1 from __future__ import unicode_literals
5 from .common import InfoExtractor
14 from .senateisvp import SenateISVPIE
17 def get_text_attr(d, attr):
18 return d.get(attr, {}).get('#text')
21 class CSpanIE(InfoExtractor):
22 _VALID_URL = r'http://(?:www\.)?c-span\.org/video/\?(?P<id>[0-9a-f]+)'
25 'url': 'http://www.c-span.org/video/?313572-1/HolderonV',
26 'md5': '94b29a4f131ff03d23471dd6f60b6a1d',
30 'title': 'Attorney General Eric Holder on Voting Rights Act Decision',
31 'description': 'Attorney General Eric Holder speaks 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.',
33 'skip': 'Regularly fails on travis, for unknown reasons',
35 'url': 'http://www.c-span.org/video/?c4486943/cspan-international-health-care-models',
36 'md5': '8e5fbfabe6ad0f89f3012a7943c1287b',
40 'title': 'CSPAN - International Health Care Models',
41 'description': 'md5:7a985a2d595dba00af3d9c9f0783c967',
44 'url': 'http://www.c-span.org/video/?318608-1/gm-ignition-switch-recall',
45 'md5': '2ae5051559169baadba13fc35345ae74',
49 'title': 'General Motors Ignition Switch Recall',
51 'description': 'md5:118081aedd24bf1d3b68b3803344e7f3'
54 # Video from senate.gov
55 'url': 'http://www.c-span.org/video/?104517-1/immigration-reforms-needed-protect-skilled-american-workers',
57 'id': 'judiciary031715',
59 'title': 'Immigration Reforms Needed to Protect Skilled American Workers',
63 def _real_extract(self, url):
64 video_id = self._match_id(url)
65 webpage = self._download_webpage(url, video_id)
66 matches = re.search(r'data-(prog|clip)id=\'([0-9]+)\'', webpage)
68 video_type, video_id = matches.groups()
69 if video_type == 'prog':
70 video_type = 'program'
72 senate_isvp_url = SenateISVPIE._search_iframe_url(webpage)
74 title = self._og_search_title(webpage)
75 surl = smuggle_url(senate_isvp_url, {'force_title': title})
76 return self.url_result(surl, 'SenateISVP', video_id, title)
78 data = self._download_json(
79 'http://www.c-span.org/assets/player/ajax-player.php?os=android&html5=%s&id=%s' % (video_type, video_id),
81 if data['@status'] != 'Success':
82 raise ExtractorError('%s said: %s' % (self.IE_NAME, get_text_attr(data, 'error')), expected=True)
84 doc = self._download_xml(
85 'http://www.c-span.org/common/services/flashXml.php?%sid=%s' % (video_type, video_id),
88 description = self._html_search_meta('description', webpage)
90 title = find_xpath_attr(doc, './/string', 'name', 'title').text
91 thumbnail = find_xpath_attr(doc, './/string', 'name', 'poster').text
94 capfile = get_text_attr(data, 'capfile')
97 for partnum, f in enumerate(files):
99 for quality in f['qualities']:
101 'format_id': '%s-%sp' % (get_text_attr(quality, 'bitrate'), get_text_attr(quality, 'height')),
102 'url': unescapeHTML(get_text_attr(quality, 'file')),
103 'height': int_or_none(get_text_attr(quality, 'height')),
104 'tbr': int_or_none(get_text_attr(quality, 'bitrate')),
106 self._sort_formats(formats)
108 'id': '%s_%d' % (video_id, partnum + 1),
110 title if len(files) == 1 else
111 '%s part %d' % (title, partnum + 1)),
113 'description': description,
114 'thumbnail': thumbnail,
115 'duration': int_or_none(get_text_attr(f, 'length')),
119 'ext': determine_ext(capfile, 'dfxp')
121 } if capfile else None,
124 if len(entries) == 1:
125 entry = dict(entries[0])
126 entry['id'] = 'c' + video_id if video_type == 'clip' else video_id
133 'id': 'c' + video_id if video_type == 'clip' else video_id,