#!/usr/bin/python
#encoding: utf8
import re, sys, urllib, time
def fetch_page(num):
sock = urllib.urlopen('http://www.habrahabr.ru/people/page%d/' % num)
html = sock.read().decode('CP1251')
sock.close()
return html
def grep_users(page):
p = re.compile(u'.*.*(.*).*.*', re.DOTALL | re.UNICODE);
m = p.match(page)
if m:
ucode = m.group(1)
else:
return []
u = re.compile('.*| (.*) | .*', re.UNICODE)
return u.findall(ucode)
users = []
for i in range(1, 117):
users = grep_users(fetch_page(i)) + users;
sys.stderr.write('Page %d processed; Total users found: %d\n' % (i, len(users)))
time.sleep(5)
c = 0
output = u''
for u in users:
c = c + 1
output = output + u
if c > 21:
output = output + '
'
c = 0