package main
import (
"fmt"
"github.com/antchfx/htmlquery"
"net/http"
"strings"
)
func getResponse(url string ) *http.Response {
client := &http.Client{}
//生成要访问的url
//提交请求
request, err := http.NewRequest("GET", url, nil)
//增加header选项
request.Header.Add("User-Agent", "golang")
if err != nil {
panic(err)
}
//处理返回结果
resp, _ := client.Do(request)
return resp
}
func main() {
var url string = "https://www.cnblogs.com/brady-wang/"
response := getResponse(url)
defer response.Body.Close()
doc,_ := htmlquery.Parse(response.Body)
list := htmlquery.Find(doc, "//*[@id='mainContent']//div[@class='postTitle']/a")
for _,item := range list{
title := htmlquery.InnerText(item)
title = strings.Replace(title, " ", "", -1)
// 去除换行符
title = strings.Replace(title, "
", "", -1)
fmt.Printf("title %s",title)
fmt.Printf("href %s
",htmlquery.SelectAttr(item,"href"))
}
}