]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/screencast.py
306869e6af109cf6971a72df14f47fe99a7b6885
[youtubedl] / youtube_dl / extractor / screencast.py
1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..utils import (
8 ExtractorError,
9 compat_parse_qs,
10 compat_urllib_request,
11 )
12
13
14 class ScreencastIE(InfoExtractor):
15 _VALID_URL = r'https?://www\.screencast\.com/t/(?P<id>[a-zA-Z0-9]+)'
16 _TESTS = [{
17 'url': 'http://www.screencast.com/t/3ZEjQXlT',
18 'md5': '917df1c13798a3e96211dd1561fded83',
19 'info_dict': {
20 'id': '3ZEjQXlT',
21 'ext': 'm4v',
22 'title': 'Color Measurement with Ocean Optics Spectrometers',
23 'description': 'md5:240369cde69d8bed61349a199c5fb153',
24 'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
25 }
26 }, {
27 'url': 'http://www.screencast.com/t/V2uXehPJa1ZI',
28 'md5': 'e8e4b375a7660a9e7e35c33973410d34',
29 'info_dict': {
30 'id': 'V2uXehPJa1ZI',
31 'ext': 'mov',
32 'title': 'The Amadeus Spectrometer',
33 'description': 're:^In this video, our friends at.*To learn more about Amadeus, visit',
34 'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
35 }
36 }, {
37 'url': 'http://www.screencast.com/t/aAB3iowa',
38 'md5': 'dedb2734ed00c9755761ccaee88527cd',
39 'info_dict': {
40 'id': 'aAB3iowa',
41 'ext': 'mp4',
42 'title': 'Google Earth Export',
43 'description': 'Provides a demo of a CommunityViz export to Google Earth, one of the 3D viewing options.',
44 'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
45 }
46 }, {
47 'url': 'http://www.screencast.com/t/X3ddTrYh',
48 'md5': '669ee55ff9c51988b4ebc0877cc8b159',
49 'info_dict': {
50 'id': 'X3ddTrYh',
51 'ext': 'wmv',
52 'title': 'Toolkit 6 User Group Webinar (2014-03-04) - Default Judgment and First Impression',
53 'description': 'md5:7b9f393bc92af02326a5c5889639eab0',
54 'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
55 }
56 },
57 ]
58
59 def _real_extract(self, url):
60 mobj = re.match(self._VALID_URL, url)
61 video_id = mobj.group('id')
62 webpage = self._download_webpage(url, video_id)
63
64 video_url = self._html_search_regex(
65 r'<embed name="Video".*?src="([^"]+)"', webpage,
66 'QuickTime embed', default=None)
67
68 if video_url is None:
69 flash_vars_s = self._html_search_regex(
70 r'<param name="flashVars" value="([^"]+)"', webpage, 'flash vars',
71 default=None)
72 if not flash_vars_s:
73 flash_vars_s = self._html_search_regex(
74 r'<param name="initParams" value="([^"]+)"', webpage, 'flash vars',
75 default=None)
76 if flash_vars_s:
77 flash_vars_s = flash_vars_s.replace(',', '&')
78 if flash_vars_s:
79 flash_vars = compat_parse_qs(flash_vars_s)
80 video_url_raw = compat_urllib_request.quote(
81 flash_vars['content'][0])
82 video_url = video_url_raw.replace('http%3A', 'http:')
83
84 if video_url is None:
85 video_meta = self._html_search_meta(
86 'og:video', webpage, default=None)
87 if video_meta:
88 video_url = self._search_regex(
89 r'src=(.*?)(?:$|&)', video_meta,
90 'meta tag video URL', default=None)
91
92 if video_url is None:
93 raise ExtractorError('Cannot find video')
94
95 title = self._og_search_title(webpage, default=None)
96 if title is None:
97 title = self._html_search_regex(
98 [r'<b>Title:</b> ([^<]*)</div>',
99 r'class="tabSeperator">></span><span class="tabText">(.*?)<'],
100 webpage, 'title')
101 thumbnail = self._og_search_thumbnail(webpage)
102 description = self._og_search_description(webpage, default=None)
103 if description is None:
104 description = self._html_search_meta('description', webpage)
105
106 return {
107 'id': video_id,
108 'url': video_url,
109 'title': title,
110 'description': description,
111 'thumbnail': thumbnail,
112 }