Merge branch 'DM-305' into 'master'

DM-305: Improved anyHeadlineContains

Closes DM-305

See merge request redactmanager/redaction-service!41

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/nodes/Section.java

@@ -83,21 +83,13 @@ public class Section implements GenericSemanticNode {
 
     public boolean anyHeadlineContainsString(String value) {
 
-        return streamChildrenOfType(NodeType.HEADLINE)//
-                .map(node -> (Headline) node).anyMatch(h -> h.containsString(value));
+        return streamAllSubNodesOfType(NodeType.HEADLINE).anyMatch(h -> h.containsString(value));
     }
 
 
     public boolean anyHeadlineContainsStringIgnoreCase(String value) {
 
-        return streamChildrenOfType(NodeType.HEADLINE)//
-                .map(node -> (Headline) node).anyMatch(h -> h.containsStringIgnoreCase(value));
-    }
-
-
-    public boolean hasParagraphs() {
-
-        return streamAllSubNodesOfType(NodeType.PARAGRAPH).findFirst().isPresent();
+        return streamAllSubNodesOfType(NodeType.HEADLINE).anyMatch(h -> h.containsStringIgnoreCase(value));
     }
 
 }

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

@@ -45,7 +45,9 @@ public class DocumineFloraTest extends AbstractRedactionIntegrationTest {
     @Disabled
     public void titleExtraction() throws IOException {
 
-        AnalyzeRequest request = uploadFileToStorage("files/Documine/Flora/ProblemDocs/15 - EVIDIS - Toxicidade oral aguda.pdf");
+
+        AnalyzeRequest request = prepareStorage("files/Documine/Flora/ProblemDocs/403-17_Fantom_ToxicidadeInalatoriaAguda.pdf",
+                "files/Documine/Flora/ProblemDocs/d75cd9358f7949552697764428183472.TABLES.json");
 
         System.out.println("Start Full integration test");
         analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId()));

redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/documine_flora.drl

@@ -422,8 +422,8 @@ rule "DOC.10.0: Batch number from CoA"
     when
         $section: Section(
             (
-                getHeadline().containsString("Analytical Report")
-                || getHeadline().containsString("Certificate of Analysis")
+                anyHeadlineContainsString("Analytical Report")
+                || anyHeadlineContainsString("Certificate of Analysis")
                 || containsStringIgnoreCase("certificate of analysis")
             )
             && (
@@ -475,7 +475,7 @@ rule "DOC.10.1: Batch number"
             && containsStringIgnoreCase("batch")
         )
     then
-        Stream.of(entityCreationService.byRegex("Batch ID ([A-Z\\d\\-]{7,14})", "batch_number", EntityType.ENTITY, $section),
+        Stream.of(entityCreationService.byRegex("Batch ID ([A-Z\\d\\-]{7,14})", "batch_number", EntityType.ENTITY, 1, $section),
                     entityCreationService.lineAfterStrings(List.of("Batch Identification",
                       "Batch number:",
                       "Batch reference number:",
@@ -875,8 +875,8 @@ rule "DOC.35.2: Animal Number 429"
      FileAttribute(label == "OECD Number", value == "429")
         $section: Section(
             (
-                getHeadline().containsStringIgnoreCase("animal")
-                || getHeadline().containsStringIgnoreCase("test system")
+                anyHeadlineContainsStringIgnoreCase("animal")
+                || anyHeadlineContainsStringIgnoreCase("test system")
             )
             && !getHeadline().containsString("selection")
             && containsStringIgnoreCase("number of animals")
@@ -1126,22 +1126,6 @@ rule "X.0.0: remove Entity contained by Entity of same type"
     end
 
 
-// Rule unit: X.1
-rule "X.1.0: merge intersecting Entities of same type"
-    salience 64
-    when
-        $first: RedactionEntity($type: type, $entityType: entityType, !resized, !skipRemoveEntitiesContainedInLarger, isActive())
-        $second: RedactionEntity(intersects($first), type == $type, entityType == $entityType, this != $first, !resized, !skipRemoveEntitiesContainedInLarger, isActive())
-    then
-        RedactionEntity mergedEntity = entityCreationService.byEntities(List.of($first, $second), $type, $entityType, document);
-        $first.remove("X.1.0", "merge intersecting Entities of same type");
-        $second.remove("X.1.0", "merge intersecting Entities of same type");
-        retract($first);
-        retract($second);
-        mergedEntity.getIntersectingNodes().forEach(node -> update(node));
-    end
-
-
 //------------------------------------ File attributes rules ------------------------------------
 
 // Rule unit: FA.1