public void DisplayTextBlocks()
{
if (PagePlotModel == null)
{
return;
}
foreach (var s in PagePlotModel.Series.Where(s => (string)s.Tag == "textblock").ToList())
{
PagePlotModel.Series.Remove(s);
}
foreach (var block in _pdfPageModel.GetTextBlocks())
{
var series1 = new LineSeries {
Tag = "textblock", Title = GetShorterText(block.Text), LineStyle = LineStyle.Solid, Color = OxyColors.Brown
};
var bbox = block.BoundingBox;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
PagePlotModel.Series.Add(series1);
}
PagePlotModel.InvalidatePlot(true);
}
public void DisplayImages()
{
if (PagePlotModel == null)
{
return;
}
foreach (var s in PagePlotModel.Series.Where(s => (string)s.Tag == "image").ToList())
{
PagePlotModel.Series.Remove(s);
}
foreach (var block in _pdfPageModel.GetImages())
{
var series1 = new LineSeries {
Tag = "image", Title = "image", LineStyle = LineStyle.Solid, Color = OxyColors.YellowGreen
};
var bbox = block.Bounds;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
PagePlotModel.Series.Add(series1);
}
PagePlotModel.InvalidatePlot(true);
}
public void DisplayLetters()
{
if (PagePlotModel == null)
{
return;
}
foreach (var s in PagePlotModel.Series.Where(s => (string)s.Tag == "letter").ToList())
{
PagePlotModel.Series.Remove(s);
}
foreach (var letter in _pdfPageModel.GetLetters())
{
var series1 = new LineSeries {
Tag = "letter", Title = GetShorterText(letter.Value), LineStyle = LineStyle.Solid, Color = OxyColors.Blue
};
var bbox = letter.GlyphRectangle;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
PagePlotModel.Series.Add(series1);
}
PagePlotModel.InvalidatePlot(true);
}
/// <summary>
/// Initializes a new instance of the <see cref="MainViewModel" /> class.
/// </summary>
public MainViewModel()
{
CustomController = new CustomPlotController();
WordExtractorList = new ObservableCollection <Type>(PdfDocumentModel.GetWordExtractors());
PageSegmenterList = new ObservableCollection <Type>(PdfDocumentModel.GetPageSegmenters());
}
public void OpenDocument(string path)
{
if (Path.GetExtension(path) != ".pdf")
{
return;
}
_pdfImageConverter = new PdfImageConverter(path);
_pdfDocumentModel = PdfDocumentModel.Open(path);
NumberOfPages = _pdfDocumentModel.NumberOfPages;
CurrentPageNumber = 1;
}
public void DisplayPaths()
{
if (PagePlotModel == null)
{
return;
}
foreach (var s in PagePlotModel.Series.Where(s => (string)s.Tag == "pdfpath").ToList())
{
PagePlotModel.Series.Remove(s);
}
foreach (var path in _pdfPageModel.GetPdfPaths())
{
foreach (var sp in path)
{
string title = ("path: " + (path.IsStroked ? "stroked " + (path.StrokeColor?.ToRGBValues()).ToString() : "") +
(path.IsFilled ? "filled " + (path.FillColor?.ToRGBValues()).ToString() : "") +
(path.IsClipping ? "clipping" : "")
).Trim();
var series1 = new LineSeries {
Tag = "pdfpath", Title = title, LineStyle = LineStyle.Solid, Color = OxyColors.Yellow
};
PdfPoint first = PdfPoint.Origin;
foreach (var c in sp.Commands)
{
if (c is Move m)
{
first = m.Location;
series1.Points.Add(PdfDocumentModel.ToDataPoint(first));
}
else if (c is Line l)
{
series1.Points.Add(PdfDocumentModel.ToDataPoint(l.From));
series1.Points.Add(PdfDocumentModel.ToDataPoint(l.To));
}
else if (c is BezierCurve bc)
{
var lines = bc.ToLines(10).ToList();
for (int i = 0; i < lines.Count; i++)
{
series1.Points.Add(PdfDocumentModel.ToDataPoint(lines[i].From));
series1.Points.Add(PdfDocumentModel.ToDataPoint(lines[i].To));
}
}
else if (c is Close)
{
series1.Points.Add(PdfDocumentModel.ToDataPoint(first));
}
else
{
throw new ArgumentException();
}
}
PagePlotModel.Series.Add(series1);
}
}
PagePlotModel.InvalidatePlot(true);
}
private void DisplayPage(int pageNo)
{
if (_pdfDocumentModel == null)
{
return;
}
var page = _pdfDocumentModel.GetPage(pageNo);
// Create the plot model
var tmp = new PlotModel {
IsLegendVisible = false
};
tmp.Axes.Add(new LinearAxis {
Position = AxisPosition.Left, Minimum = 0, Maximum = page.Height
});
tmp.Axes.Add(new LinearAxis {
Position = AxisPosition.Bottom, Minimum = 0, Maximum = page.Width
});
switch (BboxLevel)
{
case "Words":
foreach (var word in page.GetWords())
{
var series1 = new LineSeries {
Title = GetShorterText(word.Text), LineStyle = LineStyle.Solid, Color = OxyColors.Red
};
var bbox = word.BoundingBox;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
tmp.Series.Add(series1);
}
break;
case "Lines":
foreach (var line in page.GetTextBlocks().SelectMany(b => b.TextLines))
{
var series1 = new LineSeries {
Title = GetShorterText(line.Text), LineStyle = LineStyle.Solid, Color = OxyColors.Red
};
var bbox = line.BoundingBox;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
tmp.Series.Add(series1);
}
break;
case "Paragraphs":
foreach (var block in page.GetTextBlocks())
{
var series1 = new LineSeries {
Title = GetShorterText(block.Text), LineStyle = LineStyle.Solid, Color = OxyColors.Red
};
var bbox = block.BoundingBox;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
tmp.Series.Add(series1);
}
break;
default:
foreach (var letter in page.GetLetters())
{
var series1 = new LineSeries {
Title = letter.Value, LineStyle = LineStyle.Solid, Color = OxyColors.Red
};
var bbox = letter.GlyphRectangle;
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopRight));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.TopLeft));
series1.Points.Add(PdfDocumentModel.ToDataPoint(bbox.BottomLeft));
tmp.Series.Add(series1);
}
break;
}
// Add background image
try
{
using (var stream = _pdfImageConverter.GetPageStream(pageNo, 2))
{
Image = new OxyImage(stream);
}
tmp.Annotations.Add(new ImageAnnotation
{
ImageSource = Image,
Opacity = 0.5,
X = new PlotLength(0, PlotLengthUnit.Data),
Y = new PlotLength(0, PlotLengthUnit.Data),
Width = new PlotLength(page.Width, PlotLengthUnit.Data),
Height = new PlotLength(page.Height, PlotLengthUnit.Data),
HorizontalAlignment = HorizontalAlignment.Left,
VerticalAlignment = VerticalAlignment.Bottom
});
}
catch (Exception)
{
throw;
}
this.PlotModel = tmp;
}
