diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/service/document/EntityCreationService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/service/document/EntityCreationService.java index 0e1c21c8..cec16515 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/service/document/EntityCreationService.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/service/document/EntityCreationService.java @@ -208,7 +208,7 @@ public class EntityCreationService { .filter(stopBoundary -> stopBoundary.start() > startTextRange.end()) .min(Comparator.comparingInt(TextRange::start)); if (optionalStopTextRangeWithMinimalDistance.isEmpty()) { - break; + continue; } entityBoundaries.add(new TextRange(startTextRange.end(), optionalStopTextRangeWithMinimalDistance.get().start())); } @@ -223,7 +223,8 @@ public class EntityCreationService { For Example consider the text: "a this is some text. a here is more text b". If "a" is the start string and "b" is the stop string, there are two possibilities. "this is some text. a here is more text" and "here is more text". We only want to keep the latter. */ - return entityTextRanges.stream().filter(boundary -> entityTextRanges.stream().noneMatch(innerBoundary -> innerBoundary != boundary && innerBoundary.containedBy(boundary))) + return entityTextRanges.stream() + .filter(boundary -> entityTextRanges.stream().noneMatch(innerBoundary -> !innerBoundary.equals(boundary) && innerBoundary.containedBy(boundary))) .toList(); } diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/DocumineFloraTest.java b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/DocumineFloraTest.java index e56944fd..d79234ce 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/DocumineFloraTest.java +++ b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/DocumineFloraTest.java @@ -48,7 +48,7 @@ public class DocumineFloraTest extends AbstractRedactionIntegrationTest { // @Disabled public void titleExtraction() throws IOException { - AnalyzeRequest request = uploadFileToStorage("files/Documine/Flora/ProblemDocs/8.SYN524464 FS (A16148F) - Teste de Ames (1).pdf"); + AnalyzeRequest request = uploadFileToStorage("files/Documine/Flora/_000008810-0.2.pdf"); // AnalyzeRequest request = prepareStorage("files/Documine/Flora/ProblemDocs/SOLICITA_VICTRATO-GOLD-II_Item 21_Mutacao_Genica (1).pdf", // "files/Documine/Flora/ProblemDocs/SOLICITA_VICTRATO-GOLD-II_Item 21_Mutacao_Genica (1).TABLES.json"); diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Documine/Flora/_000008810-0.2.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Documine/Flora/_000008810-0.2.pdf new file mode 100644 index 00000000..2a3465ad Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Documine/Flora/_000008810-0.2.pdf differ