def get_title(domain):
domain_t = IP._getDomainStr(domain)
#print domain_t
# 根据传递进来的 domain 获取 首先直接打开网站获取到 title 然后再去修改 header的值 去对比titie
# 如果网站本来就打不开呢 打得开 获取 title 打不开直接 输出title
try:
http_domain = u'http://' + unicode(domain_t[0])
u = u'(' + http_domain + '.*?)"'
u2 = u'' + http_domain + u'(.*)'
t = u'<title>.*</title>'
_re_title = re.compile(t)
_re_url = re.compile(u2)
abc = 'http://' + domain.strip()
r = requests.get(abc, verify=False, allow_redirects=True)
url = _re_url.findall(r.url)[0]
get1 = _re_url.findall('http://' + domain)[0]
if get1 == '':
get1 = '/'
r1 = get(
domain_t[0], 80,
__http_header.replace('{host}',
domain_t[0]).replace('{url}', get1))
r2, code = x2Unicode(r1)
title = _re_title.findall(r2)[0]
print 'Get key words:\n"' + title.decode('utf-8') + '"'
u_title, code = x2Unicode(title)
return url, u_title, domain_t[0]
except Exception as a:
print a.message
#return url,'1',domain_t[0]
return 0
