Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

Merge branch 'RED-8402' into 'master'

Browse Source 
RED-8402: hotfix for fixing exception caused by layout parser changes

Closes RED-8402

See merge request redactmanager/redaction-service!376
This commit is contained in:
Yannik Hampe 2024-04-17 12:50:27 +02:00
commit 1af74c3f2e
1 changed files with 9 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/service/document/NerEntitiesAdapter.java
View File

@ -12,7 +12,8 @@ import com.iqser.red.service.redaction.v1.server.client.model.NerEntitiesModel;
import com.iqser.red.service.redaction.v1.server.model.NerEntities; import com.iqser.red.service.redaction.v1.server.model.NerEntities;
import com.iqser.red.service.redaction.v1.server.model.document.TextRange; import com.iqser.red.service.redaction.v1.server.model.document.TextRange;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Document; import com.iqser.red.service.redaction.v1.server.model.document.nodes.Document;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.Section; import com.iqser.red.service.redaction.v1.server.model.document.nodes.NodeType;
import com.iqser.red.service.redaction.v1.server.model.document.nodes.SemanticNode;
import com.iqser.red.service.redaction.v1.server.model.document.textblock.TextBlock; import com.iqser.red.service.redaction.v1.server.model.document.textblock.TextBlock;
import lombok.AccessLevel; import lombok.AccessLevel;
@ -43,7 +44,7 @@ public class NerEntitiesAdapter {
*/ */
public NerEntities toNerEntities(NerEntitiesModel nerEntitiesModel, Document document) { public NerEntities toNerEntities(NerEntitiesModel nerEntitiesModel, Document document) {
return new NerEntities(addOffsetsAndFlatten(getStringStartOffsetsForMainSections(document), return new NerEntities(addOffsetsAndFlatten(getStringStartOffsetsForMainSectionsHeadersFooters(document),
nerEntitiesModel).map(nerEntityModel -> new NerEntities.NerEntity(nerEntityModel.getValue(), nerEntitiesModel).map(nerEntityModel -> new NerEntities.NerEntity(nerEntityModel.getValue(),
new TextRange(nerEntityModel.getStartOffset(), new TextRange(nerEntityModel.getStartOffset(),
nerEntityModel.getEndOffset()), nerEntityModel.getEndOffset()),
@ -160,11 +161,11 @@ public class NerEntitiesAdapter {
} }
private static Stream<EntityRecognitionEntity> addOffsetsAndFlatten(List<Integer> stringOffsetsForMainSections, NerEntitiesModel nerEntitiesModel) { private static Stream<EntityRecognitionEntity> addOffsetsAndFlatten(List<Integer> stringOffsetsForMainSectionsHeadersFooters, NerEntitiesModel nerEntitiesModel) {
nerEntitiesModel.getData() nerEntitiesModel.getData()
.forEach((sectionNumber, listOfNerEntities) -> listOfNerEntities.forEach(entityRecognitionEntity -> { .forEach((sectionNumber, listOfNerEntities) -> listOfNerEntities.forEach(entityRecognitionEntity -> {
int newStartOffset = entityRecognitionEntity.getStartOffset() + stringOffsetsForMainSections.get(sectionNumber); int newStartOffset = entityRecognitionEntity.getStartOffset() + stringOffsetsForMainSectionsHeadersFooters.get(sectionNumber);
entityRecognitionEntity.setStartOffset(newStartOffset); entityRecognitionEntity.setStartOffset(newStartOffset);
entityRecognitionEntity.setEndOffset(newStartOffset + entityRecognitionEntity.getValue().length()); entityRecognitionEntity.setEndOffset(newStartOffset + entityRecognitionEntity.getValue().length());
})); }));
@ -174,11 +175,11 @@ public class NerEntitiesAdapter {
} }
private static List<Integer> getStringStartOffsetsForMainSections(Document document) { private static List<Integer> getStringStartOffsetsForMainSectionsHeadersFooters(Document document) {
return document.getMainSections() return document.streamChildren()
.stream() .filter(child -> (child.getType().equals(NodeType.FOOTER) ||child.getType().equals(NodeType.HEADER) ||child.getType().equals(NodeType.SECTION)))
.map(Section::getTextBlock) .map(SemanticNode::getTextBlock)
.map(TextBlock::getTextRange) .map(TextBlock::getTextRange)
.map(TextRange::start) .map(TextRange::start)
.toList(); .toList();