Merge branch 'RED-7384' into 'main'

Red 7384 See merge request fforesight/layout-parser!140
Red 7384
2024-04-23 12:13:21 +02:00 · 2024-04-23 12:13:19 +02:00 · 2024-04-19 09:21:50 +02:00 · 2024-04-18 17:42:34 +02:00 · 2024-04-18 15:08:00 +02:00 · 2024-04-18 14:39:10 +02:00
32 changed files with 570 additions and 264 deletions
@@ -42,3 +42,5 @@ gradlew.bat
 gradlew
 gradle.properties
 gradle/
+.DS_Store
+.DS_Store/
@@ -25,4 +25,5 @@ dependencies {
    implementation("com.fasterxml.jackson.datatype:jackson-datatype-jsr310:${jacksonVersion}")
    implementation("org.springframework.boot:spring-boot-starter-web:3.1.3")
    implementation("org.jgrapht:jgrapht-core:1.5.2")
+    implementation("org.tinspin:tinspin-indexes:2.1.3")
 }
@@ -47,6 +47,7 @@ import com.knecon.fforesight.service.layoutparser.processor.services.TableExtrac
 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.classification.ClarifyndClassificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.classification.DocuMineClassificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.classification.RedactManagerClassificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
@@ -88,6 +89,7 @@ public class LayoutParsingPipeline {
    LayoutGridService layoutGridService;
    ObservationRegistry observationRegistry;
    VisualLayoutParsingAdapter visualLayoutParsingAdapter;
+    ClarifyndClassificationService clarifyndClassificationService;


    public LayoutParsingFinishedEvent parseLayoutAndSaveFilesToStorage(LayoutParsingRequest layoutParsingRequest) throws IOException {
@@ -256,7 +258,7 @@ public class LayoutParsingPipeline {

            ClassificationPage classificationPage = switch (layoutParsingType) {
                case REDACT_MANAGER_OLD ->
-                        redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
+                        redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells);
                case DOCUMINE -> docuMineBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
                case REDACT_MANAGER, REDACT_MANAGER_PARAGRAPH_DEBUG -> docstrumBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells, true);
                case CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG -> docstrumBlockificationService.blockify(stripper.getTextPositionSequences(), emptyTableCells, false);
@@ -291,7 +293,7 @@ public class LayoutParsingPipeline {
            if (layoutParsingType == LayoutParsingType.REDACT_MANAGER) {
                docstrumBlockificationService.combineBlocks(classificationPage);
            } else if (layoutParsingType == LayoutParsingType.CLARIFYND) {
-                docstrumBlockificationService.mergeIntersectingBlocks(classificationPage.getTextBlocks());
+                docstrumBlockificationService.mergeIntersectingBlocks(classificationPage.getTextBlocks(), 0, 6.5f);
            }

            buildPageStatistics(classificationPage);
@@ -306,9 +308,10 @@ public class LayoutParsingPipeline {
        bodyTextFrameService.setBodyTextFrames(classificationDocument, layoutParsingType);
        log.info("Classify TextBlocks for {}", identifier);
        switch (layoutParsingType) {
-            case REDACT_MANAGER, REDACT_MANAGER_PARAGRAPH_DEBUG, REDACT_MANAGER_OLD, CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG ->
+            case REDACT_MANAGER, REDACT_MANAGER_PARAGRAPH_DEBUG, REDACT_MANAGER_OLD, CLARIFYND_PARAGRAPH_DEBUG ->
                    redactManagerClassificationService.classifyDocument(classificationDocument);
            case DOCUMINE -> docuMineClassificationService.classifyDocument(classificationDocument);
+            case CLARIFYND -> clarifyndClassificationService.classifyDocument(classificationDocument);
        }

        log.info("Building Sections for {}", identifier);
@@ -1,10 +1,10 @@
 package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;

-import java.util.ArrayList;
-import java.util.Comparator;
 import java.util.List;

 import org.springframework.stereotype.Service;
+import org.tinspin.index.Index;
+import org.tinspin.index.kdtree.KDTree;

 import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Neighbor;
@@ -13,115 +13,23 @@ import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Neigh
 public class NearestNeighbourService {

    private static final int NUMBER_OF_NEIGHBOURS = 8;
-    private static final double STEP = 16.0;


    public void findNearestNeighbors(List<Character> characters) {
+        KDTree<Character> kdTree = KDTree.create(2);
+        characters.forEach(c -> kdTree.insert(new double[]{c.getX(), c.getY()}, c));

-        if (characters.isEmpty() || characters.size() == 1) {
-            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++) {
-
-            Neighbor[] candidates = new Neighbor[maxNeighborCount + 1];
-            int neighborInsertionIndex = 0;
-            int neighborCount = 0;
-            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[neighborInsertionIndex] = new Neighbor(characters.get(start), characters.get(i));
-                    neighborCount++;
-                    if (neighborCount > maxNeighborCount) {
-                        neighborInsertionIndex = clearMostDistant(candidates);
-                        neighborCount--;
-                    } else {
-                        neighborInsertionIndex++;
-                    }
-                    newCandidatesFound = true;
-                }
-
-                while (end < characters.size() && characters.get(end).getX() - characters.get(i).getX() < searchDistance) {
-                    candidates[neighborInsertionIndex] = new Neighbor(characters.get(end), characters.get(i));
-                    neighborCount++;
-                    if (neighborCount > maxNeighborCount) {
-                        neighborInsertionIndex = clearMostDistant(candidates);
-                        neighborCount--;
-                    } else {
-                        neighborInsertionIndex++;
-                    }
-                    end++;
-                    newCandidatesFound = true;
-                }
-
-                if (newCandidatesFound && neighborCount >= maxNeighborCount) {
-                    distance = maxDistance(candidates);
-                }
+        for(Character c : characters) {
+            Index.PointIteratorKnn<Character> iterator = kdTree.queryKnn(new double[]{c.getX(), c.getY()}, NUMBER_OF_NEIGHBOURS + 1);
+            // skip the first as this is identity
+            if(iterator.hasNext()) {
+                iterator.next();
            }
-            if (neighborCount < maxNeighborCount) {
-                clearMostDistant(candidates);
-            }
-
-            List<Neighbor> candidatesList = new ArrayList<>(maxNeighborCount);
-            for (Neighbor candidate : candidates) {
-                if (candidate != null) {
-                    candidatesList.add(candidate);
-                }
-            }
-            candidatesList.sort(Comparator.comparingDouble(Neighbor::getDistance));
-            assert candidatesList.size() == maxNeighborCount;
-            characters.get(i).setNeighbors(candidatesList);
-        }
-    }
-
-
-    private double maxDistance(Neighbor[] candidates) {
-
-        double maxDistance = 0;
-        for (Neighbor candidate : candidates) {
-            if (candidate == null) {
-                continue;
-            }
-            if (candidate.getDistance() > maxDistance) {
-                maxDistance = candidate.getDistance();
+            while(iterator.hasNext()) {
+                c.getNeighbors().add(new Neighbor(iterator.next().value(), c));
            }
        }
-        return maxDistance;
-    }

-
-    private int clearMostDistant(Neighbor[] candidates) {
-
-        double maxDistance = 0;
-        int maxIndex = 0;
-        for (int i = 0; i < candidates.length; i++) {
-            Neighbor candidate = candidates[i];
-            if (candidate == null) {
-                continue;
-            }
-            if (candidate.getDistance() > maxDistance) {
-                maxDistance = candidate.getDistance();
-                maxIndex = i;
-            }
-        }
-        candidates[maxIndex] = null;
-        return maxIndex;
    }

 }
@@ -65,7 +65,7 @@ public class ZoneBuilderService {

        List<Zone> zones = new ArrayList<>();
        unionFind.getGroups().forEach(group -> {
-            zones.add(new Zone(new ArrayList<>(group)));
+            zones.add(mergeLinesInZone(new ArrayList<>(group), characterSpacing, lineSpacing));
        });

        if (zones.size() > MAX_ZONES) {
@@ -64,9 +64,24 @@ public class Document implements GenericSemanticNode {
    }


+    public List<Header> getHeaders() {
+
+        return streamChildrenOfType(NodeType.HEADER).map(node -> (Header) node)
+                .collect(Collectors.toList());
+    }
+
+
+    public List<Footer> getFooters() {
+
+        return streamChildrenOfType(NodeType.FOOTER).map(node -> (Footer) node)
+                .collect(Collectors.toList());
+    }
+
+
    public Stream<TextBlock> streamTerminalTextBlocksInOrder() {

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


@@ -13,7 +13,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.No
 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;
@@ -40,6 +39,8 @@ public class Image implements GenericSemanticNode {
    boolean transparent;
    Rectangle2D position;

+    TextBlock leafTextBlock;
+
    boolean redaction;
    boolean ignored;
    @Builder.Default
@@ -70,9 +71,7 @@ public class Image implements GenericSemanticNode {
    @Override
    public TextBlock getTextBlock() {

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


@@ -98,4 +97,11 @@ public class Image implements GenericSemanticNode {
        return bBoxPerPage;
    }

+
+    @Override
+    public boolean isLeaf() {
+
+        return true;
+    }
+
 }
@@ -208,6 +208,12 @@ public class TextPositionSequence implements CharSequence {
    }


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

        return textPositions.get(0).getHeightDir() + HEIGHT_PADDING;
@@ -44,7 +44,6 @@ 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) {
@@ -63,11 +62,6 @@ 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);
@@ -95,9 +89,6 @@ public class SectionsBuilderService {
            if (!footer.isEmpty()) {
                footers.add(new ClassificationFooter(footer));
            }
-            if (!unclassifiedText.isEmpty()) {
-                unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
-            }
        }

        ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
@@ -215,6 +206,7 @@ public class SectionsBuilderService {
                            .getX() <= xMax && image.getPosition().getY() >= yMin && image.getPosition().getY() <= yMax) {
                        section.getImages().add(image);
                        image.setAppendedToSection(true);
+                        break;
                    }
                }
                if (!image.isAppendedToSection()) {
@@ -1,26 +1,51 @@
 package com.knecon.fforesight.service.layoutparser.processor.services;

 import java.util.List;
+import java.util.stream.Collectors;
+import java.util.stream.Stream;

 import org.springframework.stereotype.Service;

 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedSectionText;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedText;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.GenericSemanticNode;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
+
+import lombok.extern.slf4j.Slf4j;

@Service
+@Slf4j
 public class SimplifiedSectionTextService {

    public SimplifiedText toSimplifiedText(Document document) {

-        List<SimplifiedSectionText> simplifiedSectionTexts = document.getMainSections().stream().map(this::toSimplifiedSectionText).toList();
-        return SimplifiedText.builder().numberOfPages(document.getNumberOfPages()).sectionTexts(simplifiedSectionTexts).build();
+        List<SimplifiedSectionText> simplifiedMainSectionsList = document.getMainSections()
+                .stream()
+                .map(this::toSimplifiedSectionText)
+                .toList();
+        List<SimplifiedSectionText> simplifiedHeadersList = document.getHeaders()
+                .stream()
+                .map(this::toSimplifiedSectionText)
+                .toList();
+        List<SimplifiedSectionText> simplifiedFootersList = document.getFooters()
+                .stream()
+                .map(this::toSimplifiedSectionText)
+                .toList();
+        List<SimplifiedSectionText> simplifiedText = Stream.of(simplifiedMainSectionsList, simplifiedHeadersList, simplifiedFootersList)
+                .flatMap(List::stream)
+                .collect(Collectors.toList());
+        return SimplifiedText.builder().numberOfPages(document.getNumberOfPages()).sectionTexts(simplifiedText).build();
    }


-    private SimplifiedSectionText toSimplifiedSectionText(Section section) {
+    private SimplifiedSectionText toSimplifiedSectionText(SemanticNode section) {

-        return SimplifiedSectionText.builder().sectionNumber(section.getTreeId().get(0)).text(section.getTextBlock().getSearchText()).build();
+        return SimplifiedSectionText.builder()
+                .sectionNumber(section.getTreeId()
+                                       .get(0))
+                .text(section.getTextBlock().getSearchText())
+                .build();
    }
+
 }
@@ -2,30 +2,28 @@ 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.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.docstrum.DocstrumSegmentationService;
 import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.DoubleUtils;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
 import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
 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.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.utils.QuickSort;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 import com.knecon.fforesight.service.layoutparser.processor.utils.RulingTextDirAdjustUtil;
-import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionSequenceComparator;

 import lombok.RequiredArgsConstructor;

@@ -41,25 +39,20 @@ public class DocstrumBlockificationService {

    public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Cell> cells, boolean xyOrder) {

-        // 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)));
-        });
+        CleanRulings usedRulings = RectangleTransformations.extractRulings(cells);

        var zones = docstrumSegmentationService.segmentPage(textPositions, xyOrder);
-        var pageBlocks = toAbstractPageBlocks(zones, usedHorizonalRulings, usedVerticalRulings);
+        var pageBlocks = toAbstractPageBlocks(zones, usedRulings.getHorizontal(), usedRulings.getVertical(), xyOrder);

-        return new ClassificationPage(pageBlocks);
+        var classificationPage = new ClassificationPage(pageBlocks);
+
+        mergeIntersectingBlocks(classificationPage.getTextBlocks(), 0, 0);
+
+        return classificationPage;
    }


-    private List<AbstractPageBlock> toAbstractPageBlocks(List<Zone> zones, List<Ruling> horizontalRulings, List<Ruling> verticalRulings) {
+    private List<AbstractPageBlock> toAbstractPageBlocks(List<Zone> zones, List<Ruling> horizontalRulings, List<Ruling> verticalRulings, boolean xyOrder) {

        List<AbstractPageBlock> abstractPageBlocks = new ArrayList<>();
        zones.forEach(zone -> {
@@ -74,20 +67,31 @@ public class DocstrumBlockificationService {
            abstractPageBlocks.addAll(splitZonesAtRulings(textPositionSequences, horizontalRulings, verticalRulings));
        });

+        if (xyOrder) {
+            abstractPageBlocks.sort(Comparator.comparing(AbstractPageBlock::getMinY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                    .thenComparing(AbstractPageBlock::getMinX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD)));
+            abstractPageBlocks.sort(new Comparator<AbstractPageBlock>() {
+                @Override
+                public int compare(AbstractPageBlock o1, AbstractPageBlock o2) {
+
+                    return Math.abs(o1.getMinY() - o2.getMinY()) < 5 && o1.getMinX() < o2.getMinX() == true ? -1 : 0;
+                }
+            });
+        }
+
        return abstractPageBlocks;
    }


    public void combineBlocks(ClassificationPage page) {

-        mergeIntersectingBlocks(page.getTextBlocks());
-
        TextPageBlock previous = new TextPageBlock();
        ListIterator<AbstractPageBlock> itty = page.getTextBlocks().listIterator();
        while (itty.hasNext()) {

            AbstractPageBlock block = itty.next();
            if (block instanceof TablePageBlock) {
+                previous = new TextPageBlock();
                continue;
            }
            TextPageBlock current = (TextPageBlock) block;
@@ -100,22 +104,22 @@ public class DocstrumBlockificationService {
                }

                if (areTheOnlyTwoBlocksOnHeightsWithBothMoreThanTwoLines(previous, current, page)) {
-                    combineBlocksAndResetIterator(previous, current, itty, true);
+                    previous = combineBlocksAndResetIterator(previous, current, itty, true);
                    continue;
                }

                if (previous.almostIntersects(current, 0, 0)) {
-                    combineBlocksAndResetIterator(previous, current, itty, previous.isToDuplicate());
+                    previous = combineBlocksAndResetIterator(previous, current, itty, previous.isToDuplicate());
                    continue;
                }

                if (isSameTopOrBottomWithPreviousHavingMaxTwoLinesAndCurrentThanOneAndMax4OtherBlocksOnHeight(previous, current, page)) {
-                    combineBlocksAndResetIterator(previous, current, itty, false);
+                    previous = combineBlocksAndResetIterator(previous, current, itty, previous.isToDuplicate());
                    continue;
                }

                if (isOnlyIntersectingYAndOnelineOrPrevoiusTwoLines(previous, current, page)) {
-                    combineBlocksAndResetIterator(previous, current, itty, false);
+                    previous = combineBlocksAndResetIterator(previous, current, itty, previous.isToDuplicate());
                    continue;
                }

@@ -123,7 +127,7 @@ public class DocstrumBlockificationService {
            previous = current;
        }

-        mergeIntersectingBlocks(page.getTextBlocks());
+        mergeIntersectingBlocks(page.getTextBlocks(), 0, 6.5f);
    }


@@ -140,7 +144,7 @@ public class DocstrumBlockificationService {
                                                                                                              ClassificationPage page) {

        return (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) //
+                && (previous.getNumberOfLines() == 1 && current.getNumberOfLines() >= 1 || previous.getNumberOfLines() >= 2 && current.getNumberOfLines() == 1) //
                && !hasBetween(current, previous, page.getTextBlocks()) && numberOfYIntersectionsOfSmallerBlocksWithOtherBlocks(current, previous, page.getTextBlocks()) <= 4;
    }

@@ -153,7 +157,7 @@ public class DocstrumBlockificationService {
    }


-    private void combineBlocksAndResetIterator(TextPageBlock previous, TextPageBlock current, ListIterator<AbstractPageBlock> itty, boolean toDuplicate) {
+    private TextPageBlock combineBlocksAndResetIterator(TextPageBlock previous, TextPageBlock current, ListIterator<AbstractPageBlock> itty, boolean toDuplicate) {

        previous.getSequences().addAll(current.getSequences());
        previous = buildTextBlock(previous.getSequences(), 0);
@@ -162,6 +166,7 @@ public class DocstrumBlockificationService {
        itty.previous();
        itty.set(previous);
        itty.next();
+        return previous;
    }


@@ -203,25 +208,23 @@ public class DocstrumBlockificationService {
    }


-    public void mergeIntersectingBlocks(List<AbstractPageBlock> blocks) {
+    public void mergeIntersectingBlocks(List<AbstractPageBlock> blocks, float xThreshold, float yThreshold) {

        ListIterator<AbstractPageBlock> itty = blocks.listIterator();
-        Set<AbstractPageBlock> toRemove = new HashSet<>();
        while (itty.hasNext()) {
            AbstractPageBlock block = itty.next();
+            if(block == null){
+                continue;
+            }
            if (block instanceof TablePageBlock) {
                continue;
            }

            TextPageBlock current = (TextPageBlock) block;

-            if (current.isToDuplicate()) {
-                continue;
-            }
-
            for (int i = 0; i < blocks.size(); i++) {

-                if (toRemove.contains(blocks.get(i))) {
+                if(blocks.get(i) == null){
                    continue;
                }
                if (blocks.get(i) == current) {
@@ -233,21 +236,24 @@ public class DocstrumBlockificationService {

                TextPageBlock inner = (TextPageBlock) blocks.get(i);

-                if (inner.isToDuplicate()) {
-                    continue;
-                }
-
-                if (current.getDir() == inner.getDir() && current.almostIntersects(inner, 0, 0)) {
+                if (current.getDir() == inner.getDir() && current.almostIntersects(inner, yThreshold, xThreshold)) {

+                    boolean toDuplicate = current.isToDuplicate() || inner.isToDuplicate();
                    current.getSequences().addAll(inner.getSequences());
-                    QuickSort.sort(current.getSequences(), new TextPositionSequenceComparator());
                    current = buildTextBlock(current.getSequences(), 0);
-                    toRemove.add(inner);
+
+                    current.setToDuplicate(toDuplicate);
+                    blocks.set(i, null);
                    itty.set(current);
                }
            }
        }
-        blocks.removeAll(toRemove);
+        var blocksIterator = blocks.iterator();
+        while(blocksIterator.hasNext()){
+            if(blocksIterator.next() == null){
+                blocksIterator.remove();
+            }
+        }
    }


@@ -13,10 +13,13 @@ 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 +34,12 @@ 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> cells) {
+
+        CleanRulings usedRulings = RectangleTransformations.extractRulings(cells);

        int indexOnPage = 0;
        List<TextPositionSequence> chunkWords = new ArrayList<>();
@@ -54,7 +57,7 @@ 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)) {
@@ -0,0 +1,114 @@
+package com.knecon.fforesight.service.layoutparser.processor.services.classification;
+
+import java.util.List;
+import java.util.regex.Pattern;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.utils.MarkedContentUtils;
+import com.knecon.fforesight.service.layoutparser.processor.utils.PositionUtils;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class ClarifyndClassificationService {
+
+    public void classifyDocument(ClassificationDocument document) {
+
+        List<Float> headlineFontSizes = document.getFontSizeCounter().getHighterThanMostPopular();
+
+        log.debug("Document FontSize counters are: {}", document.getFontSizeCounter().getCountPerValue());
+
+        for (ClassificationPage page : document.getPages()) {
+            classifyPage(page, document, headlineFontSizes);
+        }
+    }
+
+
+    private void classifyPage(ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
+
+        for (AbstractPageBlock textBlock : page.getTextBlocks()) {
+            if (textBlock instanceof TextPageBlock) {
+                classifyBlock((TextPageBlock) textBlock, page, document, headlineFontSizes);
+            }
+        }
+    }
+
+
+    private void classifyBlock(TextPageBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
+
+        var bodyTextFrame = page.getBodyTextFrame();
+
+        if (document.getFontSizeCounter().getMostPopular() == null) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH);
+            return;
+        }
+        if (MarkedContentUtils.intersects(textBlock, page.getMarkedContentBboxPerType(), MarkedContentUtils.HEADER) || PositionUtils.isOverBodyTextFrame(bodyTextFrame,
+                textBlock,
+                page.getRotation()) && (document.getFontSizeCounter().getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
+                .getMostPopular())) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH);
+
+        } else if (MarkedContentUtils.intersects(textBlock, page.getMarkedContentBboxPerType(), MarkedContentUtils.FOOTER) || PositionUtils.isUnderBodyTextFrame(bodyTextFrame,
+                textBlock,
+                page.getRotation()) && (document.getFontSizeCounter().getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
+                .getMostPopular())) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH);
+        } else if (page.getPageNumber() == 1 && (PositionUtils.getHeightDifferenceBetweenChunkWordAndDocumentWord(textBlock,
+                document.getTextHeightCounter().getMostPopular()) > 2.5 && textBlock.getHighestFontSize() > document.getFontSizeCounter().getMostPopular() || page.getTextBlocks()
+                .size() == 1)) {
+            if (!Pattern.matches("[0-9]+", textBlock.toString())) {
+                textBlock.setClassification(PageBlockType.TITLE);
+            }
+        } else if (textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter()
+                .getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && (textBlock.getMostPopularWordStyle().equals("bold") || !document.getFontStyleCounter()
+                .getCountPerValue()
+                .containsKey("bold") && textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter().getMostPopular() + 1) && textBlock.getSequences()
+                .get(0)
+                .getTextPositions()
+                .get(0)
+                .getFontSizeInPt() >= textBlock.getMostPopularWordFontSize()) {
+
+            for (int i = 1; i <= headlineFontSizes.size(); i++) {
+                if (textBlock.getMostPopularWordFontSize() == headlineFontSizes.get(i - 1)) {
+                    textBlock.setClassification(PageBlockType.getHeadlineType(i));
+                    document.setHeadlines(true);
+                }
+            }
+        } else if (!textBlock.getText().startsWith("Figure ") && PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordStyle()
+                .equals("bold") && !document.getFontStyleCounter().getMostPopular().equals("bold") && PositionUtils.getApproxLineCount(textBlock) < 2.9 && textBlock.getSequences()
+                .get(0)
+                .getTextPositions()
+                .get(0)
+                .getFontSizeInPt() >= textBlock.getMostPopularWordFontSize()) {
+            textBlock.setClassification(PageBlockType.getHeadlineType(headlineFontSizes.size() + 1));
+            document.setHeadlines(true);
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter()
+                .getMostPopular() && textBlock.getMostPopularWordStyle().equals("bold") && !document.getFontStyleCounter().getMostPopular().equals("bold")) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH_BOLD);
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFont()
+                .equals(document.getFontCounter().getMostPopular()) && textBlock.getMostPopularWordStyle()
+                .equals(document.getFontStyleCounter().getMostPopular()) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter().getMostPopular()) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH);
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter()
+                .getMostPopular() && textBlock.getMostPopularWordStyle().equals("italic") && !document.getFontStyleCounter()
+                .getMostPopular()
+                .equals("italic") && PositionUtils.getApproxLineCount(textBlock) < 2.9) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH_ITALIC);
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
+            textBlock.setClassification(PageBlockType.PARAGRAPH_UNKNOWN);
+        } else {
+            textBlock.setClassification(PageBlockType.PARAGRAPH);
+        }
+    }
+
+}
+
@@ -54,9 +54,14 @@ 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(layoutParsingType, document, context);
+        document.getPages()
+                .forEach(context::buildAndAddPageWithCounter);
+        document.getSections()
+                .stream()
+                .flatMap(section -> section.getImages()
+                        .stream())
+                .forEach(image -> context.getImages().add(image));
+        addSections(layoutParsingType, document, context, documentGraph);
        addHeaderAndFooterToEachPage(document, context);

        documentGraph.setNumberOfPages(context.pages.size());
@@ -67,9 +72,10 @@ public class DocumentGraphFactory {
    }


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

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


@@ -79,11 +85,14 @@ public class DocumentGraphFactory {

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

        page.getMainBody().add(node);
@@ -96,8 +105,9 @@ public class DocumentGraphFactory {

        if (node instanceof DuplicatedParagraph duplicatedParagraph) {
            AtomicTextBlock unsortedTextBlock = context.textBlockFactory.buildAtomicTextBlock(textBlocks.stream()
-                    .flatMap(tb -> tb.getSequences().stream())
-                    .collect(Collectors.toList()), node, context, page);
+                                                                                                      .flatMap(tb -> tb.getSequences()
+                                                                                                              .stream())
+                                                                                                      .collect(Collectors.toList()), node, context, page);
            duplicatedParagraph.setUnsortedLeafTextBlock(unsortedTextBlock);
        }

@@ -107,25 +117,29 @@ public class DocumentGraphFactory {
    }


-    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;
    }


@@ -164,11 +178,12 @@ 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);
@@ -179,7 +194,8 @@ 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);
@@ -191,7 +207,8 @@ 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);
@@ -203,7 +220,8 @@ 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);
@@ -13,6 +13,7 @@ import java.util.Set;

 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 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;
@@ -30,27 +31,48 @@ public class SectionNodeFactory {
                           GenericSemanticNode parentNode,
                           List<AbstractPageBlock> pageBlocks,
                           List<ClassifiedImage> images,
-                           DocumentGraphFactory.Context context) {
+                           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(layoutParsingType, pageBlocks, context, section);
+        addFirstHeadlineDirectlyToSection(layoutParsingType, pageBlocks, context, section, document);
        if (containsTablesAndTextBlocks(pageBlocks)) {
-            splitPageBlocksIntoSubSections(pageBlocks).forEach(subSectionPageBlocks -> addSection(layoutParsingType, section, subSectionPageBlocks, emptyList(), context));
+            splitPageBlocksIntoSubSections(pageBlocks).forEach(subSectionPageBlocks -> addSection(layoutParsingType,
+                                                                                                  section,
+                                                                                                  subSectionPageBlocks,
+                                                                                                  emptyList(),
+                                                                                                  context,
+                                                                                                  document));
        } else {
-            addTablesAndParagraphsAndHeadlinesToSection(layoutParsingType, pageBlocks, context, section);
+            addTablesAndParagraphsAndHeadlinesToSection(layoutParsingType, pageBlocks, context, section, document);
        }

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


@@ -64,10 +86,14 @@ public class SectionNodeFactory {
    }


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

        if (pageBlocks.get(0).isHeadline()) {
-            addTablesAndParagraphsAndHeadlinesToSection(layoutParsingType, List.of(pageBlocks.get(0)), context, section);
+            addTablesAndParagraphsAndHeadlinesToSection(layoutParsingType, List.of(pageBlocks.get(0)), context, section, document);
            pageBlocks.remove(0);
        }
    }
@@ -76,7 +102,8 @@ public class SectionNodeFactory {
    private void addTablesAndParagraphsAndHeadlinesToSection(LayoutParsingType layoutParsingType,
                                                             List<AbstractPageBlock> pageBlocks,
                                                             DocumentGraphFactory.Context context,
-                                                             Section section) {
+                                                             Section section,
+                                                             Document document) {

        Set<AbstractPageBlock> alreadyMerged = new HashSet<>();
        List<AbstractPageBlock> remainingBlocks = new LinkedList<>(pageBlocks);
@@ -91,7 +118,7 @@ public class SectionNodeFactory {
            if (abstractPageBlock instanceof TextPageBlock) {

                switch (layoutParsingType) {
-                    case CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG -> {
+                    case REDACT_MANAGER, CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG -> {
                        alreadyMerged.add(abstractPageBlock);
                        remainingBlocks.remove(abstractPageBlock);
                        DocumentGraphFactory.addParagraphOrHeadline(section, (TextPageBlock) abstractPageBlock, context, new ArrayList<>());
@@ -105,7 +132,7 @@ public class SectionNodeFactory {
            } else if (abstractPageBlock instanceof TablePageBlock tablePageBlock) {
                List<TablePageBlock> tablesToMerge = TableMergingUtility.findConsecutiveTablesWithSameColCountAndSameHeaders(tablePageBlock, remainingBlocks);
                alreadyMerged.addAll(tablesToMerge);
-                TableNodeFactory.addTable(layoutParsingType, section, tablesToMerge, context);
+                TableNodeFactory.addTable(layoutParsingType, section, tablesToMerge, context, document);
            } else {
                throw new RuntimeException(format("Unhandled AbstractPageBlockType %s!", abstractPageBlock.getClass()));
            }
@@ -115,7 +142,9 @@ 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);
    }


@@ -131,7 +160,9 @@ 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();
    }


@@ -152,7 +183,8 @@ 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);
    }


@@ -9,6 +9,7 @@ import java.util.stream.Collectors;

 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 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.SemanticNode;
@@ -28,23 +29,30 @@ public class TableNodeFactory {
    public final double TABLE_CELL_MERGE_CONTENTS_SIZE_THRESHOLD = 0.05;


-    public void addTable(LayoutParsingType layoutParsingType, GenericSemanticNode parentNode, List<TablePageBlock> tablesToMerge, DocumentGraphFactory.Context context) {
+    public void addTable(LayoutParsingType layoutParsingType,
+                         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(layoutParsingType, mergedRows, table, context);
+        addTableCells(layoutParsingType, mergedRows, table, context, document);

        ifTableHasNoHeadersSetFirstRowAsHeaders(table);
    }
@@ -64,7 +72,8 @@ 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()));
    }
@@ -83,28 +92,44 @@ 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(LayoutParsingType layoutParsingType, List<List<Cell>> rows, Table table, DocumentGraphFactory.Context context) {
+    private void addTableCells(LayoutParsingType layoutParsingType, 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(layoutParsingType, rows.get(rowIndex).get(colIndex), rowIndex, colIndex, table, context);
+                addTableCell(layoutParsingType,
+                             rows.get(rowIndex)
+                                     .get(colIndex),
+                             rowIndex,
+                             colIndex,
+                             table,
+                             context,
+                             document);
            }
        }
    }


    @SuppressWarnings("PMD.UnusedPrivateMethod") // PMD actually flags this wrong
-    private void addTableCell(LayoutParsingType layoutParsingType, Cell cell, int rowIndex, int colIndex, Table tableNode, DocumentGraphFactory.Context context) {
+    private void addTableCell(LayoutParsingType layoutParsingType,
+                              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);
@@ -114,16 +139,27 @@ 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(layoutParsingType, tableCell, cell.getTextBlocks().stream().map(tb -> (AbstractPageBlock) tb).toList(), emptyList(), context);
+            SectionNodeFactory.addSection(layoutParsingType,
+                                          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()));
        }
    }

@@ -136,7 +172,8 @@ public class TableNodeFactory {

    private boolean firstTextBlockIsHeadline(Cell cell) {

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

 }
@@ -129,11 +129,13 @@ 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 +144,7 @@ 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 +154,7 @@ 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()));
@@ -2,8 +2,10 @@ package com.knecon.fforesight.service.layoutparser.processor.utils;

 import static java.lang.String.format;

+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;
@@ -19,6 +21,8 @@ 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;
@@ -147,7 +151,27 @@ 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();
    }


@@ -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;
        }
    }
+
 }
@@ -1,6 +1,9 @@
 package com.knecon.fforesight.service.layoutparser.server.queue;

 import java.util.Arrays;
+import java.util.Locale;
+import java.util.Map;
+import java.util.stream.Collectors;

 import org.springframework.amqp.AmqpRejectAndDontRequeueException;
 import org.springframework.amqp.core.Message;
@@ -28,6 +31,7 @@ public class MessageHandler {
    private final LayoutParsingPipeline layoutParsingPipeline;
    private final ObjectMapper objectMapper;
    private final RabbitTemplate rabbitTemplate;
+    private final static String X_PIPELINE_PREFIX = "X-PIPE-";


    @RabbitHandler
@@ -46,14 +50,24 @@ public class MessageHandler {
                    layoutParsingRequest.identifier()));
        }
        LayoutParsingFinishedEvent layoutParsingFinishedEvent = layoutParsingPipeline.parseLayoutAndSaveFilesToStorage(layoutParsingRequest);
-        sendLayoutParsingFinishedEvent(layoutParsingFinishedEvent);
+        sendLayoutParsingFinishedEvent(layoutParsingFinishedEvent, message);
    }


-    public void sendLayoutParsingFinishedEvent(LayoutParsingFinishedEvent layoutParsingFinishedEvent) {
+    public void sendLayoutParsingFinishedEvent(LayoutParsingFinishedEvent layoutParsingFinishedEvent, Message message) {

        Arrays.stream(layoutParsingFinishedEvent.message().split("\n")).forEach(log::info);
-        rabbitTemplate.convertAndSend(LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE, layoutParsingFinishedEvent);
+        rabbitTemplate.convertAndSend(LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE, layoutParsingFinishedEvent, m -> {
+            var forwardHeaders = message.getMessageProperties()
+                    .getHeaders()
+                    .entrySet()
+                    .stream()
+                    .filter(e -> e.getKey().toUpperCase(Locale.ROOT).startsWith(X_PIPELINE_PREFIX))
+                    .collect(Collectors.toMap(Map.Entry::getKey,
+                            Map.Entry::getValue));
+            m.getMessageProperties().getHeaders().putAll(forwardHeaders);
+            return m;
+        });
    }

 }
@@ -26,9 +26,22 @@ public class LayoutparserEnd2EndTest extends AbstractTest {
    public void testLayoutParserEndToEnd() {

        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);
+    }
+
+
+    @Test
+    @SneakyThrows
+    public void testLayoutParserEndToEnd_RED_8747() {
+
+        prepareStorage("files/SinglePages/MergedEntities.pdf");
        LayoutParsingRequest layoutParsingRequest = buildDefaultLayoutParsingRequest(LayoutParsingType.REDACT_MANAGER_OLD);
        LayoutParsingFinishedEvent finishedEvent = layoutParsingPipeline.parseLayoutAndSaveFilesToStorage(layoutParsingRequest);
-        Arrays.stream(finishedEvent.message().split("\n")).forEach(log::info);
+        Arrays.stream(finishedEvent.message().split("\n"))
+                .forEach(log::info);
    }

 }
@@ -0,0 +1,78 @@
+package com.knecon.fforesight.service.layoutparser.server;
+
+import static org.assertj.core.api.AssertionsForClassTypes.assertThat;
+
+import java.io.File;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.atomic.AtomicReference;
+import java.util.stream.Collectors;
+
+import org.junit.jupiter.api.Test;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.core.io.ClassPathResource;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedSectionText;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedText;
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingFinishedEvent;
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
+import com.knecon.fforesight.service.layoutparser.processor.LayoutParsingPipeline;
+import com.knecon.fforesight.service.layoutparser.processor.LayoutParsingStorageService;
+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.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.SimplifiedSectionTextService;
+import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
+import com.knecon.fforesight.service.layoutparser.server.utils.AbstractTest;
+
+import io.micrometer.observation.Observation;
+import lombok.SneakyThrows;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public class SimplifiedTextServiceTest
+        extends AbstractTest {
+
+    @Autowired
+    private LayoutParsingPipeline layoutParsingPipeline;
+    @Autowired
+    private SimplifiedSectionTextService simplifiedSectionTextService;
+
+
+    @Test
+    @SneakyThrows
+    public void testSearchIndexing() {
+
+        File file = new ClassPathResource("files/headerFooterTest3Pages.pdf").getFile();
+        String footerExample = "Footer to search for ";
+        String headerExample ="Header to search for ";
+        Document document = buildGraph(file);
+        SimplifiedText simplifiedText = simplifiedSectionTextService.toSimplifiedText(document);
+        List<SimplifiedSectionText> sectionTexts = simplifiedText.getSectionTexts();
+        assertThat(sectionTexts.stream().filter(section -> section.getText().equals(footerExample)).collect(Collectors.toList()).size()).isGreaterThan(0);
+        assertThat(sectionTexts.stream().filter(section -> section.getText().equals(headerExample)).collect(Collectors.toList()).size()).isGreaterThan(0);
+
+
+
+    }
+
+    @SneakyThrows
+    protected Document buildGraph(File file) {
+
+        return DocumentGraphFactory.buildDocumentGraph(LayoutParsingType.REDACT_MANAGER,
+                                                       layoutParsingPipeline.parseLayout(LayoutParsingType.REDACT_MANAGER,
+                                                                                         file,
+                                                                                         new ImageServiceResponse(),
+                                                                                         new TableServiceResponse(),
+                                                                                         new VisualLayoutParsingResponse(),
+                                                                                         Map.of("file",file.toString())));
+    }
+
+
+
+}
@@ -27,7 +27,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
    @SneakyThrows
    public void testViewerDocument() {

-        String fileName = "files/S-Metolachlor_RAR_01_Volume_1_2018-09-06.pdf";
+        String fileName = "files/new/ScrambledTextAfterSorting.pdf";
        String tmpFileName = "/tmp/" + Path.of(fileName).getFileName() + "_VIEWER.pdf";

        var documentFile = new ClassPathResource(fileName).getFile();
@@ -35,7 +35,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
        LayoutGridService layoutGridService = new LayoutGridService(viewerDocumentService);

        long start = System.currentTimeMillis();
-        Document document = buildGraph(fileName, LayoutParsingType.REDACT_MANAGER);
+        Document document = buildGraph(fileName, LayoutParsingType.DOCUMINE);
        layoutGridService.addLayoutGrid(documentFile, document, new File(tmpFileName), true);
        System.out.printf("Total time: %.2fs%n", ((float) (System.currentTimeMillis() - start)) / 1000);
    }
@@ -59,7 +59,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
                new ImageServiceResponse(),
                tableResponse,
                new VisualLayoutParsingResponse(),
-                Map.of("file",Path.of(fileName).getFileName().toFile().toString()));
+                Map.of("file", Path.of(fileName).getFileName().toFile().toString()));
        ViewerDocumentService viewerDocumentService = new ViewerDocumentService(null);
        LayoutGridService layoutGridService = new LayoutGridService(viewerDocumentService);
        Document document = DocumentGraphFactory.buildDocumentGraph(LayoutParsingType.DOCUMINE, classificationDocument);
@@ -667,6 +667,19 @@ 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) {
@@ -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;
@@ -285,7 +286,7 @@ public class ViewerDocumentService {

        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
Kilian Schüttler	d38d023485	Merge branch 'RED-7384' into 'main' Red 7384 See merge request fforesight/layout-parser!140	2024-04-23 12:13:21 +02:00
Kilian Schüttler	c1afe9b11f	Red 7384	2024-04-23 12:13:19 +02:00
Dominique Eifländer	c947d552d2	Merge branch 'RED-8995-fp' into 'main' RED-8995: unclassified text might be missing from document data See merge request fforesight/layout-parser!135	2024-04-19 09:21:50 +02:00
Kilian Schuettler	f256f9b30f	RED-8995: unclassified text might be missing from document data * treat TablePageBlock.OTHER like PARAGRAPH (no special treatment)	2024-04-18 17:42:34 +02:00
Yannik Hampe	6167e3fb57	Merge branch 'RED-8402' into 'main' RED-8402: Header and footer are not indexed / searched See merge request fforesight/layout-parser!134	2024-04-18 15:08:00 +02:00
yhampe	a78fb0244a	Merge remote-tracking branch 'origin/RED-8402' into RED-8402	2024-04-18 14:39:10 +02:00
yhampe	8099a00bb6	RED-8402: Header and footer are not indexed / searched added unit test and file	2024-04-18 14:39:01 +02:00
yhampe	9bb0468b2b	RED-8402: Header and footer are not indexed / searched added unit test and file	2024-04-18 14:36:25 +02:00
Kilian Schüttler	c4d9c5df02	Merge branch 'RED-8747-fp' into 'main' RED-8747 - Entities not merged properly - fp See merge request fforesight/layout-parser!131	2024-04-09 16:30:02 +02:00
Corina Olariu	976f408237	RED-8747 - Entities not merged properly - fp - rework the extraction of rulings from the table cells	2024-04-09 14:38:48 +03:00
Corina Olariu	319268c53d	RED-8747 - Entities not merged properly - fp - update test	2024-04-09 12:24:19 +03:00
Corina Olariu	014eba9fc3	RED-8747 - Entities not merged properly - fp - fix typo - add validate table test	2024-04-09 12:14:57 +03:00
Yannik Hampe	9bd8419770	Merge branch 'RED-8402' into 'main' RED-8402: Header and footer are not indexed / searched See merge request fforesight/layout-parser!128	2024-04-08 12:28:06 +02:00
yhampe	c13ff7fbf6	RED-8402: Header and footer are not indexed / searched checkstyle added review comments	2024-04-08 12:17:49 +02:00
yhampe	5d3826e9b9	Merge remote-tracking branch 'origin/RED-8402' into RED-8402	2024-04-08 12:02:47 +02:00
yhampe	0c3194276a	RED-8402: Header and footer are not indexed / searched added headers and footers to simplifiedtext	2024-04-08 12:02:36 +02:00
yhampe	e302d9784e	RED-8402: Header and footer are not indexed / searched added headers and footers to simplifiedtext	2024-04-08 11:59:35 +02:00
Corina Olariu	f185b13f2b	RED-8747 - Entities not merged properly - fp - use the rullings from the found tables instead of all rullings as splitting rullings in the blockification service	2024-04-08 09:42:32 +03:00
Dominique Eifländer	990c376ce6	Merge branch 'RED-8873' into 'main' RED-8773 - Fix images not appearing on specific file See merge request fforesight/layout-parser!123	2024-04-05 10:11:23 +02:00
Kilian Schüttler	bf6a0d770b	Merge branch 'RED-8799' into 'main' RED-8799: LayoutGrid is wrong draw for some tables See merge request fforesight/layout-parser!126	2024-04-04 15:23:12 +02:00
Kilian Schuettler	f18bda1d4e	RED-8799: LayoutGrid is wrong draw for some tables	2024-04-04 13:33:22 +02:00
Maverick Studer	0a11992361	Merge branch 'RED-8702' into 'main' RED-8702: Explore document databases to store entityLog See merge request fforesight/layout-parser!125	2024-04-03 10:00:38 +02:00
Andrei Isvoran	456b8fe4a1	RED-8773 - Fix images not appearing on specific file	2024-04-03 10:20:46 +03:00
maverickstuder	9778ece992	RED-8702: Explore document databases to store entityLog * fix for duplicate images in document structure that are linked to multiple sections	2024-04-02 14:19:14 +02:00
Timo Bejan	8bd0de6263	Merge branch 'RED-8827' into 'main' Red 8827 See merge request fforesight/layout-parser!122	2024-03-22 12:25:36 +01:00
Timo Bejan	5c1708f97f	Issue with merging text blocks multiple times	2024-03-22 12:47:05 +02:00
Timo Bejan	a35d77be2e	ignore mc files	2024-03-22 10:22:00 +02:00
Dominique Eifländer	631160eb22	Merge branch 'RED-8627' into 'main' RED-8627: Fixed scrambled text after sorting See merge request fforesight/layout-parser!120	2024-03-19 11:09:48 +01:00
Dominique Eifländer	8e7e588d26	RED-8627: Fixed scrambled text after sorting	2024-03-19 10:58:36 +01:00
Dominique Eifländer	ac850c2626	Merge branch 'RED-7141' into 'main' RED-7141: Fixed more overlap problems See merge request fforesight/layout-parser!119	2024-03-14 16:46:10 +01:00
Dominique Eifländer	1d765a6baa	RED-7141: Fixed more overlap problems	2024-03-14 16:30:52 +01:00
Dominique Eifländer	c55984aa67	Merge branch 'RED-7141' into 'main' RED-7141: Fixed overlapping blocks See merge request fforesight/layout-parser!118	2024-03-14 09:09:52 +01:00
Dominique Eifländer	27aa418029	RED-7141: Fixed overlapping blocks	2024-03-13 16:14:55 +01:00
Dominique Eifländer	c4edff4696	Merge branch 'RED-7141' into 'main' RED-7141: Readded lost mergeLinesInZones See merge request fforesight/layout-parser!116	2024-03-12 13:49:09 +01:00
Dominique Eifländer	92fd1a72de	RED-7141: Readded lost mergeLinesInZones	2024-03-12 13:42:40 +01:00
Dominique Eifländer	0d3d25e7d7	Merge branch 'RED-7141-hotfix' into 'main' RED-7141: Align backend text sorting with Webviewer sorting See merge request fforesight/layout-parser!115	2024-03-12 11:15:41 +01:00
maverickstuder	956fbff872	RED-7141: Align backend text sorting with Webviewer sorting * hotfix for tables not being detected due to wrong x-y-sorting	2024-03-12 11:06:53 +01:00
Maverick Studer	2488009af1	Merge branch 'RED-8715' into 'main' RED-8715: Improve NearestNeighbor Algorithm in LayoutParser See merge request fforesight/layout-parser!114	2024-03-11 15:10:41 +01:00
maverickstuder	16be2467fd	RED-8715: Improve NearestNeighbor Algorithm in LayoutParser * replaced the old algorithm with an algorithm based on a kd-tree	2024-03-11 14:42:28 +01:00
Timo Bejan	f4cae8a7dc	Merge branch 'Clarifynd' into 'main' Clarifynd See merge request fforesight/layout-parser!113	2024-03-11 10:37:05 +01:00
Timo Bejan	dfc23955d7	Linespacing claryfind	2024-03-11 11:30:51 +02:00
Dominique Eifländer	d6e3d6fe22	Clarifynd	2024-03-11 11:24:58 +02:00
Timo Bejan	bef23e38b5	Merge branch 'clari-30' into 'main' CLARI-30 - forward analysis headers See merge request fforesight/layout-parser!112	2024-03-08 15:51:53 +01:00
Timo Bejan	65ab7a1912	CLARI-30 - forward analysis headers	2024-03-08 16:47:27 +02:00