private static void ThreadOCRMainEntry(object arg)
{
string fname = "???";
List <int> pgnums;
int pgnum = 0;
try
{
lock (global_vars_access_lock)
{
fname = pdf_filename;
pgnums = page_numbers;
}
Dictionary <int, WordList> word_lists = new Dictionary <int, WordList>();
Dictionary <int, bool> page_ocr_successes = new Dictionary <int, bool>();
foreach (var p in pgnums)
{
pgnum = p;
WordList word_list = DoOCR(fname, pgnum);
Logging.Info("We have an OCR word list of length {0} for page {1}", word_list?.Count, pgnum);
// Check that we have something to write
if (null != word_list && word_list.Count > 0)
{
word_lists[pgnum] = word_list;
page_ocr_successes[pgnum] = true;
}
else
{
// FAKE a word list to shut up Qiqqa for the time being!
word_lists[pgnum] = FakeEngine.ConvertToWordList(); // new WordList();
page_ocr_successes[pgnum] = false;
}
}
// Check that we have something to write
Logging.Info("Writing OCR to file {0}", ocr_output_filename);
WordList.WriteToFile(ocr_output_filename, word_lists, page_ocr_successes[pgnums[0]] ? "OCR" : "OCR-Failed");
}
catch (Exception ex)
{
Logging.Error(ex, "Problem while doing OCR for file {0} @ page {1}", fname, pgnum);
lock (global_vars_access_lock)
{
exception_ocr = ex;
}
}
finally
{
lock (global_vars_access_lock)
{
has_exited_ocr = true;
}
}
}
/// <summary>
/// Arguments are:
///
/// 1) mode: GROUP
/// 2) pdf_filename
/// 3) page number(s) - comma separated
/// 4) ocr_output_filename - where the extracted word list info is stored
/// 5) pdf_user_password - encrypted
/// 6) language - (unused)
///
/// or:
///
/// 1) mode: SINGLE
/// 2) pdf_filename
/// 3) page number - only one page per run
/// 4) ocr_output_filename - where the extracted word list info is stored
/// 5) pdf_user_password - encrypted
/// 6) language - default is 'eng'
///
/// or:
///
/// 1) mode: SINGLE-FAKE
/// 2) pdf_filename
/// 3) page number - only one page per run
/// 4) ocr_output_filename - where the extracted word list info is stored
///
/// 7) NOKILL (optional)
/// </summary>
/// <param name="args"></param>
/// <returns></returns>
private static int Main(string[] args)
{
// This is used to return any errors to the OS
int exit_code = 0;
bool no_kill = (args.Length > 6 && 0 == args[6].ToUpper().CompareTo("NOKILL"));
try
{
Thread.CurrentThread.Name = "Main";
// Check that we were given the right number of parameters
if (args.Length < 1)
{
throw new Exception("Not enough command line arguments");
}
string mode_switch = args[0];
switch (mode_switch)
{
case "GROUP":
TextExtractEngine.MainEntry(args, no_kill);
break;
case "SINGLE":
OCREngine.MainEntry(args, no_kill);
break;
case "SINGLE-FAKE":
FakeEngine.MainEntry(args, no_kill);
break;
default:
throw new Exception("Unknown mode switch: " + mode_switch);
}
}
catch (Exception ex)
{
StringBuilder sb = new StringBuilder();
sb.AppendLine("--- Parameters ---");
foreach (string arg in args)
{
sb.Append(arg);
sb.Append(" ");
}
sb.AppendLine();
sb.AppendLine("--- Exception ---");
sb.AppendLine(ex.ToString());
Logging.Error("There was an error in QiqqaOCR:\n{0}", sb.ToString());
exit_code = -1;
}
// Check if we should exit
if (no_kill)
{
Logging.Error("PAUSED");
Console.ReadKey();
}
// This must be the last line the application executes, EVAR!
Logging.ShutDown();
return(exit_code);
}