[tv5mondeplus] clean description and use stable id

This commit is contained in:
Remita Amine 2017-04-02 00:26:48 +01:00
parent 61e2331ad8
commit dbf70c489f

View File

@ -3,6 +3,7 @@
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import ( from ..utils import (
clean_html,
determine_ext, determine_ext,
extract_attributes, extract_attributes,
get_element_by_class, get_element_by_class,
@ -19,10 +20,10 @@ class TV5MondePlusIE(InfoExtractor):
'url': 'http://www.tv5mondeplus.com/toutes-les-videos/documentaire/tdah-mon-amour-tele-quebec-tdah-mon-amour-ep001-enfants', 'url': 'http://www.tv5mondeplus.com/toutes-les-videos/documentaire/tdah-mon-amour-tele-quebec-tdah-mon-amour-ep001-enfants',
'md5': '12130fc199f020673138a83466542ec6', 'md5': '12130fc199f020673138a83466542ec6',
'info_dict': { 'info_dict': {
'id': '0a774110-dc60-4037-f769-996439514f1f', 'id': 'tdah-mon-amour-tele-quebec-tdah-mon-amour-ep001-enfants',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Tdah, mon amour - Enfants', 'title': 'Tdah, mon amour - Enfants',
'description': 'md5:b65f0cc50e46947e62e5d352e9916cc4', 'description': 'md5:230e3aca23115afcf8006d1bece6df74',
'upload_date': '20170401', 'upload_date': '20170401',
'timestamp': 1491022860, 'timestamp': 1491022860,
} }
@ -65,10 +66,10 @@ def _real_extract(self, url):
self._sort_formats(formats) self._sort_formats(formats)
return { return {
'id': vpl_data.get('data-guid') or display_id, 'id': display_id,
'display_id': display_id, 'display_id': display_id,
'title': title, 'title': title,
'description': get_element_by_class('video-detail__description', webpage), 'description': clean_html(get_element_by_class('video-detail__description', webpage)),
'thumbnail': vpl_data.get('data-image'), 'thumbnail': vpl_data.get('data-image'),
'duration': int_or_none(vpl_data.get('data-duration')) or parse_duration(self._html_search_meta('duration', webpage)), 'duration': int_or_none(vpl_data.get('data-duration')) or parse_duration(self._html_search_meta('duration', webpage)),
'timestamp': parse_iso8601(self._html_search_meta('uploadDate', webpage)), 'timestamp': parse_iso8601(self._html_search_meta('uploadDate', webpage)),