/// <summary>
/// Advanced OCR fixing via replace/spelling dictionaries + some hardcoded rules
/// </summary>
/// <param name="threeLetterIsoLanguageName">E.g. eng for English</param>
/// <param name="hunspellName">Name of hunspell dictionary</param>
/// <param name="parentForm">Used for centering/show spell check dialog</param>
public OcrFixEngine(string threeLetterIsoLanguageName, string hunspellName, Form parentForm)
{
if (threeLetterIsoLanguageName == "per")
threeLetterIsoLanguageName = "fas";
_threeLetterIsoLanguageName = threeLetterIsoLanguageName;
_parentForm = parentForm;
_spellCheck = new OcrSpellCheck { StartPosition = FormStartPosition.Manual };
_spellCheck.Location = new Point(parentForm.Left + (parentForm.Width / 2 - _spellCheck.Width / 2),
parentForm.Top + (parentForm.Height / 2 - _spellCheck.Height / 2));
_ocrFixReplaceList = OcrFixReplaceList.FromLanguageId(threeLetterIsoLanguageName);
LoadSpellingDictionaries(threeLetterIsoLanguageName, hunspellName); // Hunspell etc.
AutoGuessesUsed = new List<string>();
UnknownWordsFound = new List<string>();
}
public void OcrFixReplaceListAddWord()
{
// Arrange
string fileName = Path.Combine(Directory.GetCurrentDirectory(), Guid.NewGuid() + ".xml");
var fixList = new OcrFixReplaceList(fileName);
fixList.WordReplaceList.Clear();
// Act
fixList.AddWordOrPartial("from", "to");
// Assert
Assert.IsTrue(fixList.WordReplaceList["from"] == "to");
// Clean up
try
{
File.Delete(fileName);
}
catch
{
}
}
public void OcrFixReplaceListRemovePartialLineReload()
{
// Arrange
string fileName = Path.Combine(Directory.GetCurrentDirectory(), Guid.NewGuid() + ".xml");
var fixList = new OcrFixReplaceList(fileName);
fixList.PartialLineWordBoundaryReplaceList.Clear();
fixList.AddWordOrPartial("from me", "to you");
fixList = new OcrFixReplaceList(fileName);
fixList.RemoveWordOrPartial("from me");
// Act
fixList = new OcrFixReplaceList(fileName);
// Assert
Assert.IsTrue(!fixList.WordReplaceList.ContainsKey("from me"));
// Clean up
try
{
File.Delete(fileName);
}
catch
{
}
}
private void LoadOcrFixList(bool reloadListBox)
{
var cb = comboBoxWordListLanguage.Items[comboBoxWordListLanguage.SelectedIndex] as ComboBoxLanguage;
if (cb == null)
return;
if (reloadListBox)
listBoxOcrFixList.Items.Clear();
_ocrFixReplaceList = OcrFixReplaceList.FromLanguageId(cb.CultureInfo.ThreeLetterISOLanguageName);
if (reloadListBox)
{
listBoxOcrFixList.BeginUpdate();
foreach (var pair in _ocrFixReplaceList.WordReplaceList)
{
listBoxOcrFixList.Items.Add(pair.Key + " --> " + pair.Value);
}
foreach (var pair in _ocrFixReplaceList.PartialLineWordBoundaryReplaceList)
{
listBoxOcrFixList.Items.Add(pair.Key + " --> " + pair.Value);
}
listBoxOcrFixList.Sorted = true;
listBoxOcrFixList.EndUpdate();
}
}