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

RED-6009 - Document Tree Structure

Browse Source 
* added gaps in Redaction Positions, if they are farther than five character widths apart
This commit is contained in:
Kilian Schuettler 2023-05-10 13:00:18 +02:00
parent ce0a65c06a
commit cdbccd83e3
7 changed files with 55 additions and 11 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/entity/RedactionEntity.java
View File

@ -137,7 +137,7 @@ public class RedactionEntity {
public List<RedactionPosition> getRedactionPositionsPerPage() {
if (redactionPositionsPerPage == null || redactionPositionsPerPage.isEmpty()) {
Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = deepestFullyContainingNode.buildTextBlock().getEntityPositionsPerPage(boundary);
Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = deepestFullyContainingNode.buildTextBlock().getPositionsPerPage(boundary);
Page firstPage = rectanglesPerLinePerPage.keySet()
.stream()

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

@ -379,7 +379,7 @@ public interface SemanticNode {
/**
* TODO this does not yet work for sections spanning multiple columns.
* TODO: this produces unwanted results for sections spanning multiple columns.
*
* @param bBoxPerPage initial empty BoundingBox
* @return The union of the BoundingBoxes of all children

14
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

@ -3,6 +3,7 @@ package com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.t
import static java.lang.String.format;
import java.awt.geom.Rectangle2D;
import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
@ -119,12 +120,13 @@ public class AtomicTextBlock implements TextBlock {
}
public Map<Page, List<Rectangle2D>> getEntityPositionsPerPage(Boundary stringBoundary) {
public Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary) {
List<Rectangle2D> rectanglesPerLine = stringBoundary.split(getLineBreaks().stream().map(lb -> lb + boundary.start()).filter(stringBoundary::contains).toList())
List<Rectangle2D> rectanglesPerLine = stringBoundary.split(getAllLineBreaksInBoundary(stringBoundary))
.stream()
.map(this::getPositions)
.map(RectangleTransformations::rectangleUnion)
.map(RectangleTransformations::rectangleUnionWithGaps)
.flatMap(Collection::stream)
.toList();
Map<Page, List<Rectangle2D>> rectanglePerLinePerPage = new HashMap<>();
rectanglePerLinePerPage.put(page, rectanglesPerLine);
@ -132,6 +134,12 @@ public class AtomicTextBlock implements TextBlock {
}
private List<Integer> getAllLineBreaksInBoundary(Boundary boundary) {
return getLineBreaks().stream().map(linebreak -> linebreak + this.boundary.start()).filter(boundary::contains).toList();
}
@Override
public String toString() {

10
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/textblock/ConcatenatedTextBlock.java
View File

@ -137,23 +137,23 @@ public class ConcatenatedTextBlock implements TextBlock {
@Override
public Map<Page, List<Rectangle2D>> getEntityPositionsPerPage(Boundary stringBoundary) {
public Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary) {
List<AtomicTextBlock> textBlocks = getAllAtomicTextBlocksPartiallyInStringBoundary(stringBoundary);
if (textBlocks.size() == 1) {
return textBlocks.get(0).getEntityPositionsPerPage(stringBoundary);
return textBlocks.get(0).getPositionsPerPage(stringBoundary);
}
AtomicTextBlock firstTextBlock = textBlocks.get(0);
Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = firstTextBlock.getEntityPositionsPerPage(new Boundary(stringBoundary.start(), firstTextBlock.getBoundary().end()));
Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = firstTextBlock.getPositionsPerPage(new Boundary(stringBoundary.start(), firstTextBlock.getBoundary().end()));
for (AtomicTextBlock textBlock : textBlocks.subList(1, textBlocks.size() - 1)) {
rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage, textBlock.getEntityPositionsPerPage(textBlock.getBoundary()));
rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage, textBlock.getPositionsPerPage(textBlock.getBoundary()));
}
AtomicTextBlock lastTextBlock = textBlocks.get(textBlocks.size() - 1);
rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage, lastTextBlock.getEntityPositionsPerPage(lastTextBlock.getBoundary()));
rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage, lastTextBlock.getPositionsPerPage(lastTextBlock.getBoundary()));
return rectanglesPerLinePerPage;
}

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

@ -39,7 +39,7 @@ public interface TextBlock extends CharSequence {
List<Rectangle2D> getPositions(Boundary stringBoundary);
Map<Page, List<Rectangle2D>> getEntityPositionsPerPage(Boundary stringBoundary);
Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary);
int numberOfLines();

35
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/utils/RectangleTransformations.java
View File

@ -2,6 +2,9 @@ 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;
import java.util.LinkedList;
import java.util.List;
import java.util.Set;
import java.util.function.BiConsumer;
@ -61,6 +64,38 @@ public class RectangleTransformations {
}
/**
* If two rectangles are further apart than five times the average width of a rectangle, a gap is inserted.
*
* @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) {
if (rectangle2DList.isEmpty()) {
return Collections.emptyList();
}
double split_threshold = rectangle2DList.stream().mapToDouble(RectangularShape::getWidth).average().orElse(5) * 5.0;
List<List<Rectangle2D>> rectangleListsWithGaps = new LinkedList<>();
List<Rectangle2D> rectangleListWithoutGaps = new LinkedList<>();
rectangleListsWithGaps.add(rectangleListWithoutGaps);
Rectangle2D previousRectangle = rectangle2DList.get(0);
for (Rectangle2D currentRectangle : rectangle2DList) {
if (Math.abs(currentRectangle.getMinX() - previousRectangle.getMaxX()) > split_threshold) {
rectangleListWithoutGaps = new LinkedList<>();
rectangleListWithoutGaps.add(currentRectangle);
rectangleListsWithGaps.add(rectangleListWithoutGaps);
previousRectangle = currentRectangle;
} else {
rectangleListWithoutGaps.add(currentRectangle);
previousRectangle = currentRectangle;
}
}
return rectangleListsWithGaps.stream().map(RectangleTransformations::rectangleUnion).toList();
}
private static class Rectangle2DUnion implements Collector<Rectangle2D, Area, Rectangle2D> {
@Override

1
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/RuleBuilderModelService.java
View File

@ -9,6 +9,7 @@ import com.iqser.red.service.redaction.v1.model.RuleBuilderModel;
@Service
public class RuleBuilderModelService {
// TODO: Evaluate if we need to implement this or remove it entirely.
public RuleBuilderModel getRuleBuilderModel() {
RuleBuilderModel ruleBuilderModel = new RuleBuilderModel();