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 re
   7 import time
   8
   9 from .common import InfoExtractor
  10 from ..compat import (
  11     compat_str,
  12     compat_urllib_parse_urlencode,
  13 )
  14 from ..utils import (
  15     clean_html,
  16     decode_packed_codes,
  17     get_element_by_id,
  18     get_element_by_attribute,
  19     ExtractorError,
  20     ohdave_rsa_encrypt,
  21     remove_start,
  22 )
  23
  24
  25 def md5_text(text):
  26     return hashlib.md5(text.encode('utf-8')).hexdigest()
  27
  28
  29 class IqiyiSDK(object):
  30     def __init__(self, target, ip, timestamp):
  31         self.target = target
  32         self.ip = ip
  33         self.timestamp = timestamp
  34
  35     @staticmethod
  36     def split_sum(data):
  37         return compat_str(sum(map(lambda p: int(p, 16), list(data))))
  38
  39     @staticmethod
  40     def digit_sum(num):
  41         if isinstance(num, int):
  42             num = compat_str(num)
  43         return compat_str(sum(map(int, num)))
  44
  45     def even_odd(self):
  46         even = self.digit_sum(compat_str(self.timestamp)[::2])
  47         odd = self.digit_sum(compat_str(self.timestamp)[1::2])
  48         return even, odd
  49
  50     def preprocess(self, chunksize):
  51         self.target = md5_text(self.target)
  52         chunks = []
  53         for i in range(32 // chunksize):
  54             chunks.append(self.target[chunksize * i:chunksize * (i + 1)])
  55         if 32 % chunksize:
  56             chunks.append(self.target[32 - 32 % chunksize:])
  57         return chunks, list(map(int, self.ip.split('.')))
  58
  59     def mod(self, modulus):
  60         chunks, ip = self.preprocess(32)
  61         self.target = chunks[0] + ''.join(map(lambda p: compat_str(p % modulus), ip))
  62
  63     def split(self, chunksize):
  64         modulus_map = {
  65             4: 256,
  66             5: 10,
  67             8: 100,
  68         }
  69
  70         chunks, ip = self.preprocess(chunksize)
  71         ret = ''
  72         for i in range(len(chunks)):
  73             ip_part = compat_str(ip[i] % modulus_map[chunksize]) if i < 4 else ''
  74             if chunksize == 8:
  75                 ret += ip_part + chunks[i]
  76             else:
  77                 ret += chunks[i] + ip_part
  78         self.target = ret
  79
  80     def handle_input16(self):
  81         self.target = md5_text(self.target)
  82         self.target = self.split_sum(self.target[:16]) + self.target + self.split_sum(self.target[16:])
  83
  84     def handle_input8(self):
  85         self.target = md5_text(self.target)
  86         ret = ''
  87         for i in range(4):
  88             part = self.target[8 * i:8 * (i + 1)]
  89             ret += self.split_sum(part) + part
  90         self.target = ret
  91
  92     def handleSum(self):
  93         self.target = md5_text(self.target)
  94         self.target = self.split_sum(self.target) + self.target
  95
  96     def date(self, scheme):
  97         self.target = md5_text(self.target)
  98         d = time.localtime(self.timestamp)
  99         strings = {
 100             'y': compat_str(d.tm_year),
 101             'm': '%02d' % d.tm_mon,
 102             'd': '%02d' % d.tm_mday,
 103         }
 104         self.target += ''.join(map(lambda c: strings[c], list(scheme)))
 105
 106     def split_time_even_odd(self):
 107         even, odd = self.even_odd()
 108         self.target = odd + md5_text(self.target) + even
 109
 110     def split_time_odd_even(self):
 111         even, odd = self.even_odd()
 112         self.target = even + md5_text(self.target) + odd
 113
 114     def split_ip_time_sum(self):
 115         chunks, ip = self.preprocess(32)
 116         self.target = compat_str(sum(ip)) + chunks[0] + self.digit_sum(self.timestamp)
 117
 118     def split_time_ip_sum(self):
 119         chunks, ip = self.preprocess(32)
 120         self.target = self.digit_sum(self.timestamp) + chunks[0] + compat_str(sum(ip))
 121
 122
 123 class IqiyiSDKInterpreter(object):
 124     def __init__(self, sdk_code):
 125         self.sdk_code = sdk_code
 126
 127     def run(self, target, ip, timestamp):
 128         self.sdk_code = decode_packed_codes(self.sdk_code)
 129
 130         functions = re.findall(r'input=([a-zA-Z0-9]+)\(input', self.sdk_code)
 131
 132         sdk = IqiyiSDK(target, ip, timestamp)
 133
 134         other_functions = {
 135             'handleSum': sdk.handleSum,
 136             'handleInput8': sdk.handle_input8,
 137             'handleInput16': sdk.handle_input16,
 138             'splitTimeEvenOdd': sdk.split_time_even_odd,
 139             'splitTimeOddEven': sdk.split_time_odd_even,
 140             'splitIpTimeSum': sdk.split_ip_time_sum,
 141             'splitTimeIpSum': sdk.split_time_ip_sum,
 142         }
 143         for function in functions:
 144             if re.match(r'mod\d+', function):
 145                 sdk.mod(int(function[3:]))
 146             elif re.match(r'date[ymd]{3}', function):
 147                 sdk.date(function[4:])
 148             elif re.match(r'split\d+', function):
 149                 sdk.split(int(function[5:]))
 150             elif function in other_functions:
 151                 other_functions[function]()
 152             else:
 153                 raise ExtractorError('Unknown funcion %s' % function)
 154
 155         return sdk.target
 156
 157
 158 class IqiyiIE(InfoExtractor):
 159     IE_NAME = 'iqiyi'
 160     IE_DESC = '爱奇艺'
 161
 162     _VALID_URL = r'https?://(?:(?:[^.]+\.)?iqiyi\.com|www\.pps\.tv)/.+\.html'
 163
 164     _NETRC_MACHINE = 'iqiyi'
 165
 166     _TESTS = [{
 167         'url': 'http://www.iqiyi.com/v_19rrojlavg.html',
 168         # MD5 checksum differs on my machine and Travis CI
 169         'info_dict': {
 170             'id': '9c1fb1b99d192b21c559e5a1a2cb3c73',
 171             'ext': 'mp4',
 172             'title': '美国德州空中惊现奇异云团 酷似UFO',
 173         }
 174     }, {
 175         'url': 'http://www.iqiyi.com/v_19rrhnnclk.html',
 176         'md5': 'b7dc800a4004b1b57749d9abae0472da',
 177         'info_dict': {
 178             'id': 'e3f585b550a280af23c98b6cb2be19fb',
 179             'ext': 'mp4',
 180             # This can be either Simplified Chinese or Traditional Chinese
 181             'title': r're:^(?:名侦探柯南 国语版：第752集 迫近灰原秘密的黑影 下篇|名偵探柯南 國語版：第752集 迫近灰原秘密的黑影 下篇)$',
 182         },
 183         'skip': 'Geo-restricted to China',
 184     }, {
 185         'url': 'http://www.iqiyi.com/w_19rt6o8t9p.html',
 186         'only_matching': True,
 187     }, {
 188         'url': 'http://www.iqiyi.com/a_19rrhbc6kt.html',
 189         'only_matching': True,
 190     }, {
 191         'url': 'http://yule.iqiyi.com/pcb.html',
 192         'only_matching': True,
 193     }, {
 194         # VIP-only video. The first 2 parts (6 minutes) are available without login
 195         # MD5 sums omitted as values are different on Travis CI and my machine
 196         'url': 'http://www.iqiyi.com/v_19rrny4w8w.html',
 197         'info_dict': {
 198             'id': 'f3cf468b39dddb30d676f89a91200dc1',
 199             'ext': 'mp4',
 200             'title': '泰坦尼克号',
 201         },
 202         'skip': 'Geo-restricted to China',
 203     }, {
 204         'url': 'http://www.iqiyi.com/a_19rrhb8ce1.html',
 205         'info_dict': {
 206             'id': '202918101',
 207             'title': '灌篮高手 国语版',
 208         },
 209         'playlist_count': 101,
 210     }, {
 211         'url': 'http://www.pps.tv/w_19rrbav0ph.html',
 212         'only_matching': True,
 213     }]
 214
 215     _FORMATS_MAP = {
 216         '96': 1,    # 216p, 240p
 217         '1': 2,     # 336p, 360p
 218         '2': 3,     # 480p, 504p
 219         '21': 4,    # 504p
 220         '4': 5,     # 720p
 221         '17': 5,    # 720p
 222         '5': 6,     # 1072p, 1080p
 223         '18': 7,    # 1080p
 224     }
 225
 226     def _real_initialize(self):
 227         self._login()
 228
 229     @staticmethod
 230     def _rsa_fun(data):
 231         # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
 232         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 233         e = 65537
 234
 235         return ohdave_rsa_encrypt(data, e, N)
 236
 237     def _login(self):
 238         (username, password) = self._get_login_info()
 239
 240         # No authentication to be performed
 241         if not username:
 242             return True
 243
 244         data = self._download_json(
 245             'http://kylin.iqiyi.com/get_token', None,
 246             note='Get token for logging', errnote='Unable to get token for logging')
 247         sdk = data['sdk']
 248         timestamp = int(time.time())
 249         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' % (
 250             username, self._rsa_fun(password.encode('utf-8')))
 251
 252         interp = IqiyiSDKInterpreter(sdk)
 253         sign = interp.run(target, data['ip'], timestamp)
 254
 255         validation_params = {
 256             'target': target,
 257             'server': 'BEA3AA1908656AABCCFF76582C4C6660',
 258             'token': data['token'],
 259             'bird_src': 'f8d91d57af224da7893dd397d52d811a',
 260             'sign': sign,
 261             'bird_t': timestamp,
 262         }
 263         validation_result = self._download_json(
 264             'http://kylin.iqiyi.com/validate?' + compat_urllib_parse_urlencode(validation_params), None,
 265             note='Validate credentials', errnote='Unable to validate credentials')
 266
 267         MSG_MAP = {
 268             'P00107': 'please login via the web interface and enter the CAPTCHA code',
 269             'P00117': 'bad username or password',
 270         }
 271
 272         code = validation_result['code']
 273         if code != 'A00000':
 274             msg = MSG_MAP.get(code)
 275             if not msg:
 276                 msg = 'error %s' % code
 277                 if validation_result.get('msg'):
 278                     msg += ': ' + validation_result['msg']
 279             self._downloader.report_warning('unable to log in: ' + msg)
 280             return False
 281
 282         return True
 283
 284     def get_raw_data(self, tvid, video_id):
 285         tm = int(time.time() * 1000)
 286
 287         key = 'd5fb4bd9d50c4be6948c97edd7254b0e'
 288         sc = md5_text(compat_str(tm) + key + tvid)
 289         params = {
 290             'tvid': tvid,
 291             'vid': video_id,
 292             'src': '76f90cbd92f94a2e925d83e8ccd22cb7',
 293             'sc': sc,
 294             't': tm,
 295         }
 296
 297         return self._download_json(
 298             'http://cache.m.iqiyi.com/jp/tmts/%s/%s/' % (tvid, video_id),
 299             video_id, transform_source=lambda s: remove_start(s, 'var tvInfoJs='),
 300             query=params, headers=self.geo_verification_headers())
 301
 302     def _extract_playlist(self, webpage):
 303         PAGE_SIZE = 50
 304
 305         links = re.findall(
 306             r'<a[^>]+class="site-piclist_pic_link"[^>]+href="(http://www\.iqiyi\.com/.+\.html)"',
 307             webpage)
 308         if not links:
 309             return
 310
 311         album_id = self._search_regex(
 312             r'albumId\s*:\s*(\d+),', webpage, 'album ID')
 313         album_title = self._search_regex(
 314             r'data-share-title="([^"]+)"', webpage, 'album title', fatal=False)
 315
 316         entries = list(map(self.url_result, links))
 317
 318         # Start from 2 because links in the first page are already on webpage
 319         for page_num in itertools.count(2):
 320             pagelist_page = self._download_webpage(
 321                 'http://cache.video.qiyi.com/jp/avlist/%s/%d/%d/' % (album_id, page_num, PAGE_SIZE),
 322                 album_id,
 323                 note='Download playlist page %d' % page_num,
 324                 errnote='Failed to download playlist page %d' % page_num)
 325             pagelist = self._parse_json(
 326                 remove_start(pagelist_page, 'var tvInfoJs='), album_id)
 327             vlist = pagelist['data']['vlist']
 328             for item in vlist:
 329                 entries.append(self.url_result(item['vurl']))
 330             if len(vlist) < PAGE_SIZE:
 331                 break
 332
 333         return self.playlist_result(entries, album_id, album_title)
 334
 335     def _real_extract(self, url):
 336         webpage = self._download_webpage(
 337             url, 'temp_id', note='download video page')
 338
 339         # There's no simple way to determine whether an URL is a playlist or not
 340         # So detect it
 341         playlist_result = self._extract_playlist(webpage)
 342         if playlist_result:
 343             return playlist_result
 344
 345         tvid = self._search_regex(
 346             r'data-player-tvid\s*=\s*[\'"](\d+)', webpage, 'tvid')
 347         video_id = self._search_regex(
 348             r'data-player-videoid\s*=\s*[\'"]([a-f\d]+)', webpage, 'video_id')
 349
 350         formats = []
 351         for _ in range(5):
 352             raw_data = self.get_raw_data(tvid, video_id)
 353
 354             if raw_data['code'] != 'A00000':
 355                 if raw_data['code'] == 'A00111':
 356                     self.raise_geo_restricted()
 357                 raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
 358
 359             data = raw_data['data']
 360
 361             for stream in data['vidl']:
 362                 if 'm3utx' not in stream:
 363                     continue
 364                 vd = compat_str(stream['vd'])
 365                 formats.append({
 366                     'url': stream['m3utx'],
 367                     'format_id': vd,
 368                     'ext': 'mp4',
 369                     'preference': self._FORMATS_MAP.get(vd, -1),
 370                     'protocol': 'm3u8_native',
 371                 })
 372
 373             if formats:
 374                 break
 375
 376             self._sleep(5, video_id)
 377
 378         self._sort_formats(formats)
 379         title = (get_element_by_id('widget-videotitle', webpage) or
 380                  clean_html(get_element_by_attribute('class', 'mod-play-tit', webpage)))
 381
 382         return {
 383             'id': video_id,
 384             'title': title,
 385             'formats': formats,
 386         }