Search in sources :

Example 1 with ExportedProject

use of de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject in project webanno by webanno.

the class AnnotationDocumentExporter method exportAnnotationDocumentContents.

private void exportAnnotationDocumentContents(ProjectExportRequest aRequest, ProjectExportTaskMonitor aMonitor, ExportedProject aExProject, File aStage) throws UIMAException, ClassNotFoundException, IOException {
    Project project = aRequest.getProject();
    // The export process may store project-related information in this context to ensure it
    // is looked up only once during the bulk operation and the DB is not hit too often.
    Map<Pair<Project, String>, Object> bulkOperationContext = new HashMap<>();
    List<SourceDocument> documents = documentService.listSourceDocuments(project);
    int i = 1;
    int initProgress = aMonitor.getProgress();
    // Create a map containing the annotation documents for each source document. Doing this
    // as one DB access before the main processing to avoid hammering the DB in the loops
    // below.
    Map<SourceDocument, List<AnnotationDocument>> srcToAnnIdx = documentService.listAnnotationDocuments(project).stream().collect(groupingBy(doc -> doc.getDocument(), toList()));
    // Cache user lookups to avoid constantly hitting the database
    LoadingCache<String, User> usersCache = Caffeine.newBuilder().build(key -> userRepository.get(key));
    for (SourceDocument srcDoc : documents) {
        try (CasStorageSession session = CasStorageSession.openNested()) {
            // If the initial CAS does not exist yet, it must be created before export.
            if (!documentService.existsInitialCas(srcDoc)) {
                documentService.createOrReadInitialCas(srcDoc);
            }
            File targetDir = new File(aStage, ANNOTATION_CAS_FOLDER + srcDoc.getName());
            forceMkdir(targetDir);
            File initialCasFile = documentService.getCasFile(srcDoc, INITIAL_CAS_PSEUDO_USER);
            copyFileToDirectory(initialCasFile, targetDir);
            log.info("Exported annotation document content for user [" + INITIAL_CAS_PSEUDO_USER + "] for source document [" + srcDoc.getId() + "] in project [" + project.getName() + "] with id [" + project.getId() + "]");
            // 
            // Export per-user annotation document
            // 
            // Determine which format to use for export
            String formatId = FORMAT_AUTO.equals(aRequest.getFormat()) ? srcDoc.getFormat() : aRequest.getFormat();
            FormatSupport format = importExportService.getWritableFormatById(formatId).orElseGet(() -> {
                FormatSupport fallbackFormat = new WebAnnoTsv3FormatSupport();
                aMonitor.addMessage(LogMessage.warn(this, "Annotation: [%s] No writer " + "found for original format [%s] - exporting as [%s] " + "instead.", srcDoc.getName(), formatId, fallbackFormat.getName()));
                return fallbackFormat;
            });
            // Export annotations from regular users
            for (AnnotationDocument annDoc : srcToAnnIdx.computeIfAbsent(srcDoc, key -> emptyList())) {
                // annotation document is not NEW/IGNORE
                if (usersCache.get(annDoc.getUser()) != null && !annDoc.getState().equals(AnnotationDocumentState.NEW) && !annDoc.getState().equals(AnnotationDocumentState.IGNORE)) {
                    File annSerDir = new File(aStage.getAbsolutePath() + ANNOTATION_CAS_FOLDER + srcDoc.getName());
                    File annDocDir = new File(aStage.getAbsolutePath() + ANNOTATION_ORIGINAL_FOLDER + srcDoc.getName());
                    forceMkdir(annSerDir);
                    forceMkdir(annDocDir);
                    File annSerFile = documentService.getCasFile(srcDoc, annDoc.getUser());
                    File annFile = null;
                    if (annSerFile.exists()) {
                        annFile = importExportService.exportAnnotationDocument(srcDoc, annDoc.getUser(), format, annDoc.getUser(), ANNOTATION, false, bulkOperationContext);
                    }
                    if (annSerFile.exists()) {
                        copyFileToDirectory(annSerFile, annSerDir);
                        copyFileToDirectory(annFile, annDocDir);
                        forceDelete(annFile);
                    }
                    log.info("Exported annotation document content for user [" + annDoc.getUser() + "] for source document [" + srcDoc.getId() + "] in project [" + project.getName() + "] with id [" + project.getId() + "]");
                }
            }
            // correction type projects.
            if (PROJECT_TYPE_AUTOMATION.equals(project.getMode()) || PROJECT_TYPE_CORRECTION.equals(project.getMode())) {
                File corrSerFile = documentService.getCasFile(srcDoc, CORRECTION_USER);
                if (corrSerFile.exists()) {
                    // Copy CAS - this is used when importing the project again
                    // Util WebAnno 3.4.x, the CORRECTION_USER CAS was exported to 'curation'
                    // and 'curation_ser'.
                    // Since WebAnno 3.5.x, the CORRECTION_USER CAS is exported to 'annotation'
                    // and 'annotation_ser'.
                    File curationSerDir = new File(aStage + ANNOTATION_AS_SERIALISED_CAS + srcDoc.getName());
                    forceMkdir(curationSerDir);
                    copyFileToDirectory(corrSerFile, curationSerDir);
                    // Copy secondary export format for convenience - not used during import
                    File curationDir = new File(aStage + ANNOTATION_ORIGINAL_FOLDER + srcDoc.getName());
                    forceMkdir(curationDir);
                    File corrFile = importExportService.exportAnnotationDocument(srcDoc, CORRECTION_USER, format, CORRECTION_USER, CORRECTION);
                    copyFileToDirectory(corrFile, curationDir);
                    forceDelete(corrFile);
                }
            }
        }
        aMonitor.setProgress(initProgress + (int) ceil(((double) i) / documents.size() * 80.0));
        i++;
    }
}
Also used : Enumeration(java.util.Enumeration) PROJECT_TYPE_AUTOMATION(de.tudarmstadt.ukp.clarin.webanno.api.WebAnnoConst.PROJECT_TYPE_AUTOMATION) LoggerFactory(org.slf4j.LoggerFactory) Autowired(org.springframework.beans.factory.annotation.Autowired) ProjectExportTaskMonitor(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportTaskMonitor) ANNOTATION(de.tudarmstadt.ukp.clarin.webanno.model.Mode.ANNOTATION) RepositoryProperties(de.tudarmstadt.ukp.clarin.webanno.api.RepositoryProperties) FastIOUtils.copy(de.tudarmstadt.ukp.clarin.webanno.support.io.FastIOUtils.copy) Collectors.toMap(java.util.stream.Collectors.toMap) Pair(org.apache.commons.lang3.tuple.Pair) Arrays.asList(java.util.Arrays.asList) Map(java.util.Map) FileUtils.forceMkdir(org.apache.commons.io.FileUtils.forceMkdir) ZipFile(java.util.zip.ZipFile) WebAnnoTsv3FormatSupport(de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport) FORMAT_AUTO(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportRequest.FORMAT_AUTO) Path(java.nio.file.Path) ZipEntry(java.util.zip.ZipEntry) UIMAException(org.apache.uima.UIMAException) CasStorageSession(de.tudarmstadt.ukp.clarin.webanno.api.dao.casstorage.CasStorageSession) PROJECT_TYPE_CORRECTION(de.tudarmstadt.ukp.clarin.webanno.api.WebAnnoConst.PROJECT_TYPE_CORRECTION) LoadingCache(com.github.benmanes.caffeine.cache.LoadingCache) Collections.emptyList(java.util.Collections.emptyList) PROJECT_FOLDER(de.tudarmstadt.ukp.clarin.webanno.api.ProjectService.PROJECT_FOLDER) UserDao(de.tudarmstadt.ukp.clarin.webanno.security.UserDao) Set(java.util.Set) AnnotationDocument(de.tudarmstadt.ukp.clarin.webanno.model.AnnotationDocument) Math.ceil(java.lang.Math.ceil) AnnotationDocumentState(de.tudarmstadt.ukp.clarin.webanno.model.AnnotationDocumentState) FileUtils.copyFileToDirectory(org.apache.commons.io.FileUtils.copyFileToDirectory) ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) CORRECTION(de.tudarmstadt.ukp.clarin.webanno.model.Mode.CORRECTION) List(java.util.List) SourceDocument(de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument) Function.identity(java.util.function.Function.identity) FilenameUtils(org.apache.commons.io.FilenameUtils) CORRECTION_USER(de.tudarmstadt.ukp.clarin.webanno.api.WebAnnoConst.CORRECTION_USER) System.currentTimeMillis(java.lang.System.currentTimeMillis) FormatSupport(de.tudarmstadt.ukp.clarin.webanno.api.format.FormatSupport) LogMessage(de.tudarmstadt.ukp.clarin.webanno.support.logging.LogMessage) Collectors.groupingBy(java.util.stream.Collectors.groupingBy) HashMap(java.util.HashMap) FileUtils.forceDelete(org.apache.commons.io.FileUtils.forceDelete) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) User(de.tudarmstadt.ukp.clarin.webanno.security.model.User) ProjectExporter(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExporter) Project(de.tudarmstadt.ukp.clarin.webanno.model.Project) INITIAL_CAS_PSEUDO_USER(de.tudarmstadt.ukp.clarin.webanno.api.WebAnnoConst.INITIAL_CAS_PSEUDO_USER) ProjectExportRequest(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportRequest) Caffeine(com.github.benmanes.caffeine.cache.Caffeine) DocumentService(de.tudarmstadt.ukp.clarin.webanno.api.DocumentService) Logger(org.slf4j.Logger) Files(java.nio.file.Files) IOException(java.io.IOException) ANNOTATION_FOLDER(de.tudarmstadt.ukp.clarin.webanno.api.ProjectService.ANNOTATION_FOLDER) File(java.io.File) Collectors.toList(java.util.stream.Collectors.toList) Component(org.springframework.stereotype.Component) ExportedAnnotationDocument(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedAnnotationDocument) DurationFormatUtils(org.apache.commons.lang3.time.DurationFormatUtils) DOCUMENT_FOLDER(de.tudarmstadt.ukp.clarin.webanno.api.ProjectService.DOCUMENT_FOLDER) ProjectImportRequest(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectImportRequest) ImportExportService(de.tudarmstadt.ukp.clarin.webanno.api.ImportExportService) WebAnnoTsv3FormatSupport(de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport) FormatSupport(de.tudarmstadt.ukp.clarin.webanno.api.format.FormatSupport) User(de.tudarmstadt.ukp.clarin.webanno.security.model.User) HashMap(java.util.HashMap) SourceDocument(de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument) WebAnnoTsv3FormatSupport(de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport) AnnotationDocument(de.tudarmstadt.ukp.clarin.webanno.model.AnnotationDocument) ExportedAnnotationDocument(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedAnnotationDocument) ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) Project(de.tudarmstadt.ukp.clarin.webanno.model.Project) Arrays.asList(java.util.Arrays.asList) Collections.emptyList(java.util.Collections.emptyList) List(java.util.List) ArrayList(java.util.ArrayList) Collectors.toList(java.util.stream.Collectors.toList) CasStorageSession(de.tudarmstadt.ukp.clarin.webanno.api.dao.casstorage.CasStorageSession) ZipFile(java.util.zip.ZipFile) File(java.io.File) Pair(org.apache.commons.lang3.tuple.Pair)

Example 2 with ExportedProject

use of de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject in project webanno by webanno.

the class ProjectSettingsExporter method exportData.

@Override
public void exportData(ProjectExportRequest aRequest, ProjectExportTaskMonitor aMonitor, ExportedProject aExProject, File aStage) throws Exception {
    Project project = aRequest.getProject();
    aExProject.setDescription(project.getDescription());
    // In older versions of WebAnno, the mode was an enum which was serialized as upper-case
    // during export but as lower-case in the database. This is compensating for this case.
    aExProject.setMode(StringUtils.upperCase(project.getMode(), Locale.US));
    aExProject.setScriptDirection(project.getScriptDirection());
    aExProject.setVersion(project.getVersion());
    aExProject.setDisableExport(project.isDisableExport());
    aExProject.setCreated(project.getCreated());
    aExProject.setUpdated(project.getUpdated());
    aExProject.setAnonymousCuration(project.isAnonymousCuration());
}
Also used : ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) Project(de.tudarmstadt.ukp.clarin.webanno.model.Project)

Example 3 with ExportedProject

use of de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject in project webanno by webanno.

the class CuratedDocumentsExporterTest method runImportAndFetchDocuments.

private List<Pair<SourceDocument, String>> runImportAndFetchDocuments(ZipFile aZipFile) throws Exception {
    ArgumentCaptor<SourceDocument> sourceDocCaptor = ArgumentCaptor.forClass(SourceDocument.class);
    ArgumentCaptor<String> usernameCaptor = ArgumentCaptor.forClass(String.class);
    // Import the project again
    ExportedProject exProject = ProjectExportServiceImpl.loadExportedProject(aZipFile);
    ProjectImportRequest importRequest = new ProjectImportRequest(true);
    sut.importData(importRequest, project, exProject, aZipFile);
    verify(casStorageService, atLeastOnce()).getCasFile(sourceDocCaptor.capture(), usernameCaptor.capture());
    List<Pair<SourceDocument, String>> importedCases = new ArrayList<>();
    List<SourceDocument> docs = sourceDocCaptor.getAllValues();
    List<String> users = usernameCaptor.getAllValues();
    for (int i = 0; i < docs.size(); i++) {
        importedCases.add(Pair.of(docs.get(i), users.get(i)));
    }
    return importedCases;
}
Also used : ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) SourceDocument(de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument) ArrayList(java.util.ArrayList) ProjectImportRequest(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectImportRequest) Pair(org.apache.commons.lang3.tuple.Pair)

Example 4 with ExportedProject

use of de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject in project webanno by webanno.

the class LayerExporterTest method runExportImportAndFetchLayers.

private ArgumentCaptor<AnnotationLayer> runExportImportAndFetchLayers() throws Exception {
    // Export the project
    ProjectExportRequest exportRequest = new ProjectExportRequest();
    ProjectExportTaskMonitor monitor = new ProjectExportTaskMonitor();
    exportRequest.setProject(project);
    ExportedProject exportedProject = new ExportedProject();
    sut.exportData(exportRequest, monitor, exportedProject, workFolder);
    // Import the project again
    ArgumentCaptor<AnnotationLayer> captor = ArgumentCaptor.forClass(AnnotationLayer.class);
    doNothing().when(annotationService).createOrUpdateLayer(captor.capture());
    ProjectImportRequest importRequest = new ProjectImportRequest(true);
    ZipFile zipFile = mock(ZipFile.class);
    sut.importData(importRequest, project, exportedProject, zipFile);
    return captor;
}
Also used : ProjectExportTaskMonitor(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportTaskMonitor) ZipFile(java.util.zip.ZipFile) ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) ProjectExportRequest(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportRequest) AnnotationLayer(de.tudarmstadt.ukp.clarin.webanno.model.AnnotationLayer) ProjectImportRequest(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectImportRequest)

Example 5 with ExportedProject

use of de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject in project webanno by webanno.

the class CuratedDocumentsExporter method exportData.

/**
 * Copy, if exists, curation documents to a folder that will be exported as Zip file
 *
 * @param aStage
 *            The folder where curated documents are copied to be exported as Zip File
 */
@Override
public void exportData(ProjectExportRequest aRequest, ProjectExportTaskMonitor aMonitor, ExportedProject aExProject, File aStage) throws Exception {
    Project project = aRequest.getProject();
    // The export process may store project-related information in this context to ensure it
    // is looked up only once during the bulk operation and the DB is not hit too often.
    Map<Pair<Project, String>, Object> bulkOperationContext = new HashMap<>();
    // Get all the source documents from the project
    List<SourceDocument> documents = documentService.listSourceDocuments(project);
    int initProgress = aMonitor.getProgress() - 1;
    int i = 1;
    for (SourceDocument sourceDocument : documents) {
        File curationCasDir = new File(aStage, CURATION_CAS_FOLDER + sourceDocument.getName());
        forceMkdir(curationCasDir);
        File curationDir = new File(aStage, CURATION_FOLDER + sourceDocument.getName());
        forceMkdir(curationDir);
        // finished or also the ones that are in progress
        if ((aRequest.isIncludeInProgress() && CURATION_IN_PROGRESS.equals(sourceDocument.getState())) || CURATION_FINISHED.equals(sourceDocument.getState())) {
            File curationCasFile = documentService.getCasFile(sourceDocument, CURATION_USER);
            if (curationCasFile.exists()) {
                // Copy CAS - this is used when importing the project again
                copyFileToDirectory(curationCasFile, curationCasDir);
                // Determine which format to use for export
                String formatId = FORMAT_AUTO.equals(aRequest.getFormat()) ? sourceDocument.getFormat() : aRequest.getFormat();
                FormatSupport format = importExportService.getWritableFormatById(formatId).orElseGet(() -> {
                    FormatSupport fallbackFormat = new WebAnnoTsv3FormatSupport();
                    aMonitor.addMessage(LogMessage.warn(this, "Curation: [%s] No writer" + " found for original format [%s] - exporting as [%s] " + "instead.", sourceDocument.getName(), formatId, fallbackFormat.getName()));
                    return fallbackFormat;
                });
                // Copy secondary export format for convenience - not used during import
                try {
                    File curationFile = importExportService.exportAnnotationDocument(sourceDocument, CURATION_USER, format, CURATION_USER, CURATION, true, bulkOperationContext);
                    copyFileToDirectory(curationFile, curationDir);
                    forceDelete(curationFile);
                } catch (Exception e) {
                    // ExceptionUtils.getRootCauseMessage(e) );
                    throw new ProjectExportException("Aborting due to unrecoverable error while exporting!");
                }
            }
        }
        aMonitor.setProgress(initProgress + (int) ceil(((double) i) / documents.size() * 10.0));
        i++;
    }
}
Also used : FormatSupport(de.tudarmstadt.ukp.clarin.webanno.api.format.FormatSupport) WebAnnoTsv3FormatSupport(de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport) HashMap(java.util.HashMap) SourceDocument(de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument) WebAnnoTsv3FormatSupport(de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport) ProjectExportException(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportException) IOException(java.io.IOException) ProjectExportException(de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportException) Project(de.tudarmstadt.ukp.clarin.webanno.model.Project) ExportedProject(de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject) ZipFile(java.util.zip.ZipFile) File(java.io.File) Pair(org.apache.commons.lang3.tuple.Pair)

Aggregations

ExportedProject (de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedProject)14 Project (de.tudarmstadt.ukp.clarin.webanno.model.Project)8 ZipFile (java.util.zip.ZipFile)8 File (java.io.File)7 ProjectExportException (de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportException)5 SourceDocument (de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument)5 IOException (java.io.IOException)5 ProjectImportRequest (de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectImportRequest)4 ArrayList (java.util.ArrayList)4 Pair (org.apache.commons.lang3.tuple.Pair)4 ProjectExporter (de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExporter)3 ProjectExportRequest (de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportRequest)2 ProjectExportTaskMonitor (de.tudarmstadt.ukp.clarin.webanno.api.export.ProjectExportTaskMonitor)2 FormatSupport (de.tudarmstadt.ukp.clarin.webanno.api.format.FormatSupport)2 ExportedSourceDocument (de.tudarmstadt.ukp.clarin.webanno.export.model.ExportedSourceDocument)2 User (de.tudarmstadt.ukp.clarin.webanno.security.model.User)2 WebAnnoTsv3FormatSupport (de.tudarmstadt.ukp.clarin.webanno.tsv.WebAnnoTsv3FormatSupport)2 FileNotFoundException (java.io.FileNotFoundException)2 HashMap (java.util.HashMap)2 HashSet (java.util.HashSet)2