Merge branch 'RED-7434' into 'main'

RED-7434 - Remove Section Grid entirely See merge request fforesight/layout-parser!78
RED-7434 - Remove Section Grid entirely
2023-10-20 10:07:01 +02:00 · 2023-10-20 09:09:22 +03:00 · 2023-10-18 16:12:15 +02:00 · 2023-10-17 15:52:19 +02:00
8 changed files with 28 additions and 183 deletions
@@ -36,9 +36,6 @@ public record LayoutParsingRequest(
        @Schema(description = "Path where the Simplified Text File will be stored.")//
        @NonNull String simplifiedTextStorageId,//
        @Schema(description = "Path where the Viewer Document PDF will be stored.")//
-        @NonNull String viewerDocumentStorageId,//
-        @Deprecated//
-        @Schema(description = "Path where the Section Grid will be stored.")//
-        @NonNull String sectionGridStorageId) {
+        @NonNull String viewerDocumentStorageId) {

 }
@@ -35,7 +35,6 @@ import com.knecon.fforesight.service.layoutparser.processor.python_api.model.tab
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.services.BodyTextFrameService;
 import com.knecon.fforesight.service.layoutparser.processor.services.RulingCleaningService;
-import com.knecon.fforesight.service.layoutparser.processor.services.SectionGridCreatorService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SectionsBuilderService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SimplifiedSectionTextService;
 import com.knecon.fforesight.service.layoutparser.processor.services.TableExtractionService;
@@ -65,7 +64,6 @@ public class LayoutParsingPipeline {
    private final CvTableParsingAdapter cvTableParsingAdapter;
    private final LayoutParsingStorageService layoutParsingStorageService;
    private final SectionsBuilderService sectionsBuilderService;
-    private final SectionGridCreatorService sectionGridCreatorService;
    private final TaasClassificationService taasClassificationService;
    private final RedactManagerClassificationService redactManagerClassificationService;
    private final DocuMineClassificationService docuMineClassificationService;
@@ -99,7 +97,6 @@ public class LayoutParsingPipeline {

            int numberOfPages = originDocument.getNumberOfPages();

-            layoutParsingStorageService.storeSectionGrid(layoutParsingRequest, sectionGridCreatorService.createSectionGrid(documentGraph));
            layoutParsingStorageService.storeDocumentData(layoutParsingRequest, DocumentDataMapper.toDocumentData(documentGraph));
            layoutParsingStorageService.storeSimplifiedText(layoutParsingRequest, simplifiedSectionTextService.toSimplifiedText(documentGraph));

@@ -17,7 +17,6 @@ import org.apache.pdfbox.pdmodel.PDDocument;
 import org.springframework.stereotype.Service;

 import com.fasterxml.jackson.databind.ObjectMapper;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.section.SectionGrid;
 import com.iqser.red.storage.commons.service.StorageService;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentData;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedText;
@@ -84,11 +83,6 @@ public class LayoutParsingStorageService {
    }


-    public void storeSectionGrid(LayoutParsingRequest layoutParsingRequest, SectionGrid sectionGrid) {
-
-        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.sectionGridStorageId(), sectionGrid);
-    }
-

    public void storeResearchDocumentData(LayoutParsingRequest layoutParsingRequest, ResearchDocumentData researchDocumentData) {

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

-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.section.SectionGrid;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.UnclassifiedText;

@@ -25,7 +24,6 @@ public class ClassificationDocument {
    private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
    private boolean headlines;

-    private SectionGrid sectionGrid = new SectionGrid();
    private long rulesVersion;

 }
@@ -1,146 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.services;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Objects;
-import java.util.Set;
-import java.util.function.BiConsumer;
-import java.util.function.BinaryOperator;
-import java.util.function.Function;
-import java.util.function.Supplier;
-import java.util.stream.Collector;
-import java.util.stream.Stream;
-
-import org.springframework.stereotype.Service;
-
-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.section.CellRectangle;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.section.SectionGrid;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.section.SectionRectangle;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell;
-
-import lombok.RequiredArgsConstructor;
-
-@Service
-@RequiredArgsConstructor
-public class SectionGridCreatorService {
-
-    public SectionGrid createSectionGrid(Document document) {
-
-        Map<Integer, List<SectionRectangle>> sectionBBox = document.streamAllSubNodesOfType(NodeType.SECTION).map(SemanticNode::getBBox).collect(new SectionGridCollector());
-        Map<Integer, List<SectionRectangle>> paragraphBBox = document.streamAllSubNodesOfType(NodeType.PARAGRAPH).map(SemanticNode::getBBox).collect(new SectionGridCollector());
-        Map<Integer, List<SectionRectangle>> headlineBBox = document.streamAllSubNodesOfType(NodeType.HEADLINE).map(SemanticNode::getBBox).collect(new SectionGridCollector());
-        Map<Integer, List<SectionRectangle>> tableBBox = document.streamAllSubNodesOfType(NodeType.TABLE).map(node -> (Table) node).collect(new TableGridCollector());
-        var sectionGrid = new SectionGrid();
-
-        sectionGrid.setRectanglesPerPage(mergeMapsByConcatenatingLists(//
-                mergeMapsByConcatenatingLists(paragraphBBox, headlineBBox), //
-                mergeMapsByConcatenatingLists(sectionBBox, tableBBox)));
-
-        return sectionGrid;
-    }
-
-
-    private static abstract class GridCollector<T> implements Collector<T, Map<Integer, List<SectionRectangle>>, Map<Integer, List<SectionRectangle>>> {
-
-        @Override
-        public Supplier<Map<Integer, List<SectionRectangle>>> supplier() {
-
-            return HashMap::new;
-        }
-
-
-        @Override
-        public Function<Map<Integer, List<SectionRectangle>>, Map<Integer, List<SectionRectangle>>> finisher() {
-
-            return Function.identity();
-        }
-
-
-        @Override
-        public BinaryOperator<Map<Integer, List<SectionRectangle>>> combiner() {
-
-            return SectionGridCreatorService::mergeMapsByConcatenatingLists;
-        }
-
-
-        @Override
-        public Set<Characteristics> characteristics() {
-
-            return Set.of(Characteristics.IDENTITY_FINISH, Characteristics.CONCURRENT, Characteristics.UNORDERED);
-        }
-
-    }
-
-    private static class TableGridCollector extends GridCollector<Table> {
-
-        @Override
-        public BiConsumer<Map<Integer, List<SectionRectangle>>, Table> accumulator() {
-
-            return (map, table) -> table.getPages()
-                    .forEach(page -> map.merge(page.getNumber(), List.of(toSectionRectangle(table, page, table.getPages().size())), SectionGridCreatorService::concatLists));
-        }
-
-
-        private static SectionRectangle toSectionRectangle(Table table, Page page, int numberOfParts) {
-
-            Rectangle2D rect = table.getBBox().get(page);
-            List<CellRectangle> tableCellRectangles = table.streamTableCells()
-                    .map(TableCell::getBBox)
-                    .map(map -> map.get(page))
-                    .filter(Objects::nonNull)
-                    .map(rectangle2D -> new CellRectangle(new Point((float) rectangle2D.getX(), (float) rectangle2D.getY()),
-                            (float) rectangle2D.getWidth(),
-                            (float) rectangle2D.getHeight()))
-                    .toList();
-            return new SectionRectangle(new Point((float) rect.getX(), (float) rect.getY()),
-                    (float) rect.getWidth(),
-                    (float) rect.getHeight(),
-                    1,
-                    numberOfParts,
-                    tableCellRectangles);
-        }
-
-    }
-
-    private static class SectionGridCollector extends GridCollector<Map<Page, Rectangle2D>> {
-
-        @Override
-        public BiConsumer<Map<Integer, List<SectionRectangle>>, Map<Page, Rectangle2D>> accumulator() {
-
-            return (mapToKeep, mapToMerge) -> mapToMerge.forEach((page, rectangle) -> mapToKeep.merge(page.getNumber(),
-                    List.of(toSectionRectangle(rectangle, mapToMerge.values().size())),
-                    SectionGridCreatorService::concatLists));
-
-        }
-
-
-        private static SectionRectangle toSectionRectangle(Rectangle2D rect, int numberOfParts) {
-
-            return new SectionRectangle(new Point((float) rect.getX(), (float) rect.getY()), (float) rect.getWidth(), (float) rect.getHeight(), 1, numberOfParts, null);
-        }
-
-    }
-
-
-    private static Map<Integer, List<SectionRectangle>> mergeMapsByConcatenatingLists(Map<Integer, List<SectionRectangle>> mapToKeep,
-                                                                                      Map<Integer, List<SectionRectangle>> mapToMerge) {
-
-        mapToMerge.forEach((page, rectangle) -> mapToKeep.merge(page, rectangle, SectionGridCreatorService::concatLists));
-        return mapToKeep;
-    }
-
-
-    private static List<SectionRectangle> concatLists(List<SectionRectangle> l1, List<SectionRectangle> l2) {
-
-        return Stream.concat(l1.stream(), l2.stream()).toList();
-    }
-
-}
@@ -4,18 +4,6 @@ package com.knecon.fforesight.service.layoutparser.processor.services.blockifica
 // TODO: figure out, why this fails the build
 // import static com.knecon.fforesight.service.layoutparser.processor.services.factory.SearchTextWithTextPositionFactory.HEIGHT_PADDING;

-import java.util.ArrayList;
-import java.util.HashSet;
-import java.util.Iterator;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.Set;
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
-import java.util.stream.Stream;
-
-import org.springframework.stereotype.Service;
-
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
 import com.knecon.fforesight.service.layoutparser.processor.model.Orientation;
@@ -23,6 +11,12 @@ import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
 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.RulingTextDirAdjustUtil;
+import org.springframework.stereotype.Service;
+
+import java.util.*;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+import java.util.stream.Stream;

@Service
@SuppressWarnings("all")
@@ -33,9 +27,10 @@ public class TaasBlockificationService {
    private static final float INTERSECTS_Y_THRESHOLD = 4;// 2 * HEIGHT_PADDING // This is exactly 2 times our position height padding. This is required to find boxes that are visually intersecting.
    private static final int X_GAP_SPLIT_CONSTANT = 50;
    public static final int X_ALIGNMENT_THRESHOLD = 1;
-    public static final int SMALL_Y_GAP_THRESHOLD = 5;
    public static final int NEGATIVE_X_GAP_THRESHOLD = -5;

+    private Pattern listIdentifier = Pattern.compile("^(?:(?:[1-9]|1\\d|20|[ivxlc]|[a-z])\\s*(?:[.)]))|\\uF0B7", Pattern.CASE_INSENSITIVE);
+

    /**
     * This method is building blocks by expanding the minX/maxX and minY/maxY value on each word that is not split by the conditions.
@@ -80,16 +75,29 @@ public class TaasBlockificationService {
        List<TextPageBlock> currentTextBlocksToMerge = new LinkedList<>();
        textBlocksToMerge.add(currentTextBlocksToMerge);
        TextPageBlock previousTextBlock = null;
+        Float lastLineGap = null;
        for (TextPageBlock currentTextBlock : classificationTextBlocks) {
            if (previousTextBlock == null) {
                currentTextBlocksToMerge.add(currentTextBlock);
                previousTextBlock = currentTextBlock;
                continue;
            }
+
+
+            Matcher listIdentifierPattern = listIdentifier.matcher(currentTextBlock.getText());
+            boolean isListIdentifier = listIdentifierPattern.find();
+
+            boolean yGap = Math.abs(currentTextBlock.getPdfMaxY() - previousTextBlock.getPdfMinY()) < previousTextBlock.getMostPopularWordHeight() * Y_GAP_SPLIT_HEIGHT_MODIFIER;
+
+            boolean sameFont = previousTextBlock.getMostPopularWordFont().equals(currentTextBlock.getMostPopularWordFont()) && previousTextBlock.getMostPopularWordFontSize() == currentTextBlock.getMostPopularWordFontSize();
+//            boolean yGap = previousTextBlock != null && currentTextBlock.getMinYDirAdj() - maxY > Math.min(word.getHeight(), prev.getHeight()) * Y_GAP_SPLIT_HEIGHT_MODIFIER;
+
            boolean alignsXRight = Math.abs(currentTextBlock.getPdfMaxX() - previousTextBlock.getPdfMaxX()) < X_ALIGNMENT_THRESHOLD;
-            boolean smallYGap = Math.abs(currentTextBlock.getPdfMaxY() - previousTextBlock.getPdfMinY()) < SMALL_Y_GAP_THRESHOLD;
-            if (alignsXRight && smallYGap) {
+            boolean alignsXLeft = Math.abs(currentTextBlock.getPdfMinX() - previousTextBlock.getPdfMinX()) < X_ALIGNMENT_THRESHOLD;
+//            boolean smallYGap = Math.abs(currentTextBlock.getPdfMaxY() - previousTextBlock.getPdfMinY()) < yGap;
+            if (yGap && sameFont && !isListIdentifier) {
                currentTextBlocksToMerge.add(currentTextBlock);
+
            } else {
                currentTextBlocksToMerge = new LinkedList<>();
                currentTextBlocksToMerge.add(currentTextBlock);
@@ -170,8 +178,8 @@ public class TaasBlockificationService {


    private List<TextPageBlock> constructFineGranularTextPageBlocks(List<TextPositionSequence> textPositions,
-                                                                                        List<Ruling> horizontalRulingLines,
-                                                                                        List<Ruling> verticalRulingLines) {
+                                                                    List<Ruling> horizontalRulingLines,
+                                                                    List<Ruling> verticalRulingLines) {

        int indexOnPage = 0;
        List<TextPositionSequence> wordClusterToCombine = new ArrayList<>();
@@ -180,13 +188,13 @@ public class TaasBlockificationService {
        float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
        TextPositionSequence prev = null;
        // TODO: make static final constant
-        var listIdentitifier = Pattern.compile("\\b(?:[1-9]|1\\d|20|[ivxlc]|[a-z])\\s*(?:[.)])", Pattern.CASE_INSENSITIVE);
+

        boolean wasSplitted = false;
        Float splitX1 = null;
        for (TextPositionSequence word : textPositions) {

-            Matcher listIdentifierPattern = listIdentitifier.matcher(word.toString());
+            Matcher listIdentifierPattern = listIdentifier.matcher(word.toString());

            boolean yGap = prev != null && word.getMinYDirAdj() - maxY > Math.min(word.getHeight(), prev.getHeight()) * Y_GAP_SPLIT_HEIGHT_MODIFIER;
            boolean sameLine = prev != null && equalsWithThreshold(prev.getMinYDirAdj(), word.getMinYDirAdj());
@@ -51,7 +51,6 @@ public abstract class AbstractTest {
    protected final static String POSITION_FILE_ID = "positions";
    protected final static String PAGES_FILE_ID = "pages";
    protected final static String TENANT_ID = "tenant";
-    protected final static String SECTION_GRID_ID = "section";
    protected final static String VIEWER_DOCUMENT_ID = "viewer";
    protected final static String SIMPLIFIED_ID = "simplified";

@@ -68,7 +67,6 @@ public abstract class AbstractTest {
                .positionBlockFileStorageId(POSITION_FILE_ID)
                .pageFileStorageId(PAGES_FILE_ID)
                .simplifiedTextStorageId(SIMPLIFIED_ID)
-                .sectionGridStorageId(SECTION_GRID_ID)
                .viewerDocumentStorageId(VIEWER_DOCUMENT_ID)
                .build();
    }
@@ -114,7 +112,6 @@ public abstract class AbstractTest {
                .positionBlockFileStorageId(POSITION_FILE_ID)
                .pageFileStorageId(PAGES_FILE_ID)
                .simplifiedTextStorageId(SIMPLIFIED_ID)
-                .sectionGridStorageId(SECTION_GRID_ID)
                .viewerDocumentStorageId(VIEWER_DOCUMENT_ID)
                .build();
    }
Author	SHA1	Message	Date
Kilian Schüttler	9abdc6d44d	Merge branch 'RED-7434' into 'main' RED-7434 - Remove Section Grid entirely See merge request fforesight/layout-parser!78	2023-10-20 10:07:01 +02:00
Corina Olariu	3bab61c446	RED-7434 - Remove Section Grid entirely - remove sectionGrid relation (including SectionGridCreatorService) - update junit tests	2023-10-20 09:09:22 +03:00
Dominique Eifländer	d17517d3c3	Merge branch 'hotfix-bdr-doc' into 'main' hotfix: Fixed parsing for specific taas document See merge request fforesight/layout-parser!77	2023-10-18 16:12:15 +02:00
Dominique Eifländer	567cbc178b	hotfix: Fixed parsing for specific taas document	2023-10-17 15:52:19 +02:00