C#.NET实现Word或Excel文件转为HTML文件

Word文件转html，返回相对路径

1 private string GetPathByDocToHTML(string strFile)
2     {
3         if (string.IsNullOrEmpty(strFile))
4         {
5             return "0";//没有文件
6         }
7
8         Microsoft.Office.Interop.Word.ApplicationClass word = new Microsoft.Office.Interop.Word.ApplicationClass();
9         Type wordType = word.GetType();
10         Microsoft.Office.Interop.Word.Documents docs = word.Documents;
11
12         // 打开文件
13         Type docsType = docs.GetType();
14
15         object fileName = strFile;
16
17         Microsoft.Office.Interop.Word.Document doc = (Microsoft.Office.Interop.Word.Document)docsType.InvokeMember("Open",
18         System.Reflection.BindingFlags.InvokeMethod, null, docs, new Object[] { fileName, true, true });
19
20         // 转换格式，另存为html
21         Type docType = doc.GetType();
22         //给文件重新起名
23         string filename = System.DateTime.Now.Year.ToString() + System.DateTime.Now.Month.ToString() + System.DateTime.Now.Day.ToString() +
24         System.DateTime.Now.Hour.ToString() + System.DateTime.Now.Minute.ToString() + System.DateTime.Now.Second.ToString();
25
26         string strFileFolder = "../html/";
27         DateTime dt = DateTime.Now;
28         //以yyyymmdd形式生成子文件夹名
29         string strFileSubFolder = dt.Year.ToString();
30         strFileSubFolder += (dt.Month < 10) ? ("0" + dt.Month.ToString()) : dt.Month.ToString();
31         strFileSubFolder += (dt.Day < 10) ? ("0" + dt.Day.ToString()) : dt.Day.ToString();
32         string strFilePath = strFileFolder + strFileSubFolder + "/";
33         // 判断指定目录下是否存在文件夹，如果不存在，则创建
34         if (!Directory.Exists(Server.MapPath(strFilePath)))
35         {
36             // 创建up文件夹
37             Directory.CreateDirectory(Server.MapPath(strFilePath));
38         }
39
40         //被转换的html文档保存的位置
41         // HttpContext.Current.Server.MapPath("html" + strFileSubFolder + filename + ".html")
42         string ConfigPath = Server.MapPath(strFilePath + filename + ".html");
43         object saveFileName = ConfigPath;
44
45         /*下面是Microsoft Word 9 Object Library的写法，如果是10，可能写成：
46           * docType.InvokeMember("SaveAs", System.Reflection.BindingFlags.InvokeMethod,
47           * null, doc, new object[]{saveFileName, Word.WdSaveFormat.wdFormatFilteredHTML});
48           * 其它格式：
49           * wdFormatHTML
50           * wdFormatDocument
51           * wdFormatDOSText
52           * wdFormatDOSTextLineBreaks
53           * wdFormatEncodedText
54           * wdFormatRTF
55           * wdFormatTemplate
56           * wdFormatText
57           * wdFormatTextLineBreaks
58           * wdFormatUnicodeText
59         */
60         docType.InvokeMember("SaveAs", System.Reflection.BindingFlags.InvokeMethod,
61         null, doc, new object[] { saveFileName, Microsoft.Office.Interop.Word.WdSaveFormat.wdFormatFilteredHTML });
62
63         //docType.InvokeMember("SaveAs", System.Reflection.BindingFlags.InvokeMethod,
64         //  null, doc, new object[] { saveFileName, Microsoft.Office.Interop.Word.WdSaveFormat.wdFormatFilteredHTML });
65
66         //关闭文档
67         docType.InvokeMember("Close", System.Reflection.BindingFlags.InvokeMethod,
68         null, doc, new object[] { null, null, null });
69
70         // 退出 Word
71         wordType.InvokeMember("Quit", System.Reflection.BindingFlags.InvokeMethod, null, word, null);
72         //转到新生成的页面
73         //return ("/" + filename + ".html");
74
75         //转化HTML页面统一编码格式
76         TransHTMLEncoding(ConfigPath);
77
78         return (strFilePath + filename + ".html");
79     }

Excel文件转HTML，返回相对路径

1 private string GetPathByXlsToHTML(string strFile)
2     {
3         if (string.IsNullOrEmpty(strFile))
4         {
5             return "0";//没有文件
6         }
7
8         //实例化Excel
9         Microsoft.Office.Interop.Excel.Application repExcel = new Microsoft.Office.Interop.Excel.Application();
10         Microsoft.Office.Interop.Excel.Workbook workbook = null;
11         Microsoft.Office.Interop.Excel.Worksheet worksheet = null;
12
13         //打开文件，n.FullPath是文件路径
14         workbook = repExcel.Application.Workbooks.Open(strFile, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing);
15         worksheet = (Microsoft.Office.Interop.Excel.Worksheet)workbook.Worksheets[1];
16
17         //给文件重新起名
18         string filename = System.DateTime.Now.Year.ToString() + System.DateTime.Now.Month.ToString() + System.DateTime.Now.Day.ToString() +
19         System.DateTime.Now.Hour.ToString() + System.DateTime.Now.Minute.ToString() + System.DateTime.Now.Second.ToString();
20
21         string strFileFolder = "../html/";
22         DateTime dt = DateTime.Now;
23         //以yyyymmdd形式生成子文件夹名
24         string strFileSubFolder = dt.Year.ToString();
25         strFileSubFolder += (dt.Month < 10) ? ("0" + dt.Month.ToString()) : dt.Month.ToString();
26         strFileSubFolder += (dt.Day < 10) ? ("0" + dt.Day.ToString()) : dt.Day.ToString();
27         string strFilePath = strFileFolder + strFileSubFolder + "/";
28         // 判断指定目录下是否存在文件夹，如果不存在，则创建
29         if (!Directory.Exists(Server.MapPath(strFilePath)))
30         {
31             // 创建up文件夹
32             Directory.CreateDirectory(Server.MapPath(strFilePath));
33         }
34         string ConfigPath = Server.MapPath(strFilePath + filename + ".html");
35         object savefilename = (object)ConfigPath;
36
37         object ofmt = Microsoft.Office.Interop.Excel.XlFileFormat.xlHtml;
38         //进行另存为操作
39         workbook.SaveAs(savefilename, ofmt, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Microsoft.Office.Interop.Excel.XlSaveAsAccessMode.xlNoChange, Type.Missing, Type.Missing, Type.Missing, Type.Missing, Type.Missing);
40         object osave = false;
41         //逐步关闭所有使用的对象
42         workbook.Close(osave, Type.Missing, Type.Missing);
43         repExcel.Quit();
44         System.Runtime.InteropServices.Marshal.ReleaseComObject(worksheet);
45         worksheet = null;
46         //垃圾回收
47         GC.Collect();
48         System.Runtime.InteropServices.Marshal.ReleaseComObject(workbook);
49         workbook = null;
50         GC.Collect();
51         System.Runtime.InteropServices.Marshal.ReleaseComObject(repExcel.Application.Workbooks);
52         GC.Collect();
53         System.Runtime.InteropServices.Marshal.ReleaseComObject(repExcel);
54         repExcel = null;
55         GC.Collect();
56         //依据时间杀灭进程
57         System.Diagnostics.Process[] process = System.Diagnostics.Process.GetProcessesByName("EXCEL");
58         foreach (System.Diagnostics.Process p in process)
59         {
60             if (DateTime.Now.Second - p.StartTime.Second > 0 && DateTime.Now.Second - p.StartTime.Second < 5)
61             {
62                 p.Kill();
63             }
64         }
65
66         return (strFilePath + filename + ".html");
67     }

这里可能会遇到一个问题，由于转化为HTML文件的页面编码可能使得浏览器无法正确解读，所以需要转码，转换代码如下：

1     private void TransHTMLEncoding(string strFilePath)
2     {
3         try
4         {
5             System.IO.StreamReader sr = new System.IO.StreamReader(strFilePath, Encoding.GetEncoding(0));
6             string html = sr.ReadToEnd();
7             sr.Close();
8             html = System.Text.RegularExpressions.Regex.Replace(html, @"<meta[^>]*>", "<meta http-equiv=Content-Type content='text/html; charset=gb2312'>", System.Text.RegularExpressions.RegexOptions.IgnoreCase);
9             System.IO.StreamWriter sw = new System.IO.StreamWriter(strFilePath, false, Encoding.Default);
10
11             sw.Write(html);
12             sw.Close();
13         }
14         catch (Exception ex)
15         {
16             Page.RegisterStartupScript("alt", "<script>alert('" + ex.Message + "')</script>");
17         }
18     }

这样就可以正常在页面上正常显示了