Always check dictionary rank when overriding annotations

Merge in RED/redaction-service from redactAndRecommendFix2 to master

* commit '7dfaf604e7fb0d09e66a1688b1657006a3e92156':
  Fixed recognize rules and exceptions for redactAndRecommend, fixed wrong order of rules(published information)

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

@@ -28,6 +28,14 @@ public class Dictionary {
     }
 
 
+    public int getDictionaryRank(String type){
+        if(!localAccessMap.containsKey(type)){
+            return 0;
+        }
+        return localAccessMap.get(type).getRank();
+    }
+
+
     public boolean isRecommendation(String type) {
 
         DictionaryModel model = localAccessMap.get(type);

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

@@ -62,6 +62,12 @@ public class Section {
     }
 
 
+    public boolean hasTableHeader(String headerName){
+       String cleanHeaderName = headerName.replaceAll("\n", "").replaceAll(" ", "").replaceAll("-", "");
+       return tabularData != null && tabularData.containsKey(cleanHeaderName);
+    }
+
+
     public boolean matchesType(String type) {
 
         return entities.stream().anyMatch(entity -> entity.getType().equals(type));
@@ -121,7 +127,7 @@ public class Section {
     public void addHintAnnotation(String value, String asType) {
 
         Set<Entity> found = findEntities(value.trim(), asType, true, false, 0, null, null);
-        addNewerToEntities(found);
+        EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
     }
 
 
@@ -134,14 +140,44 @@ public class Section {
             for (String value : values) {
                 if (StringUtils.isNotBlank(value)) {
                     Set<Entity> found = findEntities(value.trim(), asType, false, true, ruleNumber, reason, legalBasis);
-                    addNewerToEntities(found);
+                    EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
+
+                    if (redactEverywhere && !isLocal()) {
+                        localDictionaryAdds.computeIfAbsent(asType, (x) -> new HashSet<>()).add(value.trim());
+                    }
                 }
             }
         }
     }
 
 
-    public void redactByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber, String reason, String legalBasis) {
+    public void recommendLineAfter(String start, String asType) {
+
+        String[] values = StringUtils.substringsBetween(text, start, "\n");
+
+        if (values != null) {
+            for (String value : values) {
+
+                String trimmedValue = value.trim();
+                String cleanValue;
+                if(trimmedValue.startsWith(":")){
+                    cleanValue = trimmedValue.substring(1).trim();
+                } else {
+                    cleanValue = trimmedValue;
+                }
+
+                if (StringUtils.isNotBlank(cleanValue) && cleanValue.length() >= 3) {
+                    localDictionaryAdds.computeIfAbsent(RECOMMENDATION_PREFIX + asType, (x) -> new HashSet<>())
+                            .add(cleanValue);
+                }
+            }
+        }
+    }
+
+
+    public void redactByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber,
+                              String reason, String legalBasis) {
+
         Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
 
         Matcher matcher = compiledPattern.matcher(searchText);
@@ -150,13 +186,14 @@ public class Section {
             String match = matcher.group(group);
             if (StringUtils.isNotBlank(match)) {
                 Set<Entity> found = findEntities(match.trim(), asType, false, true, ruleNumber, reason, legalBasis);
-                addNewerToEntities(found);
+                EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
             }
         }
     }
 
 
     public void addRecommendationByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType) {
+
         Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
 
         Matcher matcher = compiledPattern.matcher(text);
@@ -164,8 +201,24 @@ public class Section {
         while (matcher.find()) {
             String match = matcher.group(group);
             if (StringUtils.isNotBlank(match) && match.length() >= 3) {
-                localDictionaryAdds.computeIfAbsent(RECOMMENDATION_PREFIX + asType, (x) -> new HashSet<>())
-                        .add(match);
+                localDictionaryAdds.computeIfAbsent(RECOMMENDATION_PREFIX + asType, (x) -> new HashSet<>()).add(match);
+            }
+        }
+    }
+
+
+    public void redactAndRecommendByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber,
+                              String reason, String legalBasis) {
+
+        Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
+
+        Matcher matcher = compiledPattern.matcher(searchText);
+
+        while (matcher.find()) {
+            String match = matcher.group(group);
+            if (StringUtils.isNotBlank(match) && match.length() >= 3) {
+                localDictionaryAdds.computeIfAbsent(RECOMMENDATION_PREFIX + asType, (x) -> new HashSet<>()).add(match);
+                localDictionaryAdds.computeIfAbsent(asType, (x) -> new HashSet<>()).add(match);
             }
         }
     }
@@ -181,7 +234,7 @@ public class Section {
                 if (StringUtils.isNotBlank(value)) {
 
                     Set<Entity> found = findEntities(value.trim(), asType, false, true, ruleNumber, reason, legalBasis);
-                    addNewerToEntities(found);
+                    EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
 
                     if (redactEverywhere && !isLocal()) {
                         localDictionaryAdds.computeIfAbsent(asType, (x) -> new HashSet<>()).add(value.trim());
@@ -192,8 +245,7 @@ public class Section {
     }
 
 
-    public void redactLinesBetween(String start, String stop, String asType, int ruleNumber,
-                                   boolean redactEverywhere,
+    public void redactLinesBetween(String start, String stop, String asType, int ruleNumber, boolean redactEverywhere,
                                    String reason, String legalBasis) {
 
         String[] values = StringUtils.substringsBetween(text, start, stop);
@@ -210,7 +262,7 @@ public class Section {
                         }
 
                         Set<Entity> found = findEntities(line.trim(), asType, false, true, ruleNumber, reason, legalBasis);
-                        addNewerToEntities(found);
+                        EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
 
                         if (redactEverywhere && !isLocal()) {
                             localDictionaryAdds.computeIfAbsent(asType, (x) -> new HashSet<>()).add(line.trim());
@@ -228,8 +280,7 @@ public class Section {
     }
 
 
-    public void redactCell(String cellHeader, int ruleNumber, String type, boolean addAsRecommendations, String
-            reason,
+    public void redactCell(String cellHeader, int ruleNumber, String type, boolean addAsRecommendations, String reason,
                            String legalBasis) {
 
         annotateCell(cellHeader, ruleNumber, type, true, addAsRecommendations, reason, legalBasis);
@@ -290,7 +341,7 @@ public class Section {
             singleEntitySet.add(entity);
             EntitySearchUtils.clearAndFindPositions(singleEntitySet, searchableText, dictionary);
 
-            addNewerToEntities(entity);
+            EntitySearchUtils.addEntitiesWithHigherRank(entities, entity, dictionary);
 
             EntitySearchUtils.removeEntitiesContainedInLarger(entities);
 
@@ -312,20 +363,6 @@ public class Section {
             }
         }
     }
-
-
-    private void addNewerToEntities(Set<Entity> found) {
-        // HashSet keeps the older value, but we want the new only.
-        entities.removeAll(found);
-        entities.addAll(found);
-    }
-
-    private void addNewerToEntities(Entity found) {
-        // HashSet keeps the older value, but we want the new only.
-        entities.remove(found);
-        entities.add(found);
-    }
-
 }
 
 

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

@@ -1,11 +1,11 @@
 package com.iqser.red.service.redaction.v1.server.redaction.service;
 
-import com.iqser.red.service.configuration.v1.api.model.RulesResponse;
-import com.iqser.red.service.redaction.v1.server.client.RulesClient;
-import com.iqser.red.service.redaction.v1.server.exception.RulesValidationException;
-import com.iqser.red.service.redaction.v1.server.redaction.model.Section;
-import lombok.Getter;
-import lombok.RequiredArgsConstructor;
+import java.io.ByteArrayInputStream;
+import java.io.InputStream;
+import java.nio.charset.StandardCharsets;
+import java.util.HashMap;
+import java.util.Map;
+
 import org.apache.commons.lang3.StringUtils;
 import org.kie.api.KieServices;
 import org.kie.api.builder.KieBuilder;
@@ -15,11 +15,12 @@ import org.kie.api.runtime.KieContainer;
 import org.kie.api.runtime.KieSession;
 import org.springframework.stereotype.Service;
 
-import java.io.ByteArrayInputStream;
-import java.io.InputStream;
-import java.nio.charset.StandardCharsets;
-import java.util.HashMap;
-import java.util.Map;
+import com.iqser.red.service.configuration.v1.api.model.RulesResponse;
+import com.iqser.red.service.redaction.v1.server.client.RulesClient;
+import com.iqser.red.service.redaction.v1.server.exception.RulesValidationException;
+import com.iqser.red.service.redaction.v1.server.redaction.model.Section;
+
+import lombok.RequiredArgsConstructor;
 
 @Service
 @RequiredArgsConstructor
@@ -29,7 +30,11 @@ public class DroolsExecutionService {
 
     private Map<String, KieContainer> kieContainers = new HashMap<>();
 
+    private Map<String, Long> rulesVersionPerRuleSetId = new HashMap<>();
+
+
     public KieContainer getKieContainer(String ruleSetId) {
+
         KieContainer container = kieContainers.get(ruleSetId);
         if (container == null) {
             return createOrUpdateKieContainer(ruleSetId);
@@ -39,9 +44,6 @@ public class DroolsExecutionService {
     }
 
 
-    @Getter
-    private long rulesVersion = -1;
-
     public Section executeRules(KieContainer kieContainer, Section section) {
 
         KieSession kieSession = kieContainer.newKieSession();
@@ -54,17 +56,24 @@ public class DroolsExecutionService {
 
     }
 
+
     public KieContainer updateRules(String ruleSetId) {
 
         long version = rulesClient.getVersion(ruleSetId);
-        if (version > rulesVersion) {
-            rulesVersion = version;
+        Long rulesVersion = rulesVersionPerRuleSetId.get(ruleSetId);
+        if (rulesVersion == null) {
+            rulesVersion = -1L;
+        }
+
+        if (version > rulesVersion.longValue()) {
+            rulesVersionPerRuleSetId.put(ruleSetId, version);
             return createOrUpdateKieContainer(ruleSetId);
         }
         return getKieContainer(ruleSetId);
 
     }
 
+
     private KieContainer createOrUpdateKieContainer(String ruleSetId) {
 
         try {
@@ -92,7 +101,9 @@ public class DroolsExecutionService {
 
     }
 
+
     private KieModule getKieModule(String ruleSetId, String rules, KieServices kieServices) {
+
         KieFileSystem kieFileSystem = kieServices.newKieFileSystem();
         InputStream input = new ByteArrayInputStream(rules.getBytes(StandardCharsets.UTF_8));
         kieFileSystem.write("src/main/resources/drools/rules" + ruleSetId + ".drl", kieServices.getResources()
@@ -104,10 +115,22 @@ public class DroolsExecutionService {
 
 
     public void testRules(String rules) {
+
         KieServices kieServices = KieServices.Factory.get();
         KieModule kieModule = getKieModule("test-rules", rules, kieServices);
         var container = kieServices.newKieContainer(kieModule.getReleaseId());
         container.newKieSession();
         container.dispose();
     }
+
+
+    public long getRulesVersion(String ruleSetId) {
+
+        Long rulesVersion = rulesVersionPerRuleSetId.get(ruleSetId);
+        if (rulesVersion == null) {
+            return -1;
+        }
+        return rulesVersion.longValue();
+    }
+
 }

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

@@ -50,7 +50,7 @@ public class EntityRedactionService {
 
         dictionaryService.updateDictionary(ruleSetId);
         KieContainer container = droolsExecutionService.updateRules(ruleSetId);
-        long rulesVersion = droolsExecutionService.getRulesVersion();
+        long rulesVersion = droolsExecutionService.getRulesVersion(ruleSetId);
 
         Dictionary dictionary = dictionaryService.getDeepCopyDictionary(ruleSetId);
 
@@ -67,10 +67,7 @@ public class EntityRedactionService {
             });
 
             Set<Entity> foundByLocal = findEntities(classifiedDoc, container, manualRedactions, dictionary, true, hintsPerSectionNumber);
-            // HashSet keeps the older value, but we want the new only.
-            documentEntities.removeAll(foundByLocal);
-            documentEntities.addAll(foundByLocal);
-
+            EntitySearchUtils.addEntitiesWithHigherRank(documentEntities, foundByLocal, dictionary);
             EntitySearchUtils.removeEntitiesContainedInLarger(documentEntities);
         }
 
@@ -159,6 +156,8 @@ public class EntityRedactionService {
                                                                Map<Integer, Set<Entity>> hintsPerSectionNumber) {
 
         List<SectionSearchableTextPair> sectionSearchableTextPairs = new ArrayList<>();
+
+        int rowNumber = 0;
         for (List<Cell> row : table.getRows()) {
             SearchableText searchableRow = new SearchableText();
             Map<String, CellValue> tabularData = new HashMap<>();
@@ -170,15 +169,15 @@ public class EntityRedactionService {
                 }
                 addSectionToManualRedactions(cell.getTextBlocks(), manualRedactions, table.getHeadline(), sectionNumber.intValue());
                 int cellStart = start;
-                cell.getHeaderCells().forEach(headerCell -> {
-                    StringBuilder headerBuilder = new StringBuilder();
-                    headerCell.getTextBlocks().forEach(textBlock -> headerBuilder.append(textBlock.getText()));
-                    String headerName = headerBuilder.toString()
-                            .replaceAll("\n", "")
-                            .replaceAll(" ", "")
-                            .replaceAll("-", "");
-                    tabularData.put(headerName, new CellValue(cell.getTextBlocks(), cellStart));
-                });
+
+                if(rowNumber != 0) {
+                    cell.getHeaderCells().forEach(headerCell -> {
+                        StringBuilder headerBuilder = new StringBuilder();
+                        headerCell.getTextBlocks().forEach(textBlock -> headerBuilder.append(textBlock.getText()));
+                        String headerName = headerBuilder.toString().replaceAll("\n", "").replaceAll(" ", "").replaceAll("-", "");
+                        tabularData.put(headerName, new CellValue(cell.getTextBlocks(), cellStart));
+                    });
+                }
 
                 for (TextBlock textBlock : cell.getTextBlocks()) {
                     // TODO avoid cell overlap merging.
@@ -206,6 +205,7 @@ public class EntityRedactionService {
                     .build(), searchableRow));
 
             sectionNumber.incrementAndGet();
+            rowNumber++;
         }
         return sectionSearchableTextPairs;
     }

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

@@ -104,4 +104,21 @@ public class EntitySearchUtils {
         }
         entities.removeAll(wordsToRemove);
     }
+
+
+    public void addEntitiesWithHigherRank(Set<Entity> entities, Set<Entity> found, Dictionary dictionary) {
+        found.forEach(f -> addEntitiesWithHigherRank(entities, f, dictionary));
+    }
+
+
+    public void addEntitiesWithHigherRank(Set<Entity> entities, Entity found, Dictionary dictionary) {
+
+        if(entities.contains(found)){
+            Entity existing = entities.stream().filter(entity -> entity.equals(found)).findFirst().get();
+            if (dictionary.getDictionaryRank(existing.getType()) <= dictionary.getDictionaryRank(found.getType())){
+                entities.remove(found);
+            }
+        }
+        entities.add(found);
+    }
 }

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

@@ -59,6 +59,16 @@ public class Table extends AbstractTextContainer {
 
         if (rows == null) {
             rows = computeRows();
+
+            // Ignore rows that does not contain any cells and values.
+            List<List<Cell>> rowsToRemove = new ArrayList<>();
+            for (List<Cell> row: rows){
+                if (row.size() == 1 && row.get(0).getTextBlocks().isEmpty()){
+                    rowsToRemove.add(row);
+                }
+            }
+            rows.removeAll(rowsToRemove);
+
             computeHeaders();
         }
 

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

@@ -20,6 +20,7 @@ import com.iqser.red.service.redaction.v1.server.client.RulesClient;
 import com.iqser.red.service.redaction.v1.server.controller.RedactionController;
 import com.iqser.red.service.redaction.v1.server.redaction.utils.ResourceLoader;
 import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities;
+
 import org.apache.commons.io.IOUtils;
 import org.junit.Before;
 import org.junit.Test;
@@ -75,6 +76,7 @@ public class RedactionIntegrationTest {
     private static final String MUST_REDACT = "must_redact";
     private static final String PUBLISHED_INFORMATION = "published_information";
     private static final String TEST_METHOD = "test_method";
+    private static final String PURITY = "purity";
 
     private static final String RECOMMENDATION_AUTHOR = "recommendation_CBI_author";
     private static final String RECOMMENDATION_ADDRESS = "recommendation_CBI_address";
@@ -97,6 +99,7 @@ public class RedactionIntegrationTest {
     private final Map<String, Boolean> hintTypeMap = new HashMap<>();
     private final Map<String, Boolean> caseInSensitiveMap = new HashMap<>();
     private final Map<String, Boolean> recommendationTypeMap = new HashMap<>();
+    private final Map<String, Integer> rankTypeMap = new HashMap<>();
     private final Colors colors = new Colors();
 
     private final static String TEST_RULESET_ID = "123";
@@ -132,7 +135,9 @@ public class RedactionIntegrationTest {
         loadDictionaryForTest();
         loadTypeForTest();
         when(dictionaryClient.getVersion(TEST_RULESET_ID)).thenReturn(0L);
-        when(dictionaryClient.getAllTypes(TEST_RULESET_ID)).thenReturn(TypeResponse.builder().types(getTypeResponse()).build());
+        when(dictionaryClient.getAllTypes(TEST_RULESET_ID)).thenReturn(TypeResponse.builder()
+                .types(getTypeResponse())
+                .build());
         when(dictionaryClient.getDictionaryForType(VERTEBRATE, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(VERTEBRATE));
         when(dictionaryClient.getDictionaryForType(ADDRESS, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(ADDRESS));
         when(dictionaryClient.getDictionaryForType(AUTHOR, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(AUTHOR));
@@ -147,6 +152,7 @@ public class RedactionIntegrationTest {
         when(dictionaryClient.getDictionaryForType(RECOMMENDATION_AUTHOR, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(RECOMMENDATION_AUTHOR));
         when(dictionaryClient.getDictionaryForType(RECOMMENDATION_ADDRESS, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(RECOMMENDATION_ADDRESS));
         when(dictionaryClient.getDictionaryForType(FALSE_POSITIVE, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(FALSE_POSITIVE));
+        when(dictionaryClient.getDictionaryForType(PURITY, TEST_RULESET_ID)).thenReturn(getDictionaryResponse(PURITY));
         when(dictionaryClient.getColors(TEST_RULESET_ID)).thenReturn(colors);
     }
 
@@ -223,6 +229,11 @@ public class RedactionIntegrationTest {
                         .stream()
                         .map(this::cleanDictionaryEntry)
                         .collect(Collectors.toSet()));
+        dictionary.computeIfAbsent(PURITY, v -> new ArrayList<>())
+                .addAll(ResourceLoader.load("dictionaries/purity.txt")
+                        .stream()
+                        .map(this::cleanDictionaryEntry)
+                        .collect(Collectors.toSet()));
     }
 
 
@@ -248,6 +259,7 @@ public class RedactionIntegrationTest {
         typeColorMap.put(RECOMMENDATION_AUTHOR, "#8df06c");
         typeColorMap.put(RECOMMENDATION_ADDRESS, "#8df06c");
         typeColorMap.put(FALSE_POSITIVE, "#ffffff");
+        typeColorMap.put(PURITY, "#ffe187");
 
         hintTypeMap.put(VERTEBRATE, true);
         hintTypeMap.put(ADDRESS, false);
@@ -263,6 +275,7 @@ public class RedactionIntegrationTest {
         hintTypeMap.put(RECOMMENDATION_AUTHOR, false);
         hintTypeMap.put(RECOMMENDATION_ADDRESS, false);
         hintTypeMap.put(FALSE_POSITIVE, true);
+        hintTypeMap.put(PURITY, false);
 
         caseInSensitiveMap.put(VERTEBRATE, true);
         caseInSensitiveMap.put(ADDRESS, false);
@@ -278,6 +291,7 @@ public class RedactionIntegrationTest {
         caseInSensitiveMap.put(RECOMMENDATION_AUTHOR, false);
         caseInSensitiveMap.put(RECOMMENDATION_ADDRESS, false);
         caseInSensitiveMap.put(FALSE_POSITIVE, false);
+        caseInSensitiveMap.put(PURITY, false);
 
         recommendationTypeMap.put(VERTEBRATE, false);
         recommendationTypeMap.put(ADDRESS, false);
@@ -293,6 +307,23 @@ public class RedactionIntegrationTest {
         recommendationTypeMap.put(RECOMMENDATION_AUTHOR, true);
         recommendationTypeMap.put(RECOMMENDATION_ADDRESS, true);
         recommendationTypeMap.put(FALSE_POSITIVE, false);
+        recommendationTypeMap.put(PURITY, false);
+
+        rankTypeMap.put(FALSE_POSITIVE, 160);
+        rankTypeMap.put(PURITY, 155);
+        rankTypeMap.put(PII, 150);
+        rankTypeMap.put(ADDRESS, 140);
+        rankTypeMap.put(AUTHOR, 130);
+        rankTypeMap.put(SPONSOR, 120);
+        rankTypeMap.put(VERTEBRATE, 110);
+        rankTypeMap.put(MUST_REDACT, 100);
+        rankTypeMap.put(REDACTION_INDICATOR, 90);
+        rankTypeMap.put(NO_REDACTION_INDICATOR, 80);
+        rankTypeMap.put(PUBLISHED_INFORMATION, 70);
+        rankTypeMap.put(TEST_METHOD, 60);
+        rankTypeMap.put(HINT_ONLY, 50);
+        rankTypeMap.put(RECOMMENDATION_AUTHOR, 40);
+        rankTypeMap.put(RECOMMENDATION_ADDRESS, 30);
 
         colors.setDefaultColor("#acfc00");
         colors.setNotRedacted("#cccccc");
@@ -312,6 +343,7 @@ public class RedactionIntegrationTest {
                         .isHint(hintTypeMap.get(typeColor.getKey()))
                         .isCaseInsensitive(caseInSensitiveMap.get(typeColor.getKey()))
                         .isRecommendation(recommendationTypeMap.get(typeColor.getKey()))
+                        .rank(rankTypeMap.get(typeColor.getKey()))
                         .build())
 
                 .collect(Collectors.toList());
@@ -326,6 +358,7 @@ public class RedactionIntegrationTest {
                 .isHint(hintTypeMap.get(type))
                 .isCaseInsensitive(caseInSensitiveMap.get(type))
                 .isRecommendation(recommendationTypeMap.get(type))
+                .rank(rankTypeMap.get(type))
                 .build();
     }
 
@@ -390,7 +423,7 @@ public class RedactionIntegrationTest {
 
         System.out.println("redactionTest");
         long start = System.currentTimeMillis();
-        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Applicant Producer Table.pdf");
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Metolachlor/S-Metolachlor_RAR_02_Volume_2_2018-09-06.pdf");
 
         RedactionRequest request = RedactionRequest.builder()
                 .ruleSetId(TEST_RULESET_ID)

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

@@ -16,6 +16,7 @@ import com.iqser.red.service.redaction.v1.server.segmentation.PdfSegmentationSer
 import org.apache.commons.io.IOUtils;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.junit.Before;
+import org.junit.Ignore;
 import org.junit.Test;
 import org.junit.runner.RunWith;
 import org.kie.api.KieServices;
@@ -412,6 +413,7 @@ public class EntityRedactionServiceTest {
 
 
     @Test
+    @Ignore
     public void testNGuideline() throws IOException {
 
         ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Empty Tabular Data.pdf");

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

@@ -1,10 +1,238 @@
-Long-term
+(Parent
+ADME - Bioanalyses, Vergèze,
+Aeration: Gentle
+Andrews mean
+Andrews pKi
+BASF study code
+Billington –
+Bond elute
+Brown Liquid
+Brown Orange
+Brown Solid
+Brown disco
+Brown discoloured
+Brown fat
 Brown liquid
 Brown solid
+Buehler (1980),
+Buehler (1980).
+Buehler 9
+Buehler Test.
+Buehler assay
+Buehler test
+Buehler)
+Buehler-like :
+Calandra lark
+Cary, NC,
+Chambersas for
+Class 4
+Class III’
+Class III’)
+Class method,
+Class2 (effectively
+ClassMethod) Vischim
+Clay =
+Cochran-Armitage Test.
+Cochran-armitage test.
+Dunnett 2
+Dunnett's "t''
+Dunnett's Test
+Dunnett's multiple
+Dunnett's t-test
+Dunnett's test
+Dunnett) **
+Dunnett) Dose
+Dunnett) Food
+Dunnett) Lower
+Dunnett, 1964)
+Dunnett’s Multiple
+Dunnett’s T
+Dunnett’s Test)
+Dunnett’s adjustment
+Dunnett’s multiple
+Dunnett’s post
+Dunnett’s t-test)
+Fine vacuolation
+Fine white
+Finewhite powder
+Fischer 344
+Fischer 344)
+Fischer F344
+Fischer rat
+Fischer) Contaminants:
+Fischer-344 acceptability
+Fischer-344/
+Fischer-344/DuCrj acceptability
+Fischer344 rats.
+Fischer344) :
+Fischerrat embryo
+Fischer’s exact
+Fisher 344
+Fisher Exact Test
+Fisher Exact test
+Fisher exact test
+Fisher's Exact
+Fisher-344 rat
+Fisher344 (F344/
+Fisher344 rat
+Fisher344 rats
+Fisher344 rats.
+Fisher344 rats/
+Fisher344) group
+Fisher’s exact.
+Freeman-Tukey transform
+Gentle aeration
+Grade 2:
+Grade 3)
+Grade 3:
+Green
+Green Alga
+Green as
+Green-, blue-,
+GreenAlga, Selenastrum
+GreenAlga.. Springborn
+Gross External
+Gross Necropsy
+Gross and
+Gross necropsies
+Gross necropsy
+Gross pathology
+Gross pathology:
+Hall Farm
+Hall Farm)
+HallFarm (396
+Ham's F12
+Ham’s F10
 Hand-held
+HarlanWinkelmann, Borchem,
+Hill International,
+Hill Top
+Hill criteria
+Hill, Bracknell,
+HillTop Chamber,
+Hillmodel was
+Hill’s model
+Japanese White
+Jerneplaque assay
+Kent. Blood
+Klimisch score
+Litchfield and Wilcoxon
+Long term
+Long-Term Toxicity
+Long-term
+Longbw/
+Major Macroscopic findings
+Major biliary
+Major defects
+Major deviations
+Major fetal
+Major metabolic
+Major metabolites
+Major route
+Major treatment-related
+Mallard (Anas
+Mallard with
 Manual-Hand held
-Manual-Hand held
-Weight:
+Marsh Frog
+Marsh harrier
+Masonjars (500
+Meier formula)
+Meier survival
+Miles anti-Canine
+Miles anti-dog
+Netherlands, 2011
+Netherlands, published:
+Netherlands. Report
+Penn State
+Porewater:
+Rattray study
+Read across
+Reed Bunting
+Reichert-Jung Cryo
+RifconGmbH. Oxon
+Ritz and
+S.P.-U/
+San Diego,
+Shaw's Farm,
 Sprague
+Strain JapaneseWhite
+Strong morphological
+Weight -
+Weight 2744
+Weight Gain
+Weight On
+Weight Range
+Weight and
 Weight and length
-Aeration: Gentle
+Weight at
+Weight gain
+Weight loss
+Weight of
+Weight only
+Weight range
+Weight, age
+Weight, feed
+Weight-of-evidence approach
+Weight:
+Weight; Adult
+Weight; Whole
+WeightGain 0-1
+WeightRange:
+Weightat dosing:
+Weightrange 200-250g.
+White (Tif:
+White 10
+White 3
+White Age
+White Powder
+White Strain
+White blood
+White cell
+White crystalline
+White guideline
+White guinea-pigs
+White hygroscopic
+White powder
+White solid
+White stork
+White swan
+White to off-white
+White wagtail
+White, solid
+WhitePowder Lot/
+Whiteguideline :
+Whitehygroscopic solid
+Whitepowder Lot/
+Whitepowder Stability
+Whitesolid Batch
+Whiteswan)
+Whiteto off-white
+Wildlife International specifications
+Wildlife Internationalspecifications Water:
+Williams E
+Williams medium
+Williams' medium
+WilliamsE medium
+Williams’ test
+Wilson's method.
+Wilson. All foetal
+Wilson.All foetal
+Wood pigeons
+York.
+Zeneca dossier);
+Zenecadossier); see
+green algae
+CTL
+No details reported
+Not given
+Not indicated
+Not reported
+Not reported.
+Not stated
+Syngenta
+Syngenta AG
+N/A
+No details reported
+Not available
+Test facility
+TBD

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

@@ -0,0 +1,315 @@
+package drools
+
+import com.iqser.red.service.redaction.v1.server.redaction.model.Section
+
+global Section section
+
+
+// --------------------------------------- CBI rules -------------------------------------------------------------------
+
+rule "1: Redact CBI Authors"
+    when
+        Section(matchesType("CBI_author") || matchesType("recommendation_CBI_author"))
+    then
+        section.redact("CBI_author", 1, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "2: Redacted because Section contains Vertebrate"
+    when
+        Section(matchesType("vertebrate"))
+    then
+        section.redact("CBI_address", 2, "Vertebrate found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "3: Not Redacted because Section contains no Vertebrate"
+    when
+        Section(!matchesType("vertebrate"))
+    then
+        section.redactNot("CBI_address", 3, "No Vertebrate found");
+    end
+
+
+rule "4: Do not redact Names and Addresses if no redaction Indicator is contained"
+    when
+        Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"))
+    then
+        section.redactNot("CBI_address", 4, "Vertebrate and No Redaction Indicator found");
+    end
+
+
+rule "5: Redact Names and Addresses if no_redaction_indicator and redaction_indicator is contained"
+    when
+        Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"), matchesType("redaction_indicator"))
+    then
+        section.redact("CBI_address", 5, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+rule "6: Redact Authors and Addresses in Reference Table if it is a Vertebrate study"
+    when
+        Section(hasTableHeader("Author(s)") && !hasTableHeader("Vertebrate study Y/N"))
+    then
+        section.redactCell("Author(s)", 6, "CBI_author", false, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "7: Redact Authors and Addresses in Reference Table if it is a Vertebrate study"
+    when
+        Section(hasTableHeader("Author") && !hasTableHeader("Vertebrate study Y/N"))
+    then
+        section.redactCell("Author", 7, "CBI_author", false, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "8: Not redacted because Vertebrate Study = N"
+    when
+        Section(rowEquals("Vertebrate study Y/N", "N") || rowEquals("Vertebrate study Y/N", "No"))
+    then
+        section.redactCell("Author(s)", 8, "CBI_author", true, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.redactNot("CBI_address", 8, "Not redacted because row is not a vertebrate study");
+        section.highlightCell("Vertebrate study Y/N", 8, "hint_only");
+    end
+
+
+rule "9: Redact if must redact entry is found"
+    when
+        Section(matchesType("must_redact"))
+    then
+        section.redact("CBI_address", 9, "must_redact entry was found.", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "10: Redact Authors and Addresses in Reference Table if it is a Vertebrate study"
+    when
+        Section(rowEquals("Vertebrate study Y/N", "Y") || rowEquals("Vertebrate study Y/N", "Yes"))
+    then
+        section.redactCell("Author(s)", 10, "CBI_author", true, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.redact("CBI_address", 10, "Redacted because row is a vertebrate study", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.highlightCell("Vertebrate study Y/N", 10, "must_redact");
+    end
+
+
+rule "11: Redact sponsor company"
+    when
+        Section(searchText.toLowerCase().contains("batches produced at"))
+    then
+        section.redactIfPrecededBy("batches produced at", "CBI_sponsor", 11, "Redacted because it represents a sponsor company", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.addHintAnnotation("batches produced at", "must_redact");
+    end
+
+
+rule "12: Redact determination of residues"
+    when
+        Section((
+            searchText.toLowerCase.contains("determination of residues") ||
+            searchText.toLowerCase.contains("determination of total residues")
+        ) && (
+            searchText.toLowerCase.contains("livestock") ||
+            searchText.toLowerCase.contains("live stock") ||
+            searchText.toLowerCase.contains("tissue") ||
+            searchText.toLowerCase.contains("tissues") ||
+            searchText.toLowerCase.contains("liver") ||
+            searchText.toLowerCase.contains("muscle") ||
+            searchText.toLowerCase.contains("bovine") ||
+            searchText.toLowerCase.contains("ruminant") ||
+            searchText.toLowerCase.contains("ruminants")
+        ))
+    then
+        section.redact("CBI_address", 12, "Determination of residues was found.", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.addHintAnnotation("determination of residues", "must_redact");
+        section.addHintAnnotation("determination of total residues", "must_redact");
+        section.addHintAnnotation("livestock", "must_redact");
+        section.addHintAnnotation("live stock", "must_redact");
+        section.addHintAnnotation("tissue", "must_redact");
+        section.addHintAnnotation("tissues", "must_redact");
+        section.addHintAnnotation("liver", "must_redact");
+        section.addHintAnnotation("muscle", "must_redact");
+        section.addHintAnnotation("bovine", "must_redact");
+        section.addHintAnnotation("ruminant", "must_redact");
+        section.addHintAnnotation("ruminants", "must_redact");
+    end
+
+
+rule "13: Redact if CTL/* or BL/*  was found"
+    when
+        Section(searchText.contains("CTL/") || searchText.contains("BL/"))
+    then
+        section.redact("CBI_address", 13, "Laboraty for vertebrate studies found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.addHintAnnotation("CTL", "must_redact");
+        section.addHintAnnotation("BL", "must_redact");
+    end
+
+
+rule "14: Redact and add recommendation for et al. author"
+    when
+        Section(searchText.contains("et al"))
+    then
+		section.redactAndRecommendByRegEx("\\b([A-ZÄÖÜ][^\\s\\.,]+( [A-ZÄÖÜ]{1,2}\\.?)?( ?[A-ZÄÖÜ]\\.?)?) et al\\.?", false, 1, "CBI_author", 14, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "15: Add recommendation for Addresses in Test Organism sections"
+    when
+        Section(searchText.contains("Species:") && searchText.contains("Source:"))
+    then
+		section.recommendLineAfter("Source:", "CBI_address");
+    end
+
+
+rule "16: Add recommendation for Addresses in Test Animals sections"
+    when
+        Section(searchText.contains("Species") && searchText.contains("Source"))
+    then
+		section.recommendLineAfter("Source", "CBI_address");
+    end
+
+
+rule "17: Do not redact Names and Addresses if no redaction Indicator is contained"
+    when
+        Section(matchesType("published_information"))
+    then
+        section.redactNot("CBI_author", 17, "Vertebrate and Published Information found");
+        section.redactNot("CBI_address", 17, "Vertebrate and Published Information found");
+    end
+
+// --------------------------------------- PII rules -------------------------------------------------------------------
+
+
+rule "18: Redacted PII Personal Identification Information"
+    when
+        Section(matchesType("PII"))
+    then
+        section.redact("PII", 18, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "19: Redact Emails by RegEx"
+    when
+        Section(searchText.contains("@"))
+    then
+        section.redactByRegEx("\\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\\.[A-Z]{2,4}\\b", true, 0, "PII", 19, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "20: Redact contact information"
+    when
+       Section(text.contains("Contact point:")
+       || text.contains("Phone:")
+       || text.contains("Fax:")
+       || text.contains("Tel.:")
+       || text.contains("Tel:")
+       || text.contains("E-mail:")
+       || text.contains("Email:")
+       || text.contains("e-mail:")
+       || text.contains("E-mail address:")
+       || text.contains("Alternative contact:")
+       || text.contains("Telephone number:")
+       || text.contains("Telephone No:")
+       || text.contains("Fax number:")
+       || text.contains("Telephone:")
+       || text.contains("European contact:"))
+    then
+          section.redactLineAfter("Contact point:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Phone:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Fax:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Tel.:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Tel:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("E-mail:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Email:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("e-mail:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("E-mail address:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Contact:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Alternative contact:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone number:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone No:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Fax number:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactBetween("No:", "Fax", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactBetween("Contact:", "Tel.:", "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("European contact:",  "PII", 20, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+     end
+
+
+rule "21: Redact contact information if applicant is found"
+    when
+        Section(headlineContainsWord("applicant") || text.contains("Applicant") || headlineContainsWord("Primary contact") || headlineContainsWord("Alternative contact") || text.contains("Telephone number:"))
+    then
+        section.redactLineAfter("Contact point:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Phone:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel.:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Email:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("e-mail:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail address:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Alternative contact:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone number:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone No:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax number:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("No:", "Fax", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("Contact:", "Tel.:", "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("European contact:",  "PII", 21, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "22: Redact contact information if Producer is found"
+    when
+        Section(text.toLowerCase().contains("producer of the plant protection") || text.toLowerCase().contains("producer of the active substance") || text.contains("Manufacturer of the active substance") || text.contains("Manufacturer:") || text.contains("Producer or producers of the active substance"))
+    then
+        section.redactLineAfter("Contact:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Phone:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax number:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone number:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel:", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("No:", "Fax", "PII", 22, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "23: Redact AUTHOR(S)"
+    when
+        Section(searchText.contains("AUTHOR(S):"))
+    then
+        section.redactLinesBetween("AUTHOR(S):", "COMPLETION DATE:", "PII", 23, true, "AUTHOR(S) was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "24: Redact PERFORMING LABORATORY"
+    when
+        Section(searchText.contains("PERFORMING LABORATORY:"))
+    then
+        section.redactBetween("PERFORMING LABORATORY:", "LABORATORY PROJECT ID:", "PII", 24, true, "PERFORMING LABORATORY was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "25: Redact On behalf of Sequani Ltd.:"
+    when
+        Section(searchText.contains("On behalf of Sequani Ltd.: Name Title"))
+    then
+        section.redactBetween("On behalf of Sequani Ltd.: Name Title", "On behalf of", "PII", 25, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+rule "26: Redact On behalf of Syngenta Ltd.:"
+    when
+        Section(searchText.contains("On behalf of Syngenta Ltd.: Name Title"))
+    then
+        section.redactBetween("On behalf of Syngenta Ltd.: Name Title", "Study dates", "PII", 26, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+
+// --------------------------------------- other rules -------------------------------------------------------------------
+
+rule "27: Redact Purity"
+    when
+        Section(searchText.contains("purity"))
+    then
+	    section.redactByRegEx("purity ?:? (([\\d\\.]+)( .{0,4}\\.)? ?%)", true, 1, "purity", 27, "Purity found", "Reg (EC) No 1107/2009 Art. 63 (2a)");
+    end

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

@@ -34,21 +34,21 @@ rule "3: Do not redact Names and Addresses if no redaction Indicator is containe
     end
 
 
-rule "4: Do not redact Names and Addresses if no redaction Indicator is contained"
-    when
-        Section(matchesType("vertebrate"), matchesType("published_information"))
-    then
-        section.redactNot("CBI_author", 4, "Vertebrate and Published Information found");
-        section.redactNot("CBI_address", 4, "Vertebrate and Published Information found");
-    end
-
-
-rule "5: Redact Names and Addresses if no_redaction_indicator and redaction_indicator is contained"
+rule "4: Redact Names and Addresses if no_redaction_indicator and redaction_indicator is contained"
     when
         Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"), matchesType("redaction_indicator"))
     then
-        section.redact("CBI_author", 5, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
-        section.redact("CBI_address", 5, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.redact("CBI_author", 4, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+        section.redact("CBI_address", 4, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "5: Do not redact Names and Addresses if no redaction Indicator is contained"
+    when
+        Section(matchesType("vertebrate"), matchesType("published_information"))
+    then
+        section.redactNot("CBI_author", 5, "Vertebrate and Published Information found");
+        section.redactNot("CBI_address", 5, "Vertebrate and Published Information found");
     end
 
 
@@ -134,33 +134,49 @@ rule "11: Redact if CTL/* or BL/*  was found"
     end
 
 
-rule "12: Add recommendation for et al. author"
+rule "12: Redact and add recommendation for et al. author"
     when
-        Section(searchText.contains("et al."))
+        Section(searchText.contains("et al"))
     then
-		section.addRecommendationByRegEx("\\b([A-ZÄÖÜ][^\\s\\.,]+( [A-ZÄÖÜ]\\.?)?( [A-ZÄÖÜ]\\.?)?) et al\\.?", false, 1, "CBI_author");
+		section.redactAndRecommendByRegEx("\\b([A-ZÄÖÜ][^\\s\\.,]+( [A-ZÄÖÜ]{1,2}\\.?)?( ?[A-ZÄÖÜ]\\.?)?) et al\\.?", false, 1, "CBI_author", 12, "Author found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
+    end
+
+
+rule "13: Add recommendation for Addresses in Test Organism sections"
+    when
+        Section(searchText.contains("Species:") && searchText.contains("Source:"))
+    then
+		section.recommendLineAfter("Source:", "CBI_address");
+    end
+
+
+rule "14: Add recommendation for Addresses in Test Animals sections"
+    when
+        Section(searchText.contains("Species") && searchText.contains("Source"))
+    then
+		section.recommendLineAfter("Source", "CBI_address");
     end
 
 // --------------------------------------- PII rules -------------------------------------------------------------------
 
 
-rule "13: Redacted PII Personal Identification Information"
+rule "14: Redacted PII Personal Identification Information"
     when
         Section(matchesType("PII"))
     then
-        section.redact("PII", 13, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redact("PII", 14, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "14: Redact Emails by RegEx"
+rule "15: Redact Emails by RegEx"
     when
         Section(searchText.contains("@"))
     then
-        section.redactByRegEx("\\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\\.[A-Z]{2,4}\\b", true, 0, "PII", 14, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactByRegEx("\\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\\.[A-Z]{2,4}\\b", true, 0, "PII", 15, "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "15: Redact contact information"
+rule "16: Redact contact information"
     when
        Section(text.contains("Contact point:")
        || text.contains("Phone:")
@@ -178,96 +194,105 @@ rule "15: Redact contact information"
        || text.contains("Telephone:")
        || text.contains("European contact:"))
     then
-          section.redactLineAfter("Contact point:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Phone:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Fax:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Tel.:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Tel:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("E-mail:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Email:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("e-mail:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("E-mail address:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Contact:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Alternative contact:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Telephone number:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Telephone No:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Fax number:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("Telephone:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactBetween("No:", "Fax", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactBetween("Contact:", "Tel.:", "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-              section.redactLineAfter("European contact:",  "PII", 15, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+          section.redactLineAfter("Contact point:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Phone:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Fax:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Tel.:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Tel:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("E-mail:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Email:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("e-mail:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("E-mail address:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Contact:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Alternative contact:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone number:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone No:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Fax number:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("Telephone:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactBetween("No:", "Fax", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactBetween("Contact:", "Tel.:", "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+              section.redactLineAfter("European contact:",  "PII", 16, true, "Contact information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
      end
 
 
-rule "16: Redact contact information if applicant is found"
+rule "17: Redact contact information if applicant is found"
     when
         Section(headlineContainsWord("applicant") || text.contains("Applicant") || headlineContainsWord("Primary contact") || headlineContainsWord("Alternative contact") || text.contains("Telephone number:"))
     then
-        section.redactLineAfter("Contact point:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Phone:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Fax:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Tel.:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Tel:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("E-mail:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Email:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("e-mail:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("E-mail address:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Contact:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Alternative contact:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Telephone number:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Telephone No:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Fax number:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Telephone:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactBetween("No:", "Fax", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactBetween("Contact:", "Tel.:", "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("European contact:",  "PII", 16, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact point:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Phone:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel.:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Email:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("e-mail:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail address:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Alternative contact:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone number:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone No:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax number:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("No:", "Fax", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("Contact:", "Tel.:", "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("European contact:",  "PII", 17, true, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "17: Redact contact information if Producer is found"
+rule "18: Redact contact information if Producer is found"
     when
         Section(text.toLowerCase().contains("producer of the plant protection") || text.toLowerCase().contains("producer of the active substance") || text.contains("Manufacturer of the active substance") || text.contains("Manufacturer:") || text.contains("Producer or producers of the active substance"))
     then
-        section.redactLineAfter("Contact:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Telephone:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Phone:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Fax:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("E-mail:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Contact:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Fax number:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Telephone number:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactLineAfter("Tel:", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
-        section.redactBetween("No:", "Fax", "PII", 17, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Phone:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("E-mail:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Contact:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Fax number:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Telephone number:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLineAfter("Tel:", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("No:", "Fax", "PII", 18, true, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "18: Redact AUTHOR(S)"
+rule "19: Redact AUTHOR(S)"
     when
         Section(searchText.contains("AUTHOR(S):"))
     then
-        section.redactLinesBetween("AUTHOR(S):", "COMPLETION DATE:", "PII", 18, true, "AUTHOR(S) was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactLinesBetween("AUTHOR(S):", "COMPLETION DATE:", "PII", 19, true, "AUTHOR(S) was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "19: Redact PERFORMING LABORATORY"
+rule "20: Redact PERFORMING LABORATORY"
     when
         Section(searchText.contains("PERFORMING LABORATORY:"))
     then
-        section.redactBetween("PERFORMING LABORATORY:", "LABORATORY PROJECT ID:", "PII", 19, true, "PERFORMING LABORATORY was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("PERFORMING LABORATORY:", "LABORATORY PROJECT ID:", "PII", 20, true, "PERFORMING LABORATORY was found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "20: Redact On behalf of Sequani Ltd.:"
+rule "21: Redact On behalf of Sequani Ltd.:"
     when
         Section(searchText.contains("On behalf of Sequani Ltd.: Name Title"))
     then
-        section.redactBetween("On behalf of Sequani Ltd.: Name Title", "On behalf of", "PII", 20, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("On behalf of Sequani Ltd.: Name Title", "On behalf of", "PII", 21, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
     end
 
 
-rule "21: Redact On behalf of Syngenta Ltd.:"
+rule "22: Redact On behalf of Syngenta Ltd.:"
     when
         Section(searchText.contains("On behalf of Syngenta Ltd.: Name Title"))
     then
-        section.redactBetween("On behalf of Syngenta Ltd.: Name Title", "Study dates", "PII", 21, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+        section.redactBetween("On behalf of Syngenta Ltd.: Name Title", "Study dates", "PII", 22, false , "PII (Personal Identification Information) found", "Reg (EC) No 1107/2009 Art. 63 (2e)");
+    end
+
+// --------------------------------------- other rules -------------------------------------------------------------------
+
+rule "25: Redact Purity"
+    when
+        Section(searchText.contains("purity"))
+    then
+	    section.redactByRegEx("purity ?:? (([\\d\\.]+)( .{0,4}\\.)? ?%)", true, 1, "purity", 17, "Purity found", "Reg (EC) No 1107/2009 Art. 63 (2a)");
     end