2 from __future__ import unicode_literals
16 from ..compat import compat_urllib_request
17 from .common import InfoExtractor
20 class DemocracynowIE(InfoExtractor):
21 _VALID_URL = r'https?://(?:www\.)?democracynow.org/?(?P<id>[^\?]*)'
22 IE_NAME = 'democracynow'
24 'url': 'http://www.democracynow.org/shows/2015/7/3',
26 'id': '2015-0703-001',
28 'title': 'July 03, 2015 - Democracy Now!',
29 'description': 'A daily independent global news hour with Amy Goodman & Juan Gonz\xe1lez "What to the Slave is 4th of July?": James Earl Jones Reads Frederick Douglass\u2019 Historic Speech : "This Flag Comes Down Today": Bree Newsome Scales SC Capitol Flagpole, Takes Down Confederate Flag : "We Shall Overcome": Remembering Folk Icon, Activist Pete Seeger in His Own Words & Songs',
30 'uploader': 'Democracy Now',
34 'url': 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree',
36 'id': '2015-0703-001',
38 'title': '"This Flag Comes Down Today": Bree Newsome Scales SC Capitol Flagpole, Takes Down Confederate Flag',
39 'description': 'md5:4d2bc4f0d29f5553c2210a4bc7761a21',
40 'uploader': 'Democracy Now',
46 def _real_extract(self, url):
47 display_id = self._match_id(url)
48 base_host = re.search(r'^(.+?://[^/]+)', url).group(1)
51 webpage = self._download_webpage(url, display_id)
52 re_desc = re.search(r'<meta property=.og:description. content=(["\'])(.+?)\1',webpage,re.DOTALL)
53 description = re_desc.group(2) if re_desc else ''
55 jstr = self._search_regex(r'({.+?"related_video_xml".+?})', webpage, 'json', default=None)
56 js = self._parse_json(jstr, display_id)
60 for key in ('caption_file','.......'):
61 # ....... = pending vtt support that doesn't clobber srt 'chapter_file':
63 if url == '' or url == None:
65 if not re.match(r'^https?://',url):
67 ext = re.search(r'\.([^\.]+)$',url).group(1)
72 for key in ('file', 'audio'):
74 if url == '' or url == None:
76 if not re.match(r'^https?://',url):
78 purl = re.search(r'/(?P<dir>[^/]+)/(?:dn)?(?P<fn>[^/]+?)\.(?P<ext>[^\.\?]+)(?P<hasparams>\?|$)',url)
80 video_id = purl.group('fn')
81 if js.get('start') != None:
82 url += '&' if purl.group('hasparams') == '?' else '?'
83 url = url + 'start='+str(js.get('start'))
85 'format_id': purl.group('dir'),
86 'ext': purl.group('ext'),
89 self._sort_formats(formats)
92 'title': js.get('title'),
93 'description': description,
94 'uploader': 'Democracy Now',
95 # 'thumbnails': thumbnails,
96 'subtitles': subtitles,