Examples with DefaultTag com.thinkbiganalytics.discovery.model.DefaultTag used on opensource projects

Example 6 with DefaultTag

use of com.thinkbiganalytics.discovery.model.DefaultTag in project kylo by Teradata.

the class FeedIT method getCreateFeedRequest.

protected FeedMetadata getCreateFeedRequest(FeedCategory category, ImportTemplate template, String name) throws Exception {
    FeedMetadata feed = new FeedMetadata();
    feed.setFeedName(name);
    feed.setSystemFeedName(name.toLowerCase());
    feed.setCategory(category);
    feed.setTemplateId(template.getTemplateId());
    feed.setTemplateName(template.getTemplateName());
    feed.setDescription("Created by functional test");
    feed.setInputProcessorType("org.apache.nifi.processors.standard.GetFile");
    List<NifiProperty> properties = new ArrayList<>();
    NifiProperty fileFilter = new NifiProperty("305363d8-015a-1000-0000-000000000000", "1f67e296-2ff8-4b5d-0000-000000000000", "File Filter", USERDATA1_CSV);
    fileFilter.setProcessGroupName("NiFi Flow");
    fileFilter.setProcessorName("Filesystem");
    fileFilter.setProcessorType("org.apache.nifi.processors.standard.GetFile");
    fileFilter.setTemplateValue("mydata\\d{1,3}.csv");
    fileFilter.setInputProperty(true);
    fileFilter.setUserEditable(true);
    properties.add(fileFilter);
    NifiProperty inputDir = new NifiProperty("305363d8-015a-1000-0000-000000000000", "1f67e296-2ff8-4b5d-0000-000000000000", "Input Directory", VAR_DROPZONE);
    inputDir.setProcessGroupName("NiFi Flow");
    inputDir.setProcessorName("Filesystem");
    inputDir.setProcessorType("org.apache.nifi.processors.standard.GetFile");
    inputDir.setInputProperty(true);
    inputDir.setUserEditable(true);
    properties.add(inputDir);
    NifiProperty loadStrategy = new NifiProperty("305363d8-015a-1000-0000-000000000000", "6aeabec7-ec36-4ed5-0000-000000000000", "Load Strategy", "FULL_LOAD");
    loadStrategy.setProcessGroupName("NiFi Flow");
    loadStrategy.setProcessorName("GetTableData");
    loadStrategy.setProcessorType("com.thinkbiganalytics.nifi.v2.ingest.GetTableData");
    properties.add(loadStrategy);
    feed.setProperties(properties);
    FeedSchedule schedule = new FeedSchedule();
    schedule.setConcurrentTasks(1);
    schedule.setSchedulingPeriod("15 sec");
    schedule.setSchedulingStrategy("TIMER_DRIVEN");
    feed.setSchedule(schedule);
    TableSetup table = new TableSetup();
    DefaultTableSchema schema = new DefaultTableSchema();
    schema.setName("test1");
    List<Field> fields = new ArrayList<>();
    fields.add(newTimestampField("registration_dttm"));
    fields.add(newBigIntField("id"));
    fields.add(newStringField("first_name"));
    fields.add(newStringField("second_name"));
    fields.add(newStringField("email"));
    fields.add(newStringField("gender"));
    fields.add(newStringField("ip_address"));
    fields.add(newBinaryField("cc"));
    fields.add(newStringField("country"));
    fields.add(newStringField("birthdate"));
    fields.add(newStringField("salary"));
    schema.setFields(fields);
    table.setTableSchema(schema);
    table.setSourceTableSchema(schema);
    table.setFeedTableSchema(schema);
    table.setTargetMergeStrategy("DEDUPE_AND_MERGE");
    table.setFeedFormat("ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.OpenCSVSerde'\n WITH SERDEPROPERTIES ( 'separatorChar' = ',' ,'escapeChar' = '\\\\' ,'quoteChar' = '\\'') STORED AS TEXTFILE");
    table.setTargetFormat("STORED AS ORC");
    List<FieldPolicy> policies = new ArrayList<>();
    policies.add(newPolicyBuilder("registration_dttm").toPolicy());
    policies.add(newPolicyBuilder("id").toPolicy());
    policies.add(newPolicyBuilder("first_name").withStandardisation(toUpperCase).withProfile().withIndex().toPolicy());
    policies.add(newPolicyBuilder("second_name").withProfile().withIndex().toPolicy());
    policies.add(newPolicyBuilder("email").withValidation(email).toPolicy());
    policies.add(newPolicyBuilder("gender").withValidation(lookup, notNull).toPolicy());
    policies.add(newPolicyBuilder("ip_address").withValidation(ipAddress).toPolicy());
    policies.add(newPolicyBuilder("cc").withStandardisation(base64EncodeBinary).withProfile().toPolicy());
    policies.add(newPolicyBuilder("country").withStandardisation(base64EncodeBinary, base64DecodeBinary, base64EncodeString, base64DecodeString).withValidation(notNull, length).withProfile().toPolicy());
    policies.add(newPolicyBuilder("birthdate").toPolicy());
    policies.add(newPolicyBuilder("salary").toPolicy());
    table.setFieldPolicies(policies);
    List<PartitionField> partitions = new ArrayList<>();
    partitions.add(byYear("registration_dttm"));
    table.setPartitions(partitions);
    TableOptions options = new TableOptions();
    options.setCompressionFormat("SNAPPY");
    options.setAuditLogging(true);
    table.setOptions(options);
    table.setTableType("SNAPSHOT");
    feed.setTable(table);
    feed.setOptions(new FeedProcessingOptions());
    feed.getOptions().setSkipHeader(true);
    feed.setDataOwner("Marketing");
    List<Tag> tags = new ArrayList<>();
    tags.add(new DefaultTag("users"));
    tags.add(new DefaultTag("registrations"));
    feed.setTags(tags);
    User owner = new User();
    owner.setSystemName("dladmin");
    owner.setDisplayName("Data Lake Admin");
    Set<String> groups = new HashSet<>();
    groups.add("admin");
    groups.add("user");
    owner.setGroups(groups);
    feed.setOwner(owner);
    return feed;
}

Example 7 with DefaultTag

use of com.thinkbiganalytics.discovery.model.DefaultTag in project kylo by Teradata.

the class FeedIT method testEditFeed.

@Test
public void testEditFeed() throws Exception {
    // Prepare environment
    prepare();
    final FeedCategory category = createCategory(CATEGORY_NAME);
    final ImportTemplate template = importDataIngestTemplate();
    // Create feed
    FeedMetadata feed = getCreateFeedRequest(category, template, createNewFeedName());
    feed.setDescription("Test feed");
    feed.setDataOwner("Some Guy");
    FeedMetadata response = createFeed(feed).getFeedMetadata();
    Assert.assertEquals(feed.getFeedName(), response.getFeedName());
    Assert.assertEquals(feed.getDataOwner(), response.getDataOwner());
    // Edit feed
    feed.setId(response.getId());
    feed.setFeedId(response.getFeedId());
    feed.setIsNew(false);
    feed.setDescription(null);
    feed.setDataOwner("Some Other Guy");
    NifiProperty fileFilter = feed.getProperties().get(0);
    fileFilter.setValue("some-file.csv");
    List<FieldPolicy> policies = feed.getTable().getFieldPolicies();
    FieldPolicy id = policies.get(1);
    // add new validator
    id.getValidation().add(notNull);
    feed.getTable().setPrimaryKeyFields("id");
    FieldPolicy firstName = policies.get(2);
    // flip profiling
    firstName.setProfile(false);
    FieldPolicy secondName = policies.get(3);
    // flip indexing
    secondName.setIndex(false);
    // add new standardiser
    secondName.getStandardization().add(toUpperCase);
    FieldPolicy email = policies.get(4);
    // remove validators
    email.setValidation(Collections.emptyList());
    FieldPolicy gender = policies.get(5);
    FieldValidationRule lookup = gender.getValidation().get(0);
    // change existing validator property
    lookup.getProperties().get(0).setValue("new value");
    // add profiling
    gender.setProfile(true);
    // add indexing
    gender.setIndex(true);
    FieldPolicy creditCard = policies.get(7);
    FieldStandardizationRule base64EncodeBinary = creditCard.getStandardization().get(0);
    // change existing standardiser property
    base64EncodeBinary.getProperties().get(0).setValue("STRING");
    feed.getOptions().setSkipHeader(false);
    feed.getTable().setTargetMergeStrategy("ROLLING_SYNC");
    feed.getTags().add(new DefaultTag("updated"));
    feed.getSchedule().setSchedulingPeriod("20 sec");
    response = createFeed(feed).getFeedMetadata();
    Assert.assertEquals(feed.getFeedName(), response.getFeedName());
    Assert.assertEquals(feed.getDescription(), response.getDescription());
    FeedVersions feedVersions = getVersions(feed.getFeedId());
    List<EntityVersion> versions = feedVersions.getVersions();
    Assert.assertEquals(2, versions.size());
    EntityVersionDifference entityDiff = getVersionDiff(feed.getFeedId(), versions.get(1).getId(), versions.get(0).getId());
    EntityDifference diff = entityDiff.getDifference();
    JsonNode patch = diff.getPatch();
    ArrayNode diffs = (ArrayNode) patch;
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/properties/0/value", "some-file.csv")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/schedule/schedulingPeriod", "20 sec")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("remove", "/description")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("add", "/tags/1")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/dataOwner", "Some Other Guy")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("add", "/table/fieldPolicies/1/validation/0")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/2/profile", "false")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/3/index", "false")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("add", "/table/fieldPolicies/3/standardization/0")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("remove", "/table/fieldPolicies/4/validation/0")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/5/profile", "true")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/5/index", "true")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/5/validation/0/properties/0/value", "new value")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/7/standardization/0/properties/0/value", "STRING")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldPolicies/8/standardization/0/properties/0/value", "STRING")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/targetMergeStrategy", "ROLLING_SYNC")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/table/fieldIndexString", "first_name,gender")));
    Assert.assertTrue(versionPatchContains(diffs, new Diff("replace", "/options/skipHeader", "false")));
}

Example 8 with DefaultTag

use of com.thinkbiganalytics.discovery.model.DefaultTag in project kylo by Teradata.

the class IntegrationTestBase method makeCreateFeedRequest.

protected FeedMetadata makeCreateFeedRequest(FeedCategory category, ImportTemplate template, String feedName, String testFile) {
    FeedMetadata feed = new FeedMetadata();
    feed.setFeedName(feedName);
    feed.setSystemFeedName(feedName.toLowerCase());
    feed.setCategory(category);
    feed.setTemplateId(template.getTemplateId());
    feed.setTemplateName(template.getTemplateName());
    feed.setDescription("Created by functional test");
    feed.setInputProcessorType("org.apache.nifi.processors.standard.GetFile");
    List<NifiProperty> properties = new ArrayList<>();
    NifiProperty fileFilter = new NifiProperty("764d053d-015e-1000-b8a2-763cd17080e1", "cffa8f24-d097-3c7a-7d04-26b7feff81ab", "File Filter", testFile);
    fileFilter.setProcessGroupName("NiFi Flow");
    fileFilter.setProcessorName("GetFile");
    fileFilter.setProcessorType("org.apache.nifi.processors.standard.GetFile");
    fileFilter.setTemplateValue("mydata\\d{1,3}.csv");
    fileFilter.setInputProperty(true);
    fileFilter.setUserEditable(true);
    properties.add(fileFilter);
    feed.setProperties(properties);
    FeedSchedule schedule = new FeedSchedule();
    schedule.setConcurrentTasks(1);
    schedule.setSchedulingPeriod("15 sec");
    schedule.setSchedulingStrategy("TIMER_DRIVEN");
    feed.setSchedule(schedule);
    feed.setDataOwner("Marketing");
    List<Tag> tags = new ArrayList<>();
    tags.add(new DefaultTag("functional tests"));
    tags.add(new DefaultTag("for category " + category.getName()));
    feed.setTags(tags);
    User owner = new User();
    owner.setSystemName("dladmin");
    owner.setDisplayName("Data Lake Admin");
    Set<String> groups = new HashSet<>();
    groups.add("admin");
    groups.add("user");
    owner.setGroups(groups);
    feed.setOwner(owner);
    return feed;
}

Example 9 with DefaultTag

use of com.thinkbiganalytics.discovery.model.DefaultTag in project kylo by Teradata.

the class FeedITBase method editFeed.

protected void editFeed() throws Exception {
    // Prepare environment
    prepare();
    final FeedCategory category = createCategory(CATEGORY_NAME);
    final ImportTemplate template = importDataIngestTemplate();
    // Create feed
    FeedMetadata feed = getCreateFeedRequest(category, template, createNewFeedName());
    feed.setDescription("Test feed");
    feed.setDataOwner("Some Guy");
    FeedMetadata response = createFeed(feed).getFeedMetadata();
    Assert.assertEquals(feed.getFeedName(), response.getFeedName());
    Assert.assertEquals(feed.getDataOwner(), response.getDataOwner());
    // Edit feed
    feed.setId(response.getId());
    feed.setFeedId(response.getFeedId());
    feed.setIsNew(false);
    feed.setDescription(null);
    feed.setDataOwner("Some Other Guy");
    NifiProperty fileFilter = feed.getProperties().get(0);
    fileFilter.setValue(getEditedFileName());
    List<FieldPolicy> policies = feed.getTable().getFieldPolicies();
    editFieldPolicies(policies);
    feed.getTable().setPrimaryKeyFields("id");
    feed.getOptions().setSkipHeader(false);
    feed.getTable().setTargetMergeStrategy("ROLLING_SYNC");
    feed.getTags().add(new DefaultTag("updated"));
    feed.getSchedule().setSchedulingPeriod("20 sec");
    response = createFeed(feed).getFeedMetadata();
    Assert.assertEquals(feed.getFeedName(), response.getFeedName());
    Assert.assertEquals(feed.getDescription(), response.getDescription());
    FeedVersions feedVersions = getVersions(feed.getFeedId());
    List<EntityVersion> versions = feedVersions.getVersions();
    Assert.assertEquals(2, versions.size());
    EntityVersionDifference entityDiff = getVersionDiff(feed.getFeedId(), versions.get(1).getId(), versions.get(0).getId());
    EntityDifference diff = entityDiff.getDifference();
    JsonNode patch = diff.getPatch();
    ArrayNode diffs = (ArrayNode) patch;
    assertEditChanges(diffs);
}

Example 10 with DefaultTag

use of com.thinkbiganalytics.discovery.model.DefaultTag in project kylo by Teradata.

the class CsvFeedIT method getTags.

protected List<Tag> getTags() {
    List<Tag> tags = new ArrayList<>();
    tags.add(new DefaultTag("users"));
    tags.add(new DefaultTag("registrations"));
    return tags;
}