1 files changed, 193 insertions, 56 deletions
diff --git a/pdf-as-lib/src/main/java/at/knowcenter/wag/egov/egiz/pdf/PDFPage.java b/pdf-as-lib/src/main/java/at/knowcenter/wag/egov/egiz/pdf/PDFPage.java
index 7de89d2..1a89b7b 100644
--- a/pdf-as-lib/src/main/java/at/knowcenter/wag/egov/egiz/pdf/PDFPage.java
+++ b/pdf-as-lib/src/main/java/at/knowcenter/wag/egov/egiz/pdf/PDFPage.java
@@ -25,15 +25,19 @@
  */
 package at.knowcenter.wag.egov.egiz.pdf;
 
+import java.awt.Rectangle;
+import java.awt.geom.GeneralPath;
 import java.io.IOException;
 import java.util.List;
 import java.util.Map;
 
+import org.apache.commons.lang.math.NumberUtils;
 import org.apache.log4j.Logger;
 import org.pdfbox.cos.COSName;
 import org.pdfbox.cos.COSStream;
 import org.pdfbox.pdmodel.PDPage;
 import org.pdfbox.pdmodel.PDResources;
+import org.pdfbox.pdmodel.common.PDRectangle;
 import org.pdfbox.pdmodel.common.PDStream;
 import org.pdfbox.pdmodel.graphics.xobject.PDXObject;
 import org.pdfbox.pdmodel.graphics.xobject.PDXObjectForm;
@@ -44,6 +48,21 @@ import org.pdfbox.util.TextPosition;
 import org.pdfbox.util.operator.OperatorProcessor;
 
 import at.knowcenter.wag.egov.egiz.cfg.ConfigLogger;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.ClosePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.CurveTo;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.CurveToReplicateFinalPoint;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.CurveToReplicateInitialPoint;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.LineTo;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.construction.MoveTo;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.CloseAndStrokePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.CloseFillEvenOddAndStrokePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.CloseFillNonZeroAndStrokePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.EndPath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.FillEvenOddAndStrokePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.FillNonZeroAndStrokePath;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.FillPathEvenOddRule;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.FillPathNonZeroWindingNumberRule;
+import at.knowcenter.wag.egov.egiz.pdf.operator.path.painting.StrokePath;
 
 /**
  * PDFPage is an inner class that is used to calculate the page length of a PDF
@@ -52,7 +71,7 @@ import at.knowcenter.wag.egov.egiz.cfg.ConfigLogger;
  * This method is called when processing the FileStream. By calling the method
  * {@link org.pdfbox.util.PDFStreamEngine#processStream(org.pdfbox.pdmodel.PDPage, org.pdfbox.pdmodel.PDResources, org.pdfbox.cos.COSStream)}
  * the implemented method showCharacter is called.
- * 
+ *
  * @author wlackner
  * @see PDFTextStripper
  */
@@ -67,41 +86,167 @@ public class PDFPage extends PDFTextStripper
    * The maximum (lowest) y position of a character.
    */
   protected float max_character_ypos = Float.NEGATIVE_INFINITY;
-  
+
   /**
    * The maximum (lowest y position of an image.
    */
   protected float max_image_ypos = Float.NEGATIVE_INFINITY;
-  
+
   /**
    * The effective page height.
    */
   protected float effectivePageHeight;
-  
+
+  /**
+   * The path currently being constructed.
+   */
+  private GeneralPath currentPath = new GeneralPath();
+
+  /**
+   * The lowest position of a drawn path (originating from top).
+   */
+  private float maxPathRelatedYPositionFromTop = Float.NEGATIVE_INFINITY;
+
   /**
    * Constructor.
-   * 
+   *
    * @param effectivePageHeight The height of the page to be evaluated. PDF elements outside this height will not be considered.
-   * 
+   *
    * @throws IOException
    */
   public PDFPage(float effectivePageHeight) throws IOException
   {
     super();
-    
+
     this.effectivePageHeight = effectivePageHeight;
-    
+
     OperatorProcessor newInvoke = new MyInvoke();
     newInvoke.setContext(this);
     operators.put("Do", newInvoke);
+
+    registerCustomPathOperators();
   }
 
+	/**
+	 * Registers operators responsible for path construction and painting in order to fix auto positioning on pages with
+	 * path elements.
+	 *
+	 * @author Datentechnik Innovation GmbH
+	 */
+	@SuppressWarnings("unchecked")
+	private void registerCustomPathOperators() {
+
+		// *** path construction
+
+		operators.put("m", new MoveTo(this));
+		operators.put("l", new LineTo(this));
+		operators.put("c", new CurveTo(this));
+		operators.put("y", new CurveToReplicateFinalPoint(this));
+		operators.put("v", new CurveToReplicateInitialPoint(this));
+		operators.put("h", new ClosePath(this));
+
+		// *** path painting
+
+		// "S": stroke path
+		operators.put("S", new StrokePath(this));
+		operators.put("s", new CloseAndStrokePath(this));
+		operators.put("f", new FillPathNonZeroWindingNumberRule(this));
+		operators.put("F", new FillPathNonZeroWindingNumberRule(this));
+		operators.put("f*", new FillPathEvenOddRule(this));
+		operators.put("b", new CloseFillNonZeroAndStrokePath(this));
+		operators.put("B", new FillNonZeroAndStrokePath(this));
+		operators.put("b*", new CloseFillEvenOddAndStrokePath(this));
+		operators.put("B*", new FillEvenOddAndStrokePath(this));
+		operators.put("n", new EndPath(this));
+
+		// Note: The graphic context (org.pdfbox.pdmodel.graphics.PDGraphicsState) of the underlying pdfbox library does
+		// not yet support clipping. This prevents feasible usage of clipping operators (W, W*).
+//		operators.put("W", new ...(this));
+//		operators.put("W*", new ...(this));
+
+	}
+
+	/**
+	 * Returns the path currently being constructed.
+	 *
+	 * @return The path currently being constructed.
+	 */
+	public GeneralPath getCurrentPath() {
+		return currentPath;
+	}
+
+	/**
+	 * Sets the current path.
+	 * @param currentPath The new current path.
+	 */
+	public void setCurrentPath(GeneralPath currentPath) {
+		this.currentPath = currentPath;
+	}
+
+	/**
+	 * Registers a rectangle that bounds the path currently being drawn.
+	 *
+	 * @param bounds
+	 *            A rectangle depicting the bounds (coordinates originating from bottom left).
+	 * @author Datentechnik Innovation GmbH
+	 */
+	public void registerPathBounds(Rectangle bounds) {
+		if (!bounds.isEmpty()) {
+			logger_.trace("Registering path bounds: " + bounds);
+
+			// vertical start of rectangle (counting from top of page)
+			float upperBoundYPositionFromTop;
+
+			// vertical end of rectangle (counting from top of page)
+			// this depicts the current end of path-related page content
+			float lowerBoundYPositionFromTop;
+
+			PDRectangle boundaryBox = page.findMediaBox();
+			float pageHeight;
+
+			switch (page.findRotation()) {
+			case 90: // CW
+				pageHeight = boundaryBox.getWidth();
+				upperBoundYPositionFromTop = (float) bounds.getMinX();
+				lowerBoundYPositionFromTop = (float) bounds.getMaxX();
+				break;
+			case 180:
+				pageHeight = boundaryBox.getHeight();
+				upperBoundYPositionFromTop = (float) bounds.getMinY();
+				lowerBoundYPositionFromTop = (float) bounds.getMaxY();
+				break;
+			case 270: // CCW
+				pageHeight = boundaryBox.getWidth();
+				upperBoundYPositionFromTop = pageHeight - (float) bounds.getMaxX();
+				lowerBoundYPositionFromTop = pageHeight - (float) bounds.getMinX();
+				break;
+			default:
+				pageHeight = boundaryBox.getHeight();
+				upperBoundYPositionFromTop = pageHeight - (float) bounds.getMaxY();
+				lowerBoundYPositionFromTop = pageHeight - (float) bounds.getMinY();
+				break;
+			}
+
+			// new maximum ?
+			if (lowerBoundYPositionFromTop > maxPathRelatedYPositionFromTop) {
+				// Is the rectangle (at least partly) located above the footer line?
+				// (effective page height := page height - footer line)
+				if (upperBoundYPositionFromTop <= effectivePageHeight) {
+					// yes: update current end of path-related page content
+					maxPathRelatedYPositionFromTop = lowerBoundYPositionFromTop;
+					logger_.trace("New max path related y position (from top): " + maxPathRelatedYPositionFromTop);
+				} else {
+					// no: rectangle is fully located below the footer line -> ignore
+					logger_.trace("Ignoring path bound below the footer line.");
+				}
+			}
+		}
+	}
 
   protected void processOperator(PDFOperator operator, List arguments) throws IOException
   {
-    //logger_.debug("operator = " + operator);
-    
-    super.processOperator(operator, arguments);
+//    logger_.debug("operator = " + operator);
+	  super.processOperator(operator, arguments);
   }
 
   // exthex
@@ -111,7 +256,7 @@ public class PDFPage extends PDFTextStripper
    * is used to calculate the latest position of a text in the page. Sorry for
    * this missinterpretation of the method, but it is the only way to do this
    * (provided by PDFBox)!!!
-   * 
+   *
    * @param text
    *          the character to be displayed -> calculate there y position.
    */
@@ -119,7 +264,7 @@ public class PDFPage extends PDFTextStripper
   {
     float current_y = text.getY();
     final String character = text.getCharacter();
-    
+
     int pageRotation = page.findRotation();
     //logger_.debug("PageRotation = " + pageRotation);
     if (pageRotation == 0)
@@ -146,7 +291,7 @@ public class PDFPage extends PDFTextStripper
       //logger_.debug("character is below footer_line. footer_line = " + this.footer_line + ", text.character=" + character + ", y=" + current_y);
       return;
     }
-    
+
     // store ypos of the char if it is not empty
     if (!character.equals(" ") && current_y > this.max_character_ypos)
     {
@@ -160,26 +305,18 @@ public class PDFPage extends PDFTextStripper
   // logger_.debug(new String(string));
   // }
 
-  /**
-   * Returns the calculated page length.
-   * 
-   * @return the max page length value
-   */
-  public float getMaxPageLength()
-  {
-    float max_ypos = Float.NEGATIVE_INFINITY;
-    
-    if (this.max_character_ypos > this.max_image_ypos)
-    {
-      max_ypos = this.max_character_ypos;
-    }
-    else
-    {
-      max_ypos = this.max_image_ypos;
-    }
-
-    return max_ypos;
-  }
+	/**
+	 * Returns the calculated page length.
+	 *
+	 * @return the max page length value
+	 */
+	public float getMaxPageLength() {
+		if (logger_.isDebugEnabled()) {
+			logger_.debug("Determining page content length: text=" + max_character_ypos + ", image=" + max_image_ypos
+					+ ", path=" + maxPathRelatedYPositionFromTop);
+		}
+		return NumberUtils.max(max_character_ypos, max_image_ypos, maxPathRelatedYPositionFromTop);
+	}
 
   public class MyInvoke extends OperatorProcessor
   {
@@ -192,26 +329,26 @@ public class PDFPage extends PDFTextStripper
 
       Map xobjects = context.getXObjects();
       PDXObject xobject = (PDXObject) xobjects.get(name.getName());
-      
+
       PDStream stream = xobject.getPDStream();
       COSStream cos_stream = stream.getStream();
-      
+
       COSName subtype = (COSName) cos_stream.getDictionaryObject(COSName.SUBTYPE);
       if (subtype.equals(COSName.IMAGE))
       {
         logger_.debug("XObject Image");
-        
+
         Matrix ctm = context.getGraphicsState().getCurrentTransformationMatrix();
         logger_.debug("ctm = " + ctm);
-        
+
         Pos [] coordinates = new Pos [] {
         new Pos(0, 0, 1),
         new Pos(1, 0, 1),
         new Pos(0, 1, 1),
         new Pos(1, 1, 1) };
-        
+
         Pos [] transformed_coordinates = transtormCoordinates(coordinates, ctm);
-        
+
         /**********************************************************
          * pdf-as fix:
          * calculating min and max point of an image to look where
@@ -219,10 +356,10 @@ public class PDFPage extends PDFTextStripper
          * fix solves problems with footer and images and
          * placement of the signature in an image only pdf document
          **********************************************************/
-        
+
         float actual_lowest_point = Float.NaN;
         float actual_starting_point = Float.NaN;
-        
+
         int pageRotation = page.findRotation();
         logger_.debug("PageRotation = " + pageRotation);
         if (pageRotation == 0)
@@ -231,7 +368,7 @@ public class PDFPage extends PDFTextStripper
           logger_.debug("min_y = " + min_y);
           float page_height = page.findMediaBox().getHeight();
           logger_.debug("page_height = " + page_height);
-          
+
           actual_lowest_point = page_height - min_y;
           actual_starting_point = page_height - findMaxY(transformed_coordinates);
         }
@@ -243,7 +380,7 @@ public class PDFPage extends PDFTextStripper
           logger_.debug("page_width = " + page_width);
 
           actual_lowest_point = max_x;
-          actual_starting_point = findMinX(transformed_coordinates);   
+          actual_starting_point = findMinX(transformed_coordinates);
         }
         if (pageRotation == 180)
         {
@@ -257,17 +394,17 @@ public class PDFPage extends PDFTextStripper
         {
           float min_x = findMinX(transformed_coordinates);
           logger_.debug("min_x = " + min_x);
-          
+
           float page_width = page.findMediaBox().getWidth();
           logger_.debug("page_width = " + page_width);
-          
+
           actual_lowest_point = page_width - min_x;
-          actual_starting_point = page_width - findMaxX(transformed_coordinates);   
+          actual_starting_point = page_width - findMaxX(transformed_coordinates);
         }
 
-        
+
         logger_.debug("actual_lowest_point = " + actual_lowest_point);
-        
+
         if (actual_lowest_point > PDFPage.this.effectivePageHeight && actual_starting_point > PDFPage.this.effectivePageHeight)
         {
           logger_.debug("image is below footer_line");
@@ -278,7 +415,7 @@ public class PDFPage extends PDFTextStripper
         {
           PDFPage.this.max_image_ypos = actual_lowest_point;
         }
-        
+
         return;
       }
 
@@ -297,7 +434,7 @@ public class PDFPage extends PDFTextStripper
       }
     }
   }
-  
+
   public static Pos [] transtormCoordinates (Pos [] coordinates, Matrix m)
   {
     Pos [] transformed = new Pos [coordinates.length];
@@ -307,18 +444,18 @@ public class PDFPage extends PDFTextStripper
     }
     return transformed;
   }
-  
+
   public static Pos transtormCoordinate (Pos pos, Matrix m)
   {
     Pos transformed = new Pos();
     transformed.x = pos.x * m.getValue(0, 0) + pos.y * m.getValue(1, 0) + pos.z * m.getValue(2, 0);
     transformed.y = pos.x * m.getValue(0, 1) + pos.y * m.getValue(1, 1) + pos.z * m.getValue(2, 1);
     transformed.z = pos.x * m.getValue(0, 2) + pos.y * m.getValue(1, 2) + pos.z * m.getValue(2, 2);
-    
+
     logger_.debug(" transformed " + pos + " --> " + transformed);
     return transformed;
   }
-  
+
   public static float findMinY (Pos [] coordinates)
   {
     float min = Float.POSITIVE_INFINITY;
@@ -331,7 +468,7 @@ public class PDFPage extends PDFTextStripper
     }
     return min;
   }
-  
+
 	public static float findMaxY(Pos[] coordinates) {
 		float max = 0;
 		for (int i = 0; i < coordinates.length; i++) {
@@ -341,7 +478,7 @@ public class PDFPage extends PDFTextStripper
 		}
 		return max;
 	}
-  
+
   public static float findMaxX (Pos [] coordinates)
   {
     float max = Float.NEGATIVE_INFINITY;