use int_or_none, check if attrib exists, remove thumbnail

This commit is contained in:
kennell 2015-10-18 20:47:42 +02:00
parent 264b23e1a4
commit 8cc83d301d
1 changed files with 5 additions and 8 deletions

View File

@ -73,19 +73,17 @@ def extract_from_xml_url(ie, video_id, xml_url):
def xml_to_thumbnails(fnode): def xml_to_thumbnails(fnode):
thumbnails = list() thumbnails = list()
for node in fnode: for node in fnode:
thumbnail = {'url': node.text}
if 'key' in node.attrib:
width_x_height = node.attrib['key'] width_x_height = node.attrib['key']
thumbnail = { thumbnail['width'] = int_or_none(width_x_height.split('x')[0])
'url': node.text, thumbnail['height'] = int_or_none(width_x_height.split('x')[1])
'width': int(width_x_height.split('x')[0]),
'height': int(width_x_height.split('x')[1])
}
thumbnails.append(thumbnail) thumbnails.append(thumbnail)
return thumbnails return thumbnails
thumbnail_nodes = doc.findall('.//teaserimages/teaserimage') thumbnail_nodes = doc.findall('.//teaserimages/teaserimage')
thumbnails = xml_to_thumbnails(thumbnail_nodes) thumbnails = xml_to_thumbnails(thumbnail_nodes)
thumbnail = thumbnails[-1]['url']
format_nodes = doc.findall('.//formitaeten/formitaet') format_nodes = doc.findall('.//formitaeten/formitaet')
formats = list(filter( formats = list(filter(
@ -98,7 +96,6 @@ def extract_from_xml_url(ie, video_id, xml_url):
'title': title, 'title': title,
'description': description, 'description': description,
'duration': duration, 'duration': duration,
'thumbnail': thumbnail,
'thumbnails': thumbnails, 'thumbnails': thumbnails,
'uploader': uploader, 'uploader': uploader,
'uploader_id': uploader_id, 'uploader_id': uploader_id,