Anonymous

def getContent(site, query, retrieveData=True):
  print '###### gC:  '+site+', '+query
  try:
    ht=httplib.HTTPConnection(site)
    ht.request("GET", query)
    rsp=ht.getresponse()
    if retrieveData: pg=rsp.read()
    else: pg=(rsp.status==200)
    ht.close()
    if retrieveData:
      v=re.search(r'Results.*?of.*?[0-9,]+', pg)
      # returns [aprox # of entries found, contents of search page minus the <head>...</head> subsection]
      return [int(re.sub(',', '', re.sub(r'Results.*of[^0-9,]+', '', pg[v.start():v.end()]))), re.findall('<a href=.*?</a>', re.sub('<head>.*</head>', '', pg))]
    else: return pg
  except:
    if retrieveData: return [0, '']
    else:  return False