def trytry():
PyMongoUtil.clean()
MemcacheUtil.clean()
SpiderBloomFilter()
queue = PyPool.get_queue()
lock = PyPool.get_lock()
listener = MyListener()
r = Regex("[a-z0-9\-\._]+@[a-z0-9\-\.]+\.[a-z]{2,4}[:,\|]*.*")
s = SpiderStrategy("http://www.leakedin.com/tag/emailpassword-dump/", 2, is_out=False, pattern=None, mode=r)
Spider(s).get_all_words(queue, lock)
listener.listen(lock, queue)
WordCount.calc_count()
return
def trytry():
PyMongoUtil.clean()
MemcacheUtil.clean()
SpiderBloomFilter()
queue = PyPool.get_queue()
lock = PyPool.get_lock()
listener = MyListener()
r = Regex("[a-z0-9\-\._]+@[a-z0-9\-\.]+\.[a-z]{2,4}[:,\|]*.*")
s = SpiderStrategy("http://www.leakedin.com/tag/emailpassword-dump/",
2,
is_out=False,
pattern=None,
mode=r)
Spider(s).get_all_words(queue, lock)
listener.listen(lock, queue)
WordCount.calc_count()
return
if not md.isdigit() or int(md) <= 0 or int(md) >= 5:
err()
continue
else:
md = int(md)
if md == 4:
regex = raw_input("please enter regex string:\n")
mode = Regex(regex)
else:
mode = Language.get_enum(md)
break
s = SpiderStrategy(url, int(depth), isOut, None, mode)
Spider(s).get_all_words(queue, lock)
listener.listen(lock, queue)
WordCount.calc_count()
while True:
query_count = raw_input("Spider Finished. please enter how many statistics you want.[All]\n")
file_path = raw_input("Please input the path you want to put result.csv.\n")
if not path.isdir(file_path):
print("Please enter the right path.\n")
continue
if query_count == None or query_count == '':
ExcelUtil.writefile(file_path, PyMongoUtil.query_result())
elif not query_count.isdigit():
print("Please enter the right query count.\n")
continue
else:
ExcelUtil.writefile(file_path, PyMongoUtil.query_result(int(query_count)))
print("Generate success. Please check the result! Bye\n")