[rte] Make more robust and add a new testcase (#4015)

This commit is contained in:
Philipp Hagemeister 2015-01-10 17:57:21 +01:00
parent 121c09c7be
commit ea1d5bdcdd
1 changed files with 21 additions and 17 deletions

View File

@ -9,29 +9,33 @@ from ..utils import (
class RteIE(InfoExtractor):
_VALID_URL = r'http?://(?:www\.)?rte\.ie/player/in/show/(?P<id>[0-9]+)/'
_VALID_URL = r'http?://(?:www\.)?rte\.ie/player/[^/]{2,3}/show/(?P<id>[0-9]+)/'
_TEST = {
'url': 'http://www.rte.ie/player/in/show/10336191/',
'url': 'http://www.rte.ie/player/de/show/10363114/',
'info_dict': {
'id': '10336191',
'id': '10363114',
'ext': 'mp4',
'title': 'Nine News',
'title': 'One News',
'thumbnail': 're:^https?://.*\.jpg$',
'description': 'The One O\'Clock News followed by Weather.',
'duration': 1622.963,
'duration': 436.844,
},
'params': {
'skip_download': 'f4m fails with --test atm'
}
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
title = self._og_search_title(webpage)
description = self._html_search_meta('description', webpage, 'description')
duration = float_or_none(self._html_search_meta(
'duration', webpage, 'duration', fatal=False), 1000)
description = self._search_regex(r'<meta name="description" content="(.*?)" />', webpage, 'description')
duration = float_or_none(self._html_search_meta('duration', webpage, 'duration'), 1000)
thumbnail_id = self._search_regex(r'<meta name="thumbnail" content="uri:irus:(.*?)" />', webpage, 'thumbnail')
thumbnail_id = self._search_regex(
r'<meta name="thumbnail" content="uri:irus:(.*?)" />', webpage, 'thumbnail')
thumbnail = 'http://img.rasset.ie/' + thumbnail_id + '.jpg'
feeds_url = self._html_search_meta("feeds-prefix", webpage, 'feeds url') + video_id