func (self *Pholcus) TryOutsource(req *context.Request) bool {
if self.IsOutsource() && req.TryOutsource() {
self.Send(*req)
return true
}
return false
}
func (self *HttpDownloader) downloadJson(p *context.Response, req *context.Request) *context.Response {
var err error
p, destbody := self.downloadFile(p, req)
if !p.IsSucc() {
return p
}
var body []byte
body = []byte(destbody)
mtype := req.GetRespType()
if mtype == "jsonp" {
tmpstr := util.JsonpToJson(destbody)
body = []byte(tmpstr)
}
var r *simplejson.Json
if r, err = simplejson.NewJson(body); err != nil {
reporter.Log.Println(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 *scheduler) Push(req *context.Request) {
is := self.Compare(req.GetUrl())
// 有重复则返回
if is {
return
}
self.SrcManage.Push(req)
}
func (self *scheduler) Push(req *context.Request) {
if self.status == STOP {
return
}
is := self.Compare(req.GetUrl() + req.GetMethod())
// 有重复则返回
if is {
return
}
self.SrcManage.Push(req)
}
func (self *SrcManage) Push(req *context.Request) {
if spiderId, ok := req.GetSpiderId(); ok {
priority := int(req.GetPriority())
if priority > MAX_PRIORITY {
priority = MAX_PRIORITY
}
for i, x := 0, priority+1-len(self.queue[spiderId]); i < x; i++ {
self.queue[spiderId] = append(self.queue[spiderId], []*context.Request{})
}
self.queue[spiderId][priority] = append(self.queue[spiderId][priority], req)
}
}
func (self *HttpDownloader) Download(req *context.Request) *context.Response {
var mtype string
var p = context.NewResponse(req)
mtype = req.GetRespType()
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:
reporter.Log.Println("error request type:" + mtype)
}
return p
}
// choose a proxy server to excute http GET/method to download
func connectByHttpProxy(p *context.Response, in_req *context.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
}
// Download file and change the charset of response charset.
func (self *HttpDownloader) downloadFile(p *context.Response, req *context.Request) (*context.Response, string) {
var err error
var urlstr string
if urlstr = req.GetUrl(); len(urlstr) == 0 {
reporter.Log.Println("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, ""
}
//b, _ := ioutil.ReadAll(resp.Body)
//fmt.Printf("Resp body %v \r\n", string(b))
p.SetHeader(resp.Header)
p.SetCookies(resp.Cookies())
// get converter to utf-8
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
}
func (self *Surfer) Download(cReq *context.Request) *context.Response {
cResp := context.NewResponse(nil)
resp, err := self.download.Download(cReq.GetMethod(), cReq.GetUrl(), cReq.GetReferer(), cReq.GetPostData(), cReq.GetHeader(), cReq.GetCookies())
cResp.SetRequest(cReq)
if err != nil {
cResp.SetStatus(true, err.Error())
return cResp
}
// get converter to utf-8
body := self.changeCharsetEncodingAuto(resp.Body, resp.Header.Get("Content-Type"))
//fmt.Printf("utf-8 body %v \r\n", bodyStr)
defer resp.Body.Close()
cResp.SetText(body)
cResp.SetStatus(false, "")
return cResp
}
// choose http GET/method to download
func connectByHttp(p *context.Response, req *context.Request) (*http.Response, error) {
client := &http.Client{
CheckRedirect: req.GetRedirectFunc(),
}
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 {
reporter.Log.Println(err.Error())
p.SetStatus(true, err.Error())
//fmt.Printf("client do error %v \r\n", err)
return nil, err
}
}
return resp, nil
}
func (self *SrcManage) Push(req *context.Request) {
if spiderId, ok := req.GetSpiderId(); ok {
self.queue[spiderId] = append(self.queue[spiderId], req)
}
}