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 *SrcManage) Push(req *context.Request) {
spiderId, ok := req.GetSpiderId()
if !ok {
return
}
// 初始化该蜘蛛的队列
if _, ok := self.queue[spiderId]; !ok {
self.mutex[spiderId] = new(sync.Mutex)
self.queue[spiderId] = make(map[int][]*context.Request)
}
priority := req.GetPriority()
// 登记该蜘蛛下该优先级队列
if _, ok := self.queue[spiderId][priority]; !ok {
self.uIndex(spiderId, priority)
}
// 添加请求到队列
self.queue[spiderId][priority] = append(self.queue[spiderId][priority], req)
}
// 添加请求到队列
func (self *scheduler) Push(req *context.Request) {
pushMutex.Lock()
defer func() {
pushMutex.Unlock()
}()
if self.status == status.STOP {
return
}
is := self.Compare(req.GetUrl() + req.GetMethod())
// 有重复则返回
if is {
return
}
// 留作未来分发请求用
// if pholcus.Self.GetRunMode() == config.SERVER || req.CanOutsource() {
// return
// }
self.SrcManage.Push(req)
}
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)
cResp.SetResponse(resp)
if err != nil {
cResp.SetStatus(true, err.Error())
return cResp
}
cResp.SetStatus(false, "")
return cResp
}
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
}