[mangoxo] fix metadata extraction
This commit is contained in:
@@ -122,18 +122,18 @@ class MangoxoAlbumExtractor(MangoxoExtractor):
|
|||||||
def metadata(self, page):
|
def metadata(self, page):
|
||||||
"""Return general metadata"""
|
"""Return general metadata"""
|
||||||
extr = text.extract_from(page)
|
extr = text.extract_from(page)
|
||||||
title = extr('<title>', '</title>')
|
title = extr('<img id="cover-img" alt="', '"')
|
||||||
count = extr('id="pic-count">', '<')
|
cid = extr('href="https://www.mangoxo.com/user/', '"')
|
||||||
cid = extr('<img alt="', '"')
|
cname = extr('<img alt="', '"')
|
||||||
cover = extr(' src="', '"')
|
cover = extr(' src="', '"')
|
||||||
cname = extr('target="_blank">', '<')
|
count = extr('id="pic-count">', '<')
|
||||||
date = extr('</i>', '<')
|
date = extr('class="fa fa-calendar"></i>', '<')
|
||||||
descr = extr('<pre>', '</pre>')
|
descr = extr('<pre>', '</pre>')
|
||||||
|
|
||||||
return {
|
return {
|
||||||
"channel": {
|
"channel": {
|
||||||
"id": cid,
|
"id": cid,
|
||||||
"name": text.unescape(cname.strip()),
|
"name": text.unescape(cname),
|
||||||
"cover": cover,
|
"cover": cover,
|
||||||
},
|
},
|
||||||
"album": {
|
"album": {
|
||||||
|
|||||||
Reference in New Issue
Block a user