yusuan/util.go

// clear
package main

import (
	"regexp"
	"strings"
)

var at = rune('&')
var ed = rune(';')
var lableMap = map[string]rune{
	"&":  rune('&'),
	" ": rune(' '),
	">":   rune('>'),
	"&lt;":   rune('<'),
}

func CleanString(input string) string {
	reg := regexp.MustCompile("[^a-zA-Z0-9\u4e00-\u9fa5]+")
	cleaned := reg.ReplaceAllString(input, "")
	return cleaned
}

// 清洗正文
func CleanDetailText(detail string) string {
	detail = regexp.MustCompile(`<!--[\w\W]*?-->`).ReplaceAllString(detail, "")
	detail = regexp.MustCompile(`<!--[\\n\\N]*?-->`).ReplaceAllString(detail, "")
	detail = CutLableStr(detail)
	return detail
}

//处理转义标签
func CutLableStr(con string) string {
	for i := 0; i < 3; i++ {
		runes := []rune{}
		pools := []rune{}
		bpool := false
		strings.IndexFunc(con, func(s rune) bool {
			if !bpool && s == at {
				bpool = true
				pools = []rune{}
			}
			if bpool {
				pools = append(pools, s)
				if s == ed { //结束
					lb := lableMap[string(pools)]
					if lb != 0 {
						runes = append(runes, lb)
					} else {
						runes = append(runes, pools...)
					}
					bpool = false
				} else if len(pools) > 6 {
					bpool = false
					runes = append(runes, pools...)
				}
			} else {
				runes = append(runes, s)
			}
			return false
		})
		str1 := string(runes)
		if i > 0 && con == str1 {
			break
		}
		con = str1
	}
	return con
}