[exhentai] metadata consistency
This commit is contained in:
parent
fea3be0aed
commit
a017cd9e63
@ -60,9 +60,12 @@ class ExhentaiGalleryExtractor(Extractor):
|
|||||||
image.update(data)
|
image.update(data)
|
||||||
image["num"] = num
|
image["num"] = num
|
||||||
text.nameext_from_url(image["url"], image)
|
text.nameext_from_url(image["url"], image)
|
||||||
if "/fullimg.php" in image[urlkey]:
|
url = image[urlkey]
|
||||||
|
del image["url"]
|
||||||
|
del image["origurl"]
|
||||||
|
if "/fullimg.php" in url:
|
||||||
self.wait((1, 2))
|
self.wait((1, 2))
|
||||||
yield Message.Url, image[urlkey], image
|
yield Message.Url, url, image
|
||||||
|
|
||||||
def get_job_metadata(self, page):
|
def get_job_metadata(self, page):
|
||||||
"""Collect metadata for extractor-job"""
|
"""Collect metadata for extractor-job"""
|
||||||
@ -100,24 +103,26 @@ class ExhentaiGalleryExtractor(Extractor):
|
|||||||
("url" , '<img id="img" src="', '"'),
|
("url" , '<img id="img" src="', '"'),
|
||||||
("title" , '<div id="i4"><div>', ' :: '),
|
("title" , '<div id="i4"><div>', ' :: '),
|
||||||
("origurl" , 'https://exhentai.org/fullimg.php', '"'),
|
("origurl" , 'https://exhentai.org/fullimg.php', '"'),
|
||||||
("gid" , 'var gid=', ';'),
|
|
||||||
("startkey" , 'var startkey="', '";'),
|
("startkey" , 'var startkey="', '";'),
|
||||||
("showkey" , 'var showkey="', '";'),
|
("showkey" , 'var showkey="', '";'),
|
||||||
))
|
))
|
||||||
data["imgkey"] = data["startkey"]
|
data["imgkey"] = data["startkey"]
|
||||||
|
|
||||||
|
request = {
|
||||||
|
"method" : "showpage",
|
||||||
|
"page" : 2,
|
||||||
|
"gid" : int(self.gid),
|
||||||
|
"imgkey" : data["imgkey-next"],
|
||||||
|
"showkey": data["showkey"],
|
||||||
|
}
|
||||||
|
del data["showkey"]
|
||||||
|
|
||||||
if data["origurl"]:
|
if data["origurl"]:
|
||||||
data["origurl"] = "https://exhentai.org/fullimg.php" + text.unescape(data["origurl"])
|
data["origurl"] = "https://exhentai.org/fullimg.php" + text.unescape(data["origurl"])
|
||||||
else:
|
else:
|
||||||
data["origurl"] = data["url"]
|
data["origurl"] = data["url"]
|
||||||
yield data
|
yield data
|
||||||
|
|
||||||
request = {
|
|
||||||
"method" : "showpage",
|
|
||||||
"page" : 2,
|
|
||||||
"gid" : int(data["gid"]),
|
|
||||||
"imgkey" : data["imgkey-next"],
|
|
||||||
"showkey": data["showkey"],
|
|
||||||
}
|
|
||||||
while True:
|
while True:
|
||||||
if data["imgkey"] == data["imgkey-next"]:
|
if data["imgkey"] == data["imgkey-next"]:
|
||||||
return
|
return
|
||||||
|
Loading…
x
Reference in New Issue
Block a user