TAAS-41: TAAS Document Structure

* added linebreaks to ParagraphData
* moved List<String> cellText to List<ParagraphData> cellTexts

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/taas/ParagraphData.java

@@ -12,6 +12,7 @@ public class ParagraphData {
     private String text;
     List<Range> boldTextBoundaries;
     List<Range> italicTextBoundaries;
+    List<Integer> linebreaks;
     private String classification;
 
     private String orientation;

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/taas/RowData.java

@@ -10,6 +10,6 @@ import lombok.Data;
 public class RowData {
 
     boolean header;
-    List<String> cellText;
+    List<ParagraphData> cellText;
     float[] bBox;
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/mapper/taas/TaasDocumentDataMapper.java

@@ -13,6 +13,7 @@ import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.Researc
 import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.RowData;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.StructureObject;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.TableData;
+import com.knecon.fforesight.service.layoutparser.processor.graph.Boundary;
 import com.knecon.fforesight.service.layoutparser.processor.graph.nodes.Document;
 import com.knecon.fforesight.service.layoutparser.processor.graph.nodes.Page;
 import com.knecon.fforesight.service.layoutparser.processor.graph.nodes.SemanticNode;
@@ -23,6 +24,7 @@ import com.knecon.fforesight.service.layoutparser.processor.graph.textblock.Text
 public class TaasDocumentDataMapper {
 
     public static ResearchDocumentData fromDocument(Document document) {
+
         AtomicInteger structureObjectNumber = new AtomicInteger();
         List<StructureObject> structureObjects = document.streamAllSubNodes()
                 .filter(node -> !node.getType().equals(NodeType.TABLE_CELL))
@@ -38,24 +40,32 @@ public class TaasDocumentDataMapper {
         return ResearchDocumentData.builder().structureObjects(structureObjects).build();
     }
 
+
     public static ParagraphData fromTextBlock(String classification, TextBlock textBlock) {
 
         return ParagraphData.builder()
                 .boldTextBoundaries(textBlock.getBoldTextBoundaries().stream().map(b -> new Range(b.start(), b.end())).toList())
                 .italicTextBoundaries(textBlock.getItalicTextBoundaries().stream().map(b -> new Range(b.start(), b.end())).toList())
                 .text(textBlock.getSearchText())
+                .linebreaks(textBlock.getLineBreaks())
                 .classification(classification)
                 .orientation(textBlock.getOrientation())
                 .textDirection(textBlock.getTextDirection())
                 .build();
     }
 
+
     public static TableData fromTable(Table table) {
 
-        List<RowData> rowData = IntStream.range(0, table.getNumberOfRows()).boxed().map(rowIdx -> table.streamRow(rowIdx).toList()).map(TaasDocumentDataMapper::fromTableCells).toList();
+        List<RowData> rowData = IntStream.range(0, table.getNumberOfRows())
+                .boxed()
+                .map(rowIdx -> table.streamRow(rowIdx).toList())
+                .map(TaasDocumentDataMapper::fromTableCells)
+                .toList();
         return new TableData(rowData, table.getNumberOfCols(), table.getNumberOfRows());
     }
 
+
     public static RowData fromTableCells(List<TableCell> tableCells) {
 
         if (tableCells.isEmpty()) {
@@ -67,8 +77,20 @@ public class TaasDocumentDataMapper {
             map1.forEach((page, rectangle) -> map2.merge(page, rectangle, (rect1, rect2) -> rect1.createUnion(rect2).getBounds2D()));
             return map2;
         }).orElseThrow().get(firstPage);
-        List<String> cellText = tableCells.stream().map(TableCell::getTextBlock).map(TextBlock::getSearchText).toList();
+        List<TextBlock> textBlocks = tableCells.stream().map(TableCell::getTextBlock).toList();
-        return new RowData(header, cellText, toFloatArray(bBox));
+        return new RowData(header, textBlocks.stream().map(textBlock -> TaasDocumentDataMapper.fromTextBlock("table_cell", textBlock)).toList(), toFloatArray(bBox));
+    }
+
+
+    private static Range toRange(Boundary boundary) {
+
+        return new Range(boundary.start(), boundary.end());
+    }
+
+
+    private static List<Range> toRange(List<Boundary> boundary) {
+
+        return boundary.stream().map(TaasDocumentDataMapper::toRange).toList();
     }
 
 
@@ -101,8 +123,10 @@ public class TaasDocumentDataMapper {
                 .build();
     }
 
+
     private static float[] toFloatArray(Rectangle2D bBox) {
 
         return new float[]{(float) bBox.getX(), (float) bBox.getY(), (float) bBox.getWidth(), (float) bBox.getHeight()};
     }
+
 }

layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/graph/DocumentDataTests.java

@@ -0,0 +1,5 @@
+package com.knecon.fforesight.service.layoutparser.server.graph;
+
+public class DocumentDataTests {
+
+}