RED-10728: Endpoint to execute full OCR on specific file

This commit is contained in:
maverickstuder 2025-01-16 14:38:17 +01:00
parent 7669d27e7b
commit 43f9db59d4
7 changed files with 28 additions and 23 deletions

View File

@ -118,11 +118,12 @@ public class ReanalysisController implements ReanalysisResource {
@PreAuthorize("hasAuthority('" + REANALYZE_FILE + "')") @PreAuthorize("hasAuthority('" + REANALYZE_FILE + "')")
public void ocrFile(@PathVariable(DOSSIER_ID) String dossierId, public void ocrFile(@PathVariable(DOSSIER_ID) String dossierId,
@PathVariable(FILE_ID) String fileId, @PathVariable(FILE_ID) String fileId,
@RequestParam(value = FORCE_PARAM, required = false, defaultValue = FALSE) boolean force) { @RequestParam(value = FORCE_PARAM, required = false, defaultValue = FALSE) boolean force,
@RequestParam(value = ALL_PAGES, required = false, defaultValue = FALSE) boolean allPages) {
accessControlService.checkDossierExistenceAndAccessPermissionsToDossier(dossierId); accessControlService.checkDossierExistenceAndAccessPermissionsToDossier(dossierId);
validateOCR(dossierId, fileId); validateOCR(dossierId, fileId);
reanalysisService.ocrFile(dossierId, fileId, force); reanalysisService.ocrFile(dossierId, fileId, force, allPages);
auditPersistenceService.audit(AuditRequest.builder() auditPersistenceService.audit(AuditRequest.builder()
.userId(KeycloakSecurity.getUserId()) .userId(KeycloakSecurity.getUserId())
.objectId(dossierId) .objectId(dossierId)
@ -140,7 +141,7 @@ public class ReanalysisController implements ReanalysisResource {
accessControlService.checkDossierExistenceAndAccessPermissionsToDossier(dossierId); accessControlService.checkDossierExistenceAndAccessPermissionsToDossier(dossierId);
fileIds.forEach(fileId -> validateOCR(dossierId, fileId)); fileIds.forEach(fileId -> validateOCR(dossierId, fileId));
reanalysisService.ocrFiles(dossierId, fileIds); reanalysisService.ocrFiles(dossierId, fileIds, false);
auditPersistenceService.audit(AuditRequest.builder() auditPersistenceService.audit(AuditRequest.builder()
.userId(KeycloakSecurity.getUserId()) .userId(KeycloakSecurity.getUserId())
.objectId(dossierId) .objectId(dossierId)

View File

@ -38,6 +38,7 @@ public interface ReanalysisResource {
String EXCLUDED_STATUS_PARAM = "excluded"; String EXCLUDED_STATUS_PARAM = "excluded";
String FORCE_PARAM = "force"; String FORCE_PARAM = "force";
String ALL_PAGES = "allPages";
@PostMapping(value = REANALYSIS_REST_PATH + DOSSIER_ID_PATH_VARIABLE) @PostMapping(value = REANALYSIS_REST_PATH + DOSSIER_ID_PATH_VARIABLE)
@ -73,7 +74,8 @@ public interface ReanalysisResource {
@ApiResponses(value = {@ApiResponse(responseCode = "204", description = "OK"), @ApiResponse(responseCode = "409", description = "Conflict"), @ApiResponse(responseCode = "404", description = "Not found"), @ApiResponse(responseCode = "403", description = "Forbidden"), @ApiResponse(responseCode = "400", description = "Cannot OCR approved file")}) @ApiResponses(value = {@ApiResponse(responseCode = "204", description = "OK"), @ApiResponse(responseCode = "409", description = "Conflict"), @ApiResponse(responseCode = "404", description = "Not found"), @ApiResponse(responseCode = "403", description = "Forbidden"), @ApiResponse(responseCode = "400", description = "Cannot OCR approved file")})
void ocrFile(@PathVariable(DOSSIER_ID) String dossierId, void ocrFile(@PathVariable(DOSSIER_ID) String dossierId,
@PathVariable(FILE_ID) String fileId, @PathVariable(FILE_ID) String fileId,
@RequestParam(value = FORCE_PARAM, required = false, defaultValue = FALSE) boolean force); @RequestParam(value = FORCE_PARAM, required = false, defaultValue = FALSE) boolean force,
@RequestParam(value = ALL_PAGES, required = false, defaultValue = FALSE) boolean allPages);
@Operation(summary = "Ocr and reanalyze multiple files for a dossier", description = "None") @Operation(summary = "Ocr and reanalyze multiple files for a dossier", description = "None")

View File

@ -57,7 +57,7 @@ public class AdminInterfaceController {
fileStatusService.validateFileIsNotDeletedAndNotApproved(fileId); fileStatusService.validateFileIsNotDeletedAndNotApproved(fileId);
fileStatusService.setStatusOcrQueued(dossierId, fileId); fileStatusService.setStatusOcrQueued(dossierId, fileId, false);
} }
@ -91,7 +91,7 @@ public class AdminInterfaceController {
if (!dryRun) { if (!dryRun) {
fileStatusService.validateFileIsNotDeletedAndNotApproved(file.getId()); fileStatusService.validateFileIsNotDeletedAndNotApproved(file.getId());
fileStatusService.setStatusOcrQueued(file.getDossierId(), file.getId()); fileStatusService.setStatusOcrQueued(file.getDossierId(), file.getId(), false);
} }
} }

View File

@ -122,7 +122,7 @@ public class FileStatusProcessingUpdateService {
} else { } else {
fileStatusService.setStatusOcrProcessing(fileId, fileStatusService.setStatusOcrProcessing(fileId,
fileEntity.getProcessingStatus().equals(ProcessingStatus.OCR_PROCESSING) ? fileEntity.getProcessingErrorCounter() + 1 : 0); fileEntity.getProcessingStatus().equals(ProcessingStatus.OCR_PROCESSING) ? fileEntity.getProcessingErrorCounter() + 1 : 0);
fileStatusService.addToOcrQueue(dossierId, fileId, 2); fileStatusService.addToOcrQueue(dossierId, fileId, 2, false);
} }
} }

View File

@ -316,7 +316,7 @@ public class FileStatusService {
} }
log.info("Add file: {} from dossier {} to OCR queue", fileId, dossierId); log.info("Add file: {} from dossier {} to OCR queue", fileId, dossierId);
setStatusOcrQueued(dossierId, fileId); setStatusOcrQueued(dossierId, fileId, false);
sendReadOnlyAnalysisEvent(dossierId, fileId, fileEntity); sendReadOnlyAnalysisEvent(dossierId, fileId, fileEntity);
return; return;
} }
@ -352,7 +352,6 @@ public class FileStatusService {
return; return;
} }
boolean forceAnalysis = false; boolean forceAnalysis = false;
if (settings.isLlmNerServiceEnabled()) { if (settings.isLlmNerServiceEnabled()) {
boolean objectExists = fileManagementStorageService.objectExists(dossierId, fileId, FileType.LLM_NER_ENTITIES); boolean objectExists = fileManagementStorageService.objectExists(dossierId, fileId, FileType.LLM_NER_ENTITIES);
@ -386,7 +385,7 @@ public class FileStatusService {
boolean reanalyse = fileModel.isReanalysisRequired() || analysisType.equals(AnalysisType.MANUAL_REDACTION_REANALYZE); boolean reanalyse = fileModel.isReanalysisRequired() || analysisType.equals(AnalysisType.MANUAL_REDACTION_REANALYZE);
MessageType messageType = calculateMessageType(reanalyse, fileModel.getProcessingStatus(), fileModel); MessageType messageType = calculateMessageType(reanalyse, fileModel.getProcessingStatus(), fileModel);
if(analysisType == AnalysisType.FORCE_ANALYSE || forceAnalysis) { if (analysisType == AnalysisType.FORCE_ANALYSE || forceAnalysis) {
messageType = MessageType.ANALYSE; messageType = MessageType.ANALYSE;
} }
@ -567,7 +566,7 @@ public class FileStatusService {
} }
public void setStatusOcrQueued(String dossierId, String fileId) { public void setStatusOcrQueued(String dossierId, String fileId, boolean allPages) {
FileEntity fileStatus = fileStatusPersistenceService.getStatus(fileId); FileEntity fileStatus = fileStatusPersistenceService.getStatus(fileId);
@ -579,7 +578,7 @@ public class FileStatusService {
updateOCRStartTime(fileId); updateOCRStartTime(fileId);
fileStatusPersistenceService.updateProcessingStatus(fileId, ProcessingStatus.OCR_PROCESSING_QUEUED); fileStatusPersistenceService.updateProcessingStatus(fileId, ProcessingStatus.OCR_PROCESSING_QUEUED);
websocketService.sendAnalysisEvent(dossierId, fileId, AnalyseStatus.OCR_PROCESSING, fileStatus.getNumberOfAnalyses() + 1); websocketService.sendAnalysisEvent(dossierId, fileId, AnalyseStatus.OCR_PROCESSING, fileStatus.getNumberOfAnalyses() + 1);
addToOcrQueue(dossierId, fileId, 2); addToOcrQueue(dossierId, fileId, 2, allPages);
} }
@ -760,13 +759,16 @@ public class FileStatusService {
} }
public void addToOcrQueue(String dossierId, String fileId, int priority) { public void addToOcrQueue(String dossierId, String fileId, int priority, boolean allPages) {
var removeWatermark = dossierTemplatePersistenceService.getDossierTemplate(dossierPersistenceService.getDossierTemplateId(dossierId)).isRemoveWatermark(); var removeWatermark = dossierTemplatePersistenceService.getDossierTemplate(dossierPersistenceService.getDossierTemplateId(dossierId)).isRemoveWatermark();
Set<AzureOcrFeature> features = new HashSet<>(); Set<AzureOcrFeature> features = new HashSet<>();
if (removeWatermark) { if (removeWatermark) {
features.add(AzureOcrFeature.REMOVE_WATERMARKS); features.add(AzureOcrFeature.REMOVE_WATERMARKS);
} }
if (allPages) {
features.add(AzureOcrFeature.ALL_PAGES);
}
if (currentApplicationTypeProvider.isDocuMine()) { if (currentApplicationTypeProvider.isDocuMine()) {
features.add(AzureOcrFeature.ROTATION_CORRECTION); features.add(AzureOcrFeature.ROTATION_CORRECTION);
features.add(AzureOcrFeature.FONT_STYLE_DETECTION); features.add(AzureOcrFeature.FONT_STYLE_DETECTION);
@ -820,7 +822,7 @@ public class FileStatusService {
fileStatusPersistenceService.updateWorkflowStatus(fileId, newWorkflowStatus, false); fileStatusPersistenceService.updateWorkflowStatus(fileId, newWorkflowStatus, false);
if(oldWorkflowStatus == WorkflowStatus.APPROVED && newWorkflowStatus != WorkflowStatus.APPROVED) { if (oldWorkflowStatus == WorkflowStatus.APPROVED && newWorkflowStatus != WorkflowStatus.APPROVED) {
fileStatusPersistenceService.clearLastDownload(fileId); fileStatusPersistenceService.clearLastDownload(fileId);
} }
} }
@ -977,7 +979,7 @@ public class FileStatusService {
if (runOcr) { if (runOcr) {
fileStatusPersistenceService.resetOcrStartAndEndDate(fileId); fileStatusPersistenceService.resetOcrStartAndEndDate(fileId);
setStatusOcrQueued(dossierId, fileId); setStatusOcrQueued(dossierId, fileId, false);
return; return;
} }
@ -1064,6 +1066,7 @@ public class FileStatusService {
addToAnalysisQueue(dossierId, fileId, priority, Sets.newHashSet(), AnalysisType.DEFAULT); addToAnalysisQueue(dossierId, fileId, priority, Sets.newHashSet(), AnalysisType.DEFAULT);
} }
@Transactional @Transactional
public void setStatusForceAnalyse(String dossierId, String fileId, boolean priority) { public void setStatusForceAnalyse(String dossierId, String fileId, boolean priority) {

View File

@ -10,7 +10,6 @@ import com.iqser.red.service.persistence.management.v1.processor.exception.Confl
import com.iqser.red.service.persistence.management.v1.processor.exception.NotFoundException; import com.iqser.red.service.persistence.management.v1.processor.exception.NotFoundException;
import com.iqser.red.service.persistence.management.v1.processor.service.persistence.DossierPersistenceService; import com.iqser.red.service.persistence.management.v1.processor.service.persistence.DossierPersistenceService;
import com.iqser.red.service.persistence.service.v1.api.shared.model.ReanalysisSettings; import com.iqser.red.service.persistence.service.v1.api.shared.model.ReanalysisSettings;
import com.iqser.red.service.persistence.service.v1.api.shared.model.analysislog.entitylog.imported.ImportedRedactions;
import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.DeleteImportedRedactionsRequest; import com.iqser.red.service.persistence.service.v1.api.shared.model.annotations.DeleteImportedRedactionsRequest;
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.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.FileType;
@ -178,11 +177,11 @@ public class ReanalysisService {
relevantFiles.stream() relevantFiles.stream()
.filter(fileStatus -> fileStatus.getOcrStartTime() == null) .filter(fileStatus -> fileStatus.getOcrStartTime() == null)
.filter(fileStatus -> fileStatus.getProcessingStatus().equals(ProcessingStatus.PROCESSED)) .filter(fileStatus -> fileStatus.getProcessingStatus().equals(ProcessingStatus.PROCESSED))
.forEach(fileStatus -> fileStatusService.setStatusOcrQueued(dossierId, fileStatus.getId())); .forEach(fileStatus -> fileStatusService.setStatusOcrQueued(dossierId, fileStatus.getId(), false));
} }
public void ocrFile(String dossierId, String fileId, boolean force) { public void ocrFile(String dossierId, String fileId, boolean force, boolean allPages) {
dossierPersistenceService.getAndValidateDossier(dossierId); dossierPersistenceService.getAndValidateDossier(dossierId);
FileModel dossierFile = fileStatusService.getStatus(fileId); FileModel dossierFile = fileStatusService.getStatus(fileId);
@ -202,18 +201,18 @@ public class ReanalysisService {
} }
if (force) { if (force) {
fileStatusService.setStatusOcrQueued(dossierId, fileId); fileStatusService.setStatusOcrQueued(dossierId, fileId, allPages);
} else { } else {
if (dossierFile.getOcrStartTime() != null) { if (dossierFile.getOcrStartTime() != null) {
throw new ConflictException("File already has been OCR processed"); throw new ConflictException("File already has been OCR processed");
} }
ocrFiles(dossierId, Sets.newHashSet(fileId)); ocrFiles(dossierId, Sets.newHashSet(fileId), allPages);
} }
} }
public void ocrFiles(String dossierId, Set<String> fileIds) { public void ocrFiles(String dossierId, Set<String> fileIds, boolean allPages) {
var relevantFiles = getRelevantFiles(dossierId, fileIds); var relevantFiles = getRelevantFiles(dossierId, fileIds);
@ -225,7 +224,7 @@ public class ReanalysisService {
relevantFiles.stream() relevantFiles.stream()
.filter(fileStatus -> fileStatus.getOcrStartTime() == null) .filter(fileStatus -> fileStatus.getOcrStartTime() == null)
.forEach(fileStatus -> fileStatusService.setStatusOcrQueued(dossierId, fileStatus.getId())); .forEach(fileStatus -> fileStatusService.setStatusOcrQueued(dossierId, fileStatus.getId(), allPages));
} }

View File

@ -56,7 +56,7 @@ public class ReanalysisTest extends AbstractPersistenceServerServiceTest {
assertThat(loadedFile.getProcessingStatus()).isEqualTo(ProcessingStatus.OCR_PROCESSING_QUEUED); assertThat(loadedFile.getProcessingStatus()).isEqualTo(ProcessingStatus.OCR_PROCESSING_QUEUED);
resetProcessingStatus(file); resetProcessingStatus(file);
reanalysisClient.ocrFile(dossier.getId(), file.getId(), true); reanalysisClient.ocrFile(dossier.getId(), file.getId(), true, false);
loadedFile = fileClient.getFileStatus(dossier.getId(), file.getId()); loadedFile = fileClient.getFileStatus(dossier.getId(), file.getId());
assertThat(loadedFile.getProcessingStatus()).isEqualTo(ProcessingStatus.OCR_PROCESSING_QUEUED); assertThat(loadedFile.getProcessingStatus()).isEqualTo(ProcessingStatus.OCR_PROCESSING_QUEUED);
resetProcessingStatus(file); resetProcessingStatus(file);