1
1
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-05 03:02:56 +01:00
youtube-dl/youtube_dl/extractor/faz.py
2015-06-22 20:11:15 +06:00

68 lines
2.4 KiB
Python

# encoding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
class FazIE(InfoExtractor):
IE_NAME = 'faz.net'
_VALID_URL = r'https?://(?:www\.)?faz\.net/(?:[^/]+/)*.*?-(?P<id>\d+)\.html'
_TESTS = [{
'url': 'http://www.faz.net/multimedia/videos/stockholm-chemie-nobelpreis-fuer-drei-amerikanische-forscher-12610585.html',
'info_dict': {
'id': '12610585',
'ext': 'mp4',
'title': 'Stockholm: Chemie-Nobelpreis für drei amerikanische Forscher',
'description': 'md5:1453fbf9a0d041d985a47306192ea253',
},
}, {
'url': 'http://www.faz.net/aktuell/politik/berlin-gabriel-besteht-zerreissprobe-ueber-datenspeicherung-13659345.html',
'only_matching': True,
}, {
'url': 'http://www.faz.net/berlin-gabriel-besteht-zerreissprobe-ueber-datenspeicherung-13659345.html',
'only_matching': True,
}, {
'url': 'http://www.faz.net/-13659345.html',
'only_matching': True,
}, {
'url': 'http://www.faz.net/aktuell/politik/-13659345.html',
'only_matching': True,
}, {
'url': 'http://www.faz.net/foobarblafasel-13659345.html',
'only_matching': True,
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
config_xml_url = self._search_regex(
r'writeFLV\(\'(.+?)\',', webpage, 'config xml url')
config = self._download_xml(
config_xml_url, video_id, 'Downloading config xml')
encodings = config.find('ENCODINGS')
formats = []
for pref, code in enumerate(['LOW', 'HIGH', 'HQ']):
encoding = encodings.find(code)
if encoding is None:
continue
encoding_url = encoding.find('FILENAME').text
formats.append({
'url': encoding_url,
'format_id': code.lower(),
'quality': pref,
})
self._sort_formats(formats)
descr = self._html_search_regex(
r'<p class="Content Copy">(.*?)</p>', webpage, 'description', fatal=False)
return {
'id': video_id,
'title': self._og_search_title(webpage),
'formats': formats,
'description': descr,
'thumbnail': config.find('STILL/STILL_BIG').text,
}