import urllib2
import sys
import bs4
import chardet
reload(sys)
sys.setdefaultencoding('utf-8')
def download(url):
  htmlfile = open('test.html','w')
  try:
    result = urllib2.urlopen(url)
    content = result.read()
    info = result.info()
    result.close()
  except Exception,e:
    print 'download error!!!'
    print e
  else:
    if content != None:
      charset1 = (chardet.detect(content))['encoding'] #real encoding type
      charset2 = info.getparam('charset') #declared encoding type
      print charset1,' ', charset2
      # case1: charset is not None.
      if charset1 != None and charset2 != None and charset1.lower() != charset2.lower():
        newcont = bs4.BeautifulSoup(content, from_encoding='GB18030')  #coding: GB18030
        for cont in newcont:
          htmlfile.write('%s\n'%cont)
      # case2: either charset is None, or charset is the same.
      else:
        #print sys.getdefaultencoding()
        htmlfile.write(content) #default coding: utf-8
  htmlfile.close()
if __name__ == "__main__":
  url = 'https://www.jb51.net'
  download(url)