func (self *HttpDownloader) downloadJson(p *robot.Page, req *robot.Request) *robot.Page {
var err error
p, destbody := self.downloadFile(p, req)
if !p.IsSucc() {
return p
}
var body []byte
body = []byte(destbody)
mtype := req.GetResponceType()
if mtype == "jsonp" {
tmpstr := goutils.JsonpToJson(destbody)
body = []byte(tmpstr)
}
var r *simplejson.Json
if r, err = simplejson.NewJson(body); err != nil {
mlog.LogInst().LogError(string(body) + "\t" + err.Error())
p.SetStatus(true, err.Error())
return p
}
// json result
p.SetBodyStr(string(body)).SetJson(r).SetStatus(false, "")
return p
}
func (self *QueueScheduler) Push(requ *robot.Request) {
self.locker.Lock()
var key [md5.Size]byte
if self.rm {
key = md5.Sum([]byte(requ.GetUrl()))
if _, ok := self.rmKey[key]; ok {
self.locker.Unlock()
return
}
}
e := self.queue.PushBack(requ)
if self.rm {
self.rmKey[key] = e
}
self.locker.Unlock()
}
func (self *RedisScheduler) Push(requ *robot.Request) {
self.locker.Lock()
defer self.locker.Unlock()
requJson, err := json.Marshal(requ)
if err != nil {
log.Println("RedisScheduler Push Error: " + err.Error())
return
}
conn := self.redisPool.Get()
defer conn.Close()
if err != nil {
log.Println("RedisScheduler Push Error: " + err.Error())
return
}
if self.forbiddenDuplicateUrl {
urlExist, err := conn.Do("HGET", self.urlList, requ.GetUrl())
if err != nil {
log.Println("RedisScheduler Push Error: " + err.Error())
return
}
if urlExist != nil {
return
}
conn.Do("MULTI")
_, err = conn.Do("HSET", self.urlList, requ.GetUrl(), 1)
if err != nil {
log.Println("RedisScheduler Push Error: " + err.Error())
conn.Do("DISCARD")
return
}
}
_, err = conn.Do("RPUSH", self.requestList, requJson)
if err != nil {
log.Println("RedisScheduler Push Error: " + err.Error())
if self.forbiddenDuplicateUrl {
conn.Do("DISCARD")
}
return
}
if self.forbiddenDuplicateUrl {
conn.Do("EXEC")
}
}
// choose a proxy server to excute http GET/method to download
func connectByHttpProxy(p *robot.Page, in_req *robot.Request) (*http.Response, error) {
request, _ := http.NewRequest("GET", in_req.GetUrl(), nil)
proxy, err := url.Parse(in_req.GetProxyHost())
if err != nil {
return nil, err
}
client := &http.Client{
Transport: &http.Transport{
Proxy: http.ProxyURL(proxy),
},
}
resp, err := client.Do(request)
if err != nil {
return nil, err
}
return resp, nil
}
func (self *HttpDownloader) Download(req *robot.Request) *robot.Page {
var mtype string
var p = robot.NewPage(req)
mtype = req.GetResponceType()
switch mtype {
case "html":
return self.downloadHtml(p, req)
case "json":
fallthrough
case "jsonp":
return self.downloadJson(p, req)
case "text":
return self.downloadText(p, req)
default:
mlog.LogInst().LogError("error request type:" + mtype)
}
return p
}
// Download file and change the charset of page charset.
func (self *HttpDownloader) downloadFile(p *robot.Page, req *robot.Request) (*robot.Page, string) {
var err error
var urlstr string
if urlstr = req.GetUrl(); len(urlstr) == 0 {
mlog.LogInst().LogError("url is empty")
p.SetStatus(true, "url is empty")
return p, ""
}
var resp *http.Response
if proxystr := req.GetProxyHost(); len(proxystr) != 0 {
//using http proxy
//fmt.Print("HttpProxy Enter ",proxystr,"\n")
resp, err = connectByHttpProxy(p, req)
} else {
//normal http download
//fmt.Print("Http Normal Enter \n",proxystr,"\n")
resp, err = connectByHttp(p, req)
}
if err != nil {
return p, ""
}
p.SetHeader(resp.Header)
p.SetCookies(resp.Cookies())
// get converter to utf-8
var bodyStr string
if resp.Header.Get("Content-Encoding") == "gzip" {
bodyStr = self.changeCharsetEncodingAutoGzipSupport(resp.Header.Get("Content-Type"), resp.Body)
} else {
bodyStr = self.changeCharsetEncodingAuto(resp.Header.Get("Content-Type"), resp.Body)
}
//fmt.Printf("utf-8 body %v \r\n", bodyStr)
defer resp.Body.Close()
return p, bodyStr
}
// choose http GET/method to download
func connectByHttp(p *robot.Page, req *robot.Request) (*http.Response, error) {
client := &http.Client{}
httpreq, err := http.NewRequest(req.GetMethod(), req.GetUrl(), strings.NewReader(req.GetPostdata()))
if header := req.GetHeader(); header != nil {
httpreq.Header = req.GetHeader()
}
if cookies := req.GetCookies(); cookies != nil {
for i := range cookies {
httpreq.AddCookie(cookies[i])
}
}
var resp *http.Response
if resp, err = client.Do(httpreq); err != nil {
if e, ok := err.(*url.Error); ok && e.Err != nil && e.Err.Error() == "normal" {
// normal
} else {
mlog.LogInst().LogError(err.Error())
p.SetStatus(true, err.Error())
//fmt.Printf("client do error %v \r\n", err)
return nil, err
}
}
return resp, nil
}
