Merge branch 'RED-8873' into 'main'

RED-8773 - Fix images not appearing on specific file See merge request fforesight/layout-parser!123
Merge branch 'RED-8799' into 'main'
2024-04-05 10:11:23 +02:00 · 2024-04-04 15:23:12 +02:00 · 2024-04-04 13:33:22 +02:00 · 2024-04-03 10:00:38 +02:00 · 2024-04-03 10:20:46 +03:00 · 2024-04-02 14:19:14 +02:00
11 changed files with 191 additions and 85 deletions
@@ -42,3 +42,5 @@ gradlew.bat
 gradlew
 gradle.properties
 gradle/
+.DS_Store
+.DS_Store/
@@ -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;
@@ -215,6 +215,7 @@ public class SectionsBuilderService {
                            .getX() <= xMax && image.getPosition().getY() >= yMin && image.getPosition().getY() <= yMax) {
                        section.getImages().add(image);
                        image.setAppendedToSection(true);
+                        break;
                    }
                }
                if (!image.isAppendedToSection()) {
@@ -221,9 +221,11 @@ public class DocstrumBlockificationService {
    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;
            }
@@ -232,7 +234,7 @@ public class DocstrumBlockificationService {

            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) {
@@ -249,13 +251,19 @@ public class DocstrumBlockificationService {
                    boolean toDuplicate = current.isToDuplicate() || inner.isToDuplicate();
                    current.getSequences().addAll(inner.getSequences());
                    current = buildTextBlock(current.getSequences(), 0);
+
                    current.setToDuplicate(toDuplicate);
-                    toRemove.add(inner);
+                    blocks.set(i, null);
                    itty.set(current);
                }
            }
        }
-        blocks.removeAll(toRemove);
+        var blocksIterator = blocks.iterator();
+        while(blocksIterator.hasNext()){
+            if(blocksIterator.next() == null){
+                blocksIterator.remove();
+            }
+        }
    }


@@ -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);
        }

@@ -109,23 +119,34 @@ public class DocumentGraphFactory {

    public void addImage(Section section, ClassifiedImage image, Context context) {

+        Image imageNode = createImage(image, context);
+        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(section, imageNode);
+        imageNode.setTreeId(treeId);
+    }
+
+
+    public void addImage(Document document, ClassifiedImage image, Context context) {
+
+        Image imageNode = createImage(image, context);
+        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(document, imageNode);
+        imageNode.setTreeId(treeId);
+    }
+
+
+    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 +185,12 @@ 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 +201,8 @@ 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 +214,8 @@ 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 +227,8 @@ 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);
@@ -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()));
@@ -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 +27,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
    @SneakyThrows
    public void testViewerDocument() {

-        String fileName = "files/100 Trinexapac-ethyl_RAR_20_Volume_3CP_B-9_ 2018-01-10.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);
    }
Author	SHA1	Message	Date
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