RED-8932 Fixed not merged headline with identifier

Merge branch 'RED-7384' into 'release/0.89.x'
RED-7384: add empty textBlock to Image to ensure continuous textranges across all SemanticNodes See merge request fforesight/layout-parser!139
2024-04-24 11:44:17 +02:00 · 2024-04-23 12:03:32 +02:00 · 2024-04-23 11:30:13 +02:00 · 2024-04-23 10:44:59 +02:00 · 2024-04-22 11:00:43 +02:00 · 2024-04-18 10:01:12 +02:00
76 changed files with 354 additions and 2415 deletions
@@ -5,7 +5,6 @@ import java.io.Serializable;
 import java.util.Arrays;
 import java.util.List;
 import java.util.Map;
-import java.util.Set;
 import java.util.stream.Stream;

 import io.swagger.v3.oas.annotations.media.Schema;
@@ -55,13 +54,6 @@ public class DocumentStructure implements Serializable {

    }

-    @Schema(description = "Object containing the extra field names, a duplicate paragraph has in its properties field.")
-    public static class DuplicateParagraphProperties implements Serializable {
-
-        public static final String UNSORTED_TEXTBLOCK_ID = "utbid";
-
-    }
-
    public static final String RECTANGLE_DELIMITER = ";";


@@ -123,8 +115,6 @@ public class DocumentStructure implements Serializable {
        Map<String, String> properties;
        @Schema(description = "All child Entries of this Entry.", example = "[1, 2, 3]")
        List<EntryData> children;
-        @Schema(description = "Describes the origin of the semantic node", example = "[ALGORITHM]")
-        Set<LayoutEngine> engines;


        @Override
@@ -1,6 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-public enum LayoutEngine {
-    ALGORITHM,
-    AI
-}
@@ -18,15 +18,11 @@ public record LayoutParsingRequest(

        @Schema(description = "Path to the original PDF file.")//
        @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,//

-        @Schema(description = "Optional Path to the the visual layout parsing service file") 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.")//
@@ -3,8 +3,5 @@ package com.knecon.fforesight.service.layoutparser.internal.api.queue;
 public enum LayoutParsingType {
    REDACT_MANAGER,
    TAAS,
-    DOCUMINE,
-
-    DOCSTRUM,
-    DOCSTRUM_XY
+    DOCUMINE
 }
@@ -34,17 +34,14 @@ import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageB
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.CvTableParsingAdapter;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.ImageServiceResponseAdapter;
-import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.VisualLayoutParsingAdapter;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableCells;
 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.layoutparser.processor.services.BodyTextFrameService;
 import com.knecon.fforesight.service.layoutparser.processor.services.RulingCleaningService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SectionsBuilderService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SimplifiedSectionTextService;
 import com.knecon.fforesight.service.layoutparser.processor.services.TableExtractionService;
-import com.knecon.fforesight.service.layoutparser.processor.services.blockification.DocstrumBlockificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.blockification.DocuMineBlockificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.blockification.RedactManagerBlockificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.blockification.TaasBlockificationService;
@@ -88,10 +85,8 @@ public class LayoutParsingPipeline {
    TaasBlockificationService taasBlockificationService;
    DocuMineBlockificationService docuMineBlockificationService;
    RedactManagerBlockificationService redactManagerBlockificationService;
-    DocstrumBlockificationService docstrumBlockificationService;
    LayoutGridService layoutGridService;
    ObservationRegistry observationRegistry;
-    VisualLayoutParsingAdapter visualLayoutParsingAdapter;


    public LayoutParsingFinishedEvent parseLayoutAndSaveFilesToStorage(LayoutParsingRequest layoutParsingRequest) throws IOException {
@@ -100,15 +95,10 @@ 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 = new VisualLayoutParsingResponse();
-        if (layoutParsingRequest.visualLayoutParsingFileId()
-                .isPresent()) {
-            visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId()
-                                                                                                         .get());
-        }

        ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
        if (layoutParsingRequest.imagesFileStorageId()
@@ -125,10 +115,9 @@ public class LayoutParsingPipeline {
        }

        ClassificationDocument classificationDocument = parseLayout(layoutParsingRequest.layoutParsingType(),
-                                                                    originFile,
-                                                                    imageServiceResponse,
-                                                                    tableServiceResponse,
-                                                                    visualLayoutParsingResponse,
+                originFile,
+                imageServiceResponse,
+                tableServiceResponse,
                layoutParsingRequest.identifier().toString());

        log.info("Building document graph for {}", layoutParsingRequest.identifier());
@@ -217,14 +206,13 @@ public class LayoutParsingPipeline {
                                              File originFile,
                                              ImageServiceResponse imageServiceResponse,
                                              TableServiceResponse tableServiceResponse,
-                                              VisualLayoutParsingResponse visualLayoutParsingResponse,
                                              String identifier) {

        PDDocument originDocument = openDocument(originFile);
        addNumberOfPagesToTrace(originDocument.getNumberOfPages(), Files.size(originFile.toPath()));
        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();
        List<ClassificationPage> classificationPages = new ArrayList<>();

@@ -263,14 +251,12 @@ public class LayoutParsingPipeline {
            PDRectangle cropbox = pdPage.getCropBox();
            CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(pdfTableCells.get(pageNumber), stripper.getRulings());

-            List<Cell> emptyTableCells = tableExtractionService.findCells(cleanRulings.getHorizontal(), cleanRulings.getVertical());
+            List<Cell> emptyTableCells = TableExtractionService.findCells(cleanRulings.getHorizontal(), cleanRulings.getVertical());

            ClassificationPage classificationPage = switch (layoutParsingType) {
-                case REDACT_MANAGER -> redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
+                case REDACT_MANAGER -> redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells);
                case TAAS -> taasBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
                case DOCUMINE -> docuMineBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
-                case DOCSTRUM -> docstrumBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells, false);
-                case DOCSTRUM_XY -> docstrumBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells, true);
            };
            classificationPage.setCleanRulings(cleanRulings);
            classificationPage.setRotation(rotation);
@@ -288,19 +274,7 @@ public class LayoutParsingPipeline {
                imageServiceResponseAdapter.findOcr(classificationPage);
            }

-            if (signatures.containsKey(pageNumber)) {
-                if (classificationPage.getImages() == null || classificationPage.getImages().size() == 0) {
-                    classificationPage.setImages(signatures.get(pageNumber));
-                } else {
-                    classificationPage.getImages().addAll(signatures.get(pageNumber));
-                }
-            }
-
-            tableExtractionService.extractTables(emptyTableCells, classificationPage);
-
-            if (layoutParsingType == LayoutParsingType.DOCSTRUM || layoutParsingType == LayoutParsingType.DOCSTRUM_XY) {
-                docstrumBlockificationService.combineBlocks(classificationPage);
-            }
+            tableExtractionService.extractTables(cleanRulings, classificationPage);

            buildPageStatistics(classificationPage);
            increaseDocumentStatistics(classificationPage, classificationDocument);
@@ -317,27 +291,11 @@ public class LayoutParsingPipeline {
            case TAAS -> taasClassificationService.classifyDocument(classificationDocument);
            case DOCUMINE -> docuMineClassificationService.classifyDocument(classificationDocument);
            case REDACT_MANAGER -> redactManagerClassificationService.classifyDocument(classificationDocument);
-            case DOCSTRUM_XY -> redactManagerClassificationService.classifyDocument(classificationDocument);
-            case DOCSTRUM -> redactManagerClassificationService.classifyDocument(classificationDocument);
        }

        log.info("Building Sections for {}", identifier);
-
-//        if (layoutParsingType == DOCSTRUM || layoutParsingType == DOCSTRUM_XY) {
-//            // Currently for debugging return paragraphs as sections, because there is a merging logic in sectionBuilder
-//            List<ClassificationSection> sections = new ArrayList<>();
-//            for (var page : classificationPages) {
-//                page.getTextBlocks().forEach(block -> {
-//                    block.setPage(page.getPageNumber());
-//                    var section = sectionsBuilderService.buildTextBlock(List.of(block), "a");
-//                    sections.add(section);
-//                });
-//            }
-//            classificationDocument.setSections(sections);
-//        } else {
        sectionsBuilderService.buildSections(classificationDocument);
        sectionsBuilderService.addImagesToSections(classificationDocument);
-//        }
        return classificationDocument;
    }

@@ -10,6 +10,8 @@ import java.nio.file.Paths;
 import java.nio.file.StandardOpenOption;
 import java.util.Optional;

+import org.apache.pdfbox.Loader;
+import org.apache.pdfbox.pdmodel.PDDocument;
 import org.springframework.stereotype.Service;

 import com.fasterxml.jackson.databind.ObjectMapper;
@@ -20,7 +22,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.Researc
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
 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.tenantcommons.TenantContext;

 import io.micrometer.observation.annotation.Observed;
@@ -78,14 +79,6 @@ public class LayoutParsingStorageService {
        }
    }

-    public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) throws IOException {
-
-        try (InputStream inputStream = getObject(storageId)) {
-            VisualLayoutParsingResponse visualLayoutParsingResponse = objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
-            return visualLayoutParsingResponse;
-        }
-    }
-

    @Observed(name = "LayoutParsingStorageService", contextualName = "store-document-data")
    public void storeDocumentData(LayoutParsingRequest layoutParsingRequest, DocumentData documentData) {
@@ -96,7 +96,7 @@ public abstract class AbstractPageBlock extends Rectangle {

        return this.minX - threshold <= apb.getMaxX() && this.maxX + threshold >= apb.getMinX();
    }
-    
+

    public abstract boolean isEmpty();

@@ -10,11 +10,11 @@ import java.util.Set;
 import java.util.stream.Collectors;
 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.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;

 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -30,9 +30,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Document implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
    Set<Page> pages;
    DocumentTree documentTree;
    Integer numberOfPages;
@@ -51,7 +48,7 @@ public class Document implements GenericSemanticNode {
    public TextBlock getTextBlock() {

        if (textBlock == null) {
-            textBlock = GenericSemanticNode.super.getTextBlock();
+            textBlock = streamTerminalTextBlocksInOrder().collect(new TextBlockCollector());
        }
        return textBlock;
    }
@@ -59,14 +56,13 @@ public class Document implements GenericSemanticNode {

    public List<Section> getMainSections() {

-        return streamChildrenOfType(NodeType.SECTION).map(node -> (Section) node)
-                .collect(Collectors.toList());
+        return streamChildrenOfType(NodeType.SECTION).map(node -> (Section) node).collect(Collectors.toList());
    }


    public Stream<TextBlock> streamTerminalTextBlocksInOrder() {

-        return streamAllNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getTextBlock);
+        return streamAllNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock);
    }


@@ -87,16 +83,13 @@ public class Document implements GenericSemanticNode {
    @Override
    public Headline getHeadline() {

-        return streamAllSubNodesOfType(NodeType.HEADLINE).map(node -> (Headline) node)
-                .findFirst()
-                .orElse(Headline.builder().build());
+        return streamAllSubNodesOfType(NodeType.HEADLINE).map(node -> (Headline) node).findFirst().orElse(Headline.builder().build());
    }


    private Stream<SemanticNode> streamAllNodes() {

-        return documentTree.allEntriesInOrder()
-                .map(DocumentTree.Entry::getNode);
+        return documentTree.allEntriesInOrder().map(DocumentTree.Entry::getNode);
    }


@@ -1,34 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.SuperBuilder;
-
-@Data
-@EqualsAndHashCode(callSuper = true)
-@SuperBuilder
-public class DuplicatedParagraph extends Paragraph {
-
-    TextBlock unsortedLeafTextBlock;
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return Stream.of(leafTextBlock, unsortedLeafTextBlock).collect(new TextBlockCollector());
-
-    }
-
-
-    @Override
-    public String toString() {
-
-        return super.toString();
-    }
-
-}
@@ -6,7 +6,6 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
@@ -27,9 +26,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Footer implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
    List<Integer> treeId;
    TextBlock leafTextBlock;

@@ -6,7 +6,6 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
@@ -27,8 +26,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Header implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
    List<Integer> treeId;
    TextBlock leafTextBlock;

@@ -6,7 +6,6 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
@@ -27,8 +26,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Headline implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
    List<Integer> treeId;
    TextBlock leafTextBlock;

@@ -8,12 +8,10 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;

 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -30,9 +28,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Image implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
    List<Integer> treeId;
    String id;

@@ -40,6 +35,8 @@ public class Image implements GenericSemanticNode {
    boolean transparent;
    Rectangle2D position;

+    TextBlock leafTextBlock;
+
    boolean redaction;
    boolean ignored;
    @Builder.Default
@@ -70,9 +67,7 @@ public class Image implements GenericSemanticNode {
    @Override
    public TextBlock getTextBlock() {

-        return streamAllSubNodes().filter(SemanticNode::isLeaf)
-                .map(SemanticNode::getLeafTextBlock)
-                .collect(new TextBlockCollector());
+        return leafTextBlock;
    }


@@ -98,4 +93,11 @@ public class Image implements GenericSemanticNode {
        return bBoxPerPage;
    }

+
+    @Override
+    public boolean isLeaf() {
+
+        return true;
+    }
+
 }
@@ -6,8 +6,6 @@ public enum ImageType {
    LOGO,
    FORMULA,
    SIGNATURE,
-
-    SIGNATURE_VISUAL,
    OTHER,
    OCR;

@@ -6,7 +6,6 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
@@ -18,17 +17,13 @@ import lombok.Builder;
 import lombok.Data;
 import lombok.EqualsAndHashCode;
 import lombok.experimental.FieldDefaults;
-import lombok.experimental.SuperBuilder;

@Data
-@SuperBuilder
+@Builder
@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PROTECTED)
+@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Paragraph implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
    List<Integer> treeId;
    TextBlock leafTextBlock;

@@ -6,11 +6,11 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;

 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -27,8 +27,6 @@ import lombok.extern.slf4j.Slf4j;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Section implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
    List<Integer> treeId;

    TextBlock textBlock;
@@ -52,8 +50,7 @@ public class Section implements GenericSemanticNode {

    public boolean hasTables() {

-        return streamAllSubNodesOfType(NodeType.TABLE).findAny()
-                .isPresent();
+        return streamAllSubNodesOfType(NodeType.TABLE).findAny().isPresent();
    }


@@ -61,7 +58,7 @@ public class Section implements GenericSemanticNode {
    public TextBlock getTextBlock() {

        if (textBlock == null) {
-            textBlock = GenericSemanticNode.super.getTextBlock();
+            textBlock = streamAllSubNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
        }
        return textBlock;
    }
@@ -12,7 +12,6 @@ import java.util.Set;
 import java.util.stream.Collectors;
 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.Boundary;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
@@ -20,7 +19,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.E
 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.RectangleTransformations;

 public interface SemanticNode {
@@ -40,10 +38,7 @@ public interface SemanticNode {
     *
     * @return TextBlock containing all AtomicTextBlocks that are located under this Node.
     */
-    default TextBlock getTextBlock() {
-
-        return streamAllSubNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getTextBlock).collect(new TextBlockCollector());
-    }
+    TextBlock getTextBlock();


    /**
@@ -340,23 +335,6 @@ public interface SemanticNode {
    }


-    /**
-     * returns the set of layoutengines.
-     *
-     * @return set of layoutengines.
-     */
-    Set<LayoutEngine> getEngines();
-
-
-    /**
-     * adds a layoutengine to the set.
-     */
-    default void addEngine(LayoutEngine engine) {
-
-        getEngines().add(engine);
-    }
-
-
    /**
     * Streams all children located directly underneath this node in the DocumentTree.
     *
@@ -448,7 +426,6 @@ public interface SemanticNode {
    /**
     * TODO: this produces unwanted results for sections spanning multiple columns.
     * Computes the Union of the bounding boxes of all children recursively.
-     *
     * @return The union of the BoundingBoxes of all children
     */
    private Map<Page, Rectangle2D> getBBoxFromChildren() {
@@ -12,7 +12,6 @@ import java.util.Set;
 import java.util.stream.IntStream;
 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.entity.RedactionEntity;
@@ -32,8 +31,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class Table implements SemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
    List<Integer> treeId;
    DocumentTree documentTree;

@@ -56,7 +53,8 @@ public class Table implements SemanticNode {
     */
    public Stream<RedactionEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) {

-        return IntStream.range(0, numberOfRows).boxed()
+        return IntStream.range(0, numberOfRows)
+                .boxed()
                .filter(row -> rowContainsStringsIgnoreCase(row, strings))
                .flatMap(this::streamRow)
                .map(TableCell::getEntities)
@@ -73,11 +71,8 @@ public class Table implements SemanticNode {
     */
    public boolean rowContainsStringsIgnoreCase(Integer row, List<String> strings) {

-        String rowText = streamRow(row).map(TableCell::getTextBlock)
-                .collect(new TextBlockCollector()).getSearchText().toLowerCase(Locale.ROOT);
-        return strings.stream()
-                .map(String::toLowerCase)
-                .allMatch(rowText::contains);
+        String rowText = streamRow(row).map(TableCell::getTextBlock).collect(new TextBlockCollector()).getSearchText().toLowerCase(Locale.ROOT);
+        return strings.stream().map(String::toLowerCase).allMatch(rowText::contains);
    }


@@ -90,13 +85,9 @@ public class Table implements SemanticNode {
     */
    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) {

-        List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header))
-                .map(TableCell::getCol)
-                .toList();
+        List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList();
        return streamTableCells().filter(tableCellNode -> vertebrateStudyCols.stream()
-                        .anyMatch(vertebrateStudyCol -> getCell(tableCellNode.getRow(), vertebrateStudyCol).containsString(value)))
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
+                .anyMatch(vertebrateStudyCol -> getCell(tableCellNode.getRow(), vertebrateStudyCol).containsString(value))).map(TableCell::getEntities).flatMap(Collection::stream);
    }


@@ -109,13 +100,9 @@ public class Table implements SemanticNode {
     */
    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) {

-        List<Integer> colsWithHeader = streamHeaders().filter(headerNode -> headerNode.containsString(header))
-                .map(TableCell::getCol)
-                .toList();
+        List<Integer> colsWithHeader = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList();
        return streamTableCells().filter(tableCellNode -> colsWithHeader.stream()
-                        .anyMatch(colWithHeader -> getCell(tableCellNode.getRow(), colWithHeader).containsAnyString(values)))
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
+                .anyMatch(colWithHeader -> getCell(tableCellNode.getRow(), colWithHeader).containsAnyString(values))).map(TableCell::getEntities).flatMap(Collection::stream);
    }


@@ -127,15 +114,12 @@ public class Table implements SemanticNode {
     */
    public Stream<RedactionEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) {

-        List<Integer> rowsWithEntityOfType = IntStream.range(0, numberOfRows).boxed()
-                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).anyMatch(existingType -> types.stream()
-                        .anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
+        List<Integer> rowsWithEntityOfType = IntStream.range(0, numberOfRows)
+                .boxed()
+                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).anyMatch(existingType -> types.stream().anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
                .toList();

-        return rowsWithEntityOfType.stream()
-                .flatMap(this::streamRow)
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
+        return rowsWithEntityOfType.stream().flatMap(this::streamRow).map(TableCell::getEntities).flatMap(Collection::stream);
    }


@@ -147,24 +131,18 @@ public class Table implements SemanticNode {
     */
    public Stream<RedactionEntity> streamEntitiesWhereRowContainsNoEntitiesOfType(List<String> types) {

-        List<Integer> rowsWithNoEntityOfType = IntStream.range(0, numberOfRows).boxed()
-                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).noneMatch(existingType -> types.stream()
-                        .anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
+        List<Integer> rowsWithNoEntityOfType = IntStream.range(0, numberOfRows)
+                .boxed()
+                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).noneMatch(existingType -> types.stream().anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
                .toList();

-        return rowsWithNoEntityOfType.stream()
-                .flatMap(this::streamRow)
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
+        return rowsWithNoEntityOfType.stream().flatMap(this::streamRow).map(TableCell::getEntities).flatMap(Collection::stream);
    }


    private Stream<String> streamEntityTypesInRow(Integer rowNumber) {

-        return streamRow(rowNumber).map(TableCell::getEntities)
-                .flatMap(Collection::stream)
-                .map(RedactionEntity::getType)
-                .distinct();
+        return streamRow(rowNumber).map(TableCell::getEntities).flatMap(Collection::stream).map(RedactionEntity::getType).distinct();
    }


@@ -181,8 +159,7 @@ public class Table implements SemanticNode {
            throw new IllegalArgumentException(format("row %d, col %d is out of bounds for number of rows of %d and number of cols %d", row, col, numberOfRows, numberOfCols));
        }
        int idx = row * numberOfCols + col;
-        return (TableCell) documentTree.getEntryById(treeId).getChildren()
-                .get(idx).getNode();
+        return (TableCell) documentTree.getEntryById(treeId).getChildren().get(idx).getNode();
    }


@@ -219,8 +196,7 @@ public class Table implements SemanticNode {
     */
    public Stream<TableCell> streamCol(int col) {

-        return IntStream.range(0, numberOfRows).boxed()
-                .map(row -> getCell(row, col));
+        return IntStream.range(0, numberOfRows).boxed().map(row -> getCell(row, col));
    }


@@ -232,11 +208,9 @@ public class Table implements SemanticNode {
     */
    public Stream<TableCell> streamRow(int row) {

-        return IntStream.range(0, numberOfCols).boxed()
-                .map(col -> getCell(row, col));
+        return IntStream.range(0, numberOfCols).boxed().map(col -> getCell(row, col));
    }

-
    /**
     * Streams all TableCells row-wise and filters them with header == true.
     *
@@ -257,8 +231,7 @@ public class Table implements SemanticNode {
     */
    public Stream<TableCell> streamHeadersForCell(int row, int col) {

-        return Stream.concat(streamRow(row), streamCol(col))
-                .filter(TableCell::isHeader);
+        return Stream.concat(streamRow(row), streamCol(col)).filter(TableCell::isHeader);
    }


@@ -331,7 +304,7 @@ public class Table implements SemanticNode {
    public TextBlock getTextBlock() {

        if (textBlock == null) {
-            textBlock = SemanticNode.super.getTextBlock();
+            textBlock = streamAllSubNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
        }
        return textBlock;
    }
@@ -342,8 +315,6 @@ public class Table implements SemanticNode {

        return treeId.toString() + ": " + NodeType.TABLE + ": #cols: " + numberOfCols + ", #rows: " + numberOfRows + ", " + this.getTextBlock().buildSummary();
    }
-
-
    @Override
    public Map<Page, Rectangle2D> getBBox() {

@@ -352,5 +323,4 @@ public class Table implements SemanticNode {
        }
        return bBoxCache;
    }
-
 }
@@ -7,7 +7,6 @@ import java.util.List;
 import java.util.Map;
 import java.util.Set;

-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.entity.RedactionEntity;
@@ -27,8 +26,6 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
 public class TableCell implements GenericSemanticNode {

-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
    List<Integer> treeId;
    int row;
    int col;
@@ -4,21 +4,18 @@ import java.awt.geom.Rectangle2D;

 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.ImageType;

-import lombok.AllArgsConstructor;
 import lombok.Data;
 import lombok.NonNull;
 import lombok.RequiredArgsConstructor;

@Data
@RequiredArgsConstructor
-@AllArgsConstructor
 public class ClassifiedImage {

    @NonNull
    private Rectangle2D position;
    @NonNull
    private ImageType imageType;
-    private boolean sourceByAi;
    private boolean isAppendedToSection;
    @NonNull
    private boolean hasTransparency;
@@ -1,7 +1,6 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.text;

 import org.apache.pdfbox.text.TextPosition;
-import org.springframework.beans.BeanUtils;

 import com.fasterxml.jackson.annotation.JsonIgnore;

@@ -50,7 +49,13 @@ public class RedTextPosition {
    public static RedTextPosition fromTextPosition(TextPosition textPosition) {

        var pos = new RedTextPosition();
-        BeanUtils.copyProperties(textPosition, pos);
+        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.setFontSizeInPt(textPosition.getFontSizeInPt());
@@ -53,9 +53,6 @@ public class TextPageBlock extends AbstractPageBlock {
    @JsonIgnore
    private PageBlockType classification;

-    @JsonIgnore
-    private boolean toDuplicate;
-

    @JsonIgnore
    public TextDirection getDir() {
@@ -76,7 +73,7 @@ public class TextPageBlock extends AbstractPageBlock {

        return sequences.get(0).getPageWidth();
    }
-
+    

    public static TextPageBlock merge(List<TextPageBlock> textBlocksToMerge) {

@@ -85,7 +82,6 @@ public class TextPageBlock extends AbstractPageBlock {
        return fromTextPositionSequences(sequences);
    }

-
    public static TextPageBlock fromTextPositionSequences(List<TextPositionSequence> wordBlockList) {

        TextPageBlock textBlock = null;
@@ -137,6 +133,7 @@ public class TextPageBlock extends AbstractPageBlock {
    }


+
    /**
     * Returns the minX value in pdf coordinate system.
     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
@@ -365,22 +362,7 @@ public class TextPageBlock extends AbstractPageBlock {
        }

        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString());
-    }

-
-    public int getNumberOfLines() {
-
-        int numberOfLines = 1;
-        TextPositionSequence previous = null;
-        for (TextPositionSequence word : sequences) {
-            if (previous != null) {
-                if (word.getMaxYDirAdj() - previous.getMaxYDirAdj() > word.getTextHeight()) {
-                    numberOfLines++;
-                }
-            }
-            previous = word;
-        }
-        return numberOfLines;
    }


@@ -55,17 +55,6 @@ public class TextPositionSequence implements CharSequence {
    }


-    public TextPositionSequence(List<RedTextPosition> textPositions, int page) {
-
-        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();
-    }
-
-
    @Override
    public int length() {

@@ -201,6 +190,12 @@ public class TextPositionSequence implements CharSequence {
    }


+    public float getTextHeightNoPadding() {
+
+        return textPositions.get(0).getHeightDir();
+    }
+
+
    public float getTextHeight() {

        return textPositions.get(0).getHeightDir() + HEIGHT_PADDING;
@@ -1,92 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.adapter;
-
-import java.awt.geom.Rectangle2D;
-import java.util.ArrayList;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.ImageType;
-import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
-import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingBox;
-import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
-import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResult;
-
-import lombok.RequiredArgsConstructor;
-import lombok.extern.slf4j.Slf4j;
-
-@Service
-@RequiredArgsConstructor
-@Slf4j
-public class VisualLayoutParsingAdapter {
-
-    private static String SIGNATURES = "signature";
-
-
-    public Map<Integer, List<VisualLayoutParsingResult>> buildExtractedTablesPerPage(VisualLayoutParsingResponse visualLayoutParsingResponse) {
-
-        Map<Integer, List<VisualLayoutParsingResult>> tableCells = new HashMap<>();
-        visualLayoutParsingResponse.getData()
-                .forEach(tableData -> tableCells.computeIfAbsent(tableData.getPage_idx(), tableCell -> new ArrayList<>()).addAll(convertTableCells(tableData.getBoxes())));
-
-        return tableCells;
-
-    }
-
-
-    public Map<Integer, List<ClassifiedImage>> buildExtractedSignaturesPerPage(VisualLayoutParsingResponse visualLayoutParsingResponse) {
-
-        Map<Integer, List<ClassifiedImage>> signatures = new HashMap<>();
-        if (visualLayoutParsingResponse.getData() != null) {
-            visualLayoutParsingResponse.getData()
-                    .forEach(tableData -> signatures.computeIfAbsent(tableData.getPage_idx() + 1, tableCell -> new ArrayList<>())
-                            .addAll(convertSignatures(tableData.getPage_idx()+1, tableData.getBoxes())));
-        }
-        return signatures;
-
-    }
-
-
-    public List<VisualLayoutParsingResult> convertTableCells(List<VisualLayoutParsingBox> tableObjects) {
-
-        List<VisualLayoutParsingResult> parsedTableCells = new ArrayList<>();
-
-        tableObjects.stream().forEach(t -> {
-            VisualLayoutParsingResult result = new VisualLayoutParsingResult();
-            result.setX0(t.getBox().getX1());
-            result.setX1(t.getBox().getX2());
-            result.setY0(t.getBox().getY1());
-            result.setY1(t.getBox().getY2());
-            result.setWidth(result.getX1() - result.getX0());
-            result.setHeight(result.getY1() - result.getY0());
-            result.setLabel(t.getLabel());
-            parsedTableCells.add(result);
-        });
-
-        return parsedTableCells;
-
-    }
-
-
-    public List<ClassifiedImage> convertSignatures(int pageNumber, List<VisualLayoutParsingBox> tableObjects) {
-
-        List<ClassifiedImage> signatures = new ArrayList<>();
-
-        tableObjects.stream().forEach(t -> {
-            if (t.getLabel().equals(SIGNATURES)) {
-                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);
-
-                signatures.add(signature);
-            }
-        });
-
-        return signatures;
-
-    }
-
-}
@@ -1,18 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class VisualLayoutParsingBox {
-
-    private VisualLayoutParsingBoxValue box;
-    private String label;
-    private float probability;
-
-}
@@ -1,19 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class VisualLayoutParsingBoxValue {
-
-    private float x1;
-    private float y1;
-    private float x2;
-    private float y2;
-
-}
@@ -1,21 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
-
-import java.util.List;
-
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class VisualLayoutParsingData {
-
-    private int page_idx;
-
-    private List<VisualLayoutParsingBox> boxes;
-
-}
@@ -1,23 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
-
-import java.util.List;
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class VisualLayoutParsingResponse {
-
-    private String dossierId;
-    private String fileId;
-    private String targetFileExtension;
-    private String responseFileExtension;
-    private String X_TENANT_ID;
-    private List<VisualLayoutParsingData> data;
-
-}
@@ -1,22 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class VisualLayoutParsingResult {
-
-    private float x0;
-    private float y0;
-    private float x1;
-    private float y1;
-    private float width;
-    private float height;
-    private String label;
-
-}
@@ -43,7 +43,6 @@ public class SectionsBuilderService {
        for (ClassificationPage page : document.getPages()) {
            List<TextPageBlock> header = new ArrayList<>();
            List<TextPageBlock> footer = new ArrayList<>();
-            List<TextPageBlock> unclassifiedText = new ArrayList<>();
            for (AbstractPageBlock current : page.getTextBlocks()) {

                if (current.getClassification() == null) {
@@ -62,11 +61,6 @@ public class SectionsBuilderService {
                    continue;
                }

-                if (current.getClassification().equals(PageBlockType.OTHER)) {
-                    unclassifiedText.add((TextPageBlock) current);
-                    continue;
-                }
-
                if (prev != null && current.getClassification().isHeadline() && !prev.getClassification().isHeadline() || !document.isHeadlines()) {
                    ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
                    chunkBlock.setHeadline(lastHeadline);
@@ -94,9 +88,6 @@ public class SectionsBuilderService {
            if (!footer.isEmpty()) {
                footers.add(new ClassificationFooter(footer));
            }
-            if (!unclassifiedText.isEmpty()) {
-                unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
-            }
        }

        ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
@@ -240,7 +231,7 @@ public class SectionsBuilderService {
    }


-    public ClassificationSection buildTextBlock(List<AbstractPageBlock> wordBlockList, String lastHeadline) {
+    private ClassificationSection buildTextBlock(List<AbstractPageBlock> wordBlockList, String lastHeadline) {

        ClassificationSection section = new ClassificationSection();

@@ -14,6 +14,7 @@ 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.table.Cell;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.Rectangle;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
@@ -40,18 +41,19 @@ public class TableExtractionService {
     * <p>
     * DirAdj (Text direction adjusted) values can not be used here.
     *
-     * @param emptyCells The cells used to build the table.
-     * @param page       Page object that contains textblocks and statistics.
+     * @param cleanRulings The lines used to build the table.
+     * @param page         Page object that contains textblocks and statistics.
     */

-    public void extractTables(List<Cell> emptyCells, ClassificationPage page) {
+    public void extractTables(CleanRulings cleanRulings, ClassificationPage page) {

+        List<Cell> cells = findCells(cleanRulings.getHorizontal(), cleanRulings.getVertical());
        // sort cells by size (height * width) ascending so that textBlocks are always assigned to the smallest cells that contain them
-        emptyCells.sort(CELL_SIZE_COMPARATOR);
+        cells.sort(CELL_SIZE_COMPARATOR);

        for (AbstractPageBlock abstractPageBlock : page.getTextBlocks()) {
            TextPageBlock textBlock = (TextPageBlock) abstractPageBlock;
-            for (Cell cell : emptyCells) {
+            for (Cell cell : cells) {
                if (cell.hasMinimumSize() && doesCellContainTextBlock(cell, textBlock)) {
                    cell.addTextBlock(textBlock);
                    break;
@@ -59,7 +61,7 @@ public class TableExtractionService {
            }
        }

-        var cells = new ArrayList<>(new HashSet<>(emptyCells));
+        cells = new ArrayList<>(new HashSet<>(cells));
        DoubleComparisons.sort(cells, Rectangle.ILL_DEFINED_ORDER);

        List<Rectangle> spreadsheetAreas = SpreadsheetFinder.findSpreadsheetsFromCells(cells);
@@ -77,7 +79,9 @@ public class TableExtractionService {
                }
            }

-            var containedCellsWithText = containedCells.stream().filter(cell -> !cell.getTextBlocks().isEmpty()).toList();
+            var containedCellsWithText = containedCells.stream()
+                    .filter(cell -> !cell.getTextBlocks().isEmpty())
+                    .toList();

            // verify if table would contain fewer cells with text than the threshold allows
            if (containedCellsWithText.size() >= MAX_TABLE_CONTAINED_CELLS_WITH_TEXT && checkIfTableCellsAreUniform(containedCells)) {
@@ -97,7 +101,11 @@ public class TableExtractionService {
            if (position != -1) {
                page.getTextBlocks().add(position, table);

-                var toBeRemoved = table.getCells().stream().map(Cell::getTextBlocks).flatMap(List::stream).toList();
+                var toBeRemoved = table.getCells()
+                        .stream()
+                        .map(Cell::getTextBlocks)
+                        .flatMap(List::stream)
+                        .toList();
                // remove text blocks from the page that were also added with the table (from its contained cells)
                page.getTextBlocks().removeAll(toBeRemoved);
            }
@@ -107,7 +115,7 @@ public class TableExtractionService {

    private boolean checkIfTableCellsAreUniform(List<Cell> containedCells) {

-        if (containedCells.size() <= 2) {
+        if(containedCells.size() <= 2) {
            return true;
        }

@@ -131,13 +139,19 @@ public class TableExtractionService {
        }
        double x0 = cell.getX();
        double y0 = cell.getY();
-        return (x >= x0 - TEXT_BLOCK_CONTAINMENT_TOLERANCE && y >= y0 - TEXT_BLOCK_CONTAINMENT_TOLERANCE && (x + w) <= x0 + cell.getWidth() + 2 * TEXT_BLOCK_CONTAINMENT_TOLERANCE && (y + h) <= y0 + cell.getHeight() + 2 * TEXT_BLOCK_CONTAINMENT_TOLERANCE);
+        return (x >= x0 - TEXT_BLOCK_CONTAINMENT_TOLERANCE
+                && y >= y0 - TEXT_BLOCK_CONTAINMENT_TOLERANCE
+                && (x + w) <= x0 + cell.getWidth() + 2 * TEXT_BLOCK_CONTAINMENT_TOLERANCE
+                && (y + h) <= y0 + cell.getHeight() + 2 * TEXT_BLOCK_CONTAINMENT_TOLERANCE);
    }


    public static List<Cell> findCells(List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {

-        return RectangularIntersectionFinder.find(horizontalRulingLines, verticalRulingLines).stream().map(Cell::new).collect(Collectors.toList());
+        return RectangularIntersectionFinder.find(horizontalRulingLines, verticalRulingLines)
+                .stream()
+                .map(Cell::new)
+                .collect(Collectors.toList());
    }

 }
@@ -1,408 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
-
-import static java.util.stream.Collectors.toSet;
-
-import java.awt.geom.Point2D;
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.HashSet;
-import java.util.List;
-import java.util.ListIterator;
-import java.util.Set;
-
-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.table.Cell;
-import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
-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;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.DocstrumSegmentationService;
-import com.knecon.fforesight.service.layoutparser.processor.utils.QuickSort;
-import com.knecon.fforesight.service.layoutparser.processor.utils.RulingTextDirAdjustUtil;
-import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionSequenceComparator;
-
-import lombok.RequiredArgsConstructor;
-
-@SuppressWarnings("all")
-@Service
-@RequiredArgsConstructor
-public class DocstrumBlockificationService {
-
-    private final DocstrumSegmentationService docstrumSegmentationService;
-
-    static final float THRESHOLD = 1f;
-
-
-    public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Cell> cells, boolean xyOder) {
-
-        // Underlined or strikethrough are also in rulings but we dont want to split blocks with them so we use cells.
-        List<Ruling> usedHorizonalRulings = new ArrayList<>();
-        List<Ruling> usedVerticalRulings = new ArrayList<>();
-
-        cells.forEach(cell -> {
-            usedHorizonalRulings.add(new Ruling(new Point2D.Float(cell.x, cell.y), new Point2D.Float(cell.x + cell.width, cell.y)));
-            usedHorizonalRulings.add(new Ruling(new Point2D.Float(cell.x, cell.y + cell.height), new Point2D.Float(cell.x + cell.width, cell.y + cell.height)));
-            usedVerticalRulings.add(new Ruling(new Point2D.Float(cell.x, cell.y), new Point2D.Float(cell.x, cell.y + cell.height)));
-            usedVerticalRulings.add(new Ruling(new Point2D.Float(cell.x + cell.width, cell.y), new Point2D.Float(cell.x + cell.width, cell.y + cell.height)));
-        });
-
-        List<AbstractPageBlock> abstractPageBlocks = new ArrayList<>();
-        var zones = docstrumSegmentationService.segmentPage(textPositions, xyOder);
-        zones.forEach(zone -> {
-
-            List<TextPositionSequence> textPositionSequences = new ArrayList<>();
-            zone.getLines().forEach(line -> {
-                line.getWords().forEach(word -> {
-                    textPositionSequences.add(new TextPositionSequence(word.getTextPositions(), word.getPage()));
-                });
-            });
-
-            abstractPageBlocks.addAll(splitZonesAtRulings(textPositionSequences, usedHorizonalRulings, usedVerticalRulings));
-//            abstractPageBlocks.add(buildTextBlock(textPositionSequences, 0));
-        });
-
-        return new ClassificationPage(abstractPageBlocks);
-    }
-
-
-    public void combineBlocks(ClassificationPage page) {
-
-        mergeZones(page.getTextBlocks());
-
-        TextPageBlock previous = new TextPageBlock();
-        ListIterator<AbstractPageBlock> itty = page.getTextBlocks().listIterator();
-        while (itty.hasNext()) {
-            AbstractPageBlock block = itty.next();
-            if (block instanceof TablePageBlock) {
-                continue;
-            }
-            TextPageBlock current = (TextPageBlock) block;
-
-            if (previous != null && !previous.getSequences().isEmpty()) {
-
-                if (current.getDir() == previous.getDir() //
-                        && previous.getNumberOfLines() >= 2 && current.getNumberOfLines() >= 2 //
-                        && previous.intersectsY(current) //
-                        && !hasBetween(current, previous, page.getTextBlocks()) //
-                        && numberOfYIntersections(current, previous, page.getTextBlocks()) == 0) {
-
-                    previous.getSequences().addAll(current.getSequences());
-                    previous = buildTextBlock(previous.getSequences(), 0);
-                    previous.setToDuplicate(true);
-                    itty.remove();
-                    itty.previous();
-                    itty.set(previous);
-                    itty.next();
-                    continue;
-                }
-
-                if (current.getDir() == previous.getDir() && (previous.almostIntersects(current, 0, 0))) {
-
-                    previous.getSequences().addAll(current.getSequences());
-                    boolean toDuplicate = previous.isToDuplicate();
-                    previous = buildTextBlock(previous.getSequences(), 0);
-                    previous.setToDuplicate(toDuplicate);
-                    itty.remove();
-                    itty.previous();
-                    itty.set(previous);
-                    itty.next();
-                    continue;
-                }
-
-                if (current.getDir() == previous.getDir() //
-                        && (Math.abs(previous.getMaxY() - current.getMaxY()) < THRESHOLD || Math.abs(previous.getMinY() - current.getMinY()) < THRESHOLD) //
-                        && (previous.getNumberOfLines() == 1 && current.getNumberOfLines() >= 1 || previous.getNumberOfLines() == 2 && current.getNumberOfLines() == 1) //
-                        && !hasBetween(current, previous, page.getTextBlocks()) && numberOfYIntersections(current, previous, page.getTextBlocks()) <= 4) {
-
-                    previous.getSequences().addAll(current.getSequences());
-                    previous = buildTextBlock(previous.getSequences(), 0);
-                    itty.remove();
-                    itty.previous();
-                    itty.set(previous);
-                    itty.next();
-                    continue;
-                }
-
-                if (current.getDir() == previous.getDir() //
-                        && current.intersectsY(previous) //
-                        && (previous.getNumberOfLines() == 1 && current.getNumberOfLines() >= 1 || previous.getNumberOfLines() == 2 && current.getNumberOfLines() == 1) //
-                        && !hasBetween(current, previous, page.getTextBlocks()) //
-                        && numberOfYIntersections(current, previous, page.getTextBlocks()) <= 0) {
-                    previous.getSequences().addAll(current.getSequences());
-                    previous = buildTextBlock(previous.getSequences(), 0);
-                    itty.remove();
-                    itty.previous();
-                    itty.set(previous);
-                    itty.next();
-                    continue;
-                }
-
-            }
-            previous = current;
-        }
-
-        mergeZones(page.getTextBlocks());
-    }
-
-
-    private boolean hasBetween(TextPageBlock block, TextPageBlock other, List<AbstractPageBlock> allBlocks) {
-
-        for (AbstractPageBlock current : allBlocks) {
-
-            if (current == other || current == block) {
-                continue;
-            }
-
-            if (other.intersectsY(current) && other.getMaxX() <= current.getMinX() && current.getMaxX() <= block.getMinX()) {
-                return true;
-            }
-        }
-
-        return false;
-    }
-
-
-    private int numberOfYIntersections(TextPageBlock block, TextPageBlock other, List<AbstractPageBlock> allBlocks) {
-
-        double minY = Math.min(block.getMinY(), other.getMinY());
-        double maxY = Math.min(block.getMaxY(), other.getMaxY());
-
-        int numberOfYIntersections = 0;
-        for (AbstractPageBlock current : allBlocks) {
-
-            if (current == other || current == block) {
-                continue;
-            }
-
-            if (minY <= current.getMaxY() && maxY >= current.getMinY()) {
-                numberOfYIntersections++;
-            }
-        }
-
-        return numberOfYIntersections;
-    }
-
-
-    private void mergeZones(List<AbstractPageBlock> zones) {
-
-        ListIterator<AbstractPageBlock> itty = zones.listIterator();
-        Set<AbstractPageBlock> toRemove = new HashSet<>();
-        while (itty.hasNext()) {
-            AbstractPageBlock block = itty.next();
-            if (block instanceof TablePageBlock) {
-                continue;
-            }
-
-            TextPageBlock current = (TextPageBlock) block;
-
-            if (current.isToDuplicate()) {
-                continue;
-            }
-
-            for (int i = 0; i < zones.size(); i++) {
-
-                if (toRemove.contains(zones.get(i))) {
-                    continue;
-                }
-                if (zones.get(i) == current) {
-                    continue;
-                }
-                if (zones.get(i) instanceof TablePageBlock) {
-                    continue;
-                }
-
-                TextPageBlock inner = (TextPageBlock) zones.get(i);
-
-                if (inner.isToDuplicate()) {
-                    continue;
-                }
-
-                if (current.getDir() == inner.getDir() && current.almostIntersects(inner, 0, 0)) {
-
-                    current.getSequences().addAll(inner.getSequences());
-                    QuickSort.sort(current.getSequences(), new TextPositionSequenceComparator());
-                    current = buildTextBlock(current.getSequences(), 0);
-                    toRemove.add(inner);
-                    itty.set(current);
-                }
-            }
-        }
-        zones.removeAll(toRemove);
-    }
-
-
-    public List<AbstractPageBlock> splitZonesAtRulings(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
-
-        int indexOnPage = 0;
-        List<TextPositionSequence> chunkWords = new ArrayList<>();
-        List<AbstractPageBlock> chunkBlockList = new ArrayList<>();
-
-        float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
-        TextPositionSequence prev = null;
-
-        for (TextPositionSequence word : textPositions) {
-
-            boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines);
-            boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
-
-            if (prev != null && (splitByDir || isSplitByRuling)) {
-
-                TextPageBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
-                indexOnPage++;
-
-                chunkBlockList.add(cb1);
-                chunkWords = new ArrayList<>();
-
-                minX = 1000;
-                maxX = 0;
-                minY = 1000;
-                maxY = 0;
-                prev = null;
-            }
-
-            chunkWords.add(word);
-
-            prev = word;
-            if (word.getMinXDirAdj() < minX) {
-                minX = word.getMinXDirAdj();
-            }
-            if (word.getMaxXDirAdj() > maxX) {
-                maxX = word.getMaxXDirAdj();
-            }
-            if (word.getMinYDirAdj() < minY) {
-                minY = word.getMinYDirAdj();
-            }
-            if (word.getMaxYDirAdj() > maxY) {
-                maxY = word.getMaxYDirAdj();
-            }
-        }
-
-        TextPageBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
-        if (cb1 != null) {
-            chunkBlockList.add(cb1);
-        }
-
-        return chunkBlockList;
-    }
-
-
-    private boolean equalsWithThreshold(float f1, float f2) {
-
-        return Math.abs(f1 - f2) < THRESHOLD;
-    }
-
-
-    private TextPageBlock buildTextBlock(List<TextPositionSequence> wordBlockList, int indexOnPage) {
-
-        TextPageBlock textBlock = null;
-
-        FloatFrequencyCounter lineHeightFrequencyCounter = new FloatFrequencyCounter();
-        FloatFrequencyCounter fontSizeFrequencyCounter = new FloatFrequencyCounter();
-        FloatFrequencyCounter spaceFrequencyCounter = new FloatFrequencyCounter();
-        StringFrequencyCounter fontFrequencyCounter = new StringFrequencyCounter();
-        StringFrequencyCounter styleFrequencyCounter = new StringFrequencyCounter();
-
-        for (TextPositionSequence wordBlock : wordBlockList) {
-
-            lineHeightFrequencyCounter.add(wordBlock.getTextHeight());
-            fontSizeFrequencyCounter.add(wordBlock.getFontSize());
-            spaceFrequencyCounter.add(wordBlock.getSpaceWidth());
-            fontFrequencyCounter.add(wordBlock.getFont());
-            styleFrequencyCounter.add(wordBlock.getFontStyle());
-
-            if (textBlock == null) {
-                textBlock = new TextPageBlock(wordBlock.getMinXDirAdj(),
-                        wordBlock.getMaxXDirAdj(),
-                        wordBlock.getMinYDirAdj(),
-                        wordBlock.getMaxYDirAdj(),
-                        wordBlockList,
-                        wordBlock.getRotation());
-            } else {
-                TextPageBlock spatialEntity = textBlock.union(wordBlock);
-                textBlock.resize(spatialEntity.getMinX(), spatialEntity.getMinY(), spatialEntity.getWidth(), spatialEntity.getHeight());
-            }
-        }
-
-        if (textBlock != null) {
-            textBlock.setMostPopularWordFont(fontFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordStyle(styleFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordFontSize(fontSizeFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordHeight(lineHeightFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordSpaceWidth(spaceFrequencyCounter.getMostPopular());
-            textBlock.setHighestFontSize(fontSizeFrequencyCounter.getHighest());
-        }
-
-        if (textBlock != null && textBlock.getSequences() != null && textBlock.getSequences().stream().map(t -> round(t.getMinYDirAdj(), 3)).collect(toSet()).size() == 1) {
-            textBlock.getSequences().sort(Comparator.comparing(TextPositionSequence::getMinXDirAdj));
-        }
-        return textBlock;
-    }
-
-
-    private boolean isSplitByRuling(float minX,
-                                    float minY,
-                                    float maxX,
-                                    float maxY,
-                                    TextPositionSequence word,
-                                    List<Ruling> horizontalRulingLines,
-                                    List<Ruling> verticalRulingLines) {
-
-        return isSplitByRuling(maxX,
-                minY,
-                word.getMinXDirAdj(),
-                word.getMinYDirAdj(),
-                verticalRulingLines,
-                word.getDir().getDegrees(),
-                word.getPageWidth(),
-                word.getPageHeight()) //
-                || isSplitByRuling(minX,
-                minY,
-                word.getMinXDirAdj(),
-                word.getMaxYDirAdj(),
-                horizontalRulingLines,
-                word.getDir().getDegrees(),
-                word.getPageWidth(),
-                word.getPageHeight()) //
-                || isSplitByRuling(maxX,
-                minY,
-                word.getMinXDirAdj(),
-                word.getMinYDirAdj(),
-                horizontalRulingLines,
-                word.getDir().getDegrees(),
-                word.getPageWidth(),
-                word.getPageHeight()) //
-                || isSplitByRuling(minX,
-                minY,
-                word.getMinXDirAdj(),
-                word.getMaxYDirAdj(),
-                verticalRulingLines,
-                word.getDir().getDegrees(),
-                word.getPageWidth(),
-                word.getPageHeight());
-    }
-
-
-    private boolean isSplitByRuling(float previousX2, float previousY1, float currentX1, float currentY1, List<Ruling> rulingLines, float dir, float pageWidth, float pageHeight) {
-
-        for (Ruling ruling : rulingLines) {
-            var line = RulingTextDirAdjustUtil.convertToDirAdj(ruling, dir, pageWidth, pageHeight);
-            if (line.intersectsLine(previousX2, previousY1, currentX1, currentY1)) {
-                return true;
-            }
-        }
-        return false;
-    }
-
-
-    private double round(float value, int decimalPoints) {
-
-        var d = Math.pow(10, decimalPoints);
-        return Math.round(value * d) / d;
-    }
-
-}
@@ -61,8 +61,8 @@ public class DocuMineBlockificationService {
            boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj();
            boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines);
            boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
-            boolean splitByOtherFontAndOtherY = prev != null && prev.getMaxYDirAdj() != word.getMaxYDirAdj() && (word.getFontStyle().contains("bold") && !prev.getFontStyle()
-                    .contains("bold") || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold"));
+            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.getMinXDirAdj()) < 5 && matcher.matches();
@@ -2,6 +2,7 @@ package com.knecon.fforesight.service.layoutparser.processor.services.blockifica

 import static java.util.stream.Collectors.toSet;

+import java.awt.geom.Point2D;
 import java.util.ArrayList;
 import java.util.Comparator;
 import java.util.Iterator;
@@ -13,10 +14,13 @@ import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBl
 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.Cell;
+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.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.utils.RectangleTransformations;
 import com.knecon.fforesight.service.layoutparser.processor.utils.RulingTextDirAdjustUtil;

@SuppressWarnings("all")
@@ -31,12 +35,12 @@ public class RedactManagerBlockificationService {
     * This method must use text direction adjusted postions (DirAdj). Where {0,0} is on the upper left. Never try to change this!
     * Rulings (TablePageBlock lines) must be adjusted to the text directions as well, when checking if a block is split by a ruling.
     *
-     * @param textPositions         The words of a page.
-     * @param horizontalRulingLines Horizontal table lines.
-     * @param verticalRulingLines   Vertical table lines.
+     * @param textPositions The words of a page.
     * @return Page object that contains the Textblock and text statistics.
     */
-    public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
+    public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Cell> emptyCells) {
+
+        CleanRulings usedRulings = RectangleTransformations.extractRulings(emptyCells);

        int indexOnPage = 0;
        List<TextPositionSequence> chunkWords = new ArrayList<>();
@@ -54,7 +58,7 @@ public class RedactManagerBlockificationService {
            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 = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines);
+            boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, usedRulings.getHorizontal(), usedRulings.getVertical());
            boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());

            if (prev != null && (lineSeparation || startFromTop || splitByX || splitByDir || isSplitByRuling)) {
@@ -1,59 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum;
-
-import java.util.ArrayList;
-import java.util.List;
-import java.util.stream.Collectors;
-
-import org.springframework.stereotype.Service;
-
-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.services.docstrum.model.Character;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Zone;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service.LineBuilderService;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service.NearestNeighbourService;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service.ReadingOrderService;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service.SpacingService;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service.ZoneBuilderService;
-
-import lombok.RequiredArgsConstructor;
-
-@Service
-@RequiredArgsConstructor
-public class DocstrumSegmentationService {
-
-    private final NearestNeighbourService nearestNeighbourService;
-    private final SpacingService spacingService;
-    private final LineBuilderService lineBuilderService;
-    private final ZoneBuilderService zoneBuilderService;
-    private final ReadingOrderService readingOrderService;
-
-
-    public List<Zone> segmentPage(List<TextPositionSequence> textPositions, boolean xyOder) {
-
-        List<Zone> zones = new ArrayList<>();
-        zones.addAll(computeZones(textPositions, TextDirection.ZERO));
-        zones.addAll(computeZones(textPositions, TextDirection.QUARTER_CIRCLE));
-        zones.addAll(computeZones(textPositions, TextDirection.HALF_CIRCLE));
-        zones.addAll(computeZones(textPositions, TextDirection.THREE_QUARTER_CIRCLE));
-
-        return readingOrderService.resolve(zones, xyOder);
-    }
-
-
-    private List<Zone> computeZones(List<TextPositionSequence> textPositions, TextDirection direction) {
-
-        var positions = textPositions.stream().filter(t -> t.getDir() == direction).map(TextPositionSequence::getTextPositions).flatMap(List::stream).toList();
-
-        var characters = positions.stream().map(Character::new).collect(Collectors.toList());
-
-        nearestNeighbourService.findNearestNeighbors(characters);
-
-        var characterSpacing = spacingService.computeCharacterSpacing(characters);
-        var lineSpacing = Math.min(spacingService.computeLineSpacing(characters), 20);
-
-        var lines = lineBuilderService.buildLines(characters, characterSpacing, lineSpacing);
-        return zoneBuilderService.buildZones(lines, characterSpacing, lineSpacing);
-    }
-
-}
@@ -1,25 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-public class AngleFilter {
-
-    protected double lowerAngle;
-    protected double upperAngle;
-
-
-    public AngleFilter(double lowerAngle, double upperAngle) {
-        
-        this.lowerAngle = lowerAngle < -Math.PI / 2 ? lowerAngle + Math.PI : lowerAngle;
-        this.upperAngle = upperAngle >= Math.PI / 2 ? upperAngle - Math.PI : upperAngle;
-    }
-
-
-    public boolean matches(Neighbor neighbor) {
-
-        if (lowerAngle <= upperAngle) {
-            return lowerAngle <= neighbor.getAngle() && neighbor.getAngle() < upperAngle;
-        } else {
-            return lowerAngle <= neighbor.getAngle() || neighbor.getAngle() < upperAngle;
-        }
-    }
-
-}
@@ -1,48 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import java.awt.geom.Rectangle2D;
-
-import lombok.Data;
-
-@Data
-public abstract class BoundingBox {
-
-    private Rectangle2D bBox;
-
-
-    public double getX() {
-
-        return bBox.getX();
-    }
-
-
-    public double getY() {
-
-        return bBox.getY();
-    }
-
-
-    public double getWidth() {
-
-        return bBox.getWidth();
-    }
-
-
-    public double getHeight() {
-
-        return bBox.getHeight();
-    }
-
-
-    public double getArea() {
-
-        return (bBox.getHeight() * bBox.getWidth());
-    }
-
-
-    public boolean contains(Rectangle2D contained, double tolerance) {
-
-        return bBox.getX() <= contained.getX() + tolerance && bBox.getY() <= contained.getY() + tolerance && bBox.getX() + bBox.getWidth() >= contained.getX() + contained.getWidth() - tolerance && bBox.getY() + bBox.getHeight() >= contained.getY() + contained.getHeight() - tolerance;
-    }
-
-}
@@ -1,85 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
-
-import lombok.Data;
-
-@Data
-public class Character {
-
-    private final double x;
-    private final double y;
-    private final RedTextPosition textPosition;
-
-    private List<Neighbor> neighbors = new ArrayList<>();
-
-
-    public Character(RedTextPosition chunk) {
-
-        this.x = chunk.getXDirAdj() + chunk.getWidthDirAdj() / 2;
-        this.y = chunk.getYDirAdj() + chunk.getHeightDir() / 2;
-        this.textPosition = chunk;
-    }
-
-
-    public double getHeight() {
-
-        return textPosition.getHeightDir();
-    }
-
-
-    public double distance(Character character) {
-
-        double dx = getX() - character.getX();
-        double dy = getY() - character.getY();
-        return Math.sqrt(dx * dx + dy * dy);
-    }
-
-
-    public double horizontalDistance(Character character) {
-
-        return Math.abs(getX() - character.getX());
-    }
-
-
-    public double verticalDistance(Character character) {
-
-        return Math.abs(getY() - character.getY());
-    }
-
-
-    public double overlappingDistance(Character other) {
-
-        double[] xs = new double[4];
-        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.getHeightDir());
-        xs[2] = c * other.x - s * other.y;
-        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);
-    }
-
-
-    public void setNeighbors(List<Neighbor> neighbors) {
-
-        this.neighbors = neighbors;
-    }
-
-
-    public double angle(Character character) {
-
-        if (getX() > character.getX()) {
-            return Math.atan2(getY() - character.getY(), getX() - character.getX());
-        } else {
-            return Math.atan2(character.getY() - getY(), character.getX() - getX());
-        }
-    }
-
-}
@@ -1,194 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import java.util.AbstractSet;
-import java.util.Collection;
-import java.util.HashMap;
-import java.util.Iterator;
-import java.util.Map;
-import java.util.NoSuchElementException;
-import java.util.Set;
-
-public class DisjointSets<E> implements Iterable<Set<E>> {
-
-    private final Map<E, Entry<E>> map = new HashMap<>();
-
-
-    public DisjointSets(Collection<? extends E> collection) {
-
-        for (E element : collection) {
-            map.put(element, new Entry<E>(element));
-        }
-    }
-
-
-    public boolean areTogether(E e1, E e2) {
-
-        return map.get(e1).findRepresentative().equals(map.get(e2).findRepresentative());
-    }
-
-
-    public void union(E e1, E e2) {
-
-        Entry<E> r1 = map.get(e1).findRepresentative();
-        Entry<E> r2 = map.get(e2).findRepresentative();
-        if (!r1.equals(r2)) {
-            if (r1.size <= r2.size) {
-                r2.mergeWith(r1);
-            } else {
-                r1.mergeWith(r2);
-            }
-        }
-    }
-
-
-    @Override
-    public Iterator<Set<E>> iterator() {
-
-        return new Iterator<>() {
-
-            private final Iterator<Entry<E>> iterator = map.values().iterator();
-            private Entry<E> nextRepresentative;
-
-            {
-                findNextRepresentative();
-            }
-
-            @Override
-            public boolean hasNext() {
-
-                return nextRepresentative != null;
-            }
-
-
-            @Override
-            public Set<E> next() {
-
-                if (nextRepresentative == null) {
-                    throw new NoSuchElementException();
-                }
-                Set<E> result = nextRepresentative.asSet();
-                findNextRepresentative();
-                return result;
-            }
-
-
-            private void findNextRepresentative() {
-
-                while (iterator.hasNext()) {
-                    Entry<E> candidate = iterator.next();
-                    if (candidate.isRepresentative()) {
-                        nextRepresentative = candidate;
-                        return;
-                    }
-                }
-                nextRepresentative = null;
-            }
-
-
-            @Override
-            public void remove() {
-
-                throw new UnsupportedOperationException();
-            }
-
-        };
-    }
-
-
-    private static class Entry<E> {
-
-        private int size = 1;
-        private final E value;
-        private Entry<E> parent = this;
-        private Entry<E> next;
-        private Entry<E> last = this;
-
-
-        Entry(E value) {
-
-            this.value = value;
-        }
-
-
-        void mergeWith(Entry<E> otherRepresentative) {
-
-            size += otherRepresentative.size;
-            last.next = otherRepresentative;
-            last = otherRepresentative.last;
-            otherRepresentative.parent = this;
-        }
-
-
-        Entry<E> findRepresentative() {
-
-            Entry<E> representative = parent;
-            while (!representative.parent.equals(representative)) {
-                representative = representative.parent;
-            }
-            for (Entry<E> entry = this; !entry.equals(representative); ) {
-                Entry<E> nextEntry = entry.parent;
-                entry.parent = representative;
-                entry = nextEntry;
-            }
-            return representative;
-        }
-
-
-        boolean isRepresentative() {
-
-            return parent.equals(this);
-        }
-
-
-        Set<E> asSet() {
-
-            return new AbstractSet<E>() {
-
-                @Override
-                public Iterator<E> iterator() {
-
-                    return new Iterator<E>() {
-
-                        private Entry<E> nextEntry = findRepresentative();
-
-
-                        @Override
-                        public boolean hasNext() {
-
-                            return nextEntry != null;
-                        }
-
-
-                        @Override
-                        public E next() {
-
-                            if (nextEntry == null) {
-                                throw new NoSuchElementException();
-                            }
-                            E result = nextEntry.value;
-                            nextEntry = nextEntry.next;
-                            return result;
-                        }
-
-
-                        @Override
-                        public void remove() {
-
-                            throw new UnsupportedOperationException();
-                        }
-
-                    };
-                }
-
-
-                @Override
-                public int size() {
-
-                    return findRepresentative().size;
-                }
-            };
-        }
-
-    }
-
-}
@@ -1,90 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-public class Histogram {
-
-    private static final double EPSILON = 1.0e-6;
-    private final double min;
-    private final double resolution;
-    private double[] frequencies;
-
-
-    public Histogram(double minValue, double maxValue, double resolution) {
-
-        this.min = minValue - EPSILON;
-        double delta = maxValue - minValue + 2 * EPSILON;
-        int size = Math.max(1, (int) Math.round((maxValue - minValue) / resolution));
-        this.resolution = delta / size;
-        this.frequencies = new double[size];
-    }
-
-
-    public void kernelSmooth(double[] kernel) {
-
-        double[] newFrequencies = new double[frequencies.length];
-        int shift = (kernel.length - 1) / 2;
-        for (int i = 0; i < kernel.length; i++) {
-            int jStart = Math.max(0, i - shift);
-            int jEnd = Math.min(frequencies.length, frequencies.length + i - shift);
-            for (int j = jStart; j < jEnd; j++) {
-                newFrequencies[j - i + shift] += kernel[i] * frequencies[j];
-            }
-        }
-        frequencies = newFrequencies;
-    }
-
-
-    public double[] createGaussianKernel(double length, double stdDeviation) {
-
-        int r = (int) Math.round(length / resolution) / 2;
-
-        int size = 2 * r + 1;
-        double[] kernel = new double[size];
-        double sum = 0;
-        double b = 2 * (stdDeviation / resolution) * (stdDeviation / resolution);
-        double a = 1 / Math.sqrt(Math.PI * b);
-        for (int i = 0; i < size; i++) {
-            kernel[i] = a * Math.exp(-(i - r) * (i - r) / b);
-            sum += kernel[i];
-        }
-        for (int i = 0; i < size; i++) {
-            kernel[i] /= sum;
-        }
-        return kernel;
-    }
-
-
-    public void gaussianSmooth(double windowLength, double stdDeviation) {
-
-        kernelSmooth(createGaussianKernel(windowLength, stdDeviation));
-    }
-
-
-    public void add(double value) {
-
-        frequencies[(int) ((value - min) / resolution)] += 1.0;
-    }
-
-
-    public int getSize() {
-
-        return frequencies.length;
-    }
-
-
-    public double getPeakValue() {
-
-        int peakIndex = 0;
-        for (int i = 1; i < frequencies.length; i++) {
-            if (frequencies[i] > frequencies[peakIndex]) {
-                peakIndex = i;
-            }
-        }
-        int peakEndIndex = peakIndex + 1;
-        final double EPS = 0.0001;
-        while (peakEndIndex < frequencies.length && Math.abs(frequencies[peakEndIndex] - frequencies[peakIndex]) < EPS) {
-            peakEndIndex++;
-        }
-        return ((double) peakIndex + peakEndIndex) / 2 * resolution + min;
-    }
-
-}
@@ -1,168 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import java.awt.geom.Rectangle2D;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
-
-import lombok.Data;
-
-@Data
-public class Line extends BoundingBox {
-
-    private static final double WORD_DISTANCE_MULTIPLIER = 0.2;
-
-    private final double x0;
-    private final double y0;
-
-    private final double x1;
-    private final double y1;
-
-    private final double height;
-
-    private final List<Character> characters;
-    private final List<TextPositionSequence> words = new ArrayList<>();
-
-
-    public Line(List<Character> characters, double wordSpacing) {
-
-        this.characters = characters;
-
-        if (characters.size() >= 2) {
-            // linear regression
-            double sx = 0.0;
-            double sxx = 0.0;
-            double sxy = 0.0;
-            double sy = 0.0;
-            for (Character character : characters) {
-                sx += character.getX();
-                sxx += character.getX() * character.getX();
-                sxy += character.getX() * character.getY();
-                sy += character.getY();
-            }
-            double b = (characters.size() * sxy - sx * sy) / (characters.size() * sxx - sx * sx);
-            double a = (sy - b * sx) / characters.size();
-
-            this.x0 = characters.get(0).getX();
-            this.y0 = a + b * this.x0;
-            this.x1 = characters.get(characters.size() - 1).getX();
-            this.y1 = a + b * this.x1;
-        } else {
-            Character character = characters.get(0);
-            double dx = character.getTextPosition().getWidthDirAdj() / 3;
-            double dy = dx * Math.tan(0);
-            this.x0 = character.getX() - dx;
-            this.x1 = character.getX() + dx;
-            this.y0 = character.getY() - dy;
-            this.y1 = character.getY() + dy;
-        }
-        height = computeHeight();
-        computeWords(wordSpacing * WORD_DISTANCE_MULTIPLIER);
-        buildBBox();
-    }
-
-
-    public double getAngle() {
-
-        return Math.atan2(y1 - y0, x1 - x0);
-    }
-
-
-    public double getLength() {
-
-        return Math.sqrt((x0 - x1) * (x0 - x1) + (y0 - y1) * (y0 - y1));
-    }
-
-
-    private double computeHeight() {
-
-        double sum = 0.0;
-        for (Character component : characters) {
-            sum += component.getHeight();
-        }
-        return sum / characters.size();
-    }
-
-
-    public double angularDifference(Line j) {
-
-        double diff = Math.abs(getAngle() - j.getAngle());
-        if (diff <= Math.PI / 2) {
-            return diff;
-        } else {
-            return Math.PI - diff;
-        }
-    }
-
-
-    public double horizontalDistance(Line other) {
-
-        double[] xs = new double[4];
-        xs[0] = x0;
-        xs[1] = x1;
-        xs[2] = other.x0;
-        xs[3] = other.x1;
-        boolean overlapping = xs[1] >= xs[2] && xs[3] >= xs[0];
-        Arrays.sort(xs);
-        return Math.abs(xs[2] - xs[1]) * (overlapping ? 1 : -1);
-    }
-
-
-    public double verticalDistance(Line other) {
-
-        double ym = (y0 + y1) / 2;
-        double yn = (other.y0 + other.y1) / 2;
-        return Math.abs(ym - yn) / Math.sqrt(1);
-    }
-
-
-    private void computeWords(double wordSpacing) {
-
-        TextPositionSequence word = new TextPositionSequence();
-        Character previous = null;
-        for (Character current : characters) {
-            if (previous != null) {
-                double dist = current.getTextPosition().getXDirAdj() - previous.getTextPosition().getXDirAdj() - previous.getTextPosition().getWidthDirAdj();
-                if (dist > wordSpacing) {
-                    words.add(word);
-                    word = new TextPositionSequence();
-                }
-            }
-            word.getTextPositions().add(current.getTextPosition());
-            previous = current;
-        }
-        words.add(word);
-    }
-
-
-    private void buildBBox() {
-
-        double minX = Double.POSITIVE_INFINITY;
-        double minY = Double.POSITIVE_INFINITY;
-        double maxX = Double.NEGATIVE_INFINITY;
-        double maxY = Double.NEGATIVE_INFINITY;
-
-        for (Character character : characters) {
-
-            minX = Math.min(minX, character.getTextPosition().getXDirAdj());
-            minY = Math.min(minY, character.getTextPosition().getYDirAdj());
-            maxX = Math.max(maxX, character.getTextPosition().getXDirAdj() + character.getTextPosition().getWidthDirAdj());
-            maxY = Math.max(maxY, character.getTextPosition().getYDirAdj() + character.getTextPosition().getHeightDir());
-
-        }
-
-        this.setBBox(new Rectangle2D.Double(minX, minY, maxX - minX, maxY - minY));
-    }
-
-
-    public String toString() {
-
-        StringBuilder sb = new StringBuilder();
-        words.forEach(word -> sb.append(word.toString()).append(" "));
-        return sb.toString().trim();
-    }
-
-}
-
@@ -1,36 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import lombok.Getter;
-
-public class Neighbor {
-
-    @Getter
-    private final double distance;
-    @Getter
-    private final double angle;
-    private final Character originCharacter;
-    @Getter
-    private final Character character;
-
-
-    public Neighbor(Character neighbor, Character origin) {
-
-        this.distance = neighbor.distance(origin);
-        this.angle = neighbor.angle(origin);
-        this.character = neighbor;
-        this.originCharacter = origin;
-    }
-
-
-    public double getHorizontalDistance() {
-
-        return character.horizontalDistance(originCharacter);
-    }
-
-
-    public double getVerticalDistance() {
-
-        return character.verticalDistance(originCharacter);
-    }
-
-}
@@ -1,51 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Comparator;
-import java.util.List;
-
-import lombok.Data;
-
-@Data
-public class Zone extends BoundingBox {
-
-    private List<Line> lines;
-
-
-    @SuppressWarnings("PMD.ConstructorCallsOverridableMethod")
-    public Zone(List<Line> lines) {
-
-        lines.sort(Comparator.comparingDouble(Line::getY));
-        this.lines = lines;
-        buildBBox();
-    }
-
-
-    public void buildBBox() {
-
-        double minX = Double.POSITIVE_INFINITY;
-        double minY = Double.POSITIVE_INFINITY;
-        double maxX = Double.NEGATIVE_INFINITY;
-        double maxY = Double.NEGATIVE_INFINITY;
-
-        for (Line line : lines) {
-
-            minX = Math.min(minX, line.getX());
-            minY = Math.min(minY, line.getY());
-            maxX = Math.max(maxX, line.getX() + line.getWidth());
-            maxY = Math.max(maxY, line.getY() + line.getHeight());
-
-        }
-
-        this.setBBox(new Rectangle2D.Double(minX, minY, maxX - minX, maxY - minY));
-    }
-
-
-    public String toString() {
-
-        StringBuilder sb = new StringBuilder();
-        lines.forEach(line -> sb.append(line.toString()).append("\n"));
-        return sb.toString().trim();
-    }
-
-}
@@ -1,51 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service;
-
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.List;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.AngleFilter;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Character;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.DisjointSets;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Line;
-
-@Service
-public class LineBuilderService {
-
-    private static final double CHARACTER_SPACING_DISTANCE_MULTIPLIER = 3.5;
-    private static final double MAX_VERTICAL_CHARACTER_DISTANCE = 0.67;
-    private static final double ANGLE_TOLERANCE = Math.PI / 6;
-
-
-    public List<Line> buildLines(List<Character> characters, double characterSpacing, double lineSpacing) {
-
-        double maxHorizontalDistance = characterSpacing * CHARACTER_SPACING_DISTANCE_MULTIPLIER;
-        double maxVerticalDistance = lineSpacing * MAX_VERTICAL_CHARACTER_DISTANCE;
-
-        DisjointSets<Character> sets = new DisjointSets<>(characters);
-        AngleFilter filter = new AngleFilter(-ANGLE_TOLERANCE, ANGLE_TOLERANCE);
-
-        characters.forEach(character -> {
-            character.getNeighbors().forEach(neighbor -> {
-                double x = neighbor.getHorizontalDistance() / maxHorizontalDistance;
-                double y = neighbor.getVerticalDistance() / maxVerticalDistance;
-                if (character.getTextPosition().getDir() == neighbor.getCharacter().getTextPosition().getDir() && filter.matches(neighbor) && Math.pow(x, 2) + Math.pow(y,
-                        2) <= 1) {
-                    sets.union(character, neighbor.getCharacter());
-                }
-            });
-        });
-
-        List<Line> lines = new ArrayList<>();
-        sets.forEach(group -> {
-            List<Character> lineCharacters = new ArrayList<>(group);
-            lineCharacters.sort(Comparator.comparingDouble(Character::getX));
-            lines.add(new Line(lineCharacters, characterSpacing));
-        });
-
-        return lines;
-    }
-
-}
@@ -1,78 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service;
-
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.List;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Character;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Neighbor;
-
-@Service
-public class NearestNeighbourService {
-
-    private static final int NUMBER_OF_NEIGHBOURS = 8;
-    private static final double STEP = 16.0;
-
-
-    public void findNearestNeighbors(List<Character> characters) {
-
-        if (characters.isEmpty()) {
-            return;
-        }
-
-        characters.sort(Comparator.comparingDouble(Character::getX));
-
-        int maxNeighborCount = NUMBER_OF_NEIGHBOURS;
-        if (characters.size() <= NUMBER_OF_NEIGHBOURS) {
-            maxNeighborCount = characters.size() - 1;
-        }
-
-        for (int i = 0; i < characters.size(); i++) {
-
-            List<Neighbor> candidates = new ArrayList<>();
-
-            int start = i;
-            int end = i + 1;
-
-            double distance = Double.POSITIVE_INFINITY;
-
-            for (double searchDistance = 0; searchDistance < distance; ) {
-
-                searchDistance += STEP;
-                boolean newCandidatesFound = false;
-
-                while (start > 0 && characters.get(i).getX() - characters.get(start - 1).getX() < searchDistance) {
-                    start--;
-                    candidates.add(new Neighbor(characters.get(start), characters.get(i)));
-                    clearLeastDistant(candidates, maxNeighborCount);
-                    newCandidatesFound = true;
-                }
-
-                while (end < characters.size() && characters.get(end).getX() - characters.get(i).getX() < searchDistance) {
-                    candidates.add(new Neighbor(characters.get(end), characters.get(i)));
-                    clearLeastDistant(candidates, maxNeighborCount);
-                    end++;
-                    newCandidatesFound = true;
-                }
-
-                if (newCandidatesFound && candidates.size() >= maxNeighborCount) {
-                    distance = candidates.get(maxNeighborCount - 1).getDistance();
-                }
-            }
-            clearLeastDistant(candidates, maxNeighborCount);
-            characters.get(i).setNeighbors(new ArrayList<>(candidates));
-        }
-    }
-
-
-    private void clearLeastDistant(List<Neighbor> candidates, int maxNeighborCount) {
-
-        if (candidates.size() > maxNeighborCount) {
-            candidates.sort(Comparator.comparingDouble(Neighbor::getDistance));
-            candidates.remove(candidates.remove(candidates.size() - 1));
-        }
-    }
-
-}
@@ -1,100 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service;
-
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.List;
-import java.util.ListIterator;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.BoundingBox;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Zone;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.utils.DoubleUtils;
-
-@Service
-public class ReadingOrderService {
-
-    private static final double THRESHOLD = 5;
-
-
-    public List<Zone> resolve(List<Zone> zones, boolean xyOrder) {
-
-        if (zones.isEmpty() || zones.size() == 1) {
-            return zones;
-        }
-
-        if (xyOrder) {
-//            QuickSort.sort(zones, new ZoneComparator());
-            zones.sort(Comparator.comparing(BoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
-                    .thenComparing(BoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, 0)));
-            return zones;
-        }
-
-        return resolveMultiColumnReadingOder(zones);
-    }
-
-
-    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
-
-        double minX = Double.POSITIVE_INFINITY;
-        double maxX = Double.NEGATIVE_INFINITY;
-
-        for (Zone zone : zones) {
-            if (zone.getX() < minX) {
-                minX = zone.getX();
-            }
-            if (zone.getX() + zone.getWidth() > maxX) {
-                maxX = zone.getX() + zone.getWidth();
-            }
-        }
-
-        double midLineXCoordinate = (minX + maxX) / 2;
-
-        List<Zone> leftOf = new ArrayList<>();
-        List<Zone> rightOf = new ArrayList<>();
-        List<Zone> middle = new ArrayList<>();
-        for (Zone zone : zones) {
-            if (zone.getX() < midLineXCoordinate && zone.getX() + zone.getWidth() < midLineXCoordinate) {
-                leftOf.add(zone);
-            } else if (zone.getX() > midLineXCoordinate && zone.getX() + zone.getWidth() > midLineXCoordinate) {
-                rightOf.add(zone);
-            } else {
-                middle.add(zone);
-            }
-        }
-
-        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> sortedZones = new ArrayList<>();
-        sortedZones.addAll(leftOf);
-        sortedZones.addAll(rightOf);
-
-        ListIterator<Zone> itty = middle.listIterator();
-
-        while (itty.hasNext()) {
-            Zone current = itty.next();
-            for (int i = 0; i < sortedZones.size(); i++) {
-                if (current.getY() < sortedZones.get(i).getY()) {
-                    sortedZones.add(i, current);
-                    itty.remove();
-                    break;
-                }
-            }
-        }
-
-        sortedZones.addAll(middle);
-
-        return sortedZones;
-    }
-
-}
@@ -1,56 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service;
-
-import java.util.List;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.AngleFilter;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Character;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Histogram;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Neighbor;
-
-@Service
-public class SpacingService {
-
-    private static final double SPACING_HISTOGRAM_RESOLUTION = 0.5;
-    private static final double SPACING_HISTOGRAM_SMOOTHING_LENGTH = 2.5;
-    private static final double SPACING_HIST_SMOOTHING_STANDARD_DEVIATION = 0.5;
-    private static final double ANGLE_TOLERANCE = Math.PI / 6;
-
-
-    public double computeCharacterSpacing(List<Character> characters) {
-
-        return computeSpacing(characters, 0);
-    }
-
-
-    public double computeLineSpacing(List<Character> characters) {
-
-        return computeSpacing(characters, Math.PI / 2);
-    }
-
-
-    private double computeSpacing(List<Character> characters, double angle) {
-
-        double maxDistance = Double.NEGATIVE_INFINITY;
-
-        for (Character character : characters) {
-            for (Neighbor neighbor : character.getNeighbors()) {
-                maxDistance = Math.max(maxDistance, neighbor.getDistance());
-            }
-        }
-        Histogram histogram = new Histogram(0, maxDistance, SPACING_HISTOGRAM_RESOLUTION);
-        AngleFilter angleFilter = new AngleFilter(angle - ANGLE_TOLERANCE, angle + ANGLE_TOLERANCE);
-        for (Character character : characters) {
-            for (Neighbor neighbor : character.getNeighbors()) {
-                if (angleFilter.matches(neighbor)) {
-                    histogram.add(neighbor.getDistance());
-                }
-            }
-        }
-
-        histogram.gaussianSmooth(SPACING_HISTOGRAM_SMOOTHING_LENGTH, SPACING_HIST_SMOOTHING_STANDARD_DEVIATION);
-        return histogram.getPeakValue();
-    }
-
-}
@@ -1,150 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.service;
-
-import java.util.ArrayList;
-import java.util.Comparator;
-import java.util.List;
-import java.util.Set;
-
-import org.springframework.stereotype.Service;
-
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Character;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.DisjointSets;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Line;
-import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Zone;
-
-@Service
-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 = -3.0;
-
-    private static final double MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER = 0.5;
-
-    private static final double MIN_LINE_SIZE_SCALE = 0.9;
-
-    private static final double MAX_LINE_SIZE_SCALE = 2.5;
-
-    private static final double ANGLE_TOLERANCE = Math.PI / 6;
-
-    private static final int MAX_ZONES = 300;
-
-    private static final double MAX_VERTICAL_MERGE_DISTANCE = 0.5;
-
-
-    public List<Zone> buildZones(List<Line> lines, double characterSpacing, double lineSpacing) {
-
-        double minHorizontalDistance = characterSpacing * MIN_HORIZONTAL_DISTANCE_MULTIPLIER;
-        double maxVerticalDistance = lineSpacing * MAX_VERTICAL_DISTANCE_MULTIPLIER;
-        double minHorizontalMergeDistance = characterSpacing * MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER;
-        double maxVerticalMergeDistance = lineSpacing * MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER;
-
-        DisjointSets<Line> sets = new DisjointSets<>(lines);
-
-        double meanHeight = calculateMeanHeight(lines);
-
-        lines.forEach(outerLine -> //
-                lines.forEach(innerLine -> {
-
-                    double scale = Math.min(outerLine.getHeight(), innerLine.getHeight()) / meanHeight;
-                    scale = Math.max(MIN_LINE_SIZE_SCALE, Math.min(scale, MAX_LINE_SIZE_SCALE));
-
-                    if (!sets.areTogether(outerLine, innerLine) && outerLine.angularDifference(innerLine) <= ANGLE_TOLERANCE) {
-
-                        double horizontalDistance = outerLine.horizontalDistance(innerLine) / scale;
-                        double verticalDistance = outerLine.verticalDistance(innerLine) / scale;
-
-                        if (minHorizontalDistance <= horizontalDistance && verticalDistance <= maxVerticalDistance //
-                                || minHorizontalMergeDistance <= horizontalDistance && verticalDistance <= maxVerticalMergeDistance) {
-                            sets.union(outerLine, innerLine);
-                        }
-                    }
-                }));
-
-        List<Zone> zones = new ArrayList<>();
-        sets.forEach(group -> {
-            zones.add(new Zone(new ArrayList<>(group)));
-        });
-
-        if (zones.size() > MAX_ZONES) {
-            List<Line> oneZoneLines = new ArrayList<>();
-            for (Zone zone : zones) {
-                oneZoneLines.addAll(zone.getLines());
-            }
-            return List.of(mergeLinesInZone(oneZoneLines, characterSpacing, lineSpacing));
-        }
-
-        return zones;
-    }
-
-
-    private double calculateMeanHeight(List<Line> lines) {
-
-        double meanHeight = 0.0;
-        double weights = 0.0;
-        for (Line line : lines) {
-            double weight = line.getLength();
-            meanHeight += line.getHeight() * weight;
-            weights += weight;
-        }
-        meanHeight /= weights;
-        return meanHeight;
-    }
-
-
-    private Zone mergeLinesInZone(List<Line> lines, double characterSpacing, double lineSpacing) {
-
-        double maxHorizontalDistance = 0;
-        double minVerticalDistance = 0;
-        double maxVerticalDistance = lineSpacing * MAX_VERTICAL_MERGE_DISTANCE;
-
-        DisjointSets<Line> sets = new DisjointSets<>(lines);
-
-        lines.forEach(outer -> {
-
-            lines.forEach(inner -> {
-                if (inner != outer) {
-
-                    double horizontalDistance = outer.horizontalDistance(inner);
-                    double verticalDistance = outer.verticalDistance(inner);
-
-                    if (horizontalDistance <= maxHorizontalDistance && minVerticalDistance <= verticalDistance && verticalDistance <= maxVerticalDistance) {
-                        sets.union(outer, inner);
-                    } else if (minVerticalDistance <= verticalDistance && verticalDistance <= maxVerticalDistance && Math.abs(horizontalDistance - Math.min(outer.getLength(),
-                            inner.getLength())) < 0.1) {
-                        boolean characterOverlap = false;
-                        int overlappingCount = 0;
-                        for (Character outerCharacter : outer.getCharacters()) {
-                            for (Character innerCharacter : inner.getCharacters()) {
-                                double characterOverlapDistance = outerCharacter.overlappingDistance(innerCharacter);
-                                if (characterOverlapDistance > 2) {
-                                    characterOverlap = true;
-                                }
-                                if (characterOverlapDistance > 0) {
-                                    overlappingCount++;
-                                }
-                            }
-                        }
-                        if (!characterOverlap && overlappingCount <= 2) {
-                            sets.union(outer, inner);
-                        }
-                    }
-                }
-            });
-        });
-
-        List<Line> outputZone = new ArrayList<>();
-        for (Set<Line> group : sets) {
-            List<Character> components = new ArrayList<>();
-            for (Line line : group) {
-                components.addAll(line.getCharacters());
-            }
-            components.sort(Comparator.comparingDouble(Character::getX));
-
-            outputZone.add(new Line(components, characterSpacing));
-        }
-        return new Zone(outputZone);
-    }
-
-}
@@ -1,15 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services.docstrum.utils;
-
-public class DoubleUtils {
-
-    public static int compareDouble(double d1, double d2, double precision) {
-
-        if (Double.isNaN(d1) || Double.isNaN(d2)) {
-            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);
-    }
-
-}
@@ -7,15 +7,12 @@ import static java.util.stream.Collectors.toList;
 import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
 import java.util.HashMap;
-import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
 import java.util.NoSuchElementException;
 import java.util.Set;
-import java.util.stream.Collectors;

-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.ClassificationDocument;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationFooter;
@@ -23,7 +20,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.Classification
 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.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.DuplicatedParagraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Footer;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.GenericSemanticNode;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Header;
@@ -53,9 +49,14 @@ public class DocumentGraphFactory {
        Document documentGraph = new Document();
        Context context = new Context(documentGraph);

-        document.getPages().forEach(context::buildAndAddPageWithCounter);
-        document.getSections().stream().flatMap(section -> section.getImages().stream()).forEach(image -> context.getImages().add(image));
-        addSections(document, context);
+        document.getPages()
+                .forEach(context::buildAndAddPageWithCounter);
+        document.getSections()
+                .stream()
+                .flatMap(section -> section.getImages()
+                        .stream())
+                .forEach(image -> context.getImages().add(image));
+        addSections(document, context, documentGraph);
        addHeaderAndFooterToEachPage(document, context);

        documentGraph.setNumberOfPages(context.pages.size());
@@ -66,9 +67,10 @@ public class DocumentGraphFactory {
    }


-    private void addSections(ClassificationDocument document, Context context) {
+    private void addSections(ClassificationDocument classificationDocument, Context context, Document document) {

-        document.getSections().forEach(section -> SectionNodeFactory.addSection(null, section.getNonEmptyPageBlocks(), section.getImages(), context));
+        classificationDocument.getSections()
+                .forEach(section -> SectionNodeFactory.addSection(null, section.getNonEmptyPageBlocks(), section.getImages(), context, document));
    }


@@ -78,11 +80,11 @@ public class DocumentGraphFactory {

        GenericSemanticNode node;
        if (originalTextBlock.isHeadline()) {
-            node = Headline.builder().documentTree(context.getDocumentTree()).build();
-        } else if (originalTextBlock.isToDuplicate()) {
-            node = DuplicatedParagraph.builder().documentTree(context.getDocumentTree()).build();
+            node = Headline.builder().documentTree(context.getDocumentTree())
+                    .build();
        } else {
-            node = Paragraph.builder().documentTree(context.getDocumentTree()).build();
+            node = Paragraph.builder().documentTree(context.getDocumentTree())
+                    .build();
        }

        page.getMainBody().add(node);
@@ -91,39 +93,35 @@ public class DocumentGraphFactory {
        textBlocks.add(originalTextBlock);
        textBlocks.addAll(textBlocksToMerge);
        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), node, context, page);
-
-        if (node instanceof DuplicatedParagraph duplicatedParagraph) {
-            AtomicTextBlock unsortedTextBlock = context.textBlockFactory.buildAtomicTextBlock(textBlocks.stream()
-                    .flatMap(tb -> tb.getSequences().stream())
-                    .collect(Collectors.toList()), node, context, page);
-            duplicatedParagraph.setUnsortedLeafTextBlock(unsortedTextBlock);
-        }
-
        List<Integer> treeId = context.documentTree.createNewChildEntryAndReturnId(parentNode, node);
        node.setLeafTextBlock(textBlock);
        node.setTreeId(treeId);
    }


-    public void addImage(Section section, ClassifiedImage image, Context context) {
+    public void addImage(GenericSemanticNode parent, ClassifiedImage image, Context context) {
+
+        Image imageNode = createImage(image, context);
+        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(parent, imageNode);
+        imageNode.setTreeId(treeId);
+        imageNode.setLeafTextBlock(context.textBlockFactory.emptyTextBlock(parent, context, context.getPage(image.getPage())));
+    }
+
+
+    private Image createImage(ClassifiedImage image, Context context) {

        Rectangle2D position = image.getPosition();
        Page page = context.getPage(image.getPage());
-        var imageBuilder = 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)
-                .documentTree(context.getDocumentTree());
-        if (image.isSourceByAi()) {
-            imageBuilder.engines(new HashSet<>(Set.of(LayoutEngine.AI)));
-        }
-        Image imageNode = imageBuilder.build();
+                .documentTree(context.getDocumentTree())
+                .build();
        page.getMainBody().add(imageNode);
-
-        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(section, imageNode);
-        imageNode.setTreeId(treeId);
+        return imageNode;
    }


@@ -162,11 +160,12 @@ public class DocumentGraphFactory {
    private void addFooter(List<TextPageBlock> textBlocks, Context context) {

        Page page = context.getPage(textBlocks.get(0).getPage());
-        Footer footer = Footer.builder().documentTree(context.getDocumentTree()).build();
+        Footer footer = Footer.builder().documentTree(context.getDocumentTree())
+                .build();
        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks),
-                footer,
-                context,
-                page);
+                                                                                  footer,
+                                                                                  context,
+                                                                                  page);
        List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(footer);
        footer.setTreeId(tocId);
        footer.setLeafTextBlock(textBlock);
@@ -177,7 +176,8 @@ public class DocumentGraphFactory {
    public void addHeader(List<TextPageBlock> textBlocks, Context context) {

        Page page = context.getPage(textBlocks.get(0).getPage());
-        Header header = Header.builder().documentTree(context.getDocumentTree()).build();
+        Header header = Header.builder().documentTree(context.getDocumentTree())
+                .build();
        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), header, 0, page);
        List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(header);
        header.setTreeId(tocId);
@@ -189,7 +189,8 @@ public class DocumentGraphFactory {
    private void addEmptyFooter(int pageIndex, Context context) {

        Page page = context.getPage(pageIndex);
-        Footer footer = Footer.builder().documentTree(context.getDocumentTree()).build();
+        Footer footer = Footer.builder().documentTree(context.getDocumentTree())
+                .build();
        AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(footer, context, page);
        List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(footer);
        footer.setTreeId(tocId);
@@ -201,7 +202,8 @@ public class DocumentGraphFactory {
    private void addEmptyHeader(int pageIndex, Context context) {

        Page page = context.getPage(pageIndex);
-        Header header = Header.builder().documentTree(context.getDocumentTree()).build();
+        Header header = Header.builder().documentTree(context.getDocumentTree())
+                .build();
        AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(header, 0, page);
        List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(header);
        header.setTreeId(tocId);
@@ -11,12 +11,13 @@ import java.util.Map;
 import java.util.Set;

 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.GenericSemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
 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.graph.nodes.GenericSemanticNode;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
 import com.knecon.fforesight.service.layoutparser.processor.utils.TableMergingUtility;

 import lombok.experimental.UtilityClass;
@@ -24,27 +25,46 @@ import lombok.experimental.UtilityClass;
@UtilityClass
 public class SectionNodeFactory {

-    public void addSection(GenericSemanticNode parentNode, List<AbstractPageBlock> pageBlocks, List<ClassifiedImage> images, DocumentGraphFactory.Context context) {
+    public void addSection(GenericSemanticNode parentNode,
+                           List<AbstractPageBlock> pageBlocks,
+                           List<ClassifiedImage> images,
+                           DocumentGraphFactory.Context context,
+                           Document document) {
+
+        // This is for the case where we have images on a page without any text/footer/header.
+        // The pageBlocks list is empty, but we still need to add those images to the document.
+        if (!images.isEmpty() && pageBlocks.isEmpty()) {
+            images.stream()
+                    .distinct()
+                    .forEach(image -> DocumentGraphFactory.addImage(document, image, context));
+            return;
+        }

        if (pageBlocks.isEmpty()) {
            return;
        }
-        Map<Integer, List<AbstractPageBlock>> blocksPerPage = pageBlocks.stream().collect(groupingBy(AbstractPageBlock::getPage));
-        Section section = Section.builder().documentTree(context.getDocumentTree()).build();
+
+        Map<Integer, List<AbstractPageBlock>> blocksPerPage = pageBlocks.stream()
+                .collect(groupingBy(AbstractPageBlock::getPage));
+        Section section = Section.builder().documentTree(context.getDocumentTree())
+                .build();

        context.getSections().add(section);
-        blocksPerPage.keySet().forEach(pageNumber -> addSectionNodeToPageNode(context, section, pageNumber));
+        blocksPerPage.keySet()
+                .forEach(pageNumber -> addSectionNodeToPageNode(context, section, pageNumber));

        section.setTreeId(getTreeId(parentNode, context, section));

-        addFirstHeadlineDirectlyToSection(pageBlocks, context, section);
+        addFirstHeadlineDirectlyToSection(pageBlocks, context, section, document);
        if (containsTablesAndTextBlocks(pageBlocks)) {
-            splitPageBlocksIntoSubSections(pageBlocks).forEach(subSectionPageBlocks -> addSection(section, subSectionPageBlocks, emptyList(), context));
+            splitPageBlocksIntoSubSections(pageBlocks).forEach(subSectionPageBlocks -> addSection(section, subSectionPageBlocks, emptyList(), context, document));
        } else {
-            addTablesAndParagraphsAndHeadlinesToSection(pageBlocks, context, section);
+            addTablesAndParagraphsAndHeadlinesToSection(pageBlocks, context, section, document);
        }

-        images.stream().distinct().forEach(image -> DocumentGraphFactory.addImage(section, image, context));
+        images.stream()
+                .distinct()
+                .forEach(image -> DocumentGraphFactory.addImage(section, image, context));
    }


@@ -58,16 +78,16 @@ public class SectionNodeFactory {
    }


-    private void addFirstHeadlineDirectlyToSection(List<AbstractPageBlock> pageBlocks, DocumentGraphFactory.Context context, Section section) {
+    private void addFirstHeadlineDirectlyToSection(List<AbstractPageBlock> pageBlocks, DocumentGraphFactory.Context context, Section section, Document document) {

        if (pageBlocks.get(0).isHeadline()) {
-            addTablesAndParagraphsAndHeadlinesToSection(List.of(pageBlocks.get(0)), context, section);
+            addTablesAndParagraphsAndHeadlinesToSection(List.of(pageBlocks.get(0)), context, section, document);
            pageBlocks.remove(0);
        }
    }


-    private void addTablesAndParagraphsAndHeadlinesToSection(List<AbstractPageBlock> pageBlocks, DocumentGraphFactory.Context context, Section section) {
+    private void addTablesAndParagraphsAndHeadlinesToSection(List<AbstractPageBlock> pageBlocks, DocumentGraphFactory.Context context, Section section, Document document) {

        Set<AbstractPageBlock> alreadyMerged = new HashSet<>();
        List<AbstractPageBlock> remainingBlocks = new LinkedList<>(pageBlocks);
@@ -86,7 +106,7 @@ public class SectionNodeFactory {
            } else if (abstractPageBlock instanceof TablePageBlock tablePageBlock) {
                List<TablePageBlock> tablesToMerge = TableMergingUtility.findConsecutiveTablesWithSameColCountAndSameHeaders(tablePageBlock, remainingBlocks);
                alreadyMerged.addAll(tablesToMerge);
-                TableNodeFactory.addTable(section, tablesToMerge, context);
+                TableNodeFactory.addTable(section, tablesToMerge, context, document);
            } else {
                throw new RuntimeException(format("Unhandled AbstractPageBlockType %s!", abstractPageBlock.getClass()));
            }
@@ -96,7 +116,9 @@ public class SectionNodeFactory {

    private boolean containsTablesAndTextBlocks(List<AbstractPageBlock> pageBlocks) {

-        return pageBlocks.stream().anyMatch(pageBlock -> pageBlock instanceof TablePageBlock) && pageBlocks.stream().anyMatch(pageBlock -> pageBlock instanceof TextPageBlock);
+        return pageBlocks.stream()
+                       .anyMatch(pageBlock -> pageBlock instanceof TablePageBlock) && pageBlocks.stream()
+                       .anyMatch(pageBlock -> pageBlock instanceof TextPageBlock);
    }


@@ -112,7 +134,9 @@ public class SectionNodeFactory {

        List<List<AbstractPageBlock>> splitList = splitIntoCoherentList(pageBlocks);
        movePrecedingHeadlineToTableList(splitList);
-        return splitList.stream().filter(list -> !list.isEmpty()).toList();
+        return splitList.stream()
+                .filter(list -> !list.isEmpty())
+                .toList();
    }


@@ -133,7 +157,8 @@ public class SectionNodeFactory {

    private boolean listIsTablesOnly(List<AbstractPageBlock> abstractPageBlocks) {

-        return abstractPageBlocks.stream().allMatch(abstractPageBlock -> abstractPageBlock instanceof TablePageBlock);
+        return abstractPageBlocks.stream()
+                .allMatch(abstractPageBlock -> abstractPageBlock instanceof TablePageBlock);
    }


@@ -171,7 +196,6 @@ public class SectionNodeFactory {
                .filter(abstractTextContainer -> abstractTextContainer.intersectsY(atc))
                .map(abstractTextContainer -> (TextPageBlock) abstractTextContainer)
                .filter(abstractTextContainer -> abstractTextContainer.getDir() == atc.getDir())
-                .filter(abstractTextContainer -> !abstractTextContainer.isToDuplicate())
                .toList();
    }

@@ -8,6 +8,7 @@ import java.util.Set;
 import java.util.stream.Collectors;

 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.Cell;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
@@ -27,23 +28,26 @@ public class TableNodeFactory {
    public final double TABLE_CELL_MERGE_CONTENTS_SIZE_THRESHOLD = 0.05;


-    public void addTable(GenericSemanticNode parentNode, List<TablePageBlock> tablesToMerge, DocumentGraphFactory.Context context) {
+    public void addTable(GenericSemanticNode parentNode, List<TablePageBlock> tablesToMerge, DocumentGraphFactory.Context context, Document document) {

        setPageNumberInCells(tablesToMerge);
-        Set<Page> pages = tablesToMerge.stream().map(AbstractPageBlock::getPage).map(context::getPage).collect(Collectors.toSet());
-        List<List<Cell>> mergedRows = tablesToMerge.stream().map(TablePageBlock::getRows).flatMap(Collection::stream).toList();
+        Set<Page> pages = tablesToMerge.stream()
+                .map(AbstractPageBlock::getPage)
+                .map(context::getPage)
+                .collect(Collectors.toSet());
+        List<List<Cell>> mergedRows = tablesToMerge.stream()
+                .map(TablePageBlock::getRows)
+                .flatMap(Collection::stream)
+                .toList();

-        Table table = Table.builder()
-                .documentTree(context.getDocumentTree())
-                .numberOfCols(mergedRows.isEmpty() ? 0 : mergedRows.get(0).size())
-                .numberOfRows(mergedRows.size())
+        Table table = Table.builder().documentTree(context.getDocumentTree()).numberOfCols(mergedRows.isEmpty() ? 0 : mergedRows.get(0).size()).numberOfRows(mergedRows.size())
                .build();

        pages.forEach(page -> addTableToPage(page, parentNode, table));

        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, table);
        table.setTreeId(treeId);
-        addTableCells(mergedRows, table, context);
+        addTableCells(mergedRows, table, context, document);

        ifTableHasNoHeadersSetFirstRowAsHeaders(table);
    }
@@ -63,7 +67,8 @@ public class TableNodeFactory {

    private void setPageNumberInTextBlocksWithPageNumberSetTo0(TablePageBlock table, Cell cell) {

-        cell.getTextBlocks().stream()//
+        cell.getTextBlocks()
+                .stream()//
                .filter(tb -> tb.getPage() == 0)//
                .forEach(tb -> tb.setPage(table.getPage()));
    }
@@ -82,28 +87,32 @@ public class TableNodeFactory {

    private void ifTableHasNoHeadersSetFirstRowAsHeaders(Table table) {

-        if (table.streamHeaders().findAny().isEmpty()) {
-            table.streamRow(0).forEach(tableCellNode -> tableCellNode.setHeader(true));
+        if (table.streamHeaders()
+                .findAny().isEmpty()) {
+            table.streamRow(0)
+                    .forEach(tableCellNode -> tableCellNode.setHeader(true));
        }
    }


-    private void addTableCells(List<List<Cell>> rows, Table table, DocumentGraphFactory.Context context) {
+    private void addTableCells(List<List<Cell>> rows, Table table, DocumentGraphFactory.Context context, Document document) {

        for (int rowIndex = 0; rowIndex < rows.size(); rowIndex++) {
            for (int colIndex = 0; colIndex < rows.get(rowIndex).size(); colIndex++) {
-                addTableCell(rows.get(rowIndex).get(colIndex), rowIndex, colIndex, table, context);
+                addTableCell(rows.get(rowIndex)
+                                     .get(colIndex), rowIndex, colIndex, table, context, document);
            }
        }
    }


    @SuppressWarnings("PMD.UnusedPrivateMethod") // PMD actually flags this wrong
-    private void addTableCell(Cell cell, int rowIndex, int colIndex, Table tableNode, DocumentGraphFactory.Context context) {
+    private void addTableCell(Cell cell, int rowIndex, int colIndex, Table tableNode, DocumentGraphFactory.Context context, Document document) {

        Page page = context.getPage(cell.getPageNumber());

-        TableCell tableCell = TableCell.builder().documentTree(context.getDocumentTree()).row(rowIndex).col(colIndex).header(cell.isHeaderCell()).bBox(cell.getBounds2D()).build();
+        TableCell tableCell = TableCell.builder().documentTree(context.getDocumentTree()).row(rowIndex).col(colIndex).header(cell.isHeaderCell()).bBox(cell.getBounds2D())
+                .build();
        page.getMainBody().add(tableCell);

        List<Integer> treeId = context.getDocumentTree().createNewTableChildEntryAndReturnId(tableNode, tableCell);
@@ -113,16 +122,26 @@ public class TableNodeFactory {
        if (cell.getTextBlocks().isEmpty()) {
            tableCell.setLeafTextBlock(context.getTextBlockFactory().emptyTextBlock(tableNode, context, page));
        } else if (cell.getTextBlocks().size() == 1) {
-            textBlock = context.getTextBlockFactory().buildAtomicTextBlock(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(tableCell, cell.getTextBlocks().stream().map(tb -> (AbstractPageBlock) tb).toList(), emptyList(), context);
+            SectionNodeFactory.addSection(tableCell,
+                                          cell.getTextBlocks()
+                                                  .stream()
+                                                  .map(tb -> (AbstractPageBlock) tb)
+                                                  .toList(),
+                                          emptyList(),
+                                          context,
+                                          document);
        } else if (cellAreaIsSmallerThanPageAreaTimesThreshold(cell, page)) {
            List<TextPositionSequence> sequences = TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(cell.getTextBlocks());
            textBlock = context.getTextBlockFactory().buildAtomicTextBlock(sequences, tableCell, context, page);
            tableCell.setLeafTextBlock(textBlock);
        } else {
-            cell.getTextBlocks().forEach(tb -> DocumentGraphFactory.addParagraphOrHeadline(tableCell, tb, context, emptyList()));
+            cell.getTextBlocks()
+                    .forEach(tb -> DocumentGraphFactory.addParagraphOrHeadline(tableCell, tb, context, emptyList()));
        }
    }

@@ -135,7 +154,8 @@ public class TableNodeFactory {

    private boolean firstTextBlockIsHeadline(Cell cell) {

-        return cell.getTextBlocks().get(0).isHeadline();
+        return cell.getTextBlocks()
+                .get(0).isHeadline();
    }

 }
@@ -2,7 +2,6 @@ package com.knecon.fforesight.service.layoutparser.processor.services.mapper;

 import java.awt.geom.Rectangle2D;
 import java.util.HashMap;
-import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
 import java.util.Set;
@@ -13,10 +12,8 @@ import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.Do
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentPositionData;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentStructure;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentTextData;
-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.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.DuplicatedParagraph;
 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.Table;
@@ -76,24 +73,17 @@ public class DocumentDataMapper {
            case TABLE -> PropertiesMapper.buildTableProperties((Table) entry.getNode());
            case TABLE_CELL -> PropertiesMapper.buildTableCellProperties((TableCell) entry.getNode());
            case IMAGE -> PropertiesMapper.buildImageProperties((Image) entry.getNode());
-            case PARAGRAPH ->
-                    entry.getNode() instanceof DuplicatedParagraph duplicatedParagraph ? PropertiesMapper.buildDuplicateParagraphProperties(duplicatedParagraph) : new HashMap<>();
            default -> new HashMap<>();
        };

-        DocumentStructure.EntryData.EntryDataBuilder documentBuilder = DocumentStructure.EntryData.builder()
+        return DocumentStructure.EntryData.builder()
                .treeId(toPrimitiveIntArray(entry.getTreeId()))
                .children(entry.getChildren().stream().map(DocumentDataMapper::toEntryData).toList())
                .type(entry.getType())
                .atomicBlockIds(atomicTextBlocks)
                .pageNumbers(entry.getNode().getPages().stream().map(Page::getNumber).map(Integer::longValue).toArray(Long[]::new))
-                .properties(properties);
-        if (entry.getNode() != null) {
-            documentBuilder.engines(entry.getNode().getEngines());
-        } else {
-            documentBuilder.engines(new HashSet<>(Set.of(LayoutEngine.ALGORITHM)));
-        }
-        return documentBuilder.build();
+                .properties(properties)
+                .build();
    }


@@ -7,14 +7,13 @@ import java.util.List;
 import java.util.Map;
 import java.util.NoSuchElementException;

-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentData;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentPage;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentPositionData;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentStructure;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentTextData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentStructure;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentPage;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
 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;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Header;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Headline;
@@ -62,7 +61,7 @@ public class DocumentGraphMapper {

            SemanticNode node = switch (entryData.getType()) {
                case SECTION -> buildSection(context);
-                case PARAGRAPH -> buildParagraph(context, entryData.getProperties());
+                case PARAGRAPH -> buildParagraph(context);
                case HEADLINE -> buildHeadline(context);
                case HEADER -> buildHeader(context);
                case FOOTER -> buildFooter(context);
@@ -141,17 +140,7 @@ public class DocumentGraphMapper {
    }


-    private Paragraph buildParagraph(Context context, Map<String, String> properties) {
-
-        if (PropertiesMapper.isDuplicateParagraph(properties)) {
-
-            DuplicatedParagraph duplicatedParagraph = DuplicatedParagraph.builder().documentTree(context.documentTree).build();
-
-            Long[] unsortedTextblockIds = PropertiesMapper.getUnsortedTextblockIds(properties);
-            duplicatedParagraph.setUnsortedLeafTextBlock(toTextBlock(unsortedTextblockIds, context, duplicatedParagraph));
-            return duplicatedParagraph;
-
-        }
+    private Paragraph buildParagraph(Context context) {

        return Paragraph.builder().documentTree(context.documentTree).build();
    }
@@ -1,19 +1,17 @@
 package com.knecon.fforesight.service.layoutparser.processor.services.mapper;

 import java.awt.geom.Rectangle2D;
-import java.util.Arrays;
+import java.util.Collections;
 import java.util.HashMap;
 import java.util.Locale;
 import java.util.Map;

 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentStructure;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.DuplicatedParagraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Image;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.ImageType;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;

 public class PropertiesMapper {

@@ -78,32 +76,6 @@ public class PropertiesMapper {
    }


-    public static Map<String, String> buildDuplicateParagraphProperties(DuplicatedParagraph duplicatedParagraph) {
-
-        Map<String, String> properties = new HashMap<>();
-        properties.put(DocumentStructure.DuplicateParagraphProperties.UNSORTED_TEXTBLOCK_ID, Arrays.toString(toAtomicTextBlockIds(duplicatedParagraph.getUnsortedLeafTextBlock())));
-        return properties;
-    }
-
-
-    public static boolean isDuplicateParagraph(Map<String, String> properties) {
-
-        return properties.containsKey(DocumentStructure.DuplicateParagraphProperties.UNSORTED_TEXTBLOCK_ID);
-    }
-
-
-    public static Long[] getUnsortedTextblockIds(Map<String, String> properties) {
-
-        return toLongArray(properties.get(DocumentStructure.DuplicateParagraphProperties.UNSORTED_TEXTBLOCK_ID));
-    }
-
-
-    public static Long[] toLongArray(String ids) {
-
-        return Arrays.stream(ids.substring(1, ids.length() - 1).trim().split(",")).map(Long::valueOf).toArray(Long[]::new);
-    }
-
-
    private static ImageType parseImageType(String imageType) {

        return switch (imageType) {
@@ -129,10 +101,4 @@ public class PropertiesMapper {
                rectangle2D.getHeight());
    }

-
-    private static Long[] toAtomicTextBlockIds(TextBlock textBlock) {
-
-        return textBlock.getAtomicTextBlocks().stream().map(AtomicTextBlock::getId).toArray(Long[]::new);
-    }
-
 }
@@ -237,13 +237,8 @@ public class PDFLinesTextStripper extends PDFTextStripper {
        int startIndex = 0;
        RedTextPosition previous = null;

-        float direction = -1;
        for (int i = 0; i <= textPositions.size() - 1; i++) {

-            if (direction == -1) {
-                direction = textPositions.get(i).getDir();
-            }
-
            if (!textPositionSequences.isEmpty()) {
                previous = textPositionSequences.get(textPositionSequences.size() - 1)
                        .getTextPositions()
@@ -255,13 +250,6 @@ public class PDFLinesTextStripper extends PDFTextStripper {
                continue;
            }

-            if (textPositions.get(i).getDir() != direction && startIndex != i) {
-                List<TextPosition> sublist = textPositions.subList(startIndex, i);
-                textPositionSequences.add(new TextPositionSequence(sublist, pageNumber, i == textPositions.size() - 1 && isParagraphStart));
-                startIndex = i;
-                direction = textPositions.get(i).getDir();
-            }
-
            // Strange but sometimes this is happening, for example: Metolachlor2.pdf
            if (checkIfCurrentPositionIsToTheRightOfPreviousPosition(i, textPositions)) {
                List<TextPosition> sublist = textPositions.subList(startIndex, i);
@@ -341,7 +329,6 @@ public class PDFLinesTextStripper extends PDFTextStripper {
                .getXDirAdj() - (previous.getXDirAdj() + previous.getWidthDirAdj()) < maximumGapSize;
    }

-
    @Override
    public String getText(PDDocument doc) throws IOException {

@@ -17,10 +17,8 @@ import java.util.stream.Stream;

 import org.springframework.stereotype.Service;

-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.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.DuplicatedParagraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
@@ -54,8 +52,6 @@ public class LayoutGridService {

    static Color INNER_LINES_COLOR = new Color(255, 175, 175);
    static Color PARAGRAPH_COLOR = new Color(70, 130, 180);
-
-    static Color DUPLICATE_PARAGRAPH_COLOR = new Color(70, 180, 101);
    static Color TABLE_COLOR = new Color(102, 205, 170);
    static Color SECTION_COLOR = new Color(50, 50, 50);
    static Color HEADLINE_COLOR = new Color(162, 56, 56);
@@ -67,58 +63,43 @@ public class LayoutGridService {
    @Observed(name = "ViewerDocumentService", contextualName = "create-viewer-document")
    public void addLayoutGrid(File originFile, Document document, File destinationFile, boolean layerVisibilityDefaultValue) {

-        Visualizations layoutGrid = this.addLayoutGrid(document, layerVisibilityDefaultValue, false);
-        Visualizations visualLayoutGrid = this.addLayoutGrid(document, layerVisibilityDefaultValue, true);
+        LayoutGrid layoutGrid = createLayoutGrid(document);

-        viewerDocumentService.addVisualizationsOnPage(originFile, destinationFile, List.of(layoutGrid, visualLayoutGrid));
+        viewerDocumentService.addVisualizationsOnPage(originFile,
+                destinationFile,
+                Visualizations.builder()
+                        .layer(ContentStreams.KNECON_LAYOUT)
+                        .visualizationsOnPages(layoutGrid.getVisualizationsPerPages())
+                        .layerVisibilityDefaultValue(layerVisibilityDefaultValue)
+                        .build());
    }


-    @SneakyThrows
-    @Observed(name = "ViewerDocumentService", contextualName = "create-viewer-document")
-    public Visualizations addLayoutGrid(Document document, boolean layerVisibilityDefaultValue, boolean visualParsingGrid) {
-
-        LayoutGrid layoutGrid = createLayoutGrid(document, visualParsingGrid);
-
-        return Visualizations.builder()
-                .layer(visualParsingGrid ? ContentStreams.KNECON_VISUAL_PARSING : ContentStreams.KNECON_LAYOUT)
-                .visualizationsOnPages(layoutGrid.getVisualizationsPerPages())
-                .layerVisibilityDefaultValue(layerVisibilityDefaultValue)
-                .build();
-    }
-
-
-    private LayoutGrid createLayoutGrid(Document document, boolean visualParsingGrid) {
+    private LayoutGrid createLayoutGrid(Document document) {

        LayoutGrid layoutGrid = new LayoutGrid(document.getNumberOfPages());
-        document.streamAllSubNodes()
-                .filter(node -> (node.getEngines().contains(LayoutEngine.AI) && visualParsingGrid) || (node.getEngines().contains(LayoutEngine.ALGORITHM) && !visualParsingGrid))
-                .forEach(semanticNode -> {
-                    Color color = switch (semanticNode.getType()) {
-                        case PARAGRAPH -> PARAGRAPH_COLOR;
-                        case TABLE -> TABLE_COLOR;
-                        case SECTION -> SECTION_COLOR;
-                        case HEADLINE -> HEADLINE_COLOR;
-                        case HEADER, FOOTER -> HEADER_COLOR;
-                        case IMAGE -> IMAGE_COLOR;
-                        default -> null;
-                    };
+        document.streamAllSubNodes().forEach(semanticNode -> {
+            Color color = switch (semanticNode.getType()) {
+                case PARAGRAPH -> PARAGRAPH_COLOR;
+                case TABLE -> TABLE_COLOR;
+                case SECTION -> SECTION_COLOR;
+                case HEADLINE -> HEADLINE_COLOR;
+                case HEADER, FOOTER -> HEADER_COLOR;
+                case IMAGE -> IMAGE_COLOR;
+                default -> null;
+            };
+            if (isNotSectionOrTableCellOrDocument(semanticNode)) {
+                addAsRectangle(semanticNode, layoutGrid, color);
+            }
+            if (semanticNode.getType().equals(NodeType.SECTION)) {
+                addSection(semanticNode, layoutGrid, color);
+            }
+            if (semanticNode.getType().equals(NodeType.TABLE)) {
+                Table table = (Table) semanticNode;
+                addInnerTableLines(table, layoutGrid);
+            }
+        });

-                    if (semanticNode instanceof DuplicatedParagraph) {
-                        color = DUPLICATE_PARAGRAPH_COLOR;
-                    }
-
-                    if (isNotSectionOrTableCellOrDocument(semanticNode)) {
-                        addAsRectangle(semanticNode, layoutGrid, color);
-                    }
-                    if (semanticNode.getType().equals(NodeType.SECTION)) {
-                        addSection(semanticNode, layoutGrid, color);
-                    }
-                    if (semanticNode.getType().equals(NodeType.TABLE)) {
-                        Table table = (Table) semanticNode;
-                        addInnerTableLines(table, layoutGrid);
-                    }
-                });
        return layoutGrid;
    }

@@ -129,11 +110,13 @@ public class LayoutGridService {
            return;
        }
        for (Page page : table.getPages()) {
+
            Optional<Integer> optionalFirstRowOnPage = table.streamCol(0).filter(tableCell -> tableCell.isOnPage(page.getNumber())).map(TableCell::getRow).findFirst();
            if (optionalFirstRowOnPage.isEmpty()) {
                continue;
            }
            int firstRowOnPage = optionalFirstRowOnPage.get();
+
            Stream<Double> xStream = switch (page.getRotation()) {
                case 90 -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMinX);
                case 180 -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMaxX);
@@ -142,6 +125,7 @@ public class LayoutGridService {
            };
            List<Double> xs = xStream.collect(Collectors.toList());
            xs.remove(0);
+
            Stream<Double> yStream = switch (page.getRotation()) {
                case 90 -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMinY);
                case 180 -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMinY);
@@ -151,7 +135,7 @@ public class LayoutGridService {
            List<Double> ys = yStream.collect(Collectors.toList());
            ys.remove(0);

-            Rectangle2D tableBBox = table.getBBox().get(table.getFirstPage());
+            Rectangle2D tableBBox = table.getBBox().get(page);
            List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(page.getNumber() - 1).getColoredLines();
            xs.forEach(x -> {
                Line2D line = new Line2D.Double(new Point2D.Double(x, tableBBox.getMaxY()), new Point2D.Double(x, tableBBox.getMinY()));
@@ -3,8 +3,10 @@ package com.knecon.fforesight.service.layoutparser.processor.utils;
 import static java.lang.String.format;

 import java.awt.geom.Area;
+import java.awt.geom.Point2D;
 import java.awt.geom.Rectangle2D;
 import java.awt.geom.RectangularShape;
+import java.util.ArrayList;
 import java.util.Collections;
 import java.util.LinkedList;
 import java.util.List;
@@ -20,6 +22,8 @@ import org.apache.pdfbox.pdmodel.common.PDRectangle;
 import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Point;
 import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;

 import lombok.AllArgsConstructor;
 import lombok.NoArgsConstructor;
@@ -149,7 +153,27 @@ public class RectangleTransformations {
                previousRectangle = currentRectangle;
            }
        }
-        return rectangleListsWithGaps.stream().map(RectangleTransformations::rectangle2DBBox).toList();
+        return rectangleListsWithGaps.stream()
+                .map(RectangleTransformations::rectangle2DBBox)
+                .toList();
+    }
+
+
+    public static CleanRulings extractRulings(List<? extends Rectangle2D.Float> rectangles) {
+
+        // Underlined or strikethrough are also in rulings but we dont want to split blocks with them so we use cells.
+        List<Ruling> horizontalRulings = new ArrayList<>();
+        List<Ruling> verticalRulings = new ArrayList<>();
+
+        rectangles.forEach(rectangle -> {
+            horizontalRulings.add(new Ruling(new Point2D.Float(rectangle.x, rectangle.y), new Point2D.Float(rectangle.x + rectangle.width, rectangle.y)));
+            horizontalRulings.add(new Ruling(new Point2D.Float(rectangle.x, rectangle.y + rectangle.height),
+                                             new Point2D.Float(rectangle.x + rectangle.width, rectangle.y + rectangle.height)));
+            verticalRulings.add(new Ruling(new Point2D.Float(rectangle.x, rectangle.y), new Point2D.Float(rectangle.x, rectangle.y + rectangle.height)));
+            verticalRulings.add(new Ruling(new Point2D.Float(rectangle.x + rectangle.width, rectangle.y),
+                                           new Point2D.Float(rectangle.x + rectangle.width, rectangle.y + rectangle.height)));
+        });
+        return CleanRulings.builder().vertical(verticalRulings).horizontal(horizontalRulings).build();
    }


@@ -3,6 +3,7 @@ package com.knecon.fforesight.service.layoutparser.processor.utils;
 import java.util.List;
 import java.util.stream.Collectors;

+
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;

@@ -28,15 +28,13 @@ import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPosit
 *
 * @author Ben Litchfield
 */
-public class TextPositionSequenceComparator implements Comparator<TextPositionSequence>
-{
+public class TextPositionSequenceComparator implements Comparator<TextPositionSequence> {
+
    @Override
-    public int compare(TextPositionSequence pos1, TextPositionSequence pos2)
-    {
+    public int compare(TextPositionSequence pos1, TextPositionSequence pos2) {
        // only compare text that is in the same direction
        int cmp1 = Float.compare(pos1.getDir().getDegrees(), pos2.getDir().getDegrees());
-        if (cmp1 != 0)
-        {
+        if (cmp1 != 0) {
            return cmp1;
        }

@@ -48,25 +46,19 @@ public class TextPositionSequenceComparator implements Comparator<TextPositionSe
        float pos2YBottom = pos2.getMaxYDirAdj();

        // note that the coordinates have been adjusted so 0,0 is in upper left
-        float pos1YTop = pos1YBottom - pos1.getTextHeight();
-        float pos2YTop = pos2YBottom - pos2.getTextHeight();
+        float pos1YTop = pos1YBottom - pos1.getTextHeightNoPadding();
+        float pos2YTop = pos2YBottom - pos2.getTextHeightNoPadding();

        float yDifference = Math.abs(pos1YBottom - pos2YBottom);

        // we will do a simple tolerance comparison
-        if (yDifference < .1 ||
-                pos2YBottom >= pos1YTop && pos2YBottom <= pos1YBottom ||
-                pos1YBottom >= pos2YTop && pos1YBottom <= pos2YBottom)
-        {
+        if (yDifference < .1 || pos2YBottom >= pos1YTop && pos2YBottom <= pos1YBottom || pos1YBottom >= pos2YTop && pos1YBottom <= pos2YBottom) {
            return Float.compare(x1, x2);
-        }
-        else if (pos1YBottom < pos2YBottom)
-        {
+        } else if (pos1YBottom < pos2YBottom) {
            return -1;
-        }
-        else
-        {
+        } else {
            return 1;
        }
    }
+
 }
@@ -27,7 +27,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Do
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
 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.layoutparser.processor.services.factory.DocumentGraphFactory;
 import com.knecon.fforesight.service.layoutparser.processor.services.mapper.TaasDocumentDataMapper;
 import com.knecon.fforesight.service.layoutparser.processor.utils.DrawingOptions;
@@ -52,7 +51,6 @@ public class BdrJsonBuildTest extends AbstractTest {
                file,
                new ImageServiceResponse(),
                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
                file.toString()));
    }

@@ -33,7 +33,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Do
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
 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.layoutparser.processor.services.factory.DocumentGraphFactory;
 import com.knecon.fforesight.tenantcommons.TenantsClient;

@@ -99,7 +98,6 @@ public class HeadlinesGoldStandardIntegrationTest {
                pdfFileResource.getFile(),
                new ImageServiceResponse(),
                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
                filePath));

        var foundHeadlines = documentGraph.streamAllSubNodes()
@@ -28,7 +28,20 @@ public class LayoutparserEnd2EndTest extends AbstractTest {
        prepareStorage("files/bdr/Wie weiter bei Kristeneinrichtungen.pdf");
        LayoutParsingRequest layoutParsingRequest = buildDefaultLayoutParsingRequest(LayoutParsingType.REDACT_MANAGER);
        LayoutParsingFinishedEvent finishedEvent = layoutParsingPipeline.parseLayoutAndSaveFilesToStorage(layoutParsingRequest);
-        Arrays.stream(finishedEvent.message().split("\n")).forEach(log::info);
+        Arrays.stream(finishedEvent.message().split("\n"))
+                .forEach(log::info);
+    }
+
+
+    @Test
+    @SneakyThrows
+    public void testLayoutParserEndToEnd_RED_8747() {
+
+        prepareStorage("files/SinglePages/MergedEntities.pdf");
+        LayoutParsingRequest layoutParsingRequest = buildDefaultLayoutParsingRequest(LayoutParsingType.REDACT_MANAGER);
+        LayoutParsingFinishedEvent finishedEvent = layoutParsingPipeline.parseLayoutAndSaveFilesToStorage(layoutParsingRequest);
+        Arrays.stream(finishedEvent.message().split("\n"))
+                .forEach(log::info);
    }

 }
@@ -16,7 +16,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsi
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 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.layoutparser.processor.services.factory.DocumentGraphFactory;
 import com.knecon.fforesight.service.layoutparser.processor.services.mapper.DocumentDataMapper;
 import com.knecon.fforesight.service.layoutparser.server.utils.BuildDocumentTest;
@@ -59,7 +58,6 @@ public class DocumentGraphJsonWritingTest extends BuildDocumentTest {
                filename.toFile(),
                new ImageServiceResponse(),
                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
                filename.toFile().toString()));

        DocumentData documentData = DocumentDataMapper.toDocumentData(documentGraph);
@@ -12,7 +12,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsi
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 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.layoutparser.processor.services.factory.DocumentGraphFactory;
 import com.knecon.fforesight.service.layoutparser.processor.services.visualization.LayoutGridService;
 import com.knecon.fforesight.service.layoutparser.server.utils.BuildDocumentTest;
@@ -26,14 +25,14 @@ public class ViewerDocumentTest extends BuildDocumentTest {
    @SneakyThrows
    public void testViewerDocument() {

-        String fileName = "files/new/270 rotated text on non rotated pages.pdf";
+        String fileName = "files/SinglePages/T5 VV-640252-Page16.pdf";
        String tmpFileName = "/tmp/" + Path.of(fileName).getFileName() + "_VIEWER.pdf";

        var documentFile = new ClassPathResource(fileName).getFile();
        ViewerDocumentService viewerDocumentService = new ViewerDocumentService(null);
        LayoutGridService layoutGridService = new LayoutGridService(viewerDocumentService);

-        Document document = buildGraph(fileName, LayoutParsingType.DOCSTRUM_XY);
+        Document document = buildGraph(fileName, LayoutParsingType.REDACT_MANAGER);
        long start = System.currentTimeMillis();
        layoutGridService.addLayoutGrid(documentFile, document, new File(tmpFileName), true);
        System.out.printf("Total time: %.2fs%n", ((float) (System.currentTimeMillis() - start)) / 1000);
@@ -54,11 +53,10 @@ public class ViewerDocumentTest extends BuildDocumentTest {
        var documentFile = new ClassPathResource(fileName).getFile();

        var classificationDocument = layoutParsingPipeline.parseLayout(LayoutParsingType.DOCUMINE,
-                documentFile,
-                new ImageServiceResponse(),
-                tableResponse,
-                new VisualLayoutParsingResponse(),
-                Path.of(fileName).getFileName().toFile().toString());
+                                                                       documentFile,
+                                                                       new ImageServiceResponse(),
+                                                                       tableResponse,
+                                                                       Path.of(fileName).getFileName().toFile().toString());
        ViewerDocumentService viewerDocumentService = new ViewerDocumentService(null);
        LayoutGridService layoutGridService = new LayoutGridService(viewerDocumentService);
        Document document = DocumentGraphFactory.buildDocumentGraph(classificationDocument);
@@ -31,7 +31,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.table.Cell;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
 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.layoutparser.processor.services.SectionsBuilderService;
 import com.knecon.fforesight.service.layoutparser.processor.services.classification.RedactManagerClassificationService;
 import com.knecon.fforesight.service.layoutparser.server.utils.AbstractTest;
@@ -60,7 +59,6 @@ public class PdfSegmentationServiceTest extends AbstractTest {
                                                                                          originDocument,
                                                                                          new ImageServiceResponse(),
                                                                                          tableServiceResponse,
-                                                                                          new VisualLayoutParsingResponse(),
                                                                                          "document");

        redactManagerClassificationService.classifyDocument(classificationDocument);
@@ -753,6 +751,19 @@ public class PdfSegmentationServiceTest extends AbstractTest {

    }

+    @Test
+    public void testMergedEntities_Page26() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/SinglePages/MergedEntities.pdf");
+
+        ClassificationDocument document = buildClassificationDocument(pdfFileResource.getFile());
+
+        validateTableSize(document, 1);
+
+        validateTable(document, 0, 6, 6, 5, 0);
+
+    }
+

    @SneakyThrows
    private void toHtml(ClassificationDocument document, String filename) {
@@ -25,7 +25,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Ta
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 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.layoutparser.processor.services.PageContentExtractor;
 import com.knecon.fforesight.service.layoutparser.processor.services.RulingCleaningService;
 import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
@@ -103,13 +102,11 @@ public class RulingCleaningServiceTest extends BuildDocumentTest {
                filename.toFile(),
                new ImageServiceResponse(),
                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
                filename.toFile().toString()));
        Document documentGraphAfter = DocumentGraphFactory.buildDocumentGraph(layoutParsingPipeline.parseLayout(LayoutParsingType.REDACT_MANAGER,
                filename.toFile(),
                new ImageServiceResponse(),
                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
                filename.toFile().toString()));
        DocumentData documentDataBefore = DocumentDataMapper.toDocumentData(documentGraphBefore);
        DocumentData documentDataAfter = DocumentDataMapper.toDocumentData(documentGraphAfter);
@@ -20,7 +20,6 @@ import org.springframework.context.annotation.Import;
 import org.springframework.context.annotation.Primary;
 import org.springframework.core.io.ClassPathResource;
 import org.springframework.test.context.junit.jupiter.SpringExtension;
-import org.xmlunit.builder.Input;

 import com.iqser.red.commons.jackson.ObjectMapperFactory;
 import com.iqser.red.storage.commons.service.StorageService;
@@ -52,8 +51,6 @@ public abstract class AbstractTest {
    private RabbitTemplate rabbitTemplate;

    protected final static String ORIGIN_FILE_ID = "origin";
-
-    protected final static String VISUAL_LAYOUT_FILE = "visual";
    protected final static String TABLE_FILE_ID = "table";
    protected final static String IMAGE_FILE_ID = "image";
    protected final static String STRUCTURE_FILE_ID = "structure";
@@ -99,7 +96,7 @@ public abstract class AbstractTest {
    @SneakyThrows
    protected LayoutParsingRequest prepareStorage(String file) {

-        return prepareStorage(file, "cv_table_parsing_response/empty.json", "image_service_response/empty.json","visual_layout_parsing_response/empty.json");
+        return prepareStorage(file, "cv_table_parsing_response/empty.json", "image_service_response/empty.json");
    }


@@ -119,7 +116,6 @@ public abstract class AbstractTest {
                .originFileStorageId(ORIGIN_FILE_ID)
                .tablesFileStorageId(Optional.of(TABLE_FILE_ID))
                .imagesFileStorageId(Optional.of(IMAGE_FILE_ID))
-                .visualLayoutParsingFileId(Optional.of(VISUAL_LAYOUT_FILE))
                .structureFileStorageId(STRUCTURE_FILE_ID)
                .textBlockFileStorageId(TEXT_FILE_ID)
                .positionBlockFileStorageId(POSITION_FILE_ID)
@@ -140,16 +136,6 @@ public abstract class AbstractTest {
        return prepareStorage(pdfFileResource.getInputStream(), cvServiceResponseFileResource.getInputStream(), imageInfoFileResource.getInputStream());
    }

-    @SneakyThrows
-    protected LayoutParsingRequest prepareStorage(String file, String cvServiceResponseFile, String imageInfoFile, String visualLayoutParsingResponseFile) {
-
-        ClassPathResource pdfFileResource = new ClassPathResource(file);
-        ClassPathResource cvServiceResponseFileResource = new ClassPathResource(cvServiceResponseFile);
-        ClassPathResource imageInfoFileResource = new ClassPathResource(imageInfoFile);
-        ClassPathResource visualLayoutParsingResponseResource = new ClassPathResource(visualLayoutParsingResponseFile);
-
-        return prepareStorage(pdfFileResource.getInputStream(), cvServiceResponseFileResource.getInputStream(), imageInfoFileResource.getInputStream(), visualLayoutParsingResponseResource.getInputStream());
-    }

    @SneakyThrows
    protected LayoutParsingRequest prepareStorage(InputStream fileStream, InputStream cvServiceResponseFileStream, InputStream imageInfoStream) {
@@ -161,17 +147,6 @@ public abstract class AbstractTest {
        return buildDefaultLayoutParsingRequest(LayoutParsingType.REDACT_MANAGER);
    }

-    @SneakyThrows
-    protected LayoutParsingRequest prepareStorage(InputStream fileStream, InputStream cvServiceResponseFileStream, InputStream imageInfoStream, InputStream visualLayoutParsingResponseFileStream) {
-
-        storageService.storeObject(TenantContext.getTenantId(), IMAGE_FILE_ID, imageInfoStream);
-        storageService.storeObject(TenantContext.getTenantId(), TABLE_FILE_ID, cvServiceResponseFileStream);
-        storageService.storeObject(TenantContext.getTenantId(), ORIGIN_FILE_ID, fileStream);
-        storageService.storeObject(TenantContext.getTenantId(),VISUAL_LAYOUT_FILE,visualLayoutParsingResponseFileStream );
-
-        return buildDefaultLayoutParsingRequest(LayoutParsingType.REDACT_MANAGER);
-    }
-

    @AfterEach
    public void cleanupStorage() {
@@ -10,7 +10,6 @@ import com.knecon.fforesight.service.layoutparser.processor.LayoutParsingPipelin
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 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.layoutparser.processor.services.factory.DocumentGraphFactory;

 import lombok.SneakyThrows;
@@ -26,12 +25,7 @@ public abstract class BuildDocumentTest extends AbstractTest {

        File fileResource = new ClassPathResource(filename).getFile();
        prepareStorage(filename);
-        return layoutParsingPipeline.parseLayout(layoutParsingType,
-                fileResource,
-                layoutParsingStorageService.getImagesFile(IMAGE_FILE_ID),
-                new TableServiceResponse(),
-                new VisualLayoutParsingResponse(),
-                filename);
+        return layoutParsingPipeline.parseLayout(layoutParsingType, fileResource, layoutParsingStorageService.getImagesFile(IMAGE_FILE_ID), new TableServiceResponse(), filename);
    }


@@ -1,8 +0,0 @@
-{
-  "dossierId": "123",
-  "fileId": "123",
-  "targetFileExtension": "ORIGIN.pdf.gz",
-  "responseFileExtension": "EXTRACTED_TABLES.json.gz",
-  "data": [],
-  "X_TENANT_ID": ""
-}
@@ -12,8 +12,6 @@ public class ContentStreams {

    public static Identifier KNECON_LAYOUT = new Identifier("Layout grid", COSName.getPDFName("KNECON_LAYOUT"), true);

-    public static Identifier KNECON_VISUAL_PARSING = new Identifier("Layout grid - visual", COSName.getPDFName("KNECON_VISUAL_PARSING"), true);
-
    public static Identifier KNECON_OCR = new Identifier("OCR", COSName.getPDFName("KNECON_OCR"), false);

    public static Identifier KNECON_OCR_TEXT_DEBUG = new Identifier("OCR Text", COSName.getPDFName("KNECON_OCR_TEXT_DEBUG"), true);
@@ -26,14 +24,7 @@ public class ContentStreams {

    public static Identifier ESCAPE_END = new Identifier("escape start", COSName.getPDFName("ESCAPE_END"), false);

-    public static List<Identifier> allContentStreams = List.of(KNECON_LAYOUT,
-                                                               KNECON_VISUAL_PARSING,
-                                                               KNECON_OCR,
-                                                               KNECON_OCR_BBOX_DEBUG,
-                                                               KNECON_OCR_TEXT_DEBUG,
-                                                               OTHER,
-                                                               ESCAPE_START,
-                                                               ESCAPE_END);
+    public static List<Identifier> allContentStreams = List.of(KNECON_LAYOUT, KNECON_OCR, KNECON_OCR_BBOX_DEBUG, KNECON_OCR_TEXT_DEBUG, OTHER, ESCAPE_START, ESCAPE_END);

    public record Identifier(String name, COSName cosName, boolean optionalContent) {

@@ -17,6 +17,7 @@ import java.util.stream.Collectors;

 import org.apache.pdfbox.Loader;
 import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.pdfwriter.compress.CompressParameters;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDDocumentCatalog;
 import org.apache.pdfbox.pdmodel.PDPage;
@@ -126,8 +127,8 @@ public class ViewerDocumentService {
                pdDocument = openPDDocument(tmpFile.toFile());
            }
        }
-        observedIncrementalSave(pdDocument, destinationFile);

+        observedIncrementalSave(pdDocument, destinationFile);
        pdDocument.close();
        assert tmpFile.toFile().delete();
    }
@@ -282,10 +283,12 @@ public class ViewerDocumentService {

    @SneakyThrows
    private void observedIncrementalSave(PDDocument pdDocument, File outputFile) {
-
+    /*
+    Sometimes the viewer document is corrupted after saving and missing the content streams on a random page, for the files we viewed it did not seem to happen with incrementalSave. It might only be a timing issue though
+     */
        Observation.createNotStarted("ViewerDocumentService", registry).contextualName("incremental-save").observe(() -> {
            try (var out = new FileOutputStream(outputFile)) {
-                pdDocument.save(out);
+                pdDocument.save(out, CompressParameters.NO_COMPRESSION);
            } catch (IOException e) {
                throw new RuntimeException(e);
            }
Author	SHA1	Message	Date
Dominique Eifländer	9eaecdf378	RED-8932 Fixed not merged headline with identifier	2024-04-24 11:44:17 +02:00
Kilian Schüttler	59745a916c	Merge branch 'RED-7384' into 'release/0.89.x' RED-7384: add empty textBlock to Image to ensure continuous textranges across all SemanticNodes See merge request fforesight/layout-parser!139	2024-04-23 12:03:32 +02:00
Kilian Schuettler	0dda309829	RED-7384: add empty textBlock to Image to ensure continuous textranges across all SemanticNodes	2024-04-23 11:30:13 +02:00
Kilian Schüttler	bfa90c2d79	Merge branch 'RED-8995' into 'release/0.89.x' RED-8995: swap incremental save for save without compression to correct wrong layers in rare cases See merge request fforesight/layout-parser!136	2024-04-23 10:44:59 +02:00
Kilian Schuettler	37f7a6a03f	RED-8995: swap incremental save for save without compression to correct wrong layers in rare cases	2024-04-22 11:00:43 +02:00
Kilian Schüttler	bdbac18169	Merge branch 'RED-8995' into 'release/0.89.x' RED-8995: unclassified text might be missing from document data See merge request fforesight/layout-parser!133	2024-04-18 10:01:12 +02:00
Kilian Schuettler	2addf63baf	RED-8995: unclassified text might be missing from document data * treat TablePageBlock.OTHER like PARAGRAPH (no special treatment)	2024-04-17 17:40:21 +02:00
Kilian Schüttler	778bae0f7f	Merge branch 'RED-8747' into 'release/0.89.x' RED-8747 - Entities not merged properly See merge request fforesight/layout-parser!130	2024-04-09 16:30:24 +02:00
Corina Olariu	a01958c842	RED-8747 - Entities not merged properly	2024-04-09 16:30:24 +02:00
Kilian Schüttler	fbe9a34343	Merge branch 'RED-8799' into 'release/0.89.x' RED-8799: LayoutGrid is wrong draw for some tables See merge request fforesight/layout-parser!127	2024-04-05 13:42:36 +02:00
Kilian Schüttler	fd7c461c8d	RED-8799: LayoutGrid is wrong draw for some tables	2024-04-05 13:42:36 +02:00
Dominique Eifländer	cafbcbefc6	Merge branch 'RED-8873-bp' into 'release/0.89.x' RED-8773 - Fix images not appearing on specific file See merge request fforesight/layout-parser!124	2024-04-05 10:11:31 +02:00
Andrei Isvoran	34b260bb60	RED-8773 - Fix images not appearing on specific file	2024-04-03 10:21:45 +03:00
Dominique Eifländer	1ca02f72c8	Merge branch 'RED-8627-4.0' into 'release/0.89.x' RED-8627: Fixed scrambled text after sorting See merge request fforesight/layout-parser!121	2024-03-19 11:26:49 +01:00
Dominique Eifländer	350513a699	RED-8627: Fixed scrambled text after sorting	2024-03-19 11:16:07 +01:00
Dominique Eifländer	ab7b2cf0d5	Merge branch 'RED-7384' into 'release/0.89.x' RED-7384: Fixes for ClassCastException See merge request fforesight/layout-parser!110	2024-03-08 12:53:14 +01:00
Kilian Schuettler	007cbfd1ee	RED-7384: Fixes for ClassCastException * changed save -> incrementalSave * always use origin file instead of reusing viewerdoc * Sometimes the viewer document is corrupted after saving and missing the contentstreams on a random page, for the files we viewed it did not seem to happen with incrementalSave.might only be a timing issue though	2024-03-08 12:42:40 +01:00
Maverick Studer	a266d98f11	Merge branch 'RED-8550-bp' into 'release/0.89.x' RED-8550: Faulty table recognition and text duplication leads to huge sections See merge request fforesight/layout-parser!107	2024-02-29 14:17:31 +01:00
Maverick Studer	33f726c689	RED-8550: Faulty table recognition and text duplication leads to huge sections (cherry picked from commit `74f55a5cbf`)	2024-02-29 13:09:50 +01:00
Maverick Studer	18a28e82d0	RED-8550: Faulty table recognition and text duplication leads to huge sections * cherrypick	2024-02-21 14:19:48 +01:00