.le.com' |
7f09a662 | 123 | } |
5c2266df | 124 | play_json_req = sanitized_Request( |
5e73a67d | 125 | 'http://api.le.com/mms/out/video/playJson?' + compat_urllib_parse.urlencode(params) |
91410c9b | 126 | ) |
63fc8000 YCH |
127 | cn_verification_proxy = self._downloader.params.get('cn_verification_proxy') |
128 | if cn_verification_proxy: | |
129 | play_json_req.add_header('Ytdl-request-proxy', cn_verification_proxy) | |
130 | ||
7f09a662 | 131 | play_json = self._download_json( |
91410c9b | 132 | play_json_req, |
576904bc | 133 | media_id, 'Downloading playJson data') |
7f09a662 YCH |
134 | |
135 | # Check for errors | |
136 | playstatus = play_json['playstatus'] | |
137 | if playstatus['status'] == 0: | |
138 | flag = playstatus['flag'] | |
139 | if flag == 1: | |
140 | msg = 'Country %s auth error' % playstatus['country'] | |
141 | else: | |
142 | msg = 'Generic error. flag = %d' % flag | |
143 | raise ExtractorError(msg, expected=True) | |
144 | ||
145 | playurl = play_json['playurl'] | |
146 | ||
147 | formats = ['350', '1000', '1300', '720p', '1080p'] | |
148 | dispatch = playurl['dispatch'] | |
149 | ||
150 | urls = [] | |
151 | for format_id in formats: | |
152 | if format_id in dispatch: | |
153 | media_url = playurl['domain'][0] + dispatch[format_id][0] | |
1e399778 YCH |
154 | media_url += '&' + compat_urllib_parse.urlencode({ |
155 | 'm3v': 1, | |
156 | 'format': 1, | |
157 | 'expect': 3, | |
158 | 'rateid': format_id, | |
7f09a662 | 159 | }) |
1e399778 YCH |
160 | |
161 | nodes_data = self._download_json( | |
162 | media_url, media_id, | |
163 | 'Download JSON metadata for format %s' % format_id) | |
164 | ||
165 | req = self._request_webpage( | |
166 | nodes_data['nodelist'][0]['location'], media_id, | |
167 | note='Downloading m3u8 information for format %s' % format_id) | |
168 | ||
169 | m3u8_data = self.decrypt_m3u8(req.read()) | |
7f09a662 YCH |
170 | |
171 | url_info_dict = { | |
05a3879f | 172 | 'url': encode_data_uri(m3u8_data, 'application/vnd.apple.mpegurl'), |
91410c9b PH |
173 | 'ext': determine_ext(dispatch[format_id][1]), |
174 | 'format_id': format_id, | |
1e399778 | 175 | 'protocol': 'm3u8', |
7f09a662 YCH |
176 | } |
177 | ||
178 | if format_id[-1:] == 'p': | |
593ddd85 | 179 | url_info_dict['height'] = int_or_none(format_id[:-1]) |
7f09a662 YCH |
180 | |
181 | urls.append(url_info_dict) | |
182 | ||
183 | publish_time = parse_iso8601(self._html_search_regex( | |
91410c9b | 184 | r'发布时间 ([^<>]+) ', page, 'publish time', default=None), |
7f09a662 | 185 | delimiter=' ', timezone=datetime.timedelta(hours=8)) |
67706359 | 186 | description = self._html_search_meta('description', page, fatal=False) |
7f09a662 YCH |
187 | |
188 | return { | |
189 | 'id': media_id, | |
190 | 'formats': urls, | |
191 | 'title': playurl['title'], | |
192 | 'thumbnail': playurl['pic'], | |
67706359 | 193 | 'description': description, |
7f09a662 YCH |
194 | 'timestamp': publish_time, |
195 | } | |
57031161 YCH |
196 | |
197 | ||
8c5a9944 | 198 | class LePlaylistIE(InfoExtractor): |
2eb25b25 YCH |
199 | _VALID_URL = r'http://[a-z]+\.le\.com/[a-z]+/(?P<id>[a-z0-9_]+)' |
200 | ||
57031161 | 201 | _TESTS = [{ |
5e73a67d | 202 | 'url': 'http://www.le.com/tv/46177.html', |
57031161 YCH |
203 | 'info_dict': { |
204 | 'id': '46177', | |
205 | 'title': '美人天下', | |
206 | 'description': 'md5:395666ff41b44080396e59570dbac01c' | |
207 | }, | |
208 | 'playlist_count': 35 | |
2eb25b25 | 209 | }, { |
5e73a67d | 210 | 'url': 'http://tv.le.com/izt/wuzetian/index.html', |
57031161 YCH |
211 | 'info_dict': { |
212 | 'id': 'wuzetian', | |
213 | 'title': '武媚娘传奇', | |
214 | 'description': 'md5:e12499475ab3d50219e5bba00b3cb248' | |
215 | }, | |
67706359 YCH |
216 | # This playlist contains some extra videos other than the drama itself |
217 | 'playlist_mincount': 96 | |
57031161 | 218 | }, { |
5e73a67d | 219 | 'url': 'http://tv.le.com/pzt/lswjzzjc/index.shtml', |
57031161 YCH |
220 | 'info_dict': { |
221 | 'id': 'lswjzzjc', | |
67706359 YCH |
222 | # The title should be "劲舞青春", but I can't find a simple way to |
223 | # determine the playlist title | |
57031161 YCH |
224 | 'title': '乐视午间自制剧场', |
225 | 'description': 'md5:b1eef244f45589a7b5b1af9ff25a4489' | |
226 | }, | |
227 | 'playlist_mincount': 7 | |
2eb25b25 YCH |
228 | }, { |
229 | 'url': 'http://www.le.com/comic/92063.html', | |
230 | 'only_matching': True, | |
231 | }, { | |
232 | 'url': 'http://list.le.com/listn/c1009_sc532002_d2_p1_o1.html', | |
233 | 'only_matching': True, | |
57031161 | 234 | }] |
5f432ac8 | 235 | |
2eb25b25 YCH |
236 | @classmethod |
237 | def suitable(cls, url): | |
8c5a9944 | 238 | return False if LeIE.suitable(url) else super(LePlaylistIE, cls).suitable(url) |
2eb25b25 YCH |
239 | |
240 | def _real_extract(self, url): | |
241 | playlist_id = self._match_id(url) | |
242 | page = self._download_webpage(url, playlist_id) | |
243 | ||
244 | # Currently old domain names are still used in playlists | |
245 | media_ids = orderedSet(re.findall( | |
246 | r'<a[^>]+href="http://www\.letv\.com/ptv/vplay/(\d+)\.html', page)) | |
8c5a9944 | 247 | entries = [self.url_result(LeIE._URL_TEMPLATE % media_id, ie='Le') |
2eb25b25 YCH |
248 | for media_id in media_ids] |
249 | ||
250 | title = self._html_search_meta('keywords', page, | |
251 | fatal=False).split(',')[0] | |
252 | description = self._html_search_meta('description', page, fatal=False) | |
253 | ||
254 | return self.playlist_result(entries, playlist_id, playlist_title=title, | |
255 | playlist_description=description) | |
256 | ||
5f432ac8 FF |
257 | |
258 | class LetvCloudIE(InfoExtractor): | |
8c5a9944 YCH |
259 | # Most of *.letv.com is changed to *.le.com on 2016/01/02 |
260 | # but yuntv.letv.com is kept, so also keep the extractor name | |
5f432ac8 | 261 | IE_DESC = '乐视云' |
73e74424 | 262 | _VALID_URL = r'https?://yuntv\.letv\.com/bcloud.html\?.+' |
5f432ac8 FF |
263 | |
264 | _TESTS = [{ | |
265 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=467623dedf', | |
266 | 'md5': '26450599afd64c513bc77030ad15db44', | |
267 | 'info_dict': { | |
268 | 'id': 'p7jnfw5hw9_467623dedf', | |
269 | 'ext': 'mp4', | |
0428106d | 270 | 'title': 'Video p7jnfw5hw9_467623dedf', |
5f432ac8 FF |
271 | }, |
272 | }, { | |
273 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=ec93197892&pu=2c7cd40209&auto_play=1&gpcflag=1&width=640&height=360', | |
7b7507d6 | 274 | 'md5': 'e03d9cc8d9c13191e1caf277e42dbd31', |
5f432ac8 FF |
275 | 'info_dict': { |
276 | 'id': 'p7jnfw5hw9_ec93197892', | |
277 | 'ext': 'mp4', | |
0428106d | 278 | 'title': 'Video p7jnfw5hw9_ec93197892', |
5f432ac8 FF |
279 | }, |
280 | }, { | |
281 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=187060b6fd', | |
7b7507d6 | 282 | 'md5': 'cb988699a776b22d4a41b9d43acfb3ac', |
5f432ac8 FF |
283 | 'info_dict': { |
284 | 'id': 'p7jnfw5hw9_187060b6fd', | |
285 | 'ext': 'mp4', | |
0428106d | 286 | 'title': 'Video p7jnfw5hw9_187060b6fd', |
5f432ac8 FF |
287 | }, |
288 | }] | |
289 | ||
7b7507d6 YCH |
290 | @staticmethod |
291 | def sign_data(obj): | |
292 | if obj['cf'] == 'flash': | |
293 | salt = '2f9d6924b33a165a6d8b5d3d42f4f987' | |
294 | items = ['cf', 'format', 'ran', 'uu', 'ver', 'vu'] | |
295 | elif obj['cf'] == 'html5': | |
296 | salt = 'fbeh5player12c43eccf2bec3300344' | |
297 | items = ['cf', 'ran', 'uu', 'bver', 'vu'] | |
298 | input_data = ''.join([item + obj[item] for item in items]) + salt | |
299 | obj['sign'] = hashlib.md5(input_data.encode('utf-8')).hexdigest() | |
300 | ||
301 | def _get_formats(self, cf, uu, vu, media_id): | |
302 | def get_play_json(cf, timestamp): | |
303 | data = { | |
304 | 'cf': cf, | |
305 | 'ver': '2.2', | |
306 | 'bver': 'firefox44.0', | |
307 | 'format': 'json', | |
308 | 'uu': uu, | |
309 | 'vu': vu, | |
310 | 'ran': compat_str(timestamp), | |
311 | } | |
312 | self.sign_data(data) | |
313 | return self._download_json( | |
314 | 'http://api.letvcloud.com/gpc.php?' + compat_urllib_parse.urlencode(data), | |
315 | media_id, 'Downloading playJson data for type %s' % cf) | |
316 | ||
317 | play_json = get_play_json(cf, time.time()) | |
318 | # The server time may be different from local time | |
319 | if play_json.get('code') == 10071: | |
320 | play_json = get_play_json(cf, play_json['timestamp']) | |
5f432ac8 | 321 | |
26de1bba YCH |
322 | if not play_json.get('data'): |
323 | if play_json.get('message'): | |
324 | raise ExtractorError('Letv cloud said: %s' % play_json['message'], expected=True) | |
325 | elif play_json.get('code'): | |
326 | raise ExtractorError('Letv cloud returned error %d' % play_json['code'], expected=True) | |
327 | else: | |
328 | raise ExtractorError('Letv cloud returned an unknwon error') | |
329 | ||
bec30224 YCH |
330 | def b64decode(s): |
331 | return base64.b64decode(s.encode('utf-8')).decode('utf-8') | |
332 | ||
10defdd0 YCH |
333 | formats = [] |
334 | for media in play_json['data']['video_info']['media'].values(): | |
335 | play_url = media['play_url'] | |
bec30224 YCH |
336 | url = b64decode(play_url['main_url']) |
337 | decoded_url = b64decode(url_basename(url)) | |
10defdd0 | 338 | formats.append({ |
bec30224 YCH |
339 | 'url': url, |
340 | 'ext': determine_ext(decoded_url), | |
10defdd0 YCH |
341 | 'format_id': int_or_none(play_url.get('vtype')), |
342 | 'format_note': str_or_none(play_url.get('definition')), | |
343 | 'width': int_or_none(play_url.get('vwidth')), | |
344 | 'height': int_or_none(play_url.get('vheight')), | |
345 | }) | |
7b7507d6 YCH |
346 | |
347 | return formats | |
348 | ||
349 | def _real_extract(self, url): | |
350 | uu_mobj = re.search('uu=([\w]+)', url) | |
351 | vu_mobj = re.search('vu=([\w]+)', url) | |
352 | ||
353 | if not uu_mobj or not vu_mobj: | |
354 | raise ExtractorError('Invalid URL: %s' % url, expected=True) | |
355 | ||
356 | uu = uu_mobj.group(1) | |
357 | vu = vu_mobj.group(1) | |
358 | media_id = uu + '_' + vu | |
359 | ||
360 | formats = self._get_formats('flash', uu, vu, media_id) + self._get_formats('html5', uu, vu, media_id) | |
5f432ac8 FF |
361 | self._sort_formats(formats) |
362 | ||
363 | return { | |
364 | 'id': media_id, | |
0428106d | 365 | 'title': 'Video %s' % media_id, |
5f432ac8 FF |
366 | 'formats': formats, |
367 | } |