清除文件HTML格式函数ClearHtml
<%
'/* 函数名称:ClearHtml ReplaceHtml
'/* 函数语言:VBScript Language
'/* 作 用:清除文件HTML格式函数
'/* 传递参数:Content (注:需要进行清除的内容)
'/* 函数作者:auuo.com
'/* 函数说明:正则匹配(正则表达式)模式进行数据匹配替换
Function ClearHtml(Content)
Content=ReplaceHtml("&#[^>]*;", "", Content)
Content=ReplaceHtml("</?marquee[^>]*>", "", Content)
Content=ReplaceHtml("</?object[^>]*>", "", Content)
Content=ReplaceHtml("</?param[^>]*>", "", Content)
Content=ReplaceHtml("</?embed[^>]*>", "", Content)
Content=ReplaceHtml("</?table[^>]*>", "", Content)
Content=ReplaceHtml(" ","",Content)
Content=ReplaceHtml("</?tr[^>]*>", "", Content)
Content=ReplaceHtml("</?th[^>]*>","",Content)
Content=ReplaceHtml("</?p[^>]*>","",Content)
Content=ReplaceHtml("</?a[^>]*>","",Content)
Content=ReplaceHtml("</?img[^>]*>","",Content)
Content=ReplaceHtml("</?tbody[^>]*>","",Content)
Content=ReplaceHtml("</?li[^>]*>","",Content)
Content=ReplaceHtml("</?span[^>]*>","",Content)
Content=ReplaceHtml("</?div[^>]*>","",Content)
Content=ReplaceHtml("</?th[^>]*>", "", Content)
Content=ReplaceHtml("</?td[^>]*>", "", Content)
Content=ReplaceHtml("</?script[^>]*>", "", Content)
Content=ReplaceHtml("(javascript|jscript|vbscript|vbs):", "", Content)
Content=ReplaceHtml("on(mouse|exit|error|click|key)", "", Content)
Content=ReplaceHtml("<\\?xml[^>]*>", "", Content)
Content=ReplaceHtml("<\/?[a-z]+:[^>]*>", "", Content)
Content=ReplaceHtml("</?font[^>]*>", "", Content)
Content=ReplaceHtml("</?b[^>]*>","",Content)
Content=ReplaceHtml("</?u[^>]*>","",Content)
Content=ReplaceHtml("</?i[^>]*>","",Content)
Content=ReplaceHtml("</?strong[^>]*>","",Content)
ClearHtml=Content
End Function
Function ReplaceHtml(patrn, strng,content)
IF IsNull(content) Then
content=""
End IF
Set regEx = New RegExp ' 建立正则表达式。
regEx.Pattern = patrn ' 设置模式。
regEx.IgnoreCase = true ' 设置忽略字符大小写。
regEx.Global = True ' 设置全局可用性。
ReplaceHtml=regEx.Replace(content,strng) ' 执行正则匹配
End Function
Function RemoveHTML(strHTML)
Dim objRegExp, Match, Matches
Set objRegExp = New Regexp
objRegExp.IgnoreCase = True
objRegExp.Global = True
'取闭合的<>
objRegExp.Pattern = "<.+?>"
'进行匹配
Set Matches = objRegExp.Execute(strHTML)
' 遍历匹配集合,并替换掉匹配的项目
For Each Match in Matches
strHtml=Replace(strHTML,Match.Value,"")
Next
RemoveHTML=strHTML
Set objRegExp = Nothing
End Function
%>