Compare commits

..
Author SHA1 Message Date
Andrei Isvoran 97291ffcae test publish stuff 2024-03-04 09:55:52 +02:00
Kevin Tumma 9a9867e739 Update .gitlab-ci.yml 2024-03-01 14:13:31 +01:00
Andrei Isvoran aa20a5c201 disable test 2024-03-01 14:56:14 +02:00
Andrei Isvoran 600394c366 change gitlab-ci 2024-03-01 14:51:59 +02:00
Andrei Isvoran 8fd9855b2c add public to it 2024-03-01 12:11:53 +02:00
Andrei Isvoran 2a8f1ecdd2 add public to it 2024-03-01 12:05:11 +02:00
Andrei Isvoran 6202e95ad8 change to pages 2024-03-01 11:59:26 +02:00
Andrei Isvoran b1fee50718 change to pages 2024-03-01 11:53:27 +02:00
Andrei Isvoran 3a59bff246 add publish keyword 2024-03-01 11:49:19 +02:00
Andrei Isvoran 57683d49b6 fix path 2024-03-01 11:48:06 +02:00
Andrei Isvoran 0fd01fa1c3 add dependsOn(delombok) 2024-03-01 11:44:37 +02:00
Andrei Isvoran 6d66e5dac3 add sleep 2024-03-01 11:24:59 +02:00
Andrei Isvoran 446c3e56a8 draft 2024-03-01 11:23:45 +02:00
Andrei Isvoran daab1f1b0a draft 2024-03-01 10:10:24 +02:00
Andrei Isvoran af70705250 draft 2024-03-01 10:06:55 +02:00
Andrei Isvoran f29f8626e9 draft 2024-03-01 09:59:31 +02:00
Andrei Isvoran 940e6e7b42 draft 2024-03-01 09:52:13 +02:00
Andrei Isvoran c08ad75893 draft 2024-03-01 09:49:55 +02:00
Andrei Isvoran b8d9b101c9 draft 2024-03-01 09:47:28 +02:00
Andrei Isvoran a1ea354225 draft 2024-03-01 09:39:53 +02:00
Andrei Isvoran e9f516580e draft 2024-02-29 17:48:02 +02:00
Andrei Isvoran b006b8c1ab draft 2024-02-29 17:36:31 +02:00
Andrei Isvoran 78912adeb6 draft 2024-02-29 17:33:23 +02:00
Andrei Isvoran 9bbff158c0 draft 2024-02-29 17:28:33 +02:00
Andrei Isvoran 6e0b343de3 draft 2024-02-29 17:21:39 +02:00
Andrei Isvoran 5bc9b60623 draft 2024-02-29 17:16:09 +02:00
Andrei Isvoran e8b32681ce draft 2024-02-29 16:40:59 +02:00
Andrei Isvoran d29b6bf816 draft 2024-02-29 16:39:09 +02:00
Corina Olariu b6ad1b5e91 Merge branch 'RED-8590' into 'master'
RED-8590 - Missing reason for added CBI Address, recategorized LOGO and signature

Closes RED-8590

See merge request redactmanager/redaction-service!293
2024-02-29 13:17:56 +01:00
Corina Olariu 5f961b0f99 RED-8590 - Missing reason for added CBI Address, recategorized LOGO and signature 2024-02-29 13:17:56 +01:00
Andrei Isvoran 89dea4138a Merge branch 'RED-8586' into 'master'
RED-8586 - Fix confidentiality rules

Closes RED-8586

See merge request redactmanager/redaction-service!294
2024-02-29 12:58:23 +01:00
Andrei Isvoran 4fb8b10429 RED-8586 - Fix confidentiality rules 2024-02-29 12:58:23 +01:00
Yannik Hampe f09a4e4d42 Merge branch 'RED-8481-rules' into 'master'
RED-8481: use visual layout parser for signature extraction

Closes RED-8481

See merge request redactmanager/redaction-service!296
2024-02-29 09:19:41 +01:00
Yannik Hampe ab5dc2f46b RED-8481: use visual layout parser for signature extraction 2024-02-29 09:19:41 +01:00
17 changed files with 1763 additions and 45 deletions
+13
View File
@@ -21,3 +21,16 @@ deploy:
- if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
- if: $CI_COMMIT_BRANCH =~ /^release/
- if: $CI_COMMIT_TAG
pages:
stage: build
tags:
- dind
script:
- echo "Generating Javadoc..."
- gradle generateJavaDoc -PjavadocDestinationDir="javadoc"
- mkdir public
- mv redaction-service-v1/redaction-service-server-v1/javadoc/* public/
artifacts:
paths:
- public
@@ -12,7 +12,7 @@ plugins {
description = "redaction-service-server-v1"
val layoutParserVersion = "0.93.0"
val layoutParserVersion = "0.94.0"
val jacksonVersion = "2.15.2"
val droolsVersion = "9.44.0.Final"
val pdfBoxVersion = "3.0.0"
@@ -84,9 +84,6 @@ tasks.test {
maxHeapSize = "1024m"
}
tasks.named<BootBuildImage>("bootBuildImage") {
environment.put("BPE_DELIM_JAVA_TOOL_OPTIONS", " ")
@@ -113,3 +110,40 @@ tasks.named<BootBuildImage>("bootBuildImage") {
tags.set(listOf(dockerTag))
}
}
fun parseDroolsImports(droolsFilePath: String): List<String> {
val imports = mutableListOf<String>()
val importPattern = Regex("^import\\s+(com\\.iqser\\.red\\.service\\.redaction\\.v1\\.[\\w.]+);")
val desiredPrefix = "com.iqser.red.service.redaction.v1"
File(droolsFilePath).forEachLine { line ->
importPattern.find(line)?.let { matchResult ->
val importPath = matchResult.groupValues[1].trim()
if (importPath.startsWith(desiredPrefix)) {
val formattedPath = importPath.replace('.', '/')
imports.add("$formattedPath.java")
}
}
}
return imports
}
val droolsImports = parseDroolsImports("redaction-service-v1/redaction-service-server-v1/src/main/resources/drools/all_rules_documine.drl")
tasks.register("generateJavaDoc", Javadoc::class) {
dependsOn("compileJava")
dependsOn("delombok")
classpath = project.sourceSets["main"].runtimeClasspath
source = fileTree("${buildDir}/generated/sources/delombok/java/main") {
include(droolsImports)
}
destinationDir = file(project.findProperty("javadocDestinationDir")?.toString() ?: "")
options.memberLevel = JavadocMemberLevel.PUBLIC
(options as StandardJavadocDocletOptions).apply {
header = "Redaction Service ${project.version}"
footer = "Redaction Service ${project.version}"
title = "API Documentation for Redaction Service ${project.version}"
}
}
@@ -135,10 +135,14 @@ public class Image implements GenericSemanticNode, IEntity {
Map<Page, Rectangle2D> bboxImage = image.getBBox();
Map<Page, Rectangle2D> bbox = this.getBBox();
Rectangle2D intersection = bboxImage.get(this.page).createIntersection(bbox.get(this.page));
double calculatedIntersection = intersection.getWidth() * intersection.getHeight();
double area = bbox.get(this.page).getWidth() * bbox.get(this.page).getHeight();
return (calculatedIntersection / area) > containmentThreshold;
//image needs to be on the same page
if(bboxImage.get(this.page) != null) {
Rectangle2D intersection = bboxImage.get(this.page).createIntersection(bbox.get(this.page));
double calculatedIntersection = intersection.getWidth() * intersection.getHeight();
double area = bbox.get(this.page).getWidth() * bbox.get(this.page).getHeight();
return (calculatedIntersection / area) > containmentThreshold;
}
return false;
}
@@ -146,7 +150,7 @@ public class Image implements GenericSemanticNode, IEntity {
Map<Page, Rectangle2D> bboxImage = image.getBBox();
Map<Page, Rectangle2D> bbox = this.getBBox();
Rectangle2D intersection = bboxImage.get(this.page).createIntersection(bbox.get(this.page));
Rectangle2D intersection = bbox.get(this.page).createIntersection(bboxImage.get(this.page));
double calculatedIntersection = intersection.getWidth() * intersection.getHeight();
double area = bbox.get(this.page).getWidth() * bbox.get(this.page).getHeight();
return (area / calculatedIntersection) > containmentThreshold;
@@ -26,10 +26,9 @@ import lombok.extern.slf4j.Slf4j;
@Slf4j
@Service
@RequiredArgsConstructor
@FieldDefaults(makeFinal=true, level= AccessLevel.PRIVATE)
@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
public class EntityChangeLogService {
@Timed("redactmanager_computeChanges")
public boolean computeChanges(List<EntityLogEntry> previousEntityLogEntries, List<EntityLogEntry> newEntityLogEntries, ManualRedactions manualRedactions, int analysisNumber) {
@@ -42,7 +41,9 @@ public class EntityChangeLogService {
boolean hasChanges = false;
for (EntityLogEntry entityLogEntry : newEntityLogEntries) {
Optional<EntityLogEntry> optionalPreviousEntity = previousEntityLogEntries.stream().filter(entry -> entry.getId().equals(entityLogEntry.getId())).findAny();
Optional<EntityLogEntry> optionalPreviousEntity = previousEntityLogEntries.stream()
.filter(entry -> entry.getId().equals(entityLogEntry.getId()))
.findAny();
if (optionalPreviousEntity.isEmpty()) {
hasChanges = true;
entityLogEntry.getChanges().add(new Change(analysisNumber, ChangeType.ADDED, now));
@@ -63,15 +64,22 @@ public class EntityChangeLogService {
private void addRemovedEntriesAsRemoved(List<EntityLogEntry> previousEntityLogEntries,
List<EntityLogEntry> newEntityLogEntries,
ManualRedactions manualRedactions,
int analysisNumber,
OffsetDateTime now) {
List<EntityLogEntry> newEntityLogEntries,
ManualRedactions manualRedactions,
int analysisNumber,
OffsetDateTime now) {
Set<String> existingIds = newEntityLogEntries.stream().map(EntityLogEntry::getId).collect(Collectors.toSet());
Set<String> existingIds = newEntityLogEntries.stream()
.map(EntityLogEntry::getId)
.collect(Collectors.toSet());
List<EntityLogEntry> removedEntries = previousEntityLogEntries.stream()
.filter(entry -> !existingIds.contains(entry.getId()))
.collect(Collectors.toList());
List<EntityLogEntry> removedDossierRedaction = removedEntries.stream()
.filter(e -> e.getState() == EntryState.REMOVED && e.getType().equals("dossier_redaction"))
.toList();
previousEntityLogEntries.removeAll(removedDossierRedaction);
removedEntries.removeAll(removedDossierRedaction);
removedEntries.forEach(entry -> entry.getChanges().add(new Change(analysisNumber, ChangeType.REMOVED, now)));
removedEntries.forEach(entry -> entry.setState(EntryState.REMOVED));
removedEntries.forEach(entry -> addManualChangeForDictionaryRemovals(entry, manualRedactions));
@@ -85,13 +93,15 @@ public class EntityChangeLogService {
return;
}
manualRedactions.getIdsToRemove().stream()
manualRedactions.getIdsToRemove()
.stream()
.filter(IdRemoval::isRemoveFromDictionary)//
.filter(removed -> removed.getAnnotationId().equals(entry.getId()))//
.findFirst()//
.ifPresent(idRemove -> entry.getManualChanges().add(ManualChangeFactory.toManualChange(idRemove, false)));
}
private ChangeType calculateChangeType(EntryState state, EntryState previousState) {
if (state.equals(previousState)) {
@@ -125,7 +125,7 @@ public class EntityLogCreatorService {
.stream()
.filter(entry -> (newEntityIds.contains(entry.getId()) || entry.getContainingNodeId().isEmpty() || sectionsToReanalyseIds.contains(entry.getContainingNodeId()
.get(0))))
.toList();
.collect(Collectors.toList());
previousEntityLog.getEntityLogEntry().removeAll(previousEntriesFromReAnalyzedSections);
boolean hasChanges = entityChangeLogService.computeChanges(previousEntriesFromReAnalyzedSections,
@@ -692,13 +692,22 @@ rule "ETC.3.1: Redact logos (vertebrate study)"
// Rule unit: ETC.5
rule "ETC.5.0: Ignore dossier_redaction entries if confidentiality is not 'confidential'"
rule "ETC.5.0: Skip dossier_redaction entries if confidentiality is 'confidential'"
when
FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.skip("ETC.5.0", "Ignore dossier_redaction when confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
end
rule "ETC.5.1: Remove dossier_redaction entries if confidentiality is not 'confidential'"
when
not FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.ignore("ETC.5.0", "Ignore dossier redactions, when not confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
$dossierRedaction.remove("ETC.5.1", "Remove dossier_redaction when not confidential");
retract($dossierRedaction);
end
@@ -848,7 +857,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -1150,13 +1150,22 @@ rule "ETC.4.2: Redact dossier dictionary entries"
// Rule unit: ETC.5
rule "ETC.5.0: Ignore dossier_redaction entries if confidentiality is not 'confidential'"
rule "ETC.5.0: Skip dossier_redaction entries if confidentiality is 'confidential'"
when
FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.skip("ETC.5.0", "Ignore dossier_redaction when confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
end
rule "ETC.5.1: Remove dossier_redaction entries if confidentiality is not 'confidential'"
when
not FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.ignore("ETC.5.0", "Ignore dossier redactions, when not confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
$dossierRedaction.remove("ETC.5.1", "Remove dossier_redaction when not confidential");
retract($dossierRedaction);
end
@@ -1411,7 +1420,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -1278,7 +1278,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -190,7 +190,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -802,13 +802,22 @@ rule "ETC.4.0: Redact dossier dictionary entries"
// Rule unit: ETC.5
rule "ETC.5.0: Ignore dossier_redaction entries if confidentiality is not 'confidential'"
rule "ETC.5.0: Skip dossier_redaction entries if confidentiality is 'confidential'"
when
FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.skip("ETC.5.0", "Ignore dossier_redaction when confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
end
rule "ETC.5.1: Remove dossier_redaction entries if confidentiality is not 'confidential'"
when
not FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.ignore("ETC.5.0", "Ignore dossier redactions, when not confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
$dossierRedaction.remove("ETC.5.1", "Remove dossier_redaction when not confidential");
retract($dossierRedaction);
end
@@ -1001,7 +1010,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -214,7 +214,7 @@ rule "MAN.3.1: Apply entity recategorization of same type"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -340,7 +340,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -240,7 +240,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -449,14 +449,22 @@ rule "ETC.3.1: Redact logos (non vertebrate study)"
// Rule unit: ETC.5
rule "ETC.5.0: Ignore dossier_redaction entries if confidentiality is not 'confidential'"
rule "ETC.5.0: Skip dossier_redaction entries if confidentiality is 'confidential'"
when
FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.skip("ETC.5.0", "Ignore dossier_redaction when confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
end
rule "ETC.5.1: Remove dossier_redaction entries if confidentiality is not 'confidential'"
when
not FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.ignore("ETC.5.0", "Ignore dossier redactions, when not confidential");
update($dossierRedaction);
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
$dossierRedaction.remove("ETC.5.1", "Remove dossier_redaction when not confidential");
retract($dossierRedaction);
end
@@ -1159,13 +1159,22 @@ rule "ETC.4.2: Redact dossier dictionary entries"
// Rule unit: ETC.5
rule "ETC.5.0: Ignore dossier_redaction entries if confidentiality is not 'confidential'"
rule "ETC.5.0: Skip dossier_redaction entries if confidentiality is 'confidential'"
when
FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.skip("ETC.5.0", "Ignore dossier_redaction when confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
end
rule "ETC.5.1: Remove dossier_redaction entries if confidentiality is not 'confidential'"
when
not FileAttribute(label == "Confidentiality", value == "confidential")
$dossierRedaction: TextEntity(type() == "dossier_redaction")
then
$dossierRedaction.ignore("ETC.5.0", "Ignore dossier redactions, when not confidential");
$dossierRedaction.getIntersectingNodes().forEach(node -> update(node));
$dossierRedaction.remove("ETC.5.1", "Remove dossier_redaction when not confidential");
retract($dossierRedaction);
end
@@ -1434,7 +1443,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
@@ -1424,7 +1424,7 @@ rule "MAN.3.2: Apply image recategorization"
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: IEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end