RED-8156: add ocr debug layers to viewer document

ocr-service-v1/ocr-service-processor/build.gradle.kts

@@ -25,6 +25,6 @@ dependencies {
     api("com.amazonaws:aws-java-sdk-kms:1.12.440")
     api("com.google.guava:guava:31.1-jre")
     api("com.iqser.red.commons:pdftron-logic-commons:2.27.0")
-    api("com.knecon.fforesight:viewer-doc-processor:0.89.0")
+    api("com.knecon.fforesight:viewer-doc-processor:0.124.0")
     testImplementation("org.junit.jupiter:junit-jupiter:5.8.1")
 }

ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/initializer/NativeLibrariesInitializer.java

@@ -6,12 +6,15 @@ import com.sun.jna.NativeLibrary;
 import jakarta.annotation.PostConstruct;
 import lombok.RequiredArgsConstructor;
 import lombok.SneakyThrows;
+import lombok.extern.slf4j.Slf4j;
+
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.stereotype.Component;
 
+@Slf4j
 @Component
 @RequiredArgsConstructor
-public class PDFNetInitializer {
+public class NativeLibrariesInitializer {
 
     @Value("${pdftron.license:}")
     private String pdftronLicense;
@@ -22,8 +25,21 @@ public class PDFNetInitializer {
     // Do not change back to application runner, if it is application runner it takes messages from the queue before PDFNet is initialized, that leads to UnsatisfiedLinkError.
     public void init() {
 
+        log.info("Initializing Native Libraries");
+        log.info("Setting pdftron license: {}", pdftronLicense);
         PDFNet.setTempPath("/tmp/pdftron");
         PDFNet.initialize(pdftronLicense);
+
+        log.info("Setting jna.library.path: {}", System.getenv("VCPKG_DYNAMIC_LIB"));
         System.setProperty("jna.library.path", System.getenv("VCPKG_DYNAMIC_LIB"));
+
+        log.info("Asserting Native Libraries loaded");
+        NativeLibrary leptonicaLib = NativeLibrary.getInstance("leptonica");
+        assert leptonicaLib != null;
+        log.info("Leptonica library loaded from {}", leptonicaLib.getFile().getAbsolutePath());
+        NativeLibrary tesseractLib = NativeLibrary.getInstance("tesseract");
+        assert tesseractLib != null;
+        log.info("Tesseract library loaded from {}", leptonicaLib.getFile().getAbsolutePath());
     }
+
 }

ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/model/OcrImage.java

@@ -93,7 +93,7 @@ public interface OcrImage {
         if (getWidth() < 200 || getHeight() < 200) {
             return ITessAPI.TessPageSegMode.PSM_SINGLE_BLOCK;
         }
-        return ITessAPI.TessPageSegMode.PSM_AUTO;
+        return ITessAPI.TessPageSegMode.PSM_SPARSE_TEXT;
     } // TODO: evaluate if PSM can be dynamically chosen to increase performance
 
 

ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/model/QuadPoint.java

@@ -1,5 +1,6 @@
 package com.knecon.fforesight.service.ocr.processor.model;
 
+import java.awt.Rectangle;
 import java.awt.geom.AffineTransform;
 import java.awt.geom.Line2D;
 import java.awt.geom.Point2D;
@@ -34,6 +35,16 @@ public record QuadPoint(Point2D a, Point2D b, Point2D c, Point2D d) {
                 new Point2D.Double(bounds.getRight(), bounds.getBottom()));
     }
 
+    public Rectangle2D getBounds2D() {
+
+        double minX = Math.min(Math.min(Math.min(a.getX(), b.getX()), c.getX()), d.getX());
+        double minY = Math.min(Math.min(Math.min(a.getY(), b.getY()), c.getY()), d.getY());
+        double maxX = Math.max(Math.max(Math.max(a.getX(), b.getX()), c.getX()), d.getX());
+        double maxY = Math.max(Math.max(Math.max(a.getY(), b.getY()), c.getY()), d.getY());
+
+        return new Rectangle2D.Double(minX, minY, maxX - minX, maxY - minY);
+    }
+
 
     public QuadPoint getTransformed(AffineTransform at) {
 

ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/service/OcrResultWriter.java

@@ -3,8 +3,12 @@ package com.knecon.fforesight.service.ocr.processor.service;
 import java.awt.Color;
 import java.awt.geom.Line2D;
 import java.awt.geom.Point2D;
+import java.awt.geom.Rectangle2D;
 import java.io.File;
+import java.io.FileInputStream;
+import java.util.ArrayList;
 import java.util.Collection;
+import java.util.HashMap;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
@@ -16,6 +20,7 @@ import java.util.stream.Stream;
 import org.apache.pdfbox.pdmodel.graphics.state.RenderingMode;
 import org.springframework.stereotype.Service;
 
+import com.iqser.red.pdftronlogic.commons.Converter;
 import com.knecon.fforesight.service.ocr.processor.model.OcrResultToWrite;
 import com.knecon.fforesight.service.ocr.processor.model.QuadPoint;
 import com.knecon.fforesight.service.ocr.processor.model.TextPositionInImage;
@@ -26,6 +31,9 @@ import com.knecon.fforesight.service.viewerdoc.model.PlacedText;
 import com.knecon.fforesight.service.viewerdoc.model.Visualizations;
 import com.knecon.fforesight.service.viewerdoc.model.VisualizationsOnPage;
 import com.knecon.fforesight.service.viewerdoc.service.ViewerDocumentService;
+import com.pdftron.pdf.PDFDoc;
+import com.pdftron.pdf.Page;
+import com.pdftron.pdf.TextExtractor;
 
 import lombok.AccessLevel;
 import lombok.RequiredArgsConstructor;
@@ -39,32 +47,62 @@ import lombok.extern.slf4j.Slf4j;
 @FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
 public class OcrResultWriter {
 
+    public static final Color REGULAR_TEXT_COLOR = Color.BLUE;
+    public static final Color BOLD_TEXT_COLOR = Color.RED;
+
+    public static final Color REGULAR_TEXT_IN_IGNORE_ZONE = Color.ORANGE;
+    public static final Color BOLD_TEXT_IN_IGNORE_ZONE = Color.MAGENTA;
+
     ViewerDocumentService viewerDocumentService;
 
 
     @SneakyThrows
     public void drawOcrResultsToPdf(File document, File viewerDocument, Map<Integer, List<OcrResultToWrite>> imagesWithResultsPerPage) {
 
-        Map<Integer, VisualizationsOnPage> ocrVisualizationsOnPages = createVisualizations(imagesWithResultsPerPage);
-        Map<Integer, VisualizationsOnPage> ocrTextDebugVisualizationsOnPages = createDebugTextVisualizations(imagesWithResultsPerPage);
-        Map<Integer, VisualizationsOnPage> ocrBBoxDebugVisualizationsOnPages = createDebugBBoxVisualizations(imagesWithResultsPerPage);
+        Map<Integer, VisualizationsOnPage> ocrVisualizationsOnPages = new HashMap<>();
+        Map<Integer, VisualizationsOnPage> ocrTextDebugVisualizationsOnPages = new HashMap<>();
+        Map<Integer, VisualizationsOnPage> ocrBBoxDebugVisualizationsOnPages = new HashMap<>();
+
+        try (var in = new FileInputStream(document); PDFDoc doc = new PDFDoc(in)) {
+
+            for (Integer pageNumber : imagesWithResultsPerPage.keySet()) {
+
+                List<Rectangle2D> textBBoxes = getTextBBoxes(doc.getPage(pageNumber));
+
+                ocrVisualizationsOnPages.put(pageNumber - 1, createVisualizations(imagesWithResultsPerPage.get(pageNumber), textBBoxes));
+                ocrTextDebugVisualizationsOnPages.put(pageNumber - 1, createDebugTextVisualizations(imagesWithResultsPerPage.get(pageNumber), textBBoxes));
+                ocrBBoxDebugVisualizationsOnPages.put(pageNumber - 1, createDebugBBoxVisualizations(imagesWithResultsPerPage.get(pageNumber)));
+            }
+        }
 
         Visualizations visualizations = new Visualizations(ContentStreams.KNECON_OCR, ocrVisualizationsOnPages, false);
 
         List<Visualizations> debugVisualizations = List.of(visualizations,
-                new Visualizations(ContentStreams.KNECON_OCR_TEXT_DEBUG, ocrTextDebugVisualizationsOnPages, false),
-                new Visualizations(ContentStreams.KNECON_OCR_BBOX_DEBUG, ocrBBoxDebugVisualizationsOnPages, false));
+                                                           new Visualizations(ContentStreams.KNECON_OCR_TEXT_DEBUG, ocrTextDebugVisualizationsOnPages, false),
+                                                           new Visualizations(ContentStreams.KNECON_OCR_BBOX_DEBUG, ocrBBoxDebugVisualizationsOnPages, false));
 
-        viewerDocumentService.addVisualizationsOnPage(document, document, visualizations);
+        viewerDocumentService.addVisualizationsOnPage(document, document, List.of(visualizations));
         viewerDocumentService.addVisualizationsOnPage(viewerDocument, viewerDocument, debugVisualizations);
     }
 
 
-    private Map<Integer, VisualizationsOnPage> createVisualizations(Map<Integer, List<OcrResultToWrite>> imagesWithResultsPerPage) {
+    private List<Rectangle2D> getTextBBoxes(Page page) {
 
-        return imagesWithResultsPerPage.keySet()
-                .stream()
-                .collect(Collectors.toMap(pageNumber1IdxTo0IdxMapper(), pageNumber -> createVisualizations(imagesWithResultsPerPage.get(pageNumber))));
+        List<Rectangle2D> textBBoxes = new ArrayList<>();
+        try (var textExtractor = new TextExtractor()) {
+            textExtractor.begin(page);
+
+            try {
+                for (TextExtractor.Line line = textExtractor.getFirstLine(); line.isValid(); line = line.getNextLine()) {
+                    for (var word = line.getFirstWord(); word.isValid(); word = word.getNextWord()) {
+                        textBBoxes.add(Converter.toRectangle2D(word.getBBox()));
+                    }
+                }
+            } catch (Exception e) {
+                log.warn("Could not get word dimension, {}", e.getMessage());
+            }
+            return textBBoxes;
+        }
     }
 
 
@@ -74,71 +112,97 @@ public class OcrResultWriter {
     }
 
 
-    private VisualizationsOnPage createVisualizations(List<OcrResultToWrite> ocrResultsToWrite) {
+    private VisualizationsOnPage createVisualizations(List<OcrResultToWrite> ocrResultsToWrite, List<Rectangle2D> ignoreZones) {
+
+        List<TextPositionInImage> words = ocrResultsToWrite.stream()
+                .map(OcrResultToWrite::textPositionInImage)
+                .flatMap(Collection::stream)
+                .filter(word -> ignoreZones.stream()
+                        .noneMatch(ignoreZone -> word.getTransformedTextBBox().getBounds2D().intersects(ignoreZone)))
+                .toList();
 
-        List<TextPositionInImage> words = ocrResultsToWrite.stream().map(OcrResultToWrite::textPositionInImage).flatMap(Collection::stream).toList();
         List<PlacedText> placedTexts = words.stream()
                 .map(word -> new PlacedText(word.getText(),
-                        null,
-                        Color.BLACK,
-                        (float) word.getFontSize(),
-                        word.getFontMetricsFactory(),
-                        Optional.of(word.getTextMatrix()),
-                        Optional.of(RenderingMode.NEITHER)))
+                                            null,
+                                            Color.BLACK,
+                                            (float) word.getFontSize(),
+                                            word.getFontMetricsFactory(),
+                                            Optional.of(word.getTextMatrix()),
+                                            Optional.of(RenderingMode.NEITHER)))
                 .toList();
         return VisualizationsOnPage.builder().placedTexts(placedTexts).build();
     }
 
 
-    private Map<Integer, VisualizationsOnPage> createDebugTextVisualizations(Map<Integer, List<OcrResultToWrite>> imagesWithResultsPerPage) {
 
-        return imagesWithResultsPerPage.keySet()
-                .stream()
-                .collect(Collectors.toMap(pageNumber1IdxTo0IdxMapper(), pageNumber -> createDebugTextVisualizations(imagesWithResultsPerPage.get(pageNumber))));
-    }
+    private VisualizationsOnPage createDebugTextVisualizations(List<OcrResultToWrite> ocrResultsToWrite, List<Rectangle2D> textBBoxes) {
 
+        List<TextPositionInImage> wordsToDraw = new ArrayList<>();
+        List<TextPositionInImage> ignoredWords = new ArrayList<>();
 
-    private VisualizationsOnPage createDebugTextVisualizations(List<OcrResultToWrite> ocrResultsToWrite) {
-
-        List<TextPositionInImage> words = ocrResultsToWrite.stream().map(OcrResultToWrite::textPositionInImage).flatMap(Collection::stream).toList();
-        List<PlacedText> placedTexts = words.stream()
+        for (OcrResultToWrite ocrResultToWrite : ocrResultsToWrite) {
+            for (TextPositionInImage textPositionInImage : ocrResultToWrite.textPositionInImage()) {
+                if (textBBoxes.stream()
+                        .anyMatch(ignoreZone -> textPositionInImage.getTransformedTextBBox().getBounds2D().intersects(ignoreZone))) {
+                    ignoredWords.add(textPositionInImage);
+                } else {
+                    wordsToDraw.add(textPositionInImage);
+                }
+            }
+        }
+        Stream<PlacedText> placedTexts = wordsToDraw.stream()
                 .map(word -> new PlacedText(word.getText(),
-                        null,
-                        word.getFontStyle().equals(FontStyle.REGULAR) ? Color.BLUE : Color.RED,
-                        (float) word.getFontSize(),
-                        word.getFontMetricsFactory(),
-                        Optional.of(word.getTextMatrix()),
-                        Optional.of(RenderingMode.FILL)))
-                .toList();
-        return VisualizationsOnPage.builder().placedTexts(placedTexts).build();
+                                            null,
+                                            word.getFontStyle().equals(FontStyle.REGULAR) ? REGULAR_TEXT_COLOR : BOLD_TEXT_COLOR,
+                                            (float) word.getFontSize(),
+                                            word.getFontMetricsFactory(),
+                                            Optional.of(word.getTextMatrix()),
+                                            Optional.of(RenderingMode.FILL)));
+
+        Stream<PlacedText> placedTexts2 = ignoredWords.stream()
+                .map(word -> new PlacedText(word.getText(),
+                                            null,
+                                            word.getFontStyle().equals(FontStyle.REGULAR) ? REGULAR_TEXT_IN_IGNORE_ZONE : BOLD_TEXT_IN_IGNORE_ZONE,
+                                            (float) word.getFontSize(),
+                                            word.getFontMetricsFactory(),
+                                            Optional.of(word.getTextMatrix()),
+                                            Optional.of(RenderingMode.FILL)));
+
+        return VisualizationsOnPage.builder()
+                .placedTexts(Stream.of(placedTexts, placedTexts2)
+                                     .flatMap(Function.identity())
+                                     .toList())
+                .build();
     }
 
 
-    private Map<Integer, VisualizationsOnPage> createDebugBBoxVisualizations(Map<Integer, List<OcrResultToWrite>> imagesWithResultsPerPage) {
-
-        return imagesWithResultsPerPage.keySet()
-                .stream()
-                .collect(Collectors.toMap(pageNumber1IdxTo0IdxMapper(), pageNumber -> createDebugBBoxVisualizations(imagesWithResultsPerPage.get(pageNumber))));
-    }
-
 
     private VisualizationsOnPage createDebugBBoxVisualizations(List<OcrResultToWrite> ocrResultsToWrite) {
 
-        List<TextPositionInImage> words = ocrResultsToWrite.stream().map(OcrResultToWrite::textPositionInImage).flatMap(Collection::stream).toList();
+        List<TextPositionInImage> words = ocrResultsToWrite.stream()
+                .map(OcrResultToWrite::textPositionInImage)
+                .flatMap(Collection::stream)
+                .toList();
         List<ColoredLine> coloredLines = Stream.concat(//
-                words.stream().map(TextPositionInImage::getTransformedTextBBox).map(this::quadPointAsLines),//
-                ocrResultsToWrite.stream().map(OcrResultToWrite::imageBoundingBox).map(this::createGrid)//
-        ).flatMap(Collection::stream).toList();
+                                                       words.stream()
+                                                               .map(TextPositionInImage::getTransformedTextBBox)
+                                                               .map(this::quadPointAsLines),//
+                                                       ocrResultsToWrite.stream()
+                                                               .map(OcrResultToWrite::imageBoundingBox)
+                                                               .map(this::createGrid)//
+                )
+                .flatMap(Collection::stream)
+                .toList();
         return VisualizationsOnPage.builder().coloredLines(coloredLines).build();
     }
 
 
     private List<ColoredLine> quadPointAsLines(QuadPoint rect) {
 
-        return List.of(new ColoredLine(new Line2D.Double(rect.a(), rect.b()), Color.ORANGE, 1),
-                new ColoredLine(new Line2D.Double(rect.b(), rect.c()), Color.BLUE, 1),
-                new ColoredLine(new Line2D.Double(rect.c(), rect.d()), Color.GREEN, 1),
-                new ColoredLine(new Line2D.Double(rect.d(), rect.a()), Color.MAGENTA, 1));
+        return List.of(new ColoredLine(new Line2D.Double(rect.a(), rect.b()), REGULAR_TEXT_IN_IGNORE_ZONE, 1),
+                       new ColoredLine(new Line2D.Double(rect.b(), rect.c()), REGULAR_TEXT_COLOR, 1),
+                       new ColoredLine(new Line2D.Double(rect.c(), rect.d()), Color.GREEN, 1),
+                       new ColoredLine(new Line2D.Double(rect.d(), rect.a()), BOLD_TEXT_IN_IGNORE_ZONE, 1));
     }
 
 

ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/service/threads/ImageProcessingThread.java

@@ -45,7 +45,7 @@ public class ImageProcessingThread extends Thread {
     final BlockingQueue<UnprocessedImage> imageInputQueue;
     final BlockingQueue<OcrImage> imageOutputQueue;
     final ITessAPI.TessBaseAPI detectionScriptHandle = initDetectionScriptHandle();
-    final L_Kernel gaussianKernel = Leptonica1.makeGaussianKernel(2, 2, 1.0f, 1);
+    final L_Kernel gaussianKernel = Leptonica1.makeGaussianKernel(2, 2, 0.7f, 1);
     final Statistics stats;
     final OcrServiceSettings settings;
     final PDDocument document;
@@ -227,7 +227,7 @@ public class ImageProcessingThread extends Thread {
         if (pix.w < 100 || pix.h < 100) {
             binarized = Leptonica1.pixThresholdToBinary(gaussian, 170);
         } else {
-            binarized = Leptonica1.pixOtsuThreshOnBackgroundNorm(gaussian, null, 50, 50, 165, 10, 100, 5, 5, 0.2f, null);
+            binarized = Leptonica1.pixOtsuThreshOnBackgroundNorm(gaussian, null, 50, 50, 165, 10, 100, 5, 5, 0.0f, null);
             if (binarized == null) { // Sometimes Otsu just fails, then we binarize directly
                 binarized = Leptonica1.pixThresholdToBinary(gaussian, 170);
             }

ocr-service-v1/ocr-service-server/src/main/resources/vcpkg.json

@@ -6,7 +6,7 @@
   "overrides": [
     {
       "name": "tesseract",
-      "version": "5.3.3"
+      "version": "5.3.4"
     },
     {
       "name": "leptonica",

ocr-service-v1/ocr-service-server/src/test/java/com/knecon/fforesight/service/ocr/v1/server/AbstractTest.java

@@ -24,10 +24,11 @@ import org.springframework.context.annotation.Primary;
 import org.springframework.test.context.junit.jupiter.SpringExtension;
 
 import com.iqser.red.commons.jackson.ObjectMapperFactory;
-import com.knecon.fforesight.service.ocr.processor.initializer.PDFNetInitializer;
+import com.knecon.fforesight.service.ocr.processor.initializer.NativeLibrariesInitializer;
 import com.iqser.red.storage.commons.StorageAutoConfiguration;
 import com.iqser.red.storage.commons.service.StorageService;
 import com.iqser.red.storage.commons.utils.FileSystemBackedStorageService;
+import com.knecon.fforesight.service.ocr.v1.server.queue.OcrMessageSender;
 import com.knecon.fforesight.tenantcommons.TenantsClient;
 import com.pdftron.pdf.PDFNet;
 
@@ -36,7 +37,7 @@ import lombok.SneakyThrows;
 
 @ExtendWith({SpringExtension.class, MockitoExtension.class})
 @SpringBootTest(classes = Application.class, webEnvironment = SpringBootTest.WebEnvironment.RANDOM_PORT)
-@Import({AbstractTest.TestConfiguration.class, PDFNetInitializer.class})
+@Import({AbstractTest.TestConfiguration.class, NativeLibrariesInitializer.class})
 @AutoConfigureObservability
 public class AbstractTest {
 
@@ -46,6 +47,9 @@ public class AbstractTest {
     @MockBean
     private TenantsClient tenantsClient;
 
+    @MockBean
+    private OcrMessageSender ocrMessageSender;
+
     @Autowired
     protected StorageService storageService;
 

ocr-service-v1/ocr-service-server/src/test/java/com/knecon/fforesight/service/ocr/v1/server/OcrServiceIntegrationTest.java

@@ -10,30 +10,46 @@ import java.io.FileOutputStream;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.nio.file.StandardCopyOption;
+import java.util.ArrayList;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.List;
+import java.util.Set;
+import java.util.TreeSet;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicInteger;
 
-import org.junit.jupiter.api.BeforeEach;
-import org.junit.jupiter.api.Disabled;
 import org.junit.jupiter.api.Test;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.boot.test.context.SpringBootTest;
 import org.springframework.core.io.ClassPathResource;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
-import com.knecon.fforesight.service.ocr.processor.service.FileStorageService;
+import com.iqser.red.pdftronlogic.commons.ClippingPathStack;
+import com.iqser.red.pdftronlogic.commons.ElementFeatures;
+import com.iqser.red.pdftronlogic.commons.MarkedContentStack;
 import com.knecon.fforesight.service.ocr.processor.service.OCRService;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;
 import com.knecon.fforesight.service.ocr.processor.service.OsUtils;
-import com.knecon.fforesight.tenantcommons.TenantContext;
+import com.pdftron.common.PDFNetException;
+import com.pdftron.pdf.ColorPt;
+import com.pdftron.pdf.ColorSpace;
+import com.pdftron.pdf.Element;
+import com.pdftron.pdf.ElementReader;
+import com.pdftron.pdf.ElementWriter;
+import com.pdftron.pdf.GState;
+import com.pdftron.pdf.PDFDoc;
+import com.pdftron.pdf.Page;
+import com.pdftron.pdf.PageIterator;
+import com.pdftron.pdf.Rect;
+import com.pdftron.sdf.Obj;
+import com.pdftron.sdf.SDFDoc;
 
 import io.micrometer.prometheus.PrometheusMeterRegistry;
 import io.micrometer.prometheus.PrometheusTimer;
+import lombok.Builder;
 import lombok.SneakyThrows;
 
-@Disabled // Ghostscript/Tesseract/Leptonica is not available on build server. If you want to run the test install these dependencies. See README.md for help.
+//@Disabled // Ghostscript/Tesseract/Leptonica is not available on build server. If you want to run the test install these dependencies. See README.md for help.
 @SpringBootTest()
 public class OcrServiceIntegrationTest extends AbstractTest {
 
@@ -54,7 +70,10 @@ public class OcrServiceIntegrationTest extends AbstractTest {
         testOCR("files/Watermark.pdf");
         testOCR("files/Watermark.pdf");
 
-        var ocrOnDocumentMeter = registry.getMeters().stream().filter(m -> m.getId().getName().equalsIgnoreCase("redactmanager_runOcrOnDocument")).findAny();
+        var ocrOnDocumentMeter = registry.getMeters()
+                .stream()
+                .filter(m -> m.getId().getName().equalsIgnoreCase("redactmanager_runOcrOnDocument"))
+                .findAny();
         assertThat(ocrOnDocumentMeter.isPresent()).isTrue();
         PrometheusTimer timer = (PrometheusTimer) ocrOnDocumentMeter.get();
         assertThat(timer.count()).isEqualTo(3);
@@ -66,7 +85,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
     @SneakyThrows
     public void testOcr() {
 
-        String text = testOCR("files/402Study.pdf");
+        String text = testOCR("files/13485 cert example 2.pdf");
     }
 
 
@@ -83,18 +102,18 @@ public class OcrServiceIntegrationTest extends AbstractTest {
         // check visually for most of the images containing text, the resulting text is kind of nonsense, just ensure it is there
         String text = testOCR("files/merge_images.pdf");
         assertThat(text).contains("Bodyweight change of dams with live young - group mean values",
-                "Control",
-                "mg/g day",
-                "10 mg/kg/day",
-                "20 mg/kg/",
-                "Days",
-                "50",
-                "-200",
-                "—250",
-                "150",
-                "200",
-                "250",
-                "—150");
+                                  "Control",
+                                  "mg/g day",
+                                  "10 mg/kg/day",
+                                  "20 mg/kg/",
+                                  "Days",
+                                  "50",
+                                  "-200",
+                                  "—250",
+                                  "150",
+                                  "200",
+                                  "250",
+                                  "—150");
     }
 
 
@@ -138,7 +157,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
     @SneakyThrows
     public void testOcrForAllDMFiles() {
 
-        String dir = "/home/kschuettler/Dokumente/TestFiles/syn-dm-testfiles/";
+        String dir = "/home/kschuettler/Dokumente/TestFiles/certificates/no-ocr";
         List<File> foundFiles = Files.walk(Path.of(dir))
                 .sorted(Comparator.comparingLong(this::getFileSize))
                 .map(Path::toFile)
@@ -148,7 +167,9 @@ public class OcrServiceIntegrationTest extends AbstractTest {
         int fileCount = foundFiles.size();
         AtomicInteger processedCount = new AtomicInteger();
         System.out.printf("Found %s files, starting OCR for each.%n%n", fileCount);
-        foundFiles.stream().peek(file -> System.out.printf("%s/%s: %s%n", processedCount.getAndIncrement(), fileCount, file)).forEach(this::testOCRForFile);
+        foundFiles.stream()
+                .peek(file -> System.out.printf("%s/%s: %s%n", processedCount.getAndIncrement(), fileCount, file))
+                .forEach(this::testOCRForFile);
     }
 
 
@@ -182,4 +203,146 @@ public class OcrServiceIntegrationTest extends AbstractTest {
         System.out.println("\n\n");
     }
 
+
+    @SneakyThrows
+    @Test
+    public void testMakeTextVisible() {
+
+        File file = new File("/home/kschuettler/Downloads/18-Curacron_ToxicidadeOcularInVitro.pdf");
+        PDFDoc doc;
+        try (var in = new FileInputStream(file)) {
+
+            doc = new PDFDoc(in);
+        }
+        execute(doc, false, false, Collections.emptySet());
+
+        try (var out = new FileOutputStream("/tmp/test.pdf")) {
+            doc.save(out, SDFDoc.SaveMode.LINEARIZED, null);
+        }
+    }
+
+
+    @SneakyThrows
+    private void execute(PDFDoc pdfDoc, boolean delta, boolean removePaths, Set<String> markedContentToIgnore) {
+
+        ElementWriter writer = new ElementWriter();
+        ElementReader reader = new ElementReader();
+        Set<Long> visitedXObjIds = new TreeSet<>();
+
+        for (PageIterator iterator = pdfDoc.getPageIterator(); iterator.hasNext(); ) {
+
+            Page page = iterator.next();
+
+            visitedXObjIds.add(page.getSDFObj().getObjNum());
+
+            Context context = Context.builder()
+                    .reader(reader)
+                    .clippingPathStack(new ClippingPathStack(page.getMediaBox()))
+                    .markedContentStack(new MarkedContentStack())
+                    .removePaths(removePaths)
+                    .delta(delta)
+                    .overlappedElements(new ArrayList<>())
+                    .visibleElements(new ArrayList<>())
+                    .visitedXObjIds(visitedXObjIds)
+                    .markedContentToIgnore(markedContentToIgnore)
+                    .build();
+
+            removeClippedElementsAndInvisibleTextAndRememberOverlappedElements(page, writer, context);
+
+        }
+        writer.destroy();
+        reader.destroy();
+
+    }
+
+
+    private void removeClippedElementsAndInvisibleTextAndRememberOverlappedElements(Page page, ElementWriter writer, Context context) throws PDFNetException {
+
+        context.reader().begin(page);
+        context.markedContentStack().clear();
+        writer.begin(page, ElementWriter.e_replacement, false, true, page.getResourceDict());
+        processElements(writer, context);
+        writer.end();
+        context.reader().end();
+    }
+
+
+    private void processElements(ElementWriter writer, Context context) throws PDFNetException {
+
+        for (Element element = context.reader().next(); element != null; element = context.reader().next()) {
+
+            if (context.markedContentStack().currentMarkedContentContainsAny(context.markedContentToIgnore()) && element.getType() != Element.e_marked_content_end) {
+                writer.writeElement(element);
+                continue;
+            }
+
+            switch (element.getType()) {
+                case Element.e_text -> processText(element, writer, context);
+                case Element.e_form -> processForm(element, writer, context);
+                default -> writer.writeElement(element);
+            }
+        }
+    }
+
+
+    private void processForm(Element formElement, ElementWriter writer, Context context) throws PDFNetException {
+
+        writer.writeElement(formElement);
+        Obj formObj = formElement.getXObject();
+
+        if (!context.visitedXObjIds().contains(formObj.getObjNum())) {
+            context.visitedXObjIds().add(formObj.getObjNum());
+            // writer needs to be newly initialized when entering a new content stream
+            // see ElementEditTest in PDFTron (https://www.pdftron.com/documentation/samples/android/java/ElementEditTest)
+            ElementWriter formWriter = new ElementWriter();
+            context.reader().formBegin();
+            formWriter.begin(formObj);
+
+            context.reader().clearChangeList();
+            formWriter.setDefaultGState(context.reader());
+
+            processElements(formWriter, context);
+            formWriter.end();
+            formWriter.destroy();
+            context.reader().end();
+        }
+    }
+
+
+    private void processText(Element textElement, ElementWriter writer, Context context) throws PDFNetException {
+
+        Rect textBBox = textElement.getBBox();
+
+        if (textBBox == null) {
+            writer.writeElement(textElement);
+            return;
+        }
+
+        GState gState = textElement.getGState();
+
+        gState.setFillColorSpace(ColorSpace.createDeviceRGB());
+        // blue for elements removed due to transparency or not rendered or same color as background
+        gState.setFillColor(new ColorPt(0, 0, 1));
+        gState.setTextRenderMode(GState.e_fill_text);
+        gState.setFillOpacity(1);
+        writer.writeElement(textElement);
+
+    }
+
+
+    @Builder
+    record Context(
+            boolean removePaths,
+            boolean delta,
+            ElementReader reader,
+            ClippingPathStack clippingPathStack,
+            MarkedContentStack markedContentStack,
+            List<ElementFeatures> overlappedElements,
+            List<ElementFeatures> visibleElements,
+            Set<Long> visitedXObjIds,
+            Set<String> markedContentToIgnore
+    ) {
+
+    }
+
 }