RSS-42: Made redactBetween rule more generic

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/Section.java

@@ -998,7 +998,17 @@ public class Section {
     private void redactBetween(String start, String stop, String asType, int ruleNumber, boolean redactEverywhere,
                                String reason, String legalBasis, boolean redaction) {
 
-        String[] values = StringUtils.substringsBetween(searchText, start, stop);
+        String[] values = new String[1];
+
+        if(start.isEmpty() && stop.isEmpty()){
+            values[0] = searchText;
+        } else if(start.isEmpty() && searchText.contains(stop)){
+            values[0] = StringUtils.substringBefore(searchText, stop);
+        } else if (stop.isEmpty() && searchText.contains(start)){
+            values[0] = StringUtils.substringAfter(searchText, start);
+        } else {
+            values = StringUtils.substringsBetween(searchText, start, stop);
+        }
 
         if (values != null) {
             for (String value : values) {

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

@@ -330,6 +330,36 @@ public class RedactionIntegrationTest {
     }
 
 
+
+    @Test
+    public void titleExtraction() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/RSS/32 - Emamectin Benzoate Technical - Acute Oral Toxicity - Mouse.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 text = redactionStorageService.getText(TEST_DOSSIER_ID, TEST_FILE_ID);
+
+        AnnotateResponse annotateResponse = annotationService.annotate(AnnotateRequest.builder()
+                .dossierId(TEST_DOSSIER_ID)
+                .fileId(TEST_FILE_ID)
+                .build());
+
+        String outputFileName = OsUtils.getTemporaryDirectory() + "/Annotated.pdf";
+
+        try (FileOutputStream fileOutputStream = new FileOutputStream(outputFileName)) {
+            fileOutputStream.write(annotateResponse.getDocument());
+        }
+
+    }
+
+
+
+
     @Test
     @Ignore
     @SneakyThrows
@@ -1064,7 +1094,7 @@ public class RedactionIntegrationTest {
     public void classificationTest() throws IOException {
 
         System.out.println("classificationTest");
-        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Single Table.pdf");
+        ClassPathResource pdfFileResource = new ClassPathResource("files/RSS/01 - CGA100251 - Acute Oral Toxicity (Up and Down Procedure) - Rat (1).pdf");
 
         AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
 

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

@@ -390,4 +390,21 @@ rule "101: Redact CAS numbers"
         Section(hasTableHeader("Sample #"))
     then
         section.redactCell("Sample #", 8, "PII", true, "Redacted because row is a vertebrate study", "Reg (EC) No 1107/2009 Art. 63 (2g)");
-    end
+    end
+
+rule "102: Extract title"
+    when
+       Section(sectionNumber <= 3 && !text.contains("AUTHOR"))
+    then
+        section.redactBetween("", "Final Report", "PII", 5, false, "Title found", "n-a");
+        section.redactBetween("", "FINAL REPORT", "PII", 5, false, "Title found", "n-a");
+        section.redactBetween("TITLE", "DATA REQUIREMENT", "PII", 5, false, "Title found", "n-a");
+        section.redactBetween("TITLE", "DATA REQUIREMENT", "PII", 5, false, "Title found", "n-a");
+     end
+
+rule "102-1: Extract title"
+    when
+       Section(sectionNumber <= 3 && text.contains("SPL"))
+    then
+        section.redactBetween("Laboratories", "SPL", "PII", 5, false, "Title found", "n-a");
+     end