How to read width of Line annotation's appearance using iTextSharp?

Question

This is related to a previous question I asked about reading an annotation's appearance stream and writing its text to the Contents. I'd like to do a similar action with a Line annotation, reading its appearance width and setting the actual width to match the appearance.

I'm having trouble figuring out how to adapt my "set text contents using appearance" function to set line width. This is the code I'm currently using for getting the text:

//main function for setting inner content to appearance value
public void changeAnnotationContentToAppearance(PdfDictionary dict)
        {

                string surface = pdfTextParser.retrieveText(dict);
                if (surface != null)
                {
                   //update CONTENTS with appearance
                   //for changing line width, I would instead modify the /BS dictionary's /W key value, i think
                   dict.Put(PdfName.CONTENTS, new PdfString(surface));
                }
        }

//get text from /AP dictionary
public string retrieveText(PdfDictionary annotDictionary)
    {

        PdfDictionary appearancesDictionary = annotDictionary.GetAsDict(PdfName.AP);
        foreach (PdfName key in appearancesDictionary.Keys)
        {
            PdfStream value = appearancesDictionary.GetAsStream(key);
            if (value != null)
            {
                String text = ExtractAnnotationText(value);
                return text;
            }
        }
        return null;

    }

//read the appearance stream and extract text contents
public String ExtractAnnotationText(PdfStream xObject)
{
    PdfDictionary resources = xObject.GetAsDict(PdfName.RESOURCES);
    ITextExtractionStrategy strategy = new LocationTextExtractionStrategy();

    PdfContentStreamProcessor processor = new PdfContentStreamProcessor(strategy);
    processor.ProcessContent(ContentByteUtils.GetContentBytesFromContentObject(xObject), resources);
    return strategy.GetResultantText();
}

ExtractAnnotationText() only seems capable of reading text, not line width, because ITextExtractionStrategy() doesn't have any methods for returning line properties. Does iTextSharp offer another extraction strategy for use in working with lines?

If I'm reading right, this question, this one, and this one suggest that I would need to implement a class, but I'm not sure which one I should subclass for getting line data, or how exactly I would go about doing that.

EDIT: I'd also like to get the appearance data for the points defining a rectangle in a textbox. Though this could be a different question, it seems closely related to this problem: retrieving non-text graphical data defining an annotation's appearance stream.

Answer 1

You need the PathRenderInfo object to get information about lines and shapes. The PathRenderInfo object was introduced in iText 7. This is a proof of concept I wrote very quickly:

public static void main(String args[]) throws IOException {
    PdfDocument document = new PdfDocument(new PdfReader(SRC));
    PdfPage page = document.getPage(1);
    PdfCanvasProcessor processor = new PdfCanvasProcessor(new IEventListener() {
        public void eventOccurred(IEventData data, EventType type) {
            if (type == EventType.RENDER_PATH) {
                PathRenderInfo renderinfo = (PathRenderInfo) data;
                int i = renderinfo.getOperation();
                switch (i) {
                    case 1:
                        System.out.print("Stroke: ");
                        break;
                    case 2:
                        System.out.print("Fill: ");
                        break;
                    default:
                        System.out.print("No: ");
                }
                for (Subpath p : renderinfo.getPath().getSubpaths()) {
                    for (IShape shape : p.getSegments()) {
                        for (Point point : shape.getBasePoints()) {
                            System.out.println(String.format("x = %s; y = %s", point.getX(), point.getY()));
                        }
                    }
                }
            }
        }
        public Set<EventType> getSupportedEvents() {
            return null;
        }
    });
    processor.processPageContent(page);
}

I ran it on a PDF with plenty of lines and this was (part of) the output:

Stroke: x = -406.0; y = -240.0
x = 406.0; y = -240.0
x = -406.0; y = -200.0
x = 406.0; y = -200.0
x = -406.0; y = -160.0
x = 406.0; y = -160.0
x = -406.0; y = -120.0
x = 406.0; y = -120.0
x = -406.0; y = -80.0
x = 406.0; y = -80.0
x = -406.0; y = -40.0
x = 406.0; y = -40.0
x = -406.0; y = 0.0
x = 406.0; y = 0.0
x = -406.0; y = 40.0
x = 406.0; y = 40.0
x = -406.0; y = 80.0

You'll have to upgrade to iText 7 to make this work and you'll also have to explore which information is contained in the PathRenderInfo, Subpath and IShape objects.

Update:

As indicated in the comments, one could wonder if you're asking the right question. Take a look at this screen shot:

If you look inside this PDF, you won't find an appearance stream:

The appearance is created by the viewer based on values such as:

• /C: the color: red = 0, green = 0, blue = 1 (hence the line is blue)
• /LE: the line endings (in this case: diamond shape)
• L: the line between (x = 20, y = 790) and (x = 575, y = 790)
• ...

Why would you parse the appearance if you have all the necessary information at hand in the annotation dictionary?

The same goes for the rectangle of a text annotation. That information is stored in the /Rect value. In this case, the annotation is dimensionless ([0 0 0 0]) because we have only one dimension and the line is defined by the values stored in /L.

Answer 2

The OP clarified in comments to @Bruno's answer

I want to parse the appearance stream and use its values to set values in the annotation dictionary.

and

when the appearance stream does not match these values (e.g. the border is black in the dictionary but red in the appearance stream), ... I want to set the dictionary contents to the appearance stream. This problem occurs in some PDFs that are created in Foxit and then opened in Adobe.

Unfortunately PDF allows many ways to create similar effects. To draw a border, e.g.,

• you can strike a path of four lines,
• or you can strike a rectangle,
• or you can fill a rectangle and fill another, slightly smaller inner rectangle with white,
• or you can fill everything with a rectangular mask,
• or you can draw a bitmap with the desired form,
• or you can ...
• ...

Thus, a truly generic solution for your problem is somewhere between extremely complicated and impossible.

Furthermore some appearances may be impossible to represent using only the limited abstract settings in the annotation dictionary. E.g. a border in the appearance stream might be created solid in the middle but fading out left and right using transparency, or it may be drawn using a color shading operator resulting in a color gradient, or its form might not be exactly rectangular but instead irregular, or, or, ...

---

If you are not looking for a generic solution, though, but merely for a solution working for annotations created by certain software products like Foxit in some versions, and if the appearances created by that software can be represented using the abstract annotation dictionary values, the task becomes feasible.

In that case you should start by analyzing sample appearance streams created by those software products. Most likely some pattern will emerge.

As soon as you have found that pattern, you can start implementing a matching iTextSharp 5.5.x IExtRenderListener or iTextSharp 7.0.x IEventListener.