]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/youporn.py
e3b56cece6bd9d8bebfe3cfffe7a1aefe2614131
[youtubedl] / youtube_dl / extractor / youporn.py
1 import json
2 import os
3 import re
4 import sys
5
6 from .common import InfoExtractor
7 from ..utils import (
8 compat_urllib_parse_urlparse,
9 compat_urllib_request,
10
11 ExtractorError,
12 unescapeHTML,
13 unified_strdate,
14 )
15 from ..aes import (
16 aes_decrypt_text
17 )
18
19 class YouPornIE(InfoExtractor):
20 _VALID_URL = r'^(?:https?://)?(?:\w+\.)?youporn\.com/watch/(?P<videoid>[0-9]+)/(?P<title>[^/]+)'
21 _TEST = {
22 u'url': u'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/',
23 u'file': u'505835.mp4',
24 u'md5': u'71ec5fcfddacf80f495efa8b6a8d9a89',
25 u'info_dict': {
26 u"upload_date": u"20101221",
27 u"description": u"Love & Sex Answers: http://bit.ly/DanAndJenn -- Is It Unhealthy To Masturbate Daily?",
28 u"uploader": u"Ask Dan And Jennifer",
29 u"title": u"Sex Ed: Is It Safe To Masturbate Daily?",
30 u"age_limit": 18,
31 }
32 }
33
34 def _print_formats(self, formats):
35 """Print all available formats"""
36 print(u'Available formats:')
37 print(u'ext\t\tformat')
38 print(u'---------------------------------')
39 for format in formats:
40 print(u'%s\t\t%s' % (format['ext'], format['format']))
41
42 def _specific(self, req_format, formats):
43 for x in formats:
44 if x["format"] == req_format:
45 return x
46 return None
47
48 def _real_extract(self, url):
49 mobj = re.match(self._VALID_URL, url)
50 video_id = mobj.group('videoid')
51
52 req = compat_urllib_request.Request(url)
53 req.add_header('Cookie', 'age_verified=1')
54 webpage = self._download_webpage(req, video_id)
55 age_limit = self._rta_search(webpage)
56
57 # Get JSON parameters
58 json_params = self._search_regex(r'var currentVideo = new Video\((.*)\);', webpage, u'JSON parameters')
59 try:
60 params = json.loads(json_params)
61 except:
62 raise ExtractorError(u'Invalid JSON')
63
64 self.report_extraction(video_id)
65 try:
66 video_title = params['title']
67 upload_date = unified_strdate(params['release_date_f'])
68 video_description = params['description']
69 video_uploader = params['submitted_by']
70 thumbnail = params['thumbnails'][0]['image']
71 except KeyError:
72 raise ExtractorError('Missing JSON parameter: ' + sys.exc_info()[1])
73
74 # Get all of the formats available
75 DOWNLOAD_LIST_RE = r'(?s)<ul class="downloadList">(?P<download_list>.*?)</ul>'
76 download_list_html = self._search_regex(DOWNLOAD_LIST_RE,
77 webpage, u'download list').strip()
78
79 # Get all of the links from the page
80 LINK_RE = r'(?s)<a href="(?P<url>[^"]+)">'
81 links = re.findall(LINK_RE, download_list_html)
82
83 # Get link of hd video if available
84 mobj = re.search(r'var encryptedQuality720URL = \'(?P<encrypted_video_url>[a-zA-Z0-9+/]+={0,2})\';', webpage)
85 if mobj != None:
86 encrypted_video_url = mobj.group(u'encrypted_video_url')
87 video_url = aes_decrypt_text(encrypted_video_url, video_title, 32).decode('utf-8')
88 links = [video_url] + links
89
90 if not links:
91 raise ExtractorError(u'ERROR: no known formats available for video')
92
93 self.to_screen(u'Links found: %d' % len(links))
94
95 formats = []
96 for link in links:
97
98 # A link looks like this:
99 # http://cdn1.download.youporn.phncdn.com/201210/31/8004515/480p_370k_8004515/YouPorn%20-%20Nubile%20Films%20The%20Pillow%20Fight.mp4?nvb=20121113051249&nva=20121114051249&ir=1200&sr=1200&hash=014b882080310e95fb6a0
100 # A path looks like this:
101 # /201210/31/8004515/480p_370k_8004515/YouPorn%20-%20Nubile%20Films%20The%20Pillow%20Fight.mp4
102 video_url = unescapeHTML( link )
103 path = compat_urllib_parse_urlparse( video_url ).path
104 extension = os.path.splitext( path )[1][1:]
105 format = path.split('/')[4].split('_')[:2]
106 # size = format[0]
107 # bitrate = format[1]
108 format = "-".join( format )
109 # title = u'%s-%s-%s' % (video_title, size, bitrate)
110
111 formats.append({
112 'id': video_id,
113 'url': video_url,
114 'uploader': video_uploader,
115 'upload_date': upload_date,
116 'title': video_title,
117 'ext': extension,
118 'format': format,
119 'thumbnail': thumbnail,
120 'description': video_description,
121 'age_limit': age_limit,
122 })
123
124 if self._downloader.params.get('listformats', None):
125 self._print_formats(formats)
126 return
127
128 req_format = self._downloader.params.get('format', 'best')
129 self.to_screen(u'Format: %s' % req_format)
130
131 if req_format is None or req_format == 'best':
132 return [formats[0]]
133 elif req_format == 'worst':
134 return [formats[-1]]
135 elif req_format in ('-1', 'all'):
136 return formats
137 else:
138 format = self._specific( req_format, formats )
139 if format is None:
140 raise ExtractorError(u'Requested format not available')
141 return [format]