RED-7375: integrate table extractor

This commit is contained in:
Kilian Schuettler 2023-08-21 12:48:46 +02:00
parent 07c5530631
commit 8a6dd4c5e8
9 changed files with 151 additions and 1 deletions

View File

@ -59,6 +59,10 @@ public class MessagingConfiguration {
public static final String CV_ANALYSIS_RESPONSE_QUEUE = "cv_analysis_response_queue";
public static final String CV_ANALYSIS_DLQ = "cv_analysis_dead_letter_queue";
public static final String TABLE_EXTRACTOR_QUEUE = "table_extractor_request_queue";
public static final String TABLE_EXTRACTOR_RESPONSE_QUEUE = "table_extractor_response_queue";
public static final String TABLE_EXTRACTOR_DLQ = "table_extractor_dead_letter_queue";
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";
@ -338,4 +342,26 @@ public class MessagingConfiguration {
return QueueBuilder.durable(LAYOUT_PARSING_DLQ).build();
}
@Bean
public Queue tableExtractorRequestQueue() {
return QueueBuilder.durable(TABLE_EXTRACTOR_QUEUE)//
.withArgument("x-dead-letter-exchange", "").withArgument("x-dead-letter-routing-key", LAYOUT_PARSING_DLQ).build();
}
@Bean
public Queue tableExtractorResponseQueue() {
return QueueBuilder.durable(TABLE_EXTRACTOR_RESPONSE_QUEUE)//
.withArgument("x-dead-letter-exchange", "").withArgument("x-dead-letter-routing-key", LAYOUT_PARSING_DLQ).build();
}
@Bean
public Queue tableExtractorDLQ() {
return QueueBuilder.durable(TABLE_EXTRACTOR_DLQ).build();
}
}

View File

@ -0,0 +1,23 @@
package com.iqser.red.service.persistence.management.v1.processor.model;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import lombok.NoArgsConstructor;
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class TableExtractorRequest {
public static final String TABLE_EXTRACTOR_FILE_EXTENSION = "EXTRACTED_TABLES.json.gz";
public static final String TARGET_FILE_EXTENSION = "ORIGIN.pdf.gz";
private String dossierId;
private String fileId;
private String targetFileExtension;
private String responseFileExtension;
}

View File

@ -0,0 +1,17 @@
package com.iqser.red.service.persistence.management.v1.processor.model;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import lombok.NoArgsConstructor;
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class TableExtractorResponse {
private String dossierId;
private String fileId;
}

View File

@ -21,6 +21,7 @@ import com.iqser.red.service.persistence.management.v1.processor.exception.Inter
import com.iqser.red.service.persistence.management.v1.processor.model.CvAnalysisServiceRequest;
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.TableExtractorRequest;
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.manualredactions.ManualRedactionProviderService;
@ -175,6 +176,12 @@ public class FileStatusService {
return;
}
if (settings.isTableExtractorEnabled() && !fileManagementStorageService.objectExists(dossierId, fileId, FileType.INVISIBLE_TABLES)) {
log.info("Add file: {} from dossier {} to Table Extractor queue", fileId, dossierId);
addToTableExtractorQueue(dossierId, fileId);
return;
}
var fileModel = MagicConverter.convert(fileEntity, FileModel.class, new FileModelMapper());
fileModel = reanalysisRequiredStatusService.enhanceFileStatusWithAnalysisRequirements(fileModel, true);
@ -230,6 +237,24 @@ public class FileStatusService {
}
private void addToTableExtractorQueue(String dossierId, String fileId) {
fileStatusPersistenceService.updateProcessingStatus(fileId, ProcessingStatus.TABLE_EXTRACTOR_ANALYZING);
rabbitTemplate.convertAndSend(MessagingConfiguration.TABLE_EXTRACTOR_QUEUE,
TableExtractorRequest.builder()
.dossierId(dossierId)
.fileId(fileId)
.targetFileExtension(TableExtractorRequest.TARGET_FILE_EXTENSION)
.responseFileExtension(TableExtractorRequest.TABLE_EXTRACTOR_FILE_EXTENSION)
.build(),
message -> {
message.getMessageProperties().setPriority(1);
return message;
});
}
@SneakyThrows
public void addToPreprocessingQueue(String dossierId, String fileId, String filename) {

View File

@ -0,0 +1,55 @@
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.model.CvAnalysisServiceResponse;
import com.iqser.red.service.persistence.management.v1.processor.model.TableExtractorResponse;
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;
import lombok.SneakyThrows;
import lombok.extern.slf4j.Slf4j;
@Slf4j
@Service
@RequiredArgsConstructor
public class TableExtractorMessageReceiver {
private final ObjectMapper objectMapper;
private final FileStatusService fileStatusService;
private final FileStatusProcessingUpdateService fileStatusProcessingUpdateService;
@SneakyThrows
@RabbitListener(queues = MessagingConfiguration.TABLE_EXTRACTOR_RESPONSE_QUEUE)
public void receive(TableExtractorResponse response) {
fileStatusService.setStatusAnalyse(response.getDossierId(), response.getFileId(), false);
log.info("Received message in {} for dossierId {} and fileId {}", MessagingConfiguration.TABLE_EXTRACTOR_RESPONSE_QUEUE, response.getDossierId(), response.getFileId());
}
@SneakyThrows
@RabbitListener(queues = MessagingConfiguration.TABLE_EXTRACTOR_DLQ)
public void handleDLQMessage(Message failedMessage) {
var response = objectMapper.readValue(failedMessage.getBody(), TableExtractorResponse.class);
log.warn("Received message from {} for dossierId {} and fileId {}", MessagingConfiguration.TABLE_EXTRACTOR_DLQ, response.getDossierId(), response.getFileId());
fileStatusProcessingUpdateService.analysisFailed(response.getDossierId(),
response.getFileId(),
new FileErrorInfo("table extractor failed", MessagingConfiguration.TABLE_EXTRACTOR_DLQ, "table-extractor", OffsetDateTime.now().truncatedTo(ChronoUnit.MILLIS)));
}
}

View File

@ -24,6 +24,7 @@ public class FileManagementServiceSettings {
private boolean imageServiceEnabled = true;
private boolean nerServiceEnabled = true;
private boolean tableExtractorEnabled = true;
private boolean storeImageFile = true;

View File

@ -30,6 +30,7 @@ cors.enabled: true
persistence-service:
imageServiceEnabled: false
nerServiceEnabled: false
tableExtractorEnabled: false
storeImageFile: false
applicationName: RedactManager
fforesight:

View File

@ -17,6 +17,7 @@ public enum FileType {
TEXT_HIGHLIGHTS(".json"),
FIGURE(".json"),
TABLES(".json"),
INVISIBLE_TABLES(".json"),
COMPONENTS(".json"),
// document is split into 4 files, all should be overridden/deleted at the same time
DOCUMENT_TEXT(".json"),

View File

@ -18,5 +18,6 @@ public enum ProcessingStatus {
PRE_PROCESSING,
PRE_PROCESSED,
FIGURE_DETECTION_ANALYZING,
TABLE_PARSING_ANALYZING
TABLE_PARSING_ANALYZING,
TABLE_EXTRACTOR_ANALYZING,
}