func main() {
req := robot.NewRequest(wkSohuUrl, "html", "index", "GET", "", nil, nil, nil, nil)
sohuSpider := robot.NewSpider(NewMyPageProcesser(), "Sohu").
AddRequest(req).
SetSleepTime("rand", 500, 1000).
SetThreadnum(2)
for i := 1; i < maxWKSouhuLayer; i++ {
url := fmt.Sprintf("http://yule.sohu.com/gossip/index_%d.shtml", 5301-i) // magic num
req := robot.NewRequest(url, "html", "index", "GET", "", nil, nil, nil, nil)
sohuSpider.AddRequest(req)
}
sohuSpider.Run()
}
func TestQueueScheduler(t *testing.T) {
var r *robot.Request
r = robot.NewRequest("http://baidu.com", "html", "", "GET", "", nil, nil, nil, nil)
fmt.Printf("%v\n", r)
var s *QueueScheduler
s = NewQueueScheduler(false)
s.Push(r)
var count int = s.Count()
if count != 1 {
t.Error("count error")
}
fmt.Println(count)
var r1 *robot.Request
r1 = s.Poll()
if r1 == nil {
t.Error("poll error")
}
fmt.Printf("%v\n", r1)
// remove duplicate
s = NewQueueScheduler(true)
r2 := robot.NewRequest("http://qq.com", "html", "", "GET", "", nil, nil, nil, nil)
s.Push(r)
s.Push(r2)
s.Push(r)
count = s.Count()
if count != 2 {
t.Error("count error")
}
fmt.Println(count)
r1 = s.Poll()
if r1 == nil {
t.Error("poll error")
}
fmt.Printf("%v\n", r1)
r1 = s.Poll()
if r1 == nil {
t.Error("poll error")
}
fmt.Printf("%v\n", r1)
}
func addRequest(p *robot.Page, tag, url, cookie, content string) {
req := robot.NewRequest(url, "json", tag, "GET", "", nil, nil, nil, content)
p.AddTargetRequestWithParams(req)
}
