470 lines
22 KiB
Plaintext

package drools
import static java.lang.String.format;
import static com.iqser.red.service.redaction.v1.server.utils.RedactionSearchUtility.anyMatch;
import static com.iqser.red.service.redaction.v1.server.utils.RedactionSearchUtility.exactMatch;
import java.util.List;
import java.util.LinkedList;
import java.util.Set;
import java.util.stream.Collectors;
import java.util.Collection;
import java.util.stream.Stream;
import java.util.Optional;
import com.iqser.red.service.redaction.v1.server.model.document.*;
import com.iqser.red.service.redaction.v1.server.model.document.TextRange;
import com.iqser.red.service.redaction.v1.server.model.document.entity.*;
import com.iqser.red.service.redaction.v1.server.model.document.entity.EntityType;
import com.iqser.red.service.redaction.v1.server.model.document.entity.MatchedRule;
import com.iqser.red.service.redaction.v1.server.model.document.entity.TextEntity
import com.iqser.red.service.redaction.v1.server.model.document.entity.MatchedRule
import com.iqser.red.service.redaction.v1.server.model.document.nodes.*;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Section;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Table;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.TableCell;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.SemanticNode;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Document;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Paragraph;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Image;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.ImageType;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Page;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Headline;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.SectionIdentifier;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Footer;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Header;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.NodeType;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.*;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.TextBlock;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.TextBlockCollector;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.AtomicTextBlock;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.ConcatenatedTextBlock;
import com.iqser.red.service.redaction.v1.server.model.NerEntities;
import com.iqser.red.service.redaction.v1.server.model.dictionary.Dictionary;
import com.iqser.red.service.redaction.v1.server.model.dictionary.DictionaryModel;
import com.iqser.red.service.redaction.v1.server.service.document.EntityCreationService;
import com.iqser.red.service.redaction.v1.server.service.ManualChangesApplicationService;
import com.iqser.red.service.redaction.v1.server.utils.RedactionSearchUtility;
import com.iqser.red.service.persistence.service.v1.api.shared.model.FileAttribute;
import com.iqser.red.service.persistence.service.v1.api.shared.model.analysislog.entitylog.Engine;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.entitymapped.ManualResizeRedaction;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.entitymapped.IdRemoval;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.entitymapped.ManualForceRedaction;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.entitymapped.ManualRecategorization;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.entitymapped.ManualLegalBasisChange;
import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
global Document document
global EntityCreationService entityCreationService
global ManualChangesApplicationService manualChangesApplicationService
global Dictionary dictionary
//------------------------------------ queries ------------------------------------
query "getFileAttributes"
$fileAttribute: FileAttribute()
end
//------------------------------------ Table extraction rules ------------------------------------
// Rule unit: TAB.0
rule "TAB.0.0: Study Type File Attribute"
when
not FileAttribute(label == "OECD Number", valueEqualsAnyOf("402","403","404","405","425","429","436","438","439","471","487"))
$section: Section(containsAnyString("DATA REQUIREMENT", "TEST GUIDELINE", "MÉTODO(S) DE REFERÊNCIA(S):")
&& containsAnyString("OECD", "EPA", "OPPTS"))
then
RedactionSearchUtility.findTextRangesByRegexIgnoreCase("(?<=OECD)(?:[\\w\\s,\\[\\]\\(\\)\\.]{1,10}|(?:.{5,40}(?:Number |Procedure |Guideline )))(4[\\d]{2})", 1 ,$section.getTextBlock()).stream()
.map(boundary -> $section.getTextBlock().subSequence(boundary).toString())
.map(value -> FileAttribute.builder().label("OECD Number").value(value).build())
.forEach(fileAttribute -> insert(fileAttribute));
RedactionSearchUtility.findTextRangesByRegexIgnoreCase("(?<=OECD).{5,40}Method (4[\\d]{2}).{1,65}(\\d{4})\\)", 1, $section.getTextBlock()).stream()
.map(boundary -> $section.getTextBlock().subSequence(boundary).toString())
.map(value -> FileAttribute.builder().label("OECD Number").value(value).build())
.forEach(fileAttribute -> insert(fileAttribute));
end
rule "TAB.0.1: Guidelines"
when
$section: Section(containsAnyString("DATA REQUIREMENT", "TEST GUIDELINE", "MÉTODO(S) DE REFERÊNCIA(S):") && containsAnyString("OECD", "EPA", "OPPTS"))
then
entityCreationService.byRegex("(?<=OECD)(?:[\\w\\s,\\[\\]\\(\\)\\.]{1,10}|.{5,40}(?:Number |Procedure |Guideline ))(4[\\d]{2})", "oecd_guideline_number", EntityType.ENTITY, 1, $section)
.forEach(guideline -> guideline.apply("TAB.0.1", "OECD Guideline no. found"));
entityCreationService.byRegex("(?<=OECD)(?:[\\w\\s,\\[\\]\\(\\)\\.]{1,10}|.{5,40}(?:Number |Procedure |Guideline ))(4[\\d]{2}),?\\s\\(?(\\d{4})\\)?", "oecd_guideline_year", EntityType.ENTITY, 2, $section)
.forEach(guideline -> guideline.apply("TAB.0.1", "OECD Guideline year found"));
entityCreationService.byRegex("(?<=OECD)[\\w\\s,\\[\\]]{1,10}\\((\\d{4})\\)\\s(4[\\d]{2})", "oecd_guideline_year", EntityType.ENTITY, 1, $section)
.forEach(guideline -> guideline.apply("TAB.0.1", "OECD Guideline year found"));
entityCreationService.byRegex("(?<=OECD).{5,40}Method (4[\\d]{2}).{1,65}(\\d{4})\\)", "oecd_guideline_number", EntityType.ENTITY, 1, $section)
.forEach(guideline -> guideline.apply("TAB.0.1", "OECD Guideline number found"));
entityCreationService.byRegex("(?<=OECD).{5,40}Method (4[\\d]{2}).{1,65}(\\d{4})\\)", "oecd_guideline_year", EntityType.ENTITY, 2, $section)
.forEach(guideline -> guideline.apply("TAB.0.1", "OECD Guideline year found"));
end
// Rule unit: TAB.6
rule "TAB.6.0: Targeted cell extraction (Experimental Stop date)"
when
$section: Section(getHeadline().containsString("Advanced Table Extraction"), containsAllStrings("female", "Female", "Survived", "Group 2"))
$table: Table(hasHeader("Group 2")) from $section.streamChildren().toList()
TableCell(containsWordIgnoreCase("Female"), $row: row) from $table.streamTableCellsWithHeader("Group 2").toList()
TableCell($row == row, containsStringIgnoreCase("Survived")) from $table.streamTableCellsWithHeader("Group 2").toList()
$femaleSurvived: TableCell($row == row) from $table.streamTableCellsWithHeader("Group 2").toList()
then
entityCreationService.bySemanticNode($femaleSurvived, "experiment_female_survived", EntityType.ENTITY)
.ifPresent(entity -> entity.apply("TAB.6.0", "Female in group to experimental start date"));
end
//------------------------------------ Manual changes rules ------------------------------------
// Rule unit: MAN.0
rule "MAN.0.0: Apply manual resize redaction"
salience 128
when
$resizeRedaction: ManualResizeRedaction($id: annotationId, $requestDate: requestDate)
not ManualResizeRedaction(annotationId == $id, requestDate.isBefore($requestDate))
$entityToBeResized: TextEntity(matchesAnnotationId($id))
then
manualChangesApplicationService.resize($entityToBeResized, $resizeRedaction);
retract($resizeRedaction);
update($entityToBeResized);
$entityToBeResized.getIntersectingNodes().forEach(node -> update(node));
end
rule "MAN.0.1: Apply manual resize redaction"
salience 128
when
$resizeRedaction: ManualResizeRedaction($id: annotationId, $requestDate: requestDate)
not ManualResizeRedaction(annotationId == $id, requestDate.isBefore($requestDate))
$imageToBeResized: Image(id == $id)
then
manualChangesApplicationService.resizeImage($imageToBeResized, $resizeRedaction);
retract($resizeRedaction);
update($imageToBeResized);
update($imageToBeResized.getParent());
end
// Rule unit: MAN.1
rule "MAN.1.0: Apply id removals that are valid and not in forced redactions to Entity"
salience 128
when
$idRemoval: IdRemoval($id: annotationId, !removeFromDictionary, !removeFromAllDossiers)
$entityToBeRemoved: TextEntity(matchesAnnotationId($id))
then
$entityToBeRemoved.getManualOverwrite().addChange($idRemoval);
update($entityToBeRemoved);
retract($idRemoval);
$entityToBeRemoved.getIntersectingNodes().forEach(node -> update(node));
end
rule "MAN.1.1: Apply id removals that are valid and not in forced redactions to Image"
salience 128
when
$idRemoval: IdRemoval($id: annotationId)
$imageEntityToBeRemoved: Image($id == id)
then
$imageEntityToBeRemoved.getManualOverwrite().addChange($idRemoval);
update($imageEntityToBeRemoved);
retract($idRemoval);
update($imageEntityToBeRemoved.getParent());
end
// Rule unit: MAN.2
rule "MAN.2.0: Apply force redaction"
salience 128
when
$force: ManualForceRedaction($id: annotationId)
$entityToForce: TextEntity(matchesAnnotationId($id))
then
$entityToForce.getManualOverwrite().addChange($force);
update($entityToForce);
$entityToForce.getIntersectingNodes().forEach(node -> update(node));
retract($force);
end
rule "MAN.2.1: Apply force redaction to images"
salience 128
when
$force: ManualForceRedaction($id: annotationId)
$imageToForce: Image(id == $id)
then
$imageToForce.getManualOverwrite().addChange($force);
update($imageToForce);
update($imageToForce.getParent());
retract($force);
end
// Rule unit: MAN.3
rule "MAN.3.0: Apply entity recategorization"
salience 128
when
$recategorization: ManualRecategorization($id: annotationId, $type: type, $requestDate: requestDate)
not ManualRecategorization($id == annotationId, requestDate.isBefore($requestDate))
$entityToBeRecategorized: TextEntity(matchesAnnotationId($id), type() != $type)
then
$entityToBeRecategorized.getIntersectingNodes().forEach(node -> update(node));
$entityToBeRecategorized.getManualOverwrite().addChange($recategorization);
update($entityToBeRecategorized);
retract($recategorization);
end
rule "MAN.3.1: Apply entity recategorization of same type"
salience 128
when
$recategorization: ManualRecategorization($id: annotationId, $type: type, $requestDate: requestDate)
not ManualRecategorization($id == annotationId, requestDate.isBefore($requestDate))
$entityToBeRecategorized: TextEntity(matchesAnnotationId($id), type() == $type)
then
$entityToBeRecategorized.getManualOverwrite().addChange($recategorization);
retract($recategorization);
end
rule "MAN.3.2: Apply image recategorization"
salience 128
when
$recategorization: ManualRecategorization($id: annotationId, $requestDate: requestDate)
not ManualRecategorization($id == annotationId, requestDate.isBefore($requestDate))
$imageToBeRecategorized: Image($id == id)
then
manualChangesApplicationService.recategorize($imageToBeRecategorized, $recategorization);
update($imageToBeRecategorized);
update($imageToBeRecategorized.getParent());
retract($recategorization);
end
rule "MAN.3.3: Apply recategorization entities by default"
salience 128
when
$entity: TextEntity(getManualOverwrite().getRecategorized().orElse(false), !dictionary.isHint(type()))
then
$entity.apply("MAN.3.3", "Recategorized entities are applied by default.", $entity.legalBasis());
end
// Rule unit: MAN.4
rule "MAN.4.0: Apply legal basis change"
salience 128
when
$legalBasisChange: ManualLegalBasisChange($id: annotationId)
$imageToBeRecategorized: Image($id == id)
then
$imageToBeRecategorized.getManualOverwrite().addChange($legalBasisChange);
update($imageToBeRecategorized)
retract($legalBasisChange)
end
rule "MAN.4.1: Apply legal basis change"
salience 128
when
$legalBasisChange: ManualLegalBasisChange($id: annotationId)
$entityToBeChanged: TextEntity(matchesAnnotationId($id))
then
$entityToBeChanged.getManualOverwrite().addChange($legalBasisChange);
update($entityToBeChanged)
retract($legalBasisChange)
end
//------------------------------------ Entity merging rules ------------------------------------
// Rule unit: X.0
rule "X.0.0: Remove Entity contained by Entity of same type"
salience 65
when
$larger: TextEntity($type: type(), $entityType: entityType, active())
$contained: TextEntity(containedBy($larger), type() == $type, entityType == $entityType, this != $larger, !hasManualChanges())
then
$contained.remove("X.0.0", "remove Entity contained by Entity of same type");
retract($contained);
end
// Rule unit: X.1
rule "X.1.0: Merge intersecting Entities of same type"
salience 64
when
$first: TextEntity($type: type(), $entityType: entityType, !resized(), active())
$second: TextEntity(intersects($first), type() == $type, entityType == $entityType, this != $first, !hasManualChanges(), active())
then
TextEntity mergedEntity = entityCreationService.mergeEntitiesOfSameType(List.of($first, $second), $type, $entityType, document);
$first.remove("X.1.0", "merge intersecting Entities of same type");
$second.remove("X.1.0", "merge intersecting Entities of same type");
retract($first);
retract($second);
mergedEntity.getIntersectingNodes().forEach(node -> update(node));
end
// Rule unit: X.2
rule "X.2.0: Remove Entity of type ENTITY when contained by FALSE_POSITIVE"
salience 64
when
$falsePositive: TextEntity($type: type(), entityType == EntityType.FALSE_POSITIVE, active())
$entity: TextEntity(containedBy($falsePositive), type() == $type, (entityType == EntityType.ENTITY || entityType == EntityType.HINT), !hasManualChanges())
then
$entity.getIntersectingNodes().forEach(node -> update(node));
$entity.remove("X.2.0", "remove Entity of type ENTITY when contained by FALSE_POSITIVE");
retract($entity)
end
// Rule unit: X.3
rule "X.3.0: Remove Entity of type RECOMMENDATION when contained by FALSE_RECOMMENDATION"
salience 64
when
$falseRecommendation: TextEntity($type: type(), entityType == EntityType.FALSE_RECOMMENDATION, active())
$recommendation: TextEntity(containedBy($falseRecommendation), type() == $type, entityType == EntityType.RECOMMENDATION, !hasManualChanges())
then
$recommendation.remove("X.3.0", "remove Entity of type RECOMMENDATION when contained by FALSE_RECOMMENDATION");
retract($recommendation);
end
// Rule unit: X.4
rule "X.4.0: Remove Entity of type RECOMMENDATION when text range equals ENTITY with same type"
salience 256
when
$entity: TextEntity($type: type(), (entityType == EntityType.ENTITY || entityType == EntityType.HINT), active())
$recommendation: TextEntity(getTextRange().equals($entity.getTextRange()), type() == $type, entityType == EntityType.RECOMMENDATION, !hasManualChanges())
then
$entity.addEngines($recommendation.getEngines());
$recommendation.remove("X.4.0", "remove Entity of type RECOMMENDATION when text range equals ENTITY with same type");
retract($recommendation);
end
// Rule unit: X.5
rule "X.5.0: Remove Entity of type RECOMMENDATION when intersected by ENTITY"
salience 256
when
$entity: TextEntity((entityType == EntityType.ENTITY || entityType == EntityType.HINT), active())
$recommendation: TextEntity(intersects($entity), entityType == EntityType.RECOMMENDATION, !hasManualChanges())
then
$recommendation.remove("X.5.0", "remove Entity of type RECOMMENDATION when intersected by ENTITY");
retract($recommendation);
end
rule "X.5.1: Remove Entity of type RECOMMENDATION when contained by RECOMMENDATION"
salience 256
when
$entity: TextEntity($type: type(), entityType == EntityType.RECOMMENDATION, active())
$recommendation: TextEntity(containedBy($entity), type() != $type, entityType == EntityType.RECOMMENDATION, !hasManualChanges())
then
$recommendation.remove("X.5.1", "remove Entity of type RECOMMENDATION when contained by RECOMMENDATION");
retract($recommendation);
end
// Rule unit: X.6
rule "X.6.0: Remove Entity of lower rank, when contained by entity of type ENTITY or HINT"
salience 32
when
$higherRank: TextEntity($type: type(), (entityType == EntityType.ENTITY || entityType == EntityType.HINT), active())
$lowerRank: TextEntity(containedBy($higherRank), type() != $type, dictionary.getDictionaryRank(type) < dictionary.getDictionaryRank($type), !hasManualChanges())
then
$lowerRank.getIntersectingNodes().forEach(node -> update(node));
$lowerRank.remove("X.6.0", "remove Entity of lower rank, when contained by entity of type ENTITY or HINT");
retract($lowerRank);
end
rule "X.6.1: remove Entity, when contained in another entity of type ENTITY or HINT with larger text range"
salience 32
when
$outer: TextEntity($type: type(), (entityType == EntityType.ENTITY || entityType == EntityType.HINT), active())
$inner: TextEntity(containedBy($outer), type() != $type, $outer.getTextRange().length > getTextRange().length(), !hasManualChanges())
then
$inner.getIntersectingNodes().forEach(node -> update(node));
$inner.remove("X.6.1", "remove Entity, when contained in another entity of type ENTITY or HINT with larger text range");
retract($inner);
end
// Rule unit: X.7
rule "X.7.0: Remove all images"
salience 512
when
$image: Image(imageType != ImageType.OCR, !hasManualChanges())
then
$image.remove("X.7.0", "remove all images");
retract($image);
end
// Rule unit: X.8
rule "X.8.0: Remove Entity when text range and type equals to imported Entity"
salience 257
when
$entity: TextEntity($type: type(), engines contains Engine.IMPORTED, active())
$other: TextEntity(getTextRange().equals($entity.getTextRange()), this != $entity, type() == $type, engines not contains Engine.IMPORTED)
then
$other.remove("X.8.0", "remove Entity when text range and type equals to imported Entity");
$entity.addEngines($other.getEngines());
retract($other);
end
rule "X.8.1: Remove Entity when intersected by imported Entity"
salience 256
when
$entity: TextEntity(engines contains Engine.IMPORTED, active())
$other: TextEntity(intersects($entity), this != $entity, engines not contains Engine.IMPORTED)
then
$other.remove("X.8.1", "remove Entity when intersected by imported Entity");
retract($other);
end
// Rule unit: X.9
rule "X.9.0: Merge mostly contained signatures"
when
$aiSignature: Image(imageType == ImageType.SIGNATURE, engines contains LayoutEngine.AI)
$signature: Image(imageType == ImageType.SIGNATURE, engines contains LayoutEngine.ALGORITHM, mostlyContains($aiSignature, 0.8))
then
$aiSignature.remove("X.9.0", "removed because already contained by alogrithm signature");
$signature.addEngine(LayoutEngine.AI);
end
// Rule unit: X.10
rule "X.10.0: remove false positives of ai"
when
$anyImage: Image(engines contains LayoutEngine.ALGORITHM)
$aiSignature: Image(imageType == ImageType.SIGNATURE, engines contains LayoutEngine.AI, !mostlyContainedBy($anyImage, 0.8))
then
$aiSignature.remove("X.10.0", "Removed because false positive");
end
//------------------------------------ File attributes rules ------------------------------------
// Rule unit: FA.1
rule "FA.1.0: Remove duplicate FileAttributes"
salience 64
when
$fileAttribute: FileAttribute($label: label, $value: value)
$duplicate: FileAttribute(this != $fileAttribute, label == $label, value == $value)
then
retract($duplicate);
end
//------------------------------------ Local dictionary search rules ------------------------------------
// Rule unit: LDS.0
rule "LDS.0.0: Run local dictionary search"
agenda-group "LOCAL_DICTIONARY_ADDS"
salience -999
when
$dictionaryModel: DictionaryModel(!localEntriesWithMatchedRules.isEmpty()) from dictionary.getDictionaryModels()
then
entityCreationService.bySearchImplementation($dictionaryModel.getLocalSearch(), $dictionaryModel.getType(), EntityType.RECOMMENDATION, document)
.forEach(entity -> {
Collection<MatchedRule> matchedRules = $dictionaryModel.getMatchedRulesForLocalDictionaryEntry(entity.getValue());
matchedRules.forEach(matchedRule -> entity.addMatchedRule(matchedRule.asSkippedIfApplied()));
});
end