RED-6725: integrate layoutparsing service

This commit is contained in:
Kilian Schuettler 2023-07-24 16:50:20 +02:00
parent e7b29fb683
commit 12703c6eda
15 changed files with 436 additions and 28 deletions

View File

@ -88,6 +88,12 @@
</exclusions>
</dependency>
<dependency>
<groupId>com.knecon.fforesight</groupId>
<artifactId>layoutparser-service-internal-api</artifactId>
<version>0.3.0</version>
</dependency>
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>search-service-api-v1</artifactId>

View File

@ -1,5 +1,9 @@
package com.iqser.red.service.persistence.management.v1.processor.configuration;
import static com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingQueueNames.LAYOUT_PARSING_DLQ;
import static com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE;
import static com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingQueueNames.LAYOUT_PARSING_REQUEST_QUEUE;
import org.springframework.amqp.core.Queue;
import org.springframework.amqp.core.QueueBuilder;
import org.springframework.context.annotation.Bean;
@ -58,6 +62,8 @@ public class MessagingConfiguration {
public static final String OCR_STATUS_UPDATE_RESPONSE_QUEUE = "ocr_status_update_response_queue";
public static final String OCR_STATUS_UPDATE_RESPONSE_DQL = "ocr_status_update_response_dql";
public static final String X_ERROR_INFO_HEADER = "x-error-message";
public static final String X_ERROR_INFO_TIMESTAMP_HEADER = "x-error-message-timestamp";
@Bean
public Queue nerRequestQueue() {
@ -311,4 +317,25 @@ public class MessagingConfiguration {
.build();
}
@Bean
public Queue layoutparsingRequestQueue() {
return QueueBuilder.durable(LAYOUT_PARSING_REQUEST_QUEUE)//
.withArgument("x-dead-letter-exchange", "").withArgument("x-dead-letter-routing-key", LAYOUT_PARSING_DLQ).build();
}
@Bean
public Queue layoutparsingResponseQueue() {
return QueueBuilder.durable(LAYOUT_PARSING_FINISHED_EVENT_QUEUE)//
.withArgument("x-dead-letter-exchange", "").withArgument("x-dead-letter-routing-key", LAYOUT_PARSING_DLQ).build();
}
@Bean
public Queue layoutparsingDLQ() {
return QueueBuilder.durable(LAYOUT_PARSING_DLQ).build();
}
}

View File

@ -40,12 +40,6 @@ public class FileStatusProcessingUpdateService {
switch (analyzeResult.getMessageType()) {
case STRUCTURE_ANALYSE:
//TODO This might be also priority depending on what was the pervious call.
fileStatusService.setStatusAnalyse(dossierId, fileId, false);
break;
case SURROUNDING_TEXT:
fileStatusService.setStatusProcessed(analyzeResult.getFileId());
manualRedactionService.updateSurroundingText(fileId, analyzeResult.getManualRedactions());
@ -75,7 +69,6 @@ public class FileStatusProcessingUpdateService {
}
@Transactional
public void preprocessingSuccessful(String dossierId, String fileId, UntouchedDocumentResponse untouchedDocumentResponse) {
fileStatusService.updateProcessingStatusPreprocessed(dossierId, fileId, untouchedDocumentResponse.isHasHighlights(), untouchedDocumentResponse.getFileSize());

View File

@ -1,5 +1,7 @@
package com.iqser.red.service.persistence.management.v1.processor.service;
import static com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingQueueNames.LAYOUT_PARSING_REQUEST_QUEUE;
import java.time.OffsetDateTime;
import java.util.ArrayList;
import java.util.List;
@ -12,7 +14,6 @@ import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.google.common.collect.Sets;
import com.iqser.red.service.pdftron.redaction.v1.api.model.DocumentRequest;
import com.iqser.red.service.pdftron.redaction.v1.api.model.ProcessUntouchedDocumentRequest;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.entity.dossier.FileAttributeEntity;
@ -22,6 +23,7 @@ import com.iqser.red.service.persistence.management.v1.processor.model.CvAnalysi
import com.iqser.red.service.persistence.management.v1.processor.model.NerServiceRequest;
import com.iqser.red.service.persistence.management.v1.processor.model.OCRStatusUpdateResponse;
import com.iqser.red.service.persistence.management.v1.processor.model.image.ImageServiceRequest;
import com.iqser.red.service.persistence.management.v1.processor.service.layoutparsing.LayoutParsingRequestFactory;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.DossierPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.FileAttributeConfigPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.FileStatusPersistenceService;
@ -74,6 +76,7 @@ public class FileStatusService {
private final ReanalysisRequiredStatusService reanalysisRequiredStatusService;
private final ViewedPagesPersistenceService viewedPagesPersistenceService;
private final FileManagementServiceSettings fileManagementServiceSettings;
private final LayoutParsingRequestFactory layoutParsingRequestFactory;
@Transactional
@ -174,7 +177,7 @@ public class FileStatusService {
}
var fileModel = MagicConverter.convert(fileEntity, FileModel.class, new FileModelMapper());
reanalysisRequiredStatusService.enhanceFileStatusWithAnalysisRequirements(fileModel, true);
fileModel = reanalysisRequiredStatusService.enhanceFileStatusWithAnalysisRequirements(fileModel, true);
var dossierTemplate = dossier.getDossierTemplate();
if (dossierTemplate.isOcrByDefault() && fileModel.getOcrEndTime() == null) {
@ -183,22 +186,20 @@ public class FileStatusService {
return;
}
MessageType messageType = null;
if (!fileManagementStorageService.objectExists(dossierId, fileId, FileType.TEXT)) {
messageType = MessageType.STRUCTURE_ANALYSE;
if (!fileManagementStorageService.objectExists(dossierId, fileId, FileType.DOCUMENT_STRUCTURE)) {
var layoutParsingRequest = layoutParsingRequestFactory.build(dossierId, fileId, priority, dossier);
rabbitTemplate.convertAndSend(LAYOUT_PARSING_REQUEST_QUEUE, layoutParsingRequest);
return;
}
if (messageType == null && settings.isNerServiceEnabled() && !fileManagementStorageService.objectExists(dossierId, fileId, FileType.NER_ENTITIES)) {
if (settings.isNerServiceEnabled() && !fileManagementStorageService.objectExists(dossierId, fileId, FileType.NER_ENTITIES)) {
log.debug("Add file: {} from dossier {} to NER queue", fileId, dossierId);
addToNerQueue(dossierId, fileId);
return;
}
if (messageType == null) {
boolean reanalyse = fileModel.isReanalysisRequired() || manualRedactionReanalyse;
messageType = calculateMessageType(reanalyse, fileModel.getProcessingStatus(), fileModel);
}
boolean reanalyse = fileModel.isReanalysisRequired() || manualRedactionReanalyse;
MessageType messageType = calculateMessageType(reanalyse, fileModel.getProcessingStatus(), fileModel);
var analyseRequest = AnalyzeRequest.builder()
.messageType(messageType)

View File

@ -10,7 +10,6 @@ import java.util.stream.Collectors;
import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.google.common.collect.Sets;
import com.iqser.red.service.pdftron.redaction.v1.api.model.ByteContentDocument;
import com.iqser.red.service.pdftron.redaction.v1.api.model.highlights.TextHighlightConversionOperation;
@ -39,7 +38,6 @@ public class ReanalysisService {
private final IndexingService indexingService;
private final PDFTronClient pDFTronRedactionClient;
private final FileManagementStorageService fileManagementStorageService;
private final ObjectMapper objectMapper;
public void reanalyzeDossier(String dossierId, boolean force) {

View File

@ -0,0 +1,44 @@
package com.iqser.red.service.persistence.management.v1.processor.service.layoutparsing;
import java.util.Map;
import java.util.Optional;
import org.springframework.stereotype.Service;
import com.iqser.red.service.persistence.management.v1.processor.entity.dossier.DossierEntity;
import com.iqser.red.service.persistence.management.v1.processor.service.FileManagementStorageService;
import com.iqser.red.service.persistence.management.v1.processor.utils.StorageIdUtils;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;
import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
import lombok.RequiredArgsConstructor;
@Service
@RequiredArgsConstructor
public class LayoutParsingRequestFactory {
private final FileManagementStorageService fileManagementStorageService;
private final LayoutParsingRequestIdentifierService layoutParsingRequestIdentifierService;
public LayoutParsingRequest build(String dossierId, String fileId, boolean priority, DossierEntity dossier) {
Optional<String> optionalImageFileId = fileManagementStorageService.objectExists(dossierId, fileId, FileType.IMAGE_INFO) //
? Optional.of(StorageIdUtils.getStorageId(dossierId, fileId, FileType.IMAGE_INFO)) : Optional.empty();
Optional<String> optionalTableFileId = fileManagementStorageService.objectExists(dossierId, fileId, FileType.TABLES) //
? Optional.of(StorageIdUtils.getStorageId(dossierId, fileId, FileType.TABLES)) : Optional.empty();
return LayoutParsingRequest.builder()
.identifier(layoutParsingRequestIdentifierService.buildIdentifier(dossierId, fileId, priority, dossier.getDossierTemplateId()))
.originFileStorageId(StorageIdUtils.getStorageId(dossierId, fileId, FileType.ORIGIN))
.imagesFileStorageId(optionalImageFileId)
.tablesFileStorageId(optionalTableFileId)
.pageFileStorageId(StorageIdUtils.getStorageId(dossierId, fileId, FileType.DOCUMENT_PAGES))
.structureFileStorageId(StorageIdUtils.getStorageId(dossierId, fileId, FileType.DOCUMENT_STRUCTURE))
.textBlockFileStorageId(StorageIdUtils.getStorageId(dossierId, fileId, FileType.DOCUMENT_TEXT))
.positionBlockFileStorageId(StorageIdUtils.getStorageId(dossierId, fileId, FileType.DOCUMENT_POSITION))
.build();
}
}

View File

@ -0,0 +1,57 @@
package com.iqser.red.service.persistence.management.v1.processor.service.layoutparsing;
import java.util.Map;
import org.springframework.stereotype.Service;
import com.iqser.red.service.persistence.management.v1.processor.entity.dossier.DossierEntity;
@Service
public class LayoutParsingRequestIdentifierService {
private enum IdentifierNames {
DOSSIER_ID,
FILE_ID,
PRIORITY,
DOSSIER_TEMPLATE_ID
}
public String parseDossierTemplateId(Map<String, String> identifiers) {
return identifiers.get(IdentifierNames.DOSSIER_TEMPLATE_ID.name());
}
public String parseDossierId(Map<String, String> identifiers) {
return identifiers.get(IdentifierNames.DOSSIER_ID.name());
}
public String parseFileId(Map<String, String> identifiers) {
return identifiers.get(IdentifierNames.FILE_ID.name());
}
public Boolean parsePriority(Map<String, String> identifiers) {
return Boolean.parseBoolean(identifiers.get(IdentifierNames.PRIORITY.name()));
}
public Map<String, String> buildIdentifier(String dossierId, String fileId, boolean priority, String dossierTemplateId) {
return Map.of(IdentifierNames.DOSSIER_TEMPLATE_ID.name(),
dossierTemplateId,
IdentifierNames.DOSSIER_ID.name(),
dossierId,
IdentifierNames.FILE_ID.name(),
fileId,
IdentifierNames.PRIORITY.name(),
String.valueOf(priority));
}
}

View File

@ -1,4 +1,4 @@
package com.iqser.red.service.persistence.management.v1.processor.service;
package com.iqser.red.service.persistence.management.v1.processor.service.queue;
import java.time.OffsetDateTime;
import java.time.temporal.ChronoUnit;
@ -10,6 +10,8 @@ import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.model.CvAnalysisServiceResponse;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusProcessingUpdateService;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileErrorInfo;
import lombok.RequiredArgsConstructor;

View File

@ -1,4 +1,4 @@
package com.iqser.red.service.persistence.management.v1.processor.service;
package com.iqser.red.service.persistence.management.v1.processor.service.queue;
import java.io.IOException;
import java.time.OffsetDateTime;
@ -13,6 +13,8 @@ import com.fasterxml.jackson.core.type.TypeReference;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusProcessingUpdateService;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import com.iqser.red.service.persistence.management.v1.processor.settings.FileManagementServiceSettings;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileErrorInfo;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;

View File

@ -0,0 +1,63 @@
package com.iqser.red.service.persistence.management.v1.processor.service.queue;
import java.time.OffsetDateTime;
import java.time.temporal.ChronoUnit;
import org.springframework.amqp.core.Message;
import org.springframework.amqp.rabbit.annotation.RabbitListener;
import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusProcessingUpdateService;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import com.iqser.red.service.persistence.management.v1.processor.service.layoutparsing.LayoutParsingRequestIdentifierService;
import com.iqser.red.service.persistence.service.v1.api.shared.model.AnalyzeRequest;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileErrorInfo;
import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingFinishedEvent;
import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingQueueNames;
import lombok.RequiredArgsConstructor;
import lombok.SneakyThrows;
import lombok.extern.slf4j.Slf4j;
@Slf4j
@Service
@RequiredArgsConstructor
public class LayoutParsingFinishedMessageReceiver {
private final FileStatusService fileStatusService;
private final FileStatusProcessingUpdateService fileStatusProcessingUpdateService;
private final ObjectMapper objectMapper;
private final LayoutParsingRequestIdentifierService layoutParsingRequestIdentifierService;
@SneakyThrows
@RabbitListener(queues = LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE)
public void receive(LayoutParsingFinishedEvent response) {
fileStatusService.setStatusAnalyse(layoutParsingRequestIdentifierService.parseDossierId(response.identifier()),
layoutParsingRequestIdentifierService.parseFileId(response.identifier()),
layoutParsingRequestIdentifierService.parsePriority(response.identifier()));
log.info("Received message {} in {}", response, MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE);
}
@SneakyThrows
@RabbitListener(queues = LayoutParsingQueueNames.LAYOUT_PARSING_DLQ)
public void handleDLQMessage(Message failedMessage) {
var analyzeRequest = objectMapper.readValue(failedMessage.getBody(), AnalyzeRequest.class);
log.info("Failed to process analyze request: {}", analyzeRequest);
String errorCause = failedMessage.getMessageProperties().getHeader(MessagingConfiguration.X_ERROR_INFO_HEADER);
OffsetDateTime timestamp = failedMessage.getMessageProperties().getHeader(MessagingConfiguration.X_ERROR_INFO_TIMESTAMP_HEADER);
timestamp = timestamp != null ? timestamp : OffsetDateTime.now().truncatedTo(ChronoUnit.MILLIS);
log.info("Failed to process layout parsing request, errorCause: {}, timestamp: {}", errorCause, timestamp);
fileStatusProcessingUpdateService.analysisFailed(analyzeRequest.getDossierId(),
analyzeRequest.getFileId(),
new FileErrorInfo(errorCause, LayoutParsingQueueNames.LAYOUT_PARSING_DLQ, "redaction-service", timestamp));
}
}

View File

@ -1,4 +1,4 @@
package com.iqser.red.service.persistence.management.v1.processor.service;
package com.iqser.red.service.persistence.management.v1.processor.service.queue;
import java.io.IOException;
import java.time.OffsetDateTime;
@ -12,6 +12,8 @@ import org.springframework.stereotype.Service;
import com.fasterxml.jackson.core.type.TypeReference;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusProcessingUpdateService;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileErrorInfo;
import lombok.RequiredArgsConstructor;

View File

@ -1,4 +1,4 @@
package com.iqser.red.service.persistence.management.v1.processor.service;
package com.iqser.red.service.persistence.management.v1.processor.service.queue;
import org.springframework.amqp.core.Message;
import org.springframework.amqp.rabbit.annotation.RabbitListener;
@ -7,6 +7,7 @@ import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.persistence.management.v1.processor.configuration.MessagingConfiguration;
import com.iqser.red.service.persistence.management.v1.processor.model.OCRStatusUpdateResponse;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import lombok.RequiredArgsConstructor;
import lombok.SneakyThrows;

View File

@ -0,0 +1,205 @@
package com.iqser.red.service.peristence.v1.server.integration.tests;
import static org.assertj.core.api.Assertions.assertThat;
import static org.mockito.ArgumentMatchers.any;
import static org.mockito.Mockito.times;
import static org.mockito.Mockito.verify;
import static org.mockito.Mockito.when;
import java.util.Collections;
import java.util.Set;
import org.junit.jupiter.api.BeforeEach;
import org.junit.jupiter.api.Test;
import org.mockito.Mock;
import org.mockito.MockitoAnnotations;
import org.springframework.amqp.rabbit.core.RabbitTemplate;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.commons.jackson.ObjectMapperFactory;
import com.iqser.red.service.pdftron.redaction.v1.api.model.highlights.TextHighlightConversionOperation;
import com.iqser.red.service.pdftron.redaction.v1.api.model.highlights.TextHighlightConversionRequest;
import com.iqser.red.service.persistence.management.v1.processor.client.pdftronredactionservice.PDFTronClient;
import com.iqser.red.service.persistence.management.v1.processor.entity.dossier.DossierEntity;
import com.iqser.red.service.persistence.management.v1.processor.entity.dossier.FileEntity;
import com.iqser.red.service.persistence.management.v1.processor.service.FileManagementStorageService;
import com.iqser.red.service.persistence.management.v1.processor.service.FileStatusService;
import com.iqser.red.service.persistence.management.v1.processor.service.IndexingService;
import com.iqser.red.service.persistence.management.v1.processor.service.ManualRedactionProviderService;
import com.iqser.red.service.persistence.management.v1.processor.service.ReanalysisRequiredStatusService;
import com.iqser.red.service.persistence.management.v1.processor.service.ReanalysisService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.DictionaryPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.DossierPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.FileAttributeConfigPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.FileStatusPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.LegalBasisMappingPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.RulesPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.ViewedPagesPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.AddRedactionPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.CommentPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.ForceRedactionPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.ImageRecategorizationPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.LegalBasisChangePersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.RemoveRedactionPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.annotations.ResizeRedactionPersistenceService;
import com.iqser.red.service.persistence.management.v1.processor.settings.FileManagementServiceSettings;
import com.iqser.red.service.persistence.management.v1.processor.utils.FileModelMapper;
import com.iqser.red.service.persistence.management.v1.processor.utils.MagicConverter;
import com.iqser.red.service.persistence.service.v1.api.shared.model.FileStatus;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.Dossier;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileModel;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.FileType;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.ProcessingStatus;
import com.iqser.red.service.persistence.service.v1.api.shared.model.dossiertemplate.dossier.file.WorkflowStatus;
class ReanalysisServiceTest {
private ReanalysisService reanalysisService;
private FileStatusService fileStatusService;
@Mock
private ReanalysisRequiredStatusService reanalysisRequiredStatusService;
@Mock
private IndexingService indexingService;
@Mock
private DossierPersistenceService dossierPersistenceService;
@Mock
private PDFTronClient pdfTronClient;
@Mock
private RabbitTemplate rabbitTemplate;
@Mock
private FileStatusPersistenceService fileStatusPersistenceService;
@Mock
private ObjectMapper objectMapper;
@Mock
private ManualRedactionProviderService manualRedactionProviderService;
@Mock
private FileManagementStorageService fileManagementStorageService;
@Mock
private LegalBasisChangePersistenceService legalBasisChangePersistenceService;
@Mock
private ImageRecategorizationPersistenceService imageRecategorizationPersistenceService;
@Mock
private CommentPersistenceService commentPersistenceService;
@Mock
private ForceRedactionPersistenceService forceRedactionPersistenceService;
@Mock
private RemoveRedactionPersistenceService removeRedactionPersistenceService;
@Mock
private AddRedactionPersistenceService addRedactionPersistenceService;
@Mock
private ResizeRedactionPersistenceService resizeRedactionPersistenceService;
@Mock
private FileAttributeConfigPersistenceService fileAttributeConfigPersistenceService;
@Mock
private FileManagementServiceSettings settings;
@Mock
private ViewedPagesPersistenceService viewedPagesPersistenceService;
@Mock
private FileManagementServiceSettings fileManagementServiceSettings;
@Mock
private DictionaryPersistenceService dictionaryPersistenceService;
@Mock
private RulesPersistenceService rulesPersistenceService;
@Mock
private LegalBasisMappingPersistenceService legalBasisMappingPersistenceService;
private static String DOSSIER_ID = "123";
private static String FILE_ID = "456";
private FileModel fileModel;
@BeforeEach
public void stubAndCreate() {
fileModel = MagicConverter.convert(getTestFileEntity(), FileModel.class, new FileModelMapper());
MockitoAnnotations.openMocks(this);
fileStatusService = new FileStatusService(fileStatusPersistenceService,
dossierPersistenceService,
rabbitTemplate,
ObjectMapperFactory.create(),
manualRedactionProviderService,
fileManagementStorageService,
legalBasisChangePersistenceService,
imageRecategorizationPersistenceService,
commentPersistenceService,
forceRedactionPersistenceService,
removeRedactionPersistenceService,
addRedactionPersistenceService,
resizeRedactionPersistenceService,
fileAttributeConfigPersistenceService,
settings,
reanalysisRequiredStatusService,
viewedPagesPersistenceService,
fileManagementServiceSettings);
reanalysisService = new ReanalysisService(fileStatusService, dossierPersistenceService, indexingService, pdfTronClient, fileManagementStorageService);
when(fileStatusPersistenceService.getStatus(FILE_ID)).thenReturn(getTestFileEntity());
when(dossierPersistenceService.getAndValidateDossier(DOSSIER_ID)).thenReturn(getDossierEntity());
when(fileManagementStorageService.objectExists(DOSSIER_ID, FILE_ID, FileType.ORIGIN)).thenReturn(true);
when(fileManagementStorageService.objectExists(DOSSIER_ID, FILE_ID, FileType.TEXT)).thenReturn(true);
when(fileManagementStorageService.objectExists(DOSSIER_ID, FILE_ID, FileType.NER_ENTITIES)).thenReturn(true);
when(settings.isFigureDetectionEnabled()).thenReturn(false);
when(settings.isCvTableParsingEnabled()).thenReturn(false);
when(settings.isImageServiceEnabled()).thenReturn(false);
when(settings.isOcrByDefault()).thenReturn(false);
when(dossierPersistenceService.findByDossierId(DOSSIER_ID)).thenReturn(getDossierEntity());
when(reanalysisRequiredStatusService.enhanceFileStatusWithAnalysisRequirements(fileModel, true)) //
.thenReturn(fileModel);
}
private Dossier getTestDossier() {
return Dossier.builder().id(DOSSIER_ID).build();
}
private static FileStatus getTestFileStatus() {
return FileStatus.builder().fileId(FILE_ID).dossierId(DOSSIER_ID).processingStatus(ProcessingStatus.PROCESSED).build();
}
private static FileEntity getTestFileEntity() {
return FileEntity.builder()
.id(FILE_ID)
.dossierId(DOSSIER_ID)
.workflowStatus(WorkflowStatus.NEW)
.fileAttributes(Collections.emptyList())
.excluded(false)
.processingStatus(ProcessingStatus.PROCESSED)
.build();
}
private static DossierEntity getDossierEntity() {
return DossierEntity.builder().id(DOSSIER_ID).build();
}
@Test
public void reanalysisTriggeredByHighLightConversionTest() {
var dossier = getTestDossier();
var fileStatus = getTestFileStatus();
var textHighlightRequest = new TextHighlightConversionRequest(dossier.getDossierId(),
fileStatus.getFileId(),
Set.of("1", "2", "3"),
TextHighlightConversionOperation.CONVERT);
when(pdfTronClient.convertTextHighlights(textHighlightRequest)).thenReturn(true);
reanalysisService.convertTextHighlights(textHighlightRequest);
assertThat(fileModel.getAnalysisVersion()).isEqualTo(1);
verify(rabbitTemplate, times(1)).convertAndSend((String) any(), (Object) any());
verify(fileStatusPersistenceService, times(1)).updateProcessingStatus(FILE_ID, ProcessingStatus.FULL_PROCESSING);
}
}

View File

@ -135,20 +135,23 @@ public class FileStatus {
private boolean hasHighlights;
@Schema(description = "Size of the optimized, internally stored file.")
private Long fileSize;
@Schema(description = "Analysis Version.")
@Schema(description = "Analysis Version.")
private int analysisVersion;
@Schema(description = "Last time the file was indexed in ES.")
private OffsetDateTime lastIndexed;
@Schema(description = "The error information for the error state of the file")
private FileErrorInfo fileErrorInfo;
@Schema(description = "Shows if this file has been OCRed by us. Last Time of OCR.")
public OffsetDateTime getLastOCRTime() {
return ocrEndTime != null ? ocrEndTime : ocrStartTime;
}
public String getId(){
public String getId() {
return fileId;
}

View File

@ -9,14 +9,18 @@ public enum FileType {
REDACTION_LOG(".json"),
SIMPLIFIED_TEXT(".json"),
SECTION_GRID(".json"),
TEXT(".json"),
TEXT(".json"), // TODO: refactor this away
NER_ENTITIES(".json"),
IMAGE_INFO(".json"),
IMPORTED_REDACTIONS(".json"),
TEXT_HIGHLIGHTS(".json"),
FIGURE(".json"),
TABLES(".json"),
COMPONENTS(".json");
COMPONENTS(".json"),
DOCUMENT_TEXT(".json"),
DOCUMENT_STRUCTURE(".json"),
DOCUMENT_POSITION(".json"),
DOCUMENT_PAGES(".json");
@Getter
private final String extension;