]>
Commit | Line | Data |
---|---|---|
265bfa2c | 1 | import datetime |
7a8fef31 | 2 | import hashlib |
57031161 | 3 | import re |
7f09a662 | 4 | import time |
7f09a662 YCH |
5 | |
6 | from .common import InfoExtractor | |
265bfa2c | 7 | from ..compat import ( |
cf282071 | 8 | compat_b64decode, |
1e399778 | 9 | compat_ord, |
7b7507d6 | 10 | compat_str, |
15707c7e | 11 | compat_urllib_parse_urlencode, |
265bfa2c PH |
12 | ) |
13 | from ..utils import ( | |
14 | determine_ext, | |
7a8fef31 | 15 | encode_data_uri, |
265bfa2c | 16 | ExtractorError, |
7a8fef31 YCH |
17 | int_or_none, |
18 | orderedSet, | |
265bfa2c | 19 | parse_iso8601, |
5f432ac8 | 20 | str_or_none, |
bec30224 | 21 | url_basename, |
30105f4a | 22 | urshift, |
265bfa2c | 23 | ) |
7f09a662 YCH |
24 | |
25 | ||
8c5a9944 | 26 | class LeIE(InfoExtractor): |
963d0ce7 | 27 | IE_DESC = '乐视网' |
0d72ff9c | 28 | _VALID_URL = r'https?://(?:www\.le\.com/ptv/vplay|(?:sports\.le|(?:www\.)?lesports)\.com/(?:match|video))/(?P<id>\d+)\.html' |
bc61c80c | 29 | _GEO_COUNTRIES = ['CN'] |
f3bc19a9 | 30 | _URL_TEMPLATE = 'http://www.le.com/ptv/vplay/%s.html' |
7f09a662 YCH |
31 | |
32 | _TESTS = [{ | |
5e73a67d | 33 | 'url': 'http://www.le.com/ptv/vplay/22005890.html', |
1e399778 | 34 | 'md5': 'edadcfe5406976f42f9f266057ee5e40', |
7f09a662 YCH |
35 | 'info_dict': { |
36 | 'id': '22005890', | |
37 | 'ext': 'mp4', | |
38 | 'title': '第87届奥斯卡颁奖礼完美落幕 《鸟人》成最大赢家', | |
67706359 | 39 | 'description': 'md5:a9cb175fd753e2962176b7beca21a47c', |
1e399778 YCH |
40 | }, |
41 | 'params': { | |
42 | 'hls_prefer_native': True, | |
43 | }, | |
7f09a662 | 44 | }, { |
5e73a67d | 45 | 'url': 'http://www.le.com/ptv/vplay/1415246.html', |
7f09a662 | 46 | 'info_dict': { |
67706359 | 47 | 'id': '1415246', |
7f09a662 | 48 | 'ext': 'mp4', |
67706359 | 49 | 'title': '美人天下01', |
5401bea2 | 50 | 'description': 'md5:28942e650e82ed4fcc8e4de919ee854d', |
67706359 | 51 | }, |
1e399778 YCH |
52 | 'params': { |
53 | 'hls_prefer_native': True, | |
54 | }, | |
91410c9b PH |
55 | }, { |
56 | 'note': 'This video is available only in Mainland China, thus a proxy is needed', | |
5e73a67d | 57 | 'url': 'http://www.le.com/ptv/vplay/1118082.html', |
1e399778 | 58 | 'md5': '2424c74948a62e5f31988438979c5ad1', |
91410c9b PH |
59 | 'info_dict': { |
60 | 'id': '1118082', | |
61 | 'ext': 'mp4', | |
62 | 'title': '与龙共舞 完整版', | |
63 | 'description': 'md5:7506a5eeb1722bb9d4068f85024e3986', | |
64 | }, | |
1e399778 YCH |
65 | 'params': { |
66 | 'hls_prefer_native': True, | |
67 | }, | |
47787efa YCH |
68 | }, { |
69 | 'url': 'http://sports.le.com/video/25737697.html', | |
70 | 'only_matching': True, | |
0d72ff9c YCH |
71 | }, { |
72 | 'url': 'http://www.lesports.com/match/1023203003.html', | |
73 | 'only_matching': True, | |
74 | }, { | |
75 | 'url': 'http://sports.le.com/match/1023203003.html', | |
76 | 'only_matching': True, | |
7f09a662 YCH |
77 | }] |
78 | ||
5401bea2 | 79 | # ror() and calc_time_key() are reversed from a embedded swf file in LetvPlayer.swf |
7f09a662 YCH |
80 | def ror(self, param1, param2): |
81 | _loc3_ = 0 | |
82 | while _loc3_ < param2: | |
30105f4a | 83 | param1 = urshift(param1, 1) + ((param1 & 1) << 31) |
7f09a662 YCH |
84 | _loc3_ += 1 |
85 | return param1 | |
86 | ||
265bfa2c | 87 | def calc_time_key(self, param1): |
5401bea2 YCH |
88 | _loc2_ = 185025305 |
89 | return self.ror(param1, _loc2_ % 17) ^ _loc2_ | |
07d7689f | 90 | |
1e399778 YCH |
91 | # see M3U8Encryption class in KLetvPlayer.swf |
92 | @staticmethod | |
93 | def decrypt_m3u8(encrypted_data): | |
94 | if encrypted_data[:5].decode('utf-8').lower() != 'vc_01': | |
95 | return encrypted_data | |
96 | encrypted_data = encrypted_data[5:] | |
97 | ||
2ebd2eac YCH |
98 | _loc4_ = bytearray(2 * len(encrypted_data)) |
99 | for idx, val in enumerate(encrypted_data): | |
100 | b = compat_ord(val) | |
101 | _loc4_[2 * idx] = b // 16 | |
102 | _loc4_[2 * idx + 1] = b % 16 | |
1e399778 YCH |
103 | idx = len(_loc4_) - 11 |
104 | _loc4_ = _loc4_[idx:] + _loc4_[:idx] | |
2ebd2eac YCH |
105 | _loc7_ = bytearray(len(encrypted_data)) |
106 | for i in range(len(encrypted_data)): | |
107 | _loc7_[i] = _loc4_[2 * i] * 16 + _loc4_[2 * i + 1] | |
1e399778 YCH |
108 | |
109 | return bytes(_loc7_) | |
110 | ||
07d7689f | 111 | def _check_errors(self, play_json): |
7f09a662 | 112 | # Check for errors |
5401bea2 | 113 | playstatus = play_json['msgs']['playstatus'] |
7f09a662 YCH |
114 | if playstatus['status'] == 0: |
115 | flag = playstatus['flag'] | |
116 | if flag == 1: | |
bc61c80c | 117 | self.raise_geo_restricted() |
7f09a662 | 118 | else: |
bc61c80c | 119 | raise ExtractorError('Generic error. flag = %d' % flag, expected=True) |
7f09a662 | 120 | |
07d7689f RA |
121 | def _real_extract(self, url): |
122 | media_id = self._match_id(url) | |
123 | page = self._download_webpage(url, media_id) | |
124 | ||
07d7689f | 125 | play_json_flash = self._download_json( |
5401bea2 | 126 | 'http://player-pc.le.com/mms/out/video/playJson', |
07d7689f RA |
127 | media_id, 'Downloading flash playJson data', query={ |
128 | 'id': media_id, | |
129 | 'platid': 1, | |
5c766952 | 130 | 'splatid': 105, |
07d7689f | 131 | 'format': 1, |
5401bea2 | 132 | 'source': 1000, |
07d7689f RA |
133 | 'tkey': self.calc_time_key(int(time.time())), |
134 | 'domain': 'www.le.com', | |
5401bea2 | 135 | 'region': 'cn', |
07d7689f RA |
136 | }, |
137 | headers=self.geo_verification_headers()) | |
138 | self._check_errors(play_json_flash) | |
7f09a662 | 139 | |
07d7689f | 140 | def get_flash_urls(media_url, format_id): |
07d7689f RA |
141 | nodes_data = self._download_json( |
142 | media_url, media_id, | |
5401bea2 YCH |
143 | 'Download JSON metadata for format %s' % format_id, |
144 | query={ | |
145 | 'm3v': 1, | |
146 | 'format': 1, | |
147 | 'expect': 3, | |
148 | 'tss': 'ios', | |
149 | }) | |
07d7689f RA |
150 | |
151 | req = self._request_webpage( | |
152 | nodes_data['nodelist'][0]['location'], media_id, | |
153 | note='Downloading m3u8 information for format %s' % format_id) | |
154 | ||
155 | m3u8_data = self.decrypt_m3u8(req.read()) | |
156 | ||
157 | return { | |
158 | 'hls': encode_data_uri(m3u8_data, 'application/vnd.apple.mpegurl'), | |
159 | } | |
160 | ||
161 | extracted_formats = [] | |
162 | formats = [] | |
5401bea2 YCH |
163 | playurl = play_json_flash['msgs']['playurl'] |
164 | play_domain = playurl['domain'][0] | |
165 | ||
166 | for format_id, format_data in playurl.get('dispatch', []).items(): | |
167 | if format_id in extracted_formats: | |
168 | continue | |
169 | extracted_formats.append(format_id) | |
170 | ||
171 | media_url = play_domain + format_data[0] | |
172 | for protocol, format_url in get_flash_urls(media_url, format_id).items(): | |
173 | f = { | |
174 | 'url': format_url, | |
175 | 'ext': determine_ext(format_data[1]), | |
176 | 'format_id': '%s-%s' % (protocol, format_id), | |
177 | 'protocol': 'm3u8_native' if protocol == 'hls' else 'http', | |
178 | 'quality': int_or_none(format_id), | |
179 | } | |
180 | ||
181 | if format_id[-1:] == 'p': | |
182 | f['height'] = int_or_none(format_id[:-1]) | |
183 | ||
184 | formats.append(f) | |
7f09a662 YCH |
185 | |
186 | publish_time = parse_iso8601(self._html_search_regex( | |
91410c9b | 187 | r'发布时间 ([^<>]+) ', page, 'publish time', default=None), |
7f09a662 | 188 | delimiter=' ', timezone=datetime.timedelta(hours=8)) |
67706359 | 189 | description = self._html_search_meta('description', page, fatal=False) |
7f09a662 YCH |
190 | |
191 | return { | |
192 | 'id': media_id, | |
07d7689f | 193 | 'formats': formats, |
7f09a662 YCH |
194 | 'title': playurl['title'], |
195 | 'thumbnail': playurl['pic'], | |
67706359 | 196 | 'description': description, |
7f09a662 | 197 | 'timestamp': publish_time, |
9f14daf2 | 198 | '_format_sort_fields': ('res', 'quality'), |
7f09a662 | 199 | } |
57031161 YCH |
200 | |
201 | ||
8c5a9944 | 202 | class LePlaylistIE(InfoExtractor): |
47787efa | 203 | _VALID_URL = r'https?://[a-z]+\.le\.com/(?!video)[a-z]+/(?P<id>[a-z0-9_]+)' |
2eb25b25 | 204 | |
57031161 | 205 | _TESTS = [{ |
5e73a67d | 206 | 'url': 'http://www.le.com/tv/46177.html', |
57031161 YCH |
207 | 'info_dict': { |
208 | 'id': '46177', | |
209 | 'title': '美人天下', | |
210 | 'description': 'md5:395666ff41b44080396e59570dbac01c' | |
211 | }, | |
212 | 'playlist_count': 35 | |
2eb25b25 | 213 | }, { |
5e73a67d | 214 | 'url': 'http://tv.le.com/izt/wuzetian/index.html', |
57031161 YCH |
215 | 'info_dict': { |
216 | 'id': 'wuzetian', | |
217 | 'title': '武媚娘传奇', | |
218 | 'description': 'md5:e12499475ab3d50219e5bba00b3cb248' | |
219 | }, | |
67706359 YCH |
220 | # This playlist contains some extra videos other than the drama itself |
221 | 'playlist_mincount': 96 | |
57031161 | 222 | }, { |
5e73a67d | 223 | 'url': 'http://tv.le.com/pzt/lswjzzjc/index.shtml', |
d58cb3ec YCH |
224 | # This series is moved to http://www.le.com/tv/10005297.html |
225 | 'only_matching': True, | |
2eb25b25 YCH |
226 | }, { |
227 | 'url': 'http://www.le.com/comic/92063.html', | |
228 | 'only_matching': True, | |
229 | }, { | |
230 | 'url': 'http://list.le.com/listn/c1009_sc532002_d2_p1_o1.html', | |
231 | 'only_matching': True, | |
57031161 | 232 | }] |
5f432ac8 | 233 | |
2eb25b25 YCH |
234 | @classmethod |
235 | def suitable(cls, url): | |
8c5a9944 | 236 | return False if LeIE.suitable(url) else super(LePlaylistIE, cls).suitable(url) |
2eb25b25 YCH |
237 | |
238 | def _real_extract(self, url): | |
239 | playlist_id = self._match_id(url) | |
240 | page = self._download_webpage(url, playlist_id) | |
241 | ||
242 | # Currently old domain names are still used in playlists | |
243 | media_ids = orderedSet(re.findall( | |
244 | r'<a[^>]+href="http://www\.letv\.com/ptv/vplay/(\d+)\.html', page)) | |
8c5a9944 | 245 | entries = [self.url_result(LeIE._URL_TEMPLATE % media_id, ie='Le') |
2eb25b25 YCH |
246 | for media_id in media_ids] |
247 | ||
248 | title = self._html_search_meta('keywords', page, | |
249 | fatal=False).split(',')[0] | |
250 | description = self._html_search_meta('description', page, fatal=False) | |
251 | ||
252 | return self.playlist_result(entries, playlist_id, playlist_title=title, | |
253 | playlist_description=description) | |
254 | ||
5f432ac8 FF |
255 | |
256 | class LetvCloudIE(InfoExtractor): | |
8c5a9944 YCH |
257 | # Most of *.letv.com is changed to *.le.com on 2016/01/02 |
258 | # but yuntv.letv.com is kept, so also keep the extractor name | |
5f432ac8 | 259 | IE_DESC = '乐视云' |
73e74424 | 260 | _VALID_URL = r'https?://yuntv\.letv\.com/bcloud.html\?.+' |
5f432ac8 FF |
261 | |
262 | _TESTS = [{ | |
263 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=467623dedf', | |
264 | 'md5': '26450599afd64c513bc77030ad15db44', | |
265 | 'info_dict': { | |
266 | 'id': 'p7jnfw5hw9_467623dedf', | |
267 | 'ext': 'mp4', | |
0428106d | 268 | 'title': 'Video p7jnfw5hw9_467623dedf', |
5f432ac8 FF |
269 | }, |
270 | }, { | |
271 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=ec93197892&pu=2c7cd40209&auto_play=1&gpcflag=1&width=640&height=360', | |
7b7507d6 | 272 | 'md5': 'e03d9cc8d9c13191e1caf277e42dbd31', |
5f432ac8 FF |
273 | 'info_dict': { |
274 | 'id': 'p7jnfw5hw9_ec93197892', | |
275 | 'ext': 'mp4', | |
0428106d | 276 | 'title': 'Video p7jnfw5hw9_ec93197892', |
5f432ac8 FF |
277 | }, |
278 | }, { | |
279 | 'url': 'http://yuntv.letv.com/bcloud.html?uu=p7jnfw5hw9&vu=187060b6fd', | |
7b7507d6 | 280 | 'md5': 'cb988699a776b22d4a41b9d43acfb3ac', |
5f432ac8 FF |
281 | 'info_dict': { |
282 | 'id': 'p7jnfw5hw9_187060b6fd', | |
283 | 'ext': 'mp4', | |
0428106d | 284 | 'title': 'Video p7jnfw5hw9_187060b6fd', |
5f432ac8 FF |
285 | }, |
286 | }] | |
287 | ||
7b7507d6 YCH |
288 | @staticmethod |
289 | def sign_data(obj): | |
290 | if obj['cf'] == 'flash': | |
291 | salt = '2f9d6924b33a165a6d8b5d3d42f4f987' | |
292 | items = ['cf', 'format', 'ran', 'uu', 'ver', 'vu'] | |
293 | elif obj['cf'] == 'html5': | |
294 | salt = 'fbeh5player12c43eccf2bec3300344' | |
295 | items = ['cf', 'ran', 'uu', 'bver', 'vu'] | |
296 | input_data = ''.join([item + obj[item] for item in items]) + salt | |
297 | obj['sign'] = hashlib.md5(input_data.encode('utf-8')).hexdigest() | |
298 | ||
299 | def _get_formats(self, cf, uu, vu, media_id): | |
300 | def get_play_json(cf, timestamp): | |
301 | data = { | |
302 | 'cf': cf, | |
303 | 'ver': '2.2', | |
304 | 'bver': 'firefox44.0', | |
305 | 'format': 'json', | |
306 | 'uu': uu, | |
307 | 'vu': vu, | |
308 | 'ran': compat_str(timestamp), | |
309 | } | |
310 | self.sign_data(data) | |
311 | return self._download_json( | |
15707c7e | 312 | 'http://api.letvcloud.com/gpc.php?' + compat_urllib_parse_urlencode(data), |
7b7507d6 YCH |
313 | media_id, 'Downloading playJson data for type %s' % cf) |
314 | ||
315 | play_json = get_play_json(cf, time.time()) | |
316 | # The server time may be different from local time | |
317 | if play_json.get('code') == 10071: | |
318 | play_json = get_play_json(cf, play_json['timestamp']) | |
5f432ac8 | 319 | |
26de1bba YCH |
320 | if not play_json.get('data'): |
321 | if play_json.get('message'): | |
322 | raise ExtractorError('Letv cloud said: %s' % play_json['message'], expected=True) | |
323 | elif play_json.get('code'): | |
324 | raise ExtractorError('Letv cloud returned error %d' % play_json['code'], expected=True) | |
325 | else: | |
843ad179 | 326 | raise ExtractorError('Letv cloud returned an unknown error') |
26de1bba | 327 | |
bec30224 | 328 | def b64decode(s): |
cf282071 | 329 | return compat_b64decode(s).decode('utf-8') |
bec30224 | 330 | |
10defdd0 YCH |
331 | formats = [] |
332 | for media in play_json['data']['video_info']['media'].values(): | |
333 | play_url = media['play_url'] | |
bec30224 YCH |
334 | url = b64decode(play_url['main_url']) |
335 | decoded_url = b64decode(url_basename(url)) | |
10defdd0 | 336 | formats.append({ |
bec30224 YCH |
337 | 'url': url, |
338 | 'ext': determine_ext(decoded_url), | |
8a370aed | 339 | 'format_id': str_or_none(play_url.get('vtype')), |
10defdd0 YCH |
340 | 'format_note': str_or_none(play_url.get('definition')), |
341 | 'width': int_or_none(play_url.get('vwidth')), | |
342 | 'height': int_or_none(play_url.get('vheight')), | |
343 | }) | |
7b7507d6 YCH |
344 | |
345 | return formats | |
346 | ||
347 | def _real_extract(self, url): | |
ec85ded8 YCH |
348 | uu_mobj = re.search(r'uu=([\w]+)', url) |
349 | vu_mobj = re.search(r'vu=([\w]+)', url) | |
7b7507d6 YCH |
350 | |
351 | if not uu_mobj or not vu_mobj: | |
352 | raise ExtractorError('Invalid URL: %s' % url, expected=True) | |
353 | ||
354 | uu = uu_mobj.group(1) | |
355 | vu = vu_mobj.group(1) | |
356 | media_id = uu + '_' + vu | |
357 | ||
358 | formats = self._get_formats('flash', uu, vu, media_id) + self._get_formats('html5', uu, vu, media_id) | |
5f432ac8 FF |
359 | |
360 | return { | |
361 | 'id': media_id, | |
0428106d | 362 | 'title': 'Video %s' % media_id, |
5f432ac8 FF |
363 | 'formats': formats, |
364 | } |