private void ContinueDisplayingResults()
{
List <string> KnownCommodityNames;
Dictionary <String, String> EconomyLevels;
try
{
KnownCommodityNames = Program.Data.getCommodityNames();
EconomyLevels = Program.Data.getEconomyLevels();
do
{
_correctionRow++;
if (_correctionRow > _commodityTexts.GetLength(0) - 1)
{
_correctionRow = 0; _correctionColumn++;
}
if (_commodityTexts.GetLength(0) == 0)
{
return;
}
if (_correctionColumn < _commodityTexts.GetLength(1))
{
//Debug.WriteLine(correctionRow + " - " + correctionColumn);
pbOcrCurrent.Image = _originalBitmaps[_correctionRow, _correctionColumn];
}
if (_correctionColumn == 0) // hacks for commodity name
{
var currentTextCamelCase =
_parent._textInfo.ToTitleCase(_commodityTexts[_correctionRow, _correctionColumn].ToLower()); // There *was* a reason why I did this...
// if the ocr have found no char so we dont need to ask Mr. Levenshtein
if (currentTextCamelCase.Trim().Length > 0)
{
if (KnownCommodityNames.Contains(
currentTextCamelCase))
{
_originalBitmapConfidences[_correctionRow, _correctionColumn] = 1;
}
else
{
var replacedCamelCase = StripPunctuationFromScannedText(currentTextCamelCase); // ignore spaces when using levenshtein to find commodity names
var lowestLevenshteinNumber = 10000;
var nextLowestLevenshteinNumber = 10000;
var lowestMatchingCommodity = "";
var lowestMatchingCommodityRef = "";
double LevenshteinLimit = 0;
foreach (var reference in KnownCommodityNames)
{
var upperRef = StripPunctuationFromScannedText(reference);
var levenshteinNumber = _parent._levenshtein.LD2(upperRef, replacedCamelCase);
//if(levenshteinNumber != _levenshtein.LD(upperRef, replacedCamelCase))
// Debug.WriteLine("Doh!");
if (upperRef != lowestMatchingCommodityRef)
{
if (levenshteinNumber < lowestLevenshteinNumber)
{
nextLowestLevenshteinNumber = lowestLevenshteinNumber;
lowestLevenshteinNumber = levenshteinNumber;
lowestMatchingCommodityRef = upperRef;
lowestMatchingCommodity = reference.ToUpper();
}
else if (levenshteinNumber < nextLowestLevenshteinNumber)
{
nextLowestLevenshteinNumber = levenshteinNumber;
}
}
}
// it's better if this depends on the length of the word - this factor works pretty good
LevenshteinLimit = Math.Round((currentTextCamelCase.Length * 0.7), 0);
if (lowestLevenshteinNumber <= LevenshteinLimit)
{
_originalBitmapConfidences[_correctionRow, _correctionColumn] = .9f;
_commodityTexts[_correctionRow, _correctionColumn] = lowestMatchingCommodity;
}
if (lowestLevenshteinNumber <= LevenshteinLimit && lowestLevenshteinNumber + 3 < nextLowestLevenshteinNumber) // INDIUM versus INDITE... could factor length in here
{
_originalBitmapConfidences[_correctionRow, _correctionColumn] = 1;
}
}
if (_commodityTexts[_correctionRow, _correctionColumn].Equals("Getreide", StringComparison.InvariantCultureIgnoreCase))
{
Debug.Print("STOP");
}
if (_commoditiesSoFar.Contains(_commodityTexts[_correctionRow, _correctionColumn].ToUpper()))
{
_commodityTexts[_correctionRow, _correctionColumn] = "";
_originalBitmapConfidences[_correctionRow, _correctionColumn] = 1;
}
// If we're doing a batch of screenshots, don't keep doing the same commodity when we keep finding it
// but only if it's sure - otherwise it will be registered later
if (_originalBitmapConfidences[_correctionRow, _correctionColumn] == 1)
{
_commoditiesSoFar.Add(_commodityTexts[_correctionRow, _correctionColumn].ToUpper());
}
}
else
{
// that was nothing
_originalBitmapConfidences[_correctionRow, _correctionColumn] = 1;
_commodityTexts[_correctionRow, _correctionColumn] = "";
}
}
else if (_correctionColumn == 5 || _correctionColumn == 7) // hacks for LOW/MED/HIGH
{
var commodityLevelUpperCase = StripPunctuationFromScannedText(_commodityTexts[_correctionRow, _correctionColumn]);
var levenshteinLow = _parent._levenshtein.LD2(EconomyLevels["LOW"].ToUpper(), commodityLevelUpperCase);
var levenshteinMed = _parent._levenshtein.LD2(EconomyLevels["MED"].ToUpper(), commodityLevelUpperCase);
var levenshteinHigh = _parent._levenshtein.LD2(EconomyLevels["HIGH"].ToUpper(), commodityLevelUpperCase);
var levenshteinBlank = _parent._levenshtein.LD2("", commodityLevelUpperCase);
//Pick the lowest levenshtein number
var lowestLevenshtein = Math.Min(Math.Min(levenshteinLow, levenshteinMed), Math.Min(levenshteinHigh, levenshteinBlank));
if (lowestLevenshtein == levenshteinLow)
{
_commodityTexts[_correctionRow, _correctionColumn] = EconomyLevels["LOW"];
}
else if (lowestLevenshtein == levenshteinMed)
{
_commodityTexts[_correctionRow, _correctionColumn] = EconomyLevels["MED"];
}
else if (lowestLevenshtein == levenshteinHigh)
{
_commodityTexts[_correctionRow, _correctionColumn] = EconomyLevels["HIGH"];
}
else // lowestLevenshtein == levenshteinBlank
{
_commodityTexts[_correctionRow, _correctionColumn] = "";
}
// we will never be challenged on low/med/high again. this doesn't get internationalized on foreign-language installs... does it? :)
_originalBitmapConfidences[_correctionRow, _correctionColumn] = 1;
}
}
// Don't pause for cells which have a high confidence, or have no commodity name
// ReSharper disable once CompareOfFloatsByEqualityOperator
while (_correctionColumn < _commodityTexts.GetLength(1) && (_originalBitmapConfidences[_correctionRow, _correctionColumn] > .9f || _originalBitmapConfidences[_correctionRow, _correctionColumn] == 0 || _commodityTexts[_correctionRow, 0] == ""));
if (_correctionColumn < _commodityTexts.GetLength(1))
{
// doing again some stateful enabling
tbCommoditiesOcrOutput.Text = _commodityTexts[_correctionRow, _correctionColumn];
tbConfidence.Text = _originalBitmapConfidences[_correctionRow, _correctionColumn].ToString(CultureInfo.InvariantCulture);
bContinueOcr.Enabled = true;
bIgnoreTrash.Enabled = true;
}
else
{
bContinueOcr.Enabled = false;
bIgnoreTrash.Enabled = false;
string finalOutput = _csvOutputSoFar;
for (int row = 0; row < _commodityTexts.GetLength(0); row++)
{
if (_commodityTexts[row, 0] != "") // don't create CSV if there's no commodity name
{
finalOutput += tbOcrSystemName.Text + ";" + tbOcrStationName.Text + ";";
for (int col = 0; col < _commodityTexts.GetLength(1); col++)
{
_commodityTexts[row, col] = _commodityTexts[row, col].Replace("\r", "").Replace("\n", "");
if (col == 3)
{
continue; // don't export cargo levels
}
finalOutput += _commodityTexts[row, col] + ";";
}
finalOutput += ocr.CurrentScreenshotDateTime.ToString("s").Substring(0, 16) + ";";
//if (cbExtendedInfoInCSV.Checked)
finalOutput += Path.GetFileName(_screenshotName) + ";";
finalOutput += _rowIds[row] + "\r\n";
}
}
_csvOutputSoFar += finalOutput;
if (pbOriginalImage.Image != null)
{
pbOriginalImage.Image.Dispose();
}
UpdateOriginalImage(null);
UpdateTrimmedImage(null, null);
if (Program.DBCon.getIniValue <Boolean>(IBE.IBESettingsView.DB_GROUPNAME, "DeleteScreenshotOnImport"))
{
File.Delete(_screenshotName);
}
Acquisition();
}
}
catch (Exception ex)
{
CErr.processError(ex);
}
}
private void bContinueOcr_Click(object sender, EventArgs e)
{
Boolean isOK = false;
Boolean finished = false;
DialogResult Answer;
string commodity;
List <string> KnownCommodityNames;
commodity = _parent._textInfo.ToTitleCase(tbCommoditiesOcrOutput.Text.ToLower().Trim());
KnownCommodityNames = Program.Data.getCommodityNames();
if (commodity.ToUpper() == "Implausible Results!".ToUpper())
{
// check results
var f = new EditOcrResults(tbFinalOcrOutput.Text);
f.onlyImplausible = true;
var q = f.ShowDialog();
if (q == DialogResult.OK)
{
tbFinalOcrOutput.Text = f.ReturnValue;
}
Acquisition(true);
isOK = false;
}
else if (commodity.ToUpper() == "Imported!".ToUpper() || commodity.ToUpper() == "Finished!".ToUpper() || commodity.ToUpper() == "No rows found...".ToUpper())
{
// its the end
isOK = true;
finished = true;
}
else if (commodity.Length == 0 || KnownCommodityNames.Contains(commodity))
{
// ok, no typing error
isOK = true;
}
else
{
// unknown commodity, is it a new one or a typing error ?
Answer = MsgBox.Show(String.Format("Do you want to add '{0}' to the known commodities ?", commodity), "Unknown commodity !",
MessageBoxButtons.OKCancel, MessageBoxIcon.Question, MessageBoxDefaultButton.Button1);
if (Answer == System.Windows.Forms.DialogResult.OK)
{
// yes, it's really new
//Program.Data.ImportCommodity(commodity);
throw new NotImplementedException();
//_Milkyway.addLocalized2RN(_commodities.Names);
isOK = true;
}
}
if (isOK)
{
if (_commodityTexts == null || _correctionColumn >= _commodityTexts.GetLength(1) || finished)
{
if (MsgBox.Show("Import this?", "Import?", MessageBoxButtons.YesNo) == DialogResult.Yes)
{
ImportFinalOcrOutput();
tbFinalOcrOutput.Text = "";
bContinueOcr.Enabled = false;
bIgnoreTrash.Enabled = false;
_commoditiesSoFar = new List <string>();
bClearOcrOutput.Enabled = false;
bEditResults.Enabled = false;
}
}
else
{
_commodityTexts[_correctionRow, _correctionColumn] = commodity.ToUpper();
_commoditiesSoFar.Add(_commodityTexts[_correctionRow, _correctionColumn].ToUpper());
ContinueDisplayingResults();
}
}
}