public VobSubNOcrEdit(NOcrDb nOcrDb, Bitmap bitmap, string fileName)
{
UiUtil.PreInitialize(this);
InitializeComponent();
VobSubEditCharacters.MakeToolStripLetters(contextMenuStripLetters, InsertLanguageCharacter);
UiUtil.FixFonts(this);
_nOcrDb = nOcrDb;
_bitmap = bitmap;
FillComboBox();
if (bitmap != null)
{
pictureBoxCharacter.Image = bitmap;
SizePictureBox();
}
labelInfo.Text = $"{(_nOcrDb.OcrCharacters.Count + _nOcrDb.OcrCharactersExpanded.Count):#,###,##0} elements in database";
labelNOcrCharInfo.Text = string.Empty;
if (!string.IsNullOrEmpty(fileName))
{
Text = "nOCR DB - " + fileName;
}
if (comboBoxTexts.Items.Count > 0)
{
comboBoxTexts.SelectedIndex = 0;
}
}
private void buttonTrain_Click(object sender, EventArgs e)
{
if (!System.IO.File.Exists(textBoxInputFile.Text))
{
return;
}
int numberOfCharactersLeaned = 0;
int numberOfCharactersSkipped = 0;
var nOcrD = new NOcrDb(textBoxNOcrDb.Text);
var lines = new List <string>();
foreach (string line in System.IO.File.ReadAllLines(textBoxInputFile.Text))
{
lines.Add(line);
}
var format = new SubRip();
var sub = new Subtitle();
format.LoadSubtitle(sub, lines, textBoxInputFile.Text);
var charactersLearned = new List <string>();
foreach (ListViewItem item in listViewFonts.Items)
{
if (item.Checked)
{
_subtitleFontName = item.Text;
_subtitleFontSize = Convert.ToInt32(comboBoxSubtitleFontSize.Items[comboBoxSubtitleFontSize.SelectedIndex].ToString());
charactersLearned = new List <string>();
foreach (Paragraph p in sub.Paragraphs)
{
foreach (char ch in p.Text)
{
string s = ch.ToString();
if (s.Trim().Length > 0)
{
if (!charactersLearned.Contains(s))
{
TrainLetter(ref numberOfCharactersLeaned, ref numberOfCharactersSkipped, nOcrD, charactersLearned, s, false);
if (checkBoxBold.Checked)
{
TrainLetter(ref numberOfCharactersLeaned, ref numberOfCharactersSkipped, nOcrD, charactersLearned, s, true);
}
}
}
}
}
}
}
nOcrD.Save();
}
internal void Initialize(Bitmap bitmap, int pixelsIsSpace, bool rightToLeft, NOcrDb nOcrDb, VobSubOcr vobSubOcr, double unItalicFactor)
{
_bitmap = bitmap;
var nbmp = new NikseBitmap(bitmap);
nbmp.ReplaceNonWhiteWithTransparent();
bitmap = nbmp.GetBitmap();
_bitmap2 = bitmap;
_nocrChars = nOcrDb.OcrCharacters;
_matchList = new List <VobSubOcr.CompareMatch>();
_vobSubOcr = vobSubOcr;
_unItalicFactor = unItalicFactor;
int minLineHeight = 6;
_imageList = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight);
// _imageList = NikseBitmapImageSplitter.SplitBitmapToLetters(nbmp, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom);
int index = 0;
while (index < _imageList.Count)
{
ImageSplitterItem item = _imageList[index];
if (item.NikseBitmap == null)
{
listBoxInspectItems.Items.Add(item.SpecialCharacter);
_matchList.Add(null);
}
else
{
nbmp = item.NikseBitmap;
nbmp.ReplaceNonWhiteWithTransparent();
item.Y += nbmp.CropTopTransparent(0);
nbmp.CropTransparentSidesAndBottom(0, true);
nbmp.ReplaceTransparentWith(Color.Black);
//get nocr matches
Nikse.SubtitleEdit.Forms.VobSubOcr.CompareMatch match = vobSubOcr.GetNOcrCompareMatchNew(item, nbmp, nOcrDb, _unItalicFactor, false, false);
if (match == null)
{
listBoxInspectItems.Items.Add("?");
_matchList.Add(null);
}
else
{
listBoxInspectItems.Items.Add(match.Text);
_matchList.Add(match);
}
}
index++;
}
}
public BinaryOcrChooseEditDb(string binaryImageDb)
{
UiUtil.PreInitialize(this);
InitializeComponent();
UiUtil.FixFonts(this);
buttonNewCharacterDatabase.Text = Configuration.Settings.Language.VobSubOcr.New;
buttonEditCharacterDatabase.Text = Configuration.Settings.Language.VobSubOcr.Edit;
buttonOK.Text = Configuration.Settings.Language.General.Ok;
buttonCancel.Text = Configuration.Settings.Language.General.Cancel;
labelImageDatabase.Text = Configuration.Settings.Language.VobSubOcr.ImageDatabase;
linkLabelOpenDictionaryFolder.Text = Configuration.Settings.Language.GetDictionaries.OpenDictionariesFolder;
Text = Configuration.Settings.Language.VobSubOcr.ImageDatabase;
var imageCompareDbName = string.Empty;
var nOcrDbName = string.Empty;
ImageCompareDatabaseName = binaryImageDb;
if (!string.IsNullOrEmpty(binaryImageDb))
{
var parts = binaryImageDb.Split('+');
if (parts.Length > 0)
{
imageCompareDbName = parts[0];
if (parts.Length > 1)
{
nOcrDbName = parts[1];
}
}
}
comboBoxNOcrLanguage.Items.Clear();
comboBoxNOcrLanguage.Items.Add(string.Empty);
foreach (string s in NOcrDb.GetDatabases())
{
comboBoxNOcrLanguage.Items.Add(s);
if (s == nOcrDbName)
{
comboBoxNOcrLanguage.SelectedIndex = comboBoxNOcrLanguage.Items.Count - 1;
}
}
comboBoxCharacterDatabase.Items.Clear();
foreach (string s in BinaryOcrDb.GetDatabases())
{
comboBoxCharacterDatabase.Items.Add(s);
if (s == imageCompareDbName)
{
comboBoxCharacterDatabase.SelectedIndex = comboBoxCharacterDatabase.Items.Count - 1;
}
}
}
public void TestNOcrSaveLoad()
{
string tempFileName = FileUtil.GetTempFileName(".nocr");
var db = new NOcrDb(tempFileName);
var nOcrChar = new NOcrChar("t");
nOcrChar.ExpandCount = 0;
nOcrChar.Italic = false;
nOcrChar.MarginTop = 2;
nOcrChar.Width = 10;
nOcrChar.Height = 10;
nOcrChar.LinesForeground.Add(new NOcrPoint(new Point(1, 1), new Point(2, 2)));
nOcrChar.LinesBackground.Add(new NOcrPoint(new Point(3, 4), new Point(5, 6)));
db.Add(nOcrChar);
var nOcrChar2 = new NOcrChar("u");
nOcrChar2.ExpandCount = 0;
nOcrChar2.Italic = false;
nOcrChar2.MarginTop = 3;
nOcrChar2.Width = 12;
nOcrChar2.Height = 12;
nOcrChar2.LinesForeground.Add(new NOcrPoint(new Point(1, 1), new Point(2, 2)));
nOcrChar2.LinesBackground.Add(new NOcrPoint(new Point(3, 4), new Point(5, 6)));
db.Add(nOcrChar2);
db.Save();
db = new NOcrDb(tempFileName);
Assert.IsTrue(db.OcrCharacters.Count == 2);
Assert.IsTrue(db.OcrCharacters[0].Text == nOcrChar2.Text);
Assert.IsTrue(db.OcrCharacters[0].Italic == nOcrChar2.Italic);
Assert.IsTrue(db.OcrCharacters[0].MarginTop == nOcrChar2.MarginTop);
Assert.IsTrue(db.OcrCharacters[0].LinesForeground.Count == nOcrChar2.LinesForeground.Count);
Assert.IsTrue(db.OcrCharacters[0].LinesForeground[0].Start.X == nOcrChar2.LinesForeground[0].Start.X);
Assert.IsTrue(db.OcrCharacters[0].LinesForeground[0].Start.Y == nOcrChar2.LinesForeground[0].Start.Y);
Assert.IsTrue(db.OcrCharacters[0].LinesBackground.Count == nOcrChar2.LinesBackground.Count);
Assert.IsTrue(db.OcrCharacters[0].LinesBackground[0].Start.X == nOcrChar2.LinesBackground[0].Start.X);
Assert.IsTrue(db.OcrCharacters[0].LinesBackground[0].Start.Y == nOcrChar2.LinesBackground[0].Start.Y);
Assert.IsTrue(db.OcrCharacters[1].Text == nOcrChar.Text);
try
{
File.Delete(tempFileName);
}
catch
{
}
}
internal void Initialize(Bitmap bitmap, int pixelsIsSpace, bool rightToLeft, NOcrDb nOcrDb, VobSubOcr vobSubOcr, bool italic, int minLineHeight, bool deepSeek)
{
_bitmap = bitmap;
var nikseBitmap = new NikseBitmap(bitmap);
nikseBitmap.ReplaceNonWhiteWithTransparent();
_nOcrChars = nOcrDb.OcrCharacters;
_nOcrDb = nOcrDb;
_matchList = new List <VobSubOcr.CompareMatch>();
_imageList = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nikseBitmap, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight, false);
int index = 0;
_indexLookup = new Dictionary <int, int>();
while (index < _imageList.Count)
{
var item = _imageList[index];
if (item.NikseBitmap == null)
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add(item.SpecialCharacter);
_matchList.Add(null);
}
else
{
var match = vobSubOcr.GetNOcrCompareMatchNew(item, nikseBitmap, nOcrDb, italic, deepSeek, index, _imageList);
if (match == null)
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add("?");
_matchList.Add(null);
}
else
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add(match.Text);
_matchList.Add(match);
if (match.ExpandCount > 0)
{
index += match.ExpandCount - 1;
}
}
}
index++;
}
}
private void buttonImport_Click(object sender, EventArgs e)
{
int importedCount = 0;
int notImportedCount = 0;
openFileDialog1.Filter = "nOCR files|*.nocr";
openFileDialog1.InitialDirectory = Configuration.DataDirectory;
openFileDialog1.FileName = string.Empty;
openFileDialog1.Title = "Import existing nOCR database into current";
if (openFileDialog1.ShowDialog(this) == DialogResult.OK)
{
var newDb = new NOcrDb(openFileDialog1.FileName);
importedCount = ImportCharacters(importedCount, ref notImportedCount, newDb.OcrCharacters);
importedCount += ImportCharacters(importedCount, ref notImportedCount, newDb.OcrCharactersExpanded);
if (importedCount > 0)
{
Changed = true;
}
MessageBox.Show($"Number of characters imported: {importedCount}\r\nNumber of characters not imported (already present): {notImportedCount}");
}
}
private void buttonTrain_Click(object sender, EventArgs e)
{
if (buttonTrain.Text == "Abort")
{
_abort = true;
return;
}
if (!File.Exists(textBoxInputFile.Text))
{
return;
}
_abort = false;
buttonTrain.Text = "Abort";
buttonOK.Enabled = false;
int numberOfCharactersLeaned = 0;
int numberOfCharactersSkipped = 0;
var nOcrD = new NOcrDb(null);
var lines = new List <string>();
foreach (string line in File.ReadAllLines(textBoxInputFile.Text))
{
lines.Add(line);
}
var format = new SubRip();
var sub = new Subtitle();
format.LoadSubtitle(sub, lines, textBoxInputFile.Text);
foreach (ListViewItem item in listViewFonts.Items)
{
if (item.Checked)
{
_subtitleFontName = item.Text;
_subtitleFontSize = Convert.ToInt32(comboBoxSubtitleFontSize.Items[comboBoxSubtitleFontSize.SelectedIndex].ToString());
var charactersLearned = new List <string>();
foreach (Paragraph p in sub.Paragraphs)
{
foreach (char ch in p.Text)
{
if (!char.IsWhiteSpace(ch))
{
var s = ch.ToString();
if (!charactersLearned.Contains(s))
{
TrainLetter(ref numberOfCharactersLeaned, ref numberOfCharactersSkipped, nOcrD, charactersLearned, s, false);
if (checkBoxBold.Checked)
{
TrainLetter(ref numberOfCharactersLeaned, ref numberOfCharactersSkipped, nOcrD, charactersLearned, s, true);
}
}
}
}
if (_abort)
{
break;
}
}
if (_abort)
{
break;
}
}
}
labelInfo.Text = "Saving...";
labelInfo.Refresh();
saveFileDialog1.Filter = "nOCR DB|*.nocr";
saveFileDialog1.InitialDirectory = Configuration.OcrDirectory;
if (saveFileDialog1.ShowDialog(this) == DialogResult.OK)
{
labelInfo.Text = $"Saving to {saveFileDialog1.FileName}...";
nOcrD.FileName = saveFileDialog1.FileName;
nOcrD.Save();
labelInfo.Text = "Training done";
}
else
{
labelInfo.Text = "Saving aborted";
}
buttonOK.Enabled = true;
buttonTrain.Text = "Start training";
_abort = false;
}
private void TrainLetter(ref int numberOfCharactersLeaned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, List <string> charactersLearned, string s, bool bold)
{
Bitmap bmp = GenerateImageFromTextWithStyle("H " + s, bold);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(280);
nbmp.CropTop(0, Color.FromArgb(0, 0, 0, 0));
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25);
if (list.Count == 3)
{
var item = list[2];
NOcrChar match = nOcrD.GetMatch(item.NikseBitmap, item.Top, false, false, 0);
if (match == null || match.Text != s)
{
labelInfo.Refresh();
Application.DoEvents();
NOcrChar nOcrChar = new NOcrChar(s)
{
Width = item.NikseBitmap.Width,
Height = item.NikseBitmap.Height,
MarginTop = item.Top,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value, checkBoxVeryAccurate.Checked, nOcrChar, item.NikseBitmap);
nOcrD.Add(nOcrChar);
charactersLearned.Add(s);
numberOfCharactersLeaned++;
labelInfo.Text = string.Format("Now training font '{1}', total characters learned is {0:#,###,###}, {2:#,###,###} skipped", numberOfCharactersLeaned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
private void TrainLetter(ref int numberOfCharactersLeaned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, List <string> charactersLearned, string s, bool bold)
{
Bitmap bmp = GenerateImageFromTextWithStyle(s, bold);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(210, 280);
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25);
if (list.Count == 1)
{
NOcrChar match = nOcrD.GetMatch(list[0].NikseBitmap);
if (match == null)
{
pictureBox1.Image = list[0].NikseBitmap.GetBitmap();
this.Refresh();
Application.DoEvents();
System.Threading.Thread.Sleep(100);
NOcrChar nOcrChar = new NOcrChar(s);
nOcrChar.Width = list[0].NikseBitmap.Width;
nOcrChar.Height = list[0].NikseBitmap.Height;
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value, checkBoxVeryAccurate.Checked, nOcrChar, list[0].NikseBitmap);
nOcrD.Add(nOcrChar);
charactersLearned.Add(s);
numberOfCharactersLeaned++;
labelInfo.Text = string.Format("Now training font '{1}', total characters leaned is {0}, {2} skipped", numberOfCharactersLeaned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
private void buttonImport_Click(object sender, EventArgs e)
{
int importedCount = 0;
int notImportedCount = 0;
openFileDialog1.Filter = "nOCR files|*.nocr";
openFileDialog1.InitialDirectory = Configuration.DataDirectory;
openFileDialog1.FileName = string.Empty;
openFileDialog1.Title = "Import existing nOCR database into current";
if (openFileDialog1.ShowDialog(this) == DialogResult.OK)
{
NOcrDb newDb = new NOcrDb(openFileDialog1.FileName);
foreach (NOcrChar newChar in newDb.OcrCharacters)
{
bool found = false;
foreach (NOcrChar oldChar in _nocrChars)
{
if (oldChar.Text == newChar.Text &&
oldChar.Width == newChar.Width &&
oldChar.Height == newChar.Height &&
oldChar.MarginTop == newChar.MarginTop &&
oldChar.ExpandCount == newChar.ExpandCount &&
oldChar.LinesForeground.Count == newChar.LinesForeground.Count &&
oldChar.LinesBackground.Count == newChar.LinesBackground.Count)
{
found = true;
for (int i = 0; i < oldChar.LinesForeground.Count; i++)
{
if (oldChar.LinesForeground[i].Start.X != newChar.LinesForeground[i].Start.X ||
oldChar.LinesForeground[i].Start.Y != newChar.LinesForeground[i].Start.Y ||
oldChar.LinesForeground[i].End.X != newChar.LinesForeground[i].End.X ||
oldChar.LinesForeground[i].End.Y != newChar.LinesForeground[i].End.Y)
{
found = false;
}
}
for (int i = 0; i < oldChar.LinesBackground.Count; i++)
{
if (oldChar.LinesBackground[i].Start.X != newChar.LinesBackground[i].Start.X ||
oldChar.LinesBackground[i].Start.Y != newChar.LinesBackground[i].Start.Y ||
oldChar.LinesBackground[i].End.X != newChar.LinesBackground[i].End.X ||
oldChar.LinesBackground[i].End.Y != newChar.LinesBackground[i].End.Y)
{
found = false;
}
}
}
}
if (!found)
{
_nocrChars.Add(newChar);
importedCount++;
}
else
{
notImportedCount++;
}
}
MessageBox.Show(string.Format("Number of characters imported: {0}\r\nNumber of characters not imported (already present): {1}", importedCount, notImportedCount));
}
}
internal void Initialize(NOcrDb _nOcrDb)
{
if (_nOcrDb != null)
{
}
}
private void TrainLetter(ref int numberOfCharactersLearned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, string s, bool bold, bool italic, bool doubleLetter)
{
var bmp = GenerateImageFromTextWithStyle("H " + s, bold, italic);
var nikseBitmap = new NikseBitmap(bmp);
nikseBitmap.MakeTwoColor(280);
nikseBitmap.CropTop(0, Color.FromArgb(0, 0, 0, 0));
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nikseBitmap, 10, false, false, 25, false);
if (list.Count == 3)
{
var item = list[2];
var match = nOcrD.GetMatch(item.NikseBitmap, item.Top, false, 25);
if (match == null || match.Text != s)
{
labelInfo.Refresh();
Application.DoEvents();
var nOcrChar = new NOcrChar(s)
{
Width = item.NikseBitmap.Width,
Height = item.NikseBitmap.Height,
MarginTop = item.Top,
Italic = italic,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + (doubleLetter ? 20 : 0), false, nOcrChar, item.NikseBitmap);
nOcrD.Add(nOcrChar);
numberOfCharactersLearned++;
labelInfo.Text = string.Format(Configuration.Settings.Language.VobSubOcr.NowTraining, numberOfCharactersLearned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
else if (!doubleLetter)
{
if (list.Count == 4 && list[2].NikseBitmap != null && list[3].NikseBitmap != null)
{
// e.g. quote (")
var expandItem = VobSubOcr.GetExpandedSelectionNew(nikseBitmap, new List <ImageSplitterItem> {
list[2], list[3]
});
var match = nOcrD.GetMatchExpanded(nikseBitmap, expandItem, 2, list);
if (match != null && match.Text == s)
{
numberOfCharactersSkipped++;
return;
}
var nOcrChar = new NOcrChar(s)
{
Width = expandItem.NikseBitmap.Width,
Height = expandItem.NikseBitmap.Height,
MarginTop = expandItem.Top,
Italic = italic,
ExpandCount = 2,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + 5, false, nOcrChar, expandItem.NikseBitmap);
nOcrD.Add(nOcrChar);
return;
}
if (list.Count == 5 && list[2].NikseBitmap != null && list[3].NikseBitmap != null && list[4].NikseBitmap != null)
{
// e.g. "%"
var expandItem = VobSubOcr.GetExpandedSelectionNew(nikseBitmap, new List <ImageSplitterItem> {
list[2], list[3], list[4]
});
var match = nOcrD.GetMatchExpanded(nikseBitmap, expandItem, 2, list);
if (match != null && match.Text == s)
{
numberOfCharactersSkipped++;
return;
}
var nOcrChar = new NOcrChar(s)
{
Width = expandItem.NikseBitmap.Width,
Height = expandItem.NikseBitmap.Height,
MarginTop = expandItem.Top,
Italic = italic,
ExpandCount = 3,
};
nOcrD.Add(nOcrChar);
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + 10, false, nOcrChar, expandItem.NikseBitmap);
return;
}
numberOfCharactersSkipped++;
}
}
