[postimg] fix extraction (#8505)

This commit is contained in:
Mike Fährmann
2025-11-05 11:15:05 +01:00
parent 617bd7a657
commit 9b03990be0
2 changed files with 26 additions and 4 deletions

View File

@@ -320,8 +320,8 @@ class PostimgImageExtractor(ImagehostImageExtractor):
def get_info(self, page):
pos = page.index(' id="download"')
url , pos = text.rextract(page, ' href="', '"', pos)
filename, pos = text.extract(page, 'class="imagename">', '<', pos)
return url, text.unescape(filename)
filename, pos = text.extract(page, ' class="my-4">', '<', pos)
return url, text.unescape(filename) if filename else None
class PostimgGalleryExtractor(ImagehostImageExtractor):
@@ -335,8 +335,16 @@ class PostimgGalleryExtractor(ImagehostImageExtractor):
def items(self):
page = self.request(self.page_url).text
data = {"_extractor": PostimgImageExtractor}
for url in text.extract_iter(page, ' class="thumb"><a href="', '"'):
title = text.extr(
page, 'property="og:title" content="', ' — Postimages"')
data = {
"_extractor" : PostimgImageExtractor,
"gallery_title": text.unescape(title),
}
for token in text.extract_iter(page, 'data-image="', '"'):
url = f"{self.root}/{token}"
yield Message.Queue, url, data