gistfile1.py

import socket
import re
import urllib

socket.setdefaulttimeout(5)

def get_content(url):
    try:
        f = urllib.urlopen(url)
    except IOError:
        return ""
    try:
        return f.read()
    finally:
        f.close()

def head(s, n):
    return "\n".join(s.splitlines()[:n])

def get_title(url):
    content = get_content(url)
    m = re.search('<title>(.*?)</title>', content, re.DOTALL|re.IGNORECASE)
    if m:
        title = m.group(1).strip()
        for enc in ('utf-8', 'euc-jp', 'shift-jis'):
            try:
                title = title.decode(enc)
                break
            except UnicodeDecodeError:
                pass
        return title
    else:
        return ""