Merge branch 'RED-3813' into 'main'

RED-3813: Recategorize same image as experimental feature

See merge request fforesight/layout-parser!155

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentStructure.java

@@ -43,6 +43,8 @@ public class DocumentStructure implements Serializable {
         public static final String POSITION = "position";
         public static final String ID = "id";
 
+        public static final String REPRESENTATION_HASH = "representationHash";
+
     }
 
     @Schema(description = "Object containing the extra field names, a table cell has in its properties field.")
@@ -67,11 +69,25 @@ public class DocumentStructure implements Serializable {
 
     public static Rectangle2D parseRectangle2D(String bBox) {
 
-        List<Float> floats = Arrays.stream(bBox.split(RECTANGLE_DELIMITER)).map(Float::parseFloat).toList();
+        List<Float> floats = Arrays.stream(bBox.split(RECTANGLE_DELIMITER))
+                .map(Float::parseFloat)
+                .toList();
         return new Rectangle2D.Float(floats.get(0), floats.get(1), floats.get(2), floats.get(3));
     }
 
 
+    public static double[] parseRepresentationVector(String representationHash) {
+
+        String[] stringArray = representationHash.split("[,\\s]+");
+        double[] doubleArray = new double[stringArray.length];
+        for (int i = 0; i < stringArray.length; i++) {
+            doubleArray[i] = Double.parseDouble(stringArray[i]);
+        }
+
+        return doubleArray;
+    }
+
+
     public EntryData get(List<Integer> tocId) {
 
         if (tocId.isEmpty()) {
@@ -87,19 +103,24 @@ public class DocumentStructure implements Serializable {
 
     public Stream<EntryData> streamAllEntries() {
 
-        return Stream.concat(Stream.of(root), root.children.stream()).flatMap(DocumentStructure::flatten);
+        return Stream.concat(Stream.of(root), root.children.stream())
+                .flatMap(DocumentStructure::flatten);
     }
 
 
     public String toString() {
 
-        return String.join("\n", streamAllEntries().map(EntryData::toString).toList());
+        return String.join("\n",
+                           streamAllEntries().map(EntryData::toString)
+                                   .toList());
     }
 
 
     private static Stream<EntryData> flatten(EntryData entry) {
 
-        return Stream.concat(Stream.of(entry), entry.children.stream().flatMap(DocumentStructure::flatten));
+        return Stream.concat(Stream.of(entry),
+                             entry.children.stream()
+                                     .flatMap(DocumentStructure::flatten));
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingPipeline.java

@@ -148,14 +148,17 @@ public class LayoutParsingPipeline {
                                                                     visualLayoutParsingResponse,
                                                                     layoutParsingRequest.identifier());
 
-        log.info("Building document graph for {}", layoutParsingRequest.identifier());
-
         Document documentGraph = observeBuildDocumentGraph(settings.getLayoutParsingTypeOverride() == null //
                                                                    ? layoutParsingRequest.layoutParsingType() : settings.getLayoutParsingTypeOverride(), classificationDocument);
 
         log.info("Creating viewer document for {}", layoutParsingRequest.identifier());
 
-        layoutGridService.addLayoutGrid(viewerDocumentFile, documentGraph, viewerDocumentFile, false, layoutParsingRequest.visualLayoutParsingFileId().isPresent());
+        layoutGridService.addLayoutGrid(viewerDocumentFile,
+                                        documentGraph,
+                                        viewerDocumentFile,
+                                        false,
+                                        layoutParsingRequest.visualLayoutParsingFileId()
+                                                .isPresent());
 
         log.info("Storing resulting files for {}", layoutParsingRequest.identifier());
 
@@ -312,7 +315,11 @@ public class LayoutParsingPipeline {
 
             pdfImages.computeIfAbsent(pageNumber, x -> new ArrayList<>())
                     .addAll(graphics.stream()
-                                    .map(g -> new ClassifiedImage(new Rectangle2D.Double(g.x1, g.y1, g.width(), g.height()), ImageType.GRAPHIC, false, stripper.getPageNumber()))
+                                    .map(g -> new ClassifiedImage(new Rectangle2D.Double(g.x1, g.y1, g.width(), g.height()),
+                                                                  ImageType.GRAPHIC,
+                                                                  false,
+                                                                  stripper.getPageNumber(),
+                                                                  ""))
                                     .toList());
 
             ClassificationPage classificationPage = switch (layoutParsingType) {

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Image.java

@@ -28,6 +28,8 @@ public class Image extends AbstractSemanticNode {
 
     String id;
 
+    String representationHash;
+
     ImageType imageType;
     boolean transparent;
     Rectangle2D position;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/image/ClassifiedImage.java

@@ -24,5 +24,7 @@ public class ClassifiedImage {
     private boolean hasTransparency;
     @NonNull
     private int page;
+    @NonNull
+    private String representation;
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/adapter/ImageServiceResponseAdapter.java

@@ -32,7 +32,7 @@ public class ImageServiceResponseAdapter {
                     .add(new ClassifiedImage(new Rectangle2D.Double(imageMetadata.getPosition().getX1(),
                             imageMetadata.getPosition().getY1(),
                             imageMetadata.getGeometry().getWidth(),
-                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber(),imageMetadata.getRepresentation()));
         });
 
         // Currently This is a copy but, it will be changed later because i don' t think that we should unclassified images.
@@ -44,7 +44,7 @@ public class ImageServiceResponseAdapter {
                     .add(new ClassifiedImage(new Rectangle2D.Double(imageMetadata.getPosition().getX1(),
                             imageMetadata.getPosition().getY1(),
                             imageMetadata.getGeometry().getWidth(),
-                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber(),imageMetadata.getRepresentation()));
         });
 
         return images;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/adapter/VisualLayoutParsingAdapter.java

@@ -79,7 +79,7 @@ public class VisualLayoutParsingAdapter {
                 ClassifiedImage signature = new ClassifiedImage(new Rectangle2D.Float(t.getBox().getX1(),
                         t.getBox().getY1(),
                         t.getBox().getX2() - t.getBox().getX1(),
-                        t.getBox().getY2() - t.getBox().getY1()), ImageType.SIGNATURE, true, false, false, pageNumber);
+                        t.getBox().getY2() - t.getBox().getY1()), ImageType.SIGNATURE, true, false, false, pageNumber,"");
 
                 signatures.add(signature);
             }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/model/image/ImageMetadata.java

@@ -12,6 +12,7 @@ import lombok.NoArgsConstructor;
 public class ImageMetadata {
 
     private Classification classification;
+    private String representation;
     private Position position;
     private Geometry geometry;
     private Filters filters;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/DocumentGraphFactory.java

@@ -45,8 +45,10 @@ import lombok.Builder;
 import lombok.Getter;
 import lombok.experimental.FieldDefaults;
 import lombok.experimental.UtilityClass;
+import lombok.extern.slf4j.Slf4j;
 
 @UtilityClass
+@Slf4j
 public class DocumentGraphFactory {
 
     public Document buildDocumentGraph(LayoutParsingType layoutParsingType, ClassificationDocument document) {
@@ -145,6 +147,7 @@ public class DocumentGraphFactory {
                 .position(position)
                 .transparent(image.isHasTransparency())
                 .page(page)
+                .representationHash(image.getRepresentation())
                 .documentTree(context.getDocumentTree())
                 .build();
         page.getMainBody().add(imageNode);

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/mapper/PropertiesMapper.java

@@ -24,6 +24,7 @@ public class PropertiesMapper {
         properties.put(DocumentStructure.ImageProperties.TRANSPARENT, String.valueOf(image.isTransparent()));
         properties.put(DocumentStructure.ImageProperties.POSITION, toString(image.getPosition()));
         properties.put(DocumentStructure.ImageProperties.ID, image.getId());
+        properties.put(DocumentStructure.ImageProperties.REPRESENTATION_HASH, image.getRepresentationHash());
         return properties;
     }
 

layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/segmentation/PdfSegmentationServiceTest.java

@@ -190,7 +190,7 @@ public class PdfSegmentationServiceTest extends AbstractTest {
                                                                         imageMetadata.getGeometry().getHeight()),
                                                  ImageType.valueOf(imageMetadata.getClassification().getLabel().toUpperCase(Locale.ROOT)),
                                                  imageMetadata.isAlpha(),
-                                                 imageMetadata.getPosition().getPageNumber())));
+                                                 imageMetadata.getPosition().getPageNumber(), "")));
 
         System.out.println("object");
     }