private int ConvertFile(string sourcefile)
{
if (!File.Exists(sourcefile))
{
return(1);
}
try
{
TextReader reader = new FilterReader(sourcefile);
using (reader)
{
// char[] buffer = new char[0x5000];
// reader.Read(buffer, 0, 0x5000);
// string context = new string(buffer);
string context = reader.ReadToEnd();
calcWords(context);
context = Regex.Replace(context, "\n\r", " ", RegexOptions.IgnoreCase);
try
{
string name = GetFileName(sourcefile);
using (StreamWriter writer = new StreamWriter((outpath + @"\" + name + ".txt").Replace(@"\\", @"\"), false, Encoding.Default))
{
writer.Write(context);
writer.Close();
}
reader.Close();
return(0);
}
catch (Exception exception)
{
reader.Close();
return(3);
}
}
}
catch (Exception e)
{
return(2);
}
}
private void btnGetText_Click(object sender, EventArgs e)
{
string dbt_wfile = "";
OpenFileDialog dlg = new OpenFileDialog();
if (dlg.ShowDialog() == DialogResult.OK)
{
dbt_wfile = dlg.FileName;
TextReader reader = new FilterReader(dbt_wfile);
using (reader)
{
MessageBox.Show(reader.ReadToEnd());
}
reader.Close();
}
}
public static int ConvertFile()
{
if (!File.Exists(sourcefile))
{
return((int)OutStatus.FileLoss);
}
try
{
TextReader reader = new FilterReader(sourcefile);
using (reader)
{
// char[] buffer = new char[0x5000];
// reader.Read(buffer, 0, 0x5000);
// string context = new string(buffer);
string context = reader.ReadToEnd();
context = Regex.Replace(context, "\n\r", " ", RegexOptions.IgnoreCase);
try
{
string txtfile = (outtxtpath + @"\" + fileid + ".txt").Replace(@"\\", @"\");
using (StreamWriter writer = new StreamWriter(txtfile, false, Encoding.Default))
{
writer.Write(context);
writer.Close();
}
reader.Close();
ExecuteRegexTxt(txtfile);
return((int)OutStatus.ConvertSuccess);
}
catch (Exception exception)
{
Console.WriteLine("保存txt文件发生异常" + exception);
return((int)OutStatus.TotxtFailed);
}
}
}
catch (Exception e)
{
Console.WriteLine("打开文件失败" + e);
return((int)OutStatus.TotxtFailed);
}
}
private static void TryReadFile(FileInfo file)
{
var stream = file.OpenRead();
FilterReader reader = null;
try
{
FilterReaderOptions filterReaderOptions = new FilterReaderOptions();
reader = new FilterReader(stream, file.Extension, filterReaderOptions);
var result = reader.ReadToEnd();
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
finally
{
reader?.Close();
stream?.Close();
}
}
/// <summary>
/// Get the textual representation of the Binary Data of the document using IFILTER
/// </summary>
/// <returns>The text of the document or null if we could not parse the document into text</returns>
public virtual string GetTextFromDocumentBinary()
{
/*
* The default is to save the binary data to a temporary location and
* use IFilter to extract the text. This should be a good catch-all for
* all files that don't have a specific mechanism for extracting the
* text of the file.
*/
// If we have no bytes then we can't do anything.
if (Bytes == null || Bytes.Length == 0)
{
// Log the problem.
log.Error("Tried to extract text from empty bytes for file " + Name);
return(null);
}
// Get the original file name without the extension
string fileNameWithoutExtension = Path.GetFileNameWithoutExtension(Name);
bool success = false;
string newFileName = "";
try
{
// Now try to generate a new temporary file name that we don't have in the temporary directory
for (int i = 0; i < 50; i++)
{
Random rand = new Random();
newFileName = "~/TemporaryFilesDirectory/" + fileNameWithoutExtension +
Convert.ToString(rand.Next(100000)) + Extension;
newFileName = HttpContext.Current.Server.MapPath(newFileName);
// Try to see if this file exists
if (!File.Exists(newFileName))
{
success = true;
break;
}
}
if (!success)
{
// We failed. Log the problem.
log.Error("Failed to create a unique file to extract data. Last file tried is " + newFileName);
return(null);
}
}
catch (Exception e)
{
// We failed. Log the problem.
log.Error("Failed to create a unique file to extract data for file " + Name, e);
return(null);
}
FileStream theFileStream = null;
try
{
// Now try to write the bytes to the newly created file
theFileStream = File.Create(newFileName);
theFileStream.Write(Bytes, 0, Bytes.Length);
theFileStream.Close();
}
catch (Exception e)
{
// We failed to write the file. Log the problem
log.Error("Failed to write bytes to new file " + newFileName, e);
// Try to close the stream, in case it is still open and delete the file
try
{
if (theFileStream != null)
{
theFileStream.Close();
}
if (File.Exists(newFileName))
{
File.Delete(newFileName);
}
}
catch
{
// We don't do anything. This is a best effort close and delete
}
return(null);
}
string text = null;
FilterReader myFilterReader = null;
// Now try to extract the text for the file
try
{
myFilterReader = new FilterReader(newFileName);
text = myFilterReader.ReadToEnd();
myFilterReader.Close();
}
catch (Exception e)
{
log.Error("Failed to parse text for file " + Name + " using IFilter", e);
// Try to close the IFilter, in case it is still open
try
{
if (myFilterReader != null)
{
myFilterReader.Close();
}
}
catch
{
// We don't do anything. This is a best effort close.
}
}
try
{
// Try to delete the temporary file
if (File.Exists(newFileName))
{
File.Delete(newFileName);
}
}
catch (Exception e)
{
log.Error("Failed to delete temporary file " + newFileName, e);
}
return(text);
}