Marmelator/layout-parser
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 289 Wiki Activity

RED-8747 - Entities not merged properly - fp

Browse Source 
- rework the extraction of rulings from the table cells
This commit is contained in:
Corina Olariu 2024-04-09 14:38:48 +03:00
parent 319268c53d
commit 976f408237
4 changed files with 33 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/BlockificationUtils.java
View File

@ -1,24 +0,0 @@
package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
import java.awt.geom.Point2D;
import java.util.List;
import com.knecon.fforesight.service.layoutparser.processor.model.table.Cell;
import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
import lombok.experimental.UtilityClass;
@UtilityClass
public class BlockificationUtils {
public void extractedRullingsFromCells(List<Cell> cells, List<Ruling> usedHorizonalRulings, List<Ruling> usedVerticalRulings) {
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)));
});
}
}

10
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/DocstrumBlockificationService.java
View File

@ -16,11 +16,13 @@ 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.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.RectangleTransformations;
import com.knecon.fforesight.service.layoutparser.processor.utils.RulingTextDirAdjustUtil;
import lombok.RequiredArgsConstructor;
@ -37,14 +39,10 @@ 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<>();
BlockificationUtils.extractedRullingsFromCells(cells, usedHorizonalRulings, usedVerticalRulings);
CleanRulings usedRulings = RectangleTransformations.extractRulings(cells);
var zones = docstrumSegmentationService.segmentPage(textPositions, xyOrder);
var pageBlocks = toAbstractPageBlocks(zones, usedHorizonalRulings, usedVerticalRulings, xyOrder);
var pageBlocks = toAbstractPageBlocks(zones, usedRulings.getHorizontal(), usedRulings.getVertical(), xyOrder);
var classificationPage = new ClassificationPage(pageBlocks);

10
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/RedactManagerBlockificationService.java
View File

@ -14,10 +14,12 @@ import com.knecon.fforesight.service.layoutparser.processor.model.Classification
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")
@ -37,11 +39,7 @@ public class RedactManagerBlockificationService {
*/
public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Cell> cells) {
// 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<>();
BlockificationUtils.extractedRullingsFromCells(cells, usedHorizonalRulings, usedVerticalRulings);
CleanRulings usedRulings = RectangleTransformations.extractRulings(cells);
int indexOnPage = 0;
List<TextPositionSequence> chunkWords = new ArrayList<>();
@ -59,7 +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, usedHorizonalRulings, usedVerticalRulings);
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)) {

26
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/utils/RectangleTransformations.java
View File

@ -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();
}