Merge remote-tracking branch 'h-collector/master'
[ytdl] / youtube_dl / extractor / fc2.py
1 #! -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
3
4 import hashlib
5
6 from .common import InfoExtractor
7 from ..compat import (
8     compat_urllib_parse,
9     compat_urllib_request,
10     compat_urlparse,
11 )
12 from ..utils import (
13     ExtractorError,
14 )
15
16
17 class FC2IE(InfoExtractor):
18     _VALID_URL = r'^http://video\.fc2\.com/(?:[^/]+/)?content/(?P<id>[^/]+)'
19     IE_NAME = 'fc2'
20     _NETRC_MACHINE = 'fc2'
21     _TESTS = [{
22         'url': 'http://video.fc2.com/en/content/20121103kUan1KHs',
23         'md5': 'a6ebe8ebe0396518689d963774a54eb7',
24         'info_dict': {
25             'id': '20121103kUan1KHs',
26             'ext': 'flv',
27             'title': 'Boxing again with Puff',
28         },
29     }, {
30         'url': 'http://video.fc2.com/en/content/20150125cEva0hDn/',
31         'info_dict': {
32             'id': '20150125cEva0hDn',
33             'ext': 'mp4',
34         },
35         'params': {
36             'username': 'ytdl@yt-dl.org',
37             'password': '(snip)',
38             'skip': 'requires actual password'
39         }
40     }]
41
42     def _login(self):
43         (username, password) = self._get_login_info()
44         if username is None or password is None:
45             return False
46
47         # Log in
48         login_form_strs = {
49             'email': username,
50             'password': password,
51             'done': 'video',
52             'Submit': ' Login ',
53         }
54
55         # Convert to UTF-8 *before* urlencode because Python 2.x's urlencode
56         # chokes on unicode
57         login_form = dict((k.encode('utf-8'), v.encode('utf-8')) for k, v in login_form_strs.items())
58         login_data = compat_urllib_parse.urlencode(login_form).encode('utf-8')
59         request = compat_urllib_request.Request(
60             'https://secure.id.fc2.com/index.php?mode=login&switch_language=en', login_data)
61
62         login_results = self._download_webpage(request, None, note='Logging in', errnote='Unable to log in')
63         if 'mode=redirect&login=done' not in login_results:
64             self.report_warning('unable to log in: bad username or password')
65             return False
66
67         # this is also needed
68         login_redir = compat_urllib_request.Request('http://id.fc2.com/?mode=redirect&login=done')
69         self._download_webpage(
70             login_redir, None, note='Login redirect', errnote='Login redirect failed')
71
72         return True
73
74     def _real_extract(self, url):
75         video_id = self._match_id(url)
76         self._login()
77         webpage = self._download_webpage(url, video_id)
78         self._downloader.cookiejar.clear_session_cookies()  # must clear
79         self._login()
80
81         title = self._og_search_title(webpage)
82         thumbnail = self._og_search_thumbnail(webpage)
83         refer = url.replace('/content/', '/a/content/')
84
85         mimi = hashlib.md5((video_id + '_gGddgPfeaf_gzyr').encode('utf-8')).hexdigest()
86
87         info_url = (
88             "http://video.fc2.com/ginfo.php?mimi={1:s}&href={2:s}&v={0:s}&fversion=WIN%2011%2C6%2C602%2C180&from=2&otag=0&upid={0:s}&tk=null&".
89             format(video_id, mimi, compat_urllib_request.quote(refer, safe='').replace('.', '%2E')))
90
91         info_webpage = self._download_webpage(
92             info_url, video_id, note='Downloading info page')
93         info = compat_urlparse.parse_qs(info_webpage)
94
95         if 'err_code' in info:
96             # most of the time we can still download wideo even if err_code is 403 or 602
97             self.report_warning(
98                 'Error code was: %s... but still trying' % info['err_code'][0])
99
100         if 'filepath' not in info:
101             raise ExtractorError('Cannot download file. Are you logged in?')
102
103         video_url = info['filepath'][0] + '?mid=' + info['mid'][0]
104         title_info = info.get('title')
105         if title_info:
106             title = title_info[0]
107
108         return {
109             'id': video_id,
110             'title': title,
111             'url': video_url,
112             'ext': 'flv',
113             'thumbnail': thumbnail,
114         }