''' <summary>
        ''' 去除HTML標記
        ''' </summary>
        ''' <param name="Htmlstring">包括HTML的源碼 </param>
        ''' <returns>已經去除後的文字</returns>
        Public Shared Function NoHTML(ByVal Htmlstring As String) As String
            '刪除腳本
            Htmlstring = Regex.Replace(Htmlstring, "<script[^>]*?>.*?</script>", "", RegexOptions.IgnoreCase)
            '刪除HTML
            Htmlstring = Regex.Replace(Htmlstring, "<(.[^>]*)>", "", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "([\r\n])[\s]+", "", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "-->", "", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "<!--.*", "", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(quot|#34);", """", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(amp|#38);", "&", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(lt|#60);", "<", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(gt|#62);", ">", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(nbsp|#160);", "   ", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(iexcl|#161);", "!", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(cent|#162);", "¢", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(pound|#163);", "£", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&(copy|#169);", "c", RegexOptions.IgnoreCase)
            Htmlstring = Regex.Replace(Htmlstring, "&#(\d+);", "", RegexOptions.IgnoreCase)
            Htmlstring.Replace("<", "")
            Htmlstring.Replace(">", "")
            Htmlstring.Replace("" & Chr(13) & "" & Chr(10) & "", "")
            Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim()
            Return Htmlstring
        End Function

        '''提取HTML代碼中文字的C#函數    
        '''   <summary>  
        '''   去除HTML標記  
        '''   </summary>  
        '''   <param   name="strHtml">包括HTML的源碼   </param>  
        '''   <returns>已經去除後的文字</returns>  
        Public Shared Function StripHTML(ByVal strHtml As String) As String
            Dim aryReg As String() = {"<script[^>]*?>.*?</script>", "<(\/\s*)?!?((\w+:)?\w+)(\w+(\s*=?\s*(([""'])(\\[""'tbnr]|[^\7])*?\7|\w+)|.{0})|\s)*?(\/\s*)?>", "([\r\n])[\s]+", "&(quot|#34);", "&(amp|#38);", "&(lt|#60);", _
             "&(gt|#62);", "&(nbsp|#160);", "&(iexcl|#161);", "&(cent|#162);", "&(pound|#163);", "&(copy|#169);", _
             "&#(\d+);", "-->", "<!--.*\n"}

            'chr(161),  
            'chr(162),  
            'chr(163),  
            'chr(169),  
            Dim aryRep As String() = {"", "", "", """", "&", "<", _
             ">", "   ", "!", "¢", "£", "c", _
             "", "" & Chr(13) & "" & Chr(10) & "", ""}

            Dim newReg As String = aryReg(0)
            Dim strOutput As String = strHtml
            For i As Integer = 0 To aryReg.Length - 1
                Dim regex As New Regex(aryReg(i), RegexOptions.IgnoreCase)
                strOutput = regex.Replace(strOutput, aryRep(i))
            Next
            strOutput.Replace("<", "")
            strOutput.Replace(">", "")
            strOutput.Replace("" & Chr(13) & "" & Chr(10) & "", "")
            Return strOutput
        End Function

        '''取出文本中的圖片位址
        '''   <summary>
        '''   取出文本中的圖片位址
        '''   </summary>
        '''   <param   name="HTMLStr">HTMLStr</param>
        Public Shared Function GetImgUrl(ByVal HTMLStr As String) As String
            Dim str As String = String.Empty
            Dim sPattern As String = "^<img\s+[^>]*>"
            Dim r As New Regex("<img\s+[^>]*\s*src\s*=\s*([']?)(?<url>\S+)'?[^>]*>", RegexOptions.Compiled)
            Dim m As Match = r.Match(HTMLStr.ToLower())
            If m.Success Then
                str = m.Result("${url}")
            End If
            Return str
        End Function

 

來源出處不可考,若有知情者,還請告知,感謝...


arrow
arrow
    全站熱搜

    Roger 發表在 痞客邦 留言(0) 人氣()