Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

RED-7082: getBBox() Performance Improvement

Browse Source
This commit is contained in:
Kilian Schüttler 2023-07-04 20:19:08 +02:00
parent ce66d9104d
commit 4d00b01514
4 changed files with 68 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/nodes/SemanticNode.java
View File

@ -465,7 +465,7 @@ public interface SemanticNode {
private Map<Page, Rectangle2D> getBBoxFromLeafTextBlock(Map<Page, Rectangle2D> bBoxPerPage) {
Map<Page, List<AtomicTextBlock>> atomicTextBlockPerPage = getTextBlock().getAtomicTextBlocks().stream().collect(Collectors.groupingBy(AtomicTextBlock::getPage));
atomicTextBlockPerPage.forEach((page, atbs) -> bBoxPerPage.put(page, RectangleTransformations.bBoxUnionAtomicTextBlock(atbs)));
atomicTextBlockPerPage.forEach((page, atbs) -> bBoxPerPage.put(page, RectangleTransformations.atomicTextBlockBBox(atbs)));
return bBoxPerPage;
}

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/textblock/AtomicTextBlock.java
View File

@ -191,7 +191,7 @@ public class AtomicTextBlock implements TextBlock {
List<Rectangle2D> rectanglesPerLine = stringBoundary.split(getAllLineBreaksInBoundary(stringBoundary))
.stream()
.map(this::getPositions)
.map(RectangleTransformations::rectangleUnionWithGaps)
.map(RectangleTransformations::rectangleBBoxWithGaps)
.flatMap(Collection::stream)
.toList();
Map<Page, List<Rectangle2D>> rectanglePerLinePerPage = new HashMap<>();

88
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/utils/RectangleTransformations.java
View File

@ -1,6 +1,5 @@
package com.iqser.red.service.redaction.v1.server.layoutparsing.document.utils;
import java.awt.geom.Area;
import java.awt.geom.Rectangle2D;
import java.awt.geom.RectangularShape;
import java.util.Collections;
@ -19,11 +18,14 @@ import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlo
import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.textblock.AtomicTextBlock;
import lombok.AllArgsConstructor;
import lombok.NoArgsConstructor;
public class RectangleTransformations {
public static PDRectangle toPDRectangleUnion(List<Rectangle> rectangles) {
public static PDRectangle toPDRectangleBBox(List<Rectangle> rectangles) {
Rectangle2D rectangle2D = RectangleTransformations.bBoxUnionRectangle(rectangles);
Rectangle2D rectangle2D = RectangleTransformations.rectangleBBox(rectangles);
PDRectangle annotationPosition = new PDRectangle();
annotationPosition.setLowerLeftX((float) rectangle2D.getMinX());
@ -34,15 +36,15 @@ public class RectangleTransformations {
}
public static Rectangle2D bBoxUnionAtomicTextBlock(List<AtomicTextBlock> atomicTextBlocks) {
public static Rectangle2D atomicTextBlockBBox(List<AtomicTextBlock> atomicTextBlocks) {
return atomicTextBlocks.stream().flatMap(atomicTextBlock -> atomicTextBlock.getPositions().stream()).collect(new Rectangle2DUnion());
return atomicTextBlocks.stream().flatMap(atomicTextBlock -> atomicTextBlock.getPositions().stream()).collect(new Rectangle2DBBoxCollector());
}
public static Rectangle2D bBoxUnionRectangle(List<Rectangle> rectangles) {
public static Rectangle2D rectangleBBox(List<Rectangle> rectangles) {
return rectangles.stream().map(RectangleTransformations::toRectangle2D).collect(new Rectangle2DUnion());
return rectangles.stream().map(RectangleTransformations::toRectangle2D).collect(new Rectangle2DBBoxCollector());
}
@ -64,9 +66,9 @@ public class RectangleTransformations {
}
public static Rectangle2D rectangleUnion(List<Rectangle2D> rectangle2DList) {
public static Rectangle2D rectangle2DBBox(List<Rectangle2D> rectangle2DList) {
return rectangle2DList.stream().collect(new Rectangle2DUnion());
return rectangle2DList.stream().collect(new Rectangle2DBBoxCollector());
}
@ -76,7 +78,7 @@ public class RectangleTransformations {
* @param rectangle2DList A list of rectangles to combine
* @return A list of rectangles which are combined if they are closer than the split threshold
*/
public static List<Rectangle2D> rectangleUnionWithGaps(List<Rectangle2D> rectangle2DList) {
public static List<Rectangle2D> rectangleBBoxWithGaps(List<Rectangle2D> rectangle2DList) {
if (rectangle2DList.isEmpty()) {
return Collections.emptyList();
@ -98,47 +100,85 @@ public class RectangleTransformations {
previousRectangle = currentRectangle;
}
}
return rectangleListsWithGaps.stream().map(RectangleTransformations::rectangleUnion).toList();
return rectangleListsWithGaps.stream().map(RectangleTransformations::rectangle2DBBox).toList();
}
private static class Rectangle2DUnion implements Collector<Rectangle2D, Area, Rectangle2D> {
private static class Rectangle2DBBoxCollector implements Collector<Rectangle2D, Rectangle2DBBoxCollector.BBox, Rectangle2D> {
@Override
public Supplier<Area> supplier() {
public Supplier<BBox> supplier() {
return Area::new;
return BBox::new;
}
@Override
public BiConsumer<Area, Rectangle2D> accumulator() {
public BiConsumer<BBox, Rectangle2D> accumulator() {
return (area, rectangle2D) -> area.add(new Area(rectangle2D));
return (bb, rect) -> bb.addRectangle(rect.getMinX(), rect.getMinY(), rect.getMaxX(), rect.getMaxY());
}
@Override
public BinaryOperator<Area> combiner() {
public BinaryOperator<BBox> combiner() {
return (area1, area2) -> {
area1.add(area2);
return area1;
};
return (b1, b2) -> new BBox(Math.min(b1.lowerLeftX, b2.lowerLeftX),
Math.min(b1.lowerLeftY, b2.lowerLeftY),
Math.max(b1.upperRightX, b2.upperRightX),
Math.max(b1.upperRightY, b2.upperRightY));
}
@Override
public Function<Area, Rectangle2D> finisher() {
public Function<BBox, Rectangle2D> finisher() {
return Area::getBounds2D;
return bb -> new Rectangle2D.Double(bb.lowerLeftX, bb.lowerLeftY, bb.upperRightX - bb.lowerLeftX, bb.upperRightY - bb.lowerLeftY);
}
@Override
public Set<Characteristics> characteristics() {
return Set.of(Characteristics.CONCURRENT, Characteristics.UNORDERED);
return Set.of(Characteristics.UNORDERED);
}
@AllArgsConstructor
@NoArgsConstructor
private static class BBox {
Double lowerLeftX;
Double lowerLeftY;
Double upperRightX;
Double upperRightY;
public void addRectangle(double lowerLeftX, double lowerLeftY, double upperRightX, double upperRightY) {
if (this.lowerLeftX == null) {
this.lowerLeftX = lowerLeftX;
} else if (this.lowerLeftX > lowerLeftX) {
this.lowerLeftX = lowerLeftX;
}
if (this.lowerLeftY == null) {
this.lowerLeftY = lowerLeftY;
} else if (this.lowerLeftY > lowerLeftY) {
this.lowerLeftY = lowerLeftY;
}
if (this.upperRightX == null) {
this.upperRightX = upperRightX;
} else if (this.upperRightX < upperRightX) {
this.upperRightX = upperRightX;
}
if (this.upperRightY == null) {
this.upperRightY = upperRightY;
} else if (this.upperRightY < upperRightY) {
this.upperRightY = upperRightY;
}
}
}
}

4
redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/annotate/AnnotationService.java
View File

@ -1,6 +1,6 @@
package com.iqser.red.service.redaction.v1.server.annotate;
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.utils.RectangleTransformations.toPDRectangleUnion;
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.utils.RectangleTransformations.toPDRectangleBBox;
import java.awt.Color;
import java.awt.geom.Rectangle2D;
@ -136,7 +136,7 @@ public class AnnotationService {
PDAnnotationTextMarkup annotation = new PDAnnotationTextMarkup(PDAnnotationTextMarkup.SUB_TYPE_HIGHLIGHT);
annotation.constructAppearances();
PDRectangle pdRectangle = toPDRectangleUnion(rectangles);
PDRectangle pdRectangle = toPDRectangleBBox(rectangles);
annotation.setRectangle(pdRectangle);
annotation.setQuadPoints(Floats.toArray(toQuadPoints(rectangles)));
if (!redactionLogEntry.isHint()) {