RED-2841: INC6207970 Rule for initials expansion should be applied only to dictionary entries without whitespaces

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

@@ -131,11 +131,27 @@ public class Section {
         return StringUtils.containsIgnoreCase(headline, word);
     }
 
+
     @ThenAction
     public void expandByRegEx(@Argument(ArgumentType.TYPE) String type,
                               @Argument(ArgumentType.REGEX) String pattern,
                               @Argument(ArgumentType.BOOLEAN) boolean patternCaseInsensitive,
                               @Argument(ArgumentType.INTEGER) int group) {
+        expandByRegEx(type, pattern, patternCaseInsensitive, group, null);
+    }
+
+    @ThenAction
+    public void expandByRegEx(@Argument(ArgumentType.TYPE) String type,
+                              @Argument(ArgumentType.REGEX) String pattern,
+                              @Argument(ArgumentType.BOOLEAN) boolean patternCaseInsensitive,
+                              @Argument(ArgumentType.INTEGER) int group,
+                              @Argument(ArgumentType.REGEX) String withoutPattern) {
+
+        Pattern compiledWithoutPattern = null;
+
+        if (withoutPattern != null) {
+            compiledWithoutPattern = Patterns.getCompiledPattern(withoutPattern, patternCaseInsensitive);
+        }
 
         Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
 
@@ -146,6 +162,13 @@ public class Section {
                 continue;
             }
 
+            if(withoutPattern != null) {
+                Matcher matcherWithout = compiledWithoutPattern.matcher(entity.getWord());
+                if (matcherWithout.find()) {
+                    continue;
+                }
+            }
+
             Matcher matcher = compiledPattern.matcher(entity.getTextAfter());
 
             while (matcher.find()) {

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

@@ -1304,6 +1304,30 @@ public class RedactionIntegrationTest {
 
     }
 
+    @Test
+    public void testExpandByRegEx() throws IOException {
+
+        System.out.println("expandByRegex");
+        long start = System.currentTimeMillis();
+
+        AnalyzeRequest request = prepareStorage("files/Metolachlor/S-Metolachlor_RAR_02_Volume_2_2018-09-06.pdf");
+        analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId()));
+        AnalyzeResult result = analyzeService.analyze(request);
+
+        AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
+                .dossierId(TEST_DOSSIER_ID)
+                .fileId(TEST_FILE_ID)
+                .build());
+
+        try (FileOutputStream fileOutputStream = new FileOutputStream(getTemporaryDirectory() + "/Annotated.pdf")) {
+            fileOutputStream.write(annotateResponse.getDocument());
+        }
+        long end = System.currentTimeMillis();
+
+        System.out.println("duration: " + (end - start));
+        System.out.println("numberOfPages: " + result.getNumberOfPages());
+    }
+
 
     private static String loadFromClassPath(String path) {
 

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

@@ -7,12 +7,20 @@ global Section section
 
 // --------------------------------------- CBI rules -------------------------------------------------------------------
 
+//rule "0: Expand CBI Authors with firstname initials"
+//    when
+//        Section(matchesType("CBI_author") || matchesType("recommendation_CBI_author"))
+//    then
+//        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+//        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+//    end
+
 rule "0: Expand CBI Authors with firstname initials"
     when
         Section(matchesType("CBI_author") || matchesType("recommendation_CBI_author"))
     then
-        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1, "[^\\s]+");
-        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1, "[^\\s]+");
     end
 
 

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

@@ -7,12 +7,20 @@ global Section section
 
 // --------------------------------------- CBI rules -------------------------------------------------------------------
 
+//rule "0: Expand CBI Authors with firstname initials"
+//    when
+//        Section(matchesType("CBI_author") || matchesType("recommendation_CBI_author"))
+//    then
+//        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+//        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+//    end
+
 rule "0: Expand CBI Authors with firstname initials"
     when
         Section(matchesType("CBI_author") || matchesType("recommendation_CBI_author"))
     then
-        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+        section.expandByRegEx("CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1, "[^\\s]+");
-        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1);
+        section.expandByRegEx("recommendation_CBI_author", "(,? [A-Z]\\.?( ?[A-Z]\\.?)?( ?[A-Z]\\.?)?\\b\\.?)", false, 1, "[^\\s]+");
     end