private void buttonOK_Click(object sender, EventArgs e)
{
string key = textBoxOcrFixKey.Text.Trim();
string value = textBoxOcrFixValue.Text.Trim();
if (key.Length == 0 || value.Length == 0 || key == value)
{
return;
}
var languageString = LanguageString;
if (languageString == null)
{
return;
}
try
{
var ci = new CultureInfo(languageString.Replace("_", "-"));
_threeLetterIsoLanguageName = ci.ThreeLetterISOLanguageName;
}
catch (CultureNotFoundException exception)
{
MessageBox.Show(exception.Message);
return;
}
var ocrFixReplaceList = OcrFixReplaceList.FromLanguageId(_threeLetterIsoLanguageName);
ocrFixReplaceList.AddWordOrPartial(key, value);
DialogResult = DialogResult.OK;
NewSource = key;
NewTarget = value;
}
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)
{
if (comboBoxWordListLanguage.Items.Count == 0 || !(comboBoxWordListLanguage.Items[comboBoxWordListLanguage.SelectedIndex] is Settings.ComboBoxLanguage cb))
{
return;
}
_ocrFixReplaceList = OcrFixReplaceList.FromLanguageId(cb.CultureInfo.GetThreeLetterIsoLanguageName());
if (reloadListBox)
{
listBoxOcrFixList.BeginUpdate();
listBoxOcrFixList.Items.Clear();
listBoxOcrFixList.Items.AddRange(_ocrFixReplaceList.WordReplaceList.Select(p => p.Key + " --> " + p.Value).ToArray<object>());
listBoxOcrFixList.Items.AddRange(_ocrFixReplaceList.PartialLineWordBoundaryReplaceList.Select(p => p.Key + " --> " + p.Value).ToArray<object>());
listBoxOcrFixList.Sorted = true;
listBoxOcrFixList.EndUpdate();
}
}
private void buttonOK_Click(object sender, EventArgs e)
{
string key = textBoxOcrFixKey.Text.RemoveControlCharacters().Trim();
string value = textBoxOcrFixValue.Text.RemoveControlCharacters().Trim();
if (key.Length == 0 || value.Length == 0 || key == value)
{
return;
}
var languageString = LanguageString;
if (languageString == null)
{
return;
}
try
{
var ci = CultureInfo.GetCultureInfo(languageString.Replace('_', '-'));
_threeLetterIsoLanguageName = ci.GetThreeLetterIsoLanguageName();
}
catch (CultureNotFoundException exception)
{
try
{
// try for fix e.g. "es-ANY"
var arr = languageString.Replace('_', '-').Split('-');
var ci2 = CultureInfo.GetCultureInfo(arr[0] + "-" + arr[0]);
_threeLetterIsoLanguageName = ci2.GetThreeLetterIsoLanguageName();
}
catch
{
MessageBox.Show(exception.Message);
return;
}
}
var ocrFixReplaceList = OcrFixReplaceList.FromLanguageId(_threeLetterIsoLanguageName);
ocrFixReplaceList.AddWordOrPartial(key, value);
DialogResult = DialogResult.OK;
NewSource = key;
NewTarget = value;
}
/// <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";
}
this.threeLetterIsoLanguageName = threeLetterIsoLanguageName;
this.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
{
}
}
