Search in sources :

Example 1 with DataSet

use of com.thinkbiganalytics.kylo.catalog.rest.model.DataSet in project kylo by Teradata.

the class DataSetIT method testUpload.

/**
 * Verify uploading files.
 */
// TODO create local-files data source
@Ignore
@Test
public void testUpload() {
    // Create a data set
    final DataSet request = new DataSet();
    request.setDataSource(new DataSource());
    request.getDataSource().setId("local-files");
    final DataSet dataSet = given(DataSetController.BASE).when().body(request).post().then().statusCode(200).extract().as(DataSet.class);
    final String id = dataSet.getId();
    // Test empty data set
    List<DataSetFile> files = listUploads(id);
    Assert.assertEquals(0, files.size());
    // Upload sample files
    uploadFile(id, getSampleFile("userdata1.csv"));
    uploadFile(id, getSampleFile("userdata2.csv"));
    files = listUploads(id);
    Assert.assertThat(files, CoreMatchers.hasItem(nameEquals("userdata1.csv")));
    Assert.assertThat(files, CoreMatchers.hasItem(nameEquals("userdata2.csv")));
    Assert.assertEquals(2, files.size());
    // Delete a file
    given(DataSetController.BASE).when().delete(id + "/uploads/userdata1.csv").then().statusCode(204);
    files = listUploads(id);
    Assert.assertThat(files, CoreMatchers.hasItem(nameEquals("userdata2.csv")));
    Assert.assertEquals(1, files.size());
}
Also used : DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) DataSetFile(com.thinkbiganalytics.kylo.catalog.rest.model.DataSetFile) DataSource(com.thinkbiganalytics.kylo.catalog.rest.model.DataSource) Ignore(org.junit.Ignore) Test(org.junit.Test)

Example 2 with DataSet

use of com.thinkbiganalytics.kylo.catalog.rest.model.DataSet in project kylo by Teradata.

the class SparkShellProxyController method addDataSourceInformation.

private DataSet addDataSourceInformation(@Nonnull DataSet dataSet) {
    DataSet fetchedDataSet = fetchDataSet(dataSet.getId());
    DataSetTemplate template = DataSetUtil.mergeTemplates(fetchedDataSet);
    fetchedDataSet.getDataSource().setTemplate(template);
    return new DataSet(fetchedDataSet);
}
Also used : DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) DataSetTemplate(com.thinkbiganalytics.kylo.catalog.rest.model.DataSetTemplate) DefaultDataSetTemplate(com.thinkbiganalytics.kylo.catalog.rest.model.DefaultDataSetTemplate)

Example 3 with DataSet

use of com.thinkbiganalytics.kylo.catalog.rest.model.DataSet in project kylo by Teradata.

the class FeedModelTransform method domainToFeedMetadata.

/**
 * Transforms the specified Metadata feed to a Feed Manager feed.
 *
 * @param domain       the Metadata feed
 * @param userFieldMap cache map from category to user-defined fields, or {@code null}
 * @return the Feed Manager feed
 */
@Nonnull
private FeedMetadata domainToFeedMetadata(@Nonnull final Feed domain, @Nullable final Map<Category, Set<UserFieldDescriptor>> userFieldMap, @Nullable ActionGroup actionGroup) {
    FeedMetadata feed = deserializeFeedMetadata(domain, false);
    feed.setId(domain.getId().toString());
    feed.setFeedId(domain.getId().toString());
    feed.setFeedName(domain.getDisplayName());
    feed.setSystemFeedName(domain.getName());
    feed.setDescription(domain.getDescription());
    feed.setAllowIndexing(domain.isAllowIndexing());
    feed.setHistoryReindexingStatus(domain.getCurrentHistoryReindexingStatus().getHistoryReindexingState().toString());
    feed.setOwner(domain.getOwner() != null ? new User(domain.getOwner().getName()) : null);
    if (domain.getCreatedTime() != null) {
        feed.setCreateDate(domain.getCreatedTime().toDate());
    }
    if (domain.getModifiedTime() != null) {
        feed.setUpdateDate(domain.getModifiedTime().toDate());
    }
    setFeedMetadataRegisteredTemplate(domain, feed);
    Category category = domain.getCategory();
    if (category != null) {
        feed.setCategory(categoryModelTransform.domainToFeedCategorySimple(category));
    }
    feed.setState(domain.getState() != null ? domain.getState().name() : null);
    // Set user-defined properties
    final Set<UserFieldDescriptor> userFields;
    if (userFieldMap == null) {
        userFields = getUserFields(category);
    } else if (userFieldMap.containsKey(category)) {
        userFields = userFieldMap.get(category);
    } else {
        userFields = getUserFields(category);
        userFieldMap.put(category, userFields);
    }
    @SuppressWarnings("unchecked") final Set<UserProperty> userProperties = UserPropertyTransform.toUserProperties(domain.getUserProperties(), userFields);
    feed.setUserProperties(userProperties);
    // Convert JCR securitygroup to DTO
    List<com.thinkbiganalytics.feedmgr.rest.model.HadoopSecurityGroup> restSecurityGroups = new ArrayList<>();
    if (domain.getSecurityGroups() != null && domain.getSecurityGroups().size() > 0) {
        for (Object group : domain.getSecurityGroups()) {
            HadoopSecurityGroup hadoopSecurityGroup = (HadoopSecurityGroup) group;
            com.thinkbiganalytics.feedmgr.rest.model.HadoopSecurityGroup restSecurityGroup = new com.thinkbiganalytics.feedmgr.rest.model.HadoopSecurityGroup();
            restSecurityGroup.setDescription(hadoopSecurityGroup.getDescription());
            restSecurityGroup.setId(hadoopSecurityGroup.getGroupId());
            restSecurityGroup.setName(hadoopSecurityGroup.getName());
            restSecurityGroups.add(restSecurityGroup);
        }
    }
    feed.setSecurityGroups(restSecurityGroups);
    feed.setTags(domain.getTags().stream().map(name -> new DefaultTag(name)).collect(Collectors.toList()));
    if (domain.getUsedByFeeds() != null) {
        final List<FeedSummary> usedByFeeds = domain.getUsedByFeeds().stream().map(this::domainToFeedSummary).collect(Collectors.toList());
        feed.setUsedByFeeds(usedByFeeds);
    }
    List<DataSet> srcDataSets = domain.getSources().stream().map(FeedConnection::getDataSet).filter(Optional::isPresent).map(Optional::get).map(dataSet -> {
        try {
            return catalogModelTransform.dataSetToRestModel().apply(dataSet);
        } catch (final AccessControlException e) {
            log.debug("Denied access to data set: {}: {}", domain.getId(), e);
            final DataSource dataSource = new DataSource();
            dataSource.setAllowedActions(new ActionGroup());
            dataSource.getAllowedActions().setActions(Collections.emptyList());
            final DataSet model = new DataSet();
            model.setId(domain.getId().toString());
            model.setDataSource(dataSource);
            return model;
        }
    }).collect(Collectors.toList());
    feed.setSourceDataSets(srcDataSets);
    if (actionGroup == null) {
        // add in access control items
        securityTransform.applyAccessControl(domain, feed);
    } else {
        feed.setAllowedActions(actionGroup);
    }
    return feed;
}
Also used : Category(com.thinkbiganalytics.metadata.api.category.Category) DataSourceProvider(com.thinkbiganalytics.metadata.api.catalog.DataSourceProvider) LoggerFactory(org.slf4j.LoggerFactory) StringUtils(org.apache.commons.lang3.StringUtils) ObjectReader(com.fasterxml.jackson.databind.ObjectReader) ObjectMapperSerializer(com.thinkbiganalytics.json.ObjectMapperSerializer) FeedProvider(com.thinkbiganalytics.metadata.api.feed.FeedProvider) DataSetProvider(com.thinkbiganalytics.metadata.api.catalog.DataSetProvider) HiveService(com.thinkbiganalytics.hive.service.HiveService) DataSource(com.thinkbiganalytics.kylo.catalog.rest.model.DataSource) Map(java.util.Map) AccessController(com.thinkbiganalytics.security.AccessController) JsonNode(com.fasterxml.jackson.databind.JsonNode) Pageable(org.springframework.data.domain.Pageable) Tag(com.thinkbiganalytics.discovery.schema.Tag) CatalogModelTransform(com.thinkbiganalytics.kylo.catalog.rest.model.CatalogModelTransform) FeedManagerTemplateProvider(com.thinkbiganalytics.metadata.api.template.FeedManagerTemplateProvider) DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) FeedVersions(com.thinkbiganalytics.feedmgr.rest.model.FeedVersions) ObjectWriter(com.fasterxml.jackson.databind.ObjectWriter) Collection(java.util.Collection) ActionGroup(com.thinkbiganalytics.security.rest.model.ActionGroup) Set(java.util.Set) DefaultTag(com.thinkbiganalytics.discovery.model.DefaultTag) Page(org.springframework.data.domain.Page) EntityVersionDifference(com.thinkbiganalytics.feedmgr.rest.model.EntityVersionDifference) Collectors(java.util.stream.Collectors) Sets(com.google.common.collect.Sets) FeedSummary(com.thinkbiganalytics.feedmgr.rest.model.FeedSummary) List(java.util.List) CategoryProvider(com.thinkbiganalytics.metadata.api.category.CategoryProvider) AccessControlException(java.security.AccessControlException) Optional(java.util.Optional) PageImpl(org.springframework.data.domain.PageImpl) JsonDiff(com.flipkart.zjsonpatch.JsonDiff) SecurityModelTransform(com.thinkbiganalytics.security.rest.controller.SecurityModelTransform) SearchResultImpl(com.thinkbiganalytics.rest.model.search.SearchResultImpl) Feed(com.thinkbiganalytics.metadata.api.feed.Feed) HistoryReindexingStatus(com.thinkbiganalytics.metadata.api.feed.reindex.HistoryReindexingStatus) FeedMetadata(com.thinkbiganalytics.feedmgr.rest.model.FeedMetadata) EntityVersion(com.thinkbiganalytics.metadata.api.versioning.EntityVersion) UserProperty(com.thinkbiganalytics.feedmgr.rest.model.UserProperty) ArrayList(java.util.ArrayList) FeedCategory(com.thinkbiganalytics.feedmgr.rest.model.FeedCategory) Inject(javax.inject.Inject) FeedConnection(com.thinkbiganalytics.metadata.api.feed.FeedConnection) UIFeed(com.thinkbiganalytics.feedmgr.rest.model.UIFeed) TemplateModelTransform(com.thinkbiganalytics.feedmgr.service.template.TemplateModelTransform) User(com.thinkbiganalytics.security.rest.model.User) RegisteredTemplate(com.thinkbiganalytics.feedmgr.rest.model.RegisteredTemplate) Nonnull(javax.annotation.Nonnull) Nullable(javax.annotation.Nullable) Logger(org.slf4j.Logger) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) IOException(java.io.IOException) UserFieldDescriptor(com.thinkbiganalytics.metadata.api.extension.UserFieldDescriptor) Maps(com.google.common.collect.Maps) SearchResult(com.thinkbiganalytics.rest.model.search.SearchResult) FeedManagerTemplate(com.thinkbiganalytics.metadata.api.template.FeedManagerTemplate) HadoopSecurityGroup(com.thinkbiganalytics.metadata.api.security.HadoopSecurityGroup) EncryptionService(com.thinkbiganalytics.security.core.encrypt.EncryptionService) CategoryModelTransform(com.thinkbiganalytics.feedmgr.service.category.CategoryModelTransform) UserPropertyTransform(com.thinkbiganalytics.feedmgr.service.UserPropertyTransform) DatasourceAccessControl(com.thinkbiganalytics.metadata.api.datasource.security.DatasourceAccessControl) JcrHadoopSecurityGroup(com.thinkbiganalytics.metadata.modeshape.security.JcrHadoopSecurityGroup) HadoopSecurityGroupProvider(com.thinkbiganalytics.metadata.api.security.HadoopSecurityGroupProvider) Collections(java.util.Collections) User(com.thinkbiganalytics.security.rest.model.User) Category(com.thinkbiganalytics.metadata.api.category.Category) FeedCategory(com.thinkbiganalytics.feedmgr.rest.model.FeedCategory) DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) FeedMetadata(com.thinkbiganalytics.feedmgr.rest.model.FeedMetadata) ArrayList(java.util.ArrayList) HadoopSecurityGroup(com.thinkbiganalytics.metadata.api.security.HadoopSecurityGroup) JcrHadoopSecurityGroup(com.thinkbiganalytics.metadata.modeshape.security.JcrHadoopSecurityGroup) Optional(java.util.Optional) UserFieldDescriptor(com.thinkbiganalytics.metadata.api.extension.UserFieldDescriptor) FeedConnection(com.thinkbiganalytics.metadata.api.feed.FeedConnection) AccessControlException(java.security.AccessControlException) DataSource(com.thinkbiganalytics.kylo.catalog.rest.model.DataSource) UserProperty(com.thinkbiganalytics.feedmgr.rest.model.UserProperty) ActionGroup(com.thinkbiganalytics.security.rest.model.ActionGroup) DefaultTag(com.thinkbiganalytics.discovery.model.DefaultTag) FeedSummary(com.thinkbiganalytics.feedmgr.rest.model.FeedSummary) Nonnull(javax.annotation.Nonnull)

Example 4 with DataSet

use of com.thinkbiganalytics.kylo.catalog.rest.model.DataSet in project kylo by Teradata.

the class App method catalogDataSetProvider.

@Bean
public CatalogDataSetProvider catalogDataSetProvider(final CatalogDataSetProviderFactory catalogDataSetProviderFactory, @Named("decryptionModule") final Module decryptionModule) throws IOException {
    final List<DataSet> dataSets;
    final String env = System.getenv("DATASETS");
    if (env != null) {
        dataSets = new ObjectMapper().registerModule(decryptionModule).readValue(env, TypeFactory.defaultInstance().constructCollectionType(List.class, DataSet.class));
    } else {
        dataSets = Collections.emptyList();
    }
    return catalogDataSetProviderFactory.getDataSetProvider(dataSets);
}
Also used : DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Bean(org.springframework.context.annotation.Bean)

Example 5 with DataSet

use of com.thinkbiganalytics.kylo.catalog.rest.model.DataSet in project kylo by Teradata.

the class DataSetProvider method createDataSet.

/**
 * Creates a new data set using the specified template.
 * @param encryptedCredentials
 *
 * @throws CatalogException if the data set is not valid
 */
@Nonnull
public DataSet createDataSet(@Nonnull final DataSet source, boolean encryptedCredentials) {
    // Find data source
    final DataSource dataSource = Optional.of(source).map(DataSet::getDataSource).map(DataSource::getId).flatMap(dataSourceProvider::findDataSource).orElseThrow(() -> new CatalogException("catalog.dataset.datasource.invalid"));
    // Validate data set
    final DataSet dataSet = new DataSet(source);
    dataSet.setDataSource(dataSource);
    validateDataSet(dataSet);
    return metadataService.commit(() -> {
        // Require admin permission if the results should include unencrypted credentials.
        accessController.checkPermission(AccessController.SERVICES, encryptedCredentials ? FeedServicesAccessControl.ACCESS_DATASOURCES : FeedServicesAccessControl.ADMIN_DATASOURCES);
        // Resolve the real data set if possible, otherwise create
        com.thinkbiganalytics.metadata.api.catalog.DataSource.ID dataSourceId = dataSourceMetadataProvider.resolveId(dataSource.getId());
        com.thinkbiganalytics.metadata.api.catalog.DataSet ds = modelTransform.buildDataSet(source, metadataProvider.build(dataSourceId));
        return modelTransform.dataSetToRestModel(encryptedCredentials).apply(ds);
    });
}
Also used : DataSet(com.thinkbiganalytics.kylo.catalog.rest.model.DataSet) CatalogException(com.thinkbiganalytics.kylo.catalog.CatalogException) DataSource(com.thinkbiganalytics.kylo.catalog.rest.model.DataSource) Nonnull(javax.annotation.Nonnull)

Aggregations

DataSet (com.thinkbiganalytics.kylo.catalog.rest.model.DataSet)21 ApiOperation (io.swagger.annotations.ApiOperation)9 DataSource (com.thinkbiganalytics.kylo.catalog.rest.model.DataSource)8 ApiResponses (io.swagger.annotations.ApiResponses)8 Path (javax.ws.rs.Path)8 CatalogException (com.thinkbiganalytics.kylo.catalog.CatalogException)7 DataSetFile (com.thinkbiganalytics.kylo.catalog.rest.model.DataSetFile)6 BadRequestException (javax.ws.rs.BadRequestException)6 DefaultDataSetTemplate (com.thinkbiganalytics.kylo.catalog.rest.model.DefaultDataSetTemplate)5 Nonnull (javax.annotation.Nonnull)5 POST (javax.ws.rs.POST)5 Test (org.junit.Test)5 Produces (javax.ws.rs.Produces)4 File (java.io.File)3 Consumes (javax.ws.rs.Consumes)3 InternalServerErrorException (javax.ws.rs.InternalServerErrorException)3 NotFoundException (javax.ws.rs.NotFoundException)3 WebApplicationException (javax.ws.rs.WebApplicationException)3 FileAlreadyExistsException (org.apache.hadoop.fs.FileAlreadyExistsException)3 ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)2