Pull request #292: RED-3059: Fixed inital expansion overlaps

Merge in RED/redaction-service from RED-3059-port to master

* commit '0e21e5b4d988d43040573a4fc71547829c815ec0':
  RED-3059: Fixed inital expansion overlaps

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

@@ -193,15 +193,35 @@ public class EntitySearchUtils {
 
         Set<Entity> result = new HashSet<>();
         if (existingEntities != null && foundEntities != null) {
-            for (Entity existingEntity : existingEntities) {
+            for (Entity foundEntity : foundEntities) {
-                for (Entity foundEntity : foundEntities) {
+
-                    if (existingEntity.getEnd() < foundEntity.getStart() || foundEntity.getEnd() < existingEntity.getStart()) {
+                if (!overlaps(existingEntities, foundEntity)) {
-                        result.add(foundEntity);
+                    result.add(foundEntity);
-                    }
                 }
+
             }
         }
         return result;
     }
 
+
+    private boolean overlaps(Set<Entity> existingEntities, Entity found) {
+
+        for (Entity existing : existingEntities) {
+
+            if(existing.getStart().equals(found.getStart())){
+                continue;
+            }
+
+            for (int i = existing.getStart(); i <= existing.getEnd(); i++) {
+                for (int j = found.getStart(); j <= found.getEnd(); j++) {
+                    if (i == j) {
+                        return true;
+                    }
+                }
+            }
+        }
+        return false;
+    }
+
 }

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

@@ -611,6 +611,23 @@ public class RedactionIntegrationTest {
     }
 
 
+    @Test
+    public void redactionExpansionOverlap() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/ExpansionTest.pdf");
+        AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
+
+        analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId()));
+        AnalyzeResult result = analyzeService.analyze(request);
+
+        var redactionLog = redactionStorageService.getRedactionLog(TEST_DOSSIER_ID, TEST_FILE_ID);
+
+        var values = redactionLog.getRedactionLogEntry().stream().map(RedactionLogEntry::getValue).collect(Collectors.toList());
+
+        assertThat(values).containsExactlyInAnyOrder("Lastname M.", "Doe", "Doe J.", "M. Mustermann", "Mustermann M.", "F. Lastname");
+    }
+
+
     @Test
     @Ignore
     public void noExceptionShouldBeThrownForAnyFiles() throws IOException {

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

@@ -151,8 +151,7 @@ public class EntitySearchUtilsTest {
         Set<Entity> result = EntitySearchUtils.findNonOverlappingMatchEntities(existingEntities, foundEntities);
 
         // Assert
-        assertThat(result.size()).isEqualTo(1);
+        assertThat(result.size()).isEqualTo(0);
-        assertThat(result).contains(foundEntities2);
 
     }
 
@@ -182,8 +181,7 @@ public class EntitySearchUtilsTest {
         Set<Entity> result = EntitySearchUtils.findNonOverlappingMatchEntities(existingEntities, foundEntities);
 
         // Assert
-        assertThat(result.size()).isEqualTo(1);
+        assertThat(result.size()).isEqualTo(0);
-        assertThat(result).contains(foundEntities2);
 
     }
 

redaction-service-v1/redaction-service-server-v1/src/test/resources/dictionaries/CBI_author.txt

@@ -8616,3 +8616,8 @@ Zoriki Hosomi Rosana
 Zuberer D
 Zubrod J
 Zwicker R.E.
+Doe
+M. Mustermann
+F. Lastname
+Mustermann
+Lastname