[radiocanada] Add fallback for title extraction

This commit is contained in:
Olivier Bilodeau 2017-09-08 10:53:24 -04:00 committed by Sergey M
parent 5113b69124
commit 931edb2ada
1 changed files with 31 additions and 14 deletions

View File

@ -20,7 +20,8 @@ from ..utils import (
class RadioCanadaIE(InfoExtractor): class RadioCanadaIE(InfoExtractor):
IE_NAME = 'radiocanada' IE_NAME = 'radiocanada'
_VALID_URL = r'(?:radiocanada:|https?://ici\.radio-canada\.ca/widgets/mediaconsole/)(?P<app_code>[^:/]+)[:/](?P<id>[0-9]+)' _VALID_URL = r'(?:radiocanada:|https?://ici\.radio-canada\.ca/widgets/mediaconsole/)(?P<app_code>[^:/]+)[:/](?P<id>[0-9]+)'
_TEST = { _TESTS = [
{
'url': 'http://ici.radio-canada.ca/widgets/mediaconsole/medianet/7184272', 'url': 'http://ici.radio-canada.ca/widgets/mediaconsole/medianet/7184272',
'info_dict': { 'info_dict': {
'id': '7184272', 'id': '7184272',
@ -29,11 +30,27 @@ class RadioCanadaIE(InfoExtractor):
'description': 'Images des caméras de surveillance fournies par la GRC montrant le parcours du tireur d\'Ottawa', 'description': 'Images des caméras de surveillance fournies par la GRC montrant le parcours du tireur d\'Ottawa',
'upload_date': '20141023', 'upload_date': '20141023',
}, },
'params': {
# m3u8 download
'skip_download': True,
}
},
{
# empty Title
'url': 'http://ici.radio-canada.ca/widgets/mediaconsole/medianet/7754998/',
'info_dict': {
'id': '7754998',
'ext': 'mp4',
'title': 'letelejournal22h',
'description': 'INTEGRALE WEB 22H-TJ',
'upload_date': '20170720',
},
'params': { 'params': {
# m3u8 download # m3u8 download
'skip_download': True, 'skip_download': True,
}, },
} }
]
def _real_extract(self, url): def _real_extract(self, url):
url, smuggled_data = unsmuggle_url(url, {}) url, smuggled_data = unsmuggle_url(url, {})
@ -145,7 +162,7 @@ class RadioCanadaIE(InfoExtractor):
return { return {
'id': video_id, 'id': video_id,
'title': get_meta('Title'), 'title': get_meta('Title') or get_meta('AV-nomEmission'),
'description': get_meta('Description') or get_meta('ShortDescription'), 'description': get_meta('Description') or get_meta('ShortDescription'),
'thumbnail': get_meta('imageHR') or get_meta('imageMR') or get_meta('imageBR'), 'thumbnail': get_meta('imageHR') or get_meta('imageMR') or get_meta('imageBR'),
'duration': int_or_none(get_meta('length')), 'duration': int_or_none(get_meta('length')),