// 添加请求到队列
func (self *scheduler) Push(req *context.Request) {
// 初始化该蜘蛛的队列
spiderId, ok := req.GetSpiderId()
if !ok {
return
}
self.RLock()
defer self.RUnlock()
if self.status == status.STOP {
return
}
// 当req不可重复时，有重复则返回
if !req.GetDuplicatable() && self.Deduplicate(req.GetUrl()+req.GetMethod()) {
return
}
// 初始化该蜘蛛下该优先级队列
priority := req.GetPriority()
if !self.foundPriority(spiderId, priority) {
self.addPriority(spiderId, priority)
}
defer func() {
recover()
}()
// 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 *Matrix) Push(req *context.Request) {
sdl.RLock()
defer sdl.RUnlock()
if sdl.status == status.STOP ||
self.maxPage >= 0 ||
// 当req不可重复下载时，已存在成功记录则返回
!req.IsReloadable() && !UpsertSuccess(req) {
return
}
// 大致限制加入队列的请求量，并发情况下应该会比maxPage多
atomic.AddInt64(&self.maxPage, 1)
priority := req.GetPriority()
// 初始化该蜘蛛下该优先级队列
if _, found := self.reqs[priority]; !found {
self.priorities = append(self.priorities, priority)
sort.Ints(self.priorities) // 从小到大排序
self.reqs[priority] = []*context.Request{}
}
// 添加请求到队列
self.reqs[priority] = append(self.reqs[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)
}