ServerStatusParser

gistfile1.py

import urllib2, time, re

link = 'http://host/server-status'
data = []
period = 1 # sec

def get(link):
    while True:
        response = urllib2.urlopen(link)
        html = response.read()
        links = re.findall (r'<td nowrap>.*</td>', html)
        for lnk in links:
            lnk = lnk.replace("<td nowrap>","").replace("</td>GET ","").replace(" HTTP/1.0</td>","").replace("</td>HEAD ","").replace("</td>POST ","")
            if lnk not in data:
                with open("links.txt", "a") as f:
                    f.write(lnk + "\n")
                data.append(lnk.replace("<td nowrap>","").replace("</td>GET ","").replace(" HTTP/1.0</td>","").replace("</td>HEAD ","").replace("</td>POST ",""))
        time.sleep(period)
        print "Total links:" + str(len(data))
get(link)