Python抓取淘宝IP地址数据

清华大佬耗费三个月吐血整理的几百G的资源,免费分享!....>>>

def fetch(ip):
    url = 'http://ip.taobao.com/service/getIpInfo.php?ip=' + ip
    result = []
    try:
        response = urllib.urlopen(url).read()
        jsondata = json.loads(response)
        if jsondata[u'code'] == 0:
            result.append(jsondata[u'data'][u'ip'].encode('utf-8'))            
            result.append(jsondata[u'data'][u'country'].encode('utf-8'))
            result.append(jsondata[u'data'][u'country_id'].encode('utf-8'))
            result.append(jsondata[u'data'][u'area'].encode('utf-8'))
            result.append(jsondata[u'data'][u'area_id'].encode('utf-8'))
            result.append(jsondata[u'data'][u'region'].encode('utf-8'))
            result.append(jsondata[u'data'][u'region_id'].encode('utf-8'))
            result.append(jsondata[u'data'][u'city'].encode('utf-8'))
            result.append(jsondata[u'data'][u'city_id'].encode('utf-8'))
            result.append(jsondata[u'data'][u'county'].encode('utf-8'))
            result.append(jsondata[u'data'][u'county_id'].encode('utf-8'))
            result.append(jsondata[u'data'][u'isp'].encode('utf-8'))
            result.append(jsondata[u'data'][u'isp_id'].encode('utf-8'))            
        else:
            return 0, result
    except:
        logging.exception("Url open failed:" + url)
        return 0, result
    return 1, result

def worker(ratelimit, jobs, results, progress):
    global cancel
    while not cancel:
        try:
            ratelimit.ratecontrol()
            ip = jobs.get(timeout=2) # Wait 2 seconds
            ok, result = fetch(ip)
            if not ok:
                logging.error("Fetch information failed, ip:{}".format(ip))
                progress.put("") # Notify the progress even it failed
            elif result is not None:
                results.put(" ".join(result))
            jobs.task_done()    # Notify one item
        except Queue.Empty:
            pass
        except:
            logging.exception("Unknown Error!")