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,
  18     compat_urllib_parse_urlparse,
  19 )
  20 from ..utils import (
  21     base_n,
  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 decode_eval_codes(self):
 134         self.sdk_code = self.sdk_code[5:-3]
 135
 136         mobj = re.search(
 137             r"'([^']+)',62,(\d+),'([^']+)'\.split\('\|'\),[^,]+,{}",
 138             self.sdk_code)
 139         obfucasted_code, count, symbols = mobj.groups()
 140         count = int(count)
 141         symbols = symbols.split('|')
 142         symbol_table = {}
 143
 144         while count:
 145             count -= 1
 146             b62count = base_n(count, 62)
 147             symbol_table[b62count] = symbols[count] or b62count
 148
 149         self.sdk_code = re.sub(
 150             r'\b(\w+)\b', lambda mobj: symbol_table[mobj.group(0)],
 151             obfucasted_code)
 152
 153     def run(self, target, ip, timestamp):
 154         self.decode_eval_codes()
 155
 156         functions = re.findall(r'input=([a-zA-Z0-9]+)\(input', self.sdk_code)
 157
 158         sdk = IqiyiSDK(target, ip, timestamp)
 159
 160         other_functions = {
 161             'handleSum': sdk.handleSum,
 162             'handleInput8': sdk.handle_input8,
 163             'handleInput16': sdk.handle_input16,
 164             'splitTimeEvenOdd': sdk.split_time_even_odd,
 165             'splitTimeOddEven': sdk.split_time_odd_even,
 166             'splitIpTimeSum': sdk.split_ip_time_sum,
 167             'splitTimeIpSum': sdk.split_time_ip_sum,
 168         }
 169         for function in functions:
 170             if re.match(r'mod\d+', function):
 171                 sdk.mod(int(function[3:]))
 172             elif re.match(r'date[ymd]{3}', function):
 173                 sdk.date(function[4:])
 174             elif re.match(r'split\d+', function):
 175                 sdk.split(int(function[5:]))
 176             elif function in other_functions:
 177                 other_functions[function]()
 178             else:
 179                 raise ExtractorError('Unknown funcion %s' % function)
 180
 181         return sdk.target
 182
 183
 184 class IqiyiIE(InfoExtractor):
 185     IE_NAME = 'iqiyi'
 186     IE_DESC = '爱奇艺'
 187
 188     _VALID_URL = r'http://(?:[^.]+\.)?iqiyi\.com/.+\.html'
 189
 190     _NETRC_MACHINE = 'iqiyi'
 191
 192     _TESTS = [{
 193         'url': 'http://www.iqiyi.com/v_19rrojlavg.html',
 194         'md5': '2cb594dc2781e6c941a110d8f358118b',
 195         'info_dict': {
 196             'id': '9c1fb1b99d192b21c559e5a1a2cb3c73',
 197             'title': '美国德州空中惊现奇异云团 酷似UFO',
 198             'ext': 'f4v',
 199         }
 200     }, {
 201         'url': 'http://www.iqiyi.com/v_19rrhnnclk.html',
 202         'info_dict': {
 203             'id': 'e3f585b550a280af23c98b6cb2be19fb',
 204             'title': '名侦探柯南第752集',
 205         },
 206         'playlist': [{
 207             'info_dict': {
 208                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part1',
 209                 'ext': 'f4v',
 210                 'title': '名侦探柯南第752集',
 211             },
 212         }, {
 213             'info_dict': {
 214                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part2',
 215                 'ext': 'f4v',
 216                 'title': '名侦探柯南第752集',
 217             },
 218         }, {
 219             'info_dict': {
 220                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part3',
 221                 'ext': 'f4v',
 222                 'title': '名侦探柯南第752集',
 223             },
 224         }, {
 225             'info_dict': {
 226                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part4',
 227                 'ext': 'f4v',
 228                 'title': '名侦探柯南第752集',
 229             },
 230         }, {
 231             'info_dict': {
 232                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part5',
 233                 'ext': 'f4v',
 234                 'title': '名侦探柯南第752集',
 235             },
 236         }, {
 237             'info_dict': {
 238                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part6',
 239                 'ext': 'f4v',
 240                 'title': '名侦探柯南第752集',
 241             },
 242         }, {
 243             'info_dict': {
 244                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part7',
 245                 'ext': 'f4v',
 246                 'title': '名侦探柯南第752集',
 247             },
 248         }, {
 249             'info_dict': {
 250                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part8',
 251                 'ext': 'f4v',
 252                 'title': '名侦探柯南第752集',
 253             },
 254         }],
 255         'params': {
 256             'skip_download': True,
 257         },
 258     }, {
 259         'url': 'http://www.iqiyi.com/w_19rt6o8t9p.html',
 260         'only_matching': True,
 261     }, {
 262         'url': 'http://www.iqiyi.com/a_19rrhbc6kt.html',
 263         'only_matching': True,
 264     }, {
 265         'url': 'http://yule.iqiyi.com/pcb.html',
 266         'only_matching': True,
 267     }, {
 268         # VIP-only video. The first 2 parts (6 minutes) are available without login
 269         # MD5 sums omitted as values are different on Travis CI and my machine
 270         'url': 'http://www.iqiyi.com/v_19rrny4w8w.html',
 271         'info_dict': {
 272             'id': 'f3cf468b39dddb30d676f89a91200dc1',
 273             'title': '泰坦尼克号',
 274         },
 275         'playlist': [{
 276             'info_dict': {
 277                 'id': 'f3cf468b39dddb30d676f89a91200dc1_part1',
 278                 'ext': 'f4v',
 279                 'title': '泰坦尼克号',
 280             },
 281         }, {
 282             'info_dict': {
 283                 'id': 'f3cf468b39dddb30d676f89a91200dc1_part2',
 284                 'ext': 'f4v',
 285                 'title': '泰坦尼克号',
 286             },
 287         }],
 288         'expected_warnings': ['Needs a VIP account for full video'],
 289     }, {
 290         'url': 'http://www.iqiyi.com/a_19rrhb8ce1.html',
 291         'info_dict': {
 292             'id': '202918101',
 293             'title': '灌篮高手 国语版',
 294         },
 295         'playlist_count': 101,
 296     }]
 297
 298     _FORMATS_MAP = [
 299         ('1', 'h6'),
 300         ('2', 'h5'),
 301         ('3', 'h4'),
 302         ('4', 'h3'),
 303         ('5', 'h2'),
 304         ('10', 'h1'),
 305     ]
 306
 307     def _real_initialize(self):
 308         self._login()
 309
 310     @staticmethod
 311     def _rsa_fun(data):
 312         # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
 313         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 314         e = 65537
 315
 316         return ohdave_rsa_encrypt(data, e, N)
 317
 318     def _login(self):
 319         (username, password) = self._get_login_info()
 320
 321         # No authentication to be performed
 322         if not username:
 323             return True
 324
 325         data = self._download_json(
 326             'http://kylin.iqiyi.com/get_token', None,
 327             note='Get token for logging', errnote='Unable to get token for logging')
 328         sdk = data['sdk']
 329         timestamp = int(time.time())
 330         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' % (
 331             username, self._rsa_fun(password.encode('utf-8')))
 332
 333         interp = IqiyiSDKInterpreter(sdk)
 334         sign = interp.run(target, data['ip'], timestamp)
 335
 336         validation_params = {
 337             'target': target,
 338             'server': 'BEA3AA1908656AABCCFF76582C4C6660',
 339             'token': data['token'],
 340             'bird_src': 'f8d91d57af224da7893dd397d52d811a',
 341             'sign': sign,
 342             'bird_t': timestamp,
 343         }
 344         validation_result = self._download_json(
 345             'http://kylin.iqiyi.com/validate?' + compat_urllib_parse.urlencode(validation_params), None,
 346             note='Validate credentials', errnote='Unable to validate credentials')
 347
 348         MSG_MAP = {
 349             'P00107': 'please login via the web interface and enter the CAPTCHA code',
 350             'P00117': 'bad username or password',
 351         }
 352
 353         code = validation_result['code']
 354         if code != 'A00000':
 355             msg = MSG_MAP.get(code)
 356             if not msg:
 357                 msg = 'error %s' % code
 358                 if validation_result.get('msg'):
 359                     msg += ': ' + validation_result['msg']
 360             self._downloader.report_warning('unable to log in: ' + msg)
 361             return False
 362
 363         return True
 364
 365     def _authenticate_vip_video(self, api_video_url, video_id, tvid, _uuid, do_report_warning):
 366         auth_params = {
 367             # version and platform hard-coded in com/qiyi/player/core/model/remote/AuthenticationRemote.as
 368             'version': '2.0',
 369             'platform': 'b6c13e26323c537d',
 370             'aid': tvid,
 371             'tvid': tvid,
 372             'uid': '',
 373             'deviceId': _uuid,
 374             'playType': 'main',  # XXX: always main?
 375             'filename': os.path.splitext(url_basename(api_video_url))[0],
 376         }
 377
 378         qd_items = compat_parse_qs(compat_urllib_parse_urlparse(api_video_url).query)
 379         for key, val in qd_items.items():
 380             auth_params[key] = val[0]
 381
 382         auth_req = sanitized_Request(
 383             'http://api.vip.iqiyi.com/services/ckn.action',
 384             urlencode_postdata(auth_params))
 385         # iQiyi server throws HTTP 405 error without the following header
 386         auth_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
 387         auth_result = self._download_json(
 388             auth_req, video_id,
 389             note='Downloading video authentication JSON',
 390             errnote='Unable to download video authentication JSON')
 391         if auth_result['code'] == 'Q00506':  # requires a VIP account
 392             if do_report_warning:
 393                 self.report_warning('Needs a VIP account for full video')
 394             return False
 395
 396         return auth_result
 397
 398     def construct_video_urls(self, data, video_id, _uuid, tvid):
 399         def do_xor(x, y):
 400             a = y % 3
 401             if a == 1:
 402                 return x ^ 121
 403             if a == 2:
 404                 return x ^ 72
 405             return x ^ 103
 406
 407         def get_encode_code(l):
 408             a = 0
 409             b = l.split('-')
 410             c = len(b)
 411             s = ''
 412             for i in range(c - 1, -1, -1):
 413                 a = do_xor(int(b[c - i - 1], 16), i)
 414                 s += chr(a)
 415             return s[::-1]
 416
 417         def get_path_key(x, format_id, segment_index):
 418             mg = ')(*&^flash@#$%a'
 419             tm = self._download_json(
 420                 'http://data.video.qiyi.com/t?tn=' + str(random.random()), video_id,
 421                 note='Download path key of segment %d for format %s' % (segment_index + 1, format_id)
 422             )['t']
 423             t = str(int(math.floor(int(tm) / (600.0))))
 424             return md5_text(t + mg + x)
 425
 426         video_urls_dict = {}
 427         need_vip_warning_report = True
 428         for format_item in data['vp']['tkl'][0]['vs']:
 429             if 0 < int(format_item['bid']) <= 10:
 430                 format_id = self.get_format(format_item['bid'])
 431             else:
 432                 continue
 433
 434             video_urls = []
 435
 436             video_urls_info = format_item['fs']
 437             if not format_item['fs'][0]['l'].startswith('/'):
 438                 t = get_encode_code(format_item['fs'][0]['l'])
 439                 if t.endswith('mp4'):
 440                     video_urls_info = format_item['flvs']
 441
 442             for segment_index, segment in enumerate(video_urls_info):
 443                 vl = segment['l']
 444                 if not vl.startswith('/'):
 445                     vl = get_encode_code(vl)
 446                 is_vip_video = '/vip/' in vl
 447                 filesize = segment['b']
 448                 base_url = data['vp']['du'].split('/')
 449                 if not is_vip_video:
 450                     key = get_path_key(
 451                         vl.split('/')[-1].split('.')[0], format_id, segment_index)
 452                     base_url.insert(-1, key)
 453                 base_url = '/'.join(base_url)
 454                 param = {
 455                     'su': _uuid,
 456                     'qyid': uuid.uuid4().hex,
 457                     'client': '',
 458                     'z': '',
 459                     'bt': '',
 460                     'ct': '',
 461                     'tn': str(int(time.time()))
 462                 }
 463                 api_video_url = base_url + vl
 464                 if is_vip_video:
 465                     api_video_url = api_video_url.replace('.f4v', '.hml')
 466                     auth_result = self._authenticate_vip_video(
 467                         api_video_url, video_id, tvid, _uuid, need_vip_warning_report)
 468                     if auth_result is False:
 469                         need_vip_warning_report = False
 470                         break
 471                     param.update({
 472                         't': auth_result['data']['t'],
 473                         # cid is hard-coded in com/qiyi/player/core/player/RuntimeData.as
 474                         'cid': 'afbe8fd3d73448c9',
 475                         'vid': video_id,
 476                         'QY00001': auth_result['data']['u'],
 477                     })
 478                 api_video_url += '?' if '?' not in api_video_url else '&'
 479                 api_video_url += compat_urllib_parse.urlencode(param)
 480                 js = self._download_json(
 481                     api_video_url, video_id,
 482                     note='Download video info of segment %d for format %s' % (segment_index + 1, format_id))
 483                 video_url = js['l']
 484                 video_urls.append(
 485                     (video_url, filesize))
 486
 487             video_urls_dict[format_id] = video_urls
 488         return video_urls_dict
 489
 490     def get_format(self, bid):
 491         matched_format_ids = [_format_id for _bid, _format_id in self._FORMATS_MAP if _bid == str(bid)]
 492         return matched_format_ids[0] if len(matched_format_ids) else None
 493
 494     def get_bid(self, format_id):
 495         matched_bids = [_bid for _bid, _format_id in self._FORMATS_MAP if _format_id == format_id]
 496         return matched_bids[0] if len(matched_bids) else None
 497
 498     def get_raw_data(self, tvid, video_id, enc_key, _uuid):
 499         tm = str(int(time.time()))
 500         tail = tm + tvid
 501         param = {
 502             'key': 'fvip',
 503             'src': md5_text('youtube-dl'),
 504             'tvId': tvid,
 505             'vid': video_id,
 506             'vinfo': 1,
 507             'tm': tm,
 508             'enc': md5_text(enc_key + tail),
 509             'qyid': _uuid,
 510             'tn': random.random(),
 511             'um': 0,
 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, swf_url, video_id):
 522         # TODO: automatic key extraction
 523         # last update at 2016-01-22 for Zombie::bite
 524         enc_key = '6ab6d0280511493ba85594779759d4ed'
 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         swf_url = self._search_regex(
 575             r'(http://[^\'"]+MainPlayer[^.]+\.swf)', webpage, 'swf player URL')
 576         _uuid = uuid.uuid4().hex
 577
 578         enc_key = self.get_enc_key(swf_url, video_id)
 579
 580         raw_data = self.get_raw_data(tvid, video_id, enc_key, _uuid)
 581
 582         if raw_data['code'] != 'A000000':
 583             raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
 584
 585         data = raw_data['data']
 586
 587         title = data['vi']['vn']
 588
 589         # generate video_urls_dict
 590         video_urls_dict = self.construct_video_urls(
 591             data, video_id, _uuid, tvid)
 592
 593         # construct info
 594         entries = []
 595         for format_id in video_urls_dict:
 596             video_urls = video_urls_dict[format_id]
 597             for i, video_url_info in enumerate(video_urls):
 598                 if len(entries) < i + 1:
 599                     entries.append({'formats': []})
 600                 entries[i]['formats'].append(
 601                     {
 602                         'url': video_url_info[0],
 603                         'filesize': video_url_info[-1],
 604                         'format_id': format_id,
 605                         'preference': int(self.get_bid(format_id))
 606                     }
 607                 )
 608
 609         for i in range(len(entries)):
 610             self._sort_formats(entries[i]['formats'])
 611             entries[i].update(
 612                 {
 613                     'id': '%s_part%d' % (video_id, i + 1),
 614                     'title': title,
 615                 }
 616             )
 617
 618         if len(entries) > 1:
 619             info = {
 620                 '_type': 'multi_video',
 621                 'id': video_id,
 622                 'title': title,
 623                 'entries': entries,
 624             }
 625         else:
 626             info = entries[0]
 627             info['id'] = video_id
 628             info['title'] = title
 629
 630         return info