go 去除html标签,golang去除html标签

package main

import (

"fmt"

"io/ioutil"

"os"

"regexp"

"strings"

)

func main() {

if len(os.Args) == 1 {

fmt.Println("请输入参 数")

return

}

b, err := ioutil.ReadFile(os.Args[1])

if err != nil {

println("readfile error")

return

}

str := string(b)

fmt.Println(trimHtml(str))

}

func trimHtml(src string) string {

//将HTML标签全转换成小写

re, _ := regexp.Compile("\\")

src = re.ReplaceAllStringFunc(src, strings.ToLower)

//去除STYLE

re, _ = regexp.Compile("\\

src = re.ReplaceAllString(src, "")

//去除SCRIPT

re, _ = regexp.Compile("\\

src = re.ReplaceAllString(src, "")

//去除所有尖括号内的HTML代码，并换成换行符

re, _ = regexp.Compile("\\<[\\S\\s]+?\\>")

src = re.ReplaceAllString(src, "\n")

//去除连续的换行符

re, _ = regexp.Compile("\\s{2,}")

src = re.ReplaceAllString(src, "\n")

return strings.TrimSpace(src)

}