RED-419: Avoid duplicate entries

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/utils/IdBuilder.java

@@ -18,7 +18,7 @@ public class IdBuilder {
 
         StringBuilder sb = new StringBuilder();
         crossSequenceParts.forEach(sequencePart -> sequencePart.getTextPositions().forEach(textPosition -> {
-            sb.append(textPosition.getTextMatrix());
+            sb.append(textPosition.getTextMatrix()).append(sequencePart.getPage());
         }));
 
         return hashFunction.hashString(sb.toString(), StandardCharsets.UTF_8).toString();

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/visualization/service/AnnotationHighlightService.java

@@ -95,6 +95,10 @@ public class AnnotationHighlightService {
 
         List<PDAnnotation> annotations = pdPage.getAnnotations();
 
+        // Duplicates can exist due table extraction colums over multiple rows.
+        Set<String> processedIds = new HashSet<>();
+
+        entityLoop:
         for (Entity entity : classifiedDoc.getEntities().get(page)) {
 
             if (flatRedaction && !isRedactionType(entity)) {
@@ -104,7 +108,15 @@ public class AnnotationHighlightService {
             RedactionLogEntry redactionLogEntry = createRedactionLogEntry(entity);
             boolean requestedToRemove = false;
             List<Comment> comments = null;
+
             for (EntityPositionSequence entityPositionSequence : entity.getPositionSequences()) {
+                if (processedIds.contains(entityPositionSequence.getId())) {
+
+                    // TODO refactor this outer loop jump as soon as we have the time.
+                    continue entityLoop;
+                } else {
+                    processedIds.add(entityPositionSequence.getId());
+                }
 
                 if (manualRedactions != null && !manualRedactions.getIdsToRemove().isEmpty()) {
                     for (IdRemoval manualRemoval : manualRedactions.getIdsToRemove()) {
@@ -142,13 +154,18 @@ public class AnnotationHighlightService {
                     }
 
                     redactionLogEntry.getPositions().addAll(rectanglesPerLine);
+
                     annotations.addAll(createAnnotation(rectanglesPerLine, entityPositionSequence.getId(), createAnnotationContent(entity), getColor(entity, requestedToRemove), comments, !isHint(entity)));
                 }
 
                 redactionLogEntry.setId(entityPositionSequence.getId());
             }
+
+            // FIXME ids should never be null. Figure out why this happens.
+            if (redactionLogEntry.getId() != null) {
                 classifiedDoc.getRedactionLogEntities().add(redactionLogEntry);
             }
+        }
 
     }
 

redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java

@@ -51,6 +51,7 @@ import com.iqser.red.service.redaction.v1.model.ManualRedactionEntry;
 import com.iqser.red.service.redaction.v1.model.ManualRedactions;
 import com.iqser.red.service.redaction.v1.model.Point;
 import com.iqser.red.service.redaction.v1.model.Rectangle;
+import com.iqser.red.service.redaction.v1.model.RedactionLogEntry;
 import com.iqser.red.service.redaction.v1.model.RedactionRequest;
 import com.iqser.red.service.redaction.v1.model.RedactionResult;
 import com.iqser.red.service.redaction.v1.model.Status;
@@ -262,7 +263,16 @@ public class RedactionIntegrationTest {
                     .document(IOUtils.toByteArray(new FileInputStream(path)))
                     .build();
             System.out.println("Redacting file : " + path.getName());
-            redactionController.redact(request);
+            RedactionResult result = redactionController.redact(request);
+
+            Map<String, List<RedactionLogEntry>> duplicates = new HashMap<>();
+            result.getRedactionLog().getRedactionLogEntry().forEach(entry -> {
+                duplicates.computeIfAbsent(entry.getId(), v -> new ArrayList<>()).add(entry);
+            });
+
+            duplicates.entrySet().forEach(entry -> {
+                assertThat(entry.getValue().size()).isEqualTo(1);
+            });
         }
 
     }
@@ -358,7 +368,6 @@ public class RedactionIntegrationTest {
         manualRedactions.getComments().put("0836727c3508a0b2ea271da69c04cc2f", List.of(comment));
         manualRedactions.getComments().put(manualAddId, List.of(comment));
 
-
         ManualRedactionEntry manualRedactionEntry = new ManualRedactionEntry();
         manualRedactionEntry.setId(manualAddId);
         manualRedactionEntry.setStatus(Status.REQUESTED);