CLARI-002: fix some stuff with DocumentDataParser

* still todo, exlude semanticNodes inside TableCells

buildSrc/src/main/kotlin/com.knecon.fforesight.java-conventions.gradle.kts

@@ -42,15 +42,6 @@ tasks.jacocoTestReport {
 }
 
 allprojects {
-
-    tasks.withType<Javadoc> {
-        options {
-            this as StandardJavadocDocletOptions
-            addBooleanOption("Xdoclint:none", true)
-            addStringOption("Xmaxwarns", "1")
-        }
-    }
-
     publishing {
         publications {
             create<MavenPublication>(name) {
@@ -73,7 +64,6 @@ java {
     withJavadocJar()
 }
 
-
 repositories {
     mavenLocal()
     mavenCentral()

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/clarifynd/IndexData.java

@@ -0,0 +1,11 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data.clarifynd;
+
+import java.util.List;
+import java.util.Map;
+
+public class IndexData {
+
+    Map<String, String> identifier;
+    List<TextChunk> textChunks;
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/clarifynd/TextChunk.java

@@ -0,0 +1,18 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data.clarifynd;
+
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.Range;
+
+import lombok.AccessLevel;
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.experimental.FieldDefaults;
+
+@Builder
+@AllArgsConstructor
+@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
+public class TextChunk {
+
+    String text;
+}

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

@@ -43,8 +43,6 @@ 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.")
@@ -69,25 +67,11 @@ 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()) {
@@ -103,24 +87,19 @@ 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-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/SimplifiedText.java

@@ -19,16 +19,6 @@ public class SimplifiedText {
     @Schema(description = "Number of pages in the entire document.")
     private int numberOfPages;
     @Schema(description = "A List of simplified Sections, which contains almost exclusively text.")
-    @Builder.Default
     private List<SimplifiedSectionText> sectionTexts = new ArrayList<>();
-    @Schema(description = "A list of the main section numbers ")
-    @Builder.Default
-    private List<String> mainSectionNumbers = new ArrayList<>();
-    @Schema(description = "A list of the header section numbers ")
-    @Builder.Default
-    private List<String> headerSectionNumbers = new ArrayList<>();
-    @Schema(description = "A list of the footer section numbers ")
-    @Builder.Default
-    private List<String> footerSectionNumbers = new ArrayList<>();
 
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/taas/StructureObject.java

@@ -1,7 +1,5 @@
 package com.knecon.fforesight.service.layoutparser.internal.api.data.taas;
 
-import java.util.List;
-
 import io.swagger.v3.oas.annotations.media.Schema;
 import lombok.AllArgsConstructor;
 import lombok.Builder;
@@ -15,8 +13,6 @@ public class StructureObject {
 
     @Schema(description = "The ID of this StructureObject.")
     Integer structureObjectNumber;
-    @Schema(description = "The Tree ID of this StructureObject.")
-    List<Integer> treeId;
     @Schema(description = "This value indicates the start of the string offsets in this Object, with respect to the reading order.")
     int page;
     @Schema(description = "This stringOffset indicates the start of the string offsets in this Object, with respect to the reading order of the entire document. It is equal to the previous' StructureObject stringOffset + its length.")

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingQueueNames.java

@@ -2,9 +2,6 @@ package com.knecon.fforesight.service.layoutparser.internal.api.queue;
 
 public class LayoutParsingQueueNames {
 
-    public static final String LAYOUT_PARSING_REQUEST_QUEUE_PREFIX = "layout_parsing_request";
-    public static final String LAYOUT_PARSING_REQUEST_EXCHANGE = "layout_parsing_request_exchange";
-    public static final String LAYOUT_PARSING_RESPONSE_QUEUE_PREFIX = "layout_parsing_response";
-    public static final String LAYOUT_PARSING_RESPONSE_EXCHANGE = "layout_parsing_response_exchange";
-    public static final String LAYOUT_PARSING_DLQ = "layout_parsing_error";
+    public static final String LAYOUT_PARSING_REQUEST_QUEUE = "layout_parsing_request_queue";
+    public static final String LAYOUT_PARSING_FINISHED_EVENT_QUEUE = "layout_parsing_response_queue";
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingRequest.java

@@ -10,37 +10,24 @@ import lombok.NonNull;
 @Builder
 @Schema(description = "Object containing all storage paths the service needs to know.")
 public record LayoutParsingRequest(
-        @Schema(description = "Enum specifying the type of layout parsing to be performed.", allowableValues = "{RedactManager, DocuMine, TAAS}")//
         @NonNull LayoutParsingType layoutParsingType,
 
-        @Schema(description = "General purpose identifiers. They are not changed by the service at all and are returned as is in the response queue.")//
         Map<String, String> identifier,
 
-        @Schema(description = "Path to the original PDF file.")//
-        @NonNull String originFileStorageId,//
+        @NonNull String originFileStorageId,
 
-        @Schema(description = "Optional Path to the table extraction file.")//
-        Optional<String> tablesFileStorageId,//
-        @Schema(description = "Optional Path to the image classification file.")//
-        Optional<String> imagesFileStorageId,//
+        Optional<String> tablesFileStorageId,
+        Optional<String> imagesFileStorageId,
 
-        @Schema(description = "Optional Path to the the visual layout parsing service file") Optional<String> visualLayoutParsingFileId,//
+        Optional<String> visualLayoutParsingFileId,
 
-        @Schema(description = "Path where the Document Structure File will be stored.")//
-        @NonNull String structureFileStorageId,//
-        @Schema(description = "Path where the Research Data File will be stored.")//
-        String researchDocumentStorageId,//
-        @Schema(description = "Path where the Document Text File will be stored.")//
-        @NonNull String textBlockFileStorageId,//
-        @Schema(description = "Path where the Document Positions File will be stored.")//
-        @NonNull String positionBlockFileStorageId,//
-        @Schema(description = "Path where the Document Pages File will be stored.")//
-        @NonNull String pageFileStorageId,//
-        @Schema(description = "Path where the Document Markdown File will be stored.")//
-        Optional<String> documentMarkdownFileStorageId,//
-        @Schema(description = "Path where the Simplified Text File will be stored.")//
-        @NonNull String simplifiedTextStorageId,//
-        @Schema(description = "Path where the Viewer Document PDF will be stored.")//
+        @NonNull String structureFileStorageId,
+        String researchDocumentStorageId,
+        String markdownDocumentStorageId,
+        @NonNull String textBlockFileStorageId,
+        @NonNull String positionBlockFileStorageId,
+        @NonNull String pageFileStorageId,
+        @NonNull String simplifiedTextStorageId,
         @NonNull String viewerDocumentStorageId
 ) {
 

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingType.java

@@ -8,5 +8,6 @@ public enum LayoutParsingType {
     DOCUMINE,
     DOCUMINE_OLD,
     CLARIFYND,
-    CLARIFYND_PARAGRAPH_DEBUG
+    CLARIFYND_PARAGRAPH_DEBUG,
+    MARKDOWN
 }

layoutparser-service/layoutparser-service-processor/build.gradle.kts

@@ -16,7 +16,7 @@ dependencies {
         exclude("org.springframework.boot", "spring-boot-starter-security")
         exclude("org.springframework.boot", "spring-boot-starter-validation")
     }
-    implementation("com.knecon.fforesight:tenant-commons:0.29.0")
+    implementation("com.knecon.fforesight:tenant-commons:0.21.0")
     implementation("com.iqser.red.commons:storage-commons:2.45.0")
 
     implementation("org.apache.pdfbox:pdfbox:${pdfBoxVersion}")
@@ -28,7 +28,8 @@ dependencies {
     implementation("org.tinspin:tinspin-indexes:2.1.3")
     implementation("org.commonmark:commonmark:0.22.0")
     implementation("org.commonmark:commonmark-ext-gfm-tables:0.22.0")
-    implementation("com.pdftron:PDFNet:10.11.0")
-    implementation("org.apache.commons:commons-text:1.12.0")
+    implementation("com.didalgo:gpt3-tokenizer:0.1.8")
 
+    implementation("org.mapstruct:mapstruct:1.5.5.Final")
+    annotationProcessor("org.mapstruct:mapstruct-processor:1.5.5.Final")
 }

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

@@ -4,12 +4,12 @@ import static java.lang.String.format;
 
 import java.awt.geom.Point2D;
 import java.awt.geom.Rectangle2D;
+import java.io.ByteArrayInputStream;
 import java.io.File;
 import java.io.IOException;
 import java.nio.file.Files;
 import java.util.ArrayList;
 import java.util.HashMap;
-import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.atomic.AtomicReference;
@@ -19,13 +19,15 @@ import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.apache.pdfbox.pdmodel.common.PDRectangle;
 import org.apache.pdfbox.pdmodel.documentinterchange.markedcontent.PDMarkedContent;
+import org.commonmark.ext.gfm.tables.TablesExtension;
+import org.commonmark.renderer.markdown.MarkdownRenderer;
 import org.springframework.stereotype.Service;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingFinishedEvent;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
-import com.knecon.fforesight.service.layoutparser.processor.markdown.MarkdownMapper;
+import com.knecon.fforesight.service.layoutparser.processor.markdown.DocumentDataParser;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
@@ -70,7 +72,6 @@ import com.knecon.fforesight.service.layoutparser.processor.services.parsing.PDF
 import com.knecon.fforesight.service.layoutparser.processor.services.visualization.LayoutGridService;
 import com.knecon.fforesight.service.layoutparser.processor.utils.MarkedContentUtils;
 import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
-import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionOperations;
 
 import io.micrometer.observation.Observation;
 import io.micrometer.observation.ObservationRegistry;
@@ -119,14 +120,23 @@ public class LayoutParsingPipeline {
         log.info("Starting layout parsing for {}", layoutParsingRequest.identifier());
 
         File originFile = layoutParsingStorageService.getOriginFile(layoutParsingRequest.originFileStorageId());
-        File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId()).orElse(originFile);
+//        File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId()).orElse(originFile);
+        File viewerDocumentFile = originFile;
 
-        VisualLayoutParsingResponse visualLayoutParsingResponse = layoutParsingRequest.visualLayoutParsingFileId()
-                .map(layoutParsingStorageService::getVisualLayoutParsingFile).orElse(new VisualLayoutParsingResponse());
-        ImageServiceResponse imageServiceResponse = layoutParsingRequest.imagesFileStorageId()
-                .map(layoutParsingStorageService::getImagesFile).orElse(new ImageServiceResponse());
-        TableServiceResponse tableServiceResponse = layoutParsingRequest.tablesFileStorageId()
-                .map(layoutParsingStorageService::getTablesFile).orElse(new TableServiceResponse());
+        VisualLayoutParsingResponse visualLayoutParsingResponse = new VisualLayoutParsingResponse();
+        if (layoutParsingRequest.visualLayoutParsingFileId().isPresent()) {
+            visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId().get());
+        }
+
+        ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
+        if (layoutParsingRequest.imagesFileStorageId().isPresent()) {
+            imageServiceResponse = layoutParsingStorageService.getImagesFile(layoutParsingRequest.imagesFileStorageId().get());
+        }
+
+        TableServiceResponse tableServiceResponse = new TableServiceResponse();
+        if (layoutParsingRequest.tablesFileStorageId().isPresent()) {
+            tableServiceResponse = layoutParsingStorageService.getTablesFile(layoutParsingRequest.tablesFileStorageId().get());
+        }
 
         ClassificationDocument classificationDocument = parseLayout(settings.getLayoutParsingTypeOverride() == null //
                                                                             ? layoutParsingRequest.layoutParsingType() : settings.getLayoutParsingTypeOverride(),
@@ -148,9 +158,6 @@ public class LayoutParsingPipeline {
         log.info("Storing resulting files for {}", layoutParsingRequest.identifier());
 
         layoutParsingStorageService.storeDocumentData(layoutParsingRequest, DocumentDataMapper.toDocumentData(documentGraph));
-        if (layoutParsingRequest.documentMarkdownFileStorageId().isPresent()) {
-            layoutParsingStorageService.storeMarkdownFile(layoutParsingRequest.documentMarkdownFileStorageId().get(), new MarkdownMapper().toMarkdownContent(documentGraph));
-        }
         layoutParsingStorageService.storeSimplifiedText(layoutParsingRequest, simplifiedSectionTextService.toSimplifiedText(documentGraph));
         layoutParsingStorageService.storeViewerDocument(layoutParsingRequest, viewerDocumentFile);
 
@@ -160,6 +167,16 @@ public class LayoutParsingPipeline {
             layoutParsingStorageService.storeResearchDocumentData(layoutParsingRequest, researchDocumentData);
         }
 
+        if (layoutParsingRequest.markdownDocumentStorageId() != null) {
+            log.info("Rendering document data as markdown for {}", layoutParsingRequest.identifier());
+            var markdownDocument = DocumentDataParser.parse(documentGraph.streamAllSubNodes());
+            MarkdownRenderer renderer = MarkdownRenderer.builder().extensions(List.of(TablesExtension.create())).build();
+            String markdown = renderer.render(markdownDocument);
+            try (var in = new ByteArrayInputStream(markdown.getBytes())) {
+                layoutParsingStorageService.storeObject(layoutParsingRequest.markdownDocumentStorageId(), in);
+            }
+        }
+
         if (!viewerDocumentFile.equals(originFile)) {
             viewerDocumentFile.delete();
         }
@@ -235,18 +252,17 @@ public class LayoutParsingPipeline {
         Map<Integer, List<TableCells>> pdfTableCells = cvTableParsingAdapter.buildCvParsedTablesPerPage(tableServiceResponse);
         Map<Integer, List<ClassifiedImage>> pdfImages = imageServiceResponseAdapter.buildClassifiedImagesPerPage(imageServiceResponse);
         Map<Integer, List<ClassifiedImage>> signatures = visualLayoutParsingAdapter.buildExtractedSignaturesPerPage(visualLayoutParsingResponse);
-
         ClassificationDocument classificationDocument = new ClassificationDocument();
 
         if (settings.isDebug() || identifier.containsKey("debug")) {
-            classificationDocument.getLayoutDebugLayer().setActive(true);
+            classificationDocument.getVisualizations().setActive(true);
         }
 
         List<ClassificationPage> classificationPages = new ArrayList<>();
         OutlineObject lastProcessedOutlineObject = null;
 
         // parsing the structure elements could be useful as well
-        if (layoutParsingType != LayoutParsingType.REDACT_MANAGER_OLD) {
+        if (layoutParsingType != LayoutParsingType.REDACT_MANAGER_OLD && layoutParsingType != LayoutParsingType.DOCUMINE_OLD) {
             classificationDocument.setOutlineObjectTree(outlineExtractorService.getOutlineObjectTree(originDocument));
         }
 
@@ -272,14 +288,12 @@ public class LayoutParsingPipeline {
             stripper.setStartPage(pageNumber);
             stripper.setEndPage(pageNumber);
             stripper.setPdpage(pdPage);
+            if (layoutParsingType.equals(LayoutParsingType.DOCUMINE_OLD)) {
+                stripper.setSortByPosition(true);
+            }
             stripper.getText(originDocument);
             List<TextPositionSequence> words = stripper.getTextPositionSequences();
-            if (layoutParsingType.equals(LayoutParsingType.DOCUMINE_OLD)) {
-                var lines = TextPositionOperations.groupByLine(new HashSet<>(words));
-                classificationDocument.getLayoutDebugLayer().addLineVisualizationsFromNestedTextPosition(lines, pageNumber);
-                words = TextPositionOperations.sortLines(lines);
-            }
-            classificationDocument.getLayoutDebugLayer().addTextVisualizations(words, pageNumber);
+            classificationDocument.getVisualizations().addTextVisualizations(words, pageNumber);
 
             PDRectangle pdr = pdPage.getMediaBox();
 
@@ -287,34 +301,33 @@ public class LayoutParsingPipeline {
             boolean isLandscape = pdr.getWidth() > pdr.getHeight() && (rotation == 0 || rotation == 180) || pdr.getHeight() > pdr.getWidth() && (rotation == 90 || rotation == 270);
 
             PDRectangle cropbox = pdPage.getCropBox();
-            classificationDocument.getLayoutDebugLayer().addRulingVisualization(stripper.getRulings(), pageNumber);
+            classificationDocument.getVisualizations().addRulingVisualization(stripper.getRulings(), pageNumber);
             CleanRulings cleanRulings = rulingCleaningService.deduplicateAndStraightenRulings(pdfTableCells.get(pageNumber), stripper.getRulings());
 
             PageInformation pageInformation = PageInformation.fromPDPage(pageNumber, pdPage);
             List<Cell> emptyTableCells = TableExtractionService.findCells(cleanRulings.getHorizontals(), cleanRulings.getVerticals(), pageInformation);
-            classificationDocument.getLayoutDebugLayer().addCellVisualizations(emptyTableCells, pageNumber);
+            classificationDocument.getVisualizations().addCellVisualizations(emptyTableCells, pageNumber);
 
             TextRulingsClassifier.classifyUnderlinedAndStrikethroughText(words, cleanRulings);
 
-            List<Box> graphics = graphicExtractorService.extractPathElementGraphics(originDocument, pdPage, pageNumber, cleanRulings, stripper.getTextPositionSequences(), false);
+            List<Box> graphics = graphicExtractorService.extractPathElementGraphics(originDocument, pdPage, pageNumber, cleanRulings, stripper.getTextPositionSequences(),
+
+                                                                                    false);
 
             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) {
                 case REDACT_MANAGER_OLD ->
-                        redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings, classificationDocument.getLayoutDebugLayer());
+                        redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings, classificationDocument.getVisualizations());
                 case DOCUMINE_OLD -> docuMineBlockificationService.blockify(words, cleanRulings);
                 case DOCUMINE, REDACT_MANAGER, REDACT_MANAGER_PARAGRAPH_DEBUG, REDACT_MANAGER_WITHOUT_DUPLICATE_PARAGRAPH ->
-                        docstrumBlockificationService.blockify(words, cleanRulings, true, classificationDocument.getLayoutDebugLayer(), layoutParsingType);
+                        docstrumBlockificationService.blockify(words, cleanRulings, true, classificationDocument.getVisualizations(), layoutParsingType);
                 case CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG ->
-                        docstrumBlockificationService.blockify(words, cleanRulings, false, classificationDocument.getLayoutDebugLayer(), layoutParsingType);
+                        docstrumBlockificationService.blockify(words, cleanRulings, false, classificationDocument.getVisualizations(), layoutParsingType);
+                default -> throw new IllegalArgumentException("Unexpected LayoutParsingType: " + layoutParsingType);
             };
 
             classificationPage.setCleanRulings(cleanRulings);
@@ -324,22 +337,21 @@ public class LayoutParsingPipeline {
             classificationPage.setPageWidth(cropbox.getWidth());
             classificationPage.setPageHeight(cropbox.getHeight());
 
-            if (layoutParsingType != LayoutParsingType.REDACT_MANAGER_OLD) {
-                List<OutlineObject> outlineObjects = classificationDocument.getOutlineObjectTree().getOutlineObjectsPerPage().getOrDefault(pageNumber, new ArrayList<>());
+            if (layoutParsingType != LayoutParsingType.REDACT_MANAGER_OLD && layoutParsingType != LayoutParsingType.DOCUMINE_OLD) {
+                List<OutlineObject> outlineObjects = classificationDocument.getOutlineObjectTree().getOutlineObjectsPerPage().getOrDefault(pageNumber - 1, new ArrayList<>());
 
                 OutlineObject notFoundOutlineObject = null;
                 if (lastProcessedOutlineObject != null && !lastProcessedOutlineObject.isFound()) {
-                    lastProcessedOutlineObject.resetPoint();
+                    lastProcessedOutlineObject.setPoint(new Point2D.Float(0, cropbox.getHeight()));
                     notFoundOutlineObject = lastProcessedOutlineObject;
                 }
                 if (!outlineObjects.isEmpty()) {
                     classificationPage.setOutlineObjects(outlineObjects);
                     lastProcessedOutlineObject = blockificationPostprocessingService.sanitizeOutlineBlocks(classificationPage, notFoundOutlineObject);
                 }
-                classificationDocument.getLayoutDebugLayer().addOutlineObjects(outlineObjects, pageInformation);
             }
 
-            classificationDocument.getLayoutDebugLayer().addMarkedContentVisualizations(stripper.getMarkedContents(), pageNumber);
+            classificationDocument.getVisualizations().addMarkedContentVisualizations(stripper.getMarkedContents(), pageNumber);
             // MarkedContent needs to be converted at this point, otherwise it leads to GC Problems in Pdfbox.
             classificationPage.setMarkedContentBboxPerType(convertMarkedContents(stripper.getMarkedContents()));
 
@@ -370,7 +382,7 @@ public class LayoutParsingPipeline {
         log.info("Calculating BodyTextFrame for {}", identifier);
         bodyTextFrameService.setBodyTextFrames(classificationDocument, layoutParsingType);
         for (ClassificationPage page : classificationDocument.getPages()) {
-            classificationDocument.getLayoutDebugLayer().addCleanRulingVisualization(page.getCleanRulings(), page.getPageNumber());
+            classificationDocument.getVisualizations().addCleanRulingVisualization(page.getCleanRulings(), page.getPageNumber());
         }
         log.info("Classify TextBlocks for {}", identifier);
         switch (layoutParsingType) {
@@ -380,12 +392,6 @@ public class LayoutParsingPipeline {
             case CLARIFYND -> clarifyndClassificationService.classifyDocument(classificationDocument);
         }
 
-        if (layoutParsingType.equals(LayoutParsingType.DOCUMINE_OLD)) {
-            for (ClassificationPage page : classificationDocument.getPages()) {
-                docuMineBlockificationService.mergeblocks(page, page.getCleanRulings().withoutTextRulings(), 0, 10);
-            }
-        }
-
         List<TextPageBlock> headlines = classificationDocument.getPages()
                 .stream()
                 .flatMap(classificationPage -> classificationPage.getTextBlocks()

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

@@ -5,7 +5,7 @@ import org.springframework.context.annotation.Bean;
 import org.springframework.context.annotation.ComponentScan;
 import org.springframework.context.annotation.Configuration;
 
-import com.knecon.fforesight.service.viewerdoc.service.PDFTronViewerDocumentService;
+import com.knecon.fforesight.service.viewerdoc.service.ViewerDocumentService;
 
 import io.micrometer.observation.ObservationRegistry;
 
@@ -13,11 +13,12 @@ import io.micrometer.observation.ObservationRegistry;
 @ComponentScan
 public class LayoutParsingServiceProcessorConfiguration {
 
+
     @Bean
     @Autowired
-    public PDFTronViewerDocumentService viewerDocumentService(ObservationRegistry registry) {
+    public ViewerDocumentService viewerDocumentService(ObservationRegistry registry) {
 
-        return new PDFTronViewerDocumentService(registry);
+        return new ViewerDocumentService(registry);
     }
 
 }

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

@@ -1,11 +1,9 @@
 package com.knecon.fforesight.service.layoutparser.processor;
 
-import java.io.ByteArrayInputStream;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
-import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.nio.file.Paths;
@@ -23,7 +21,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsi
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
-import com.knecon.fforesight.service.viewerdoc.service.ViewerDocVersioningUtility;
 import com.knecon.fforesight.tenantcommons.TenantContext;
 
 import io.micrometer.observation.annotation.Observed;
@@ -39,7 +36,6 @@ public class LayoutParsingStorageService {
     private final StorageService storageService;
     private final ObjectMapper objectMapper;
 
-
     @Observed(name = "LayoutParsingStorageService", contextualName = "get-origin-file")
     public File getOriginFile(String storageId) throws IOException {
 
@@ -57,18 +53,11 @@ public class LayoutParsingStorageService {
         }
         File tempFile = createTempFile("viewerDocument", ".pdf");
         storageService.downloadTo(TenantContext.getTenantId(), storageId, tempFile);
-
-        if (!ViewerDocVersioningUtility.isCurrentVersion(tempFile)) {
-            assert tempFile.delete();
-            return Optional.empty();
-        }
-
         return Optional.of(tempFile);
     }
 
 
-    @SneakyThrows
-    public ImageServiceResponse getImagesFile(String storageId) {
+    public ImageServiceResponse getImagesFile(String storageId) throws IOException {
 
         try (InputStream inputStream = getObject(storageId)) {
 
@@ -79,8 +68,7 @@ public class LayoutParsingStorageService {
     }
 
 
-    @SneakyThrows
-    public TableServiceResponse getTablesFile(String storageId) {
+    public TableServiceResponse getTablesFile(String storageId) throws IOException {
 
         try (var tableClassificationStream = getObject(storageId)) {
 
@@ -90,12 +78,11 @@ public class LayoutParsingStorageService {
         }
     }
 
-
-@SneakyThrows
-    public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) {
+    public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) throws IOException {
 
         try (InputStream inputStream = getObject(storageId)) {
-            return objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
+            VisualLayoutParsingResponse visualLayoutParsingResponse = objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
+            return visualLayoutParsingResponse;
         }
     }
 
@@ -115,6 +102,11 @@ public class LayoutParsingStorageService {
         storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.researchDocumentStorageId(), researchDocumentData);
     }
 
+    public void storeObject(String storageId, InputStream in) {
+
+        storageService.storeObject(TenantContext.getTenantId(), storageId, in);
+    }
+
 
     private File createTempFile(String filenamePrefix, String filenameSuffix) throws IOException {
 
@@ -167,16 +159,4 @@ public class LayoutParsingStorageService {
         }
     }
 
-
-    @SneakyThrows
-    @Observed(name = "LayoutParsingStorageService", contextualName = "store-markdown-file")
-    public void storeMarkdownFile(String markdownFileStorageId, String markdownContent) {
-
-        try (InputStream inputStream = new ByteArrayInputStream(markdownContent.getBytes(StandardCharsets.UTF_8))) {
-
-            storageService.storeObject(TenantContext.getTenantId(), markdownFileStorageId, inputStream);
-        }
-
-    }
-
 }

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

@@ -17,5 +17,4 @@ public class LayoutparserSettings {
 
     boolean debug;
     LayoutParsingType layoutParsingTypeOverride;
-    String pdftronLicense;
 }

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

@@ -0,0 +1,74 @@
+package com.knecon.fforesight.service.layoutparser.processor;
+
+import java.nio.charset.StandardCharsets;
+import java.util.List;
+
+import org.commonmark.Extension;
+import org.commonmark.ext.gfm.tables.TablesExtension;
+import org.commonmark.node.Document;
+import org.commonmark.node.Node;
+import org.commonmark.parser.Parser;
+import org.commonmark.renderer.Renderer;
+import org.commonmark.renderer.markdown.MarkdownRenderer;
+import org.springframework.stereotype.Service;
+
+import com.iqser.red.storage.commons.service.StorageService;
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingFinishedEvent;
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
+import com.knecon.fforesight.service.layoutparser.processor.markdown.MarkdownChunker;
+import com.knecon.fforesight.tenantcommons.TenantContext;
+
+import io.micrometer.observation.annotation.Observed;
+import lombok.AccessLevel;
+import lombok.RequiredArgsConstructor;
+import lombok.SneakyThrows;
+import lombok.experimental.FieldDefaults;
+
+@Service
+@RequiredArgsConstructor
+@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
+public class MarkdownParsingPipeline {
+
+    StorageService storageService;
+
+
+    @SneakyThrows
+    @Observed(name = "MarkdownParsingPipeline", contextualName = "parse-markdown")
+    public LayoutParsingFinishedEvent parseMarkdownAndSaveToStorage(LayoutParsingRequest request) {
+
+        long start = System.currentTimeMillis();
+        String markdown;
+        try (var in = storageService.getObject(TenantContext.getTenantId(), request.originFileStorageId()).getInputStream()) {
+            markdown = new String(in.readAllBytes(), StandardCharsets.UTF_8);
+        }
+        Parser parser = buildParser();
+        Node node = parser.parse(markdown);
+
+        MarkdownChunker chunker = new MarkdownChunker(600);
+
+        node.accept(chunker);
+
+        Renderer renderer = buildRenderer();
+        List<Document> markdownChunks = chunker.getResult();
+        for (Document markdownChunk : markdownChunks) {
+
+        }
+
+        return LayoutParsingFinishedEvent.builder().identifier(request.identifier()).numberOfPages(1).duration(System.currentTimeMillis() - start).build();
+    }
+
+
+    public static Parser buildParser() {
+
+        List<Extension> extensions = List.of(TablesExtension.create());
+        return Parser.builder().extensions(extensions).build();
+    }
+
+
+    public static MarkdownRenderer buildRenderer() {
+
+        List<Extension> extensions = List.of(TablesExtension.create());
+        return MarkdownRenderer.builder().extensions(extensions).build();
+    }
+
+}

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.docstrum;
 
 import java.util.ArrayList;
-import java.util.EnumMap;
 import java.util.List;
 import java.util.stream.Collectors;
 
@@ -19,7 +18,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRul
 import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
-import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutparsingVisualizations;
 
 import lombok.RequiredArgsConstructor;
 
@@ -27,7 +26,6 @@ import lombok.RequiredArgsConstructor;
 @RequiredArgsConstructor
 public class DocstrumSegmentationService {
 
-    public static final double SAME_DIRECTION_THRESHOLD = 0.9;
     private final NearestNeighbourService nearestNeighbourService;
     private final SpacingService spacingService;
     private final LineBuilderService lineBuilderService;
@@ -35,50 +33,19 @@ public class DocstrumSegmentationService {
     private final ReadingOrderService readingOrderService;
 
 
-    public List<Zone> segmentPage(List<TextPositionSequence> textPositions, boolean xyOrder, CleanRulings usedRulings, LayoutDebugLayer visualizations) {
+    public List<Zone> segmentPage(List<TextPositionSequence> textPositions, boolean xyOrder, CleanRulings usedRulings, LayoutparsingVisualizations visualizations) {
 
-        EnumMap<TextDirection, Integer> directionCounts = new EnumMap<>(TextDirection.class);
+        List<Zone> zones = new ArrayList<>();
+        zones.addAll(computeZones(textPositions, usedRulings, visualizations, TextDirection.ZERO));
+        zones.addAll(computeZones(textPositions, usedRulings, visualizations, TextDirection.QUARTER_CIRCLE));
+        zones.addAll(computeZones(textPositions, usedRulings, visualizations, TextDirection.HALF_CIRCLE));
+        zones.addAll(computeZones(textPositions, usedRulings, visualizations, TextDirection.THREE_QUARTER_CIRCLE));
 
-        List<Zone> newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.ZERO);
-        directionCounts.put(TextDirection.ZERO, newZones.size());
-        List<Zone> zones = new ArrayList<>(newZones);
-
-        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.QUARTER_CIRCLE);
-        directionCounts.put(TextDirection.QUARTER_CIRCLE, newZones.size());
-        zones.addAll(newZones);
-
-        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.HALF_CIRCLE);
-        directionCounts.put(TextDirection.HALF_CIRCLE, newZones.size());
-        zones.addAll(newZones);
-
-        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.THREE_QUARTER_CIRCLE);
-        directionCounts.put(TextDirection.THREE_QUARTER_CIRCLE, newZones.size());
-        zones.addAll(newZones);
-
-        return readingOrderService.resolve(zones, xyOrder, mostSameDirection(directionCounts));
+        return readingOrderService.resolve(zones, xyOrder);
     }
 
 
-    private boolean mostSameDirection(EnumMap<TextDirection, Integer> directionCounts) {
-
-        int total = directionCounts.values()
-                .stream()
-                .mapToInt(i -> i).sum();
-
-        if ((double) directionCounts.get(TextDirection.ZERO) / total > SAME_DIRECTION_THRESHOLD) {
-            return true;
-        } else if ((double) directionCounts.get(TextDirection.QUARTER_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
-            return true;
-        } else if ((double) directionCounts.get(TextDirection.HALF_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
-            return true;
-        } else if ((double) directionCounts.get(TextDirection.THREE_QUARTER_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
-            return true;
-        }
-        return false;
-    }
-
-
-    private List<Zone> computeZones(List<TextPositionSequence> textPositions, CleanRulings rulings, LayoutDebugLayer visualizations, TextDirection direction) {
+    private List<Zone> computeZones(List<TextPositionSequence> textPositions, CleanRulings rulings, LayoutparsingVisualizations visualizations, TextDirection direction) {
 
         List<RedTextPosition> positions = textPositions.stream()
                 .filter(t -> t.getDir() == direction)

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/AngleFilter.java

@@ -15,16 +15,10 @@ public class AngleFilter {
 
     public boolean matches(Neighbor neighbor) {
 
-        return matches(neighbor.getAngle());
-    }
-
-
-    public boolean matches(double angle) {
-
         if (lowerAngle <= upperAngle) {
-            return lowerAngle <= angle && angle < upperAngle;
+            return lowerAngle <= neighbor.getAngle() && neighbor.getAngle() < upperAngle;
         } else {
-            return lowerAngle <= angle || angle < upperAngle;
+            return lowerAngle <= neighbor.getAngle() || neighbor.getAngle() < upperAngle;
         }
     }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/BoundingBox.java

@@ -7,12 +7,8 @@ import java.util.List;
 import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.SuperBuilder;
 
 @Data
-@SuperBuilder
-@NoArgsConstructor
 public abstract class BoundingBox {
 
     // Java coordinate system: (0, 0) is always upper left, x is increasing left to right and y is increasing from top to bottom.
@@ -23,7 +19,7 @@ public abstract class BoundingBox {
     // This rotates completely in 90 degree steps with page rotation.
     // Needs to be used when writing to a PDF.
     // Also, these are definitely correct and should be used whenever possible.
-    protected Rectangle2D bBoxPdf;
+    protected Rectangle2D bBoxInitialUserSpace;
 
     protected static final float VERTICAL_COMPARISON_THRESHOLD = 0.4f;
 
@@ -54,25 +50,25 @@ public abstract class BoundingBox {
 
     public double getPdfMinX() {
 
-        return bBoxPdf.getMinX();
+        return bBoxInitialUserSpace.getMinX();
     }
 
 
     public double getPdfMaxX() {
 
-        return bBoxPdf.getMaxX();
+        return bBoxInitialUserSpace.getMaxX();
     }
 
 
     public double getPdfMinY() {
 
-        return bBoxPdf.getMinY();
+        return bBoxInitialUserSpace.getMinY();
     }
 
 
     public double getPdfMaxY() {
 
-        return bBoxPdf.getMaxY();
+        return bBoxInitialUserSpace.getMaxY();
     }
 
 
@@ -133,31 +129,13 @@ public abstract class BoundingBox {
     }
 
 
-    public boolean intersectsX(BoundingBox other, float threshold) {
-
-        return this.getX() - threshold <= other.getMaxX() && this.getMaxX() + threshold >= other.getX();
-    }
-
-
-    public boolean intersectsPdf(BoundingBox other) {
-
-        return this.intersectsXPdf(other) && this.intersectsYPdf(other);
-    }
-
-
-    public boolean intersectsPdf(BoundingBox other, float yThreshold, float xThreshold) {
-
-        return this.intersectsXPdf(other, xThreshold) && this.intersectsYPdf(other, yThreshold);
-    }
-
-
-    public boolean intersectsYPdf(BoundingBox other) {
+    public boolean intersectsY(BoundingBox other) {
 
         return this.getPdfMinY() <= other.getPdfMaxY() && this.getPdfMaxY() >= other.getPdfMinY();
     }
 
 
-    public boolean intersectsY(BoundingBox other) {
+    public boolean intersectsYJava(BoundingBox other) {
 
         return this.getY() <= other.getMaxY() && this.getMaxY() >= other.getY();
     }
@@ -165,31 +143,25 @@ public abstract class BoundingBox {
 
     public boolean intersectsY(BoundingBox other, float threshold) {
 
-        return this.getY() - threshold <= other.getMaxY() && this.getMaxY() + threshold >= other.getY();
-    }
-
-
-    public boolean intersectsYPdf(BoundingBox other, float threshold) {
-
         return this.getPdfMinY() - threshold <= other.getPdfMaxY() && this.getPdfMaxY() + threshold >= other.getPdfMinY();
     }
 
 
-    public boolean intersectsXPdf(BoundingBox other) {
-
-        return this.getPdfMinX() <= other.getPdfMaxX() && this.getPdfMaxX() >= other.getPdfMinX();
-    }
-
-
     public boolean intersectsX(BoundingBox other) {
 
+        return this.getPdfMinX() <= other.getPdfMaxX() && this.getPdfMaxX() >= other.getPdfMinX();
+    }
+
+
+    public boolean intersectsXJava(BoundingBox other) {
+
         return this.getX() <= other.getMaxX() && this.getMaxX() >= other.getMinX();
     }
 
 
-    public boolean intersectsXPdf(BoundingBox other, float threshold) {
+    public boolean intersectsX(BoundingBox other, float threshold) {
 
-        return this.getPdfMinX() - threshold <= other.getPdfMaxX() && this.getMaxX() + threshold >= other.getPdfMinX();
+        return this.getPdfMinX() - threshold <= other.getMaxX() && this.getMaxX() + threshold >= other.getPdfMinX();
     }
 
 
@@ -198,8 +170,8 @@ public abstract class BoundingBox {
         this.bBox = components.stream()
                 .map(BoundingBox::getBBox)
                 .collect(RectangleTransformations.collectBBox());
-        this.bBoxPdf = components.stream()
-                .map(BoundingBox::getBBoxPdf)
+        this.bBoxInitialUserSpace = components.stream()
+                .map(BoundingBox::getBBoxInitialUserSpace)
                 .collect(RectangleTransformations.collectBBox());
     }
 
@@ -257,25 +229,25 @@ public abstract class BoundingBox {
 
     public boolean rightOf(BoundingBox other) {
 
-        return this.intersectsY(other) && other.getMaxX() <= this.getMinX();
+        return this.intersectsYJava(other) && other.getMaxX() <= this.getMinX();
     }
 
 
     public boolean leftOf(BoundingBox other) {
 
-        return this.intersectsY(other) && other.getMinX() >= this.getMaxX();
+        return this.intersectsYJava(other) && other.getMinX() >= this.getMaxX();
     }
 
 
     public boolean isAbove(BoundingBox other) {
 
-        return this.intersectsX(other) && other.getMinY() >= this.getMaxY();
+        return this.intersectsXJava(other) && other.getMinY() >= this.getMaxY();
     }
 
 
     public boolean isBelow(BoundingBox other) {
 
-        return this.intersectsX(other) && this.getMinY() >= other.getMaxY();
+        return this.intersectsXJava(other) && this.getMinY() >= other.getMaxY();
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Character.java

@@ -35,7 +35,7 @@ public class Character {
 
     public double getHeight() {
 
-        return textPosition.getHeightDirAdj();
+        return textPosition.getHeightDir();
     }
 
 
@@ -65,9 +65,9 @@ public class Character {
         double s = Math.sin(-0);
         double c = Math.cos(-0);
         xs[0] = c * x - s * y;
-        xs[1] = c * (x + textPosition.getWidthDirAdj()) - s * (y + textPosition.getHeightDirAdj());
+        xs[1] = c * (x + textPosition.getWidthDirAdj()) - s * (y + textPosition.getHeightDir());
         xs[2] = c * other.x - s * other.y;
-        xs[3] = c * (other.x + other.textPosition.getWidthDirAdj()) - s * (other.y + other.textPosition.getHeightDirAdj());
+        xs[3] = c * (other.x + other.textPosition.getWidthDirAdj()) - s * (other.y + other.textPosition.getHeightDir());
         boolean overlapping = xs[1] >= xs[2] && xs[3] >= xs[0];
         Arrays.sort(xs);
         return Math.abs(xs[2] - xs[1]) * (overlapping ? 1 : -1);

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Line.java

@@ -1,28 +1,18 @@
 package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
 
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence.BOLD;
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence.BOLD_ITALIC;
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence.ITALIC;
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence.STANDARD;
-
 import java.util.ArrayList;
 import java.util.Arrays;
-import java.util.Comparator;
-import java.util.EnumMap;
 import java.util.List;
-import java.util.Map;
-import java.util.concurrent.atomic.AtomicInteger;
 
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.FastAtan2;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.FontStyle;
 
 import lombok.Data;
 import lombok.EqualsAndHashCode;
 
 @Data
-@EqualsAndHashCode(onlyExplicitlyIncluded = true, callSuper = false)
-public class Line extends TextBoundingBox {
+@EqualsAndHashCode(onlyExplicitlyIncluded = true)
+public class Line extends BoundingBox {
 
     private static final double WORD_DISTANCE_MULTIPLIER = 0.18;
 
@@ -38,8 +28,6 @@ public class Line extends TextBoundingBox {
 
     private final double height;
 
-    private FontStyle fontStyle;
-
     private final List<Character> characters;
     private final List<TextPositionSequence> words = new ArrayList<>();
 
@@ -79,29 +67,6 @@ public class Line extends TextBoundingBox {
         height = computeHeight();
         computeWords(wordSpacing * WORD_DISTANCE_MULTIPLIER);
         buildBBox();
-        computeFontStyle();
-    }
-
-
-    private void computeFontStyle() {
-
-        EnumMap<FontStyle, AtomicInteger> fontStyleCounter = new EnumMap<>(FontStyle.class);
-        for (FontStyle fontStyle : FontStyle.values()) {
-            fontStyleCounter.put(fontStyle, new AtomicInteger(0));
-        }
-        for (TextPositionSequence word : words) {
-            switch (word.getFontStyle()) {
-                case STANDARD -> fontStyleCounter.get(FontStyle.REGULAR).getAndIncrement();
-                case BOLD -> fontStyleCounter.get(FontStyle.BOLD).getAndIncrement();
-                case ITALIC -> fontStyleCounter.get(FontStyle.ITALIC).getAndIncrement();
-                case BOLD_ITALIC -> fontStyleCounter.get(FontStyle.BOLD_ITALIC).getAndIncrement();
-            }
-        }
-        fontStyle = fontStyleCounter.entrySet()
-                .stream()
-                .max(Comparator.comparing(entry -> entry.getValue().get()))
-                .map(Map.Entry::getKey)
-                .orElse(FontStyle.REGULAR);
     }
 
 
@@ -179,8 +144,8 @@ public class Line extends TextBoundingBox {
     private void buildBBox() {
 
         this.setToBBoxOfComponents(characters.stream()
-                                           .map(Character::getTextPosition)
-                                           .toList());
+                                       .map(Character::getTextPosition)
+                                       .toList());
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/TextBoundingBox.java

@@ -1,102 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
-
-import java.awt.geom.Rectangle2D;
-import java.util.List;
-import java.util.Set;
-import java.util.stream.Collectors;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
-import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
-
-import lombok.EqualsAndHashCode;
-import lombok.Getter;
-import lombok.NoArgsConstructor;
-import lombok.Setter;
-import lombok.experimental.SuperBuilder;
-
-@Getter
-@Setter
-@SuperBuilder
-@NoArgsConstructor
-@EqualsAndHashCode(callSuper = false)
-public abstract class TextBoundingBox extends BoundingBox {
-
-    protected Rectangle2D bBoxDirAdj;
-
-    protected TextDirection dir;
-
-
-    @Override
-    public void setToBBoxOfComponents(List<? extends BoundingBox> components) {
-
-        super.setToBBoxOfComponents(components);
-        this.bBoxDirAdj = components.stream()
-                .filter(c -> c instanceof TextBoundingBox)
-                .map(c -> (TextBoundingBox) c)
-                .map(TextBoundingBox::getBBoxDirAdj)
-                .collect(RectangleTransformations.collectBBox());
-
-        Set<TextDirection> textDirections = components.stream()
-                .filter(c -> c instanceof TextBoundingBox)
-                .map(c -> (TextBoundingBox) c)
-                .map(TextBoundingBox::getDir)
-                .collect(Collectors.toSet());
-
-        if (textDirections.isEmpty()) {
-            dir = TextDirection.ZERO;
-        } else if (textDirections.size() > 1) {
-            throw new IllegalArgumentException("More than one text direction found");
-        } else {
-            dir = textDirections.iterator().next();
-        }
-    }
-
-
-    public double getXDirAdj() {
-
-        return this.bBoxDirAdj.getX();
-    }
-
-
-    public double getYDirAdj() {
-
-        return this.bBoxDirAdj.getY();
-    }
-
-
-    public double getWidthDirAdj() {
-
-        return this.bBoxDirAdj.getWidth();
-    }
-
-
-    public double getHeightDirAdj() {
-
-        return this.bBoxDirAdj.getHeight();
-    }
-
-
-    public double getMaxXDirAdj() {
-
-        return this.bBoxDirAdj.getMaxX();
-    }
-
-
-    public double getMaxYDirAdj() {
-
-        return this.bBoxDirAdj.getMaxY();
-    }
-
-
-    public double getCenterYDirAdj() {
-
-        return this.bBoxDirAdj.getCenterY();
-    }
-
-
-    public double getCenterXDirAdj() {
-
-        return this.bBoxDirAdj.getCenterX();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Zone.java

@@ -6,11 +6,9 @@ import java.util.List;
 import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.Data;
-import lombok.EqualsAndHashCode;
 
 @Data
-@EqualsAndHashCode(callSuper = false)
-public class Zone extends TextBoundingBox {
+public class Zone extends BoundingBox {
 
     private List<Line> lines;
 

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

@@ -1,17 +1,15 @@
 package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
 
-import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
 import java.util.Comparator;
 import java.util.HashMap;
 import java.util.List;
 import java.util.ListIterator;
 import java.util.Map;
-import java.util.stream.Collectors;
 
 import org.springframework.stereotype.Service;
 
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.TextBoundingBox;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.DoubleUtils;
 
@@ -21,30 +19,21 @@ public class ReadingOrderService {
     private static final double THRESHOLD = 5;
     public static final double MULTI_COLUMN_DETECTION_THRESHOLD = 1.5;
 
-    private static final Comparator<TextBoundingBox> COMPARATOR = //
-            Comparator.comparing(TextBoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
-                    .thenComparing(TextBoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD));
 
-    private static final Comparator<TextBoundingBox> COMPARATOR_DIR_ADJ = //
-            Comparator.comparing(TextBoundingBox::getYDirAdj, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
-                    .thenComparing(TextBoundingBox::getXDirAdj, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD));
-
-
-    public List<Zone> resolve(List<Zone> zones, boolean xyReadingOrder, boolean useDirAdjCoords) {
+    public List<Zone> resolve(List<Zone> zones, boolean xyReadingOrder) {
 
         if (zones.isEmpty() || zones.size() == 1) {
             return zones;
         }
 
         if (xyReadingOrder) {
-            return resolveSingleColumnReadingOrder(zones, useDirAdjCoords);
+            return resolveSingleColumnReadingOrder(zones);
         }
 
         Map<Long, Integer> histogram = new HashMap<>();
         for (Zone zone : zones) {
-            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
-            long minY = Math.round(bbox.getMinY());
-            long maxY = Math.round(bbox.getMaxY());
+            long minY = Math.round(zone.getBBox().getMinY());
+            long maxY = Math.round(zone.getBBox().getMaxY());
             for (long i = minY; i <= maxY; i++) {
                 histogram.put(i, histogram.getOrDefault(i, 0) + 1);
             }
@@ -54,32 +43,24 @@ public class ReadingOrderService {
                     .stream()
                     .mapToInt(Integer::intValue).average()
                     .orElse(1) < MULTI_COLUMN_DETECTION_THRESHOLD) {
-            return resolveSingleColumnReadingOrder(zones, useDirAdjCoords);
+            return resolveSingleColumnReadingOrder(zones);
         } else {
 
-            return resolveMultiColumnReadingOder(zones, useDirAdjCoords);
+            return resolveMultiColumnReadingOder(zones);
         }
 
     }
 
 
-    private static List<Zone> resolveSingleColumnReadingOrder(List<Zone> zones, boolean useDirAdjCoords) {
+    private static List<Zone> resolveSingleColumnReadingOrder(List<Zone> zones) {
 
-        if (useDirAdjCoords) {
-            return zones.stream()
-                    .collect(Collectors.groupingBy(TextBoundingBox::getDir)).values()
-                    .stream()
-                    .flatMap(words -> words.stream()
-                            .sorted(COMPARATOR_DIR_ADJ))
-                    .toList();
-        }
-
-        zones.sort(COMPARATOR);
+        zones.sort(Comparator.comparing(BoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                           .thenComparing(BoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
         return zones;
     }
 
 
-    private List<Zone> resolveMultiColumnReadingOder(List<Zone> zones, boolean useDirAdjCoords) {
+    private List<Zone> resolveMultiColumnReadingOder(List<Zone> zones) {
 
         // Simple reading order resolver for multi column page layout as described here : https://pub.towardsai.net/advanced-rag-02-unveiling-pdf-parsing-b84ae866344e
         // TODO implement a more fancy reading order resolver see https://github.com/BobLd/DocumentLayoutAnalysis/blob/master/README.md#reading-order
@@ -88,12 +69,11 @@ public class ReadingOrderService {
         double maxX = Double.NEGATIVE_INFINITY;
 
         for (Zone zone : zones) {
-            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
-            if (bbox.getX() < minX) {
-                minX = zone.getXDirAdj();
+            if (zone.getX() < minX) {
+                minX = zone.getX();
             }
-            if (bbox.getMaxX() > maxX) {
-                maxX = zone.getMaxXDirAdj();
+            if (zone.getX() + zone.getWidth() > maxX) {
+                maxX = zone.getX() + zone.getWidth();
             }
         }
 
@@ -102,27 +82,24 @@ public class ReadingOrderService {
         List<Zone> leftOf = new ArrayList<>();
         List<Zone> rightOf = new ArrayList<>();
         List<Zone> middle = new ArrayList<>();
-
         for (Zone zone : zones) {
-            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
-            if (bbox.getX() < midLineXCoordinate && bbox.getX() + bbox.getWidth() < midLineXCoordinate) {
+            if (zone.getX() < midLineXCoordinate && zone.getX() + zone.getWidth() < midLineXCoordinate) {
                 leftOf.add(zone);
-            } else if (bbox.getX() > midLineXCoordinate && bbox.getX() + bbox.getWidth() > midLineXCoordinate) {
+            } else if (zone.getX() > midLineXCoordinate && zone.getX() + zone.getWidth() > midLineXCoordinate) {
                 rightOf.add(zone);
             } else {
                 middle.add(zone);
             }
         }
 
-        if (useDirAdjCoords) {
-            leftOf.sort(COMPARATOR_DIR_ADJ);
-            rightOf.sort(COMPARATOR_DIR_ADJ);
-            middle.sort(COMPARATOR_DIR_ADJ);
-        } else {
-            leftOf.sort(COMPARATOR);
-            rightOf.sort(COMPARATOR);
-            middle.sort(COMPARATOR);
-        }
+        leftOf.sort(Comparator.comparing(BoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                            .thenComparing(BoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
+
+        rightOf.sort(Comparator.comparing(BoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                             .thenComparing(BoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
+
+        middle.sort(Comparator.comparing(BoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                            .thenComparing(BoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
 /*
         List<Zone> leftNotIntersecting = new ArrayList<>();
         for (Zone leftZone : leftOf) {
@@ -174,9 +151,8 @@ public class ReadingOrderService {
 
         while (itty.hasNext()) {
             Zone current = itty.next();
-            Rectangle2D bbox = useDirAdjCoords ? current.getBBoxDirAdj() : current.getBBox();
             for (int i = 0; i < sortedZones.size(); i++) {
-                if (bbox.getY() < sortedZones.get(i).getY()) {
+                if (current.getY() < sortedZones.get(i).getY()) {
                     sortedZones.add(i, current);
                     itty.remove();
                     break;

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

@@ -1,7 +1,5 @@
 package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
 
-import static com.knecon.fforesight.service.layoutparser.processor.model.SectionIdentifier.numericalIdentifierPattern;
-
 import java.util.ArrayList;
 import java.util.Comparator;
 import java.util.HashSet;
@@ -23,7 +21,7 @@ public class ZoneBuilderService {
     private static final double MIN_HORIZONTAL_DISTANCE_MULTIPLIER = -0.5;
     private static final double MAX_VERTICAL_DISTANCE_MULTIPLIER = 1.2;
 
-    private static final double MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER = -7;
+    private static final double MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER = -3.0;
 
     private static final double MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER = 0.5;
 
@@ -40,7 +38,7 @@ public class ZoneBuilderService {
 
         double minHorizontalDistance = characterSpacing * MIN_HORIZONTAL_DISTANCE_MULTIPLIER;
         double maxVerticalDistance = lineSpacing * MAX_VERTICAL_DISTANCE_MULTIPLIER;
-        double minHorizontalMergeDistance = lineSpacing * MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER;
+        double minHorizontalMergeDistance = characterSpacing * MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER;
         double maxVerticalMergeDistance = lineSpacing * MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER;
 
         UnionFind<Line> unionFind = new UnionFind<>(new HashSet<>(lines));
@@ -56,26 +54,11 @@ public class ZoneBuilderService {
                     return;
                 }
 
-//                if (!innerLine.getFontStyle().equals(outerLine.getFontStyle()) //
-//                    && !outerLine.intersectsY(innerLine, -2f)) {
-//                    return;
-//                }
+                double scale = Math.min(outerLine.getHeight(), innerLine.getHeight()) / meanHeight;
+                scale = Math.max(MIN_LINE_SIZE_SCALE, Math.min(scale, MAX_LINE_SIZE_SCALE));
 
-                double horizontalScale = Math.min(outerLine.getHeightDirAdj(), innerLine.getHeightDirAdj()) / meanHeight;
-                horizontalScale = Math.max(MIN_LINE_SIZE_SCALE, Math.min(horizontalScale, MAX_LINE_SIZE_SCALE));
-                double verticalScale = horizontalScale;
-
-//                if (innerLine.toString().endsWith(":")
-//                    || outerLine.toString().endsWith(":")
-//                    || numericalIdentifierPattern.matcher(innerLine.toString()).matches()
-//                    || numericalIdentifierPattern.matcher(outerLine.toString()).matches()) {
-//
-//                    horizontalScale *= 5;
-//                    verticalScale /= 10;
-//                }
-
-                double horizontalDistance = outerLine.horizontalDistance(innerLine) / horizontalScale;
-                double verticalDistance = outerLine.verticalDistance(innerLine) / verticalScale;
+                double horizontalDistance = outerLine.horizontalDistance(innerLine) / scale;
+                double verticalDistance = outerLine.verticalDistance(innerLine) / scale;
 
                 if ((!(minHorizontalDistance <= horizontalDistance) || !(verticalDistance <= maxVerticalDistance)) //
                     && (!(minHorizontalMergeDistance <= horizontalDistance) || !(verticalDistance <= maxVerticalMergeDistance))) {
@@ -104,7 +87,7 @@ public class ZoneBuilderService {
         double weights = 0.0;
         for (Line line : lines) {
             double weight = line.getLength();
-            meanHeight += line.getHeightDirAdj() * weight;
+            meanHeight += line.getHeight() * weight;
             weights += weight;
         }
         meanHeight /= weights;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/utils/DoubleUtils.java

@@ -7,12 +7,9 @@ public class DoubleUtils {
         if (Double.isNaN(d1) || Double.isNaN(d2)) {
             return Double.compare(d1, d2);
         }
-
-        if (Math.abs(d1 - d2) < precision) {
-            return 0;
-        }
-
-        return Double.compare(d1, d2);
+        long i1 = Math.round(d1 / (precision == 0 ? 1 : precision));
+        long i2 = Math.round(d2 / (precision == 0 ? 1 : precision));
+        return Long.compare(i1, i2);
     }
 
 }

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

@@ -5,89 +5,69 @@ import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
+import java.util.Objects;
 import java.util.Set;
 import java.util.TreeMap;
+import java.util.stream.Stream;
 
-import org.commonmark.Extension;
 import org.commonmark.ext.gfm.tables.TableBlock;
 import org.commonmark.ext.gfm.tables.TableBody;
 import org.commonmark.ext.gfm.tables.TableCell;
 import org.commonmark.ext.gfm.tables.TableHead;
 import org.commonmark.ext.gfm.tables.TableRow;
-import org.commonmark.ext.gfm.tables.TablesExtension;
-import org.commonmark.node.Block;
 import org.commonmark.node.Document;
 import org.commonmark.node.Emphasis;
 import org.commonmark.node.HardLineBreak;
 import org.commonmark.node.Heading;
 import org.commonmark.node.Node;
-import org.commonmark.node.Paragraph;
+import org.commonmark.node.SoftLineBreak;
 import org.commonmark.node.StrongEmphasis;
 import org.commonmark.node.Text;
-import org.commonmark.renderer.markdown.MarkdownRenderer;
 
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.AbstractNodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.TextRange;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Headline;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.FontStyle;
 
-public class MarkdownMapper extends AbstractNodeVisitor {
+import lombok.experimental.UtilityClass;
 
-    Document markdownDocument = new Document();
+@UtilityClass
+public class DocumentDataParser {
 
+    public Document parse(Stream<SemanticNode> semanticNodes) {
 
-    public String toMarkdownContent(com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document document) {
-
-        visit(document);
-
-        return buildRenderer().render(this.markdownDocument);
+        Document document = new Document();
+        semanticNodes.map(DocumentDataParser::parseNode)
+                .filter(Objects::nonNull)
+                .forEach(document::appendChild);
+        return document;
     }
 
 
-    @Override
-    public void visit(Headline headline) {
+    private Node parseNode(SemanticNode semanticNode) {
 
-        markdownDocument.appendChild(parseHeadline(headline));
+        return switch (semanticNode.getType()) {
+            case HEADLINE -> parseHeadline((Headline) semanticNode);
+            case PARAGRAPH -> parseParagraph((Paragraph) semanticNode);
+            case TABLE -> parseTable((Table) semanticNode);
+            default -> null;
+        };
     }
 
 
-    @Override
-    public void visit(com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph paragraph) {
-
-        markdownDocument.appendChild(parseParagraph(paragraph));
-    }
-
-
-    @Override
-    public void visit(Table table) {
-
-        markdownDocument.appendChild(parseTable(table));
-    }
-
-
-    private static MarkdownRenderer buildRenderer() {
-
-        List<Extension> extensions = List.of(TablesExtension.create());
-        return MarkdownRenderer.builder().extensions(extensions).build();
-    }
-
-
-    private Block parseTable(Table table) {
-
-//        if (table.getNumberOfRows() == 1 && table.getNumberOfCols() == 1) {
-//            org.commonmark.node.Paragraph markdownParagraph = new org.commonmark.node.Paragraph();
-//            parseTextBlock(table.getTextBlock(), true).forEach(markdownParagraph::appendChild);
-//            return markdownParagraph;
-//        }
+    private TableBlock parseTable(Table table) {
 
         TableBlock tableNode = new TableBlock();
         TableHead head = new TableHead();
         TableRow tableRow = createTableRow(table, 0);
         head.appendChild(tableRow);
         int row = 1;
+        for (; row < table.getNumberOfRows() && table.streamRow(row)
+                .allMatch(com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell::isHeader); row++) {
+            head.appendChild(createTableRow(table, row));
+        }
         tableNode.appendChild(head);
         TableBody tableBody = new TableBody();
         for (; row < table.getNumberOfRows(); row++) {
@@ -102,7 +82,7 @@ public class MarkdownMapper extends AbstractNodeVisitor {
 
         TableRow tableRow = new TableRow();
         table.streamRow(row)
-                .map(this::createTableCell)
+                .map(DocumentDataParser::createTableCell)
                 .forEach(tableRow::appendChild);
         return tableRow;
     }
@@ -111,22 +91,16 @@ public class MarkdownMapper extends AbstractNodeVisitor {
     private Node createTableCell(com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell tc) {
 
         var cell = new TableCell();
-        List<SemanticNode> childNodes = tc.streamChildren()
-                .toList();
-        if (childNodes.isEmpty()) {
-            parseTextBlock(tc.getTextBlock(), false).forEach(cell::appendChild);
-        } else {
-            childNodes.forEach(semanticNode -> parseTextBlock(semanticNode.getTextBlock(), false).forEach(cell::appendChild));
-        }
+        parseTextBlock(tc.getTextBlock()).forEach(cell::appendChild);
         return cell;
     }
 
 
-    private Paragraph parseParagraph(com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph paragraph) {
+    private org.commonmark.node.Paragraph parseParagraph(Paragraph paragraph) {
 
-        org.commonmark.node.Paragraph markdownParagraph = new org.commonmark.node.Paragraph();
-        parseTextBlock(paragraph.getTextBlock(), true).forEach(markdownParagraph::appendChild);
-        return markdownParagraph;
+        org.commonmark.node.Paragraph heading = new org.commonmark.node.Paragraph();
+        parseTextBlockWithLineBreaks(paragraph.getTextBlock()).forEach(heading::appendChild);
+        return heading;
     }
 
 
@@ -134,90 +108,86 @@ public class MarkdownMapper extends AbstractNodeVisitor {
 
         Heading heading = new Heading();
         heading.setLevel(headline.getTreeId().size());
-        heading.appendChild(parseTextBlockAsText(headline.getTextBlock()));
+        parseTextBlockWithLineBreaks(headline.getTextBlock()).forEach(heading::appendChild);
         return heading;
 
     }
 
-
-    private Text parseTextBlockAsText(TextBlock textBlock) {
-
-        return new Text(textBlock.getSearchText());
-    }
-
-
-    private List<Node> parseTextBlock(TextBlock textBlock, boolean includeLineBreaks) {
+    private List<Node> parseTextBlockWithLineBreaks(TextBlock textBlock) {
 
         LinkedList<Node> result = new LinkedList<>();
         List<TextRangeWithTextType> textRanges = mergeTextStyles(textBlock);
-
-        String fullText = getText(textBlock, textBlock.getTextRange(), includeLineBreaks);
-        List<Integer> lineTextSizes = getLineTextSizes(fullText);
-        int idx = 0;
-        int charCount = 0;
         for (TextRangeWithTextType textRange : textRanges) {
-            String text = getText(textBlock, textRange.textRange(), includeLineBreaks);
+            if (textBlock.subSequenceWithLineBreaks(textRange.textRange()).equals("\n")) {
+                result.add(new HardLineBreak());
+            }
+            String text = textBlock.subSequenceWithLineBreaks(textRange.textRange());
             String[] lines = text.split("\n");
             for (String line : lines) {
-                charCount += line.length();
+                String cleanedLine = line.trim();
+                if (cleanedLine.isEmpty()) {
+                    result.add(new HardLineBreak());
+                    continue;
+                }
                 switch (textRange.fontStyle()) {
-                    case REGULAR -> result.add(new Text(line));
+                    case REGULAR -> result.add(new Text(cleanedLine));
                     case BOLD -> {
                         StrongEmphasis boldBlock = new StrongEmphasis();
-                        boldBlock.appendChild(new Text(line));
+                        boldBlock.appendChild(new Text(cleanedLine));
                         result.add(boldBlock);
                     }
                     case ITALIC -> {
-                        Emphasis italicBlock = new Emphasis();
-                        italicBlock.appendChild(new Text(line));
+                        Emphasis italicBlock = new Emphasis("_");
+                        italicBlock.appendChild(new Text(cleanedLine));
                         result.add(italicBlock);
                     }
                     case BOLD_ITALIC -> {
-                        Emphasis italicBlock = new Emphasis();
+                        Emphasis italicBlock = new Emphasis("_");
 
                         StrongEmphasis boldBlock = new StrongEmphasis();
-                        boldBlock.appendChild(new Text(line));
+                        boldBlock.appendChild(new Text(cleanedLine));
 
                         italicBlock.appendChild(boldBlock);
                         result.add(italicBlock);
                     }
                 }
-                if (includeLineBreaks && lineTextSizes.get(idx).equals(charCount)) {
-                    result.add(new HardLineBreak());
-                    idx++;
-                }
-
+                result.add(new HardLineBreak());
             }
         }
-        if (!result.isEmpty() && result.getLast() instanceof HardLineBreak) {
-            result.removeLast();
-        }
+        result.removeLast();
         return result;
     }
 
 
-    private static List<Integer> getLineTextSizes(String input) {
+    private List<Node> parseTextBlock(TextBlock textBlock) {
 
-        String[] parts = input.split("\n");
-        List<Integer> textSizes = new ArrayList<>();
+        List<Node> result = new ArrayList<>();
+        List<TextRangeWithTextType> textRanges = mergeTextStyles(textBlock);
+        for (TextRangeWithTextType textRange : textRanges) {
+            switch (textRange.fontStyle()) {
+                case REGULAR -> result.add(new Text(textBlock.subSequence(textRange.textRange()).toString()));
+                case BOLD -> {
+                    StrongEmphasis boldBlock = new StrongEmphasis();
+                    boldBlock.appendChild(new Text(textBlock.subSequence(textRange.textRange()).toString()));
+                    result.add(boldBlock);
+                }
+                case ITALIC -> {
+                    Emphasis italicBlock = new Emphasis("_");
+                    italicBlock.appendChild(new Text(textBlock.subSequence(textRange.textRange()).toString()));
+                    result.add(italicBlock);
+                }
+                case BOLD_ITALIC -> {
+                    Emphasis italicBlock = new Emphasis("_");
 
-        int size = 0;
-        for (int i = 0; i < parts.length; i++) {
-            size += parts[i].length();
-            textSizes.add(size);
+                    StrongEmphasis boldBlock = new StrongEmphasis();
+                    boldBlock.appendChild(new Text(textBlock.subSequence(textRange.textRange()).toString()));
+
+                    italicBlock.appendChild(boldBlock);
+                    result.add(italicBlock);
+                }
+            }
         }
-
-        if (textSizes.isEmpty()) {
-            textSizes.add(0);
-        }
-
-        return textSizes;
-    }
-
-
-    private static String getText(TextBlock textBlock, TextRange textRange, boolean includeLineBreaks) {
-
-        return includeLineBreaks ? textBlock.subSequenceWithLineBreaks(textRange) : textBlock.subSequence(textRange).toString();
+        return result;
     }
 
 
@@ -280,7 +250,9 @@ public class MarkdownMapper extends AbstractNodeVisitor {
             result.add(new TextRangeWithTextType(new TextRange(start, textBlock.getTextRange().end()), style));
         }
 
-        return result;
+        return result.stream()
+                .filter(t -> t.textRange.length() > 1)
+                .toList();
     }
 
 
@@ -298,6 +270,12 @@ public class MarkdownMapper extends AbstractNodeVisitor {
     }
 
 
+    enum FontStyle {
+        REGULAR,
+        BOLD,
+        ITALIC,
+        BOLD_ITALIC;
+    }
 
     record FontStyleChange(boolean enter, FontStyle style) {
 
@@ -323,4 +301,5 @@ public class MarkdownMapper extends AbstractNodeVisitor {
     record TextRangeWithTextType(TextRange textRange, FontStyle fontStyle) {
 
     }
+
 }

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

@@ -0,0 +1,13 @@
+package com.knecon.fforesight.service.layoutparser.processor.markdown;
+
+import org.commonmark.node.IndentedCodeBlock;
+import org.commonmark.node.Paragraph;
+
+import com.knecon.fforesight.service.layoutparser.processor.utils.TokenCounter;
+
+public class ExtraTokens {
+
+    public static int INDENTED_CODE_BLOCK = 10;
+    public static int PARAGRAPH = 10;
+
+}

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

@@ -0,0 +1,527 @@
+package com.knecon.fforesight.service.layoutparser.processor.markdown;
+
+import static com.knecon.fforesight.service.layoutparser.processor.MarkdownParsingPipeline.buildRenderer;
+import static com.knecon.fforesight.service.layoutparser.processor.utils.TokenCounter.countTokens;
+
+import java.text.BreakIterator;
+import java.util.Collections;
+import java.util.Deque;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Locale;
+import java.util.stream.Collectors;
+
+import org.commonmark.ext.gfm.tables.TableBlock;
+import org.commonmark.ext.gfm.tables.TableBody;
+import org.commonmark.node.AbstractVisitor;
+import org.commonmark.node.BlockQuote;
+import org.commonmark.node.BulletList;
+import org.commonmark.node.Code;
+import org.commonmark.node.CustomBlock;
+import org.commonmark.node.CustomNode;
+import org.commonmark.node.Document;
+import org.commonmark.node.Emphasis;
+import org.commonmark.node.FencedCodeBlock;
+import org.commonmark.node.HardLineBreak;
+import org.commonmark.node.Heading;
+import org.commonmark.node.HtmlBlock;
+import org.commonmark.node.HtmlInline;
+import org.commonmark.node.Image;
+import org.commonmark.node.IndentedCodeBlock;
+import org.commonmark.node.Link;
+import org.commonmark.node.LinkReferenceDefinition;
+import org.commonmark.node.ListBlock;
+import org.commonmark.node.ListItem;
+import org.commonmark.node.Node;
+import org.commonmark.node.OrderedList;
+import org.commonmark.node.Paragraph;
+import org.commonmark.node.SoftLineBreak;
+import org.commonmark.node.StrongEmphasis;
+import org.commonmark.node.Text;
+import org.commonmark.node.ThematicBreak;
+import org.commonmark.renderer.Renderer;
+
+import com.knecon.fforesight.service.layoutparser.processor.utils.TokenCounter;
+
+import lombok.AccessLevel;
+import lombok.experimental.FieldDefaults;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@FieldDefaults(level = AccessLevel.PRIVATE)
+public class MarkdownChunker extends AbstractVisitor {
+
+    NodeCopier nodeCopier = NodeCopier.INSTANCE;
+
+    final int tokenLimit;
+    List<Document> allChunks;
+
+    Deque<Heading> currentHeadings;
+    Document chunk;
+    boolean validChunk;
+
+
+    public MarkdownChunker(int tokenLimit) {
+
+        this.tokenLimit = tokenLimit;
+        allChunks = new LinkedList<>();
+        currentHeadings = new LinkedList<>();
+        startNewChunk();
+    }
+
+
+    public List<Document> getResult() {
+
+        for (Document chunk : allChunks) {
+            if (countTokens(chunk) > tokenLimit) {
+                throwUnsplittableNodeError(chunk);
+            }
+        }
+        return allChunks;
+    }
+
+
+    @Override
+    public void visit(Heading heading) {
+
+        if (heading.getLevel() > 4) {
+            addToChunk(heading);
+        }
+        if (currentHeadings.isEmpty() || currentHeadings.peek().getLevel() < heading.getLevel()) {
+            currentHeadings.push(heading);
+        } else {
+            while (!currentHeadings.isEmpty() && currentHeadings.peek().getLevel() >= heading.getLevel()) {
+                currentHeadings.pop();
+            }
+            currentHeadings.push(heading);
+        }
+
+        startNewChunk();
+    }
+
+
+    private void startNewChunk() {
+
+        if (!validChunk && !allChunks.isEmpty()) {
+            allChunks.remove(allChunks.size() - 1);
+        }
+        validChunk = false;
+        chunk = buildNewChunk();
+        allChunks.add(chunk);
+    }
+
+
+    private Document buildNewChunk() {
+
+        Document document = new Document();
+        List<Node> headingCopies = currentHeadings.stream()
+                .map(nodeCopier::copyNodeWithChildren)
+                .collect(Collectors.toList());
+        Collections.reverse(headingCopies);
+        headingCopies.forEach(document::appendChild);
+        return document;
+    }
+
+
+    public int currentTokenCount() {
+
+        return TokenCounter.countTokens(chunk);
+    }
+
+
+    public boolean fitsTokenLimit(Node node) {
+
+        Document document = buildNewChunk();
+        document.appendChild(nodeCopier.copyNodeWithChildren(node));
+        return TokenCounter.countTokens(document) <= tokenLimit;
+    }
+
+
+    private void addToChunk(Node node) {
+
+        chunk.appendChild(node);
+
+        if (currentTokenCount() <= tokenLimit) {
+            return;
+        }
+
+        node.unlink();
+        startNewChunk();
+        chunk.appendChild(node);
+
+        if (currentTokenCount() > tokenLimit) { // node is too large and won't fit in tokenLimit, split is necessary
+            node.unlink();
+            startNewChunk();
+            splitNodeAndAddToChunk(node);
+            return;
+        }
+
+        validChunk = true;
+
+    }
+
+
+    private void splitNodeAndAddToChunk(Node node) {
+
+        if (node instanceof TableBlock tableBlock) {
+            splitTable(tableBlock);
+            return;
+        } else if (node instanceof BulletList bulletList) {
+            splitList(bulletList);
+            return;
+        } else if (node instanceof OrderedList orderedList) {
+            splitList(orderedList);
+            return;
+        } else if (node instanceof Paragraph paragraph) {
+            splitParagraph(paragraph);
+            return;
+        } else if (node instanceof IndentedCodeBlock indentedCodeBlock) {
+            splitCodeBlock(indentedCodeBlock);
+            return;
+        }
+        throwUnsplittableNodeError(node);
+    }
+
+
+    private void splitCodeBlock(IndentedCodeBlock indentedCodeBlock) {
+
+        List<IndentedCodeBlock> splitBlocks = new LinkedList<>();
+        StringBuilder sb = new StringBuilder();
+        BreakIterator lineIterator = BreakIterator.getLineInstance(Locale.ENGLISH);
+        lineIterator.setText(indentedCodeBlock.getLiteral());
+        int start = lineIterator.first();
+        for (int end = lineIterator.next(); end != BreakIterator.DONE; start = end, end = lineIterator.next()) {
+            String sentence = indentedCodeBlock.getLiteral().substring(start, end);
+            if (!fitsTokenLimit(buildIndentedCodeBlock(sb.toString()))) {
+                sb.replace(sb.length() - sentence.length(), sb.length(), "");
+                IndentedCodeBlock block = buildIndentedCodeBlock(sb.toString());
+                splitBlocks.add(block);
+                sb = new StringBuilder();
+            }
+            sb.append(sentence);
+        }
+        if (!sb.isEmpty()) {
+            if (fitsTokenLimit(buildIndentedCodeBlock(sb.toString()))) {
+                splitBlocks.add(buildIndentedCodeBlock(sb.toString()));
+            } else {
+                int mid = sb.length() / 2;
+                splitBlocks.add(buildIndentedCodeBlock(sb.substring(0, mid)));
+                splitBlocks.add(buildIndentedCodeBlock(sb.substring(mid, sb.length())));
+            }
+        }
+
+        splitBlocks.forEach(this::addToChunk);
+    }
+
+
+    private static IndentedCodeBlock buildIndentedCodeBlock(String string) {
+
+        IndentedCodeBlock block = new IndentedCodeBlock();
+        block.setLiteral(string);
+        return block;
+    }
+
+
+    private void splitParagraph(Paragraph paragraph) {
+
+        if (fitsTokenLimit(paragraph)) {
+            addToChunk(paragraph);
+            return;
+        }
+
+        List<Node> children = collectAllChildNodes(paragraph);
+
+        if (children.size() == 1) {
+            if (children.get(0) instanceof Text text) {
+                List<Text> splitTexts = splitText(text);
+                for (Text splitText : splitTexts) {
+                    Paragraph paragraph1 = new Paragraph();
+                    paragraph1.appendChild(splitText);
+                    addToChunk(paragraph1);
+                }
+                return;
+            }
+            throwUnsplittableNodeError(children.get(0));
+        }
+
+        Paragraph paragraph1 = new Paragraph();
+        Paragraph paragraph2 = new Paragraph();
+
+        int mid = children.size() / 2;
+        children.subList(0, mid)
+                .forEach(paragraph1::appendChild);
+        children.subList(mid, children.size())
+                .forEach(paragraph2::appendChild);
+
+        splitParagraph(paragraph1);
+        splitParagraph(paragraph2);
+    }
+
+
+    private void throwUnsplittableNodeError(Node node) {
+
+        Renderer renderer = buildRenderer();
+        String renderedNode = renderer.render(node);
+        log.error(renderedNode);
+        throw new IllegalArgumentException(String.format("Node %s exceeds token limit (%d/%d) and can't be split!", node, countTokens(renderedNode), tokenLimit));
+    }
+
+
+    private static List<Node> collectAllChildNodes(Node parent) {
+
+        List<Node> children = new LinkedList<>();
+        Node next;
+        for (Node child = parent.getFirstChild(); child != null; child = next) {
+            next = child.getNext();
+            children.add(child);
+        }
+        return children;
+    }
+
+
+    private List<Text> splitText(Text text) {
+
+        List<Text> splitTexts = new LinkedList<>();
+        StringBuilder sb = new StringBuilder();
+        BreakIterator sentenceIterator = BreakIterator.getSentenceInstance(Locale.ENGLISH);
+        sentenceIterator.setText(text.getLiteral());
+        int start = sentenceIterator.first();
+        for (int end = sentenceIterator.next(); end != BreakIterator.DONE; start = end, end = sentenceIterator.next()) {
+            String sentence = text.getLiteral().substring(start, end);
+            if (!fitsTokenLimit(buildParagraphWithText(sb))) {
+                sb.replace(sb.length() - sentence.length(), sb.length(), "");
+                splitTexts.add(new Text(sb.toString()));
+                sb = new StringBuilder();
+            }
+            sb.append(sentence);
+        }
+        if (!sb.isEmpty()) {
+            if (fitsTokenLimit(buildParagraphWithText(sb))) {
+                splitTexts.add(new Text(sb.toString()));
+            } else {
+                int mid = sb.length() / 2;
+                splitTexts.add(new Text(sb.substring(0, mid)));
+                splitTexts.add(new Text(sb.substring(mid, sb.length())));
+            }
+        }
+        return splitTexts;
+    }
+
+
+    private static Paragraph buildParagraphWithText(StringBuilder sb) {
+
+        Paragraph paragraph = new Paragraph();
+        paragraph.appendChild(new Text(sb.toString()));
+        return paragraph;
+    }
+
+
+    private void splitList(BulletList bulletList) {
+
+        if (fitsTokenLimit(bulletList)) {
+            addToChunk(bulletList);
+            return;
+        }
+
+        BulletList list1 = new BulletList();
+        BulletList list2 = new BulletList();
+
+        splitLists(bulletList, list1, list2);
+    }
+
+
+    private void splitList(OrderedList orderedList) {
+
+        if (fitsTokenLimit(orderedList)) {
+            addToChunk(orderedList);
+            return;
+        }
+
+        OrderedList list1 = new OrderedList();
+        OrderedList list2 = new OrderedList();
+
+        splitLists(orderedList, list1, list2);
+    }
+
+
+    private void splitLists(ListBlock originList, ListBlock list1, ListBlock list2) {
+
+        List<Node> listItems = collectAllChildNodes(originList);
+
+        if (listItems.size() == 1) {
+            collectAllChildNodes(listItems.get(0)).forEach(this::addToChunk);
+        }
+
+        int mid = listItems.size() / 2;
+        listItems.subList(0, mid)
+                .forEach(list1::appendChild);
+        listItems.subList(mid, listItems.size())
+                .forEach(list2::appendChild);
+
+        splitNodeAndAddToChunk(list1);
+        splitNodeAndAddToChunk(list2);
+    }
+
+
+    private void splitTable(TableBlock tableBlock) {
+
+        if (fitsTokenLimit(tableBlock)) {
+            addToChunk(tableBlock);
+            return;
+        }
+
+        TableBlock tableBlock1 = new TableBlock();
+        TableBlock tableBlock2 = new TableBlock();
+
+        tableBlock1.appendChild(nodeCopier.copy(tableBlock.getFirstChild()));
+        tableBlock2.appendChild(nodeCopier.copy(tableBlock.getFirstChild()));
+
+        TableBody tableBody1 = new TableBody();
+        TableBody tableBody2 = new TableBody();
+
+        List<Node> tableRows = collectAllChildNodes(tableBlock.getLastChild());
+
+        if (tableRows.isEmpty()) {
+            throw new IllegalArgumentException("The table headers already exceeds the token limit");
+        }
+        if (tableRows.size() == 1) {
+            throw new IllegalArgumentException("A single table row already exceeds the token limit");
+        }
+
+        int mid = tableRows.size() / 2;
+        tableRows.subList(0, mid)
+                .forEach(tableBody1::appendChild);
+        tableRows.subList(mid, tableRows.size())
+                .forEach(tableBody2::appendChild);
+
+        splitTable(tableBlock1);
+        splitTable(tableBlock2);
+    }
+
+
+    public void visit(BlockQuote blockQuote) {
+
+        this.addToChunk(blockQuote);
+    }
+
+
+    public void visit(BulletList bulletList) {
+
+        this.addToChunk(bulletList);
+    }
+
+
+    public void visit(Code code) {
+
+        this.addToChunk(code);
+    }
+
+
+    public void visit(Emphasis emphasis) {
+
+        this.addToChunk(emphasis);
+    }
+
+
+    public void visit(FencedCodeBlock fencedCodeBlock) {
+
+        this.addToChunk(fencedCodeBlock);
+    }
+
+
+    public void visit(HardLineBreak hardLineBreak) {
+
+        this.addToChunk(hardLineBreak);
+    }
+
+
+    public void visit(ThematicBreak thematicBreak) {
+
+        this.addToChunk(thematicBreak);
+    }
+
+
+    public void visit(HtmlInline htmlInline) {
+
+        this.addToChunk(htmlInline);
+    }
+
+
+    public void visit(HtmlBlock htmlBlock) {
+
+        this.addToChunk(htmlBlock);
+    }
+
+
+    public void visit(Image image) {
+
+        this.addToChunk(image);
+    }
+
+
+    public void visit(IndentedCodeBlock indentedCodeBlock) {
+
+        this.addToChunk(indentedCodeBlock);
+    }
+
+
+    public void visit(Link link) {
+
+        this.addToChunk(link);
+    }
+
+
+    public void visit(ListItem listItem) {
+
+        this.addToChunk(listItem);
+    }
+
+
+    public void visit(OrderedList orderedList) {
+
+        this.addToChunk(orderedList);
+    }
+
+
+    public void visit(Paragraph paragraph) {
+
+        this.addToChunk(paragraph);
+    }
+
+
+    public void visit(SoftLineBreak softLineBreak) {
+
+        this.addToChunk(softLineBreak);
+    }
+
+
+    public void visit(StrongEmphasis strongEmphasis) {
+
+        this.addToChunk(strongEmphasis);
+    }
+
+
+    public void visit(Text text) {
+
+        this.addToChunk(text);
+    }
+
+
+    public void visit(LinkReferenceDefinition linkReferenceDefinition) {
+
+        this.addToChunk(linkReferenceDefinition);
+    }
+
+
+    public void visit(CustomBlock customBlock) {
+
+        this.addToChunk(customBlock);
+    }
+
+
+    public void visit(CustomNode customNode) {
+
+        this.addToChunk(customNode);
+    }
+
+}

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

@@ -0,0 +1,171 @@
+package com.knecon.fforesight.service.layoutparser.processor.markdown;
+
+import org.commonmark.ext.gfm.tables.TableBlock;
+import org.commonmark.ext.gfm.tables.TableBody;
+import org.commonmark.ext.gfm.tables.TableCell;
+import org.commonmark.ext.gfm.tables.TableHead;
+import org.commonmark.ext.gfm.tables.TableRow;
+import org.commonmark.node.BlockQuote;
+import org.commonmark.node.BulletList;
+import org.commonmark.node.Code;
+import org.commonmark.node.Document;
+import org.commonmark.node.Emphasis;
+import org.commonmark.node.FencedCodeBlock;
+import org.commonmark.node.HardLineBreak;
+import org.commonmark.node.Heading;
+import org.commonmark.node.HtmlBlock;
+import org.commonmark.node.HtmlInline;
+import org.commonmark.node.Image;
+import org.commonmark.node.IndentedCodeBlock;
+import org.commonmark.node.Link;
+import org.commonmark.node.LinkReferenceDefinition;
+import org.commonmark.node.ListItem;
+import org.commonmark.node.Node;
+import org.commonmark.node.OrderedList;
+import org.commonmark.node.Paragraph;
+import org.commonmark.node.SoftLineBreak;
+import org.commonmark.node.StrongEmphasis;
+import org.commonmark.node.Text;
+import org.commonmark.node.ThematicBreak;
+import org.mapstruct.Mapper;
+import org.mapstruct.Mapping;
+import org.mapstruct.factory.Mappers;
+
+@Mapper
+public interface NodeCopier {
+
+    NodeCopier INSTANCE = Mappers.getMapper(NodeCopier.class);
+
+
+    default Node copyNodeWithChildren(Node node) {
+
+        Node copy = copy(node);
+
+        Node next;
+        for (Node child = node.getFirstChild(); child != null; child = next) {
+            next = child.getNext();
+            copy.appendChild(copyNodeWithChildren(child));
+        }
+        return copy;
+    }
+
+
+    default Node copy(Node node) {
+
+        return switch (node.getClass().getSimpleName()) {
+            case "BlockQuote" -> copy((BlockQuote) node);
+            case "BulletList" -> copy((BulletList) node);
+            case "Code" -> copy((Code) node);
+            case "Document" -> copy((Document) node);
+            case "Emphasis" -> copy((Emphasis) node);
+            case "FencedCodeBlock" -> copy((FencedCodeBlock) node);
+            case "HardLineBreak" -> copy((HardLineBreak) node);
+            case "Heading" -> copy((Heading) node);
+            case "HtmlBlock" -> copy((HtmlBlock) node);
+            case "HtmlInline" -> copy((HtmlInline) node);
+            case "Image" -> copy((Image) node);
+            case "IndentedCodeBlock" -> copy((IndentedCodeBlock) node);
+            case "Link" -> copy((Link) node);
+            case "LinkReferenceDefinition" -> copy((LinkReferenceDefinition) node);
+            case "ListItem" -> copy((ListItem) node);
+            case "OrderedList" -> copy((OrderedList) node);
+            case "Paragraph" -> copy((Paragraph) node);
+            case "SoftLineBreak" -> copy((SoftLineBreak) node);
+            case "StrongEmphasis" -> copy((StrongEmphasis) node);
+            case "Text" -> copy((Text) node);
+            case "ThematicBreak" -> copy((ThematicBreak) node);
+            case "TableBlock" -> copy((TableBlock) node);
+            case "TableBody" -> copy((TableBody) node);
+            case "TableCell" -> copy((TableCell) node);
+            case "TableHead" -> copy((TableHead) node);
+            case "TableRow" -> copy((TableRow) node);
+            default -> throw new IllegalArgumentException("No copy method found for class: " + node.getClass().getName());
+        };
+    }
+
+
+    BlockQuote copy(BlockQuote blockQuote);
+
+
+    @Mapping(target = "bulletMarker", ignore = true)
+    BulletList copy(BulletList bulletList);
+
+
+    Code copy(Code code);
+
+
+    Document copy(Document document);
+
+
+    @Mapping(target = "delimiter", source = "openingDelimiter")
+    Emphasis copy(Emphasis emphasis);
+
+
+    @Mapping(target = "fenceChar", ignore = true)
+    @Mapping(target = "fenceLength", ignore = true)
+    FencedCodeBlock copy(FencedCodeBlock fencedCodeBlock);
+
+
+    HardLineBreak copy(HardLineBreak hardLineBreak);
+
+
+    Heading copy(Heading heading);
+
+
+    HtmlBlock copy(HtmlBlock htmlBlock);
+
+
+    HtmlInline copy(HtmlInline htmlInline);
+
+
+    Image copy(Image image);
+
+
+    IndentedCodeBlock copy(IndentedCodeBlock indentedCodeBlock);
+
+
+    Link copy(Link link);
+
+
+    LinkReferenceDefinition copy(LinkReferenceDefinition linkReferenceDefinition);
+
+
+    ListItem copy(ListItem listItem);
+
+
+    @Mapping(target = "startNumber", ignore = true)
+    @Mapping(target = "delimiter", ignore = true)
+    OrderedList copy(OrderedList orderedList);
+
+
+    Paragraph copy(Paragraph paragraph);
+
+
+    SoftLineBreak copy(SoftLineBreak softLineBreak);
+
+
+    @Mapping(target = "delimiter", source = "openingDelimiter")
+    StrongEmphasis copy(StrongEmphasis strongEmphasis);
+
+
+    Text copy(Text text);
+
+
+    ThematicBreak copy(ThematicBreak thematicBreak);
+
+
+    TableBlock copy(TableBlock tableBlock);
+
+
+    TableBody copy(TableBody tableBody);
+
+
+    TableCell copy(TableCell tableCell);
+
+
+    TableHead copy(TableHead tableHead);
+
+
+    TableRow copy(TableRow tableRow);
+
+}

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

@@ -0,0 +1,65 @@
+package com.knecon.fforesight.service.layoutparser.processor.markdown;
+
+import java.lang.reflect.Field;
+
+import org.commonmark.node.Node;
+
+public class ReflectionNodeCopier {
+
+    NodeCopier mapperNodeCopier;
+
+
+
+    public static Node copyNode(Node node) {
+
+        Node copy = deepCopy(node);
+        copyChildren(node, copy);
+        return copy;
+    }
+
+
+    private static void copyChildren(Node nodeToCopy, Node copy) {
+
+        Node next;
+        for (Node node = nodeToCopy.getFirstChild(); node != null; node = next) {
+            next = node.getNext();
+            copy.appendChild(copyNode(node));
+        }
+    }
+
+
+    private static <T> T deepCopy(T object) {
+
+        try {
+            Class<?> clazz = object.getClass();
+            T copy = (T) clazz.getDeclaredConstructor().newInstance();
+
+            for (Field field : clazz.getDeclaredFields()) {
+                field.setAccessible(true);
+                Object value = field.get(object);
+                if (isPrimitiveOrWrapper(field.getType()) || field.getType().equals(String.class)) {
+                    field.set(copy, value);
+                }
+            }
+            return copy;
+        } catch (Exception e) {
+            e.printStackTrace();
+            return null;
+        }
+    }
+
+
+    private static boolean isPrimitiveOrWrapper(Class<?> type) {
+
+        return type.isPrimitive()
+               || type == Boolean.class
+               || type == Byte.class
+               || type == Character.class
+               || type == Double.class
+               || type == Float.class
+               || type == Integer.class
+               || type == Long.class
+               || type == Short.class;
+    }
+
+}

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

@@ -7,7 +7,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.outline.Outlin
 import com.knecon.fforesight.service.layoutparser.processor.model.outline.TableOfContents;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.UnclassifiedText;
-import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutparsingVisualizations;
 
 import lombok.Data;
 import lombok.NoArgsConstructor;
@@ -25,7 +25,7 @@ public class ClassificationDocument {
     private FloatFrequencyCounter fontSizeCounter = new FloatFrequencyCounter();
     private StringFrequencyCounter fontCounter = new StringFrequencyCounter();
     private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
-    private LayoutDebugLayer layoutDebugLayer = new LayoutDebugLayer();
+    private LayoutparsingVisualizations visualizations = new LayoutparsingVisualizations();
     private boolean headlines;
 
     private long rulesVersion;

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

@@ -18,7 +18,6 @@ import lombok.RequiredArgsConstructor;
 
 @Data
 @RequiredArgsConstructor
-
 public class ClassificationPage {
 
     @NonNull
@@ -26,7 +25,7 @@ public class ClassificationPage {
 
     private List<OutlineObject> outlineObjects = new ArrayList<>();
 
-    private List<AbstractPageBlock> headlines = new ArrayList<>();
+    private List<AbstractPageBlock> headlines  = new ArrayList<>();
 
     private List<ClassifiedImage> images = new ArrayList<>();
 
@@ -45,7 +44,7 @@ public class ClassificationPage {
     private float pageWidth;
     private float pageHeight;
 
-    private CleanRulings cleanRulings;
+    CleanRulings cleanRulings;
 
     private Map<String, List<Rectangle2D>> markedContentBboxPerType = new HashMap<>();
 

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

@@ -7,15 +7,17 @@ import java.util.List;
 import java.util.Map;
 import java.util.stream.Collectors;
 
+import com.google.common.base.Functions;
+
 import lombok.Getter;
 
 @Getter
 public class FloatFrequencyCounter {
 
-    Map<Double, Integer> countPerValue = new HashMap<>();
+    Map<Float, Integer> countPerValue = new HashMap<>();
 
 
-    public void add(double value) {
+    public void add(float value) {
 
         if (!countPerValue.containsKey(value)) {
             countPerValue.put(value, 1);
@@ -25,9 +27,9 @@ public class FloatFrequencyCounter {
     }
 
 
-    public void addAll(Map<Double, Integer> otherCounter) {
+    public void addAll(Map<Float, Integer> otherCounter) {
 
-        for (Map.Entry<Double, Integer> entry : otherCounter.entrySet()) {
+        for (Map.Entry<Float, Integer> entry : otherCounter.entrySet()) {
             if (countPerValue.containsKey(entry.getKey())) {
                 countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
             } else {
@@ -37,10 +39,10 @@ public class FloatFrequencyCounter {
     }
 
 
-    public Double getMostPopular() {
+    public Float getMostPopular() {
 
-        Map.Entry<Double, Integer> mostPopular = null;
-        for (Map.Entry<Double, Integer> entry : countPerValue.entrySet()) {
+        Map.Entry<Float, Integer> mostPopular = null;
+        for (Map.Entry<Float, Integer> entry : countPerValue.entrySet()) {
             if (mostPopular == null || entry.getValue() >= mostPopular.getValue()) {
                 mostPopular = entry;
             }
@@ -49,24 +51,26 @@ public class FloatFrequencyCounter {
     }
 
 
-    public List<Double> getHigherThanMostPopular() {
+    public List<Float> getHighterThanMostPopular() {
 
-        Double mostPopular = getMostPopular();
-        List<Double> higher = new ArrayList<>();
-        for (Double value : countPerValue.keySet()) {
+        Float mostPopular = getMostPopular();
+        List<Float> higher = new ArrayList<>();
+        for (Float value : countPerValue.keySet()) {
             if (value > mostPopular) {
                 higher.add(value);
             }
         }
 
-        return higher.stream().sorted(Collections.reverseOrder()).collect(Collectors.toList());
+        return higher.stream()
+                .sorted(Collections.reverseOrder())
+                .collect(Collectors.toList());
     }
 
 
-    public Double getHighest() {
+    public Float getHighest() {
 
-        Double highest = null;
-        for (Double value : countPerValue.keySet()) {
+        Float highest = null;
+        for (Float value : countPerValue.keySet()) {
             if (highest == null || value > highest) {
                 highest = value;
             }
@@ -74,4 +78,16 @@ public class FloatFrequencyCounter {
         return highest;
     }
 
+
+    public double getAverage() {
+
+        double sum = countPerValue.keySet()
+                .stream()
+                .mapToDouble(fontSize -> fontSize * countPerValue.get(fontSize)).sum();
+        double count = countPerValue.values()
+                .stream()
+                .mapToInt(Integer::intValue).sum();
+        return sum / count;
+    }
+
 }

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

@@ -15,7 +15,7 @@ import lombok.experimental.FieldDefaults;
 @FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
 public class SectionIdentifier {
 
-    public static Pattern numericalIdentifierPattern = Pattern.compile("^[\\s]?(\\d+)[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?");
+    static Pattern numericalIdentifierPattern = Pattern.compile("^[\\s]?(\\d+)[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?");
 
     public enum Format {
         EMPTY,

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

@@ -1,94 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Footer;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Header;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Headline;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Image;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SuperSection;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell;
-
-public abstract class AbstractNodeVisitor implements NodeVisitor {
-
-    @Override
-    public void visit(Document document) {
-
-        visitChildren(document);
-    }
-
-
-    @Override
-    public void visit(SuperSection superSection) {
-
-        visitChildren(superSection);
-    }
-
-
-    @Override
-    public void visit(Section section) {
-
-        visitChildren(section);
-    }
-
-
-    @Override
-    public void visit(Headline headline) {
-
-        visitChildren(headline);
-    }
-
-
-    @Override
-    public void visit(Paragraph paragraph) {
-
-        visitChildren(paragraph);
-    }
-
-
-    @Override
-    public void visit(Footer footer) {
-
-        visitChildren(footer);
-    }
-
-
-    @Override
-    public void visit(Header header) {
-
-        visitChildren(header);
-    }
-
-
-    @Override
-    public void visit(Image image) {
-
-        visitChildren(image);
-    }
-
-
-    @Override
-    public void visit(Table table) {
-
-        visitChildren(table);
-    }
-
-
-    @Override
-    public void visit(TableCell tableCell) {
-
-        visitChildren(tableCell);
-    }
-
-
-    private void visitChildren(SemanticNode semanticNode) {
-
-        semanticNode.streamChildren()
-                .forEach(node -> node.accept(this));
-
-    }
-
-}

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

@@ -2,7 +2,6 @@ package com.knecon.fforesight.service.layoutparser.processor.model.graph;
 
 import static java.lang.String.format;
 
-import java.util.ArrayList;
 import java.util.Collections;
 import java.util.LinkedList;
 import java.util.List;
@@ -40,10 +39,7 @@ public class DocumentTree {
 
     public TextBlock buildTextBlock() {
 
-        return allEntriesInOrder().map(Entry::getNode)
-                .filter(SemanticNode::isLeaf)
-                .map(SemanticNode::getLeafTextBlock)
-                .collect(new TextBlockCollector());
+        return allEntriesInOrder().map(Entry::getNode).filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
     }
 
 
@@ -117,16 +113,13 @@ public class DocumentTree {
 
     public Stream<SemanticNode> childNodes(List<Integer> treeId) {
 
-        return getEntryById(treeId).children.stream()
-                .map(Entry::getNode);
+        return getEntryById(treeId).children.stream().map(Entry::getNode);
     }
 
 
     public Stream<SemanticNode> childNodesOfType(List<Integer> treeId, NodeType nodeType) {
 
-        return getEntryById(treeId).children.stream()
-                .filter(entry -> entry.node.getType().equals(nodeType))
-                .map(Entry::getNode);
+        return getEntryById(treeId).children.stream().filter(entry -> entry.node.getType().equals(nodeType)).map(Entry::getNode);
     }
 
 
@@ -163,32 +156,26 @@ public class DocumentTree {
 
     public Stream<Entry> allEntriesInOrder() {
 
-        return Stream.of(root)
-                .flatMap(DocumentTree::flatten);
+        return Stream.of(root).flatMap(DocumentTree::flatten);
     }
 
 
     public Stream<Entry> allSubEntriesInOrder(List<Integer> parentId) {
 
-        return getEntryById(parentId).children.stream()
-                .flatMap(DocumentTree::flatten);
+        return getEntryById(parentId).children.stream().flatMap(DocumentTree::flatten);
     }
 
 
     @Override
     public String toString() {
 
-        return String.join("\n",
-                           allEntriesInOrder().map(Entry::toString)
-                                   .toList());
+        return String.join("\n", allEntriesInOrder().map(Entry::toString).toList());
     }
 
 
     private static Stream<Entry> flatten(Entry entry) {
 
-        return Stream.concat(Stream.of(entry),
-                             entry.children.stream()
-                                     .flatMap(DocumentTree::flatten));
+        return Stream.concat(Stream.of(entry), entry.children.stream().flatMap(DocumentTree::flatten));
     }
 
 

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

@@ -1,45 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Footer;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Header;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Headline;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Image;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SuperSection;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell;
-
-public interface NodeVisitor {
-
-    void visit(Document document);
-
-
-    void visit(SuperSection superSection);
-
-
-    void visit(Section section);
-
-
-    void visit(Headline headline);
-
-
-    void visit(Paragraph paragraph);
-
-
-    void visit(Footer footer);
-
-
-    void visit(Header header);
-
-
-    void visit(Image image);
-
-
-    void visit(Table table);
-
-
-    void visit(TableCell tableCell);
-
-}

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

@@ -79,12 +79,6 @@ public class TextRange implements Comparable<TextRange> {
 
     public boolean contains(int index) {
 
-        return start <= index && index <= end;
-    }
-
-
-    public boolean containsExclusive(int index) {
-
         return start <= index && index < end;
     }
 
@@ -97,13 +91,8 @@ public class TextRange implements Comparable<TextRange> {
 
     public List<TextRange> split(List<Integer> splitIndices) {
 
-        if (splitIndices.stream()
-                .anyMatch(idx -> !this.containsExclusive(idx))) {
-            throw new IndexOutOfBoundsException(format("%s splitting indices are out of range for %s",
-                                                       splitIndices.stream()
-                                                               .filter(idx -> !this.containsExclusive(idx))
-                                                               .toList(),
-                                                       this));
+        if (splitIndices.stream().anyMatch(idx -> !this.contains(idx))) {
+            throw new IndexOutOfBoundsException(format("%s splitting indices are out of range for %s", splitIndices.stream().filter(idx -> !this.contains(idx)).toList(), this));
         }
         List<TextRange> splitBoundaries = new LinkedList<>();
         int previousIndex = start;
@@ -120,23 +109,15 @@ public class TextRange implements Comparable<TextRange> {
         return splitBoundaries;
     }
 
-
     public IntStream intStream() {
 
         return IntStream.range(start, end);
     }
 
-
     public static TextRange merge(Collection<TextRange> boundaries) {
 
-        int minStart = boundaries.stream()
-                .mapToInt(TextRange::start)
-                .min()
-                .orElseThrow(IllegalArgumentException::new);
-        int maxEnd = boundaries.stream()
-                .mapToInt(TextRange::end)
-                .max()
-                .orElseThrow(IllegalArgumentException::new);
+        int minStart = boundaries.stream().mapToInt(TextRange::start).min().orElseThrow(IllegalArgumentException::new);
+        int maxEnd = boundaries.stream().mapToInt(TextRange::end).max().orElseThrow(IllegalArgumentException::new);
         return new TextRange(minStart, maxEnd);
     }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/entity/RedactionEntity.java

@@ -28,7 +28,7 @@ import lombok.experimental.FieldDefaults;
 @AllArgsConstructor
 @FieldDefaults(level = AccessLevel.PRIVATE)
 @EqualsAndHashCode(onlyExplicitlyIncluded = true)
-public class TextEntity {
+public class RedactionEntity {
 
     // initial values
     @EqualsAndHashCode.Include
@@ -47,7 +47,7 @@ public class TextEntity {
     boolean dictionaryEntry;
     boolean dossierDictionaryEntry;
     Set<Engine> engines;
-    Set<TextEntity> references;
+    Set<RedactionEntity> references;
     @Builder.Default
     Deque<Integer> matchedRules = new LinkedList<>();
     String redactionReason;
@@ -66,9 +66,9 @@ public class TextEntity {
     SemanticNode deepestFullyContainingNode;
 
 
-    public static TextEntity initialEntityNode(TextRange textRange, String type, EntityType entityType) {
+    public static RedactionEntity initialEntityNode(TextRange textRange, String type, EntityType entityType) {
 
-        return TextEntity.builder().type(type).entityType(entityType).textRange(textRange).engines(new HashSet<>()).references(new HashSet<>()).build();
+        return RedactionEntity.builder().type(type).entityType(entityType).textRange(textRange).engines(new HashSet<>()).references(new HashSet<>()).build();
     }
 
 
@@ -155,21 +155,21 @@ public class TextEntity {
     }
 
 
-    public boolean containedBy(TextEntity textEntity) {
+    public boolean containedBy(RedactionEntity redactionEntity) {
 
-        return this.textRange.containedBy(textEntity.getTextRange());
+        return this.textRange.containedBy(redactionEntity.getTextRange());
     }
 
 
-    public boolean contains(TextEntity textEntity) {
+    public boolean contains(RedactionEntity redactionEntity) {
 
-        return this.textRange.contains(textEntity.getTextRange());
+        return this.textRange.contains(redactionEntity.getTextRange());
     }
 
 
-    public boolean intersects(TextEntity textEntity) {
+    public boolean intersects(RedactionEntity redactionEntity) {
 
-        return this.textRange.intersects(textEntity.getTextRange());
+        return this.textRange.intersects(redactionEntity.getTextRange());
     }
 
 
@@ -185,13 +185,13 @@ public class TextEntity {
     }
 
 
-    public void addReference(TextEntity reference) {
+    public void addReference(RedactionEntity reference) {
 
         references.add(reference);
     }
 
 
-    public void addReferences(List<TextEntity> references) {
+    public void addReferences(List<RedactionEntity> references) {
 
         this.references.addAll(references);
     }

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

@@ -8,7 +8,7 @@ import java.util.Set;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.TextEntity;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -39,7 +39,7 @@ public abstract class AbstractSemanticNode implements GenericSemanticNode {
 
     @Builder.Default
     @EqualsAndHashCode.Exclude
-    Set<TextEntity> entities = new HashSet<>();
+    Set<RedactionEntity> entities = new HashSet<>();
 
     @EqualsAndHashCode.Exclude
     Map<Page, Rectangle2D> bBoxCache;

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

@@ -11,9 +11,8 @@ import java.util.stream.Stream;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutparsingVisualizations;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -34,7 +33,7 @@ public class Document extends AbstractSemanticNode {
     Set<Page> pages;
     Integer numberOfPages;
 
-    LayoutDebugLayer layoutDebugLayer;
+    LayoutparsingVisualizations visualizations;
 
 
     @Override
@@ -102,7 +101,8 @@ public class Document extends AbstractSemanticNode {
     public Headline getHeadline() {
 
         return streamAllSubNodesOfType(NodeType.HEADLINE).map(node -> (Headline) node)
-                .findFirst().orElse(Headline.builder().build());
+                .findFirst()
+                .orElse(Headline.builder().build());
     }
 
 
@@ -163,11 +163,4 @@ public class Document extends AbstractSemanticNode {
         return bBox;
     }
 
-
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
 }

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -35,12 +34,6 @@ public class Footer extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-        visitor.visit(this);
-    }
-
-
     @Override
     public TextBlock getTextBlock() {
 

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -28,13 +27,6 @@ public class Header extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
-
     @Override
     public NodeType getType() {
 

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -35,13 +34,6 @@ public class Headline extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
-
     @Override
     public TextBlock getTextBlock() {
 

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

@@ -7,7 +7,6 @@ import java.util.Map;
 import java.util.Set;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -29,8 +28,6 @@ public class Image extends AbstractSemanticNode {
 
     String id;
 
-    String representationHash;
-
     ImageType imageType;
     boolean transparent;
     Rectangle2D position;
@@ -87,29 +84,10 @@ public class Image extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
-
     @Override
     public boolean isLeaf() {
 
         return true;
     }
 
-
-    public double getArea() {
-
-        return position.getWidth() * position.getHeight();
-    }
-
-
-    public boolean isFullPageImage() {
-
-        return imageType.equals(ImageType.OCR) || getArea() >= 0.5 * page.getArea();
-    }
-
 }

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

@@ -6,6 +6,7 @@ public enum ImageType {
     LOGO,
     FORMULA,
     SIGNATURE,
+
     SIGNATURE_VISUAL,
     OTHER,
     OCR,

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

@@ -1,15 +1,12 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
-import java.util.Comparator;
 import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Set;
-import java.util.stream.Stream;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.TextEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
 
@@ -32,8 +29,9 @@ public class Page {
     Integer height;
     Integer width;
     Integer rotation;
+
     @EqualsAndHashCode.Exclude
-    List<AtomicTextBlock> textBlocksOnPage;
+    List<SemanticNode> mainBody;
     @EqualsAndHashCode.Exclude
     Header header;
     @EqualsAndHashCode.Exclude
@@ -41,7 +39,7 @@ public class Page {
 
     @Builder.Default
     @EqualsAndHashCode.Exclude
-    Set<TextEntity> entities = new HashSet<>();
+    Set<RedactionEntity> entities = new HashSet<>();
 
     @Builder.Default
     @EqualsAndHashCode.Exclude
@@ -55,40 +53,14 @@ public class Page {
                 .width((int) classificationPage.getPageWidth())
                 .number(classificationPage.getPageNumber())
                 .rotation(classificationPage.getRotation())
-                .textBlocksOnPage(new LinkedList<>())
+                .mainBody(new LinkedList<>())
                 .build();
     }
 
 
-    /**
-     * Constructs and returns a {@link TextBlock} representing the concatenated text of all leaf semantic nodes in the main body.
-     *
-     * @return The main body text block.
-     */
     public TextBlock getMainBodyTextBlock() {
 
-        return textBlocksOnPage.stream()
-                .collect(new TextBlockCollector());
-    }
-
-
-    public List<SemanticNode> getMainBody() {
-
-        return textBlocksOnPage.stream()
-                .map(AtomicTextBlock::getParent)
-                .map(this::getHighestParentOnPage)
-                .distinct()
-                .toList();
-    }
-
-
-    private SemanticNode getHighestParentOnPage(SemanticNode node) {
-
-        SemanticNode currentNode = node;
-        while (currentNode.hasParent() && currentNode.getParent().onlyOnPage(this)) {
-            currentNode = currentNode.getParent();
-        }
-        return currentNode;
+        return mainBody.stream().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
     }
 
 
@@ -112,10 +84,4 @@ public class Page {
         return o instanceof Page && o.hashCode() == this.hashCode();
     }
 
-
-    public double getArea() {
-
-        return height * width;
-    }
-
 }

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 
 import lombok.AccessLevel;
@@ -35,13 +34,6 @@ public class Paragraph extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
-
     @Override
     public TextBlock getTextBlock() {
 

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -27,20 +26,15 @@ public class Section extends AbstractSemanticNode {
     public Headline getHeadline() {
 
         return streamChildrenOfType(NodeType.HEADLINE).map(node -> (Headline) node)
-                .findFirst().orElseGet(() -> getParent().getHeadline());
-    }
-
-
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
+                .findFirst()
+                .orElseGet(() -> getParent().getHeadline());
     }
 
 
     public boolean hasTables() {
 
-        return streamAllSubNodesOfType(NodeType.TABLE).findAny().isPresent();
+        return streamAllSubNodesOfType(NodeType.TABLE).findAny()
+                .isPresent();
     }
 
 

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

@@ -14,16 +14,14 @@ import java.util.stream.Stream;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.TextRange;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.EntityType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.TextEntity;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-import com.knecon.fforesight.service.layoutparser.processor.utils.BBoxMergingUtility;
 import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 
 public interface SemanticNode {
 
@@ -56,7 +54,7 @@ public interface SemanticNode {
      *
      * @return Set of all Entities associated with this Node
      */
-    Set<TextEntity> getEntities();
+    Set<RedactionEntity> getEntities();
 
 
     /**
@@ -223,7 +221,7 @@ public interface SemanticNode {
      * @param type string representing the type of entities to return
      * @return List of RedactionEntities of any the type
      */
-    default List<TextEntity> getEntitiesOfType(String type) {
+    default List<RedactionEntity> getEntitiesOfType(String type) {
 
         return getEntities().stream()
                 .filter(redactionEntity -> redactionEntity.getType().equals(type))
@@ -237,7 +235,7 @@ public interface SemanticNode {
      * @param types A list of strings representing the types of entities to return
      * @return List of RedactionEntities of any provided type
      */
-    default List<TextEntity> getEntitiesOfType(List<String> types) {
+    default List<RedactionEntity> getEntitiesOfType(List<String> types) {
 
         return getEntities().stream()
                 .filter(redactionEntity -> redactionEntity.isAnyType(types))
@@ -255,7 +253,8 @@ public interface SemanticNode {
 
         TextBlock textBlock = getTextBlock();
         if (!textBlock.getAtomicTextBlocks().isEmpty()) {
-            return getTextBlock().getAtomicTextBlocks().get(0).getNumberOnPage();
+            return getTextBlock().getAtomicTextBlocks()
+                    .get(0).getNumberOnPage();
         } else {
             return -1;
         }
@@ -340,19 +339,19 @@ public interface SemanticNode {
      * This function is used during insertion of EntityNodes into the graph, it checks if the boundary of the RedactionEntity intersects or even contains the RedactionEntity.
      * It sets the fields accordingly and recursively calls this function on all its children.
      *
-     * @param textEntity RedactionEntity, which is being inserted into the graph
+     * @param redactionEntity RedactionEntity, which is being inserted into the graph
      */
-    default void addThisToEntityIfIntersects(TextEntity textEntity) {
+    default void addThisToEntityIfIntersects(RedactionEntity redactionEntity) {
 
         TextBlock textBlock = getTextBlock();
-        if (textBlock.getTextRange().intersects(textEntity.getTextRange())) {
-            if (textBlock.containsBoundary(textEntity.getTextRange())) {
-                textEntity.setDeepestFullyContainingNode(this);
+        if (textBlock.getTextRange().intersects(redactionEntity.getTextRange())) {
+            if (textBlock.containsBoundary(redactionEntity.getTextRange())) {
+                redactionEntity.setDeepestFullyContainingNode(this);
             }
 
-            textEntity.addIntersectingNode(this);
-            streamChildren().filter(semanticNode -> semanticNode.getBoundary().intersects(textEntity.getTextRange()))
-                    .forEach(node -> node.addThisToEntityIfIntersects(textEntity));
+            redactionEntity.addIntersectingNode(this);
+            streamChildren().filter(semanticNode -> semanticNode.getBoundary().intersects(redactionEntity.getTextRange()))
+                    .forEach(node -> node.addThisToEntityIfIntersects(redactionEntity));
         }
     }
 
@@ -473,19 +472,34 @@ public interface SemanticNode {
      */
     private Map<Page, Rectangle2D> getBBoxFromChildren() {
 
-        List<Map<Page, Rectangle2D>> childrenBBoxes = streamChildren().filter(child -> !isFullPageImage(child))
+        Map<Page, Rectangle2D> bBoxPerPage = new HashMap<>();
+
+        List<Map<Page, Rectangle2D>> childrenBBoxes = streamChildren() //
+                .filter(SemanticNode::isNotOcrImage)
                 .map(SemanticNode::getBBox)
                 .toList();
-        return BBoxMergingUtility.mergeBBoxes(childrenBBoxes);
+
+        Set<Page> pages = childrenBBoxes.stream()
+                .flatMap(map -> map.keySet()
+                        .stream())
+                .collect(Collectors.toSet());
+        for (Page page : pages) {
+            Rectangle2D bBoxOnPage = childrenBBoxes.stream()
+                    .filter(childBboxPerPage -> childBboxPerPage.containsKey(page))
+                    .map(childBboxPerPage -> childBboxPerPage.get(page))
+                    .collect(RectangleTransformations.collectBBox());
+            bBoxPerPage.put(page, bBoxOnPage);
+        }
+        return bBoxPerPage;
     }
 
 
-    private static boolean isFullPageImage(SemanticNode child) {
+    private static boolean isNotOcrImage(SemanticNode node) {
 
-        if (!child.getType().equals(NodeType.IMAGE)) {
-            return false;
+        if (!node.getType().equals(NodeType.IMAGE)) {
+            return true;
         }
-        return ((Image) child).isFullPageImage();
+        return false;
     }
 
 
@@ -502,20 +516,4 @@ public interface SemanticNode {
         return bBoxPerPage;
     }
 
-
-    void accept(NodeVisitor visitor);
-
-
-    /**
-     * Checks wether this SemanticNode appears on a single page only, and if that page is the provided one.
-     *
-     * @param page the page to check
-     * @return true, when SemanticNode is on a single page only and the page is the provided page. Otherwise, false.
-     */
-    default boolean onlyOnPage(Page page) {
-
-        Set<Page> pages = getPages();
-        return pages.size() == 1 && pages.contains(page);
-    }
-
 }

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -27,14 +26,8 @@ public class SuperSection extends AbstractSemanticNode {
     public Headline getHeadline() {
 
         return streamChildrenOfType(NodeType.HEADLINE).map(node -> (Headline) node)
-                .findFirst().orElseGet(() -> getParent().getHeadline());
-    }
-
-
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
+                .findFirst()
+                .orElseGet(() -> getParent().getHeadline());
     }
 
 

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

@@ -15,8 +15,7 @@ import java.util.stream.Stream;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.TextEntity;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
 
@@ -44,7 +43,7 @@ public class Table implements SemanticNode {
 
     @Builder.Default
     @EqualsAndHashCode.Exclude
-    Set<TextEntity> entities = new HashSet<>();
+    Set<RedactionEntity> entities = new HashSet<>();
 
     @EqualsAndHashCode.Exclude
     Map<Page, Rectangle2D> bBoxCache;
@@ -55,7 +54,7 @@ public class Table implements SemanticNode {
      * @param strings Strings to check whether a row contains them
      * @return Stream of all entities in this table, that appear in a row, which contains any of the provided strings
      */
-    public Stream<TextEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) {
+    public Stream<RedactionEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) {
 
         return IntStream.range(0, numberOfRows).boxed()
                 .filter(row -> rowContainsStringsIgnoreCase(row, strings))
@@ -89,7 +88,7 @@ public class Table implements SemanticNode {
      * @param value  the string which the table cell should contain
      * @return a stream of all entities, which appear in a row where at least one cell has the provided header and the provided value.
      */
-    public Stream<TextEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) {
+    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) {
 
         List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header))
                 .map(TableCell::getCol)
@@ -108,7 +107,7 @@ public class Table implements SemanticNode {
      * @param values the strings which the table cell should contain
      * @return a stream of all entities, which appear in a row where at least one cell has the provided header and any provided value.
      */
-    public Stream<TextEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) {
+    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) {
 
         List<Integer> colsWithHeader = streamHeaders().filter(headerNode -> headerNode.containsString(header))
                 .map(TableCell::getCol)
@@ -126,7 +125,7 @@ public class Table implements SemanticNode {
      * @param types type strings to check whether a row contains an entity like them
      * @return Stream of all entities in this table, that appear in a row, which contains at least one entity with any of the provided types.
      */
-    public Stream<TextEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) {
+    public Stream<RedactionEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) {
 
         List<Integer> rowsWithEntityOfType = IntStream.range(0, numberOfRows).boxed()
                 .filter(rowNumber -> streamEntityTypesInRow(rowNumber).anyMatch(existingType -> types.stream()
@@ -146,7 +145,7 @@ public class Table implements SemanticNode {
      * @param types type strings to check whether a row doesn't contain an entity like it
      * @return Stream of all entities in this table, that appear in a row, which does not contain any entity with any of the provided types.
      */
-    public Stream<TextEntity> streamEntitiesWhereRowContainsNoEntitiesOfType(List<String> types) {
+    public Stream<RedactionEntity> streamEntitiesWhereRowContainsNoEntitiesOfType(List<String> types) {
 
         List<Integer> rowsWithNoEntityOfType = IntStream.range(0, numberOfRows).boxed()
                 .filter(rowNumber -> streamEntityTypesInRow(rowNumber).noneMatch(existingType -> types.stream()
@@ -164,7 +163,7 @@ public class Table implements SemanticNode {
 
         return streamRow(rowNumber).map(TableCell::getEntities)
                 .flatMap(Collection::stream)
-                .map(TextEntity::getType)
+                .map(RedactionEntity::getType)
                 .distinct();
     }
 
@@ -305,12 +304,12 @@ public class Table implements SemanticNode {
      * Finds all entities of the provided type, which appear in the same row that the provided entity appears in.
      *
      * @param type            the type of entities to search for
-     * @param textEntity the entity, which appears in the row to search
+     * @param redactionEntity the entity, which appears in the row to search
      * @return List of all entities of the provided type, which appear in the same row that the provided entity appears in.
      */
-    public List<TextEntity> getEntitiesOfTypeInSameRow(String type, TextEntity textEntity) {
+    public List<RedactionEntity> getEntitiesOfTypeInSameRow(String type, RedactionEntity redactionEntity) {
 
-        return textEntity.getIntersectingNodes()
+        return redactionEntity.getIntersectingNodes()
                 .stream()
                 .filter(node -> node instanceof TableCell)
                 .map(node -> (TableCell) node)
@@ -354,10 +353,4 @@ public class Table implements SemanticNode {
         return bBoxCache;
     }
 
-
-    @Override
-    public void accept(NodeVisitor visitor) {
-        visitor.visit(this);
-    }
-
 }

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

@@ -5,7 +5,6 @@ import java.util.HashMap;
 import java.util.Map;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.NodeVisitor;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
 
@@ -43,13 +42,6 @@ public class TableCell extends AbstractSemanticNode {
     }
 
 
-    @Override
-    public void accept(NodeVisitor visitor) {
-
-        visitor.visit(this);
-    }
-
-
     @Override
     public NodeType getType() {
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/AtomicTextBlock.java

@@ -74,7 +74,7 @@ public class AtomicTextBlock implements TextBlock {
         }
 
         Set<Integer> lbInBoundary = lineBreaks.stream()
-                .map(i -> i + this.textRange.start())
+                .map(i -> i + stringTextRange.start())
                 .filter(stringTextRange::contains)
                 .collect(Collectors.toSet());
         if (stringTextRange.end() == getTextRange().end()) {
@@ -177,9 +177,6 @@ public class AtomicTextBlock implements TextBlock {
             throw new IndexOutOfBoundsException(format("line %d out of range for AtomicTextBlock with %d lines", lineNumber, numberOfLines()));
         }
         if (lineNumber == 0) {
-            if (lineBreaks.isEmpty()) {
-                return searchText;
-            }
             return subSequence(textRange.start(), lineBreaks.get(0) + textRange.start());
         } else if (lineNumber == numberOfLines() - 1) {
             return subSequence(lineBreaks.get(lineBreaks.size() - 1) + textRange.start(), textRange.end());
@@ -257,7 +254,7 @@ public class AtomicTextBlock implements TextBlock {
     }
 
 
-    protected List<Integer> getAllLineBreaksInBoundary(TextRange textRange) {
+    private List<Integer> getAllLineBreaksInBoundary(TextRange textRange) {
 
         return getLineBreaks().stream()
                 .map(linebreak -> linebreak + this.textRange.start())

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/ConcatenatedTextBlock.java

@@ -69,7 +69,7 @@ public class ConcatenatedTextBlock implements TextBlock {
     private AtomicTextBlock getAtomicTextBlockByStringIndex(int stringIdx) {
 
         return atomicTextBlocks.stream()
-                .filter(textBlock -> textBlock.getTextRange().containsExclusive(stringIdx))
+                .filter(textBlock -> textBlock.getTextRange().contains(stringIdx))
                 .findAny()
                 .orElseThrow(IndexOutOfBoundsException::new);
     }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/TextBlock.java

@@ -115,7 +115,7 @@ public interface TextBlock extends CharSequence {
 
     default boolean containsIndex(int stringIndex) {
 
-        return getTextRange().containsExclusive(stringIndex);
+        return getTextRange().contains(stringIndex);
     }
 
 

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

@@ -10,8 +10,8 @@ import lombok.NonNull;
 import lombok.RequiredArgsConstructor;
 
 @Data
-@AllArgsConstructor
 @RequiredArgsConstructor
+@AllArgsConstructor
 public class ClassifiedImage {
 
     @NonNull
@@ -20,19 +20,9 @@ public class ClassifiedImage {
     private ImageType imageType;
     private boolean sourceByAi;
     private boolean isAppendedToSection;
-    private boolean hasTransparency;
-    private int page;
     @NonNull
-    private String representation;
-
-
-    public ClassifiedImage(@NonNull Rectangle2D position, @NonNull ImageType imageType, boolean hasTransparency, int page, @NonNull String representation) {
-
-        this.position = position;
-        this.imageType = imageType;
-        this.hasTransparency = hasTransparency;
-        this.page = page;
-        this.representation = representation;
-    }
+    private boolean hasTransparency;
+    @NonNull
+    private int page;
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineExtractorService.java

@@ -1,6 +1,5 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.outline;
 
-import java.awt.geom.AffineTransform;
 import java.awt.geom.Point2D;
 import java.io.IOException;
 import java.util.ArrayList;
@@ -27,9 +26,6 @@ import org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDDocume
 import org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDOutlineItem;
 import org.springframework.stereotype.Service;
 
-import com.knecon.fforesight.service.layoutparser.processor.utils.CoordinateTransforms;
-import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
-
 import lombok.SneakyThrows;
 import lombok.extern.slf4j.Slf4j;
 
@@ -86,20 +82,11 @@ public class OutlineExtractorService {
 
         String title = item.getTitle();
 
-        PDPage page;
-        try {
-            // Can throw: "Error: can't convert to Destination COSArray" for some OCR'd PDFs
-            page = item.findDestinationPage(document);
-            if (page == null) {
-                return Optional.empty();
-            }
-        } catch (IOException e) {
-            log.info(String.format("Error occurred during position resolution for outline item with title %s: " + e, title));
+        PDPage page = item.findDestinationPage(document);
+        if (page == null) {
             return Optional.empty();
         }
-
-        int pageNumber = document.getPages().indexOf(page) + 1;
-        AffineTransform userSpaceToPageCoords = CoordinateTransforms.calculateInitialUserSpaceCoordsToPageCoords(PageInformation.fromPDPage(pageNumber, page));
+        int pageNumber = document.getPages().indexOf(page);
 
         Optional<Point2D> outlinePosition = Optional.empty();
 
@@ -128,15 +115,8 @@ public class OutlineExtractorService {
             log.info(String.format("Error occurred during position resolution for outline item on page %s with title %s: " + e, pageNumber, title));
         }
 
-        return Optional.of(new OutlineObjectTreeNode(new OutlineObject(title,
-                                                                       pageNumber,
-                                                                       transformPointToPageCoords(outlinePosition, userSpaceToPageCoords), depth)));
-    }
+        return Optional.of(new OutlineObjectTreeNode(new OutlineObject(title, pageNumber, outlinePosition.orElse(new Point2D.Float(0, 0)), depth)));
 
-
-    private static Point2D transformPointToPageCoords(Optional<Point2D> outlinePosition, AffineTransform userSpaceToPageCoords) {
-
-        return outlinePosition.map(point -> userSpaceToPageCoords.transform(point, null)).orElse(null);
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineObject.java

@@ -1,34 +1,27 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.outline;
 
 import java.awt.geom.Point2D;
-import java.util.Optional;
 
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
-
-import lombok.Getter;
-import lombok.Setter;
+import lombok.AllArgsConstructor;
+import lombok.Data;
+import lombok.RequiredArgsConstructor;
 
+@Data
+@RequiredArgsConstructor
+@AllArgsConstructor
 public class OutlineObject {
 
-    @Getter
     private final String title;
-    @Getter
     private final int pageNumber;
-    @Getter
+    private Point2D point;
     private final int treeDepth;
 
-    private Point2D point; // java coordinates, (0, 0) is always top left
-
-    @Getter
-    @Setter
     private boolean found;
 
 
     public OutlineObject(String title, int pageNumber, Point2D point2D, int depth) {
 
-        this.title = title;
-        this.pageNumber = pageNumber;
-        this.treeDepth = depth;
+        this(title, pageNumber, depth);
         this.point = point2D;
     }
 
@@ -39,39 +32,4 @@ public class OutlineObject {
         return "OutlineObject{" + "title='" + title + '\'' + '}';
     }
 
-
-    public Optional<Point2D> getPoint() {
-
-        return Optional.ofNullable(point);
-    }
-
-
-    public boolean isAbove(BoundingBox boundingBox) {
-
-        if (point == null) {
-            return true;
-        }
-        return point.getY() <= boundingBox.getMaxY();
-    }
-
-
-    public double distance(BoundingBox boundingBox) {
-
-        if (point == null) {
-            return 0;
-        }
-        if (boundingBox.getBBox().contains(point)) {
-            return 0;
-        }
-        double deltaX = Math.min(Math.abs(boundingBox.getMinX() - point.getX()), Math.abs(boundingBox.getMaxX() - point.getX()));
-        double deltaY = Math.min(Math.abs(boundingBox.getMinY() - point.getY()), Math.abs(boundingBox.getMaxY() - point.getY()));
-        return Math.sqrt(deltaX * deltaX + deltaY * deltaY);
-    }
-
-
-    public void resetPoint() {
-
-        this.point = null;
-    }
-
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineObjectTree.java

@@ -39,28 +39,4 @@ public class OutlineObjectTree {
         }
     }
 
-    @Override
-    public String toString() {
-
-        StringBuilder sb = new StringBuilder();
-        sb.append("OutlineObjectTree(\n");
-        for (OutlineObjectTreeNode node : rootNodes) {
-            buildString(node, sb, 1);
-        }
-        sb.append(")");
-        return sb.toString();
-    }
-
-    private void buildString(OutlineObjectTreeNode node, StringBuilder sb, int depth) {
-
-        for (int i = 0; i < depth; i++) {
-            sb.append("  ");
-        }
-        sb.append(node.getOutlineObject().getTitle()).append("\n");
-
-        for (OutlineObjectTreeNode child : node.getChildren()) {
-            buildString(child, sb, depth + 1);
-        }
-    }
-
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineValidationService.java

@@ -12,14 +12,12 @@ import org.springframework.stereotype.Service;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 
-import io.micrometer.observation.annotation.Observed;
 import lombok.extern.slf4j.Slf4j;
 
 @Service
 @Slf4j
 public class OutlineValidationService {
 
-    @Observed(name = "OutlineValidationService", contextualName = "create-toc")
     public TableOfContents createToC(List<TextPageBlock> headlines) {
 
         List<TableOfContentItem> mainSections = new ArrayList<>();

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/Cell.java

@@ -35,14 +35,14 @@ public class Cell extends BoundingBox {
 
     public Cell(Point2D topLeft, Point2D bottomRight) {
 
-        this.bBoxPdf = new Rectangle2D.Double(topLeft.getX(), topLeft.getY(), (bottomRight.getX() - topLeft.getX()), (bottomRight.getY() - topLeft.getY()));
-        this.bBox = bBoxPdf;
+        this.bBoxInitialUserSpace = new Rectangle2D.Double(topLeft.getX(), topLeft.getY(), (bottomRight.getX() - topLeft.getX()), (bottomRight.getY() - topLeft.getY()));
+        this.bBox = bBoxInitialUserSpace;
     }
 
 
     public Cell(Rectangle2D bBoxInitialUserSpace, AffineTransform initialUserSpaceToJava) {
 
-        this.bBoxPdf = bBoxInitialUserSpace;
+        this.bBoxInitialUserSpace = bBoxInitialUserSpace;
         this.bBox = initialUserSpaceToJava.createTransformedShape(bBoxInitialUserSpace).getBounds2D();
     }
 
@@ -50,7 +50,7 @@ public class Cell extends BoundingBox {
     public static Cell copy(Cell cell) {
 
         Cell copy = new Cell();
-        copy.bBoxPdf = cell.bBoxPdf;
+        copy.bBoxInitialUserSpace = cell.bBoxInitialUserSpace;
         copy.bBox = cell.bBox;
         return copy;
     }
@@ -87,7 +87,7 @@ public class Cell extends BoundingBox {
 
         }
 
-        return TextNormalizationUtilities.cleanString(sb.toString());
+        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString()).replaceAll("\n", " ").replaceAll(" {2}", " ");
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/CleanRulings.java

@@ -70,7 +70,7 @@ public class CleanRulings {
 
     public boolean lineBetween(BoundingBox a, BoundingBox b) {
 
-        return lineBetween(a.getBBoxPdf(), b.getBBoxPdf());
+        return lineBetween(a.getBBoxInitialUserSpace(), b.getBBoxInitialUserSpace());
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/TablePageBlock.java

@@ -263,8 +263,8 @@ public class TablePageBlock extends AbstractPageBlock {
 
                         cells.stream()
                                 .map(originalCell -> new CellWithIntersection(originalCell,
-                                                                              RectangleTransformations.calculateIntersectedArea(cellFromGridStructure.getBBoxPdf(),
-                                                                                                                                originalCell.getBBoxPdf())))
+                                                                              RectangleTransformations.calculateIntersectedArea(cellFromGridStructure.getBBoxInitialUserSpace(),
+                                                                                                                                originalCell.getBBoxInitialUserSpace())))
                                 .filter(cellWithIntersection -> cellWithIntersection.intersectedArea > 0)
                                 .filter(cellWithIntersection -> cellWithIntersection.originalCell.getArea() > cellWithIntersection.intersectedArea * CELL_AREA_CONTAINED_THRESHOLD)
                                 .max(Comparator.comparing(CellWithIntersection::intersectedArea))

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/FontStyle.java

@@ -1,9 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.text;
-
-
-public enum FontStyle {
-    REGULAR,
-    BOLD,
-    ITALIC,
-    BOLD_ITALIC;
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/RedTextPosition.java

@@ -5,50 +5,64 @@ import java.awt.geom.Rectangle2D;
 
 import org.apache.pdfbox.text.TextPosition;
 
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.TextBoundingBox;
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.FastAtan2;
+import com.fasterxml.jackson.annotation.JsonIgnore;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 
-import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
+import lombok.Builder;
 import lombok.Data;
-import lombok.EqualsAndHashCode;
 import lombok.NoArgsConstructor;
 import lombok.SneakyThrows;
-import lombok.experimental.FieldDefaults;
-import lombok.experimental.SuperBuilder;
 
 @Data
-@SuperBuilder
+@Builder
 @NoArgsConstructor
 @AllArgsConstructor
-@EqualsAndHashCode(callSuper = true)
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class RedTextPosition extends TextBoundingBox {
+public class RedTextPosition extends BoundingBox {
 
     public final static int HEIGHT_PADDING = 2;
 
-    String unicode;
+    private Rectangle2D.Float bBoxDirAdj; // adjusted to text rotation
 
-    // estimated using the TextMatrix in radians
-    float exactDir;
+    @JsonIgnore
+    private int rotation;
 
-    float widthOfSpace;
+    @JsonIgnore
+    private float pageHeight;
 
-    float fontSizeInPt;
+    @JsonIgnore
+    private float pageWidth;
 
-    String fontName;
+    private String unicode;
+
+    @JsonIgnore
+    private float dir;
+
+    // not used in reanalysis
+    @JsonIgnore
+    private float widthOfSpace;
+
+    // not used in reanalysis
+    @JsonIgnore
+    private float fontSizeInPt;
+
+    // not used in reanalysis
+    @JsonIgnore
+    private String fontName;
 
 
     @SneakyThrows
     public static RedTextPosition fromTextPosition(TextPosition textPosition) {
 
         var pos = new RedTextPosition();
+        pos.setRotation(textPosition.getRotation());
+        pos.setPageHeight(textPosition.getPageHeight());
+        pos.setPageWidth(textPosition.getPageWidth());
         pos.setUnicode(textPosition.getUnicode());
+        pos.setDir(textPosition.getDir());
         pos.setWidthOfSpace(textPosition.getWidthOfSpace());
         pos.setFontSizeInPt(textPosition.getFontSizeInPt());
         pos.setFontName(textPosition.getFont().getName());
-        pos.setExactDir((float) FastAtan2.fastAtan2(textPosition.getTextMatrix().getShearY(), textPosition.getTextMatrix().getScaleX()));
-        pos.setDir(TextDirection.fromDegrees(textPosition.getDir()));
 
         //TODO: There is a mismatch in the java coords of the text and the rulings,
         // I guess if we start with the initial user space positions and transform them the same way we do the rulings it would work.
@@ -59,18 +73,18 @@ public class RedTextPosition extends TextBoundingBox {
                                                                  textPosition.getYDirAdj() - textHeight,
                                                                  textPosition.getWidthDirAdj(),
                                                                  textHeight + HEIGHT_PADDING);
-
         pos.setBBoxDirAdj(dirAdjPosition);
 
         AffineTransform affineTransform = getRotationMatrix(TextDirection.fromDegrees(textPosition.getDir()), textPosition.getPageWidth(), textPosition.getPageHeight());
         Rectangle2D bBoxInitialUserSpace = affineTransform.createTransformedShape(dirAdjPosition).getBounds2D();
 
-        pos.setBBoxPdf(bBoxInitialUserSpace); // These are definitely correct
+        pos.setBBoxInitialUserSpace(bBoxInitialUserSpace); // These are definitely correct
 
         return pos;
     }
 
 
+
     private static AffineTransform getRotationMatrix(TextDirection textDirection, float pageWidth, float pageHeight) {
 
         AffineTransform transform = new AffineTransform();
@@ -89,4 +103,32 @@ public class RedTextPosition extends TextBoundingBox {
         return transform;
     }
 
+
+    @JsonIgnore
+    public float getXDirAdj() {
+
+        return this.bBoxDirAdj.x;
+    }
+
+
+    @JsonIgnore
+    public float getYDirAdj() {
+
+        return this.bBoxDirAdj.y;
+    }
+
+
+    @JsonIgnore
+    public float getWidthDirAdj() {
+
+        return this.bBoxDirAdj.width;
+    }
+
+
+    @JsonIgnore
+    public float getHeightDir() {
+
+        return this.bBoxDirAdj.height;
+    }
+
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/SearchableText.java

@@ -2,7 +2,6 @@ package com.knecon.fforesight.service.layoutparser.processor.model.text;
 
 import java.util.ArrayList;
 import java.util.List;
-
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextNormalizationUtilities;
 
 import lombok.Getter;
@@ -39,7 +38,11 @@ public class SearchableText {
             sb.append(word);
             sb.append(' ');
         }
-        return TextNormalizationUtilities.cleanString(sb.toString());
+        String text = sb.toString();
+        text = TextNormalizationUtilities.removeHyphenLineBreaks(text);
+        text = TextNormalizationUtilities.removeLineBreaks(text);
+        text = TextNormalizationUtilities.removeRepeatingWhitespaces(text);
+        return text;
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextDirection.java

@@ -44,15 +44,4 @@ public enum TextDirection {
 
         throw new IllegalArgumentException(String.format("A value of %f is not supported by TextDirection", degrees));
     }
-
-
-    public int getRotation() {
-
-        return switch (this) {
-            case ZERO -> 0;
-            case QUARTER_CIRCLE -> 1;
-            case HALF_CIRCLE -> 2;
-            case THREE_QUARTER_CIRCLE -> 3;
-        };
-    }
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextPageBlock.java

@@ -8,7 +8,6 @@ import com.fasterxml.jackson.annotation.JsonIgnore;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType;
-import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextNormalizationUtilities;
 
 import lombok.AllArgsConstructor;
@@ -27,19 +26,17 @@ public class TextPageBlock extends AbstractPageBlock {
     @Builder.Default
     private List<TextPositionSequence> sequences = new ArrayList<>();
 
-    private Rectangle2D bBoxDirAdj;
-
     private String mostPopularWordFont;
 
     private String mostPopularWordStyle;
 
-    private double mostPopularWordFontSize;
+    private float mostPopularWordFontSize;
 
-    private double mostPopularWordHeight;
+    private float mostPopularWordHeight;
 
-    private double mostPopularWordSpaceWidth;
+    private float mostPopularWordSpaceWidth;
 
-    private double highestFontSize;
+    private float highestFontSize;
 
     private PageBlockType classification;
 
@@ -49,31 +46,39 @@ public class TextPageBlock extends AbstractPageBlock {
     public TextPageBlock(List<TextPositionSequence> sequences) {
 
         this.sequences = sequences;
-        if (!sequences.isEmpty()) {
-            calculateFrequencyCounters();
-        }
+        calculateFrequencyCounters();
         calculateBBox();
     }
 
 
+    @JsonIgnore
     public TextDirection getDir() {
 
         return sequences.get(0).getDir();
     }
 
 
+    @JsonIgnore
+    public float getPageHeight() {
+
+        return sequences.get(0).getPageHeight();
+    }
+
+
+    @JsonIgnore
+    public float getPageWidth() {
+
+        return sequences.get(0).getPageWidth();
+    }
+
+
     private void calculateBBox() {
 
         if (sequences == null) {
             this.bBox = new Rectangle2D.Double();
-            this.bBoxPdf = new Rectangle2D.Double();
-            this.bBoxDirAdj = new Rectangle2D.Double();
+            this.bBoxInitialUserSpace = new Rectangle2D.Double();
             return;
         }
-        this.bBoxDirAdj = sequences.stream()
-                .map(TextPositionSequence::getBBoxDirAdj)
-                .collect(RectangleTransformations.collectBBox());
-
         setToBBoxOfComponents(sequences);
     }
 
@@ -213,7 +218,7 @@ public class TextPageBlock extends AbstractPageBlock {
             previous = word;
         }
 
-        return TextNormalizationUtilities.removeHyphenLinebreaks(sb.toString());
+        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString());
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextPositionSequence.java

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.text;
 
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition.HEIGHT_PADDING;
-
+import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.Locale;
@@ -9,7 +8,8 @@ import java.util.stream.Collectors;
 
 import org.apache.pdfbox.text.TextPosition;
 
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.TextBoundingBox;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.AllArgsConstructor;
 import lombok.Builder;
@@ -23,21 +23,23 @@ import lombok.extern.slf4j.Slf4j;
 @Builder
 @NoArgsConstructor
 @AllArgsConstructor
-@EqualsAndHashCode(onlyExplicitlyIncluded = true, callSuper = true) // needs the bbox to be unique
-public class TextPositionSequence extends TextBoundingBox implements CharSequence {
+@EqualsAndHashCode(onlyExplicitlyIncluded = true, callSuper = false)
+public class TextPositionSequence extends BoundingBox implements CharSequence {
 
-    public static final String STANDARD = "standard";
-    public static final String BOLD_ITALIC = "bold, italic";
-    public static final String BOLD = "bold";
-    public static final String ITALIC = "italic";
+    public static final int HEIGHT_PADDING = 2;
 
     @EqualsAndHashCode.Include
     private int page;
 
     @EqualsAndHashCode.Include
-    @Builder.Default
     private List<RedTextPosition> textPositions = new ArrayList<>();
 
+    private Rectangle2D bBoxDirAdj;
+    @EqualsAndHashCode.Include
+    private TextDirection dir;
+    private int rotation;
+    private float pageHeight;
+    private float pageWidth;
     private boolean isParagraphStart;
     private boolean strikethrough;
     private boolean underline;
@@ -49,6 +51,10 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
                 .map(RedTextPosition::fromTextPosition)
                 .collect(Collectors.toList());
         this.page = pageNumber;
+        this.dir = TextDirection.fromDegrees(textPositions.get(0).getDir());
+        this.rotation = textPositions.get(0).getRotation();
+        this.pageHeight = textPositions.get(0).getPageHeight();
+        this.pageWidth = textPositions.get(0).getPageWidth();
         this.isParagraphStart = isParagraphStart;
         calculateBBox();
     }
@@ -56,6 +62,9 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
 
     private void calculateBBox() {
 
+        this.bBoxDirAdj = textPositions.stream()
+                .map(RedTextPosition::getBBoxDirAdj)
+                .collect(RectangleTransformations.collectBBox());
         setToBBoxOfComponents(getTextPositions());
     }
 
@@ -64,6 +73,10 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
 
         this.textPositions = textPositions;
         this.page = page;
+        this.dir = TextDirection.fromDegrees(textPositions.get(0).getDir());
+        this.rotation = textPositions.get(0).getRotation();
+        this.pageHeight = textPositions.get(0).getPageHeight();
+        this.pageWidth = textPositions.get(0).getPageWidth();
         calculateBBox();
     }
 
@@ -99,6 +112,9 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
         textPositionSequence.textPositions = textPositions.subList(start, end);
         textPositionSequence.page = page;
         textPositionSequence.dir = dir;
+        textPositionSequence.rotation = rotation;
+        textPositionSequence.pageHeight = pageHeight;
+        textPositionSequence.pageWidth = pageWidth;
         textPositionSequence.setToBBoxOfComponents(getTextPositions());
         return textPositionSequence;
     }
@@ -125,6 +141,10 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
 
         this.textPositions.add(textPosition);
         this.page = textPositionSequence.getPage();
+        this.dir = textPositionSequence.getDir();
+        this.rotation = textPositionSequence.getRotation();
+        this.pageHeight = textPositionSequence.getPageHeight();
+        this.pageWidth = textPositionSequence.getPageWidth();
         calculateBBox();
     }
 
@@ -132,18 +152,79 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
     public void add(TextPosition textPosition) {
 
         this.textPositions.add(RedTextPosition.fromTextPosition(textPosition));
+        this.dir = TextDirection.fromDegrees(textPositions.get(0).getDir());
+        this.rotation = textPositions.get(0).getRotation();
+        this.pageHeight = textPositions.get(0).getPageHeight();
+        this.pageWidth = textPositions.get(0).getPageWidth();
         calculateBBox();
     }
 
-    public double getTextHeightNoPadding() {
 
-        return textPositions.get(0).getHeightDirAdj();
+    /**
+     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
+     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
+     *
+     * @return the text direction adjusted minX value
+     */
+
+    public float getMinXDirAdj() {
+
+        return textPositions.get(0).getXDirAdj();
+
     }
 
 
-    public double getTextHeight() {
+    /**
+     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
+     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
+     *
+     * @return the text direction adjusted maxX value
+     */
 
-        return textPositions.get(0).getHeightDirAdj() + HEIGHT_PADDING;
+    public float getMaxXDirAdj() {
+
+        return textPositions.get(textPositions.size() - 1).getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidthDirAdj() + HEIGHT_PADDING;
+
+    }
+
+
+    /**
+     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
+     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
+     *
+     * @return the text direction adjusted minY value. The upper border of the bounding box of the word.
+     */
+
+    public float getMinYDirAdj() {
+
+        return textPositions.get(0).getYDirAdj() - getTextHeight();
+
+    }
+
+
+    /**
+     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
+     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
+     *
+     * @return the text direction adjusted maxY value. The lower border of the bounding box of the word.
+     */
+
+    public float getMaxYDirAdj() {
+
+        return textPositions.get(0).getYDirAdj();
+
+    }
+
+
+    public float getTextHeightNoPadding() {
+
+        return textPositions.get(0).getHeightDir();
+    }
+
+
+    public float getTextHeight() {
+
+        return textPositions.get(0).getHeightDir() + HEIGHT_PADDING;
     }
 
 
@@ -159,18 +240,18 @@ public class TextPositionSequence extends TextBoundingBox implements CharSequenc
     public String getFontStyle() {
 
         if (textPositions.get(0).getFontName() == null) {
-            return STANDARD;
+            return "standard";
         }
         String lowercaseFontName = textPositions.get(0).getFontName().toLowerCase(Locale.ROOT);
 
-        if (lowercaseFontName.contains(BOLD) && lowercaseFontName.contains(ITALIC)) {
-            return BOLD_ITALIC;
-        } else if (lowercaseFontName.contains(BOLD)) {
-            return BOLD;
-        } else if (lowercaseFontName.contains(ITALIC)) {
-            return ITALIC;
+        if (lowercaseFontName.contains("bold") && lowercaseFontName.contains("italic")) {
+            return "bold, italic";
+        } else if (lowercaseFontName.contains("bold")) {
+            return "bold";
+        } else if (lowercaseFontName.contains("italic")) {
+            return "italic";
         } else {
-            return STANDARD;
+            return "standard";
         }
     }
 

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.getRepresentation()));
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
         });
 
         // 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.getRepresentation()));
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
         });
 
         return images;
@@ -56,7 +56,7 @@ public class ImageServiceResponseAdapter {
         classificationPage.getImages().forEach(image -> {
             if (image.getImageType().equals(ImageType.OTHER)) {
                 for (AbstractPageBlock textblock : classificationPage.getTextBlocks()) {
-                    if (image.getPosition().contains(textblock.getBBoxPdf())) {
+                    if (image.getPosition().contains(textblock.getBBoxInitialUserSpace())) {
                         image.setImageType(ImageType.OCR);
                         return;
                     }

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/Classification.java

@@ -14,7 +14,6 @@ import lombok.NoArgsConstructor;
 @AllArgsConstructor
 public class Classification {
 
-    @Builder.Default
     private Map<String, Float> probabilities = new HashMap<>();
     private String label;
 

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

@@ -12,7 +12,6 @@ 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/python_api/model/image/ImageServiceResponse.java

@@ -22,10 +22,8 @@ public class ImageServiceResponse {
 
     @JsonProperty(value = "imageMetadata")
     @JsonAlias("data")
-    @Builder.Default
     private List<ImageMetadata> data = new ArrayList<>();
 
-    @Builder.Default
     private List<ImageMetadata> dataCV = new ArrayList<>();
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/model/table/TableData.java

@@ -15,7 +15,6 @@ import lombok.NoArgsConstructor;
 public class TableData {
 
     private PageInfo pageInfo;
-    @Builder.Default
     private List<TableCells> tableCells = new ArrayList<>();
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/model/table/TableServiceResponse.java

@@ -19,7 +19,7 @@ public class TableServiceResponse {
     private String operation;
     private String targetFileExtension;
     private String responseFileExtension;
-    @Builder.Default
+
     private List<TableData> data = new ArrayList<>();
 
 }

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

@@ -33,7 +33,7 @@ public class BodyTextFrameService {
         for (ClassificationPage page : classificationDocument.getPages()) {
             var updatedBodyTextFrame = getBodyTextFrameFromRulings(page, bodyTextFrame, landscapeBodyTextFrame);
             setBodyTextFrameAdjustedToPage(page, bodyTextFrame, landscapeBodyTextFrame);
-            classificationDocument.getLayoutDebugLayer().addMainBodyVisualization(page.isLandscape() ? landscapeBodyTextFrame : bodyTextFrame, page.getPageNumber());
+            classificationDocument.getVisualizations().addMainBodyVisualization(page.isLandscape() ? landscapeBodyTextFrame : bodyTextFrame, page.getPageNumber());
         }
     }
 

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

@@ -6,6 +6,7 @@ import java.util.List;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.GapInformation;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.AllArgsConstructor;
 import lombok.experimental.UtilityClass;
@@ -37,7 +38,7 @@ public class GapDetectionService {
         for (TextPositionSequence currentTextPosition : sortedTextPositionSequences.subList(1, sortedTextPositionSequences.size())) {
 
             double yDifference = Math.abs(currentTextPosition.getMaxYDirAdj() - previousTextPosition.getMaxYDirAdj());
-            double xGap = Math.abs(previousTextPosition.getMaxXDirAdj() - currentTextPosition.getXDirAdj());
+            double xGap = Math.abs(previousTextPosition.getMaxXDirAdj() - currentTextPosition.getMinXDirAdj());
             Rectangle2D previousTextPositionBBox = toRectangle2D(previousTextPosition);
             Rectangle2D currentTextPositionBBox = toRectangle2D(currentTextPosition);
 

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

@@ -71,7 +71,7 @@ public class LineDetectionService {
 
     private static boolean isXGap(TextPositionSequence currentTextPosition, TextPositionSequence previousTextPosition, double avgTextPositionHeight) {
 
-        return Math.abs(previousTextPosition.getMaxXDirAdj() - currentTextPosition.getXDirAdj()) > (avgTextPositionHeight * X_GAP_FACTOR);
+        return Math.abs(previousTextPosition.getMaxXDirAdj() - currentTextPosition.getMinXDirAdj()) > (avgTextPositionHeight * X_GAP_FACTOR);
     }
 
 
@@ -83,7 +83,7 @@ public class LineDetectionService {
 
     private static boolean isNewLine(TextPositionSequence currentTextPosition, TextPositionSequence previousTextPosition, double avgTextPositionHeight) {
 
-        return Math.abs(previousTextPosition.getYDirAdj() - currentTextPosition.getYDirAdj()) > avgTextPositionHeight;
+        return Math.abs(previousTextPosition.getMinYDirAdj() - currentTextPosition.getMinYDirAdj()) > avgTextPositionHeight;
     }
 
 

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

@@ -34,22 +34,7 @@ public class SimplifiedSectionTextService {
         List<SimplifiedSectionText> simplifiedText = Stream.of(simplifiedMainSectionsList, simplifiedHeadersList, simplifiedFootersList)
                 .flatMap(List::stream)
                 .collect(Collectors.toList());
-        return SimplifiedText.builder()
-                .numberOfPages(document.getNumberOfPages())
-                .sectionTexts(simplifiedText)
-                .mainSectionNumbers(document.getAllSections()
-                                            .stream()
-                                            .map(this::getSectionNumber)
-                                            .toList())
-                .headerSectionNumbers(document.getHeaders()
-                                              .stream()
-                                              .map(this::getSectionNumber)
-                                              .toList())
-                .footerSectionNumbers(document.getFooters()
-                                              .stream()
-                                              .map(this::getSectionNumber)
-                                              .toList())
-                .build();
+        return SimplifiedText.builder().numberOfPages(document.getNumberOfPages()).sectionTexts(simplifiedText).build();
     }
 
 
@@ -64,13 +49,4 @@ public class SimplifiedSectionTextService {
                 .build();
     }
 
-
-    private String getSectionNumber(SemanticNode semanticNode) {
-
-        return semanticNode.getTreeId()
-                .stream()
-                .map(String::valueOf)
-                .collect(Collectors.joining("."));
-    }
-
 }

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

@@ -78,7 +78,7 @@ public class TableExtractionService {
 
             List<Cell> containedCells = new ArrayList<>();
             for (Cell c : cells) {
-                if (c.hasMinimumSize() && area.contains(c.getBBoxPdf())) {
+                if (c.hasMinimumSize() && area.contains(c.getBBoxInitialUserSpace())) {
                     containedCells.add(c);
                 }
             }

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

@@ -31,13 +31,13 @@ public class TextRulingsClassifier {
 
     private static void handleVerticalText(CleanRulings cleanRulings, TextPositionSequence word) {
 
-        float lowerY = (float) (word.getBBoxPdf().getMinY() + TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
-        float upperY = (float) (word.getBBoxPdf().getMaxY() - TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
+        float lowerY = (float) (word.getBBoxInitialUserSpace().getMinY() + TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
+        float upperY = (float) (word.getBBoxInitialUserSpace().getMaxY() - TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
 
-        float strikethroughCenterX = (float) word.getBBoxPdf().getCenterX();
+        float strikethroughCenterX = (float) word.getBBoxInitialUserSpace().getCenterX();
         float strikethroughBoxHeight = (float) ((word.getBBoxDirAdj().getHeight() * STRIKETHROUGH_ZONE) / 2);
 
-        float underlineCenterX = (float) (word.getDir().equals(TextDirection.QUARTER_CIRCLE) ? word.getBBoxPdf().getMaxX() : word.getBBoxPdf().getMinX());
+        float underlineCenterX = (float) (word.getDir().equals(TextDirection.QUARTER_CIRCLE) ? word.getBBoxInitialUserSpace().getMaxX() : word.getBBoxInitialUserSpace().getMinX());
         float underlineBoxHeight = (float) ((word.getBBoxDirAdj().getHeight() * UNDERLINE_ZONE) / 2);
 
         float leftX = Math.min(underlineCenterX - underlineBoxHeight, strikethroughCenterX - strikethroughBoxHeight);
@@ -65,13 +65,13 @@ public class TextRulingsClassifier {
 
     private static void handleHorizontalText(CleanRulings cleanRulings, TextPositionSequence word) {
 
-        float leftX = (float) (word.getBBoxPdf().getMinX() + TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
-        float rightX = (float) (word.getBBoxPdf().getMaxX() - TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
+        float leftX = (float) (word.getBBoxInitialUserSpace().getMinX() + TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
+        float rightX = (float) (word.getBBoxInitialUserSpace().getMaxX() - TEXT_BBOX_THRESHOLD_FACTOR * word.getWidth());
 
-        float strikethroughCenterY = (float) word.getBBoxPdf().getCenterY();
+        float strikethroughCenterY = (float) word.getBBoxInitialUserSpace().getCenterY();
         float strikethroughBoxHeight = (float) ((word.getBBoxDirAdj().getHeight() * STRIKETHROUGH_ZONE) / 2);
 
-        float underlineCenterY = (float) (word.getDir().equals(TextDirection.ZERO) ? word.getBBoxPdf().getMinY() : word.getBBoxPdf().getMaxY());
+        float underlineCenterY = (float) (word.getDir().equals(TextDirection.ZERO) ? word.getBBoxInitialUserSpace().getMinY() : word.getBBoxInitialUserSpace().getMaxY());
         float underlineBoxHeight = (float) ((word.getBBoxDirAdj().getHeight() * UNDERLINE_ZONE) / 2);
 
         float lowerY = Math.min(underlineCenterY - underlineBoxHeight, strikethroughCenterY - strikethroughBoxHeight);

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/BlockificationPostprocessingService.java

@@ -2,13 +2,14 @@ package com.knecon.fforesight.service.layoutparser.processor.services.blockifica
 
 import static com.knecon.fforesight.service.layoutparser.processor.services.blockification.DocstrumBlockificationService.buildTextBlock;
 
+import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.ListIterator;
 import java.util.Locale;
+import java.util.function.Function;
 
 import org.apache.commons.lang3.StringUtils;
-import org.apache.commons.text.similarity.LevenshteinDistance;
 import org.springframework.stereotype.Service;
 
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
@@ -18,13 +19,23 @@ import com.knecon.fforesight.service.layoutparser.processor.model.SectionIdentif
 import com.knecon.fforesight.service.layoutparser.processor.model.outline.OutlineObject;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.services.factory.SearchTextWithTextPositionFactory;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.Data;
 
 @Service
 public class BlockificationPostprocessingService {
 
-    private static final float STRING_SIMILARITY_THRESHOLD = 0.1f;
+    private static final float BLOCK_TO_OUTLINE_DISTANCE_THRESHOLD = 5.0f;
+
+    private static final Function<TextPageBlock, Rectangle2D> blockToBoundingBox = (abstractPageBlock) -> abstractPageBlock.getSequences()
+            .stream()
+            .map(textPositionSequence -> textPositionSequence.getTextPositions()
+                    .stream()
+                    .map(tp -> SearchTextWithTextPositionFactory.mapRedTextPositionToInitialUserSpace(tp, textPositionSequence))
+                    .collect(RectangleTransformations.collectBBox()))
+            .collect(RectangleTransformations.collectBBox());
 
 
     public OutlineObject sanitizeOutlineBlocks(ClassificationPage classificationPage, OutlineObject notFoundOutlineObject) {
@@ -35,18 +46,20 @@ public class BlockificationPostprocessingService {
             return null;
         }
 
+        float pageHeight = classificationPage.getPageHeight();
+
         ListIterator<OutlineObject> outlineObjectListIterator = outlineObjects.listIterator();
 
         if (notFoundOutlineObject != null) {
             OutlineProcessionContext notFoundOutlineObjectProcessionContext = new OutlineProcessionContext(notFoundOutlineObject);
-            processTextBlocks(getTextPageBlocks(classificationPage), notFoundOutlineObjectProcessionContext);
+            processTextBlocks(getTextPageBlocks(classificationPage), pageHeight, notFoundOutlineObjectProcessionContext);
 
             OutlineObject firstOutlineObject = null;
             OutlineProcessionContext firstOutlineObjectProcessionContext = null;
             if (outlineObjectListIterator.hasNext()) {
                 firstOutlineObject = outlineObjectListIterator.next();
                 firstOutlineObjectProcessionContext = new OutlineProcessionContext(firstOutlineObject);
-                processTextBlocks(getTextPageBlocks(classificationPage), firstOutlineObjectProcessionContext);
+                processTextBlocks(getTextPageBlocks(classificationPage), pageHeight, firstOutlineObjectProcessionContext);
             }
 
             if (!contextsOverlap(notFoundOutlineObjectProcessionContext, firstOutlineObjectProcessionContext)) {
@@ -55,7 +68,7 @@ public class BlockificationPostprocessingService {
             if (firstOutlineObject != null) {
                 // re-create the context for the updated blocks
                 firstOutlineObjectProcessionContext = new OutlineProcessionContext(firstOutlineObject);
-                processTextBlocks(getTextPageBlocks(classificationPage), firstOutlineObjectProcessionContext);
+                processTextBlocks(getTextPageBlocks(classificationPage), pageHeight, firstOutlineObjectProcessionContext);
                 firstOutlineObject.setFound(selectMatch(classificationPage, firstOutlineObjectProcessionContext));
             }
 
@@ -63,7 +76,7 @@ public class BlockificationPostprocessingService {
 
         outlineObjectListIterator.forEachRemaining(outlineObject -> {
             OutlineProcessionContext outlineObjectProcessionContext = new OutlineProcessionContext(outlineObject);
-            processTextBlocks(getTextPageBlocks(classificationPage), outlineObjectProcessionContext);
+            processTextBlocks(getTextPageBlocks(classificationPage), pageHeight, outlineObjectProcessionContext);
             outlineObject.setFound(selectMatch(classificationPage, outlineObjectProcessionContext));
         });
 
@@ -103,7 +116,8 @@ public class BlockificationPostprocessingService {
 
         double maxYFirst = blocksOfFirstOutline.stream()
                 .mapToDouble(TextPageBlock::getPdfMaxY)
-                .max().orElse(Double.NEGATIVE_INFINITY);
+                .max()
+                .orElse(Double.NEGATIVE_INFINITY);
 
         return blocksOfNotFoundOutline.stream()
                 .mapToDouble(TextPageBlock::getPdfMaxY)
@@ -125,13 +139,13 @@ public class BlockificationPostprocessingService {
     }
 
 
-    private void processTextBlocks(List<TextPageBlock> textBlocks, OutlineProcessionContext context) {
+    private void processTextBlocks(List<TextPageBlock> textBlocks, float pageHeight, OutlineProcessionContext context) {
 
         OutlineObject outlineObject = context.getOutlineObject();
         ListIterator<TextPageBlock> iterator = textBlocks.listIterator();
         while (iterator.hasNext()) {
             TextPageBlock pageBlock = iterator.next();
-            if (outlineObject.isAbove(pageBlock)) {
+            if (pageHeight - outlineObject.getPoint().getY() - BLOCK_TO_OUTLINE_DISTANCE_THRESHOLD <= pageBlock.getMaxY()) {
                 break;
             }
         }
@@ -176,7 +190,8 @@ public class BlockificationPostprocessingService {
             for (List<TextPageBlock> combination : combinations) {
                 double averageDistance = combination.stream()
                         .map(block -> calculateDistance(outlineObject, block))
-                        .mapToDouble(Double::doubleValue).average().orElse(Double.MAX_VALUE);
+                        .mapToDouble(Double::doubleValue).average()
+                        .orElse(Double.MAX_VALUE);
                 if (distanceToBestMergeCandidates > averageDistance) {
                     distanceToBestMergeCandidates = averageDistance;
                     bestMergeCandidateCombination = combination;
@@ -258,10 +273,6 @@ public class BlockificationPostprocessingService {
         List<TextPositionSequence> postSequence = new ArrayList<>();
         StringBuilder currentSequence = new StringBuilder();
 
-        if (target.isBlank()) {
-            return new WordSequenceResult();
-        }
-
         for (TextPositionSequence sequence : textPositionSequences) {
 
             currentSequence.append(sanitizeString(sequence.toString()));
@@ -321,7 +332,7 @@ public class BlockificationPostprocessingService {
         TextPositionSequence in = null;
         TextPositionSequence out;
 
-        String currentSequence = sequence.toString().toLowerCase(Locale.ROOT);
+        String currentSequence = sequence.toString();
         int index = currentSequence.indexOf(toRemove);
         int endIndex = index + toRemove.length();
 
@@ -405,7 +416,9 @@ public class BlockificationPostprocessingService {
 
     private double calculateDistance(OutlineObject outlineObject, TextPageBlock pageBlock) {
 
-        return outlineObject.distance(pageBlock);
+        double deltaX = outlineObject.getPoint().getX() - pageBlock.getMinX();
+        double deltaY = pageBlock.getPageHeight() - outlineObject.getPoint().getY() - pageBlock.getMinY();
+        return Math.sqrt(deltaX * deltaX + deltaY * deltaY);
     }
 
 
@@ -422,13 +435,6 @@ public class BlockificationPostprocessingService {
         String blockText = sanitizeString(pageBlock.getText());
         String outlineTitle = sanitizeString(outlineObject.getTitle());
 
-        int threshold = (int) (Math.min(blockText.length(), outlineTitle.length()) * STRING_SIMILARITY_THRESHOLD) + 1;
-        int distance = new LevenshteinDistance(threshold).apply(blockText, outlineTitle);
-        if (distance >= 0 && distance < threshold) {
-            context.directMatch = pageBlock;
-            return true;
-        }
-
         boolean blockTextContainsOutlineTitle = blockText.contains(outlineTitle);
         boolean outlineTitleContainsBlockText = outlineTitle.contains(blockText);
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/DocstrumBlockificationService.java

@@ -1,6 +1,7 @@
 package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
 
 import java.util.ArrayList;
+import java.util.Comparator;
 import java.util.List;
 import java.util.ListIterator;
 
@@ -9,13 +10,14 @@ import org.springframework.stereotype.Service;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.DocstrumSegmentationService;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.DoubleUtils;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
-import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutparsingVisualizations;
 
 import lombok.RequiredArgsConstructor;
 
@@ -33,12 +35,12 @@ public class DocstrumBlockificationService {
     public ClassificationPage blockify(List<TextPositionSequence> textPositions,
                                        CleanRulings rulings,
                                        boolean xyOrder,
-                                       LayoutDebugLayer visualizations,
+                                       LayoutparsingVisualizations visualizations,
                                        LayoutParsingType layoutParsingType) {
 
         CleanRulings usedRulings = rulings.withoutTextRulings();
 
-        List<Zone> zones = docstrumSegmentationService.segmentPage(textPositions, xyOrder, usedRulings, visualizations);
+        var zones = docstrumSegmentationService.segmentPage(textPositions, xyOrder, usedRulings, visualizations);
 
         if (!textPositions.isEmpty()) {
             visualizations.addZoneVisualizations(zones, textPositions.get(0).getPage());
@@ -46,7 +48,11 @@ public class DocstrumBlockificationService {
             visualizations.addCharactersWithNeighbours(zones, textPositions.get(0).getPage());
         }
 
-        var pageBlocks = toAbstractPageBlocks(zones);
+        var pageBlocks = toAbstractPageBlocks(zones, xyOrder, usedRulings);
+
+        if (xyOrder) {
+            sortPageBlocksXThenY(pageBlocks);
+        }
 
         var classificationPage = new ClassificationPage(pageBlocks);
         classificationPage.setCleanRulings(rulings);
@@ -67,7 +73,21 @@ public class DocstrumBlockificationService {
     }
 
 
-    private List<AbstractPageBlock> toAbstractPageBlocks(List<Zone> zones) {
+    private static void sortPageBlocksXThenY(List<AbstractPageBlock> pageBlocks) {
+
+        pageBlocks.sort(Comparator.comparing(AbstractPageBlock::getMinY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                                .thenComparing(AbstractPageBlock::getMinX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
+        pageBlocks.sort(new Comparator<AbstractPageBlock>() {
+            @Override
+            public int compare(AbstractPageBlock o1, AbstractPageBlock o2) {
+
+                return Math.abs(o1.getMinY() - o2.getMinY()) < 5 && o1.getMinX() < o2.getMinX() == true ? -1 : 0;
+            }
+        });
+    }
+
+
+    private List<AbstractPageBlock> toAbstractPageBlocks(List<Zone> zones, boolean xyOrder, CleanRulings usedRulings) {
 
         List<AbstractPageBlock> abstractPageBlocks = new ArrayList<>();
         zones.forEach(zone -> {
@@ -276,10 +296,6 @@ public class DocstrumBlockificationService {
                     continue;
                 }
 
-//                if (!current.getMostPopularWordStyle().equals(inner.getMostPopularWordStyle())) {
-//                    continue;
-//                }
-
                 if (current.getDir() == inner.getDir() && current.intersects(inner, yThreshold, xThreshold)) {
 
                     boolean toDuplicate = current.isToDuplicate() || inner.isToDuplicate();

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/DocuMineBlockificationService.java

@@ -1,24 +1,25 @@
 package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
 
+import static java.util.stream.Collectors.toSet;
+
 import java.util.ArrayList;
+import java.util.Comparator;
 import java.util.List;
-import java.util.ListIterator;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
 
 import org.springframework.stereotype.Service;
 
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.Orientation;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
-import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
 
-@SuppressWarnings("all")
 @Service
 public class DocuMineBlockificationService {
 
@@ -43,33 +44,31 @@ public class DocuMineBlockificationService {
 
         CleanRulings usedRulings = cleanRulings.withoutTextRulings();
 
-        double minX = 1000;
-        double maxX = 0;
-        double minY = 1000;
-        double maxY = 0;
+        float minX = 1000;
+        float maxX = 0;
+        float minY = 1000;
+        float maxY = 0;
         TextPositionSequence prev = null;
 
         boolean wasSplitted = false;
-        Double splitX1 = null;
+        Float splitX1 = null;
         for (TextPositionSequence word : textPositions) {
 
-            boolean lineSeparation = prev != null && word.getYDirAdj() - prev.getMaxYDirAdj() > Math.min(word.getHeight(), prev.getHeight()) * 1.1;
-            boolean startFromTop = prev != null && word.getYDirAdj() < prev.getYDirAdj() - prev.getTextHeight();
-            boolean splitByX = prev != null && maxX + 50 < word.getXDirAdj() && prev.getYDirAdj() == word.getYDirAdj();
-            boolean negativeXGap = prev != null && word.getXDirAdj() - minX < -5;
-            boolean newLineAfterSplit = prev != null && word.getYDirAdj() != prev.getYDirAdj() && wasSplitted && splitX1 != word.getXDirAdj();
+            boolean lineSeparation = prev != null && word.getMinYDirAdj() - maxY > Math.min(word.getHeight(), prev.getHeight()) * 1.25;
+            boolean startFromTop = prev != null && word.getMinYDirAdj() < prev.getMinYDirAdj() - prev.getTextHeight();
+            boolean splitByX = prev != null && maxX + 50 < word.getMinXDirAdj() && prev.getMinYDirAdj() == word.getMinYDirAdj();
+            boolean negativeXGap = prev != null && word.getMinXDirAdj() - minX < -5;
+            boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj();
             boolean isSplitByRuling = prev != null && usedRulings.lineBetween(prev, word);
             boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
-            boolean splitByOtherFontAndOtherY = prev != null && Math.abs(prev.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight() * 0.2 //
-                                                && (word.getFontStyle().contains("bold") && !prev.getFontStyle().contains("bold")
-                                                    //
-                                                    || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold")
-                                                    || Math.abs(prev.getFontSize() - word.getFontSize()) >= 1
-                                                    || Math.abs(word.getTextHeight() - prev.getTextHeight()) > 0.8);
+            boolean splitByOtherFontAndOtherY = prev != null && Math.abs(prev.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight() * 0.2 && (word.getFontStyle()
+                                                                                                                                                               .contains("bold")
+                                                                                                                                                       && !prev.getFontStyle()
+                    .contains("bold") || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold"));
 
             Matcher matcher = pattern.matcher(chunkWords.stream()
                                                       .collect(Collectors.joining(" ")).toString());
-            boolean startsOnSameX = Math.abs(minX - word.getXDirAdj()) < 5 && matcher.matches();
+            boolean startsOnSameX = Math.abs(minX - word.getMinXDirAdj()) < 5 && matcher.matches();
 
             if (prev != null && (lineSeparation || startFromTop || splitByDir || isSplitByRuling || splitByOtherFontAndOtherY || negativeXGap || startsOnSameX)) {
 
@@ -85,7 +84,7 @@ public class DocuMineBlockificationService {
                 if (splitByX && !isSplitByRuling) {
                     wasSplitted = true;
                     cb1.setOrientation(Orientation.LEFT);
-                    splitX1 = word.getXDirAdj();
+                    splitX1 = word.getMinXDirAdj();
                 } else if (newLineAfterSplit && !isSplitByRuling) {
                     wasSplitted = false;
                     cb1.setOrientation(Orientation.RIGHT);
@@ -108,14 +107,14 @@ public class DocuMineBlockificationService {
             chunkWords.add(word);
 
             prev = word;
-            if (word.getXDirAdj() < minX) {
-                minX = word.getXDirAdj();
+            if (word.getMinXDirAdj() < minX) {
+                minX = word.getMinXDirAdj();
             }
             if (word.getMaxXDirAdj() > maxX) {
                 maxX = word.getMaxXDirAdj();
             }
-            if (word.getYDirAdj() < minY) {
-                minY = word.getYDirAdj();
+            if (word.getMinYDirAdj() < minY) {
+                minY = word.getMinYDirAdj();
             }
             if (word.getMaxYDirAdj() > maxY) {
                 maxY = word.getMaxYDirAdj();
@@ -128,76 +127,6 @@ public class DocuMineBlockificationService {
     }
 
 
-    public void mergeblocks(ClassificationPage page, CleanRulings usedRulings, float xThreshold, float yThreshold) {
-
-        var blocks = page.getTextBlocks();
-        ListIterator<AbstractPageBlock> itty = blocks.listIterator();
-        while (itty.hasNext()) {
-            AbstractPageBlock block = itty.next();
-            if (block == null) {
-                continue;
-            }
-            if (block instanceof TablePageBlock) {
-                continue;
-            }
-
-            TextPageBlock current = (TextPageBlock) block;
-
-            for (int i = 0; i < blocks.size(); i++) {
-
-                AbstractPageBlock abstractPageBlock = blocks.get(i);
-                if (abstractPageBlock == null) {
-                    continue;
-                }
-                if (abstractPageBlock == current) {
-                    continue;
-                }
-                if (abstractPageBlock instanceof TablePageBlock) {
-                    continue;
-                }
-
-                if (isHeadlineFromOutline(current) || isHeadlineFromOutline(abstractPageBlock)) {
-                    continue;
-                }
-
-                TextPageBlock inner = (TextPageBlock) abstractPageBlock;
-
-                if (usedRulings.lineBetween(current, blocks.get(i))) {
-                    continue;
-                }
-
-                if (current.getDir() == inner.getDir() && current.intersects(inner, yThreshold, xThreshold) && (current.getClassification() == null || current.getClassification()
-                        .equals(inner.getClassification()))) {
-
-                    boolean toDuplicate = current.isToDuplicate() || inner.isToDuplicate();
-                    current.getSequences().addAll(inner.getSequences());
-                    current = buildTextBlock(current.getSequences(), 0);
-                    current.setClassification(inner.getClassification());
-                    current.setToDuplicate(toDuplicate);
-                    blocks.set(i, null);
-                    itty.set(current);
-                }
-            }
-        }
-        var blocksIterator = blocks.iterator();
-        while (blocksIterator.hasNext()) {
-            if (blocksIterator.next() == null) {
-                blocksIterator.remove();
-            }
-        }
-    }
-
-
-    private boolean isHeadlineFromOutline(AbstractPageBlock abstractPageBlock) {
-
-        return abstractPageBlock.getEngines().contains(LayoutEngine.OUTLINE) && abstractPageBlock.getClassification() != null && abstractPageBlock.getClassification().isHeadline();
-    }
-
-
-    public static TextPageBlock buildTextBlock(List<TextPositionSequence> wordBlockList, int indexOnPage) {
-
-        return new TextPageBlock(wordBlockList);
-    }
 
 }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/RedactManagerBlockificationService.java

@@ -1,6 +1,9 @@
 package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
 
+import static java.util.stream.Collectors.toSet;
+
 import java.util.ArrayList;
+import java.util.Comparator;
 import java.util.Iterator;
 import java.util.List;
 
@@ -8,11 +11,13 @@ import org.springframework.stereotype.Service;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.Orientation;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
-import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutparsingVisualizations;
 
 @SuppressWarnings("all")
 @Service
@@ -30,7 +35,7 @@ public class RedactManagerBlockificationService {
      * @param visualizations
      * @return Page object that contains the Textblock and text statistics.
      */
-    public ClassificationPage blockify(List<TextPositionSequence> textPositions, CleanRulings cleanRulings, LayoutDebugLayer visualizations) {
+    public ClassificationPage blockify(List<TextPositionSequence> textPositions, CleanRulings cleanRulings, LayoutparsingVisualizations visualizations) {
 
         CleanRulings usedRulings = cleanRulings.withoutTextRulings();
 
@@ -38,18 +43,18 @@ public class RedactManagerBlockificationService {
         List<TextPositionSequence> chunkWords = new ArrayList<>();
         List<AbstractPageBlock> chunkBlockList = new ArrayList<>();
 
-        double minX = 1000, maxX = 0, minY = 1000, maxY = 0;
+        float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
         TextPositionSequence prev = null;
 
         boolean wasSplitted = false;
-        Double splitX1 = null;
+        Float splitX1 = null;
         for (TextPositionSequence word : textPositions) {
 
-            boolean lineSeparation = word.getYDirAdj() - maxY > word.getHeight() * 1.25;
-            boolean startFromTop = prev != null && word.getYDirAdj() < prev.getYDirAdj() - prev.getTextHeight();
-            boolean splitByX = prev != null && maxX + 50 < word.getXDirAdj() && prev.getYDirAdj() == word.getYDirAdj();
-            boolean xIsBeforeFirstX = prev != null && word.getXDirAdj() < minX;
-            boolean newLineAfterSplit = prev != null && word.getYDirAdj() != prev.getYDirAdj() && wasSplitted && splitX1 != word.getXDirAdj();
+            boolean lineSeparation = word.getMinYDirAdj() - maxY > word.getHeight() * 1.25;
+            boolean startFromTop = prev != null && word.getMinYDirAdj() < prev.getMinYDirAdj() - prev.getTextHeight();
+            boolean splitByX = prev != null && maxX + 50 < word.getMinXDirAdj() && prev.getMinYDirAdj() == word.getMinYDirAdj();
+            boolean xIsBeforeFirstX = prev != null && word.getMinXDirAdj() < minX;
+            boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj();
             boolean isSplitByRuling = prev != null && usedRulings.lineBetween(prev, word);
             boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
 
@@ -69,7 +74,7 @@ public class RedactManagerBlockificationService {
                 if (splitByX && !isSplitByRuling) {
                     wasSplitted = true;
                     cb1.setOrientation(Orientation.LEFT);
-                    splitX1 = word.getXDirAdj();
+                    splitX1 = word.getMinXDirAdj();
                 } else if (newLineAfterSplit && !isSplitByRuling) {
                     wasSplitted = false;
                     cb1.setOrientation(Orientation.RIGHT);
@@ -92,14 +97,14 @@ public class RedactManagerBlockificationService {
             chunkWords.add(word);
 
             prev = word;
-            if (word.getXDirAdj() < minX) {
-                minX = word.getXDirAdj();
+            if (word.getMinXDirAdj() < minX) {
+                minX = word.getMinXDirAdj();
             }
             if (word.getMaxXDirAdj() > maxX) {
                 maxX = word.getMaxXDirAdj();
             }
-            if (word.getYDirAdj() < minY) {
-                minY = word.getYDirAdj();
+            if (word.getMinYDirAdj() < minY) {
+                minY = word.getMinYDirAdj();
             }
             if (word.getMaxYDirAdj() > maxY) {
                 maxY = word.getMaxYDirAdj();

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/classification/ClarifyndClassificationService.java

@@ -23,7 +23,7 @@ public class ClarifyndClassificationService {
 
     public void classifyDocument(ClassificationDocument document) {
 
-        List<Double> headlineFontSizes = document.getFontSizeCounter().getHigherThanMostPopular();
+        List<Float> headlineFontSizes = document.getFontSizeCounter().getHighterThanMostPopular();
 
         log.debug("Document FontSize counters are: {}", document.getFontSizeCounter().getCountPerValue());
 
@@ -35,7 +35,7 @@ public class ClarifyndClassificationService {
     }
 
 
-    private void classifyPage(HeadlineClassificationService headlineClassificationService, ClassificationPage page, ClassificationDocument document, List<Double> headlineFontSizes) {
+    private void classifyPage(HeadlineClassificationService headlineClassificationService, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
 
         for (AbstractPageBlock textBlock : page.getTextBlocks()) {
             if (textBlock instanceof TextPageBlock) {
@@ -45,7 +45,7 @@ public class ClarifyndClassificationService {
     }
 
 
-    private void classifyBlock(HeadlineClassificationService headlineClassificationService, TextPageBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Double> headlineFontSizes) {
+    private void classifyBlock(HeadlineClassificationService headlineClassificationService, TextPageBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
 
         var bodyTextFrame = page.getBodyTextFrame();
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/classification/DocuMineClassificationService.java

@@ -31,7 +31,7 @@ public class DocuMineClassificationService {
 
     public void classifyDocument(ClassificationDocument document) {
 
-        List<Double> headlineFontSizes = document.getFontSizeCounter().getHigherThanMostPopular();
+        List<Float> headlineFontSizes = document.getFontSizeCounter().getHighterThanMostPopular();
 
         log.debug("Document FontSize counters are: {}", document.getFontSizeCounter().getCountPerValue());
 
@@ -46,7 +46,7 @@ public class DocuMineClassificationService {
     private void classifyPage(HeadlineClassificationService headlineClassificationService,
                               ClassificationPage page,
                               ClassificationDocument document,
-                              List<Double> headlineFontSizes) {
+                              List<Float> headlineFontSizes) {
 
         for (AbstractPageBlock textBlock : page.getTextBlocks()) {
             if (textBlock instanceof TextPageBlock) {
@@ -60,7 +60,7 @@ public class DocuMineClassificationService {
                                TextPageBlock textBlock,
                                ClassificationPage page,
                                ClassificationDocument document,
-                               List<Double> headlineFontSizes) {
+                               List<Float> headlineFontSizes) {
 
         log.debug("headlineFontSizes: {}", headlineFontSizes);
         var bodyTextFrame = page.getBodyTextFrame();
@@ -74,7 +74,7 @@ public class DocuMineClassificationService {
             return;
         }
         if (document.getFontSizeCounter().getMostPopular() == null) {
-            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            textBlock.setClassification(PageBlockType.OTHER);
             return;
         }
         if (MarkedContentUtils.intersects(textBlock, page.getMarkedContentBboxPerType(), MarkedContentUtils.HEADER) //
@@ -108,10 +108,7 @@ public class DocuMineClassificationService {
                        && Character.isDigit(textBlock.toString().charAt(0))
                        && atLeast3Matcher.reset().find()
                        && !textBlock.toString().contains(":") //
-                       || textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT))
-                          && atLeast3Matcher.reset().find()
-                          && !textBlock.toString().contains(":")
-                          && !textBlock.toString().startsWith("(")//
+                       || textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)) && atLeast3Matcher.reset().find() && !textBlock.toString().contains(":") //
                        || textBlock.toString().startsWith("APPENDIX") //
                        || textBlock.toString().startsWith("FIGURE") //
                        || textBlock.toString().startsWith("Continued TABLE") //
@@ -146,9 +143,9 @@ public class DocuMineClassificationService {
                    && PositionUtils.getApproxLineCount(textBlock) < 2.9) {
             textBlock.setClassification(PageBlockType.PARAGRAPH_ITALIC);
         } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
-            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            textBlock.setClassification(PageBlockType.PARAGRAPH_UNKNOWN);
         } else {
-            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            textBlock.setClassification(PageBlockType.OTHER);
         }
     }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/classification/RedactManagerClassificationService.java

@@ -25,7 +25,7 @@ public class RedactManagerClassificationService {
 
     public void classifyDocument(ClassificationDocument document) {
 
-        List<Double> headlineFontSizes = document.getFontSizeCounter().getHigherThanMostPopular();
+        List<Float> headlineFontSizes = document.getFontSizeCounter().getHighterThanMostPopular();
 
         log.debug("Document FontSize counters are: {}", document.getFontSizeCounter().getCountPerValue());
 
@@ -37,7 +37,7 @@ public class RedactManagerClassificationService {
     }
 
 
-    private void classifyPage(HeadlineClassificationService headlineClassificationService, ClassificationPage page, ClassificationDocument document, List<Double> headlineFontSizes) {
+    private void classifyPage(HeadlineClassificationService headlineClassificationService, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
 
         for (AbstractPageBlock textBlock : page.getTextBlocks()) {
             if (textBlock instanceof TextPageBlock) {
@@ -47,7 +47,7 @@ public class RedactManagerClassificationService {
     }
 
 
-    private void classifyBlock(HeadlineClassificationService headlineClassificationService, TextPageBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Double> headlineFontSizes) {
+    private void classifyBlock(HeadlineClassificationService headlineClassificationService, TextPageBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
 
         var bodyTextFrame = page.getBodyTextFrame();
 
@@ -56,7 +56,7 @@ public class RedactManagerClassificationService {
             return;
         }
         if (document.getFontSizeCounter().getMostPopular() == null) {
-            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            textBlock.setClassification(PageBlockType.OTHER);
             return;
         }
 
@@ -129,7 +129,7 @@ public class RedactManagerClassificationService {
         } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
             textBlock.setClassification(PageBlockType.PARAGRAPH_UNKNOWN);
         } else {
-            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            textBlock.setClassification(PageBlockType.OTHER);
         }
     }
 

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

@@ -6,7 +6,6 @@ import static java.util.stream.Collectors.toList;
 
 import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
-import java.util.Collection;
 import java.util.HashMap;
 import java.util.LinkedList;
 import java.util.List;
@@ -16,7 +15,6 @@ import java.util.Optional;
 import java.util.Set;
 import java.util.stream.Collectors;
 
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
@@ -24,7 +22,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.Classification
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationHeader;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.AbstractSemanticNode;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.DuplicatedParagraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Footer;
@@ -34,9 +31,8 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.He
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Image;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.AbstractSemanticNode;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
 import com.knecon.fforesight.service.layoutparser.processor.model.outline.TableOfContentItem;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
@@ -49,17 +45,15 @@ 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) {
 
         Document documentGraph = new Document();
 
-        documentGraph.setLayoutDebugLayer(document.getLayoutDebugLayer());
+        documentGraph.setVisualizations(document.getVisualizations());
 
         Context context = new Context(documentGraph);
 
@@ -72,25 +66,10 @@ public class DocumentGraphFactory {
         documentGraph.setPages(context.pages.keySet());
         documentGraph.setDocumentTree(context.documentTree);
         documentGraph.setTextBlock(documentGraph.getTextBlock());
-        addTextBlocksToPages(documentGraph);
-
         return documentGraph;
     }
 
 
-    private void addTextBlocksToPages(Document documentGraph) {
-
-        documentGraph.streamAllSubNodes()
-                .filter(SemanticNode::isLeaf)
-                .filter(node -> !node.getType().equals(NodeType.HEADER))
-                .filter(node -> !node.getType().equals(NodeType.FOOTER))
-                .map(SemanticNode::getTextBlock)
-                .map(TextBlock::getAtomicTextBlocks)
-                .flatMap(Collection::stream)
-                .forEach(atb -> atb.getPage().getTextBlocksOnPage().add(atb));
-    }
-
-
     private void addSections(LayoutParsingType layoutParsingType, ClassificationDocument classificationDocument, Context context, Document document) {
 
         for (TableOfContentItem tocItem : classificationDocument.getTableOfContents()) {
@@ -124,17 +103,19 @@ public class DocumentGraphFactory {
             node = Paragraph.builder().documentTree(context.getDocumentTree()).build();
         }
 
+        page.getMainBody().add(node);
+
         List<TextPageBlock> textBlocks = new ArrayList<>();
         textBlocks.add(originalTextBlock);
         textBlocks.addAll(textBlocksToMerge);
 
-        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock2(TextPositionOperations.mergeAndSort(textBlocks), node, context, page);
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), node, context, page);
 
         if (node instanceof DuplicatedParagraph duplicatedParagraph) {
-            AtomicTextBlock unsortedTextBlock = context.textBlockFactory.buildAtomicTextBlock2(textBlocks.stream()
-                                                                                                       .flatMap(tb -> tb.getSequences()
-                                                                                                               .stream())
-                                                                                                       .collect(Collectors.toList()), node, context, page);
+            AtomicTextBlock unsortedTextBlock = context.textBlockFactory.buildAtomicTextBlock(textBlocks.stream()
+                                                                                                      .flatMap(tb -> tb.getSequences()
+                                                                                                              .stream())
+                                                                                                      .collect(Collectors.toList()), node, context, page);
             duplicatedParagraph.setUnsortedLeafTextBlock(unsortedTextBlock);
         }
 
@@ -158,15 +139,16 @@ public class DocumentGraphFactory {
 
         Rectangle2D position = image.getPosition();
         Page page = context.getPage(image.getPage());
-        return Image.builder()
+        Image imageNode = Image.builder()
                 .id(IdBuilder.buildId(Set.of(page), List.of(position)))
                 .imageType(image.getImageType())
                 .position(position)
                 .transparent(image.isHasTransparency())
                 .page(page)
-                .representationHash(image.getRepresentation())
                 .documentTree(context.getDocumentTree())
                 .build();
+        page.getMainBody().add(imageNode);
+        return imageNode;
     }
 
 
@@ -206,7 +188,7 @@ public class DocumentGraphFactory {
 
         Page page = context.getPage(textBlocks.get(0).getPage());
         Footer footer = Footer.builder().documentTree(context.getDocumentTree()).build();
-        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock2(TextPositionOperations.merge(textBlocks), footer, context, page);
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeTextPositionSequence(textBlocks), footer, context, page);
         List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(footer);
         footer.setTreeId(tocId);
         footer.setLeafTextBlock(textBlock);
@@ -218,7 +200,7 @@ public class DocumentGraphFactory {
 
         Page page = context.getPage(textBlocks.get(0).getPage());
         Header header = Header.builder().documentTree(context.getDocumentTree()).build();
-        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.merge(textBlocks), header, 0, page);
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeTextPositionSequence(textBlocks), header, 0, page);
         List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(header);
         header.setTreeId(tocId);
         header.setLeafTextBlock(textBlock);
@@ -295,8 +277,7 @@ public class DocumentGraphFactory {
             return pages.keySet()
                     .stream()
                     .filter(page -> page.getNumber() == pageIndex)
-                    .findFirst()
-                    .orElseThrow(() -> new NoSuchElementException(format("ClassificationPage with number %d not found", pageIndex)));
+                    .findFirst().orElseThrow(() -> new NoSuchElementException(format("ClassificationPage with number %d not found", pageIndex)));
         }
 
     }

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

@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.services.factory;
 
-import static com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition.HEIGHT_PADDING;
-
+import java.awt.geom.AffineTransform;
 import java.awt.geom.Rectangle2D;
 import java.util.Collection;
 import java.util.Collections;
@@ -12,6 +11,7 @@ import java.util.Objects;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.TextRange;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
 
 import lombok.experimental.UtilityClass;
@@ -19,13 +19,13 @@ import lombok.experimental.UtilityClass;
 @UtilityClass
 public class SearchTextWithTextPositionFactory {
 
+    public final int HEIGHT_PADDING = 2;
     // when checking for a hyphen linebreak, we need to check after a linebreak if the last hyphen was less than three symbols away.
     // We detect a linebreak as either a "\n" character or if two adjacent symbol's position differ in y-coordinates by at least one character height.
     // If there is a hyphen linebreak, the hyphen will be 1 position in front of a "\n" or 2 positions in front of the character which has a lower y-coordinate
     // This is why, we need to initialize this to < -2, otherwise, if the very first symbol is a \n we would detect a hyphen linebreak that isn't there.
     // Also, Integer.MIN_VALUE is a bad idea due to potential overflow during arithmetic operations. This is why the default should be -3.
     public final int MAX_HYPHEN_LINEBREAK_DISTANCE = 3;
-    public static final double LINEBREAK_DELTA_TOLERANCE = 1.5;
 
 
     public SearchTextWithTextPositionDto buildSearchTextToTextPositionDto(List<TextPositionSequence> sequences) {
@@ -37,13 +37,15 @@ public class SearchTextWithTextPositionFactory {
 
         Context context = new Context();
 
-        RedTextPosition currentTextPosition = sequences.get(0).getTextPositions().get(0);
+        RedTextPosition currentTextPosition = sequences.get(0).getTextPositions()
+                .get(0);
         RedTextPosition previousTextPosition = RedTextPosition.builder().unicode(" ").bBoxDirAdj(currentTextPosition.getBBoxDirAdj()).build();
 
         for (TextPositionSequence word : sequences) {
             for (int i = 0; i < word.getTextPositions().size(); ++i) {
 
-                currentTextPosition = word.getTextPositions().get(i);
+                currentTextPosition = word.getTextPositions()
+                        .get(i);
                 if (isLineBreak(currentTextPosition, previousTextPosition)) {
                     removeHyphenLinebreaks(context);
                     context.lineBreaksStringIdx.add(context.stringIdx);
@@ -65,10 +67,11 @@ public class SearchTextWithTextPositionFactory {
             ++context.stringIdx;
         }
 
+
         List<Rectangle2D> positions = sequences.stream()
                 .map(TextPositionSequence::getTextPositions)
                 .flatMap(Collection::stream)
-                .map(RedTextPosition::getBBoxPdf)
+                .map(RedTextPosition::getBBoxInitialUserSpace)
                 .toList();
 
         assert context.stringBuilder.length() == context.stringIdxToPositionIdx.size();
@@ -157,8 +160,8 @@ public class SearchTextWithTextPositionFactory {
             return false;
         }
 
-        double deltaY = (Math.abs(currentPosition.getYDirAdj() - previousPosition.getYDirAdj()) * LINEBREAK_DELTA_TOLERANCE) + (2 * HEIGHT_PADDING);
-        return deltaY >= currentPosition.getHeightDirAdj() || deltaY >= previousPosition.getHeightDirAdj();
+        double deltaY = Math.abs(currentPosition.getYDirAdj() - previousPosition.getYDirAdj());
+        return deltaY >= currentPosition.getHeightDir();
     }
 
 
@@ -184,6 +187,32 @@ public class SearchTextWithTextPositionFactory {
     }
 
 
+    public Rectangle2D mapRedTextPositionToInitialUserSpace(RedTextPosition textPosition, TextPositionSequence sequence) {
+
+        float textHeight = sequence.getTextHeight() + HEIGHT_PADDING;
+        Rectangle2D rectangle2D = new Rectangle2D.Double(textPosition.getXDirAdj(),
+                textPosition.getYDirAdj() - textHeight,
+                textPosition.getWidthDirAdj(),
+                textHeight + HEIGHT_PADDING);
+
+        AffineTransform transform = new AffineTransform();
+
+        if (sequence.getDir() == TextDirection.ZERO || sequence.getDir() == TextDirection.HALF_CIRCLE) {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageHeight() / 2f);
+            transform.translate(0f, sequence.getPageHeight());
+        } else if (sequence.getDir() == TextDirection.QUARTER_CIRCLE) {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageWidth() / 2f);
+            transform.translate(0f, sequence.getPageWidth());
+        } else {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageHeight() / 2f, sequence.getPageHeight() / 2f);
+            transform.translate(0f, sequence.getPageWidth());
+        }
+        transform.scale(1., -1.);
+
+        return transform.createTransformedShape(rectangle2D).getBounds2D();
+    }
+
+
     private class Context {
 
         List<Integer> stringIdxToPositionIdx = new LinkedList<>();

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

@@ -51,6 +51,9 @@ public class SectionNodeFactory {
             return Optional.empty();
         }
 
+        Map<Integer, List<AbstractPageBlock>> blocksPerPage = pageBlocks.stream()
+                .collect(groupingBy(AbstractPageBlock::getPage));
+
         AbstractSemanticNode section;
         boolean containsTablesAndTextBlocks = containsTablesAndTextBlocks(pageBlocks);
         if (isLeaf && !containsTablesAndTextBlocks) {
@@ -60,6 +63,8 @@ public class SectionNodeFactory {
         }
 
         context.getSections().add(section);
+        blocksPerPage.keySet()
+                .forEach(pageNumber -> addSectionNodeToPageNode(context, section, pageNumber));
 
         section.setTreeId(getTreeId(parentNode, context, section));
 
@@ -229,7 +234,7 @@ public class SectionNodeFactory {
                 .filter(abstractTextContainer -> !abstractTextContainer.equals(atc))
                 .filter(abstractTextContainer -> abstractTextContainer.getPage() == atc.getPage())
                 .filter(abstractTextContainer -> abstractTextContainer instanceof TextPageBlock)
-                .filter(abstractTextContainer -> abstractTextContainer.intersectsYPdf(atc))
+                .filter(abstractTextContainer -> abstractTextContainer.intersectsY(atc))
                 .map(abstractTextContainer -> (TextPageBlock) abstractTextContainer)
                 .filter(abstractTextContainer -> abstractTextContainer.getDir() == atc.getDir())
                 .filter(abstractTextContainer -> !abstractTextContainer.isToDuplicate())
@@ -237,5 +242,10 @@ public class SectionNodeFactory {
     }
 
 
+    private void addSectionNodeToPageNode(DocumentGraphFactory.Context context, AbstractSemanticNode section, Integer pageNumber) {
+
+        Page page = context.getPage(pageNumber);
+        page.getMainBody().add(section);
+    }
 
 }

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

@@ -51,6 +51,8 @@ public class TableNodeFactory {
                 .numberOfRows(mergedRows.size())
                 .build();
 
+        pages.forEach(page -> addTableToPage(page, parentNode, table));
+
         List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, table);
         table.setTreeId(treeId);
         addTableCells(layoutParsingType, mergedRows, table, context, document);
@@ -80,6 +82,17 @@ public class TableNodeFactory {
     }
 
 
+    @SuppressWarnings("PMD.UnusedPrivateMethod") // PMD actually flags this wrong
+    private void addTableToPage(Page page, SemanticNode parentNode, Table table) {
+
+        if (!page.getMainBody().contains(parentNode)) {
+            parentNode.getPages().add(page);
+        }
+
+        page.getMainBody().add(table);
+    }
+
+
     private void ifTableHasNoHeadersSetFirstRowAsHeaders(Table table) {
 
         if (table.streamHeaders()
@@ -94,7 +107,14 @@ public class TableNodeFactory {
 
         for (int rowIndex = 0; rowIndex < rows.size(); rowIndex++) {
             for (int colIndex = 0; colIndex < rows.get(rowIndex).size(); colIndex++) {
-                addTableCell(layoutParsingType, rows.get(rowIndex).get(colIndex), rowIndex, colIndex, table, context, document);
+                addTableCell(layoutParsingType,
+                             rows.get(rowIndex)
+                                     .get(colIndex),
+                             rowIndex,
+                             colIndex,
+                             table,
+                             context,
+                             document);
             }
         }
     }
@@ -111,7 +131,14 @@ public class TableNodeFactory {
 
         Page page = context.getPage(cell.getPageNumber());
 
-        TableCell tableCell = TableCell.builder().documentTree(context.getDocumentTree()).row(rowIndex).col(colIndex).header(cell.isHeaderCell()).bBox(cell.getBBoxPdf()).build();
+        TableCell tableCell = TableCell.builder()
+                .documentTree(context.getDocumentTree())
+                .row(rowIndex)
+                .col(colIndex)
+                .header(cell.isHeaderCell())
+                .bBox(cell.getBBoxInitialUserSpace())
+                .build();
+        page.getMainBody().add(tableCell);
 
         List<Integer> treeId = context.getDocumentTree().createNewTableChildEntryAndReturnId(tableNode, tableCell);
         tableCell.setTreeId(treeId);
@@ -120,7 +147,9 @@ public class TableNodeFactory {
         if (cell.getTextBlocks().isEmpty()) {
             tableCell.setLeafTextBlock(context.getTextBlockFactory().emptyTextBlock(tableNode, context, page));
         } else if (cell.getTextBlocks().size() == 1) {
-            textBlock = context.getTextBlockFactory().buildAtomicTextBlock2(cell.getTextBlocks().get(0).getSequences(), tableCell, context, page);
+            textBlock = context.getTextBlockFactory()
+                    .buildAtomicTextBlock(cell.getTextBlocks()
+                                                  .get(0).getSequences(), tableCell, context, page);
             tableCell.setLeafTextBlock(textBlock);
         } else if (firstTextBlockIsHeadline(cell)) {
             SectionNodeFactory.addSection(layoutParsingType,
@@ -134,8 +163,8 @@ public class TableNodeFactory {
                                           context,
                                           document);
         } else if (cellAreaIsSmallerThanPageAreaTimesThreshold(cell, page)) {
-            List<TextPositionSequence> sequences = TextPositionOperations.mergeAndSort(cell.getTextBlocks());
-            textBlock = context.getTextBlockFactory().buildAtomicTextBlock2(sequences, tableCell, context, page);
+            List<TextPositionSequence> sequences = TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(cell.getTextBlocks());
+            textBlock = context.getTextBlockFactory().buildAtomicTextBlock(sequences, tableCell, context, page);
             tableCell.setLeafTextBlock(textBlock);
         } else {
             cell.getTextBlocks()
@@ -152,7 +181,8 @@ public class TableNodeFactory {
 
     private boolean firstTextBlockIsHeadline(Cell cell) {
 
-        return cell.getTextBlocks().get(0).isHeadline();
+        return cell.getTextBlocks()
+                .get(0).isHeadline();
     }
 
 }

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

@@ -17,7 +17,7 @@ public class TextBlockFactory {
     long textBlockIdx;
 
 
-    public AtomicTextBlock buildAtomicTextBlock2(List<TextPositionSequence> sequences, SemanticNode parent, DocumentGraphFactory.Context context, Page page) {
+    public AtomicTextBlock buildAtomicTextBlock(List<TextPositionSequence> sequences, SemanticNode parent, DocumentGraphFactory.Context context, Page page) {
 
         Integer numberOnPage = context.getAndIncrementTextBlockNumberOnPage(page);
         return buildAtomicTextBlock(sequences, parent, numberOnPage, page);
@@ -32,34 +32,35 @@ public class TextBlockFactory {
         long idx = textBlockIdx;
         textBlockIdx++;
         String orientation;
-        int textRotation;
+        int textDirection;
         if (sequences.isEmpty()) {
             orientation = null;
-            textRotation = 0;
+            textDirection = 0;
         } else {
             orientation = sequences.get(0).getDir().toString();
-            textRotation = sequences.get(0).getDir().getRotation();
+            textDirection = sequences.get(0).getRotation();
         }
-        var atb = AtomicTextBlock.fromSearchTextWithTextPosition(searchTextWithTextPositionDto.getSearchText(),
-                                                                 searchTextWithTextPositionDto.getLineBreaks(),
-                                                                 searchTextWithTextPositionDto.getBoldTextBoundaries(),
-                                                                 searchTextWithTextPositionDto.getItalicTextBoundaries(),
-                                                                 searchTextWithTextPositionDto.getPositions(),
-                                                                 searchTextWithTextPositionDto.getStringIdxToPositionIdx(),
-                                                                 idx,
-                                                                 parent,
-                                                                 numberOnPage,
-                                                                 page,
-                                                                 offset,
-                                                                 orientation,
-                                                                 textRotation);
-        return atb;
+        return AtomicTextBlock.fromSearchTextWithTextPosition(searchTextWithTextPositionDto.getSearchText(),
+                searchTextWithTextPositionDto.getLineBreaks(),
+                searchTextWithTextPositionDto.getBoldTextBoundaries(),
+                searchTextWithTextPositionDto.getItalicTextBoundaries(),
+                searchTextWithTextPositionDto.getPositions(),
+                searchTextWithTextPositionDto.getStringIdxToPositionIdx(),
+                idx,
+                parent,
+                numberOnPage,
+                page,
+                offset,
+                orientation,
+                textDirection);
     }
 
 
     public AtomicTextBlock emptyTextBlock(SemanticNode parent, DocumentGraphFactory.Context context, Page page) {
 
-        return emptyTextBlock(parent, context.getAndIncrementTextBlockNumberOnPage(page), page);
+        long idx = textBlockIdx;
+        textBlockIdx++;
+        return AtomicTextBlock.empty(idx, stringOffset, page, context.getAndIncrementTextBlockNumberOnPage(page), parent);
     }
 
 
@@ -67,8 +68,7 @@ public class TextBlockFactory {
 
         long idx = textBlockIdx;
         textBlockIdx++;
-        var atb = AtomicTextBlock.empty(idx, stringOffset, page, numberOnPage, parent);
-        return atb;
+        return AtomicTextBlock.empty(idx, stringOffset, page, numberOnPage, parent);
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/graphics/GraphicExtractorService.java

@@ -8,11 +8,12 @@ import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.springframework.stereotype.Service;
 
-import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.services.factory.SearchTextWithTextPositionFactory;
 import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 
 import lombok.RequiredArgsConstructor;
 import lombok.SneakyThrows;
@@ -66,7 +67,10 @@ public class GraphicExtractorService {
     private List<Box> getCharacterBBoxes(List<TextPositionSequence> textPositionSequences) {
 
         return textPositionSequences.stream()
-                .map(BoundingBox::getBBoxPdf)
+                .map(pos -> pos.getTextPositions()
+                        .stream()
+                        .map(tp -> SearchTextWithTextPositionFactory.mapRedTextPositionToInitialUserSpace(tp, pos))
+                        .collect(RectangleTransformations.collectBBox()))
                 .map(Box::new)
                 .collect(Collectors.toList());
     }

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

@@ -41,9 +41,7 @@ public class DocumentGraphMapper {
         DocumentTree documentTree = new DocumentTree(document);
         Context context = new Context(documentData, documentTree);
 
-        context.pages.addAll(Arrays.stream(documentData.getDocumentPages())
-                                     .map(DocumentGraphMapper::buildPage)
-                                     .toList());
+        context.pages.addAll(Arrays.stream(documentData.getDocumentPages()).map(DocumentGraphMapper::buildPage).toList());
 
         context.documentTree.getRoot().getChildren().addAll(buildEntries(documentData.getDocumentStructure().getRoot().getChildren(), context));
 
@@ -61,9 +59,7 @@ public class DocumentGraphMapper {
         List<DocumentTree.Entry> newEntries = new LinkedList<>();
         for (DocumentStructure.EntryData entryData : entries) {
 
-            List<Page> pages = Arrays.stream(entryData.getPageNumbers())
-                    .map(pageNumber -> getPage(pageNumber, context))
-                    .toList();
+            List<Page> pages = Arrays.stream(entryData.getPageNumbers()).map(pageNumber -> getPage(pageNumber, context)).toList();
 
             SemanticNode node = switch (entryData.getType()) {
                 case SECTION -> buildSection(context);
@@ -81,17 +77,16 @@ public class DocumentGraphMapper {
             if (entryData.getAtomicBlockIds().length > 0) {
                 TextBlock textBlock = toTextBlock(entryData.getAtomicBlockIds(), context, node);
                 node.setLeafTextBlock(textBlock);
-                switch (entryData.getType()) {
-                    case HEADER -> pages.forEach(page -> page.setHeader((Header) node));
-                    case FOOTER -> pages.forEach(page -> page.setFooter((Footer) node));
-                    default -> textBlock.getAtomicTextBlocks()
-                            .forEach(atb -> atb.getPage().getTextBlocksOnPage().add(atb));
-                }
             }
-            List<Integer> treeId = Arrays.stream(entryData.getTreeId()).boxed()
-                    .toList();
+            List<Integer> treeId = Arrays.stream(entryData.getTreeId()).boxed().toList();
             node.setTreeId(treeId);
 
+            switch (entryData.getType()) {
+                case HEADER -> pages.forEach(page -> page.setHeader((Header) node));
+                case FOOTER -> pages.forEach(page -> page.setFooter((Footer) node));
+                default -> pages.forEach(page -> page.getMainBody().add(node));
+            }
+
             newEntries.add(DocumentTree.Entry.builder().treeId(treeId).children(buildEntries(entryData.getChildren(), context)).node(node).build());
         }
         return newEntries;
@@ -147,7 +142,6 @@ public class DocumentGraphMapper {
         return Section.builder().documentTree(context.documentTree).build();
     }
 
-
     private SuperSection buildSuperSection(Context context) {
 
         return SuperSection.builder().documentTree(context.documentTree).build();
@@ -172,24 +166,22 @@ public class DocumentGraphMapper {
 
     private TextBlock toTextBlock(Long[] atomicTextBlockIds, Context context, SemanticNode parent) {
 
-        return Arrays.stream(atomicTextBlockIds)
-                .map(atomicTextBlockId -> getAtomicTextBlock(context, parent, atomicTextBlockId))
-                .collect(new TextBlockCollector());
+        return Arrays.stream(atomicTextBlockIds).map(atomicTextBlockId -> getAtomicTextBlock(context, parent, atomicTextBlockId)).collect(new TextBlockCollector());
     }
 
 
     private AtomicTextBlock getAtomicTextBlock(Context context, SemanticNode parent, Long atomicTextBlockId) {
 
         return AtomicTextBlock.fromAtomicTextBlockData(context.documentTextDataBlockData.get(Math.toIntExact(atomicTextBlockId)),
-                                                       context.atomicPositionBlockData.get(Math.toIntExact(atomicTextBlockId)),
-                                                       parent,
-                                                       getPage(context.documentTextDataBlockData.get(Math.toIntExact(atomicTextBlockId)).getPage(), context));
+                context.atomicPositionBlockData.get(Math.toIntExact(atomicTextBlockId)),
+                parent,
+                getPage(context.documentTextDataBlockData.get(Math.toIntExact(atomicTextBlockId)).getPage(), context));
     }
 
 
     private Page buildPage(DocumentPage p) {
 
-        return Page.builder().rotation(p.getRotation()).height(p.getHeight()).width(p.getWidth()).number(p.getNumber()).textBlocksOnPage(new LinkedList<>()).build();
+        return Page.builder().rotation(p.getRotation()).height(p.getHeight()).width(p.getWidth()).number(p.getNumber()).mainBody(new LinkedList<>()).build();
     }
 
 
@@ -214,10 +206,8 @@ public class DocumentGraphMapper {
 
             this.documentTree = documentTree;
             this.pages = new LinkedList<>();
-            this.documentTextDataBlockData = Arrays.stream(documentData.getDocumentTextData())
-                    .toList();
-            this.atomicPositionBlockData = Arrays.stream(documentData.getDocumentPositions())
-                    .toList();
+            this.documentTextDataBlockData = Arrays.stream(documentData.getDocumentTextData()).toList();
+            this.atomicPositionBlockData = Arrays.stream(documentData.getDocumentPositions()).toList();
 
         }
 

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

@@ -24,7 +24,6 @@ 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-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/mapper/TaasDocumentDataMapper.java

@@ -99,7 +99,6 @@ public class TaasDocumentDataMapper {
         Page page = semanticNode.getFirstPage();
         Rectangle2D bBox = semanticNode.getBBox().get(page);
         return StructureObject.builder()
-                .treeId(semanticNode.getTreeId())
                 .structureObjectNumber(structureObjectNumber)
                 .boundingBox(toFloatArray(bBox))
                 .stringOffset(semanticNode.getBoundary().start())
@@ -115,7 +114,6 @@ public class TaasDocumentDataMapper {
         Page page = table.getFirstPage();
         Rectangle2D bBox = table.getBBox().get(page);
         return StructureObject.builder()
-                .treeId(table.getTreeId())
                 .structureObjectNumber(structureObjectNumber)
                 .boundingBox(toFloatArray(bBox))
                 .stringOffset(table.getBoundary().start())

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/parsing/LegacyPDFStreamEngine.java

@@ -82,7 +82,6 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
 
     private int pageRotation;
     private PDRectangle pageSize;
-    private Matrix translateMatrix;
     private final GlyphList glyphList;
     private final Map<COSDictionary, Float> fontHeightMap = new WeakHashMap<COSDictionary, Float>();
 
@@ -134,12 +133,6 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
         this.pageRotation = page.getRotation();
         this.pageSize = page.getCropBox();
 
-        if (pageSize.getLowerLeftX() == 0 && pageSize.getLowerLeftY() == 0) {
-            translateMatrix = null;
-        } else {
-            // translation matrix for cropbox
-            translateMatrix = Matrix.getTranslateInstance(-pageSize.getLowerLeftX(), -pageSize.getLowerLeftY());
-        }
         super.processPage(page);
     }
 
@@ -264,22 +257,13 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
                 return;
             }
         }
-        // adjust for cropbox if needed
-        Matrix translatedTextRenderingMatrix;
-        if (translateMatrix == null) {
-            translatedTextRenderingMatrix = textRenderingMatrix;
-        } else {
-            translatedTextRenderingMatrix = Matrix.concatenate(translateMatrix, textRenderingMatrix);
-            nextX -= pageSize.getLowerLeftX();
-            nextY -= pageSize.getLowerLeftY();
-        }
 
         // This is a hack for unicode letter with 2 chars e.g. RA see unicodeProblem.pdf
         if (unicodeMapping.length() == 2) {
             processTextPosition(new TextPosition(pageRotation,
                                                  pageSize.getWidth(),
                                                  pageSize.getHeight(),
-                                                 translatedTextRenderingMatrix,
+                                                 textRenderingMatrix,
                                                  nextX,
                                                  nextY,
                                                  Math.abs(dyDisplay),
@@ -293,7 +277,7 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
             processTextPosition(new TextPosition(pageRotation,
                                                  pageSize.getWidth(),
                                                  pageSize.getHeight(),
-                                                 translatedTextRenderingMatrix,
+                                                 textRenderingMatrix,
                                                  nextX,
                                                  nextY,
                                                  Math.abs(dyDisplay),
@@ -309,7 +293,7 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
             processTextPosition(new TextPosition(pageRotation,
                                                  pageSize.getWidth(),
                                                  pageSize.getHeight(),
-                                                 translatedTextRenderingMatrix,
+                                                 textRenderingMatrix,
                                                  nextX,
                                                  nextY,
                                                  Math.abs(dyDisplay),