修改www.goubanjia.com抓取IP的代码

ProxyGetter/getFreeProxy.py

@@ -97,9 +97,12 @@ class GetFreeProxy(object):
         """
         url = "http://www.goubanjia.com/free/gngn/index.shtml"
         tree = getHtmlTree(url)
-        proxy_list = tree.xpath('.//td[@class="ip"]')
-        for proxy in proxy_list:
-            yield ''.join(proxy.xpath('.//text()'))
+        # 现在每天最多放15个（一页）
+        for i in xrange(15):
+            d = tree.xpath('.//table[@class="table"]/tbody/tr[{}]/td'.format(i + 1))[0]
+            o = d.xpath('.//span/text() | .//div/text()')
+            a = ''.join(o[:-1]) + ':' + o[-1]
+            yield ''.join(o[:-1]) + ':' + o[-1]
 
 
 if __name__ == '__main__':