C#讀取HTML檔案內容寫入記事本

C#讀取HTML檔案內容寫入記事本

try
            {
int totalFile = 0;
//string dirPath = @"E:\chfuMetarnet\BSC6810 alarm\";
                if (this.textBox1.Text.Trim() == "")
                {
                    MessageBox.Show("請輸入HTML檔案路徑！");
                }
else
                {
string dirPath = this.textBox1.Text.Trim();
if (!dirPath.Substring(dirPath.Length - 1).Contains("\\"))
                    {
                        dirPath = dirPath+"\\";
                    }
                    StreamWriter sw;
                    DirectoryInfo dirInfo = new DirectoryInfo(dirPath);
                    FileInfo[] files = dirInfo.GetFiles();
string filename = dirPath + "警示經驗庫資訊.txt";
if (File.Exists(filename))
                    {
                        sw = File.AppendText(filename);
                    }
else
                    {
                        sw = File.CreateText(filename);
                    }
foreach (FileInfo fileinfo in files)
                    {
if (fileinfo.Extension.Equals(".htm"))//遍曆所有htm檔案
                        {
                            totalFile = totalFile + 1;
                            WebRequest myWebRequest = WebRequest.Create(dirPath + fileinfo.Name);
                            WebResponse myWebResponse = myWebRequest.GetResponse();
                            Stream myStream = myWebResponse.GetResponseStream();
                            Encoding encode = System.Text.Encoding.GetEncoding("gb2312");
                            StreamReader myStreamReader = new StreamReader(myStream, encode);
string strhtml = myStreamReader.ReadToEnd();
                            myWebResponse.Close();
string stroutput = strhtml;
                            Regex regex = new Regex(@"<[^>]+>|</[^>]+>");//去掉HTML標記的Regex
                            string tmpStr = "<h4>([^<]*)</h4>";        //擷取<h4>之間內容的運算式
                            Match TitleMatch = Regex.Match(strhtml, tmpStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
string causename = TitleMatch.Value.ToString();//包含<h4>和</h4>標記
                            causename = Regex.Replace(causename, "[\n|\r|\t]", " ");//去掉換行和TAB鍵符號
                            causename = causename.Trim();
string cause = causename.Substring(4, causename.Length - 9);//得到警示原因
                            string titleStr = "<title>([^<]*)</title>";
                            TitleMatch = Regex.Match(strhtml, titleStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
string titlename = TitleMatch.Value.ToString();
                            titlename = Regex.Replace(titlename, "[\n|\r|\t]", "");//去掉換行和TAB鍵符號
                            titlename = titlename.Trim();
string regexStr = "<ul><li>(?<key>.*?)</ul>";//擷取<ul><li>後邊的內容，直到</ul>結尾
                            Regex r = new Regex(regexStr, RegexOptions.None);
                            strhtml = Regex.Replace(strhtml, "[\n|\r|\t]", "");//去掉換行和TAB鍵符號
                            Match mc = r.Match(strhtml);
string dataStr = mc.Groups["key"].Value;
                            dataStr = "<ul><li>" + dataStr + "</ul>";//得到完整的<ul></ul>之間的源碼
                            strhtml = strhtml.Replace(dataStr, "");//將去掉分行符號和tab鍵的源碼中去除<ul></ul>部分源碼
                            strhtml = strhtml.Replace(titlename, "");//去掉<title></title>
                            strhtml = regex.Replace(strhtml, " ");//過濾掉HTML標記
                            strhtml = strhtml.Replace("&nbsp;", "");//去掉空白字元
                            string[] arr = cause.Split(' ');
string zhCause = arr[arr.Length - 1];//擷取數組最後一個元素：警示原因
                            sw.WriteLine("第" + totalFile + "個檔案：" + fileinfo.Name);
                            sw.WriteLine("-----警示原因------：");
//sw.WriteLine(cause);// ALM-1 網元啟動
                            zhCause=this.chinaString(zhCause);
                            sw.WriteLine(zhCause);//網元啟動
                            sw.WriteLine("-----處理經驗------：");
                            sw.WriteLine(strhtml);
                            sw.WriteLine();
                            sw.Flush();

                        }
                    }
                    sw.Close();
                    MessageBox.Show("操作成功！", "提示", MessageBoxButtons.OK, MessageBoxIcon.Information);
                }
            }
catch (Exception ee)
            {
                MessageBox.Show("操作失敗：" + ee.Message);
            }