Created attachment 22524 [details]
Document with 2 shapes (the second is a group)

Office Art shapes (rectangles, circles, arrows, shape groups, etc.) are coded into Word documents in a PlexOfCps that starts at fib.getFcPlcspaMom(), ends at fib.getLcbPlcspaMom() and is made of structures 26 bytes wide. I've written some patch code to access these data by:


1) adding to "usermodel" package the class

package org.apache.poi.hwpf.usermodel;

import org.apache.poi.hwpf.model.GenericPropertyNode;
import org.apache.poi.util.LittleEndian;

public class Shape {
	int _id, _left, _right, _top, _bottom;
	boolean _inDoc; //true if the Shape bounds are within document (for example, it's false if the image left corner is outside the doc, like for embedded documents)
	
	public Shape(GenericPropertyNode nodo) {
		byte [] contenuto = nodo.getBytes();
		_id = LittleEndian.getInt(contenuto);
		_left = LittleEndian.getInt(contenuto, 4);
		_top = LittleEndian.getInt(contenuto, 8);
		_right = LittleEndian.getInt(contenuto, 12);
		_bottom = LittleEndian.getInt(contenuto, 16);
		_inDoc = (_left >= 0 && _right >= 0 && _top >= 0 && _bottom >= 0);
	}
	
	public int getId() {
		return _id;
	}
	
	public int getLeft() {
		return _left;
	}
	
	public int getRight() {
		return _right;
	}
	
	public int getTop() {
		return _top;
	}
	
	public int getBottom() {
		return _bottom;
	}
	
	public int getWidth() {
		return _right - _left + 1;
	}
	
	public int getHeight() {
		return _bottom - _top + 1;
	}
	
	public boolean isWithinDocument() {
		return _inDoc;
	}
}

2) adding to "model" package the class

package org.apache.poi.hwpf.model;

import java.util.ArrayList;
import java.util.List;

import org.apache.poi.hwpf.usermodel.Shape;

public class ShapesTable {
	private List<Shape> _shapes;
	private List<Shape> _shapesVisibili;  //holds visible shapes
	
	public ShapesTable(byte [] tblStream, FileInformationBlock fib) {
		PlexOfCps binTable = new PlexOfCps(tblStream, fib.getFcPlcspaMom(), fib.getLcbPlcspaMom(), 26);
		
		_shapes = new ArrayList<Shape>();
		_shapesVisibili = new ArrayList<Shape>();
		

		for(int i = 0; i < binTable.length(); i++) {
			GenericPropertyNode nodo = binTable.getProperty(i);
			
			Shape sh = new Shape(nodo);
			_shapes.add(sh);
			if(sh.isWithinDocument())
				_shapesVisibili.add(sh);
		}
	}
	
	public List<Shape> getAllShapes() {
		return _shapes;
	}
	
	public List<Shape> getVisibleShapes() {
		return _shapesVisibili;
	}
}

3) editing the HWPFDocument class, adding in the constructor

public HWPFDocument(DirectoryNode directory, POIFSFileSystem pfilesystem) throws IOException

the declaration

/** Holds Office Art objects */
protected ShapesTable _officeArts;

and the line (for example after the _pictures stream reading line)

_officeArts = new ShapesTable(_tableStream, _fib);

which reads shapes data from doc table stream

4) I've also written a simple main method to test this

public static void main(String[] args)
  {
          try
          {

        	  JFileChooser jfc = new JFileChooser();

                int esito = jfc.showOpenDialog(null);

                if(esito != JFileChooser.APPROVE_OPTION)
                {
                        JOptionPane.showMessageDialog(null, "No file selected");
                }
                else
                {
                        String path = jfc.getSelectedFile().getAbsolutePath();

                        HWPFDocument doc = new HWPFDocument(new FileInputStream(path));
                        Range r = doc.getRange();
                        
                        List<Shape> shapes = doc.getShapesTable().getAllShapes();
                        
                        for(Shape shp: shapes)
                        	System.out.println("SHAPE " + shp.getWidth() + "x" + shp.getHeight() + ", WITHIN DOC=" + shp.isWithinDocument());
                }
          }
          catch(Exception er)
          {
                  er.printStackTrace();
          }
  }

that can be tried with the attached doc

5) Perhaps the MS Open Specification Promise initiative can be helpful? They have published all Office doc specs...

Hope this helps :D