RSS-86: Added new rule function redactLineAfterAcrossColumns

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

@@ -9,7 +9,11 @@ import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizati
 
 import java.util.ArrayList;
 import java.util.Collections;
+import java.util.Comparator;
 import java.util.List;
+import java.util.stream.Collectors;
+
+import lombok.Getter;
 
 public class SearchableText {
 
@@ -17,6 +21,7 @@ public class SearchableText {
     @JsonIgnore
     @JsonAttribute(ignore = true)
     private transient String stringRepresentation;
+    @Getter
     private final List<TextPositionSequence> sequences = new ArrayList<>();
 
 
@@ -218,6 +223,31 @@ public class SearchableText {
     }
 
 
+    public String getAsStringWithLinebreaksSorted() {
+
+
+        var sorted = sequences.stream().sorted(Comparator.comparing(a -> a.getTextPositions().get(0).getXDirAdj())).sorted(Comparator.comparing(a -> a.getTextPositions().get(0).getYDirAdj())).sorted(Comparator.comparing(a -> a.getPage())).collect(Collectors.toList());
+
+        StringBuilder sb = new StringBuilder();
+
+        TextPositionSequence previous = null;
+        for (TextPositionSequence word : sorted) {
+
+            if (previous != null) {
+                if (Math.abs(previous.getRotationAdjustedY() - word.getRotationAdjustedY()) > word.getTextHeight()) {
+                    sb.append('\n');
+                } else {
+                    sb.append(' ');
+                }
+            }
+            sb.append(word.toString());
+            previous = word;
+        }
+        return sb.append("\n").toString();
+    }
+
+
+
     public String getAsStringWithLinebreaks() {
 
         StringBuilder sb = new StringBuilder();

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

@@ -1013,6 +1013,29 @@ public class Section {
     }
 
 
+
+    public void redactLineAfterAcrossColumns(String start, String asType, int ruleNumber, boolean redactEverywhere, String reason,
+                                 String legalBasis) {
+
+
+        String[] values = StringUtils.substringsBetween(searchableText.getAsStringWithLinebreaksSorted(), start, "\n");
+
+        if (values != null) {
+            for (String value : values) {
+                if (StringUtils.isNotBlank(value)) {
+                    Set<Entity> found = findEntities(value.trim(), asType, false, true, ruleNumber, reason, legalBasis, Engine.RULE, false);
+                    EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
+
+                    if (redactEverywhere && !isLocal()) {
+                        localDictionaryAdds.computeIfAbsent(asType, x -> new HashSet<>()).add(value.trim());
+                    }
+                }
+            }
+        }
+    }
+
+
+
     private void redactByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber,
                                String reason, String legalBasis, boolean redaction) {
 

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

@@ -407,4 +407,23 @@ rule "102-1: Extract title"
        Section(sectionNumber <= 3 && text.contains("SPL"))
     then
         section.redactBetween("Laboratories", "SPL", "PII", 5, false, "Title found", "n-a");
-     end
+     end
+
+
+rule "30: Vehicle components"
+
+    when
+
+        Section(text.contains("vehicle"))
+
+    then
+
+        //section.redactLineAfter("Name:","vehicle",30,true,"Vehicle components found.","n-a");
+
+        section.redactLineAfterAcrossColumns("Name:","PII",30,false,"bla","n-a");
+
+        section.redactLineAfter("Vehicle:","vehicle",30,true,"Vehicle keyword found.","n-a");
+
+        section.redactLineAfter("The vehicle was","vehicle",30,true,"Vehicle section found.","n-a");
+
+    end