Pull request #470: RSS-107: Added funktion to redact Headline

Merge in RED/redaction-service from RSS-107 to master

* commit 'd92757cda462d8af9ac852b88c867032c753a3b0':
  RSS-107: Added funktion to redact Headline
This commit is contained in:
Dominique Eiflaender 2022-09-13 12:12:35 +02:00
commit c478935b86

View File

@ -530,52 +530,72 @@ public class Section {
@Deprecated
@ThenAction
@SuppressWarnings("unused")
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason, @Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactBetween(start, stop,false, false, asType, ruleNumber, redactEverywhere, false, reason, legalBasis, true, false, false);
}
@ThenAction
@SuppressWarnings("unused")
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine, @Argument(ArgumentType.STRING) String reason,
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactBetween(start, stop,false,false, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, false, false);
redactBetween(start, stop, false, false, asType, ruleNumber, redactEverywhere, false, reason, legalBasis, true, false, false);
}
@ThenAction
@SuppressWarnings("unused")
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine, @Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis, @Argument(ArgumentType.BOOLEAN) boolean skipRemoveEntitiesContainedInLarger, @Argument(ArgumentType.BOOLEAN) boolean sortedResult) {
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactBetween(start, stop,false, false, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, skipRemoveEntitiesContainedInLarger, sortedResult);
redactBetween(start, stop, false, false, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, false, false);
}
@ThenAction
@SuppressWarnings("unused")
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop,@Argument(ArgumentType.BOOLEAN) boolean includeStart, @Argument(ArgumentType.BOOLEAN) boolean includeStop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine, @Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis, @Argument(ArgumentType.BOOLEAN) boolean skipRemoveEntitiesContainedInLarger, @Argument(ArgumentType.BOOLEAN) boolean sortedResult) {
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis,
@Argument(ArgumentType.BOOLEAN) boolean skipRemoveEntitiesContainedInLarger,
@Argument(ArgumentType.BOOLEAN) boolean sortedResult) {
redactBetween(start, stop,includeStart, includeStop, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, skipRemoveEntitiesContainedInLarger, sortedResult);
redactBetween(start, stop, false, false, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, skipRemoveEntitiesContainedInLarger, sortedResult);
}
@ThenAction
@SuppressWarnings("unused")
public void redactBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.BOOLEAN) boolean includeStart,
@Argument(ArgumentType.BOOLEAN) boolean includeStop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis,
@Argument(ArgumentType.BOOLEAN) boolean skipRemoveEntitiesContainedInLarger,
@Argument(ArgumentType.BOOLEAN) boolean sortedResult) {
redactBetween(start, stop, includeStart, includeStop, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, legalBasis, true, skipRemoveEntitiesContainedInLarger, sortedResult);
}
@Deprecated
@ThenAction
@SuppressWarnings("unused")
public void redactNotBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
public void redactNotBetween(@Argument(ArgumentType.STRING) String start,
@Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason) {
redactBetween(start, stop, asType, ruleNumber, redactEverywhere, false, reason, null, false, false);
@ -584,9 +604,12 @@ public class Section {
@ThenAction
@SuppressWarnings("unused")
public void redactNotBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine, @Argument(ArgumentType.STRING) String reason) {
public void redactNotBetween(@Argument(ArgumentType.STRING) String start,
@Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.BOOLEAN) boolean excludeHeadLine,
@Argument(ArgumentType.STRING) String reason) {
redactBetween(start, stop, asType, ruleNumber, redactEverywhere, excludeHeadLine, reason, null, false, false);
}
@ -594,20 +617,25 @@ public class Section {
@ThenAction
@SuppressWarnings("unused")
public void redactLinesBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason, @Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
public void redactLinesBetween(@Argument(ArgumentType.STRING) String start,
@Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactLinesBetween(start, stop, asType, ruleNumber, redactEverywhere, reason, legalBasis, true);
}
@ThenAction
@SuppressWarnings("unused")
public void redactNotLinesBetween(@Argument(ArgumentType.STRING) String start, @Argument(ArgumentType.STRING) String stop, @Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
public void redactNotLinesBetween(@Argument(ArgumentType.STRING) String start,
@Argument(ArgumentType.STRING) String stop,
@Argument(ArgumentType.TYPE) String asType,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.BOOLEAN) boolean redactEverywhere,
@Argument(ArgumentType.STRING) String reason) {
redactLinesBetween(start, stop, asType, ruleNumber, redactEverywhere, reason, null, false);
@ -750,7 +778,8 @@ public class Section {
@SuppressWarnings("unused")
public void ignoreRecommendations(@Argument(ArgumentType.TYPE) String type) {
entities.removeIf(entity -> entity.getType().equals(type) && entity.getEntityType().equals(EntityType.RECOMMENDATION));
entities.removeIf(entity -> entity.getType().equals(type) && entity.getEntityType()
.equals(EntityType.RECOMMENDATION));
}
@ -855,7 +884,9 @@ public class Section {
@ThenAction
@SuppressWarnings("unused")
public void redactSectionText(@Argument(ArgumentType.TYPE) String type, @Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.STRING) String reason,
public void redactSectionText(@Argument(ArgumentType.TYPE) String type,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactBetween("", "", type, ruleNumber, false, false, reason, legalBasis, true, false);
@ -864,16 +895,34 @@ public class Section {
@ThenAction
@SuppressWarnings("unused")
public void redactSectionTextWithoutHeadLine(@Argument(ArgumentType.TYPE) String type, @Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.STRING) String reason, @Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
public void redactSectionTextWithoutHeadLine(@Argument(ArgumentType.TYPE) String type,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
redactBetween("", "", type, ruleNumber, false, true, reason, legalBasis, true, false);
}
@ThenAction
public void redactHeadline(@Argument(ArgumentType.TYPE) String type,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
if (searchText.contains(headline)) {
Set<Entity> found = findEntities(headline.trim(), type, false, true, ruleNumber, reason, legalBasis, Engine.RULE, false);
EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
}
}
@ThenAction
@SuppressWarnings("unused")
public void redactSection(@Argument(ArgumentType.TYPE) String type, @Argument(ArgumentType.RULE_NUMBER) int ruleNumber, @Argument(ArgumentType.STRING) String reason,
public void redactSection(@Argument(ArgumentType.TYPE) String type,
@Argument(ArgumentType.RULE_NUMBER) int ruleNumber,
@Argument(ArgumentType.STRING) String reason,
@Argument(ArgumentType.LEGAL_BASIS) String legalBasis) {
for (SectionArea sectionArea : sectionAreas) {
@ -906,8 +955,8 @@ public class Section {
}
private void redactAndRecommendByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber, String reason, String legalBasis,
boolean redaction) {
private void redactAndRecommendByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType,
int ruleNumber, String reason, String legalBasis, boolean redaction) {
Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
Matcher matcher = compiledPattern.matcher(searchText);
@ -1037,10 +1086,8 @@ public class Section {
}
public void redactLineAfterAcrossColumns(String start, String asType, int ruleNumber, boolean redactEverywhere, String reason,
String legalBasis) {
public void redactLineAfterAcrossColumns(String start, String asType, int ruleNumber, boolean redactEverywhere,
String reason, String legalBasis) {
String[] values = StringUtils.substringsBetween(searchableText.getAsStringWithLinebreaksSorted(), start, "\n");
@ -1059,7 +1106,6 @@ public class Section {
}
private void redactByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType, int ruleNumber,
String reason, String legalBasis, boolean redaction) {
@ -1077,8 +1123,10 @@ public class Section {
}
private void redactBetween(String start, String stop,boolean includeStart, boolean includeStop, String asType, int ruleNumber, boolean redactEverywhere, boolean excludeHeadLine, String reason, String legalBasis,
boolean redaction, boolean skipRemoveEntitiesContainedInLarger, boolean sortedResult) {
private void redactBetween(String start, String stop, boolean includeStart, boolean includeStop, String asType,
int ruleNumber, boolean redactEverywhere, boolean excludeHeadLine, String reason,
String legalBasis, boolean redaction, boolean skipRemoveEntitiesContainedInLarger,
boolean sortedResult) {
String[] values = new String[1];
@ -1106,15 +1154,19 @@ public class Section {
searchString = start + searchString;
}
if(!stop.isEmpty() && includeStop){
if (!stop.isEmpty() && includeStop) {
searchString = searchString + stop;
}
Set<Entity> found = findEntities(searchString.trim(), asType, false, redaction, ruleNumber, reason, legalBasis, Engine.RULE, false);
found.forEach(f -> {
f.setSkipRemoveEntitiesContainedInLarger(skipRemoveEntitiesContainedInLarger);
if(sortedResult){
f.setWord(searchableText.getAsStringWithLinebreaksSorted(f.getPositionSequences().stream().map(EntityPositionSequence::getSequences).flatMap(Collection::stream).collect(Collectors.toList())));
if (sortedResult) {
f.setWord(searchableText.getAsStringWithLinebreaksSorted(f.getPositionSequences()
.stream()
.map(EntityPositionSequence::getSequences)
.flatMap(Collection::stream)
.collect(Collectors.toList())));
}
});