lastId has to be string, cache kg files for 24 hours
This commit is contained in:
parent
8e2565ccf8
commit
c0d9627e69
1 changed files with 3 additions and 3 deletions
|
@ -78,6 +78,7 @@ def getId(url):
|
|||
return url.split("=")[-1]
|
||||
|
||||
def getIds(lastId = 20):
|
||||
lastId = '%s' % lastId
|
||||
ids = []
|
||||
page = 0
|
||||
while True:
|
||||
|
@ -86,14 +87,13 @@ def getIds(lastId = 20):
|
|||
ids.append(id)
|
||||
if lastId in ids:
|
||||
break
|
||||
print page, len(ids)
|
||||
page += 1
|
||||
return map(lambda id: str(id), sorted(map(lambda id: int(id), set(ids))))
|
||||
|
||||
def getIdsByPage(page):
|
||||
ids = []
|
||||
url = 'http://karagarga.net/browse.php?page=%s&cat=1&sort=added&d=DESC' % page
|
||||
html = getUrlUnicode(url, timeout = -1)
|
||||
html = getUrlUnicode(url, timeout = 23*60*60) #get new ids once per day
|
||||
strings = html.split('<td width="42" style="padding:0px;">')
|
||||
strings.pop(0)
|
||||
for string in strings:
|
||||
|
@ -111,4 +111,4 @@ def parseTable(html, title):
|
|||
|
||||
if __name__ == "__main__":
|
||||
print getIds("79317")
|
||||
print getData("79317")
|
||||
print getData("79317")
|
||||
|
|
Loading…
Reference in a new issue