Merge branch 'webofstories' of https://github.com/dufferzafar/youtube-dl into dufferz...
[youtube-dl] / youtube_dl / extractor / shared.py
1 from __future__ import unicode_literals
2
3 import re
4 import base64
5
6 from .common import InfoExtractor
7 from ..compat import (
8     compat_urllib_parse,
9     compat_urllib_request,
10 )
11 from ..utils import (
12     ExtractorError,
13     int_or_none,
14 )
15
16
17 class SharedIE(InfoExtractor):
18     _VALID_URL = r'http://shared\.sx/(?P<id>[\da-z]{10})'
19
20     _TEST = {
21         'url': 'http://shared.sx/0060718775',
22         'md5': '106fefed92a8a2adb8c98e6a0652f49b',
23         'info_dict': {
24             'id': '0060718775',
25             'ext': 'mp4',
26             'title': 'Bmp4',
27         },
28     }
29
30     def _real_extract(self, url):
31         video_id = self._match_id(url)
32         webpage = self._download_webpage(url, video_id)
33
34         if '>File does not exist<' in webpage:
35             raise ExtractorError(
36                 'Video %s does not exist' % video_id, expected=True)
37
38         download_form = self._form_hidden_inputs(webpage)
39         request = compat_urllib_request.Request(
40             url, compat_urllib_parse.urlencode(download_form))
41         request.add_header('Content-Type', 'application/x-www-form-urlencoded')
42
43         video_page = self._download_webpage(
44             request, video_id, 'Downloading video page')
45
46         video_url = self._html_search_regex(
47             r'data-url="([^"]+)"', video_page, 'video URL')
48         title = base64.b64decode(self._html_search_meta(
49             'full:title', webpage, 'title').encode('utf-8')).decode('utf-8')
50         filesize = int_or_none(self._html_search_meta(
51             'full:size', webpage, 'file size', fatal=False))
52         thumbnail = self._html_search_regex(
53             r'data-poster="([^"]+)"', video_page, 'thumbnail', default=None)
54
55         return {
56             'id': video_id,
57             'url': video_url,
58             'ext': 'mp4',
59             'filesize': filesize,
60             'title': title,
61             'thumbnail': thumbnail,
62         }