jyInfo/rpc/util/htmlclear.go

package util

import (
	mc "app.yhyue.com/moapp/jybase/common"
	"regexp"
	"strings"
)

//
type Cut struct {
	scriptTag     *regexp.Regexp
	sa            *regexp.Regexp
	replBlankLine *regexp.Regexp
	replStartWrap *regexp.Regexp
	multiCR       *regexp.Regexp
	tag           *regexp.Regexp
	//annotate      *regexp.Regexp
	//tag           *regexp.Regexp
	//inputag       *regexp.Regexp
	//isborder      *regexp.Regexp
	//hiddentag     *regexp.Regexp
	//styletag      *regexp.Regexp
	//colstag       *regexp.Regexp
	//rowstag       *regexp.Regexp
	//display       *regexp.Regexp
	//replTags2CR   []string
	//retainTags2CR []string
}

//
func NewCut() *Cut {
	scs := regexp.MustCompile("(?s)<(script|style)[^>]*>.+?</(script|style)>")
	sa := regexp.MustCompile("<[a|A]\\s*[^>]*>(.*?)</[a|A]>")
	bl := regexp.MustCompile("\\s+[\r\n]")
	sw := regexp.MustCompile("^[\u3000\u2003\u00a0\\s]+|[\u3000\u2003\u00a0\\s]+$")
	m, _ := regexp.Compile("([\r\n][\u3000\u2003\u00a0\\s]*)+|[\r\n]+")
	t, _ := regexp.Compile("<[^>]+>")

	//t, _ := regexp.Compile("<[^>]+>")
	////sc, _ := regexp.Compile("\\<script[^\\>]*\\>*[^\\>]+\\</script\\>")
	////ss, _ := regexp.Compile("\\<style[^\\>]*\\>*[^\\>]+\\</style\\>")
	//at := regexp.MustCompile("(?s)<(!%-%-|!--).*?(%-%-|--)>") //注释 css
	//hiddentag := regexp.MustCompile(`<\s*input[^<]*type=("|')hidden("|')[^<]*>`)
	//input := regexp.MustCompile(`<\s*input[^<]*value=("|')([^>"']*)[^<]*>`)
	//cols, _ := regexp.Compile(`colspan="\d+"`)
	//rows, _ := regexp.Compile(`rowspan="\d+"`)
	//border, _ := regexp.Compile(`(border="(\d+)")|(cellpadding="(\d+)")|(cellspacing="(\d+)")`)
	//dis, _ := regexp.Compile(`display:none`)
	return &Cut{
		scriptTag:     scs,
		sa:            sa,
		replBlankLine: bl,
		replStartWrap: sw,
		multiCR:       m,
		tag:           t,
		//annotate:      at,
		//tag:           t,
		//hiddentag:     hiddentag,
		//inputag:       input,
		//colstag:       cols,
		//isborder:      border,
		//rowstag:       rows,
		//display:       dis,
		//replTags2CR:   []string{"div", "p", "br", "h1", "h2", "h3", "h4", "h5"},
		//retainTags2CR: []string{"table", "thead", "tfoot", "tbody", "th", "td", "tr"},
	}
}

//清理HTML标签
func (c *Cut) ClearHtml(src string) string {
	src = strings.Replace(src, ">\n", ">", -1)
	src = strings.Replace(src, " ", "", -1)
	//标签全转小写
	src = c.tag.ReplaceAllStringFunc(src, strings.ToLower)
	//清script,style
	src = c.scriptTag.ReplaceAllString(src, "")
	src = c.sa.ReplaceAllString(src, "$1")
	src = c.replStartWrap.ReplaceAllString(src, "")
	src = c.replBlankLine.ReplaceAllString(src, "\n")
	//清除多余换行
	c.multiCR.ReplaceAllString(src, "\n")
	return strings.Replace(src, "\n", "<br/>", -1)
}

//判断table是否加表格线
func isHasBoder(con string, reg *regexp.Regexp) bool {
	res := reg.FindAllStringSubmatch(con, -1)
	hasBorder := false
	for _, v := range res {
		for k, val := range v {
			if k > 0 && k%2 == 0 && mc.IntAll(val) > 0 {
				hasBorder = true
				break
			}
		}
	}
	return hasBorder
}