def gtaskManager(self,urls,extractSearchResults,proxy_flag = 0,ua_flag = 0):
task_log = None
gtaskpool.setlogging(logging.INFO,task_log)
purl1 = ["http://192.168.120.17:8014/proxy/get_http_proxy_list"]
uurl1 = "http://192.168.120.17:8014/proxy/get_useragent_list"
limited_urls = [
('^https{0,1}://', 0)
]
global proxymgr
if proxy_flag == 1:
proxymgr = ProxyManager(get_http_proxies, limited_urls,
{'refresh': True, 'interval': 30 * 60, 'delay': 8 * 60}, *purl1)
else:
proxymgr = None
print proxymgr
global useragents
if ua_flag == 1:
useragents = get_useragents(uurl1)
else:
useragents = [None]
if useragents == []:
useragents = [None]
gtaskpool.runtasks(AccessUrls.taskGenerator(self,urls,extractSearchResults))
def gtaskmanager(self,engine_type):
#task_log = 'task_log.log'
task_log = None
gtaskpool.setlogging(logging.INFO,task_log)
purl1 = ["http://192.168.120.185:5500/get_google_http_proxy_list"]
uurl1 = "http://192.168.120.17:8014/proxy/get_useragent_list"
limited_urls = [
('^https://search\.disconnect\.me', 1)
]
global proxymgr
proxymgr = ProxyManager(get_http_proxies, limited_urls,
{'refresh': True, 'interval': 30 * 60, 'delay': 8 * 60}, *purl1)
global useragents
useragents = get_useragents(uurl1)
if useragents == []:
useragents = [None]
gtaskpool.runtasks(self.task_generator(self,engine_type))
def gtaskmanager(self, engine_type):
#task_log = 'task_log.log'
task_log = None
gtaskpool.setlogging(logging.INFO, task_log)
purl1 = ["http://192.168.120.185:5500/get_google_http_proxy_list"]
uurl1 = "http://192.168.120.17:8014/proxy/get_useragent_list"
limited_urls = [('^https://search\.disconnect\.me', 1)]
global proxymgr
proxymgr = ProxyManager(get_http_proxies, limited_urls, {
'refresh': True,
'interval': 30 * 60,
'delay': 8 * 60
}, *purl1)
global useragents
useragents = get_useragents(uurl1)
if useragents == []:
useragents = [None]
gtaskpool.runtasks(self.task_generator(self, engine_type))
gtaskpool.setlogging(logging.INFO)
purl1 = ["http://192.168.120.17:8014/proxy/get_http_proxy_list"]
purl2 = ["http://192.168.1.14:5500/get_http_proxy_list"]
uurl1 = "http://192.168.120.17:8014/proxy/get_useragent_list"
uurl2 = "http://192.168.1.14:5500/get_useragent_list"
# Create a ProxyManager if you need
limited_urls = [("^http://www\.baidu\.com/s\?wd=apple&pn=\d+$", 1)]
proxymgr = ProxyManager(
get_http_proxies, limited_urls, {"refresh": True, "interval": 30 * 60, "delay": 8 * 60}, *purl2
)
# Or if you don't want refresh proxies periodcally
# proxymgr = ProxyManager(get_http_proxies, *purl2, limited_urls, \
# {'refresh': False}, *purl2)
# A useragent list for http request if you need
useragents = get_useragents(uurl2)
if useragents == []:
useragents = [None]
fresult = open("result.txt", "w")
fleft = open("left.txt", "w")
# Optional args:
# @max_ongoing_tasks (default to 1000)
gtaskpool.runtasks(task_generator())
fresult.close()
fleft.close()
while trycnt != max_try:
res = task()
res['try_idx'] = trycnt + 1
if trycnt+1 == max_try or res['finish']:
res['last_try'] = True
else:
res['last_try'] = False
log_task_result(res, task_log)
if res['finish']:
return
trycnt += 1
def log_task_result(result, filehandle):
result['ts'] = str(datetime.now())
jstr = json.dumps(result, ensure_ascii=False).encode('utf-8')
filehandle.write(jstr + "\n")
def runtasks(task_generator, task_log, max_try=10):
def gen_task():
while True:
try:
task = task_generator.next()
except StopIteration, e:
return
yield gtaskpool.Task(retry_task, [task, task_log, max_try])
gtaskpool.runtasks(gen_task())
while trycnt != max_try:
res = task()
res['try_idx'] = trycnt + 1
if trycnt + 1 == max_try or res['finish']:
res['last_try'] = True
else:
res['last_try'] = False
log_task_result(res, task_log)
if res['finish']:
return
trycnt += 1
def log_task_result(result, filehandle):
result['ts'] = str(datetime.now())
jstr = json.dumps(result, ensure_ascii=False).encode('utf-8')
filehandle.write(jstr + "\n")
def runtasks(task_generator, task_log, max_try=10):
def gen_task():
while True:
try:
task = task_generator.next()
except StopIteration, e:
return
yield gtaskpool.Task(retry_task, [task, task_log, max_try])
gtaskpool.runtasks(gen_task())
#!/usr/bin/env python
# encoding: utf-8
import gtaskpool
import requests
import logging
def task(n1, n2):
logging.info("task(%s, %s): called", n1, n2)
r = requests.get("http://www.baidu.com")
print "task(%s, %s): response (len=%s): %s..." % \
(n1, n2, len(r.text), r.text[:min(100, len(r.text))])
logging.info("task(%s, %s): finished", n1, n2)
def task_generator():
task_num = 10
for i in xrange(1, task_num + 1):
yield gtaskpool.Task(task, [i, i])
if __name__ == "__main__":
gtaskpool.setlogging(logging.INFO)
gtaskpool.runtasks(task_generator())