Extraction: Use accessibility data to get timestamp and to get views for recommended videos

This commit is contained in:
James Taylor 2019-12-18 20:53:11 -08:00
parent f6bf5213a5
commit 004e14a538

View File

@ -386,6 +386,16 @@ def extract_item_info(item, additional_info={}):
if primary_type == 'video':
info['id'] = item.get('videoId')
info['view_count'] = extract_int(item.get('viewCountText'))
# dig into accessibility data to get view_count for videos marked as recommended, and to get time_published
accessibility_label = deep_get(item, 'title', 'accessibility', 'accessibilityData', 'label', default='')
timestamp = re.search(r'(\d+ \w+ ago)', accessibility_label)
if timestamp:
conservative_update(info, 'time_published', timestamp.group(1))
view_count = re.search(r'(\d+) views', accessibility_label.replace(',', ''))
if view_count:
conservative_update(info, 'view_count', int(view_count.group(1)))
if info['view_count']:
info['approx_view_count'] = '{:,}'.format(info['view_count'])
else: