Java Code Examples for org.apache.hadoop.hive.metastore.api.StorageDescriptor#setCols()

@Test
public void testCheckStructTableSchemaTypeInvalid() throws MetaException {
  TableDescription description = getHashRangeTable();

  Table table = new Table();
  Map<String, String> parameters = Maps.newHashMap();
  parameters.put(DynamoDBConstants.DYNAMODB_COLUMN_MAPPING, "col1:dynamo_col1$," +
      "col2:dynamo_col2#,hashKey:hashKey");
  table.setParameters(parameters);
  StorageDescriptor sd = new StorageDescriptor();
  List<FieldSchema> cols = Lists.newArrayList();
  cols.add(new FieldSchema("col1", "struct<bignum:bigint,smallnum:tinyint>", ""));
  cols.add(new FieldSchema("col2", "array<map<string,bigint>>", ""));
  cols.add(new FieldSchema("hashKey", "string", ""));
  sd.setCols(cols);
  table.setSd(sd);

  exceptionRule.expect(MetaException.class);
  exceptionRule.expectMessage("The hive type struct<bignum:bigint,smallnum:tinyint> is not " +
      "supported in DynamoDB");
  storageHandler.checkTableSchemaType(description, table);
}
 

public static StorageDescriptor convertStorageDescriptor(com.amazonaws.services.glue.model.StorageDescriptor catalogSd) {
  StorageDescriptor hiveSd = new StorageDescriptor();
  hiveSd.setCols(convertFieldSchemaList(catalogSd.getColumns()));
  hiveSd.setLocation(catalogSd.getLocation());
  hiveSd.setInputFormat(catalogSd.getInputFormat());
  hiveSd.setOutputFormat(catalogSd.getOutputFormat());
  hiveSd.setCompressed(catalogSd.getCompressed());
  hiveSd.setNumBuckets(catalogSd.getNumberOfBuckets());
  hiveSd.setSerdeInfo(convertSerDeInfo(catalogSd.getSerdeInfo()));
  hiveSd.setBucketCols(firstNonNull(catalogSd.getBucketColumns(), Lists.<String>newArrayList()));
  hiveSd.setSortCols(convertOrderList(catalogSd.getSortColumns()));
  hiveSd.setParameters(firstNonNull(catalogSd.getParameters(), Maps.<String, String>newHashMap()));
  hiveSd.setSkewedInfo(convertSkewedInfo(catalogSd.getSkewedInfo()));
  hiveSd.setStoredAsSubDirectories(catalogSd.getStoredAsSubDirectories());

  return hiveSd;
}
 

@Test
public void testCheckTableSchemaMappingMissingColumnMapping() throws MetaException {
  TableDescription description = getHashRangeTable();

  Table table = new Table();
  Map<String, String> parameters = Maps.newHashMap();
  parameters.put(DynamoDBConstants.DYNAMODB_COLUMN_MAPPING, "col1:dynamo_col1$," +
   "col2:dynamo_col2#,hashKey:hashKey,hashMap:hashMap");
  table.setParameters(parameters);
  StorageDescriptor sd = new StorageDescriptor();
  List<FieldSchema> cols = Lists.newArrayList();
  cols.add(new FieldSchema("col1", "string", ""));
  cols.add(new FieldSchema("hashMap", "map<string,string>", ""));
  sd.setCols(cols);
  table.setSd(sd);

  exceptionRule.expect(MetaException.class);
  exceptionRule.expectMessage("Could not find column(s) for column mapping(s): ");
  exceptionRule.expectMessage("col2:dynamo_col2#");
  exceptionRule.expectMessage("hashkey:hashKey");
  storageHandler.checkTableSchemaMapping(description, table);
}
 

public Table makeMetastoreTableObject(HiveMetaStoreClient client,
    String dbName, String tabName, List<FieldSchema> cols) throws Exception {
  Table tbl = new Table();
  tbl.setDbName(dbName);
  tbl.setTableName(tabName);
  StorageDescriptor sd = new StorageDescriptor();
  tbl.setSd(sd);
  tbl.setParameters(new HashMap<String, String>());
  sd.setCols(cols);
  sd.setCompressed(false);
  sd.setParameters(new HashMap<String, String>());
  sd.setSerdeInfo(new SerDeInfo());
  sd.getSerdeInfo().setName(tbl.getTableName());
  sd.getSerdeInfo().setParameters(new HashMap<String, String>());
  sd.getSerdeInfo().getParameters()
      .put(serdeConstants.SERIALIZATION_FORMAT, "1");
  sd.setSortCols(new ArrayList<Order>());
  return tbl;
}
 

static Table createPartitionedTable(HiveMetaStoreClient metaStoreClient, String database, String table, File location)
  throws Exception {

  Table hiveTable = new Table();
  hiveTable.setDbName(database);
  hiveTable.setTableName(table);
  hiveTable.setTableType(TableType.EXTERNAL_TABLE.name());
  hiveTable.putToParameters("EXTERNAL", "TRUE");

  hiveTable.setPartitionKeys(PARTITION_COLUMNS);

  StorageDescriptor sd = new StorageDescriptor();
  sd.setCols(DATA_COLUMNS);
  sd.setLocation(location.toURI().toString());
  sd.setParameters(new HashMap<>());
  sd.setSerdeInfo(new SerDeInfo());

  hiveTable.setSd(sd);

  metaStoreClient.createTable(hiveTable);

  return hiveTable;
}
 

@Test
public void testCheckTableSchemaTypeValid() throws MetaException {
  TableDescription description = getHashRangeTable();

  Table table = new Table();
  Map<String, String> parameters = Maps.newHashMap();
  parameters.put(DynamoDBConstants.DYNAMODB_COLUMN_MAPPING, "col1:dynamo_col1$," +
      "col2:dynamo_col2#,hashKey:hashKey");
  table.setParameters(parameters);
  StorageDescriptor sd = new StorageDescriptor();
  List<FieldSchema> cols = Lists.newArrayList();
  cols.add(new FieldSchema("col1", "string", ""));
  cols.add(new FieldSchema("col2", "bigint", ""));
  cols.add(new FieldSchema("hashKey", "string", ""));
  sd.setCols(cols);
  table.setSd(sd);
  // This check is expected to pass for the given input
  storageHandler.checkTableSchemaType(description, table);
}
 

public Partition addTestPartition(Table tbl, List<String> values, int createTime) throws Exception {
  StorageDescriptor partitionSd = new StorageDescriptor();
  if (StringUtils.isNotBlank(tbl.getSd().getLocation())) {
    partitionSd.setLocation(tbl.getSd().getLocation() + values);
  } else {
    partitionSd.setLocation("/tmp/" + tbl.getTableName() + "/part1");
  }

  partitionSd.setSerdeInfo(
      new SerDeInfo("name", "serializationLib", ImmutableMap.of(HiveAvroSerDeManager.SCHEMA_URL, "/tmp/dummy")));
  partitionSd.setCols(tbl.getPartitionKeys());
  Partition partition =
      new Partition(values, tbl.getDbName(), tbl.getTableName(), 1, 1, partitionSd, new HashMap<String, String>());
  partition.setCreateTime(createTime);
  return this.getLocalMetastoreClient().add_partition(partition);

}
 

@Test
public void testCheckTableSchemaMappingMissingColumn() throws MetaException {
  TableDescription description = getHashRangeTable();

  Table table = new Table();
  Map<String, String> parameters = Maps.newHashMap();
  parameters.put(DynamoDBConstants.DYNAMODB_COLUMN_MAPPING, "col1:dynamo_col1$,hashMap:hashMap");
  table.setParameters(parameters);
  StorageDescriptor sd = new StorageDescriptor();
  List<FieldSchema> cols = Lists.newArrayList();
  cols.add(new FieldSchema("col1", "string", ""));
  cols.add(new FieldSchema("col2", "tinyint", ""));
  cols.add(new FieldSchema("col3", "string", ""));
  cols.add(new FieldSchema("hashMap", "map<string,string>", ""));
  sd.setCols(cols);
  table.setSd(sd);

  exceptionRule.expect(MetaException.class);
  exceptionRule.expectMessage("Could not find column mapping for column: col2");
  storageHandler.checkTableSchemaMapping(description, table);
}
 

public static StorageDescriptor create(String location) {
  StorageDescriptor storageDescriptor = new StorageDescriptor();
  storageDescriptor.setInputFormat(AVRO_INPUT_FORMAT);
  storageDescriptor.setOutputFormat(AVRO_OUTPUT_FORMAT);
  storageDescriptor.setLocation(location);
  storageDescriptor.setCols(emptyList());

  SerDeInfo serdeInfo = new SerDeInfo();
  serdeInfo.setSerializationLib(AVRO_SERDE);
  storageDescriptor.setSerdeInfo(serdeInfo);

  return storageDescriptor;
}
 

private Partition getTestPartition(Table table) throws HiveException {
  Partition partition = new Partition(table, ImmutableMap.of("partition_key", "1"), null);
  StorageDescriptor sd = new StorageDescriptor();
  sd.setSerdeInfo(new SerDeInfo("avro", AvroSerDe.class.getName(), null));
  sd.setCols(Lists.newArrayList(new FieldSchema("foo", "int", null)));
  partition.getTPartition().setSd(sd);
  return partition;
}
 

@Before
public void init() {
  partition = new Partition();
  partition.setDbName("database");
  partition.setTableName("table");
  partition.setValues(ImmutableList.of("part"));

  Map<String, List<PrivilegeGrantInfo>> userPrivileges = new HashMap<>();
  userPrivileges.put("read", ImmutableList.of(new PrivilegeGrantInfo()));
  PrincipalPrivilegeSet privileges = new PrincipalPrivilegeSet();
  privileges.setUserPrivileges(userPrivileges);
  partition.setPrivileges(privileges);

  StorageDescriptor storageDescriptor = new StorageDescriptor();
  storageDescriptor.setCols(Arrays.asList(new FieldSchema("a", "int", null)));
  storageDescriptor.setInputFormat("input_format");
  storageDescriptor.setOutputFormat("output_format");
  storageDescriptor.setSerdeInfo(new SerDeInfo("serde", "lib", new HashMap<String, String>()));
  storageDescriptor.setSkewedInfo(new SkewedInfo());
  storageDescriptor.setParameters(new HashMap<String, String>());
  storageDescriptor.setLocation("database/table/part/");
  partition.setSd(storageDescriptor);

  Map<String, String> parameters = new HashMap<>();
  parameters.put("com.company.parameter", "abc");
  partition.setParameters(parameters);
}
 

public static StorageDescriptor newStorageDescriptor(File location, String... columns) {
  StorageDescriptor sd = new StorageDescriptor();
  List<FieldSchema> cols = new ArrayList<>(columns.length);
  for (String name : columns) {
    cols.add(newFieldSchema(name));
  }
  sd.setCols(cols);
  sd.setSerdeInfo(new SerDeInfo());
  sd.setLocation(location.toURI().toString());
  return sd;
}
 

public static Table newTable(String database, String tableName) {
  Table table = new Table();
  table.setDbName(database);
  table.setTableName(tableName);
  table.setTableType(TABLE_TYPE);
  table.setOwner(OWNER);
  table.setCreateTime(CREATE_TIME);
  table.setRetention(RETENTION);

  Map<String, List<PrivilegeGrantInfo>> userPrivileges = new HashMap<>();
  userPrivileges.put("read", ImmutableList.of(new PrivilegeGrantInfo()));
  PrincipalPrivilegeSet privileges = new PrincipalPrivilegeSet();
  privileges.setUserPrivileges(userPrivileges);
  table.setPrivileges(privileges);

  StorageDescriptor storageDescriptor = new StorageDescriptor();
  storageDescriptor.setCols(COLS);
  storageDescriptor.setInputFormat(INPUT_FORMAT);
  storageDescriptor.setOutputFormat(OUTPUT_FORMAT);
  storageDescriptor.setSerdeInfo(new SerDeInfo(SERDE_INFO_NAME, SERIALIZATION_LIB, new HashMap<String, String>()));
  storageDescriptor.setSkewedInfo(new SkewedInfo());
  storageDescriptor.setParameters(new HashMap<String, String>());
  storageDescriptor.setLocation(DATABASE + "/" + tableName + "/");
  table.setSd(storageDescriptor);

  Map<String, String> parameters = new HashMap<>();
  parameters.put("com.company.parameter", "abc");
  table.setParameters(parameters);

  return table;
}
 

@Before
public void before() throws TException, IOException {
  Table table = new Table();
  table.setDbName(DATABASE);
  table.setTableName("source_" + TABLE);
  table.setTableType(TableType.EXTERNAL_TABLE.name());
  table.putToParameters("EXTERNAL", "TRUE");

  StorageDescriptor sd = new StorageDescriptor();
  sd.setCols(Arrays.asList(new FieldSchema("col1", "string", null)));
  sd.setSerdeInfo(new SerDeInfo());
  table.setSd(sd);

  hive.client().createTable(table);
}
 

private void createSourceTable() throws Exception {
  File partitionEurope = new File(sourceTableUri, "local_date=2000-01-01");
  File partitionUk = new File(partitionEurope, "local_hour=0");
  File dataFileUk = new File(partitionUk, PART_00000);
  FileUtils.writeStringToFile(dataFileUk, "1\tadam\tlondon\n2\tsusan\tglasgow\n");

  File partitionAsia = new File(sourceTableUri, "local_date=2000-01-02");
  File partitionChina = new File(partitionAsia, "local_hour=0");
  File dataFileChina = new File(partitionChina, PART_00000);
  String data = "1\tchun\tbeijing\n2\tshanghai\tmilan\n";
  FileUtils.writeStringToFile(dataFileChina, data);

  HiveMetaStoreClient sourceClient = catalog.client();

  Table source = new Table();
  source.setDbName(DATABASE);
  source.setTableName(SOURCE_TABLE);
  source.setTableType(TableType.EXTERNAL_TABLE.name());
  Map<String, String> parameters = new HashMap<>();
  parameters.put("comment", "comment source");
  source.setParameters(parameters);

  List<FieldSchema> partitionColumns = Arrays.asList(new FieldSchema("local_date", "string", ""),
      new FieldSchema("local_hour", "string", ""));
  source.setPartitionKeys(partitionColumns);

  List<FieldSchema> dataColumns = Arrays.asList(new FieldSchema("id", "bigint", ""),
      new FieldSchema("name", "string", ""), new FieldSchema("city", "string", ""));

  StorageDescriptor sd = new StorageDescriptor();
  sd.setCols(dataColumns);
  sd.setLocation(sourceTableUri.toURI().toString());
  sd.setParameters(new HashMap<String, String>());
  sd.setSerdeInfo(new SerDeInfo());

  source.setSd(sd);

  sourceClient.createTable(source);
  LOG.info(">>>> Partitions added: {}",
      +sourceClient
          .add_partitions(Arrays.asList(newPartition(SOURCE_TABLE, sd, Arrays.asList("2000-01-01", "0"), partitionUk),
              newPartition(SOURCE_TABLE, sd, Arrays.asList("2000-01-02", "0"), partitionChina))));
}
 

public static void alterColumns(StorageDescriptor sd, CatalogTable catalogTable) {
	List<FieldSchema> allCols = HiveTableUtil.createHiveColumns(catalogTable.getSchema());
	List<FieldSchema> nonPartCols = allCols.subList(0, allCols.size() - catalogTable.getPartitionKeys().size());
	sd.setCols(nonPartCols);
}
 

private void createReplicaTable() throws Exception {
  File partitionEurope = new File(replicaTableUri, "local_date=2000-01-01");
  File partitionUk = new File(partitionEurope, "local_hour=0");
  File dataFileUk = new File(partitionUk, PART_00000);
  FileUtils.writeStringToFile(dataFileUk, "1\tadam\tlondon\tuk\n2\tsusan\tglasgow\tuk\n");

  File partitionAsia = new File(replicaTableUri, "local_date=2000-01-02");
  File partitionChina = new File(partitionAsia, "local_hour=0");
  File dataFileChina = new File(partitionChina, PART_00000);
  String data = "1\tchun\tbeijing\tchina\n2\tshanghai\tmilan\titaly\n";
  FileUtils.writeStringToFile(dataFileChina, data);

  HiveMetaStoreClient replicaClient = catalog.client();

  Table replica = new Table();
  replica.setDbName(DATABASE);
  replica.setTableName(REPLICA_TABLE);
  replica.setTableType(TableType.EXTERNAL_TABLE.name());
  Map<String, String> parameters = new HashMap<>();
  parameters.put("comment", "comment replica");
  replica.setParameters(parameters);
  List<FieldSchema> partitionColumns = Arrays.asList(new FieldSchema("local_date", "string", ""),
      new FieldSchema("local_hour", "string", ""));
  replica.setPartitionKeys(partitionColumns);

  List<FieldSchema> dataColumns = Arrays.asList(new FieldSchema("id", "bigint", ""),
      new FieldSchema("name", "string", ""), new FieldSchema("city", "string", ""),
      new FieldSchema("country", "string", ""));

  StorageDescriptor sd = new StorageDescriptor();
  sd.setCols(dataColumns);
  sd.setLocation(replicaTableUri.toURI().toString());
  sd.setParameters(new HashMap<String, String>());
  sd.setSerdeInfo(new SerDeInfo());

  replica.setSd(sd);

  replicaClient.createTable(replica);
  LOG.info(">>>> Partitions added: {}",
      +replicaClient.add_partitions(
          Arrays.asList(newPartition(REPLICA_TABLE, sd, Arrays.asList("2000-01-01", "0"), partitionUk),
              newPartition(REPLICA_TABLE, sd, Arrays.asList("2000-01-02", "0"), partitionChina))));
}
 

public static Table createPartitionedTable(
    HiveMetaStoreClient metaStoreClient,
    String database,
    String table,
    URI location,
    List<FieldSchema> columns,
    List<FieldSchema> partitionKeys,
    String serializationLib,
    String inputFormatClassName,
    String outputFormatClassName)
    throws Exception {

  Table hiveTable = new Table();
  hiveTable.setDbName(database);
  hiveTable.setTableName(table);
  hiveTable.setTableType(TableType.EXTERNAL_TABLE.name());
  hiveTable.putToParameters("EXTERNAL", "TRUE");

  hiveTable.setPartitionKeys(partitionKeys);

  StorageDescriptor sd = new StorageDescriptor();
  sd.setCols(columns);
  sd.setLocation(location.toString());
  sd.setParameters(new HashMap<String, String>());
  sd.setInputFormat(inputFormatClassName);
  sd.setOutputFormat(outputFormatClassName);
  sd.setSerdeInfo(new SerDeInfo());
  sd.getSerdeInfo().setSerializationLib(serializationLib);

  hiveTable.setSd(sd);

  metaStoreClient.createTable(hiveTable);

  ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(true, database, table);
  ColumnStatisticsData statsData = new ColumnStatisticsData(_Fields.LONG_STATS, new LongColumnStatsData(1L, 2L));
  ColumnStatisticsObj cso1 = new ColumnStatisticsObj("id", "bigint", statsData);
  List<ColumnStatisticsObj> statsObj = Collections.singletonList(cso1);
  metaStoreClient.updateTableColumnStatistics(new ColumnStatistics(statsDesc, statsObj));

  return hiveTable;
}
 

@Test
public void getTableMetadataWithMultipleTables() throws Exception {

    fetcher = new HiveMetadataFetcher(context, mockConfigurationFactory, fakeHiveClientWrapper);

    String tablePattern = "*";
    String dbPattern = "*";
    String dbName = "default";
    String tableNameBase = "regulartable";
    String pattern = dbPattern + "." + tablePattern;

    List<String> dbNames = new ArrayList<>(Collections.singletonList(dbName));
    List<String> tableNames = new ArrayList<>();

    // Prepare for tables
    List<FieldSchema> fields = new ArrayList<>();
    fields.add(new FieldSchema("field1", "string", null));
    fields.add(new FieldSchema("field2", "int", null));
    StorageDescriptor sd = new StorageDescriptor();
    sd.setCols(fields);
    sd.setInputFormat("org.apache.hadoop.mapred.TextInputFormat");

    // Mock hive tables returned from hive client
    for (int index = 1; index <= 2; index++) {
        String tableName = tableNameBase + index;
        tableNames.add(tableName);
        Table hiveTable = new Table();
        hiveTable.setTableType("MANAGED_TABLE");
        hiveTable.setSd(sd);
        hiveTable.setPartitionKeys(new ArrayList<>());
        when(mockHiveClient.getTable(dbName, tableName)).thenReturn(hiveTable);
    }

    // Mock database and table names return from hive client
    when(mockHiveClient.getDatabases(dbPattern)).thenReturn(dbNames);
    when(mockHiveClient.getTables(dbName, tablePattern)).thenReturn(tableNames);

    // Get metadata
    metadataList = fetcher.getMetadata(pattern);
    assertEquals(2, metadataList.size());

    for (int index = 1; index <= 2; index++) {
        Metadata metadata = metadataList.get(index - 1);
        assertEquals(dbName + "." + tableNameBase + index, metadata.getItem().toString());
        List<Metadata.Field> resultFields = metadata.getFields();
        assertNotNull(resultFields);
        assertEquals(2, resultFields.size());
        Metadata.Field field = resultFields.get(0);
        assertEquals("field1", field.getName());
        assertEquals("text", field.getType().getTypeName()); // converted type
        field = resultFields.get(1);
        assertEquals("field2", field.getName());
        assertEquals("int4", field.getType().getTypeName());
    }
}
 

/**
 * {@inheritDoc}
 */
@Override
public Partition metacatToHivePartition(final PartitionDto partitionDto, @Nullable final TableDto tableDto) {
    final Partition result = new Partition();

    final QualifiedName name = partitionDto.getName();
    List<String> values = Lists.newArrayListWithCapacity(16);
    String databaseName = null;
    String tableName = null;
    if (name != null) {
        if (name.getPartitionName() != null) {
            //
            // Unescape the partition name to get the right partition values.
            // Partition name always are escaped where as the parition values are not.
            //
            values = getPartValsFromName(tableDto, name.getPartitionName());
        }

        if (name.getDatabaseName() != null) {
            databaseName = name.getDatabaseName();
        }

        if (name.getTableName() != null) {
            tableName = name.getTableName();
        }
    }
    result.setValues(values);
    result.setDbName(databaseName);
    result.setTableName(tableName);

    Map<String, String> metadata = partitionDto.getMetadata();
    if (metadata == null) {
        metadata = Maps.newHashMap();
    }
    result.setParameters(metadata);

    result.setSd(fromStorageDto(partitionDto.getSerde(), tableName));
    final StorageDescriptor sd = result.getSd();
    if (tableDto != null) {
        if (sd.getSerdeInfo() != null && tableDto.getSerde() != null && Strings.isNullOrEmpty(
            sd.getSerdeInfo().getSerializationLib())) {
            sd.getSerdeInfo().setSerializationLib(tableDto.getSerde().getSerializationLib());
        }

        final List<FieldDto> fields = tableDto.getFields();
        if (fields == null) {
            sd.setCols(Collections.emptyList());
        } else {
            sd.setCols(fields.stream()
                .filter(field -> !field.isPartition_key())
                .map(this::metacatToHiveField)
                .collect(Collectors.toList()));
        }
    }

    final AuditDto auditDto = partitionDto.getAudit();
    if (auditDto != null) {
        if (auditDto.getCreatedDate() != null) {
            result.setCreateTime(dateToEpochSeconds(auditDto.getCreatedDate()));
        }
        if (auditDto.getLastModifiedDate() != null) {
            result.setLastAccessTime(dateToEpochSeconds(auditDto.getLastModifiedDate()));
        }
    }

    return result;
}