RED-6126: performance-test

*fixed NullPointerException
*fixed StackOverFlowError by ignoring very small images and moving to while loop instead of recursion

ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/ImagePositionRetrievalService.java

@@ -15,6 +15,7 @@ import com.pdftron.common.Matrix2D;
 import com.pdftron.common.PDFNetException;
 import com.pdftron.pdf.Element;
 import com.pdftron.pdf.ElementReader;
+import com.pdftron.pdf.Image;
 import com.pdftron.pdf.PDFDoc;
 import com.pdftron.pdf.Page;
 import com.pdftron.pdf.Rect;
@@ -27,9 +28,12 @@ public class ImagePositionRetrievalService {
 
     private static final double TOLERANCE = 1e-1;
 
+    // any image with smaller height and width than this gets thrown out, see everyPointInDashedLineIsImage.pdf
+    private static final int PIXEL_THRESHOLD = 10;
+
 
     /**
-     * Iterates over all elements in a PDF Document and retrieves the bounding box for each image,
+     * Iterates over all elements in a PDF Document and retrieves the bounding box for each image, that is larger than the pixel threshold of 10 in both dimensions,
      * Then it adjusts the bounding boxes for the page rotation.
      * If the mirrorY flag is set, the Y Coordinates are mirrored and moved up by the page height. This is required for PDFTrons OCRModule.
      *
@@ -63,7 +67,13 @@ public class ImagePositionRetrievalService {
         Element element;
         while ((element = reader.next()) != null) {
             switch (element.getType()) {
-                case Element.e_image, Element.e_inline_image -> imagePositions.addRect(toRotationAdjustedRect(element.getBBox(), currentPage, mirrorY));
+                case Element.e_image, Element.e_inline_image -> {
+                    Image image = new Image(element.getXObject());
+                    // see everyPointInDashedLineIsImage.pdf TestFile
+                    if (image.getImageHeight() > PIXEL_THRESHOLD && image.getImageWidth() > PIXEL_THRESHOLD) {
+                        imagePositions.addRect(toRotationAdjustedRect(element.getBBox(), currentPage, mirrorY));
+                    }
+                }
                 case Element.e_form -> {
                     reader.formBegin();
                     findImagePositionsOnPage(reader, imagePositions, currentPage, mirrorY);
@@ -77,38 +87,38 @@ public class ImagePositionRetrievalService {
     @SneakyThrows
     public RectCollection mergeOverlappingRects(RectCollection imagePositions) {
 
-        if (imagePositions.getNumRects() == 1) {
+        if (imagePositions.getNumRects() < 2) {
             return imagePositions;
         }
 
         List<Rectangle2D> rectangleList = toSortedRectangleList(imagePositions);
 
-        rectangleList = mergeRectangleListRecursive(rectangleList, 0);
+        mergeRectangleList(rectangleList);
+
         return toRectCollection(rectangleList);
     }
 
 
     // Sometimes images are split up into stripes, here we merge the positions of aligned and intersecting rectangles into one larger rectangle
-    private List<Rectangle2D> mergeRectangleListRecursive(List<Rectangle2D> rectangleList, int currentIdx) {
+    private void mergeRectangleList(List<Rectangle2D> rectangleList) {
+        int idx = 0;
 
-        if (rectangleList.size() < currentIdx + 2) {
-            return rectangleList;
-        }
+        while (rectangleList.size() >= idx + 2) {
 
-        var rect1 = rectangleList.get(currentIdx);
-        var rect2 = rectangleList.get(currentIdx + 1);
+            var rect1 = rectangleList.get(idx);
+            var rect2 = rectangleList.get(idx + 1);
 
-        boolean isAlignedX = Math.abs(rect1.getMinX() - rect2.getMinX()) < TOLERANCE && Math.abs(rect1.getMaxX() - rect2.getMaxX()) < TOLERANCE;
-        boolean isAlignedY = Math.abs(rect1.getMinY() - rect2.getMinY()) < TOLERANCE && Math.abs(rect1.getMaxY() - rect2.getMaxY()) < TOLERANCE;
-        boolean intersects = rect1.intersects(rect2.getMinX() - TOLERANCE, rect2.getMinY() - TOLERANCE, rect2.getWidth() + (2 * TOLERANCE), rect2.getHeight() + (2 * TOLERANCE));
+            boolean isAlignedX = Math.abs(rect1.getMinX() - rect2.getMinX()) < TOLERANCE && Math.abs(rect1.getMaxX() - rect2.getMaxX()) < TOLERANCE;
+            boolean isAlignedY = Math.abs(rect1.getMinY() - rect2.getMinY()) < TOLERANCE && Math.abs(rect1.getMaxY() - rect2.getMaxY()) < TOLERANCE;
+            boolean intersects = rect1.intersects(rect2.getMinX() - TOLERANCE, rect2.getMinY() - TOLERANCE, rect2.getWidth() + (2 * TOLERANCE), rect2.getHeight() + (2 * TOLERANCE));
 
-        if (intersects && (isAlignedX || isAlignedY)) {
-            rectangleList.remove(currentIdx + 1);
-            rectangleList.remove(currentIdx);
-            rectangleList.add(currentIdx, rect1.createUnion(rect2));
-            return mergeRectangleListRecursive(rectangleList, currentIdx);
-        } else {
-            return mergeRectangleListRecursive(rectangleList, currentIdx + 1);
+            if (intersects && (isAlignedX || isAlignedY)) {
+                rectangleList.remove(idx + 1);
+                rectangleList.remove(idx);
+                rectangleList.add(idx, rect1.createUnion(rect2));
+            } else {
+                ++idx;
+            }
         }
     }
 

ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/InvisibleElementRemovalService.java

@@ -76,6 +76,8 @@ public class InvisibleElementRemovalService {
             Page page = iterator.next();
 
             visitedXObjIds.add(page.getSDFObj().getObjNum());
+
+
             InvisibleElementRemovalContext context = InvisibleElementRemovalContext.builder()
                     .reader(reader)
                     .clippingPathStack(new ClippingPathStack(page.getMediaBox()))
@@ -221,8 +223,14 @@ public class InvisibleElementRemovalService {
 
 
     private void processPath(Element pathElement, ElementWriter writer, InvisibleElementRemovalContext context) throws PDFNetException {
+        PathData pathData = pathElement.getPathData();
 
-        GeneralPath linePath = convertToGeneralPath(pathElement.getPathData());
+        if (pathData.getOperators().length == 0 && pathData.getPoints().length == 0) {
+            writer.writeGStateChanges(pathElement);
+            return;
+        }
+
+        GeneralPath linePath = convertToGeneralPath(pathData);
 
         //transform path to initial user space
         var ctm = pathElement.getCTM();

ocr-service-v1/ocr-service-server-v1/src/test/java/com/iqser/red/service/ocr/v1/server/service/ImagePositionRetrievalServiceTest.java

@@ -122,6 +122,14 @@ class ImagePositionRetrievalServiceTest {
         assertThat(allRectCoords.size()).isEqualTo(48);
     }
 
+    @Test
+    @SneakyThrows
+    public void testEveryPointInDashedLineIsImage() {
+        String fileName = "everyPointInDashedLineIsImage";
+        List<int[]> allRectCoords = testImagePositionDetection(fileName);
+        assertThat(allRectCoords.size()).isEqualTo(0);
+    }
+
 
     private List<int[]> testImagePositionDetection(String fileName) throws IOException, PDFNetException {