func trimHtml(src string) string { //将HTML标签全转换成小写 re,_ := regexp.Compile("\\<[\\S\\s]+?\\>") src = re.ReplaceAllStringFunc(src,strings.ToLower) //去除STYLE re,_ = regexp.Compile("\\<style[\\S\\s]+?\\</style\\>") src = re.ReplaceAllString(src,"") //去除SCRIPT re,_ = regexp.Compile("\\<script[\\S\\s]+?\\</script\\>") 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) }
原文链接:https://www.f2er.com/go/188940.html