[readcomiconline] fix extraction

This commit is contained in:
enduser420
2025-07-07 02:32:23 +05:30
parent d349c248c0
commit 8c1628ea4e
2 changed files with 9 additions and 10 deletions

View File

@@ -80,21 +80,20 @@ class ReadcomiconlineIssueExtractor(ReadcomiconlineBase, ChapterExtractor):
def images(self, page):
results = []
referer = {"_http_headers": {"Referer": self.page_url}}
root = text.extr(page, "return baeu(l, '", "'")
root, pos = text.extract(page, "return baeu(l, '", "'")
_ , pos = text.extract(page, "var pth = '", "", pos)
var , pos = text.extract(page, "var ", "= '", pos)
replacements = re.findall(
r"l = l\.replace\(/([^/]+)/g, [\"']([^\"']*)", page)
for block in page.split("\t\tpht = '")[1:]:
pth = text.extr(block, "", "'")
for path in page.split(var)[2:]:
path = text.extr(path, "= '", "'")
for needle, repl in re.findall(
r"pth = pth\.replace\(/([^/]+)/g, [\"']([^\"']*)", block):
pth = pth.replace(needle, repl)
for needle, repl in replacements:
pth = pth.replace(needle, repl)
path = path.replace(needle, repl)
results.append((baeu(pth, root), referer))
results.append((baeu(path, root), referer))
return results

View File

@@ -22,7 +22,7 @@ __tests__ = (
"#category": ("", "readcomiconline", "comic"),
"#class" : readcomiconline.ReadcomiconlineComicExtractor,
"#sha1_url" : "74eb8b9504b4084fcc9367b341300b2c52260918",
"#sha1_metadata": "3986248e4458fa44a201ec073c3684917f48ee0c",
"#sha1_metadata": "574051aaf7a5c92dafed9e94baa40a1a93db5c90",
},
{
@@ -30,7 +30,7 @@ __tests__ = (
"#category": ("", "readcomiconline", "comic"),
"#class" : readcomiconline.ReadcomiconlineComicExtractor,
"#sha1_url" : "2f66a467a772df4d4592e97a059ddbc3e8991799",
"#sha1_metadata": "f5ba5246cd787bb750924d9690cb1549199bd516",
"#sha1_metadata": "9563a19454e1b4e0da5b7a28112bf00a3e8069a8",
},
)