[tagesschau] Add support for width/height

This commit is contained in:
Philipp Hagemeister 2014-06-07 15:14:20 +02:00
parent ebdb64d605
commit 4a5b4d34dc

View file

@ -7,7 +7,7 @@ from .common import InfoExtractor
class TagesschauIE(InfoExtractor): class TagesschauIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tagesschau\.de/multimedia/video/video(?P<id>-?\d+)\.html' _VALID_URL = r'https?://(?:www\.)?tagesschau\.de/multimedia/video/video(?P<id>-?[0-9]+)\.html'
_TESTS = [{ _TESTS = [{
'url': 'http://www.tagesschau.de/multimedia/video/video1399128.html', 'url': 'http://www.tagesschau.de/multimedia/video/video1399128.html',
@ -25,12 +25,18 @@ class TagesschauIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '196', 'id': '196',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Ukraine-Konflikt: Klitschko in Kiew als B\xfcrgermeister vereidigt', 'title': 'Ukraine-Konflikt: Klitschko in Kiew als Bürgermeister vereidigt',
'description': 'md5:f22e4af75821d174fa6c977349682691', 'description': 'md5:f22e4af75821d174fa6c977349682691',
'thumbnail': 're:http://.*\.jpg', 'thumbnail': 're:http://.*\.jpg',
}, },
}] }]
_FORMATS = {
's': {'width': 256, 'height': 144, 'quality': 1},
'm': {'width': 512, 'height': 288, 'quality': 2},
'l': {'width': 960, 'height': 544, 'quality': 3},
}
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id') video_id = mobj.group('id')
@ -43,31 +49,22 @@ class TagesschauIE(InfoExtractor):
webpage = self._download_webpage(url, display_id) webpage = self._download_webpage(url, display_id)
playerpage = self._download_webpage( playerpage = self._download_webpage(
'http://www.tagesschau.de/multimedia/video/video%s~player_autoplay-true.html' % video_id, display_id, 'Downloading player page') 'http://www.tagesschau.de/multimedia/video/video%s~player_autoplay-true.html' % video_id,
display_id, 'Downloading player page')
medias = re.findall(r'"(http://media.+?)", type:"video/(.+?)", quality:"(.+?)"', playerpage) medias = re.findall(
r'"(http://media.+?)", type:"video/(.+?)", quality:"(.+?)"',
playerpage)
formats = [] formats = []
for url, ext, res in medias: for url, ext, res in medias:
f = {
if res == 's': 'format_id': res + '_' + ext,
res = 'small'
quality = 0
elif res == 'm':
res = 'medium'
quality = 1
elif res == 'l':
res = 'large'
quality = 2
else:
quality = 0
formats.append({
'format_id': res+'_'+ext,
'url': url, 'url': url,
'quality': quality,
'ext': ext, 'ext': ext,
}) }
f.update(self._FORMATS.get(res, {}))
formats.append(f)
self._sort_formats(formats) self._sort_formats(formats)
@ -76,7 +73,7 @@ class TagesschauIE(InfoExtractor):
return { return {
'id': display_id, 'id': display_id,
'title': self._og_search_title(webpage).strip(), 'title': self._og_search_title(webpage).strip(),
'thumbnail': 'http://www.tagesschau.de'+thumbnail, 'thumbnail': 'http://www.tagesschau.de' + thumbnail,
'formats': formats, 'formats': formats,
'description': self._og_search_description(webpage).strip(), 'description': self._og_search_description(webpage).strip(),
} }