Python-сообщество

# -*- coding: cp1251 -*-
import urllib, urlparse, re
from BeautifulSoup import BeautifulSoup
from BeautifulSoup import BeautifulStoneSoup

url_start = "http://site.ru"
url_end = "sitemap.xml"
url = urlparse.urljoin(url_start, url_end)

print u"Ждите ..."

link = urllib.urlopen(url)
sup = BeautifulStoneSoup(link)
supik = str(sup)
k = 1

p = re.compile(r"<loc>(.*?)</loc>", re.S | re.I)

def starting(gg):
    global k
    link1 = urllib.urlopen(gg)
    soup = BeautifulSoup(link1, fromEncoding="utf-8")
    a = soup(attrs={"name": "Keywords"})[0]["content"]
    f1 = open("file1.txt", "a")
    f1.write(a.encode("cp1251") + "\n")
    f1.close()
    link1.close()
    print u"Спарсили ключевик № " + str(k)
    k += 1

q = raw_input("Parse ? y/n: ")

if q == "y":
    f = open("file1.txt", "w")
    for i in p.findall(supik):
        starting(i)
    f.close()
    link.close()
else:
    print "Okay ..."

raw_input()

# -*- coding: cp1251 -*-
import urllib, urlparse, re, Queue, threading, time
from BeautifulSoup import BeautifulSoup
from BeautifulSoup import BeautifulStoneSoup

queue = Queue.Queue()
lock = threading.RLock()
threads_count = 100

url_start = "http://site.ru"
url_end = "sitemap.xml"
url = urlparse.urljoin(url_start, url_end)

print u"Ждите ..."

link = urllib.urlopen(url)
sup = BeautifulStoneSoup(link)
supik = str(sup)
k = 1

p = re.compile(r"<loc>(.*?)</loc>", re.S | re.I)

def worker():
    global queue
    while True:
        try:
            target = queue.get_nowait()
        except Queue.Empty:
            return
        starting(target)
		
def write(data, s):
    global k
    lock.acquire()
    if s == True:
        f1 = open("file1.txt", "a")
        f1.write(data + "\n")
        f1.close()
    elif s != True:
        pass
    lock.release()
    print u"Спарсили ключевик № " + str(k)
    k += 1

def starting(gg):
    try:
        link1 = urllib.urlopen(gg)
        soup = BeautifulSoup(link1, fromEncoding="utf-8")
        a = soup(attrs={"name": "Keywords"})[0]["content"]
        b = a.encode("cp1251")
        write(b, True)
        link1.close()
    except:
        write(b, False)
		
def main():
    global link
    f = open("file1.txt", "w")
    for i in p.findall(supik):
        queue.put(i)
    for _ in xrange(threads_count):
        thread_ = threading.Thread(target=worker)
        thread_.start()
        while threading.active_count() >1:
            time.sleep(1)
    f.close()
    link.close()   

q = raw_input("Parse ? y/n: ")

if q == "y":
    main()
else:
    print "Okay ..."

raw_input()

link1 = urllib.urlopen(gg)