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     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'http://(?:[^.]+\.)?iqiyi\.com/.+\.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
 278     _FORMATS_MAP = [
 279         ('1', 'h6'),
 280         ('2', 'h5'),
 281         ('3', 'h4'),
 282         ('4', 'h3'),
 283         ('5', 'h2'),
 284         ('10', 'h1'),
 285     ]
 286
 287     def _real_initialize(self):
 288         self._login()
 289
 290     @staticmethod
 291     def _rsa_fun(data):
 292         # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
 293         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 294         e = 65537
 295
 296         return ohdave_rsa_encrypt(data, e, N)
 297
 298     def _login(self):
 299         (username, password) = self._get_login_info()
 300
 301         # No authentication to be performed
 302         if not username:
 303             return True
 304
 305         data = self._download_json(
 306             'http://kylin.iqiyi.com/get_token', None,
 307             note='Get token for logging', errnote='Unable to get token for logging')
 308         sdk = data['sdk']
 309         timestamp = int(time.time())
 310         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' % (
 311             username, self._rsa_fun(password.encode('utf-8')))
 312
 313         interp = IqiyiSDKInterpreter(sdk)
 314         sign = interp.run(target, data['ip'], timestamp)
 315
 316         validation_params = {
 317             'target': target,
 318             'server': 'BEA3AA1908656AABCCFF76582C4C6660',
 319             'token': data['token'],
 320             'bird_src': 'f8d91d57af224da7893dd397d52d811a',
 321             'sign': sign,
 322             'bird_t': timestamp,
 323         }
 324         validation_result = self._download_json(
 325             'http://kylin.iqiyi.com/validate?' + compat_urllib_parse.urlencode(validation_params), None,
 326             note='Validate credentials', errnote='Unable to validate credentials')
 327
 328         MSG_MAP = {
 329             'P00107': 'please login via the web interface and enter the CAPTCHA code',
 330             'P00117': 'bad username or password',
 331         }
 332
 333         code = validation_result['code']
 334         if code != 'A00000':
 335             msg = MSG_MAP.get(code)
 336             if not msg:
 337                 msg = 'error %s' % code
 338                 if validation_result.get('msg'):
 339                     msg += ': ' + validation_result['msg']
 340             self._downloader.report_warning('unable to log in: ' + msg)
 341             return False
 342
 343         return True
 344
 345     def _authenticate_vip_video(self, api_video_url, video_id, tvid, _uuid, do_report_warning):
 346         auth_params = {
 347             # version and platform hard-coded in com/qiyi/player/core/model/remote/AuthenticationRemote.as
 348             'version': '2.0',
 349             'platform': 'b6c13e26323c537d',
 350             'aid': tvid,
 351             'tvid': tvid,
 352             'uid': '',
 353             'deviceId': _uuid,
 354             'playType': 'main',  # XXX: always main?
 355             'filename': os.path.splitext(url_basename(api_video_url))[0],
 356         }
 357
 358         qd_items = compat_parse_qs(compat_urllib_parse_urlparse(api_video_url).query)
 359         for key, val in qd_items.items():
 360             auth_params[key] = val[0]
 361
 362         auth_req = sanitized_Request(
 363             'http://api.vip.iqiyi.com/services/ckn.action',
 364             urlencode_postdata(auth_params))
 365         # iQiyi server throws HTTP 405 error without the following header
 366         auth_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
 367         auth_result = self._download_json(
 368             auth_req, video_id,
 369             note='Downloading video authentication JSON',
 370             errnote='Unable to download video authentication JSON')
 371         if auth_result['code'] == 'Q00506':  # requires a VIP account
 372             if do_report_warning:
 373                 self.report_warning('Needs a VIP account for full video')
 374             return False
 375
 376         return auth_result
 377
 378     def construct_video_urls(self, data, video_id, _uuid, tvid):
 379         def do_xor(x, y):
 380             a = y % 3
 381             if a == 1:
 382                 return x ^ 121
 383             if a == 2:
 384                 return x ^ 72
 385             return x ^ 103
 386
 387         def get_encode_code(l):
 388             a = 0
 389             b = l.split('-')
 390             c = len(b)
 391             s = ''
 392             for i in range(c - 1, -1, -1):
 393                 a = do_xor(int(b[c - i - 1], 16), i)
 394                 s += chr(a)
 395             return s[::-1]
 396
 397         def get_path_key(x, format_id, segment_index):
 398             mg = ')(*&^flash@#$%a'
 399             tm = self._download_json(
 400                 'http://data.video.qiyi.com/t?tn=' + str(random.random()), video_id,
 401                 note='Download path key of segment %d for format %s' % (segment_index + 1, format_id)
 402             )['t']
 403             t = str(int(math.floor(int(tm) / (600.0))))
 404             return md5_text(t + mg + x)
 405
 406         video_urls_dict = {}
 407         need_vip_warning_report = True
 408         for format_item in data['vp']['tkl'][0]['vs']:
 409             if 0 < int(format_item['bid']) <= 10:
 410                 format_id = self.get_format(format_item['bid'])
 411             else:
 412                 continue
 413
 414             video_urls = []
 415
 416             video_urls_info = format_item['fs']
 417             if not format_item['fs'][0]['l'].startswith('/'):
 418                 t = get_encode_code(format_item['fs'][0]['l'])
 419                 if t.endswith('mp4'):
 420                     video_urls_info = format_item['flvs']
 421
 422             for segment_index, segment in enumerate(video_urls_info):
 423                 vl = segment['l']
 424                 if not vl.startswith('/'):
 425                     vl = get_encode_code(vl)
 426                 is_vip_video = '/vip/' in vl
 427                 filesize = segment['b']
 428                 base_url = data['vp']['du'].split('/')
 429                 if not is_vip_video:
 430                     key = get_path_key(
 431                         vl.split('/')[-1].split('.')[0], format_id, segment_index)
 432                     base_url.insert(-1, key)
 433                 base_url = '/'.join(base_url)
 434                 param = {
 435                     'su': _uuid,
 436                     'qyid': uuid.uuid4().hex,
 437                     'client': '',
 438                     'z': '',
 439                     'bt': '',
 440                     'ct': '',
 441                     'tn': str(int(time.time()))
 442                 }
 443                 api_video_url = base_url + vl
 444                 if is_vip_video:
 445                     api_video_url = api_video_url.replace('.f4v', '.hml')
 446                     auth_result = self._authenticate_vip_video(
 447                         api_video_url, video_id, tvid, _uuid, need_vip_warning_report)
 448                     if auth_result is False:
 449                         need_vip_warning_report = False
 450                         break
 451                     param.update({
 452                         't': auth_result['data']['t'],
 453                         # cid is hard-coded in com/qiyi/player/core/player/RuntimeData.as
 454                         'cid': 'afbe8fd3d73448c9',
 455                         'vid': video_id,
 456                         'QY00001': auth_result['data']['u'],
 457                     })
 458                 api_video_url += '?' if '?' not in api_video_url else '&'
 459                 api_video_url += compat_urllib_parse.urlencode(param)
 460                 js = self._download_json(
 461                     api_video_url, video_id,
 462                     note='Download video info of segment %d for format %s' % (segment_index + 1, format_id))
 463                 video_url = js['l']
 464                 video_urls.append(
 465                     (video_url, filesize))
 466
 467             video_urls_dict[format_id] = video_urls
 468         return video_urls_dict
 469
 470     def get_format(self, bid):
 471         matched_format_ids = [_format_id for _bid, _format_id in self._FORMATS_MAP if _bid == str(bid)]
 472         return matched_format_ids[0] if len(matched_format_ids) else None
 473
 474     def get_bid(self, format_id):
 475         matched_bids = [_bid for _bid, _format_id in self._FORMATS_MAP if _format_id == format_id]
 476         return matched_bids[0] if len(matched_bids) else None
 477
 478     def get_raw_data(self, tvid, video_id, enc_key, _uuid):
 479         tm = str(int(time.time()))
 480         tail = tm + tvid
 481         param = {
 482             'key': 'fvip',
 483             'src': md5_text('youtube-dl'),
 484             'tvId': tvid,
 485             'vid': video_id,
 486             'vinfo': 1,
 487             'tm': tm,
 488             'enc': md5_text(enc_key + tail),
 489             'qyid': _uuid,
 490             'tn': random.random(),
 491             'um': 0,
 492             'authkey': md5_text(md5_text('') + tail),
 493             'k_tag': 1,
 494         }
 495
 496         api_url = 'http://cache.video.qiyi.com/vms' + '?' + \
 497             compat_urllib_parse.urlencode(param)
 498         raw_data = self._download_json(api_url, video_id)
 499         return raw_data
 500
 501     def get_enc_key(self, video_id):
 502         # TODO: automatic key extraction
 503         # last update at 2016-01-22 for Zombie::bite
 504         enc_key = '6ab6d0280511493ba85594779759d4ed'
 505         return enc_key
 506
 507     def _extract_playlist(self, webpage):
 508         PAGE_SIZE = 50
 509
 510         links = re.findall(
 511             r'<a[^>]+class="site-piclist_pic_link"[^>]+href="(http://www\.iqiyi\.com/.+\.html)"',
 512             webpage)
 513         if not links:
 514             return
 515
 516         album_id = self._search_regex(
 517             r'albumId\s*:\s*(\d+),', webpage, 'album ID')
 518         album_title = self._search_regex(
 519             r'data-share-title="([^"]+)"', webpage, 'album title', fatal=False)
 520
 521         entries = list(map(self.url_result, links))
 522
 523         # Start from 2 because links in the first page are already on webpage
 524         for page_num in itertools.count(2):
 525             pagelist_page = self._download_webpage(
 526                 'http://cache.video.qiyi.com/jp/avlist/%s/%d/%d/' % (album_id, page_num, PAGE_SIZE),
 527                 album_id,
 528                 note='Download playlist page %d' % page_num,
 529                 errnote='Failed to download playlist page %d' % page_num)
 530             pagelist = self._parse_json(
 531                 remove_start(pagelist_page, 'var tvInfoJs='), album_id)
 532             vlist = pagelist['data']['vlist']
 533             for item in vlist:
 534                 entries.append(self.url_result(item['vurl']))
 535             if len(vlist) < PAGE_SIZE:
 536                 break
 537
 538         return self.playlist_result(entries, album_id, album_title)
 539
 540     def _real_extract(self, url):
 541         webpage = self._download_webpage(
 542             url, 'temp_id', note='download video page')
 543
 544         # There's no simple way to determine whether an URL is a playlist or not
 545         # So detect it
 546         playlist_result = self._extract_playlist(webpage)
 547         if playlist_result:
 548             return playlist_result
 549
 550         tvid = self._search_regex(
 551             r'data-player-tvid\s*=\s*[\'"](\d+)', webpage, 'tvid')
 552         video_id = self._search_regex(
 553             r'data-player-videoid\s*=\s*[\'"]([a-f\d]+)', webpage, 'video_id')
 554         _uuid = uuid.uuid4().hex
 555
 556         enc_key = self.get_enc_key(video_id)
 557
 558         raw_data = self.get_raw_data(tvid, video_id, enc_key, _uuid)
 559
 560         if raw_data['code'] != 'A000000':
 561             raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
 562
 563         data = raw_data['data']
 564
 565         title = data['vi']['vn']
 566
 567         # generate video_urls_dict
 568         video_urls_dict = self.construct_video_urls(
 569             data, video_id, _uuid, tvid)
 570
 571         # construct info
 572         entries = []
 573         for format_id in video_urls_dict:
 574             video_urls = video_urls_dict[format_id]
 575             for i, video_url_info in enumerate(video_urls):
 576                 if len(entries) < i + 1:
 577                     entries.append({'formats': []})
 578                 entries[i]['formats'].append(
 579                     {
 580                         'url': video_url_info[0],
 581                         'filesize': video_url_info[-1],
 582                         'format_id': format_id,
 583                         'preference': int(self.get_bid(format_id))
 584                     }
 585                 )
 586
 587         for i in range(len(entries)):
 588             self._sort_formats(entries[i]['formats'])
 589             entries[i].update(
 590                 {
 591                     'id': '%s_part%d' % (video_id, i + 1),
 592                     'title': title,
 593                 }
 594             )
 595
 596         if len(entries) > 1:
 597             info = {
 598                 '_type': 'multi_video',
 599                 'id': video_id,
 600                 'title': title,
 601                 'entries': entries,
 602             }
 603         else:
 604             info = entries[0]
 605             info['id'] = video_id
 606             info['title'] = title
 607
 608         return info