Raphaël G. Git Repositories - youtubedl/blob - 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     ExtractorError,
  22     ohdave_rsa_encrypt,
  23     remove_start,
  24     sanitized_Request,
  25     urlencode_postdata,
  26     url_basename,
  27 )
  28
  29
  30 def md5_text(text):
  31     return hashlib.md5(text.encode('utf-8')).hexdigest()
  32
  33
  34 class IqiyiSDK(object):
  35     def __init__(self, target, ip, timestamp):
  36         self.target = target
  37         self.ip = ip
  38         self.timestamp = timestamp
  39
  40     @staticmethod
  41     def split_sum(data):
  42         return compat_str(sum(map(lambda p: int(p, 16), list(data))))
  43
  44     @staticmethod
  45     def digit_sum(num):
  46         if isinstance(num, int):
  47             num = compat_str(num)
  48         return compat_str(sum(map(int, num)))
  49
  50     def even_odd(self):
  51         even = self.digit_sum(compat_str(self.timestamp)[::2])
  52         odd = self.digit_sum(compat_str(self.timestamp)[1::2])
  53         return even, odd
  54
  55     def preprocess(self, chunksize):
  56         self.target = md5_text(self.target)
  57         chunks = []
  58         for i in range(32 // chunksize):
  59             chunks.append(self.target[chunksize * i:chunksize * (i + 1)])
  60         if 32 % chunksize:
  61             chunks.append(self.target[32 - 32 % chunksize:])
  62         return chunks, list(map(int, self.ip.split('.')))
  63
  64     def mod(self, modulus):
  65         chunks, ip = self.preprocess(32)
  66         self.target = chunks[0] + ''.join(map(lambda p: compat_str(p % modulus), ip))
  67
  68     def split(self, chunksize):
  69         modulus_map = {
  70             4: 256,
  71             5: 10,
  72             8: 100,
  73         }
  74
  75         chunks, ip = self.preprocess(chunksize)
  76         ret = ''
  77         for i in range(len(chunks)):
  78             ip_part = compat_str(ip[i] % modulus_map[chunksize]) if i < 4 else ''
  79             if chunksize == 8:
  80                 ret += ip_part + chunks[i]
  81             else:
  82                 ret += chunks[i] + ip_part
  83         self.target = ret
  84
  85     def handle_input16(self):
  86         self.target = md5_text(self.target)
  87         self.target = self.split_sum(self.target[:16]) + self.target + self.split_sum(self.target[16:])
  88
  89     def handle_input8(self):
  90         self.target = md5_text(self.target)
  91         ret = ''
  92         for i in range(4):
  93             part = self.target[8 * i:8 * (i + 1)]
  94             ret += self.split_sum(part) + part
  95         self.target = ret
  96
  97     def handleSum(self):
  98         self.target = md5_text(self.target)
  99         self.target = self.split_sum(self.target) + self.target
 100
 101     def date(self, scheme):
 102         self.target = md5_text(self.target)
 103         d = time.localtime(self.timestamp)
 104         strings = {
 105             'y': compat_str(d.tm_year),
 106             'm': '%02d' % d.tm_mon,
 107             'd': '%02d' % d.tm_mday,
 108         }
 109         self.target += ''.join(map(lambda c: strings[c], list(scheme)))
 110
 111     def split_time_even_odd(self):
 112         even, odd = self.even_odd()
 113         self.target = odd + md5_text(self.target) + even
 114
 115     def split_time_odd_even(self):
 116         even, odd = self.even_odd()
 117         self.target = even + md5_text(self.target) + odd
 118
 119     def split_ip_time_sum(self):
 120         chunks, ip = self.preprocess(32)
 121         self.target = compat_str(sum(ip)) + chunks[0] + self.digit_sum(self.timestamp)
 122
 123     def split_time_ip_sum(self):
 124         chunks, ip = self.preprocess(32)
 125         self.target = self.digit_sum(self.timestamp) + chunks[0] + compat_str(sum(ip))
 126
 127
 128 class IqiyiSDKInterpreter(object):
 129     BASE62_TABLE = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
 130
 131     def __init__(self, sdk_code):
 132         self.sdk_code = sdk_code
 133
 134     @classmethod
 135     def base62(cls, num):
 136         if num == 0:
 137             return '0'
 138         ret = ''
 139         while num:
 140             ret = cls.BASE62_TABLE[num % 62] + ret
 141             num = num // 62
 142         return ret
 143
 144     def decode_eval_codes(self):
 145         self.sdk_code = self.sdk_code[5:-3]
 146
 147         mobj = re.search(
 148             r"'([^']+)',62,(\d+),'([^']+)'\.split\('\|'\),[^,]+,{}",
 149             self.sdk_code)
 150         obfucasted_code, count, symbols = mobj.groups()
 151         count = int(count)
 152         symbols = symbols.split('|')
 153         symbol_table = {}
 154
 155         while count:
 156             count -= 1
 157             b62count = self.base62(count)
 158             symbol_table[b62count] = symbols[count] or b62count
 159
 160         self.sdk_code = re.sub(
 161             r'\b(\w+)\b', lambda mobj: symbol_table[mobj.group(0)],
 162             obfucasted_code)
 163
 164     def run(self, target, ip, timestamp):
 165         self.decode_eval_codes()
 166
 167         functions = re.findall(r'input=([a-zA-Z0-9]+)\(input', self.sdk_code)
 168
 169         sdk = IqiyiSDK(target, ip, timestamp)
 170
 171         other_functions = {
 172             'handleSum': sdk.handleSum,
 173             'handleInput8': sdk.handle_input8,
 174             'handleInput16': sdk.handle_input16,
 175             'splitTimeEvenOdd': sdk.split_time_even_odd,
 176             'splitTimeOddEven': sdk.split_time_odd_even,
 177             'splitIpTimeSum': sdk.split_ip_time_sum,
 178             'splitTimeIpSum': sdk.split_time_ip_sum,
 179         }
 180         for function in functions:
 181             if re.match(r'mod\d+', function):
 182                 sdk.mod(int(function[3:]))
 183             elif re.match(r'date[ymd]{3}', function):
 184                 sdk.date(function[4:])
 185             elif re.match(r'split\d+', function):
 186                 sdk.split(int(function[5:]))
 187             elif function in other_functions:
 188                 other_functions[function]()
 189             else:
 190                 raise ExtractorError('Unknown funcion %s' % function)
 191
 192         return sdk.target
 193
 194
 195 class IqiyiIE(InfoExtractor):
 196     IE_NAME = 'iqiyi'
 197     IE_DESC = '爱奇艺'
 198
 199     _VALID_URL = r'http://(?:[^.]+\.)?iqiyi\.com/.+\.html'
 200
 201     _NETRC_MACHINE = 'iqiyi'
 202
 203     _TESTS = [{
 204         'url': 'http://www.iqiyi.com/v_19rrojlavg.html',
 205         'md5': '2cb594dc2781e6c941a110d8f358118b',
 206         'info_dict': {
 207             'id': '9c1fb1b99d192b21c559e5a1a2cb3c73',
 208             'title': '美国德州空中惊现奇异云团 酷似UFO',
 209             'ext': 'f4v',
 210         }
 211     }, {
 212         'url': 'http://www.iqiyi.com/v_19rrhnnclk.html',
 213         'info_dict': {
 214             'id': 'e3f585b550a280af23c98b6cb2be19fb',
 215             'title': '名侦探柯南第752集',
 216         },
 217         'playlist': [{
 218             'info_dict': {
 219                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part1',
 220                 'ext': 'f4v',
 221                 'title': '名侦探柯南第752集',
 222             },
 223         }, {
 224             'info_dict': {
 225                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part2',
 226                 'ext': 'f4v',
 227                 'title': '名侦探柯南第752集',
 228             },
 229         }, {
 230             'info_dict': {
 231                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part3',
 232                 'ext': 'f4v',
 233                 'title': '名侦探柯南第752集',
 234             },
 235         }, {
 236             'info_dict': {
 237                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part4',
 238                 'ext': 'f4v',
 239                 'title': '名侦探柯南第752集',
 240             },
 241         }, {
 242             'info_dict': {
 243                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part5',
 244                 'ext': 'f4v',
 245                 'title': '名侦探柯南第752集',
 246             },
 247         }, {
 248             'info_dict': {
 249                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part6',
 250                 'ext': 'f4v',
 251                 'title': '名侦探柯南第752集',
 252             },
 253         }, {
 254             'info_dict': {
 255                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part7',
 256                 'ext': 'f4v',
 257                 'title': '名侦探柯南第752集',
 258             },
 259         }, {
 260             'info_dict': {
 261                 'id': 'e3f585b550a280af23c98b6cb2be19fb_part8',
 262                 'ext': 'f4v',
 263                 'title': '名侦探柯南第752集',
 264             },
 265         }],
 266         'params': {
 267             'skip_download': True,
 268         },
 269     }, {
 270         'url': 'http://www.iqiyi.com/w_19rt6o8t9p.html',
 271         'only_matching': True,
 272     }, {
 273         'url': 'http://www.iqiyi.com/a_19rrhbc6kt.html',
 274         'only_matching': True,
 275     }, {
 276         'url': 'http://yule.iqiyi.com/pcb.html',
 277         'only_matching': True,
 278     }, {
 279         # VIP-only video. The first 2 parts (6 minutes) are available without login
 280         # MD5 sums omitted as values are different on Travis CI and my machine
 281         'url': 'http://www.iqiyi.com/v_19rrny4w8w.html',
 282         'info_dict': {
 283             'id': 'f3cf468b39dddb30d676f89a91200dc1',
 284             'title': '泰坦尼克号',
 285         },
 286         'playlist': [{
 287             'info_dict': {
 288                 'id': 'f3cf468b39dddb30d676f89a91200dc1_part1',
 289                 'ext': 'f4v',
 290                 'title': '泰坦尼克号',
 291             },
 292         }, {
 293             'info_dict': {
 294                 'id': 'f3cf468b39dddb30d676f89a91200dc1_part2',
 295                 'ext': 'f4v',
 296                 'title': '泰坦尼克号',
 297             },
 298         }],
 299         'expected_warnings': ['Needs a VIP account for full video'],
 300     }, {
 301         'url': 'http://www.iqiyi.com/a_19rrhb8ce1.html',
 302         'info_dict': {
 303             'id': '202918101',
 304             'title': '灌篮高手 国语版',
 305         },
 306         'playlist_count': 101,
 307     }]
 308
 309     _FORMATS_MAP = [
 310         ('1', 'h6'),
 311         ('2', 'h5'),
 312         ('3', 'h4'),
 313         ('4', 'h3'),
 314         ('5', 'h2'),
 315         ('10', 'h1'),
 316     ]
 317
 318     def _real_initialize(self):
 319         self._login()
 320
 321     @staticmethod
 322     def _rsa_fun(data):
 323         # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
 324         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 325         e = 65537
 326
 327         return ohdave_rsa_encrypt(data, e, N)
 328
 329     def _login(self):
 330         (username, password) = self._get_login_info()
 331
 332         # No authentication to be performed
 333         if not username:
 334             return True
 335
 336         data = self._download_json(
 337             'http://kylin.iqiyi.com/get_token', None,
 338             note='Get token for logging', errnote='Unable to get token for logging')
 339         sdk = data['sdk']
 340         timestamp = int(time.time())
 341         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' % (
 342             username, self._rsa_fun(password.encode('utf-8')))
 343
 344         interp = IqiyiSDKInterpreter(sdk)
 345         sign = interp.run(target, data['ip'], timestamp)
 346
 347         validation_params = {
 348             'target': target,
 349             'server': 'BEA3AA1908656AABCCFF76582C4C6660',
 350             'token': data['token'],
 351             'bird_src': 'f8d91d57af224da7893dd397d52d811a',
 352             'sign': sign,
 353             'bird_t': timestamp,
 354         }
 355         validation_result = self._download_json(
 356             'http://kylin.iqiyi.com/validate?' + compat_urllib_parse.urlencode(validation_params), None,
 357             note='Validate credentials', errnote='Unable to validate credentials')
 358
 359         MSG_MAP = {
 360             'P00107': 'please login via the web interface and enter the CAPTCHA code',
 361             'P00117': 'bad username or password',
 362         }
 363
 364         code = validation_result['code']
 365         if code != 'A00000':
 366             msg = MSG_MAP.get(code)
 367             if not msg:
 368                 msg = 'error %s' % code
 369                 if validation_result.get('msg'):
 370                     msg += ': ' + validation_result['msg']
 371             self._downloader.report_warning('unable to log in: ' + msg)
 372             return False
 373
 374         return True
 375
 376     def _authenticate_vip_video(self, api_video_url, video_id, tvid, _uuid, do_report_warning):
 377         auth_params = {
 378             # version and platform hard-coded in com/qiyi/player/core/model/remote/AuthenticationRemote.as
 379             'version': '2.0',
 380             'platform': 'b6c13e26323c537d',
 381             'aid': tvid,
 382             'tvid': tvid,
 383             'uid': '',
 384             'deviceId': _uuid,
 385             'playType': 'main',  # XXX: always main?
 386             'filename': os.path.splitext(url_basename(api_video_url))[0],
 387         }
 388
 389         qd_items = compat_parse_qs(compat_urllib_parse_urlparse(api_video_url).query)
 390         for key, val in qd_items.items():
 391             auth_params[key] = val[0]
 392
 393         auth_req = sanitized_Request(
 394             'http://api.vip.iqiyi.com/services/ckn.action',
 395             urlencode_postdata(auth_params))
 396         # iQiyi server throws HTTP 405 error without the following header
 397         auth_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
 398         auth_result = self._download_json(
 399             auth_req, video_id,
 400             note='Downloading video authentication JSON',
 401             errnote='Unable to download video authentication JSON')
 402         if auth_result['code'] == 'Q00506':  # requires a VIP account
 403             if do_report_warning:
 404                 self.report_warning('Needs a VIP account for full video')
 405             return False
 406
 407         return auth_result
 408
 409     def construct_video_urls(self, data, video_id, _uuid, tvid):
 410         def do_xor(x, y):
 411             a = y % 3
 412             if a == 1:
 413                 return x ^ 121
 414             if a == 2:
 415                 return x ^ 72
 416             return x ^ 103
 417
 418         def get_encode_code(l):
 419             a = 0
 420             b = l.split('-')
 421             c = len(b)
 422             s = ''
 423             for i in range(c - 1, -1, -1):
 424                 a = do_xor(int(b[c - i - 1], 16), i)
 425                 s += chr(a)
 426             return s[::-1]
 427
 428         def get_path_key(x, format_id, segment_index):
 429             mg = ')(*&^flash@#$%a'
 430             tm = self._download_json(
 431                 'http://data.video.qiyi.com/t?tn=' + str(random.random()), video_id,
 432                 note='Download path key of segment %d for format %s' % (segment_index + 1, format_id)
 433             )['t']
 434             t = str(int(math.floor(int(tm) / (600.0))))
 435             return md5_text(t + mg + x)
 436
 437         video_urls_dict = {}
 438         need_vip_warning_report = True
 439         for format_item in data['vp']['tkl'][0]['vs']:
 440             if 0 < int(format_item['bid']) <= 10:
 441                 format_id = self.get_format(format_item['bid'])
 442             else:
 443                 continue
 444
 445             video_urls = []
 446
 447             video_urls_info = format_item['fs']
 448             if not format_item['fs'][0]['l'].startswith('/'):
 449                 t = get_encode_code(format_item['fs'][0]['l'])
 450                 if t.endswith('mp4'):
 451                     video_urls_info = format_item['flvs']
 452
 453             for segment_index, segment in enumerate(video_urls_info):
 454                 vl = segment['l']
 455                 if not vl.startswith('/'):
 456                     vl = get_encode_code(vl)
 457                 is_vip_video = '/vip/' in vl
 458                 filesize = segment['b']
 459                 base_url = data['vp']['du'].split('/')
 460                 if not is_vip_video:
 461                     key = get_path_key(
 462                         vl.split('/')[-1].split('.')[0], format_id, segment_index)
 463                     base_url.insert(-1, key)
 464                 base_url = '/'.join(base_url)
 465                 param = {
 466                     'su': _uuid,
 467                     'qyid': uuid.uuid4().hex,
 468                     'client': '',
 469                     'z': '',
 470                     'bt': '',
 471                     'ct': '',
 472                     'tn': str(int(time.time()))
 473                 }
 474                 api_video_url = base_url + vl
 475                 if is_vip_video:
 476                     api_video_url = api_video_url.replace('.f4v', '.hml')
 477                     auth_result = self._authenticate_vip_video(
 478                         api_video_url, video_id, tvid, _uuid, need_vip_warning_report)
 479                     if auth_result is False:
 480                         need_vip_warning_report = False
 481                         break
 482                     param.update({
 483                         't': auth_result['data']['t'],
 484                         # cid is hard-coded in com/qiyi/player/core/player/RuntimeData.as
 485                         'cid': 'afbe8fd3d73448c9',
 486                         'vid': video_id,
 487                         'QY00001': auth_result['data']['u'],
 488                     })
 489                 api_video_url += '?' if '?' not in api_video_url else '&'
 490                 api_video_url += compat_urllib_parse.urlencode(param)
 491                 js = self._download_json(
 492                     api_video_url, video_id,
 493                     note='Download video info of segment %d for format %s' % (segment_index + 1, format_id))
 494                 video_url = js['l']
 495                 video_urls.append(
 496                     (video_url, filesize))
 497
 498             video_urls_dict[format_id] = video_urls
 499         return video_urls_dict
 500
 501     def get_format(self, bid):
 502         matched_format_ids = [_format_id for _bid, _format_id in self._FORMATS_MAP if _bid == str(bid)]
 503         return matched_format_ids[0] if len(matched_format_ids) else None
 504
 505     def get_bid(self, format_id):
 506         matched_bids = [_bid for _bid, _format_id in self._FORMATS_MAP if _format_id == format_id]
 507         return matched_bids[0] if len(matched_bids) else None
 508
 509     def get_raw_data(self, tvid, video_id, enc_key, _uuid):
 510         tm = str(int(time.time()))
 511         tail = tm + tvid
 512         param = {
 513             'key': 'fvip',
 514             'src': md5_text('youtube-dl'),
 515             'tvId': tvid,
 516             'vid': video_id,
 517             'vinfo': 1,
 518             'tm': tm,
 519             'enc': md5_text(enc_key + tail),
 520             'qyid': _uuid,
 521             'tn': random.random(),
 522             'um': 0,
 523             'authkey': md5_text(md5_text('') + tail),
 524             'k_tag': 1,
 525         }
 526
 527         api_url = 'http://cache.video.qiyi.com/vms' + '?' + \
 528             compat_urllib_parse.urlencode(param)
 529         raw_data = self._download_json(api_url, video_id)
 530         return raw_data
 531
 532     def get_enc_key(self, swf_url, video_id):
 533         # TODO: automatic key extraction
 534         # last update at 2016-01-22 for Zombie::bite
 535         enc_key = '6ab6d0280511493ba85594779759d4ed'
 536         return enc_key
 537
 538     def _extract_playlist(self, webpage):
 539         PAGE_SIZE = 50
 540
 541         links = re.findall(
 542             r'<a[^>]+class="site-piclist_pic_link"[^>]+href="(http://www\.iqiyi\.com/.+\.html)"',
 543             webpage)
 544         if not links:
 545             return
 546
 547         album_id = self._search_regex(
 548             r'albumId\s*:\s*(\d+),', webpage, 'album ID')
 549         album_title = self._search_regex(
 550             r'data-share-title="([^"]+)"', webpage, 'album title', fatal=False)
 551
 552         entries = list(map(self.url_result, links))
 553
 554         # Start from 2 because links in the first page are already on webpage
 555         for page_num in itertools.count(2):
 556             pagelist_page = self._download_webpage(
 557                 'http://cache.video.qiyi.com/jp/avlist/%s/%d/%d/' % (album_id, page_num, PAGE_SIZE),
 558                 album_id,
 559                 note='Download playlist page %d' % page_num,
 560                 errnote='Failed to download playlist page %d' % page_num)
 561             pagelist = self._parse_json(
 562                 remove_start(pagelist_page, 'var tvInfoJs='), album_id)
 563             vlist = pagelist['data']['vlist']
 564             for item in vlist:
 565                 entries.append(self.url_result(item['vurl']))
 566             if len(vlist) < PAGE_SIZE:
 567                 break
 568
 569         return self.playlist_result(entries, album_id, album_title)
 570
 571     def _real_extract(self, url):
 572         webpage = self._download_webpage(
 573             url, 'temp_id', note='download video page')
 574
 575         # There's no simple way to determine whether an URL is a playlist or not
 576         # So detect it
 577         playlist_result = self._extract_playlist(webpage)
 578         if playlist_result:
 579             return playlist_result
 580
 581         tvid = self._search_regex(
 582             r'data-player-tvid\s*=\s*[\'"](\d+)', webpage, 'tvid')
 583         video_id = self._search_regex(
 584             r'data-player-videoid\s*=\s*[\'"]([a-f\d]+)', webpage, 'video_id')
 585         swf_url = self._search_regex(
 586             r'(http://[^\'"]+MainPlayer[^.]+\.swf)', webpage, 'swf player URL')
 587         _uuid = uuid.uuid4().hex
 588
 589         enc_key = self.get_enc_key(swf_url, video_id)
 590
 591         raw_data = self.get_raw_data(tvid, video_id, enc_key, _uuid)
 592
 593         if raw_data['code'] != 'A000000':
 594             raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
 595
 596         data = raw_data['data']
 597
 598         title = data['vi']['vn']
 599
 600         # generate video_urls_dict
 601         video_urls_dict = self.construct_video_urls(
 602             data, video_id, _uuid, tvid)
 603
 604         # construct info
 605         entries = []
 606         for format_id in video_urls_dict:
 607             video_urls = video_urls_dict[format_id]
 608             for i, video_url_info in enumerate(video_urls):
 609                 if len(entries) < i + 1:
 610                     entries.append({'formats': []})
 611                 entries[i]['formats'].append(
 612                     {
 613                         'url': video_url_info[0],
 614                         'filesize': video_url_info[-1],
 615                         'format_id': format_id,
 616                         'preference': int(self.get_bid(format_id))
 617                     }
 618                 )
 619
 620         for i in range(len(entries)):
 621             self._sort_formats(entries[i]['formats'])
 622             entries[i].update(
 623                 {
 624                     'id': '%s_part%d' % (video_id, i + 1),
 625                     'title': title,
 626                 }
 627             )
 628
 629         if len(entries) > 1:
 630             info = {
 631                 '_type': 'multi_video',
 632                 'id': video_id,
 633                 'title': title,
 634                 'entries': entries,
 635             }
 636         else:
 637             info = entries[0]
 638             info['id'] = video_id
 639             info['title'] = title
 640
 641         return info