[update] 代理校验修改

abop · Apr 3, 2018 · ae09f0d · ae09f0d
1 parent 24a9f09
commit ae09f0d
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 24 deletions.
diff --git a/DB/SsdbClient.py b/DB/SsdbClient.py
@@ -63,7 +63,7 @@ def put(self, proxy, num=1):
         :param num:
         :return:
         """
-        data = self.__conn.hincrby(self.name, proxy, num)
+        data = self.__conn.hset(self.name, proxy, num)
         return data
 
     def delete(self, key):

diff --git a/Schedule/ProxyCheck.py b/Schedule/ProxyCheck.py
@@ -22,7 +22,7 @@
 from Manager.ProxyManager import ProxyManager
 from Util.LogHandler import LogHandler
 
-FAIL_COUNT = 1  # 校验失败次数， 超过次数删除代理
+FAIL_COUNT = 2  # 校验失败次数， 超过次数删除代理
 
 
 class ProxyCheck(ProxyManager, Thread):
@@ -34,27 +34,21 @@ def __init__(self):
     def run(self):
         self.db.changeTable(self.useful_proxy_queue)
         while True:
-            proxy_item = self.db.pop()
-            while proxy_item:
-                proxy = proxy_item.get('proxy')
-                counter = proxy_item.get('value', 1)
+            for proxy, count in self.db.getAll().items():
                 if validUsefulProxy(proxy):
-                    # 验证通过计数器加1
-                    if counter and int(counter) < 1:
-                        self.db.put(proxy, num=int(counter) + 1)
+                    # 验证通过计数器减1
+                    if count and int(count) > 0:
+                        self.db.put(proxy, num=int(count) - 1)
                     else:
-                        self.db.put(proxy)
+                        pass
                     self.log.info('ProxyCheck: {} validation pass'.format(proxy))
                 else:
                     self.log.info('ProxyCheck: {} validation fail'.format(proxy))
-                    # 验证失败，计数器减1
-                    if counter and int(counter) <= FAIL_COUNT:
+                    if count and int(count) > FAIL_COUNT:
                         self.log.info('ProxyCheck: {} fail too many, delete!'.format(proxy))
                         self.db.delete(proxy)
                     else:
-                        self.db.put(proxy, num=int(counter) - 1)
-
-                proxy_item = self.db.pop()
+                        self.db.put(proxy, num=int(count) + 1)
             sleep(60 * 5)
 
 

diff --git a/Schedule/ProxyRefreshSchedule.py b/Schedule/ProxyRefreshSchedule.py
@@ -95,9 +95,9 @@ def main(process_num=30):
 
 def run():
     main()
-    sched = BlockingScheduler()
-    sched.add_job(main, 'interval', minutes=10)  # 每10分钟抓取一次
-    sched.start()
+    sch = BlockingScheduler()
+    sch.add_job(main, 'interval', minutes=10)  # 每10分钟抓取一次
+    sch.start()
 
 
 if __name__ == '__main__':

diff --git a/Util/utilFunction.py b/Util/utilFunction.py
@@ -18,7 +18,7 @@
 from Util.LogHandler import LogHandler
 from Util.WebRequest import WebRequest
 
-logger = LogHandler(__name__, stream=False)
+# logger = LogHandler(__name__, stream=False)
 
 
 # noinspection PyPep8Naming
@@ -27,8 +27,9 @@ def decorate(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except Exception as e:
-            logger.info(u"sorry, 抓取出错。错误原因:")
-            logger.info(e)
+            pass
+            # logger.info(u"sorry, 抓取出错。错误原因:")
+            # logger.info(e)
 
     return decorate
 
@@ -98,10 +99,10 @@ def validUsefulProxy(proxy):
     proxies = {"http": "http://{proxy}".format(proxy=proxy)}
     try:
         # 超过20秒的代理就不要了
-        r = requests.get('http://httpbin.org/ip', proxies=proxies, timeout=20, verify=False)
+        r = requests.get('http://httpbin.org/ip', proxies=proxies, timeout=10, verify=False)
         if r.status_code == 200:
-            logger.info('%s is ok' % proxy)
+            # logger.info('%s is ok' % proxy)
             return True
     except Exception as e:
-        logger.debug(e)
+        # logger.error(str(e))
         return False