github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[zerochan] fix metadata extraction

Browse Source 
author, path, tags
This commit is contained in:
Mike Fährmann
2023-11-24 21:21:14 +01:00
parent f9dac43be9
commit adc3aa0b77
1 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
gallery_dl/extractor/zerochan.py
View File

@@ -63,14 +63,14 @@ class ZerochanExtractor(BooruExtractor):
data = {
"id" : text.parse_int(entry_id),
"author" : extr('"author": "', '"'),
"author" : text.parse_unicode_escapes(extr(' "name": "', '"')),
"file_url": extr('"contentUrl": "', '"'),
"date" : text.parse_datetime(extr('"datePublished": "', '"')),
"width" : text.parse_int(extr('"width": "', ' ')),
"height" : text.parse_int(extr('"height": "', ' ')),
"size" : text.parse_bytes(extr('"contentSize": "', 'B')),
"path" : text.split_html(extr(
'class="breadcrumbs', '</p>'))[2:],
'class="breadcrumbs', '</nav>'))[2:],
"uploader": extr('href="/user/', '"'),
"tags" : extr('<ul id="tags"', '</ul>'),
"source" : extr('<h2>Source</h2>', '</p><h2>').rpartition(
@@ -80,9 +80,9 @@ class ZerochanExtractor(BooruExtractor):
html = data["tags"]
tags = data["tags"] = []
for tag in html.split("<li class=")[1:]:
category = text.extr(tag, 'alt="', '"')
name = text.extr(tag, ">-->", "</a>")
tags.append(category + ":" + name.strip())
category = text.extr(tag, 'data-type="', '"')
name = text.extr(tag, 'data-tag="', '"')
tags.append(category.capitalize() + ":" + name)
return data