From f881ba4164b21c391df2964c6a15a9020640e976 Mon Sep 17 00:00:00 2001
From: Q5nmvkg4x <1678550324@qq.com>
Date: Sun, 7 Nov 2021 06:09:30 +0800
Subject: [PATCH] ADD file via upload

---
 ipproxy.py | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 ipproxy.py

diff --git a/ipproxy.py b/ipproxy.py
new file mode 100644
index 0000000..cc064a2
--- /dev/null
+++ b/ipproxy.py
@@ -0,0 +1,33 @@
+import re
+
+import scrapy
+
+
+class IpproxySpider(scrapy.Spider):
+    name = 'ipproxy'
+    allowed_domains = ['kuaidaili.com']
+    start_urls = ['http://kuaidaili.com/free']
+    custom_settings={
+        'DOWNLOADER_MIDDLEWARES' :{
+            'spider.middlewares.IPDownloaderMiddleware': 543
+        },
+        'ITEM_PIPELINES' : {
+            'spider.pipelines.IPProxyPipeline': 500
+        },
+        'DOWNLOAD_DELAY' : 5
+    }
+
+    def parse(self, response):
+
+        for i in range(1,16):
+            IpPool = {}
+            ip = response.xpath('//*[@id="list"]/table/tbody/tr[{}]/td[1]'.format(i)).extract_first()
+            port = response.xpath('//*[@id="list"]/table/tbody/tr[{}]/td[2]'.format(i)).extract_first()
+            print(ip,port)
+            ip = re.findall('>.*?<',ip)[0][1:-1]
+            port = re.findall('>.*?<',port)[0][1:-1]
+            print(ip, port)
+            IpPool['ip']='http://'+ip+':'+port
+            print(IpPool['ip'])
+            yield IpPool
+