C#读取HTML文件内容写入记事本代码,CSharp代码库,德仔网

代码语言
代码分类
【CSharp】 C#读取HTML文件内容写入记事本
作者:Dezai.CN / 发布于2011/11/29/ 544
<div><span style="color: rgb(0,0,255)">try</span>
 {
 <span style="color: rgb(0,0,255)">int</span> totalFile = <span style="color: rgb(128,0,128)">0</span>;
 <span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">string dirPath = @"E:\chfuMetarnet\BSC6810 alarm\";</span><span style="color: rgb(0,128,0)">
</span> <span style="color: rgb(0,0,255)">if</span> (<span style="color: rgb(0,0,255)">this</span>.textBox1.Text.Trim() == <span style="color: rgb(128,0,0)">""</span>)
 {
 MessageBox.Show(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">请输入HTML文件路径！</span><span style="color: rgb(128,0,0)">"</span>);
 }
 <span style="color: rgb(0,0,255)">else</span>
 {
 <span style="color: rgb(0,0,255)">string</span> dirPath = <span style="color: rgb(0,0,255)">this</span>.textBox1.Text.Trim();
 <span style="color: rgb(0,0,255)">if</span> (!dirPath.Substring(dirPath.Length - <span style="color: rgb(128,0,128)">1</span>).Contains(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">\\</span><span style="color: rgb(128,0,0)">"</span>))
 {
 dirPath = dirPath+<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">\\</span><span style="color: rgb(128,0,0)">"</span>;
 }
 StreamWriter sw;
 DirectoryInfo dirInfo = <span style="color: rgb(0,0,255)">new</span> DirectoryInfo(dirPath);
 FileInfo[] files = dirInfo.GetFiles();
 <span style="color: rgb(0,0,255)">string</span> filename = dirPath + <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">告警经验库信息.txt</span><span style="color: rgb(128,0,0)">"</span>;
 <span style="color: rgb(0,0,255)">if</span> (File.Exists(filename))
 {
 sw = File.AppendText(filename);
 }
 <span style="color: rgb(0,0,255)">else</span>
 {
 sw = File.CreateText(filename);
 }
 <span style="color: rgb(0,0,255)">foreach</span> (FileInfo fileinfo <span style="color: rgb(0,0,255)">in</span> files)
 {
 <span style="color: rgb(0,0,255)">if</span> (fileinfo.Extension.Equals(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">.htm</span><span style="color: rgb(128,0,0)">"</span>))<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">遍历所有htm文件</span><span style="color: rgb(0,128,0)">
</span> {
 totalFile = totalFile + <span style="color: rgb(128,0,128)">1</span>;
 WebRequest myWebRequest = WebRequest.Create(dirPath + fileinfo.Name);
 WebResponse myWebResponse = myWebRequest.GetResponse();
 Stream myStream = myWebResponse.GetResponseStream();
 Encoding encode = System.Text.Encoding.GetEncoding(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">gb2312</span><span style="color: rgb(128,0,0)">"</span>);
 StreamReader myStreamReader = <span style="color: rgb(0,0,255)">new</span> StreamReader(myStream, encode);
 <span style="color: rgb(0,0,255)">string</span> strhtml = myStreamReader.ReadToEnd();
 myWebResponse.Close();
 <span style="color: rgb(0,0,255)">string</span> stroutput = strhtml;
 Regex regex = <span style="color: rgb(0,0,255)">new</span> Regex(<span style="color: rgb(128,0,0)">@"</span><span style="color: rgb(128,0,0)"><[^>]+>|</[^>]+></span><span style="color: rgb(128,0,0)">"</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉HTML标记的正则表达式</span><span style="color: rgb(0,128,0)">
</span> <span style="color: rgb(0,0,255)">string</span> tmpStr = <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"><h4>([^<]*)</h4></span><span style="color: rgb(128,0,0)">"</span>; <span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">获取<h4>之间内容的表达式</span><span style="color: rgb(0,128,0)">
</span> Match TitleMatch = Regex.Match(strhtml, tmpStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
 <span style="color: rgb(0,0,255)">string</span> causename = TitleMatch.Value.ToString();<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">包含<h4>和</h4>标记</span><span style="color: rgb(0,128,0)">
</span> causename = Regex.Replace(causename, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">[\n|\r|\t]</span><span style="color: rgb(128,0,0)">"</span>, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"> </span><span style="color: rgb(128,0,0)">"</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉换行和TAB键符号</span><span style="color: rgb(0,128,0)">
</span> causename = causename.Trim();
 <span style="color: rgb(0,0,255)">string</span> cause = causename.Substring(<span style="color: rgb(128,0,128)">4</span>, causename.Length - <span style="color: rgb(128,0,128)">9</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">得到告警原因</span><span style="color: rgb(0,128,0)">
</span> <span style="color: rgb(0,0,255)">string</span> titleStr = <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"><title>([^<]*)</title></span><span style="color: rgb(128,0,0)">"</span>;
 TitleMatch = Regex.Match(strhtml, titleStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
 <span style="color: rgb(0,0,255)">string</span> titlename = TitleMatch.Value.ToString();
 titlename = Regex.Replace(titlename, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">[\n|\r|\t]</span><span style="color: rgb(128,0,0)">"</span>, <span style="color: rgb(128,0,0)">""</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉换行和TAB键符号</span><span style="color: rgb(0,128,0)">
</span> titlename = titlename.Trim();
 <span style="color: rgb(0,0,255)">string</span> regexStr = <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"><ul><li>(?<key>.*?)</ul></span><span style="color: rgb(128,0,0)">"</span>;<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">获取<ul><li>后边的内容，直到</ul>结尾</span><span style="color: rgb(0,128,0)">
</span> Regex r = <span style="color: rgb(0,0,255)">new</span> Regex(regexStr, RegexOptions.None);
 strhtml = Regex.Replace(strhtml, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">[\n|\r|\t]</span><span style="color: rgb(128,0,0)">"</span>, <span style="color: rgb(128,0,0)">""</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉换行和TAB键符号</span><span style="color: rgb(0,128,0)">
</span> Match mc = r.Match(strhtml);
 <span style="color: rgb(0,0,255)">string</span> dataStr = mc.Groups[<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">key</span><span style="color: rgb(128,0,0)">"</span>].Value;
 dataStr = <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"><ul><li></span><span style="color: rgb(128,0,0)">"</span> + dataStr + <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"></ul></span><span style="color: rgb(128,0,0)">"</span>;<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">得到完整的<ul></ul>之间的源码</span><span style="color: rgb(0,128,0)">
</span> strhtml = strhtml.Replace(dataStr, <span style="color: rgb(128,0,0)">""</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">将去掉换行符和tab键的源码中去除<ul></ul>部分源码</span><span style="color: rgb(0,128,0)">
</span> strhtml = strhtml.Replace(titlename, <span style="color: rgb(128,0,0)">""</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉<title></title></span><span style="color: rgb(0,128,0)">
</span> strhtml = regex.Replace(strhtml, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)"> </span><span style="color: rgb(128,0,0)">"</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">过滤掉HTML标记</span><span style="color: rgb(0,128,0)">
</span> strhtml = strhtml.Replace(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">&nbsp;</span><span style="color: rgb(128,0,0)">"</span>, <span style="color: rgb(128,0,0)">""</span>);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">去掉空格字符</span><span style="color: rgb(0,128,0)">
</span> <span style="color: rgb(0,0,255)">string</span>[] arr = cause.Split(<span style="color: rgb(128,0,0)">'</span><span style="color: rgb(128,0,0)"> </span><span style="color: rgb(128,0,0)">'</span>);
 <span style="color: rgb(0,0,255)">string</span> zhCause = arr[arr.Length - <span style="color: rgb(128,0,128)">1</span>];<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">获取数组最后一个元素：告警原因</span><span style="color: rgb(0,128,0)">
</span> sw.WriteLine(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">第</span><span style="color: rgb(128,0,0)">"</span> + totalFile + <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">个文件：</span><span style="color: rgb(128,0,0)">"</span> + fileinfo.Name);
 sw.WriteLine(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">-----告警原因------：</span><span style="color: rgb(128,0,0)">"</span>);
 <span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">sw.WriteLine(cause);</span><span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)"> ALM-1 网元启动</span><span style="color: rgb(0,128,0)">
</span> zhCause=<span style="color: rgb(0,0,255)">this</span>.chinaString(zhCause);
 sw.WriteLine(zhCause);<span style="color: rgb(0,128,0)">//</span><span style="color: rgb(0,128,0)">网元启动</span><span style="color: rgb(0,128,0)">
</span> sw.WriteLine(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">-----处理经验------：</span><span style="color: rgb(128,0,0)">"</span>);
 sw.WriteLine(strhtml);
 sw.WriteLine();
 sw.Flush();

 }
 }
 sw.Close();
 MessageBox.Show(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">操作成功！</span><span style="color: rgb(128,0,0)">"</span>, <span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">提示</span><span style="color: rgb(128,0,0)">"</span>, MessageBoxButtons.OK, MessageBoxIcon.Information);
 }
 }
 <span style="color: rgb(0,0,255)">catch</span> (Exception ee)
 {
 MessageBox.Show(<span style="color: rgb(128,0,0)">"</span><span style="color: rgb(128,0,0)">操作失败：</span><span style="color: rgb(128,0,0)">"</span> + ee.Message);
 }
试试其它关键字
　读取HTML文件内容　
同语言下
可能有用的
Dezai.CN贡献的其它代码(4037)