feat:detail 标签清理

rpc/util/htmlclear.go

@@ -8,8 +8,11 @@ import (
 
 //
 type Cut struct {
-	scripttag *regexp.Regexp
-	sa        *regexp.Regexp
+	scripttag     *regexp.Regexp
+	sa            *regexp.Regexp
+	replBlankLine *regexp.Regexp
+	replStartWrap *regexp.Regexp
+	multiCR       *regexp.Regexp
 	//annotate      *regexp.Regexp
 	//tag           *regexp.Regexp
 	//inputag       *regexp.Regexp
@@ -19,9 +22,6 @@ type Cut struct {
 	//colstag       *regexp.Regexp
 	//rowstag       *regexp.Regexp
 	//display       *regexp.Regexp
-	//multiCR       *regexp.Regexp
-	//replBlankLine *regexp.Regexp
-	//replStartWrap *regexp.Regexp
 	//replTags2CR   []string
 	//retainTags2CR []string
 }
@@ -30,9 +30,11 @@ type Cut struct {
 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("<[^>]+>")
-	//m, _ := regexp.Compile("([\r\n][\u3000\u2003\u00a0\\s]*)+|[\r\n]+")
 	////sc, _ := regexp.Compile("\\<script[^\\>]*\\>*[^\\>]+\\</script\\>")
 	////ss, _ := regexp.Compile("\\<style[^\\>]*\\>*[^\\>]+\\</style\\>")
 	//at := regexp.MustCompile("(?s)<(!%-%-|!--).*?(%-%-|--)>") //注释 css
@@ -43,8 +45,11 @@ func NewCut() *Cut {
 	//border, _ := regexp.Compile(`(border="(\d+)")|(cellpadding="(\d+)")|(cellspacing="(\d+)")`)
 	//dis, _ := regexp.Compile(`display:none`)
 	return &Cut{
-		scripttag: scs,
-		sa:        sa,
+		scripttag:     scs,
+		sa:            sa,
+		replBlankLine: bl,
+		replStartWrap: sw,
+		multiCR:       m,
 		//annotate:      at,
 		//tag:           t,
 		//hiddentag:     hiddentag,
@@ -53,9 +58,6 @@ func NewCut() *Cut {
 		//isborder:      border,
 		//rowstag:       rows,
 		//display:       dis,
-		//multiCR:       m,
-		//replBlankLine: regexp.MustCompile("\\s+[\r\n]"),
-		//replStartWrap: regexp.MustCompile("^[\u3000\u2003\u00a0\\s]+|[\u3000\u2003\u00a0\\s]+$"),
 		//replTags2CR:   []string{"div", "p", "br", "h1", "h2", "h3", "h4", "h5"},
 		//retainTags2CR: []string{"table", "thead", "tfoot", "tbody", "th", "td", "tr"},
 	}
@@ -66,6 +68,10 @@ func (c *Cut) ClearHtml(src string) string {
 	//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)
 }