]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/iqiyi.py
Initiate new release.
[youtubedl] / youtube_dl / extractor / iqiyi.py
index 2df1da3f00a9d9655a0c9f6a447510dcf9e2caf0..ddcb3c916e6a0610484dc5ceddbd84b507e761fd 100644 (file)
 from __future__ import unicode_literals
 
 import hashlib
+import itertools
 import math
+import os
 import random
+import re
 import time
 import uuid
 
 from .common import InfoExtractor
-from ..compat import compat_urllib_parse
-from ..utils import ExtractorError
+from ..compat import (
+    compat_parse_qs,
+    compat_str,
+    compat_urllib_parse_urlencode,
+    compat_urllib_parse_urlparse,
+)
+from ..utils import (
+    decode_packed_codes,
+    ExtractorError,
+    ohdave_rsa_encrypt,
+    remove_start,
+    sanitized_Request,
+    urlencode_postdata,
+    url_basename,
+)
+
+
+def md5_text(text):
+    return hashlib.md5(text.encode('utf-8')).hexdigest()
+
+
+class IqiyiSDK(object):
+    def __init__(self, target, ip, timestamp):
+        self.target = target
+        self.ip = ip
+        self.timestamp = timestamp
+
+    @staticmethod
+    def split_sum(data):
+        return compat_str(sum(map(lambda p: int(p, 16), list(data))))
+
+    @staticmethod
+    def digit_sum(num):
+        if isinstance(num, int):
+            num = compat_str(num)
+        return compat_str(sum(map(int, num)))
+
+    def even_odd(self):
+        even = self.digit_sum(compat_str(self.timestamp)[::2])
+        odd = self.digit_sum(compat_str(self.timestamp)[1::2])
+        return even, odd
+
+    def preprocess(self, chunksize):
+        self.target = md5_text(self.target)
+        chunks = []
+        for i in range(32 // chunksize):
+            chunks.append(self.target[chunksize * i:chunksize * (i + 1)])
+        if 32 % chunksize:
+            chunks.append(self.target[32 - 32 % chunksize:])
+        return chunks, list(map(int, self.ip.split('.')))
+
+    def mod(self, modulus):
+        chunks, ip = self.preprocess(32)
+        self.target = chunks[0] + ''.join(map(lambda p: compat_str(p % modulus), ip))
+
+    def split(self, chunksize):
+        modulus_map = {
+            4: 256,
+            5: 10,
+            8: 100,
+        }
+
+        chunks, ip = self.preprocess(chunksize)
+        ret = ''
+        for i in range(len(chunks)):
+            ip_part = compat_str(ip[i] % modulus_map[chunksize]) if i < 4 else ''
+            if chunksize == 8:
+                ret += ip_part + chunks[i]
+            else:
+                ret += chunks[i] + ip_part
+        self.target = ret
+
+    def handle_input16(self):
+        self.target = md5_text(self.target)
+        self.target = self.split_sum(self.target[:16]) + self.target + self.split_sum(self.target[16:])
+
+    def handle_input8(self):
+        self.target = md5_text(self.target)
+        ret = ''
+        for i in range(4):
+            part = self.target[8 * i:8 * (i + 1)]
+            ret += self.split_sum(part) + part
+        self.target = ret
+
+    def handleSum(self):
+        self.target = md5_text(self.target)
+        self.target = self.split_sum(self.target) + self.target
+
+    def date(self, scheme):
+        self.target = md5_text(self.target)
+        d = time.localtime(self.timestamp)
+        strings = {
+            'y': compat_str(d.tm_year),
+            'm': '%02d' % d.tm_mon,
+            'd': '%02d' % d.tm_mday,
+        }
+        self.target += ''.join(map(lambda c: strings[c], list(scheme)))
+
+    def split_time_even_odd(self):
+        even, odd = self.even_odd()
+        self.target = odd + md5_text(self.target) + even
+
+    def split_time_odd_even(self):
+        even, odd = self.even_odd()
+        self.target = even + md5_text(self.target) + odd
+
+    def split_ip_time_sum(self):
+        chunks, ip = self.preprocess(32)
+        self.target = compat_str(sum(ip)) + chunks[0] + self.digit_sum(self.timestamp)
+
+    def split_time_ip_sum(self):
+        chunks, ip = self.preprocess(32)
+        self.target = self.digit_sum(self.timestamp) + chunks[0] + compat_str(sum(ip))
+
+
+class IqiyiSDKInterpreter(object):
+    def __init__(self, sdk_code):
+        self.sdk_code = sdk_code
+
+    def run(self, target, ip, timestamp):
+        self.sdk_code = decode_packed_codes(self.sdk_code)
+
+        functions = re.findall(r'input=([a-zA-Z0-9]+)\(input', self.sdk_code)
+
+        sdk = IqiyiSDK(target, ip, timestamp)
+
+        other_functions = {
+            'handleSum': sdk.handleSum,
+            'handleInput8': sdk.handle_input8,
+            'handleInput16': sdk.handle_input16,
+            'splitTimeEvenOdd': sdk.split_time_even_odd,
+            'splitTimeOddEven': sdk.split_time_odd_even,
+            'splitIpTimeSum': sdk.split_ip_time_sum,
+            'splitTimeIpSum': sdk.split_time_ip_sum,
+        }
+        for function in functions:
+            if re.match(r'mod\d+', function):
+                sdk.mod(int(function[3:]))
+            elif re.match(r'date[ymd]{3}', function):
+                sdk.date(function[4:])
+            elif re.match(r'split\d+', function):
+                sdk.split(int(function[5:]))
+            elif function in other_functions:
+                other_functions[function]()
+            else:
+                raise ExtractorError('Unknown funcion %s' % function)
+
+        return sdk.target
 
 
 class IqiyiIE(InfoExtractor):
     IE_NAME = 'iqiyi'
     IE_DESC = '爱奇艺'
 
-    _VALID_URL = r'http://(?:www\.)iqiyi.com/v_.+?\.html'
+    _VALID_URL = r'https?://(?:(?:[^.]+\.)?iqiyi\.com|www\.pps\.tv)/.+\.html'
+
+    _NETRC_MACHINE = 'iqiyi'
 
     _TESTS = [{
         'url': 'http://www.iqiyi.com/v_19rrojlavg.html',
@@ -84,6 +235,47 @@ class IqiyiIE(InfoExtractor):
         'params': {
             'skip_download': True,
         },
+    }, {
+        'url': 'http://www.iqiyi.com/w_19rt6o8t9p.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.iqiyi.com/a_19rrhbc6kt.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://yule.iqiyi.com/pcb.html',
+        'only_matching': True,
+    }, {
+        # VIP-only video. The first 2 parts (6 minutes) are available without login
+        # MD5 sums omitted as values are different on Travis CI and my machine
+        'url': 'http://www.iqiyi.com/v_19rrny4w8w.html',
+        'info_dict': {
+            'id': 'f3cf468b39dddb30d676f89a91200dc1',
+            'title': '泰坦尼克号',
+        },
+        'playlist': [{
+            'info_dict': {
+                'id': 'f3cf468b39dddb30d676f89a91200dc1_part1',
+                'ext': 'f4v',
+                'title': '泰坦尼克号',
+            },
+        }, {
+            'info_dict': {
+                'id': 'f3cf468b39dddb30d676f89a91200dc1_part2',
+                'ext': 'f4v',
+                'title': '泰坦尼克号',
+            },
+        }],
+        'expected_warnings': ['Needs a VIP account for full video'],
+    }, {
+        'url': 'http://www.iqiyi.com/a_19rrhb8ce1.html',
+        'info_dict': {
+            'id': '202918101',
+            'title': '灌篮高手 国语版',
+        },
+        'playlist_count': 101,
+    }, {
+        'url': 'http://www.pps.tv/w_19rrbav0ph.html',
+        'only_matching': True,
     }]
 
     _FORMATS_MAP = [
@@ -95,11 +287,112 @@ class IqiyiIE(InfoExtractor):
         ('10', 'h1'),
     ]
 
+    AUTH_API_ERRORS = {
+        # No preview available (不允许试看鉴权失败)
+        'Q00505': 'This video requires a VIP account',
+        # End of preview time (试看结束鉴权失败)
+        'Q00506': 'Needs a VIP account for full video',
+    }
+
+    def _real_initialize(self):
+        self._login()
+
     @staticmethod
-    def md5_text(text):
-        return hashlib.md5(text.encode('utf-8')).hexdigest()
+    def _rsa_fun(data):
+        # public key extracted from http://static.iqiyi.com/js/qiyiV2/20160129180840/jobs/i18n/i18nIndex.js
+        N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
+        e = 65537
+
+        return ohdave_rsa_encrypt(data, e, N)
+
+    def _login(self):
+        (username, password) = self._get_login_info()
+
+        # No authentication to be performed
+        if not username:
+            return True
+
+        data = self._download_json(
+            'http://kylin.iqiyi.com/get_token', None,
+            note='Get token for logging', errnote='Unable to get token for logging')
+        sdk = data['sdk']
+        timestamp = int(time.time())
+        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' % (
+            username, self._rsa_fun(password.encode('utf-8')))
+
+        interp = IqiyiSDKInterpreter(sdk)
+        sign = interp.run(target, data['ip'], timestamp)
+
+        validation_params = {
+            'target': target,
+            'server': 'BEA3AA1908656AABCCFF76582C4C6660',
+            'token': data['token'],
+            'bird_src': 'f8d91d57af224da7893dd397d52d811a',
+            'sign': sign,
+            'bird_t': timestamp,
+        }
+        validation_result = self._download_json(
+            'http://kylin.iqiyi.com/validate?' + compat_urllib_parse_urlencode(validation_params), None,
+            note='Validate credentials', errnote='Unable to validate credentials')
+
+        MSG_MAP = {
+            'P00107': 'please login via the web interface and enter the CAPTCHA code',
+            'P00117': 'bad username or password',
+        }
 
-    def construct_video_urls(self, data, video_id, _uuid):
+        code = validation_result['code']
+        if code != 'A00000':
+            msg = MSG_MAP.get(code)
+            if not msg:
+                msg = 'error %s' % code
+                if validation_result.get('msg'):
+                    msg += ': ' + validation_result['msg']
+            self._downloader.report_warning('unable to log in: ' + msg)
+            return False
+
+        return True
+
+    def _authenticate_vip_video(self, api_video_url, video_id, tvid, _uuid, do_report_warning):
+        auth_params = {
+            # version and platform hard-coded in com/qiyi/player/core/model/remote/AuthenticationRemote.as
+            'version': '2.0',
+            'platform': 'b6c13e26323c537d',
+            'aid': tvid,
+            'tvid': tvid,
+            'uid': '',
+            'deviceId': _uuid,
+            'playType': 'main',  # XXX: always main?
+            'filename': os.path.splitext(url_basename(api_video_url))[0],
+        }
+
+        qd_items = compat_parse_qs(compat_urllib_parse_urlparse(api_video_url).query)
+        for key, val in qd_items.items():
+            auth_params[key] = val[0]
+
+        auth_req = sanitized_Request(
+            'http://api.vip.iqiyi.com/services/ckn.action',
+            urlencode_postdata(auth_params))
+        # iQiyi server throws HTTP 405 error without the following header
+        auth_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
+        auth_result = self._download_json(
+            auth_req, video_id,
+            note='Downloading video authentication JSON',
+            errnote='Unable to download video authentication JSON')
+
+        code = auth_result.get('code')
+        msg = self.AUTH_API_ERRORS.get(code) or auth_result.get('msg') or code
+        if code == 'Q00506':
+            if do_report_warning:
+                self.report_warning(msg)
+            return False
+        if 'data' not in auth_result:
+            if msg is not None:
+                raise ExtractorError('%s said: %s' % (self.IE_NAME, msg), expected=True)
+            raise ExtractorError('Unexpected error from Iqiyi auth API')
+
+        return auth_result['data']
+
+    def construct_video_urls(self, data, video_id, _uuid, tvid):
         def do_xor(x, y):
             a = y % 3
             if a == 1:
@@ -125,9 +418,10 @@ class IqiyiIE(InfoExtractor):
                 note='Download path key of segment %d for format %s' % (segment_index + 1, format_id)
             )['t']
             t = str(int(math.floor(int(tm) / (600.0))))
-            return self.md5_text(t + mg + x)
+            return md5_text(t + mg + x)
 
         video_urls_dict = {}
+        need_vip_warning_report = True
         for format_item in data['vp']['tkl'][0]['vs']:
             if 0 < int(format_item['bid']) <= 10:
                 format_id = self.get_format(format_item['bid'])
@@ -146,11 +440,13 @@ class IqiyiIE(InfoExtractor):
                 vl = segment['l']
                 if not vl.startswith('/'):
                     vl = get_encode_code(vl)
-                key = get_path_key(
-                    vl.split('/')[-1].split('.')[0], format_id, segment_index)
+                is_vip_video = '/vip/' in vl
                 filesize = segment['b']
                 base_url = data['vp']['du'].split('/')
-                base_url.insert(-1, key)
+                if not is_vip_video:
+                    key = get_path_key(
+                        vl.split('/')[-1].split('.')[0], format_id, segment_index)
+                    base_url.insert(-1, key)
                 base_url = '/'.join(base_url)
                 param = {
                     'su': _uuid,
@@ -161,8 +457,23 @@ class IqiyiIE(InfoExtractor):
                     'ct': '',
                     'tn': str(int(time.time()))
                 }
-                api_video_url = base_url + vl + '?' + \
-                    compat_urllib_parse.urlencode(param)
+                api_video_url = base_url + vl
+                if is_vip_video:
+                    api_video_url = api_video_url.replace('.f4v', '.hml')
+                    auth_result = self._authenticate_vip_video(
+                        api_video_url, video_id, tvid, _uuid, need_vip_warning_report)
+                    if auth_result is False:
+                        need_vip_warning_report = False
+                        break
+                    param.update({
+                        't': auth_result['t'],
+                        # cid is hard-coded in com/qiyi/player/core/player/RuntimeData.as
+                        'cid': 'afbe8fd3d73448c9',
+                        'vid': video_id,
+                        'QY00001': auth_result['u'],
+                    })
+                api_video_url += '?' if '?' not in api_video_url else '&'
+                api_video_url += compat_urllib_parse_urlencode(param)
                 js = self._download_json(
                     api_video_url, video_id,
                     note='Download video info of segment %d for format %s' % (segment_index + 1, format_id))
@@ -186,58 +497,96 @@ class IqiyiIE(InfoExtractor):
         tail = tm + tvid
         param = {
             'key': 'fvip',
-            'src': self.md5_text('youtube-dl'),
+            'src': md5_text('youtube-dl'),
             'tvId': tvid,
             'vid': video_id,
             'vinfo': 1,
             'tm': tm,
-            'enc': self.md5_text(enc_key + tail),
+            'enc': md5_text(enc_key + tail),
             'qyid': _uuid,
             'tn': random.random(),
-            'um': 0,
-            'authkey': self.md5_text(self.md5_text('') + tail),
+            # In iQiyi's flash player, um is set to 1 if there's a logged user
+            # Some 1080P formats are only available with a logged user.
+            # Here force um=1 to trick the iQiyi server
+            'um': 1,
+            'authkey': md5_text(md5_text('') + tail),
+            'k_tag': 1,
         }
 
         api_url = 'http://cache.video.qiyi.com/vms' + '?' + \
-            compat_urllib_parse.urlencode(param)
+            compat_urllib_parse_urlencode(param)
         raw_data = self._download_json(api_url, video_id)
         return raw_data
 
-    def get_enc_key(self, swf_url, video_id):
+    def get_enc_key(self, video_id):
         # TODO: automatic key extraction
-        # last update at 2015-10-22 for Zombie::bite
-        # '7223c67061dbea1259d0ceb44f44b6d62288f4f80c972170de5201d2321060270e05'[2:66][0::2]
-        enc_key = '2c76de15dcb44bd28ff0927d50d31620'
+        # last update at 2016-01-22 for Zombie::bite
+        enc_key = '4a1caba4b4465345366f28da7c117d20'
         return enc_key
 
+    def _extract_playlist(self, webpage):
+        PAGE_SIZE = 50
+
+        links = re.findall(
+            r'<a[^>]+class="site-piclist_pic_link"[^>]+href="(http://www\.iqiyi\.com/.+\.html)"',
+            webpage)
+        if not links:
+            return
+
+        album_id = self._search_regex(
+            r'albumId\s*:\s*(\d+),', webpage, 'album ID')
+        album_title = self._search_regex(
+            r'data-share-title="([^"]+)"', webpage, 'album title', fatal=False)
+
+        entries = list(map(self.url_result, links))
+
+        # Start from 2 because links in the first page are already on webpage
+        for page_num in itertools.count(2):
+            pagelist_page = self._download_webpage(
+                'http://cache.video.qiyi.com/jp/avlist/%s/%d/%d/' % (album_id, page_num, PAGE_SIZE),
+                album_id,
+                note='Download playlist page %d' % page_num,
+                errnote='Failed to download playlist page %d' % page_num)
+            pagelist = self._parse_json(
+                remove_start(pagelist_page, 'var tvInfoJs='), album_id)
+            vlist = pagelist['data']['vlist']
+            for item in vlist:
+                entries.append(self.url_result(item['vurl']))
+            if len(vlist) < PAGE_SIZE:
+                break
+
+        return self.playlist_result(entries, album_id, album_title)
+
     def _real_extract(self, url):
         webpage = self._download_webpage(
             url, 'temp_id', note='download video page')
+
+        # There's no simple way to determine whether an URL is a playlist or not
+        # So detect it
+        playlist_result = self._extract_playlist(webpage)
+        if playlist_result:
+            return playlist_result
+
         tvid = self._search_regex(
             r'data-player-tvid\s*=\s*[\'"](\d+)', webpage, 'tvid')
         video_id = self._search_regex(
             r'data-player-videoid\s*=\s*[\'"]([a-f\d]+)', webpage, 'video_id')
-        swf_url = self._search_regex(
-            r'(http://[^\'"]+MainPlayer[^.]+\.swf)', webpage, 'swf player URL')
         _uuid = uuid.uuid4().hex
 
-        enc_key = self.get_enc_key(swf_url, video_id)
+        enc_key = self.get_enc_key(video_id)
 
         raw_data = self.get_raw_data(tvid, video_id, enc_key, _uuid)
 
         if raw_data['code'] != 'A000000':
             raise ExtractorError('Unable to load data. Error code: ' + raw_data['code'])
 
-        if not raw_data['data']['vp']['tkl']:
-            raise ExtractorError('No support iQiqy VIP video')
-
         data = raw_data['data']
 
         title = data['vi']['vn']
 
         # generate video_urls_dict
         video_urls_dict = self.construct_video_urls(
-            data, video_id, _uuid)
+            data, video_id, _uuid, tvid)
 
         # construct info
         entries = []