RED-8932 Fixed not merged headline with identifier

RED-7384: add empty textBlock to Image to ensure continuous textranges across all SemanticNodes

See merge request fforesight/layout-parser!139

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Image.java

@@ -12,7 +12,6 @@ import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.No
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -36,6 +35,8 @@ public class Image implements GenericSemanticNode {
     boolean transparent;
     Rectangle2D position;
 
+    TextBlock leafTextBlock;
+
     boolean redaction;
     boolean ignored;
     @Builder.Default
@@ -66,7 +67,7 @@ public class Image implements GenericSemanticNode {
     @Override
     public TextBlock getTextBlock() {
 
-        return streamAllSubNodes().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
+        return leafTextBlock;
     }
 
 
@@ -92,4 +93,11 @@ public class Image implements GenericSemanticNode {
         return bBoxPerPage;
     }
 
+
+    @Override
+    public boolean isLeaf() {
+
+        return true;
+    }
+
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/SectionsBuilderService.java

@@ -43,7 +43,6 @@ public class SectionsBuilderService {
         for (ClassificationPage page : document.getPages()) {
             List<TextPageBlock> header = new ArrayList<>();
             List<TextPageBlock> footer = new ArrayList<>();
-            List<TextPageBlock> unclassifiedText = new ArrayList<>();
             for (AbstractPageBlock current : page.getTextBlocks()) {
 
                 if (current.getClassification() == null) {
@@ -62,11 +61,6 @@ public class SectionsBuilderService {
                     continue;
                 }
 
-                if (current.getClassification().equals(PageBlockType.OTHER)) {
-                    unclassifiedText.add((TextPageBlock) current);
-                    continue;
-                }
-
                 if (prev != null && current.getClassification().isHeadline() && !prev.getClassification().isHeadline() || !document.isHeadlines()) {
                     ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
                     chunkBlock.setHeadline(lastHeadline);
@@ -94,9 +88,6 @@ public class SectionsBuilderService {
             if (!footer.isEmpty()) {
                 footers.add(new ClassificationFooter(footer));
             }
-            if (!unclassifiedText.isEmpty()) {
-                unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
-            }
         }
 
         ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/blockification/DocuMineBlockificationService.java

@@ -61,8 +61,8 @@ public class DocuMineBlockificationService {
             boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj();
             boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines);
             boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
-            boolean splitByOtherFontAndOtherY = prev != null && prev.getMaxYDirAdj() != word.getMaxYDirAdj() && (word.getFontStyle().contains("bold") && !prev.getFontStyle()
-                    .contains("bold") || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold"));
+            boolean splitByOtherFontAndOtherY = prev != null && Math.abs(prev.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight() * 0.2 && (word.getFontStyle()
+                    .contains("bold") && !prev.getFontStyle().contains("bold") || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold"));
 
             Matcher matcher = pattern.matcher(chunkWords.stream().collect(Collectors.joining(" ")).toString());
             boolean startsOnSameX = Math.abs(minX - word.getMinXDirAdj()) < 5 && matcher.matches();

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/DocumentGraphFactory.java

@@ -18,8 +18,6 @@ import com.knecon.fforesight.service.layoutparser.processor.model.Classification
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationFooter;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationHeader;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
-import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Footer;
@@ -31,6 +29,8 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Pa
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
 import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.utils.IdBuilder;
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionOperations;
 
@@ -99,19 +99,12 @@ public class DocumentGraphFactory {
     }
 
 
-    public void addImage(Section section, ClassifiedImage image, Context context) {
+    public void addImage(GenericSemanticNode parent, ClassifiedImage image, Context context) {
 
         Image imageNode = createImage(image, context);
-        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(section, imageNode);
-        imageNode.setTreeId(treeId);
-    }
-
-
-    public void addImage(Document document, ClassifiedImage image, Context context) {
-
-        Image imageNode = createImage(image, context);
-        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(document, imageNode);
+        List<Integer> treeId = context.getDocumentTree().createNewChildEntryAndReturnId(parent, imageNode);
         imageNode.setTreeId(treeId);
+        imageNode.setLeafTextBlock(context.textBlockFactory.emptyTextBlock(parent, context, context.getPage(image.getPage())));
     }
 
 

layoutparser-service/viewer-doc-processor/src/main/java/com/knecon/fforesight/service/viewerdoc/service/ViewerDocumentService.java

@@ -17,6 +17,7 @@ import java.util.stream.Collectors;
 
 import org.apache.pdfbox.Loader;
 import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.pdfwriter.compress.CompressParameters;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDDocumentCatalog;
 import org.apache.pdfbox.pdmodel.PDPage;
@@ -287,7 +288,7 @@ public class ViewerDocumentService {
      */
         Observation.createNotStarted("ViewerDocumentService", registry).contextualName("incremental-save").observe(() -> {
             try (var out = new FileOutputStream(outputFile)) {
-                pdDocument.saveIncremental(out);
+                pdDocument.save(out, CompressParameters.NO_COMPRESSION);
             } catch (IOException e) {
                 throw new RuntimeException(e);
             }