func (this *PageProcesserHtml) Process(p *page.Page) {
if !p.IsSucc() {
println(p.Errormsg())
return
}
result := map[string]string{}
for k, _ := range this.rule {
result[k] = ""
}
query := p.GetHtmlParser()
var urls []string
query.Find(this.page["rule"]).Each(func(i int, s *goquery.Selection) {
href := ""
if this.page["fun"] == "text" {
href = s.Text()
} else {
href, _ = s.Attr(this.page["fun"])
}
urls = append(urls, this.page["pre"]+href)
})
p.AddMyTargetRequests(urls, this.conf["texttype"], "", this.conf["resqType"], this.conf["postdata"], this.conf["proxy"], this.conf["heardefile"], this.conf["cookie"])
for k, v := range this.rule {
if this.num[k] == "ALL" {
var items []string
query.Find(v).Each(func(i int, s *goquery.Selection) {
item := ""
if this.fun[k] == "text" {
item = s.Text()
} else {
item, _ = s.Attr(this.fun[k])
}
items = append(items, item)
})
result[k] = strings.Join(items, "|")
} else {
if this.fun[k] == "text" {
result[k] = query.Find(v).Text()
} else {
result[k], _ = query.Find(v).Attr(this.fun[k])
}
result[k] = strings.Trim(result[k], " \t\n")
}
if result[k] == "" {
p.SetSkip(true)
}
p.AddField(k, result[k])
}
for k, v := range p.GetPageItems().GetAll() {
println(k, v)
}
}
func TestDownloadHtml(t *testing.T) {
//return
//request := request.NewRequest("http://live.sina.com.cn/zt/api/l/get/finance/globalnews1/index.htm?format=json&callback=t13975294&id=23521&pagesize=45&dire=f&dpc=1")
var req *request.Request
req = request.NewRequest("http://live.sina.com.cn/zt/l/v/finance/globalnews1/", "html", "", "GET", "", nil, nil, nil, nil)
var dl downloader.Downloader
dl = downloader.NewHttpDownloader()
var p *page.Page
p = dl.Download(req)
var doc *goquery.Document
doc = p.GetHtmlParser()
//fmt.Println(doc)
//body := p.GetBodyStr()
//fmt.Println(body)
var s *goquery.Selection
s = doc.Find("body")
if s.Length() < 1 {
t.Error("html parse failed!")
}
/*
doc, err := goquery.NewDocument("http://live.sina.com.cn/zt/l/v/finance/globalnews1/")
if err != nil {
fmt.Printf("%v",err)
}
s := doc.Find("meta");
fmt.Println(s.Length())
resp, err := http.Get("http://live.sina.com.cn/zt/l/v/finance/globalnews1/")
if err != nil {
fmt.Printf("%v",err)
}
defer resp.Body.Close()
doc, err = goquery.NewDocumentFromReader(resp.Body)
s = doc.Find("meta");
fmt.Println(s.Length())
*/
}