RED-7141: Improved basic block combination logic

This commit is contained in:
Dominique Eifländer 2024-02-22 16:57:24 +01:00
parent d06933ed17
commit d0e1af3a44
2 changed files with 17 additions and 3 deletions

View File

@ -45,6 +45,12 @@ public abstract class AbstractPageBlock {
}
public boolean containsBlock(TextPageBlock other, float threshold) {
return this.minX <= other.getMinX() + threshold && this.maxX >= other.getMaxX() - threshold && this.minY <= other.getMinY() + threshold && this.maxY >= other.getMaxY() - threshold;
}
public boolean contains(AbstractPageBlock other) {
return this.minX <= other.minX && this.maxX >= other.maxX && this.minY >= other.minY && this.maxY <= other.maxY;

View File

@ -7,6 +7,7 @@ import java.util.Comparator;
import java.util.List;
import java.util.ListIterator;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import org.springframework.stereotype.Service;
@ -68,24 +69,31 @@ public class DocstrumBlockificationService {
if (previous != null) {
if ((Math.abs(previous.getMaxY() - current.getMaxY()) < THRESHOLD || Math.abs(previous.getMinY() - current.getMinY()) < THRESHOLD) && (previous.getNumberOfLines() == 1 && current.getNumberOfLines() >= 1 && current.getNumberOfLines() <= 5 || previous.getNumberOfLines() == 2 && current.getNumberOfLines() == 1)) {
if ((Math.abs(previous.getMaxY() - current.getMaxY()) < THRESHOLD || Math.abs(previous.getMinY() - current.getMinY()) < THRESHOLD) && (previous.getNumberOfLines() == 1 && current.getNumberOfLines() >= 1 /* && current.getNumberOfLines() <= 10 */ || previous.getNumberOfLines() == 2 && current.getNumberOfLines() == 1)) {
previous.getSequences().addAll(current.getSequences());
previous = buildTextBlock(previous.getSequences(), 0);
itty.remove();
continue;
}
if ((Math.abs(previous.getMaxY() - current.getMaxY()) < THRESHOLD || Math.abs(previous.getMinY() - current.getMinY()) < THRESHOLD) && (previous.getNumberOfLines() >= 2 && current.getNumberOfLines() >= 2 && current.getNumberOfLines() <= 5 && previous.getNumberOfLines() <= current.getNumberOfLines())) {
if ((Math.abs(previous.getMaxY() - current.getMaxY()) < THRESHOLD || Math.abs(previous.getMinY() - current.getMinY()) < THRESHOLD) && (previous.getNumberOfLines() >= 2 && current.getNumberOfLines() >= 2 /* && current.getNumberOfLines() <= 10 */ && previous.getNumberOfLines() <= current.getNumberOfLines())) {
previous.getSequences().addAll(current.getSequences());
previous = buildTextBlock(previous.getSequences(), 0);
itty.remove();
// Might be a left/right mapping add one sorted as well
var sortedDublicate = buildTextBlock(previous.getSequences().stream().sorted(new TextPositionSequenceComparator()).toList(), 0);
var sortedDublicate = buildTextBlock(previous.getSequences().stream().sorted(new TextPositionSequenceComparator()).collect(Collectors.toList()), 0);
itty.add(sortedDublicate);
continue;
}
if (previous.containsBlock(current, THRESHOLD)) {
previous.getSequences().addAll(current.getSequences());
previous = buildTextBlock(previous.getSequences().stream().sorted(new TextPositionSequenceComparator()).collect(Collectors.toList()), 0);
itty.remove();
continue;
}
}
previous = current;
}