/// <summary>
/// Returns a new IOpenSDKReader that can read the contents
/// of the provided format document. If the file format
/// is not supported, then an InvalidDataException will
/// be thrown. It is possible to do a pre-check using
/// the IsValidFile method.
/// </summary>
/// <param name="path"></param>
/// <returns></returns>
public static IOpenSDKReader GetNew(string path)
{
string extension = Path.GetExtension(path).ToUpper();
IOpenSDKReader reader = null;
if (_WordOpenSDKExtensions.Contains(extension))
{
reader = new WordReader(path);
return(reader);
}
if (_PowerPointOpenSDKExtensions.Contains(extension))
{
reader = new PowerPointReader(path);
return(reader);
}
if (_ExcelOpenSDKExtensions.Contains(extension))
{
reader = new ExcelReader(path);
return(reader);
}
throw new System.IO.InvalidDataException(NOT_VALID_FILE_MESSAGE);
}
public string ReadContents()
{
CheckInternalState();
string contents = string.Empty;
//Stopwatch stopwatch = Stopwatch.StartNew();
if (IsReadablePdf(_InternalFilePath))
{
contents = ReadPdfContents();
}
if (BinaryReader.IsValidFile(_InternalFilePath))
{
IBinaryReader reader = BinaryReader.GetNew(_InternalFilePath);
contents = reader.ReadContents();
}
if (OpenSDKReader.IsValidFile(_InternalFilePath))
{
IOpenSDKReader reader = OpenSDKReader.GetNew(_InternalFilePath);
contents = reader.ReadContents();
}
if (OutlookReader.IsValidFile(_InternalFilePath))
{
IOutlookReader reader = OutlookReader.GetNew(_InternalFilePath);
contents = reader.ReadContents();
if (IncludeEmbeddedFiles)
{
StringBuilder builder = new StringBuilder(contents);
foreach (string tempEmbeddedFile in reader.GetEmbeddedFiles())
{
//NOTE: the GetEmbeddedFiles method of OutlookDataReader creates and
//returns temp files for the attachments, if they exist, therefore
//the caller method is responsible for disposing of them here when
//done
IFileReader embeddedFileReader = new FileReader();
if (embeddedFileReader.IsReadable(tempEmbeddedFile))
{
embeddedFileReader.Open(tempEmbeddedFile);
builder.Append(embeddedFileReader.ReadContents());
}
try
{
File.Delete(tempEmbeddedFile);
}
catch (Exception) { }
}
contents = builder.ToString();
}
}
//stopwatch.Stop();
//if (stopwatch.Elapsed.TotalMinutes > 1)
//{
// //took a long time to read this file, log it
// System.IO.File.AppendAllText("C:\\temp\\crawler\\logs\\LongRunning.log",
// _InternalFilePath + "," + stopwatch.Elapsed.TotalMinutes.ToString("0.00") + Environment.NewLine);
//}
return(contents);
}
