]>
Commit | Line | Data |
---|---|---|
ca9e7922 PH |
1 | from __future__ import unicode_literals |
2 | ||
aa0c8739 JMF |
3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
6 | from ..utils import ( | |
aea6e7fc | 7 | int_or_none, |
ca9e7922 | 8 | unescapeHTML, |
009a3408 | 9 | find_xpath_attr, |
aa0c8739 JMF |
10 | ) |
11 | ||
ca9e7922 | 12 | |
aa0c8739 | 13 | class CSpanIE(InfoExtractor): |
11a15be4 | 14 | _VALID_URL = r'http://(?:www\.)?c-span\.org/video/\?(?P<id>[0-9a-f]+)' |
ca9e7922 | 15 | IE_DESC = 'C-SPAN' |
11a15be4 | 16 | _TESTS = [{ |
009a3408 | 17 | 'url': 'http://www.c-span.org/video/?313572-1/HolderonV', |
ca9e7922 PH |
18 | 'md5': '8e44ce11f0f725527daccc453f553eb0', |
19 | 'info_dict': { | |
009a3408 JMF |
20 | 'id': '315139', |
21 | 'ext': 'mp4', | |
ca9e7922 | 22 | 'title': 'Attorney General Eric Holder on Voting Rights Act Decision', |
009a3408 | 23 | 'description': 'Attorney General Eric Holder spoke to reporters following the Supreme Court decision in Shelby County v. Holder in which the court ruled that the preclearance provisions of the Voting Rights Act could not be enforced until Congress established new guidelines for review.', |
6f5ac90c | 24 | }, |
11577ec0 | 25 | 'skip': 'Regularly fails on travis, for unknown reasons', |
11a15be4 PH |
26 | }, { |
27 | 'url': 'http://www.c-span.org/video/?c4486943/cspan-international-health-care-models', | |
6f13b055 | 28 | # For whatever reason, the served video alternates between |
11a15be4 PH |
29 | # two different ones |
30 | #'md5': 'dbb0f047376d457f2ab8b3929cbb2d0c', | |
31 | 'info_dict': { | |
32 | 'id': '340723', | |
33 | 'ext': 'mp4', | |
34 | 'title': 'International Health Care Models', | |
35 | 'description': 'md5:7a985a2d595dba00af3d9c9f0783c967', | |
36 | } | |
22a6f150 PH |
37 | }, { |
38 | 'url': 'http://www.c-span.org/video/?318608-1/gm-ignition-switch-recall', | |
39 | 'info_dict': { | |
40 | 'id': '342759', | |
41 | 'title': 'General Motors Ignition Switch Recall', | |
42 | }, | |
43 | 'playlist_duration_sum': 14855, | |
11a15be4 | 44 | }] |
aa0c8739 JMF |
45 | |
46 | def _real_extract(self, url): | |
47 | mobj = re.match(self._VALID_URL, url) | |
009a3408 JMF |
48 | page_id = mobj.group('id') |
49 | webpage = self._download_webpage(url, page_id) | |
11a15be4 | 50 | video_id = self._search_regex(r'progid=\'?([0-9]+)\'?>', webpage, 'video id') |
009a3408 JMF |
51 | |
52 | description = self._html_search_regex( | |
53 | [ | |
54 | # The full description | |
55 | r'<div class=\'expandable\'>(.*?)<a href=\'#\'', | |
56 | # If the description is small enough the other div is not | |
57 | # present, otherwise this is a stripped version | |
58 | r'<p class=\'initial\'>(.*?)</p>' | |
59 | ], | |
60 | webpage, 'description', flags=re.DOTALL) | |
ca9e7922 PH |
61 | |
62 | info_url = 'http://c-spanvideo.org/videoLibrary/assets/player/ajax-player.php?os=android&html5=program&id=' + video_id | |
009a3408 | 63 | data = self._download_json(info_url, video_id) |
ca9e7922 | 64 | |
aea6e7fc PH |
65 | doc = self._download_xml( |
66 | 'http://www.c-span.org/common/services/flashXml.php?programid=' + video_id, | |
009a3408 JMF |
67 | video_id) |
68 | ||
aea6e7fc PH |
69 | title = find_xpath_attr(doc, './/string', 'name', 'title').text |
70 | thumbnail = find_xpath_attr(doc, './/string', 'name', 'poster').text | |
71 | ||
72 | files = data['video']['files'] | |
73 | ||
74 | entries = [{ | |
75 | 'id': '%s_%d' % (video_id, partnum + 1), | |
76 | 'title': ( | |
77 | title if len(files) == 1 else | |
78 | '%s part %d' % (title, partnum + 1)), | |
79 | 'url': unescapeHTML(f['path']['#text']), | |
80 | 'description': description, | |
81 | 'thumbnail': thumbnail, | |
82 | 'duration': int_or_none(f.get('length', {}).get('#text')), | |
83 | } for partnum, f in enumerate(files)] | |
009a3408 | 84 | |
ca9e7922 | 85 | return { |
aea6e7fc PH |
86 | '_type': 'playlist', |
87 | 'entries': entries, | |
88 | 'title': title, | |
ca9e7922 | 89 | 'id': video_id, |
ca9e7922 | 90 | } |