]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/iqiyi.py
ddcb3c916e6a0610484dc5ceddbd84b507e761fd
[youtubedl] / youtube_dl / extractor / iqiyi.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import hashlib
5 import itertools
6 import math
7 import os
8 import random
9 import re
10 import time
11 import uuid
12
13 from .common import InfoExtractor
14 from ..compat import (
15 compat_parse_qs,
16 compat_str,
17 compat_urllib_parse_urlencode,
18 compat_urllib_parse_urlparse,
19 )
20 from ..utils import (
21 decode_packed_codes,
22 ExtractorError,
23 ohdave_rsa_encrypt,
24 remove_start,
25 sanitized_Request,
26 urlencode_postdata,
27 url_basename,
28 )
29
30
31 def md5_text(text):
32 return hashlib.md5(text.encode('utf-8')).hexdigest()
33
34
35 class IqiyiSDK(object):
36 def __init__(self, target, ip, timestamp):
37 self.target = target
38 self.ip = ip
39 self.timestamp = timestamp
40
41 @staticmethod
42 def split_sum(data):
43 return compat_str(sum(map(lambda p: int(p, 16), list(data))))
44
45 @staticmethod
46 def digit_sum(num):
47 if isinstance(num, int):
48 num = compat_str(num)
49 return compat_str(sum(map(int, num)))
50
51 def even_odd(self):
52 even = self.digit_sum(compat_str(self.timestamp)[::2])
53 odd = self.digit_sum(compat_str(self.timestamp)[1::2])
54 return even, odd
55
56 def preprocess(self, chunksize):
57 self.target = md5_text(self.target)
58 chunks = []
59 for i in range(32 // chunksize):
60 chunks.append(self.target[chunksize * i:chunksize * (i + 1)])
61 if 32 % chunksize:
62 chunks.append(self.target[32 - 32 % chunksize:])
63 return chunks, list(map(int, self.ip.split('.')))
64
65 def mod(self, modulus):
66 chunks, ip = self.preprocess(32)
67 self.target = chunks[0] + ''.join(map(lambda p: compat_str(p % modulus), ip))
68
69 def split(self, chunksize):
70 modulus_map = {
71 4: 256,
72 5: 10,
73 8: 100,
74 }
75
76 chunks, ip = self.preprocess(chunksize)
77 ret = ''
78 for i in range(len(chunks)):
79 ip_part = compat_str(ip[i] % modulus_map[chunksize]) if i < 4 else ''
80 if chunksize == 8:
81 ret += ip_part + chunks[i]
82 else:
83 ret += chunks[i] + ip_part
84 self.target = ret
85
86 def handle_input16(self):
87 self.target = md5_text(self.target)
88 self.target = self.split_sum(self.target[:16]) + self.target + self.split_sum(self.target[16:])
89
90 def handle_input8(self):
91 self.target = md5_text(self.target)
92 ret = ''
93 for i in range(4):
94 part = self.target[8 * i:8 * (i + 1)]
95 ret += self.split_sum(part) + part
96 self.target = ret
97
98 def handleSum(self):
99 self.target = md5_text(self.target)
100 self.target = self.split_sum(self.target) + self.target
101
102 def date(self, scheme):
103 self.target = md5_text(self.target)
104 d = time.localtime(self.timestamp)
105 strings = {
106 'y': compat_str(d.tm_year),
107 'm': '%02d' % d.tm_mon,
108 'd': '%02d' % d.tm_mday,
109 }
110 self.target += ''.join(map(lambda c: strings[c], list(scheme)))
111
112 def split_time_even_odd(self):
113 even, odd = self.even_odd()
114 self.target = odd + md5_text(self.target) + even
115
116 def split_time_odd_even(self):
117 even, odd = self.even_odd()
118 self.target = even + md5_text(self.target) + odd
119
120 def split_ip_time_sum(self):
121 chunks, ip = self.preprocess(32)
122 self.target = compat_str(sum(ip)) + chunks[0] + self.digit_sum(self.timestamp)
123
124 def split_time_ip_sum(self):
125 chunks, ip = self.preprocess(32)
126 self.target = self.digit_sum(self.timestamp) + chunks[0] + compat_str(sum(ip))
127
128
129 class IqiyiSDKInterpreter(object):
130 def __init__(self, sdk_code):
131 self.sdk_code = sdk_code
132
133 def run(self, target, ip, timestamp):
134 self.sdk_code = decode_packed_codes(self.sdk_code)
135
136 functions = re.findall(r'input=([a-zA-Z0-9]+)\(input', self.sdk_code)
137
138 sdk = IqiyiSDK(target, ip, timestamp)
139
140 other_functions = {
141 'handleSum': sdk.handleSum,
142 'handleInput8': sdk.handle_input8,
143 'handleInput16': sdk.handle_input16,
144 'splitTimeEvenOdd': sdk.split_time_even_odd,
145 'splitTimeOddEven': sdk.split_time_odd_even,
146 'splitIpTimeSum': sdk.split_ip_time_sum,
147 'splitTimeIpSum': sdk.split_time_ip_sum,
148 }
149 for function in functions:
150 if re.match(r'mod\d+', function):
151 sdk.mod(int(function[3:]))
152 elif re.match(r'date[ymd]{3}', function):
153 sdk.date(function[4:])
154 elif re.match(r'split\d+', function):
155 sdk.split(int(function[5:]))
156 elif function in other_functions:
157 other_functions[function]()
158 else:
159 raise ExtractorError('Unknown funcion %s' % function)
160
161 return sdk.target
162
163
164 class IqiyiIE(InfoExtractor):
165 IE_NAME = 'iqiyi'
166 IE_DESC = '爱奇艺'
167
168 _VALID_URL = r'https?://(?:(?:[^.]+\.)?iqiyi\.com|www\.pps\.tv)/.+\.html'
169
170 _NETRC_MACHINE = 'iqiyi'
171
172 _TESTS = [{
173 'url': 'http://www.iqiyi.com/v_19rrojlavg.html',
174 'md5': '2cb594dc2781e6c941a110d8f358118b',
175 'info_dict': {
176 'id': '9c1fb1b99d192b21c559e5a1a2cb3c73',
177 'title': '美国德州空中惊现奇异云团 酷似UFO',
178 'ext': 'f4v',
179 }
180 }, {
181 'url': 'http://www.iqiyi.com/v_19rrhnnclk.html',
182 'info_dict': {
183 'id': 'e3f585b550a280af23c98b6cb2be19fb',
184 'title': '名侦探柯南第752集',
185 },
186 'playlist': [{
187 'info_dict': {
188 'id': 'e3f585b550a280af23c98b6cb2be19fb_part1',
189 'ext': 'f4v',
190 'title': '名侦探柯南第752集',
191 },
192 }, {
193 'info_dict': {
194 'id': 'e3f585b550a280af23c98b6cb2be19fb_part2',
195 'ext': 'f4v',
196 'title': '名侦探柯南第752集',
197 },
198 }, {
199 'info_dict': {
200 'id': 'e3f585b550a280af23c98b6cb2be19fb_part3',
201 'ext': 'f4v',
202 'title': '名侦探柯南第752集',
203 },
204 }, {
205 'info_dict': {
206 'id': 'e3f585b550a280af23c98b6cb2be19fb_part4',
207 'ext': 'f4v',
208 'title': '名侦探柯南第752集',
209 },
210 }, {
211 'info_dict': {
212 'id': 'e3f585b550a280af23c98b6cb2be19fb_part5',
213 'ext': 'f4v',
214 'title': '名侦探柯南第752集',
215 },
216 }, {
217 'info_dict': {
218 'id': 'e3f585b550a280af23c98b6cb2be19fb_part6',
219 'ext': 'f4v',
220 'title': '名侦探柯南第752集',
221 },
222 }, {
223 'info_dict': {
224 'id': 'e3f585b550a280af23c98b6cb2be19fb_part7',
225 'ext': 'f4v',
226 'title': '名侦探柯南第752集',
227 },
228 }, {
229 'info_dict': {
230 'id': 'e3f585b550a280af23c98b6cb2be19fb_part8',
231 'ext': 'f4v',
232 'title': '名侦探柯南第752集',
233 },
234 }],
235 'params': {
236 'skip_download': True,
237 },
238 }, {
239 'url': 'http://www.iqiyi.com/w_19rt6o8t9p.html',
240 'only_matching': True,
241 }, {
242 'url': 'http://www.iqiyi.com/a_19rrhbc6kt.html',
243 'only_matching': True,
244 }, {
245 'url': 'http://yule.iqiyi.com/pcb.html',
246 'only_matching': True,
247 }, {
248 # VIP-only video. The first 2 parts (6 minutes) are available without login
249 # MD5 sums omitted as values are different on Travis CI and my machine
250 'url': 'http://www.iqiyi.com/v_19rrny4w8w.html',
251 'info_dict': {
252 'id': 'f3cf468b39dddb30d676f89a91200dc1',
253 'title': '泰坦尼克号',
254 },
255 'playlist': [{
256 'info_dict': {
257 'id': 'f3cf468b39dddb30d676f89a91200dc1_part1',
258 'ext': 'f4v',
259 'title': '泰坦尼克号',
260 },
261 }, {
262 'info_dict': {
263 'id': 'f3cf468b39dddb30d676f89a91200dc1_part2',
264 'ext': 'f4v',
265 'title': '泰坦尼克号',
266 },
267 }],
268 'expected_warnings': ['Needs a VIP account for full video'],
269 }, {
270 'url': 'http://www.iqiyi.com/a_19rrhb8ce1.html',
271 'info_dict': {
272 'id': '202918101',
273 'title': '灌篮高手 国语版',
274 },
275 'playlist_count': 101,
276 }, {
277 'url': 'http://www.pps.tv/w_19rrbav0ph.html',
278 'only_matching': True,
279 }]
280
281 _FORMATS_MAP = [
282 ('1', 'h6'),
283 ('2', 'h5'),
284 ('3', 'h4'),
285 ('4', 'h3'),
286 ('5', 'h2'),
287 ('10', 'h1'),
288 ]
289
290 AUTH_API_ERRORS = {
291 # No preview available (不允许试看鉴权失败)
292 'Q00505': 'This video requires a VIP account',
293 # End of preview time (试看结束鉴权失败)
294 'Q00506': 'Needs a VIP account for full video',
295 }
296
297 def _real_initialize(self):
298 self._login()
299
300 @staticmethod
301 def _rsa_fun(data):
302 # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
303 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
304 e = 65537
305
306 return ohdave_rsa_encrypt(data, e, N)
307
308 def _login(self):
309 (username, password) = self._get_login_info()
310
311 # No authentication to be performed
312 if not username:
313 return True
314
315 data = self._download_json(
316 'http://kylin.iqiyi.com/get_token', None,
317 note='Get token for logging', errnote='Unable to get token for logging')
318 sdk = data['sdk']
319 timestamp = int(time.time())
320 target = '/apis/reglogin/login.action?lang=zh_TW&area_code=null&email=%s&passwd=%s&agenttype=1&from=undefined&keeplogin=0&piccode=&fromurl=&_pos=1' % (
321 username, self._rsa_fun(password.encode('utf-8')))
322
323 interp = IqiyiSDKInterpreter(sdk)
324 sign = interp.run(target, data['ip'], timestamp)
325
326 validation_params = {
327 'target': target,
328 'server': 'BEA3AA1908656AABCCFF76582C4C6660',
329 'token': data['token'],
330 'bird_src': 'f8d91d57af224da7893dd397d52d811a',
331 'sign': sign,
332 'bird_t': timestamp,
333 }
334 validation_result = self._download_json(
335 'http://kylin.iqiyi.com/validate?' + compat_urllib_parse_urlencode(validation_params), None,
336 note='Validate credentials', errnote='Unable to validate credentials')
337
338 MSG_MAP = {
339 'P00107': 'please login via the web interface and enter the CAPTCHA code',
340 'P00117': 'bad username or password',
341 }
342
343 code = validation_result['code']
344 if code != 'A00000':
345 msg = MSG_MAP.get(code)
346 if not msg:
347 msg = 'error %s' % code
348 if validation_result.get('msg'):
349 msg += ': ' + validation_result['msg']
350 self._downloader.report_warning('unable to log in: ' + msg)
351 return False
352
353 return True
354
355 def _authenticate_vip_video(self, api_video_url, video_id, tvid, _uuid, do_report_warning):
356 auth_params = {
357 # version and platform hard-coded in com/qiyi/player/core/model/remote/AuthenticationRemote.as
358 'version': '2.0',
359 'platform': 'b6c13e26323c537d',
360 'aid': tvid,
361 'tvid': tvid,
362 'uid': '',
363 'deviceId': _uuid,
364 'playType': 'main', # XXX: always main?
365 'filename': os.path.splitext(url_basename(api_video_url))[0],
366 }
367
368 qd_items = compat_parse_qs(compat_urllib_parse_urlparse(api_video_url).query)
369 for key, val in qd_items.items():
370 auth_params[key] = val[0]
371
372 auth_req = sanitized_Request(
373 'http://api.vip.iqiyi.com/services/ckn.action',
374 urlencode_postdata(auth_params))
375 # iQiyi server throws HTTP 405 error without the following header
376 auth_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
377 auth_result = self._download_json(
378 auth_req, video_id,
379 note='Downloading video authentication JSON',
380 errnote='Unable to download video authentication JSON')
381
382 code = auth_result.get('code')
383 msg = self.AUTH_API_ERRORS.get(code) or auth_result.get('msg') or code
384 if code == 'Q00506':
385 if do_report_warning:
386 self.report_warning(msg)
387 return False
388 if 'data' not in auth_result:
389 if msg is not None:
390 raise ExtractorError('%s said: %s' % (self.IE_NAME, msg), expected=True)
391 raise ExtractorError('Unexpected error from Iqiyi auth API')
392
393 return auth_result['data']
394
395 def construct_video_urls(self, data, video_id, _uuid, tvid):
396 def do_xor(x, y):
397 a = y % 3
398 if a == 1:
399 return x ^ 121
400 if a == 2:
401 return x ^ 72
402 return x ^ 103
403
404 def get_encode_code(l):
405 a = 0
406 b = l.split('-')
407 c = len(b)
408 s = ''
409 for i in range(c - 1, -1, -1):
410 a = do_xor(int(b[c - i - 1], 16), i)
411 s += chr(a)
412 return s[::-1]
413
414 def get_path_key(x, format_id, segment_index):
415 mg = ')(*&^flash@#$%a'
416 tm = self._download_json(
417 'http://data.video.qiyi.com/t?tn=' + str(random.random()), video_id,
418 note='Download path key of segment %d for format %s' % (segment_index + 1, format_id)
419 )['t']
420 t = str(int(math.floor(int(tm) / (600.0))))
421 return md5_text(t + mg + x)
422
423 video_urls_dict = {}
424 need_vip_warning_report = True
425 for format_item in data['vp']['tkl'][0]['vs']:
426 if 0 < int(format_item['bid']) <= 10:
427 format_id = self.get_format(format_item['bid'])
428 else:
429 continue
430
431 video_urls = []
432
433 video_urls_info = format_item['fs']
434 if not format_item['fs'][0]['l'].startswith('/'):
435 t = get_encode_code(format_item['fs'][0]['l'])
436 if t.endswith('mp4'):
437 video_urls_info = format_item['flvs']
438
439 for segment_index, segment in enumerate(video_urls_info):
440 vl = segment['l']
441 if not vl.startswith('/'):
442 vl = get_encode_code(vl)
443 is_vip_video = '/vip/' in vl
444 filesize = segment['b']
445 base_url = data['vp']['du'].split('/')
446 if not is_vip_video:
447 key = get_path_key(
448 vl.split('/')[-1].split('.')[0], format_id, segment_index)
449 base_url.insert(-1, key)
450 base_url = '/'.join(base_url)
451 param = {
452 'su': _uuid,
453 'qyid': uuid.uuid4().hex,
454 'client': '',
455 'z': '',
456 'bt': '',
457 'ct': '',
458 'tn': str(int(time.time()))
459 }
460 api_video_url = base_url + vl
461 if is_vip_video:
462 api_video_url = api_video_url.replace('.f4v', '.hml')
463 auth_result = self._authenticate_vip_video(
464 api_video_url, video_id, tvid, _uuid, need_vip_warning_report)
465 if auth_result is False:
466 need_vip_warning_report = False
467 break
468 param.update({
469 't': auth_result['t'],
470 # cid is hard-coded in com/qiyi/player/core/player/RuntimeData.as
471 'cid': 'afbe8fd3d73448c9',
472 'vid': video_id,
473 'QY00001': auth_result['u'],
474 })
475 api_video_url += '?' if '?' not in api_video_url else '&'
476 api_video_url += compat_urllib_parse_urlencode(param)
477 js = self._download_json(
478 api_video_url, video_id,
479 note='Download video info of segment %d for format %s' % (segment_index + 1, format_id))
480 video_url = js['l']
481 video_urls.append(
482 (video_url, filesize))
483
484 video_urls_dict[format_id] = video_urls
485 return video_urls_dict
486
487 def get_format(self, bid):
488 matched_format_ids = [_format_id for _bid, _format_id in self._FORMATS_MAP if _bid == str(bid)]
489 return matched_format_ids[0] if len(matched_format_ids) else None
490
491 def get_bid(self, format_id):
492 matched_bids = [_bid for _bid, _format_id in self._FORMATS_MAP if _format_id == format_id]
493 return matched_bids[0] if len(matched_bids) else None
494
495 def get_raw_data(self, tvid, video_id, enc_key, _uuid):
496 tm = str(int(time.time()))
497 tail = tm + tvid
498 param = {
499 'key': 'fvip',
500 'src': md5_text('youtube-dl'),
501 'tvId': tvid,
502 'vid': video_id,
503 'vinfo': 1,
504 'tm': tm,
505 'enc': md5_text(enc_key + tail),
506 'qyid': _uuid,
507 'tn': random.random(),
508 # In iQiyi's flash player, um is set to 1 if there's a logged user
509 # Some 1080P formats are only available with a logged user.
510 # Here force um=1 to trick the iQiyi server
511 'um': 1,
512 'authkey': md5_text(md5_text('') + tail),
513 'k_tag': 1,
514 }
515
516 api_url = 'http://cache.video.qiyi.com/vms' + '?' + \
517 compat_urllib_parse_urlencode(param)
518 raw_data = self._download_json(api_url, video_id)
519 return raw_data
520
521 def get_enc_key(self, video_id):
522 # TODO: automatic key extraction
523 # last update at 2016-01-22 for Zombie::bite
524 enc_key = '4a1caba4b4465345366f28da7c117d20'
525 return enc_key
526
527 def _extract_playlist(self, webpage):
528 PAGE_SIZE = 50
529
530 links = re.findall(
531 r'<a[^>]+class="site-piclist_pic_link"[^>]+href="(http://www\.iqiyi\.com/.+\.html)"',
532 webpage)
533 if not links:
534 return
535
536 album_id = self._search_regex(
537 r'albumId\s*:\s*(\d+),', webpage, 'album ID')
538 album_title = self._search_regex(
539 r'data-share-title="([^"]+)"', webpage, 'album title', fatal=False)
540
541 entries = list(map(self.url_result, links))
542
543 # Start from 2 because links in the first page are already on webpage
544 for page_num in itertools.count(2):
545 pagelist_page = self._download_webpage(
546 'http://cache.video.qiyi.com/jp/avlist/%s/%d/%d/' % (album_id, page_num, PAGE_SIZE),
547 album_id,
548 note='Download playlist page %d' % page_num,
549 errnote='Failed to download playlist page %d' % page_num)
550 pagelist = self._parse_json(
551 remove_start(pagelist_page, 'var tvInfoJs='), album_id)
552 vlist = pagelist['data']['vlist']
553 for item in vlist:
554 entries.append(self.url_result(item['vurl']))
555 if len(vlist) < PAGE_SIZE:
556 break
557
558 return self.playlist_result(entries, album_id, album_title)
559
560 def _real_extract(self, url):
561 webpage = self._download_webpage(
562 url, 'temp_id', note='download video page')
563
564 # There's no simple way to determine whether an URL is a playlist or not
565 # So detect it
566 playlist_result = self._extract_playlist(webpage)
567 if playlist_result:
568 return playlist_result
569
570 tvid = self._search_regex(
571 r'data-player-tvid\s*=\s*[\'"](\d+)', webpage, 'tvid')
572 video_id = self._search_regex(
573 r'data-player-videoid\s*=\s*[\'"]([a-f\d]+)', webpage, 'video_id')
574 _uuid = uuid.uuid4().hex
575
576 enc_key = self.get_enc_key(video_id)
577
578 raw_data = self.get_raw_data(tvid, video_id, enc_key, _uuid)
579
580 if raw_data['code'] != 'A000000':
581 raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
582
583 data = raw_data['data']
584
585 title = data['vi']['vn']
586
587 # generate video_urls_dict
588 video_urls_dict = self.construct_video_urls(
589 data, video_id, _uuid, tvid)
590
591 # construct info
592 entries = []
593 for format_id in video_urls_dict:
594 video_urls = video_urls_dict[format_id]
595 for i, video_url_info in enumerate(video_urls):
596 if len(entries) < i + 1:
597 entries.append({'formats': []})
598 entries[i]['formats'].append(
599 {
600 'url': video_url_info[0],
601 'filesize': video_url_info[-1],
602 'format_id': format_id,
603 'preference': int(self.get_bid(format_id))
604 }
605 )
606
607 for i in range(len(entries)):
608 self._sort_formats(entries[i]['formats'])
609 entries[i].update(
610 {
611 'id': '%s_part%d' % (video_id, i + 1),
612 'title': title,
613 }
614 )
615
616 if len(entries) > 1:
617 info = {
618 '_type': 'multi_video',
619 'id': video_id,
620 'title': title,
621 'entries': entries,
622 }
623 else:
624 info = entries[0]
625 info['id'] = video_id
626 info['title'] = title
627
628 return info