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

Example 6 with DefaultTableSchema

use of com.thinkbiganalytics.discovery.model.DefaultTableSchema 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;
}