Pull request #121: Add hint annotation by reg-ex and expand to hint annotation by reg-ex

Merge in RED/redaction-service from dismissGen to master

* commit '0f263c69b8f2eb7201f062cc263c8d06b7a710a1':
  Add hint annotation by reg-ex and expand to hint annotation by reg-ex

bamboo-specs/src/main/java/buildjob/PlanSpec.java

@@ -72,14 +72,6 @@ public class PlanSpec {
                         .jobs(new Job("Default Job",
                                 new BambooKey("JOB1"))
                                 .tasks(
-                                        new ScriptTask()
-                                                .description("Clean")
-                                                .inlineBody("#!/bin/bash\n" +
-                                                        "set -e\n" +
-                                                        "rm -rf ./*"),
-                                        new VcsCheckoutTask()
-                                                .description("Checkout Default Repository")
-                                                .checkoutItems(new CheckoutItem().defaultRepository()),
                                         new ScriptTask()
                                                 .description("Clean")
                                                 .inlineBody("#!/bin/bash\n" +

redaction-service-v1/pom.xml

@@ -5,7 +5,7 @@
     <parent>
         <artifactId>platform-dependency</artifactId>
         <groupId>com.iqser.red</groupId>
-        <version>1.0.2</version>
+        <version>1.0.3</version>
     </parent>
     <modelVersion>4.0.0</modelVersion>
 

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

@@ -140,6 +140,48 @@ public class Section {
     }
 
 
+    public void expandToHintAnnotationByRegEx(String type, String pattern, boolean patternCaseInsensitive, int group, String asType) {
+
+        Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
+
+        Set<Entity> expanded = new HashSet<>();
+        for (Entity entity : entities) {
+
+            if (!entity.getType().equals(type) || entity.getTextAfter() == null) {
+                continue;
+            }
+
+            Matcher matcher = compiledPattern.matcher(entity.getTextAfter());
+
+            while (matcher.find()) {
+                String match = matcher.group(group);
+                if (StringUtils.isNotBlank(match)) {
+                    expanded.addAll(findEntities(entity.getWord() + match, asType, false, false, 0, null, null));
+                }
+            }
+        }
+
+        EntitySearchUtils.addEntitiesWithHigherRank(entities, expanded, dictionary);
+        EntitySearchUtils.removeEntitiesContainedInLarger(entities);
+    }
+
+
+    public void addHintAnnotationByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType) {
+
+        Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
+
+        Matcher matcher = compiledPattern.matcher(searchText);
+
+        while (matcher.find()) {
+            String match = matcher.group(group);
+            if (StringUtils.isNotBlank(match)) {
+                Set<Entity> found = findEntities(match.trim(), asType, false, false, 0, null, null);
+                EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
+            }
+        }
+    }
+
+
     public void redactIfPrecededBy(String prefix, String type, int ruleNumber, String reason, String legalBasis) {
 
         entities.forEach(entity -> {

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

@@ -316,9 +316,9 @@ rule "26: Redact On behalf of Syngenta Ltd.:"
 
 // --------------------------------------- other rules -------------------------------------------------------------------
 
-rule "27: Redact Purity"
+rule "27: Purity Hint"
     when
-        Section(searchText.contains("purity"))
+        Section(searchText.toLowerCase().contains("purity"))
     then
-	    section.redactByRegEx("purity ?:? (([\\d\\.]+)( .{0,4}\\.)? ?%)", true, 1, "purity", 27, "Purity found", "Reg (EC) No 1107/2009 Art. 63 (2a)");
+	    section.addHintAnnotationByRegEx("(purity( of)?( ?:)?) [\\d\\.]+( .{0,4}\\.)? ?%", true, 1, "hint_only");
     end

redaction-service-v1/redaction-service-server-v1/src/test/resources/log4j2-test.xml

@@ -10,7 +10,7 @@
         <Root level="warn">
             <AppenderRef ref="CONSOLE" />
         </Root>
-        <Logger name="com.iqser.gin4" level="info" />
+        <Logger name="com.iqser" level="info" />
     </Loggers>
 
 </Configuration>