Das deutsche Python-Forum

Code: Alles auswählen

import urllib
import re


class ProxyListMaker():
    def __init__(self):
        self.counter = 1
        self.std_URL = "http://www.samair.ru/proxy/proxy-NR.htm"
        self.finder = re.compile(r"[0-9]{1,}.[0-9]{1,}.[0-9]{1,}.[0-9]{1,}:[0-9]{2,}")

    def getData(self):
        counter_ = "0" + str(self.counter)
        URL = self.std_URL.replace("NR", str(counter_))

        sock = urllib.urlopen(URL)
        data = sock.read()
        sock.close()

        return data

    def getAndWriteIPs(self):
        data = self.getData()
        proxies = re.findall(self.finder, data)

        writer = open("proxies.txt", "a")

        for proxy in proxies:
            writer.write(proxy + "\n")

        writer.close()

    def collectProxies(self):
        for self.counter in range(1, 40, 1):
            print "Getting Proxies from page: %s" % str(self.counter),
            self.getAndWriteIPs()
            self.counter += 1
            print "Getted!"

Code: Alles auswählen

import urllib
import re
from functools import partial

STD_URL = "http://www.samair.ru/proxy/proxy-{number}.htm"
IP_REGEXP = re.compile(r"[0-9]{1,}.[0-9]{1,}.[0-9]{1,}.[0-9]{1,}:[0-9]{2,}")

def get_proxies(url):
    proxies = list()
    for number in xrange(1,40):
        url = url.format(number=number)
        with urllib.urlopen(url) as socket:
            proxies.extend(re.findall(IP_REGEXP, socket.read())

    return proxies

def main(outfile, url=STD_URL):
    proxies = get_proxies(url)
    with open(outfile, 'w') as out:
        for proxy in proxies:
            out.write("{0}\n".format(proxy)

if __name__ == '__main__':
    import sys
    modes = { 2 : partial(main, sys.argv[1]),
              3 : partial(main, sys.argv[1], sys.argv[2])
            }
    try:
        modes[sys.argv]()
    except KeyError:
        #print usage info

Code: Alles auswählen

>>> IP_REGEXP = re.compile(r"[0-9]{1,}.[0-9]{1,}.[0-9]{1,}.[0-9]{1,}:[0-9]{2,}")
>>> print IP_REGEXP.match("1a2b3c4:42")
<_sre.SRE_Match object at 0x7f5a085e56b0>

Code: Alles auswählen

>>> print IP_REGEXP.match("11111111111111111111111111.2.3.4:42")
<_sre.SRE_Match object at 0x7f5a085e56b0>

Code: Alles auswählen

^((25[0-5]|2[0-4]\d|1\d{2}|\d{1,2})\.){3}(25[0-5]|2[0-4]\d|1\d{2}|\d{1,2}):[0-9]{2,}$