midgard/youtube-dl
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[thisamericanlife] Improve and simplify

Browse source
This commit is contained in:
Sergey M․ 2015-07-04 05:42:53 +06:00
parent 95506e37af
commit e56a4c9e9b
1 changed files with 22 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

42
youtube_dl/extractor/thisamericanlife.py
View file

@ -1,38 +1,40 @@
# coding: utf-8
from __future__ import unicode_literals from __future__ import unicode_literals
from .common import InfoExtractor from .common import InfoExtractor
class ThisAmericanLifeIE(InfoExtractor): class ThisAmericanLifeIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?thisamericanlife\.org/radio-archives/episode/(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.)?thisamericanlife\.org/(?:radio-archives/episode/|play_full\.php\?play=)(?P<id>\d+)'
_TESTS = [{
_TEST = {
'url': 'http://www.thisamericanlife.org/radio-archives/episode/487/harper-high-school-part-one', 'url': 'http://www.thisamericanlife.org/radio-archives/episode/487/harper-high-school-part-one',
'md5': '5cda28076c9f9d1fd0b0f5cff5959948', 'md5': '8f7d2da8926298fdfca2ee37764c11ce',
'info_dict': { 'info_dict': {
'id': '487', 'id': '487',
'ext': 'm4a',
'title': '487: Harper High School, Part One', 'title': '487: Harper High School, Part One',
'url' : 'http://stream.thisamericanlife.org/487/stream/487_64k.m3u8', 'description': 'md5:ee40bdf3fb96174a9027f76dbecea655',
'ext': 'aac', 'thumbnail': 're:^https?://.*\.jpg$',
'thumbnail': 'http://www.thisamericanlife.org/sites/default/files/imagecache/large_square/episodes/487_lg_2.jpg',
'description': 'We spent five months at Harper High School in Chicago, where last year alone 29 current and recent students were shot. 29. We went to get a sense of what it means to live in the midst of all this gun violence, how teens and adults navigate a world of funerals and Homecoming dances.',
}, },
'params': { }, {
# m38u download 'url': 'http://www.thisamericanlife.org/play_full.php?play=487',
'skip_download': True, 'only_matching': True,
}, }]
}
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
webpage = self._download_webpage(
'http://www.thisamericanlife.org/radio-archives/episode/%s' % video_id, video_id)
return { return {
'id': video_id, 'id': video_id,
'title': self._html_search_regex(r'<meta property="twitter:title" content="(.*?)"', webpage, 'title'), 'url': 'http://stream.thisamericanlife.org/{0}/stream/{0}_64k.m3u8'.format(video_id),
'url': 'http://stream.thisamericanlife.org/' + video_id + '/stream/' + video_id + '_64k.m3u8', 'protocol': 'm3u8_native',
'ext': 'aac', 'ext': 'm4a',
'thumbnail': self._html_search_regex(r'<meta property="og:image" content="(.*?)"', webpage, 'thumbnail'), 'acodec': 'aac',
'description': self._html_search_regex(r'<meta name="description" content="(.*?)"', webpage, 'description'), 'vcodec': 'none',
'abr': 64,
'title': self._html_search_meta(r'twitter:title', webpage, 'title', fatal=True),
'description': self._html_search_meta(r'description', webpage, 'description'),
'thumbnail': self._og_search_thumbnail(webpage),
} }