[leeco] Fix extraction (closes #12974)

Seems on mobile devices a similar API is used, but I always get an AD
with mimicking that API.
This commit is contained in:
Yen Chi Hsuan 2017-05-04 03:18:56 +08:00
parent 7a6d33a9a5
commit 5401bea27f
No known key found for this signature in database
GPG Key ID: 7F902A182457CA23
2 changed files with 39 additions and 70 deletions

View File

@ -1,3 +1,9 @@
version <unreleased>
Extractors
* [leeco] Fix extraction (#12974)
version 2017.05.01 version 2017.05.01
Core Core

View File

@ -23,7 +23,6 @@
str_or_none, str_or_none,
url_basename, url_basename,
urshift, urshift,
update_url_query,
) )
@ -51,7 +50,7 @@ class LeIE(InfoExtractor):
'id': '1415246', 'id': '1415246',
'ext': 'mp4', 'ext': 'mp4',
'title': '美人天下01', 'title': '美人天下01',
'description': 'md5:f88573d9d7225ada1359eaf0dbf8bcda', 'description': 'md5:28942e650e82ed4fcc8e4de919ee854d',
}, },
'params': { 'params': {
'hls_prefer_native': True, 'hls_prefer_native': True,
@ -69,7 +68,6 @@ class LeIE(InfoExtractor):
'params': { 'params': {
'hls_prefer_native': True, 'hls_prefer_native': True,
}, },
'skip': 'Only available in China',
}, { }, {
'url': 'http://sports.le.com/video/25737697.html', 'url': 'http://sports.le.com/video/25737697.html',
'only_matching': True, 'only_matching': True,
@ -81,7 +79,7 @@ class LeIE(InfoExtractor):
'only_matching': True, 'only_matching': True,
}] }]
# ror() and calc_time_key() are reversed from a embedded swf file in KLetvPlayer.swf # ror() and calc_time_key() are reversed from a embedded swf file in LetvPlayer.swf
def ror(self, param1, param2): def ror(self, param1, param2):
_loc3_ = 0 _loc3_ = 0
while _loc3_ < param2: while _loc3_ < param2:
@ -90,15 +88,8 @@ def ror(self, param1, param2):
return param1 return param1
def calc_time_key(self, param1): def calc_time_key(self, param1):
_loc2_ = 773625421 _loc2_ = 185025305
_loc3_ = self.ror(param1, _loc2_ % 13) return self.ror(param1, _loc2_ % 17) ^ _loc2_
_loc3_ = _loc3_ ^ _loc2_
_loc3_ = self.ror(_loc3_, _loc2_ % 17)
return _loc3_
# reversed from http://jstatic.letvcdn.com/sdk/player.js
def get_mms_key(self, time):
return self.ror(time, 8) ^ 185025305
# see M3U8Encryption class in KLetvPlayer.swf # see M3U8Encryption class in KLetvPlayer.swf
@staticmethod @staticmethod
@ -122,7 +113,7 @@ def decrypt_m3u8(encrypted_data):
def _check_errors(self, play_json): def _check_errors(self, play_json):
# Check for errors # Check for errors
playstatus = play_json['playstatus'] playstatus = play_json['msgs']['playstatus']
if playstatus['status'] == 0: if playstatus['status'] == 0:
flag = playstatus['flag'] flag = playstatus['flag']
if flag == 1: if flag == 1:
@ -134,59 +125,32 @@ def _real_extract(self, url):
media_id = self._match_id(url) media_id = self._match_id(url)
page = self._download_webpage(url, media_id) page = self._download_webpage(url, media_id)
play_json_h5 = self._download_json(
'http://api.le.com/mms/out/video/playJsonH5',
media_id, 'Downloading html5 playJson data', query={
'id': media_id,
'platid': 3,
'splatid': 304,
'format': 1,
'tkey': self.get_mms_key(int(time.time())),
'domain': 'www.le.com',
'tss': 'no',
},
headers=self.geo_verification_headers())
self._check_errors(play_json_h5)
play_json_flash = self._download_json( play_json_flash = self._download_json(
'http://api.le.com/mms/out/video/playJson', 'http://player-pc.le.com/mms/out/video/playJson',
media_id, 'Downloading flash playJson data', query={ media_id, 'Downloading flash playJson data', query={
'id': media_id, 'id': media_id,
'platid': 1, 'platid': 1,
'splatid': 101, 'splatid': 101,
'format': 1, 'format': 1,
'source': 1000,
'tkey': self.calc_time_key(int(time.time())), 'tkey': self.calc_time_key(int(time.time())),
'domain': 'www.le.com', 'domain': 'www.le.com',
'region': 'cn',
}, },
headers=self.geo_verification_headers()) headers=self.geo_verification_headers())
self._check_errors(play_json_flash) self._check_errors(play_json_flash)
def get_h5_urls(media_url, format_id):
location = self._download_json(
media_url, media_id,
'Download JSON metadata for format %s' % format_id, query={
'format': 1,
'expect': 3,
'tss': 'no',
})['location']
return {
'http': update_url_query(location, {'tss': 'no'}),
'hls': update_url_query(location, {'tss': 'ios'}),
}
def get_flash_urls(media_url, format_id): def get_flash_urls(media_url, format_id):
media_url += '&' + compat_urllib_parse_urlencode({ nodes_data = self._download_json(
media_url, media_id,
'Download JSON metadata for format %s' % format_id,
query={
'm3v': 1, 'm3v': 1,
'format': 1, 'format': 1,
'expect': 3, 'expect': 3,
'rateid': format_id, 'tss': 'ios',
}) })
nodes_data = self._download_json(
media_url, media_id,
'Download JSON metadata for format %s' % format_id)
req = self._request_webpage( req = self._request_webpage(
nodes_data['nodelist'][0]['location'], media_id, nodes_data['nodelist'][0]['location'], media_id,
note='Downloading m3u8 information for format %s' % format_id) note='Downloading m3u8 information for format %s' % format_id)
@ -199,8 +163,7 @@ def get_flash_urls(media_url, format_id):
extracted_formats = [] extracted_formats = []
formats = [] formats = []
for play_json, get_urls in ((play_json_h5, get_h5_urls), (play_json_flash, get_flash_urls)): playurl = play_json_flash['msgs']['playurl']
playurl = play_json['playurl']
play_domain = playurl['domain'][0] play_domain = playurl['domain'][0]
for format_id, format_data in playurl.get('dispatch', []).items(): for format_id, format_data in playurl.get('dispatch', []).items():
@ -209,7 +172,7 @@ def get_flash_urls(media_url, format_id):
extracted_formats.append(format_id) extracted_formats.append(format_id)
media_url = play_domain + format_data[0] media_url = play_domain + format_data[0]
for protocol, format_url in get_urls(media_url, format_id).items(): for protocol, format_url in get_flash_urls(media_url, format_id).items():
f = { f = {
'url': format_url, 'url': format_url,
'ext': determine_ext(format_data[1]), 'ext': determine_ext(format_data[1]),