private void ProcessFile(string filename, Extractor item)
{
m_stderr.WriteLine("Processing " + filename);
try
{
int flags = isys_docfilters.IGR_BODY_AND_META;
if (m_html)
{
flags |= isys_docfilters.IGR_FORMAT_HTML;
}
item.Open(flags);
// Extract the text and return it to stdout
while (!item.getEOF())
{
String t = item.GetText(4096);
// Cleanup the text
t = t.Replace('\u000E', '\n');
t = t.Replace('\r', '\n');
m_stdout.Write(t);
}
m_stdout.WriteLine("");
// Extract the HTML generated images into the current folder
if (m_html)
{
SubFile image = item.GetFirstImage();
while (image != null)
{
m_stderr.WriteLine("Extracting image " + image.getName());
image.CopyTo(image.getName());
image.Close();
// Move onto the next image
image = item.GetNextImage();
}
}
// Extract the sub files (if any) and process recursively
if (m_subFiles)
{
SubFile child = item.GetFirstSubFile();
while (child != null)
{
ProcessFile(filename + ">" + child.getName(), child);
// Move onto the next sub file
child = item.GetNextSubFile();
}
}
}
catch (Exception e)
{
m_stderr.WriteLine("Error Processing " + filename);
m_stderr.WriteLine(" - " + e.ToString());
}
finally
{
item.Close();
}
}
