org.apache.hadoop.hive.metastore.api.StorageDescriptor#getSerdeInfo

Source File: ThriftMetastoreUtil.java From presto with Apache License 2.0

6 votes

private static void fromMetastoreApiStorageDescriptor(
        Map<String, String> tableParameters,
        StorageDescriptor storageDescriptor,
        Storage.Builder builder,
        String tablePartitionName)
{
    SerDeInfo serdeInfo = storageDescriptor.getSerdeInfo();
    if (serdeInfo == null) {
        throw new PrestoException(HIVE_INVALID_METADATA, "Table storage descriptor is missing SerDe info");
    }

    builder.setStorageFormat(StorageFormat.createNullable(serdeInfo.getSerializationLib(), storageDescriptor.getInputFormat(), storageDescriptor.getOutputFormat()))
            .setLocation(nullToEmpty(storageDescriptor.getLocation()))
            .setBucketProperty(HiveBucketProperty.fromStorageDescriptor(tableParameters, storageDescriptor, tablePartitionName))
            .setSkewed(storageDescriptor.isSetSkewedInfo() && storageDescriptor.getSkewedInfo().isSetSkewedColNames() && !storageDescriptor.getSkewedInfo().getSkewedColNames().isEmpty())
            .setSerdeParameters(serdeInfo.getParameters() == null ? ImmutableMap.of() : serdeInfo.getParameters());
}

Source File: HiveTableUtil.java From flink with Apache License 2.0

6 votes

/**
 * Create properties info to initialize a SerDe.
 * @param storageDescriptor
 * @return
 */
public static Properties createPropertiesFromStorageDescriptor(StorageDescriptor storageDescriptor) {
	SerDeInfo serDeInfo = storageDescriptor.getSerdeInfo();
	Map<String, String> parameters = serDeInfo.getParameters();
	Properties properties = new Properties();
	properties.setProperty(
			serdeConstants.SERIALIZATION_FORMAT,
			parameters.get(serdeConstants.SERIALIZATION_FORMAT));
	List<String> colTypes = new ArrayList<>();
	List<String> colNames = new ArrayList<>();
	List<FieldSchema> cols = storageDescriptor.getCols();
	for (FieldSchema col: cols){
		colTypes.add(col.getType());
		colNames.add(col.getName());
	}
	properties.setProperty(serdeConstants.LIST_COLUMNS, StringUtils.join(colNames, String.valueOf(SerDeUtils.COMMA)));
	// Note: serdeConstants.COLUMN_NAME_DELIMITER is not defined in previous Hive. We use a literal to save on shim
	properties.setProperty("column.name.delimite", String.valueOf(SerDeUtils.COMMA));
	properties.setProperty(serdeConstants.LIST_COLUMN_TYPES, StringUtils.join(colTypes, DEFAULT_LIST_COLUMN_TYPES_SEPARATOR));
	properties.setProperty(serdeConstants.SERIALIZATION_NULL_FORMAT, "NULL");
	properties.putAll(parameters);
	return properties;
}

Source File: HiveConnectorTableService.java From metacat with Apache License 2.0

6 votes

private HiveStorageFormat extractHiveStorageFormat(final Table table) throws MetaException {
    final StorageDescriptor descriptor = table.getSd();
    if (descriptor == null) {
        throw new MetaException("Table is missing storage descriptor");
    }
    final SerDeInfo serdeInfo = descriptor.getSerdeInfo();
    if (serdeInfo == null) {
        throw new MetaException(
            "Table storage descriptor is missing SerDe info");
    }
    final String outputFormat = descriptor.getOutputFormat();
    final String serializationLib = serdeInfo.getSerializationLib();

    for (HiveStorageFormat format : HiveStorageFormat.values()) {
        if (format.getOutputFormat().equals(outputFormat) && format.getSerde().equals(serializationLib)) {
            return format;
        }
    }
    throw new MetaException(
        String.format("Output format %s with SerDe %s is not supported", outputFormat, serializationLib));
}

Source File: HiveConnectorInfoConverter.java From metacat with Apache License 2.0

6 votes

private StorageInfo toStorageInfo(final StorageDescriptor sd, final String owner) {
    if (sd == null) {
        return new StorageInfo();
    }
    if (sd.getSerdeInfo() != null) {
        return StorageInfo.builder().owner(owner)
            .uri(sd.getLocation())
            .inputFormat(sd.getInputFormat())
            .outputFormat(sd.getOutputFormat())
            .parameters(sd.getParameters())
            .serializationLib(sd.getSerdeInfo().getSerializationLib())
            .serdeInfoParameters(sd.getSerdeInfo().getParameters())
            .build();
    }
    return StorageInfo.builder().owner(owner).uri(sd.getLocation()).inputFormat(sd.getInputFormat())
        .outputFormat(sd.getOutputFormat()).parameters(sd.getParameters()).build();
}

Source File: HiveConvertersImpl.java From metacat with Apache License 2.0

6 votes

private StorageDto toStorageDto(@Nullable final StorageDescriptor sd, final String owner) {
    final StorageDto result = new StorageDto();
    if (sd != null) {
        result.setOwner(owner);
        result.setUri(sd.getLocation());
        result.setInputFormat(sd.getInputFormat());
        result.setOutputFormat(sd.getOutputFormat());
        result.setParameters(sd.getParameters());
        final SerDeInfo serde = sd.getSerdeInfo();
        if (serde != null) {
            result.setSerializationLib(serde.getSerializationLib());
            result.setSerdeInfoParameters(serde.getParameters());
        }
    }
    return result;
}

Source File: AvroHiveTableStrategyTest.java From data-highway with Apache License 2.0

5 votes

@Test
public void newHiveTable() throws URISyntaxException {
  when(uriResolver.resolve(schema1, TABLE, 1))
      .thenReturn(new URI("https://s3.amazonaws.com/road-schema-bucket/roads/table/schemas/1/table_v1.avsc"));
  doReturn(Instant.ofEpochSecond(1526462225L)).when(clock).instant();

  Table result = underTest.newHiveTable(DATABASE, TABLE, PARTITION_COLUMN, LOCATION, schema1, 1);

  assertThat(result.getDbName(), is(DATABASE));
  assertThat(result.getTableName(), is(TABLE));
  assertThat(result.getTableType(), is(TableType.EXTERNAL_TABLE.toString()));
  Map<String, String> parameters = result.getParameters();
  assertThat(parameters.get("EXTERNAL"), is("TRUE"));
  assertThat(parameters.get("data-highway.version"), is(DataHighwayVersion.VERSION));
  assertThat(parameters.get("data-highway.last-revision"), is("2018-05-16T09:17:05Z"));
  assertThat(parameters.get(AvroHiveTableStrategy.AVRO_SCHEMA_URL),
      is("https://s3.amazonaws.com/road-schema-bucket/roads/table/schemas/1/table_v1.avsc"));
  assertThat(parameters.get(AvroHiveTableStrategy.AVRO_SCHEMA_VERSION), is("1"));
  List<FieldSchema> partitionKeys = result.getPartitionKeys();
  assertThat(partitionKeys.size(), is(1));
  assertThat(partitionKeys.get(0), is(new FieldSchema(PARTITION_COLUMN, "string", null)));
  StorageDescriptor storageDescriptor = result.getSd();
  assertThat(storageDescriptor.getInputFormat(), is(AvroStorageDescriptorFactory.AVRO_INPUT_FORMAT));
  assertThat(storageDescriptor.getOutputFormat(), is(AvroStorageDescriptorFactory.AVRO_OUTPUT_FORMAT));
  assertThat(storageDescriptor.getLocation(), is(LOCATION));
  assertThat(storageDescriptor.getCols().size(), is(0));
  SerDeInfo serdeInfo = storageDescriptor.getSerdeInfo();
  assertThat(serdeInfo.getSerializationLib(), is(AvroStorageDescriptorFactory.AVRO_SERDE));
}

Source File: AvroHiveTableStrategyTest.java From data-highway with Apache License 2.0

5 votes

@Test
public void alterHiveTable() throws URISyntaxException {
  when(uriResolver.resolve(schema1, TABLE, 1))
      .thenReturn(new URI("https://s3.amazonaws.com/road-schema-bucket/roads/table/schemas/1/table_v1.avsc"));
  when(uriResolver.resolve(schema2, TABLE, 2))
      .thenReturn(new URI("https://s3.amazonaws.com/road-schema-bucket/roads/table/schemas/2/table_v2.avsc"));
  doReturn(Instant.ofEpochSecond(1526462225L)).when(clock).instant();

  Table table = underTest.newHiveTable(DATABASE, TABLE, PARTITION_COLUMN, LOCATION, schema1, 1);

  Table result = underTest.alterHiveTable(table, schema2, 2);

  assertThat(result.getDbName(), is(DATABASE));
  assertThat(result.getTableName(), is(TABLE));
  assertThat(result.getTableType(), is(TableType.EXTERNAL_TABLE.toString()));
  Map<String, String> parameters = result.getParameters();
  assertThat(parameters.get("EXTERNAL"), is("TRUE"));
  assertThat(parameters.get("data-highway.version"), is(DataHighwayVersion.VERSION));
  assertThat(parameters.get("data-highway.last-revision"), is("2018-05-16T09:17:05Z"));
  assertThat(parameters.get(AvroHiveTableStrategy.AVRO_SCHEMA_URL),
      is("https://s3.amazonaws.com/road-schema-bucket/roads/table/schemas/2/table_v2.avsc"));
  assertThat(parameters.get(AvroHiveTableStrategy.AVRO_SCHEMA_VERSION), is("2"));
  List<FieldSchema> partitionKeys = result.getPartitionKeys();
  assertThat(partitionKeys.size(), is(1));
  assertThat(partitionKeys.get(0), is(new FieldSchema(PARTITION_COLUMN, "string", null)));
  StorageDescriptor storageDescriptor = result.getSd();
  assertThat(storageDescriptor.getInputFormat(), is(AvroStorageDescriptorFactory.AVRO_INPUT_FORMAT));
  assertThat(storageDescriptor.getOutputFormat(), is(AvroStorageDescriptorFactory.AVRO_OUTPUT_FORMAT));
  assertThat(storageDescriptor.getLocation(), is(LOCATION));
  assertThat(storageDescriptor.getCols().size(), is(0));
  SerDeInfo serdeInfo = storageDescriptor.getSerdeInfo();
  assertThat(serdeInfo.getSerializationLib(), is(AvroStorageDescriptorFactory.AVRO_SERDE));
}

Source File: ThriftMetastoreUtil.java From presto with Apache License 2.0

5 votes

private static SerDeInfo getSerdeInfo(org.apache.hadoop.hive.metastore.api.Table table)
{
    StorageDescriptor storageDescriptor = table.getSd();
    if (storageDescriptor == null) {
        throw new PrestoException(HIVE_INVALID_METADATA, "Table does not contain a storage descriptor: " + table);
    }
    SerDeInfo serdeInfo = storageDescriptor.getSerdeInfo();
    if (serdeInfo == null) {
        throw new PrestoException(HIVE_INVALID_METADATA, "Table storage descriptor is missing SerDe info");
    }

    return serdeInfo;
}

Source File: HiveClientWrapper.java From pxf with Apache License 2.0

5 votes

private String getSerdeParameter(StorageDescriptor sd, String parameterKey) {
    String parameterValue = null;
    if (sd != null && sd.getSerdeInfo() != null && sd.getSerdeInfo().getParameters() != null && sd.getSerdeInfo().getParameters().get(parameterKey) != null) {
        parameterValue = sd.getSerdeInfo().getParameters().get(parameterKey);
    }

    return parameterValue;
}

Source File: HiveConnectorFastPartitionService.java From metacat with Apache License 2.0

5 votes

private void copyTableSdToPartitionInfoSd(final PartitionInfo partitionInfo, final Table table) {
    StorageInfo sd = partitionInfo.getSerde();
    //
    // Partitions can be provided in the request without the storage information.
    //
    if (sd == null) {
        sd = new StorageInfo();
        partitionInfo.setSerde(sd);
    }
    final StorageDescriptor tableSd = table.getSd();

    if (StringUtils.isBlank(sd.getInputFormat())) {
        sd.setInputFormat(tableSd.getInputFormat());
    }
    if (StringUtils.isBlank(sd.getOutputFormat())) {
        sd.setOutputFormat(tableSd.getOutputFormat());
    }
    if (sd.getParameters() == null || sd.getParameters().isEmpty()) {
        sd.setParameters(tableSd.getParameters());
    }
    final SerDeInfo tableSerde = tableSd.getSerdeInfo();
    if (tableSerde != null) {
        if (StringUtils.isBlank(sd.getSerializationLib())) {
            sd.setSerializationLib(tableSerde.getSerializationLib());
        }
        if (sd.getSerdeInfoParameters() == null || sd.getSerdeInfoParameters().isEmpty()) {
            sd.setSerdeInfoParameters(tableSerde.getParameters());
        }
    }
}

Source File: HiveConnectorPartitionService.java From metacat with Apache License 2.0

5 votes

private void copyTableSdToPartitionSd(final List<Partition> hivePartitions, final Table table) {
    //
    // Update the partition info based on that of the table.
    //
    for (Partition partition : hivePartitions) {
        final StorageDescriptor sd = partition.getSd();
        final StorageDescriptor tableSdCopy = table.getSd().deepCopy();
        if (tableSdCopy.getSerdeInfo() == null) {
            final SerDeInfo serDeInfo = new SerDeInfo(null, null, new HashMap<>());
            tableSdCopy.setSerdeInfo(serDeInfo);
        }

        tableSdCopy.setLocation(sd.getLocation());
        if (!Strings.isNullOrEmpty(sd.getInputFormat())) {
            tableSdCopy.setInputFormat(sd.getInputFormat());
        }
        if (!Strings.isNullOrEmpty(sd.getOutputFormat())) {
            tableSdCopy.setOutputFormat(sd.getOutputFormat());
        }
        if (sd.getParameters() != null && !sd.getParameters().isEmpty()) {
            tableSdCopy.setParameters(sd.getParameters());
        }
        if (sd.getSerdeInfo() != null) {
            if (!Strings.isNullOrEmpty(sd.getSerdeInfo().getName())) {
                tableSdCopy.getSerdeInfo().setName(sd.getSerdeInfo().getName());
            }
            if (!Strings.isNullOrEmpty(sd.getSerdeInfo().getSerializationLib())) {
                tableSdCopy.getSerdeInfo().setSerializationLib(sd.getSerdeInfo().getSerializationLib());
            }
            if (sd.getSerdeInfo().getParameters() != null && !sd.getSerdeInfo().getParameters().isEmpty()) {
                tableSdCopy.getSerdeInfo().setParameters(sd.getSerdeInfo().getParameters());
            }
        }
        partition.setSd(tableSdCopy);
    }
}

Source File: HiveMetaStoreBridge.java From atlas with Apache License 2.0

4 votes

private AtlasEntity toStorageDescEntity(StorageDescriptor storageDesc, String tableQualifiedName, String sdQualifiedName, AtlasObjectId tableId ) throws AtlasHookException {
    AtlasEntity ret = new AtlasEntity(HiveDataTypes.HIVE_STORAGEDESC.getName());

    ret.setRelationshipAttribute(ATTRIBUTE_TABLE, AtlasTypeUtil.getAtlasRelatedObjectId(tableId, RELATIONSHIP_HIVE_TABLE_STORAGE_DESC));
    ret.setAttribute(ATTRIBUTE_QUALIFIED_NAME, sdQualifiedName);
    ret.setAttribute(ATTRIBUTE_PARAMETERS, storageDesc.getParameters());
    ret.setAttribute(ATTRIBUTE_LOCATION, HdfsNameServiceResolver.getPathWithNameServiceID(storageDesc.getLocation()));
    ret.setAttribute(ATTRIBUTE_INPUT_FORMAT, storageDesc.getInputFormat());
    ret.setAttribute(ATTRIBUTE_OUTPUT_FORMAT, storageDesc.getOutputFormat());
    ret.setAttribute(ATTRIBUTE_COMPRESSED, storageDesc.isCompressed());
    ret.setAttribute(ATTRIBUTE_NUM_BUCKETS, storageDesc.getNumBuckets());
    ret.setAttribute(ATTRIBUTE_STORED_AS_SUB_DIRECTORIES, storageDesc.isStoredAsSubDirectories());

    if (storageDesc.getBucketCols().size() > 0) {
        ret.setAttribute(ATTRIBUTE_BUCKET_COLS, storageDesc.getBucketCols());
    }

    if (storageDesc.getSerdeInfo() != null) {
        SerDeInfo serdeInfo = storageDesc.getSerdeInfo();

        LOG.debug("serdeInfo = {}", serdeInfo);
        // SkewedInfo skewedInfo = storageDesc.getSkewedInfo();

        AtlasStruct serdeInfoStruct = new AtlasStruct(HiveDataTypes.HIVE_SERDE.getName());

        serdeInfoStruct.setAttribute(ATTRIBUTE_NAME, serdeInfo.getName());
        serdeInfoStruct.setAttribute(ATTRIBUTE_SERIALIZATION_LIB, serdeInfo.getSerializationLib());
        serdeInfoStruct.setAttribute(ATTRIBUTE_PARAMETERS, serdeInfo.getParameters());

        ret.setAttribute(ATTRIBUTE_SERDE_INFO, serdeInfoStruct);
    }

    if (CollectionUtils.isNotEmpty(storageDesc.getSortCols())) {
        List<AtlasStruct> sortColsStruct = new ArrayList<>();

        for (Order sortcol : storageDesc.getSortCols()) {
            String hiveOrderName = HiveDataTypes.HIVE_ORDER.getName();
            AtlasStruct colStruct = new AtlasStruct(hiveOrderName);
            colStruct.setAttribute("col", sortcol.getCol());
            colStruct.setAttribute("order", sortcol.getOrder());

            sortColsStruct.add(colStruct);
        }

        ret.setAttribute(ATTRIBUTE_SORT_COLS, sortColsStruct);
    }

    return ret;
}

Source File: HiveConvertersImpl.java From metacat with Apache License 2.0

4 votes

/**
 * {@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;
}

Source File: HiveMetaStoreBridge.java From incubator-atlas with Apache License 2.0

4 votes

public Referenceable fillStorageDesc(StorageDescriptor storageDesc, String tableQualifiedName,
    String sdQualifiedName, Id tableId) throws AtlasHookException {
    LOG.debug("Filling storage descriptor information for {}", storageDesc);

    Referenceable sdReferenceable = new Referenceable(HiveDataTypes.HIVE_STORAGEDESC.getName());
    sdReferenceable.set(AtlasClient.REFERENCEABLE_ATTRIBUTE_NAME, sdQualifiedName);

    SerDeInfo serdeInfo = storageDesc.getSerdeInfo();
    LOG.debug("serdeInfo = {}", serdeInfo);
    // SkewedInfo skewedInfo = storageDesc.getSkewedInfo();

    String serdeInfoName = HiveDataTypes.HIVE_SERDE.getName();
    Struct serdeInfoStruct = new Struct(serdeInfoName);

    serdeInfoStruct.set(AtlasClient.NAME, serdeInfo.getName());
    serdeInfoStruct.set("serializationLib", serdeInfo.getSerializationLib());
    serdeInfoStruct.set(PARAMETERS, serdeInfo.getParameters());

    sdReferenceable.set("serdeInfo", serdeInfoStruct);
    sdReferenceable.set(STORAGE_NUM_BUCKETS, storageDesc.getNumBuckets());
    sdReferenceable
            .set(STORAGE_IS_STORED_AS_SUB_DIRS, storageDesc.isStoredAsSubDirectories());

    List<Struct> sortColsStruct = new ArrayList<>();
    for (Order sortcol : storageDesc.getSortCols()) {
        String hiveOrderName = HiveDataTypes.HIVE_ORDER.getName();
        Struct colStruct = new Struct(hiveOrderName);
        colStruct.set("col", sortcol.getCol());
        colStruct.set("order", sortcol.getOrder());

        sortColsStruct.add(colStruct);
    }
    if (sortColsStruct.size() > 0) {
        sdReferenceable.set("sortCols", sortColsStruct);
    }

    sdReferenceable.set(LOCATION, storageDesc.getLocation());
    sdReferenceable.set("inputFormat", storageDesc.getInputFormat());
    sdReferenceable.set("outputFormat", storageDesc.getOutputFormat());
    sdReferenceable.set("compressed", storageDesc.isCompressed());

    if (storageDesc.getBucketCols().size() > 0) {
        sdReferenceable.set("bucketCols", storageDesc.getBucketCols());
    }

    sdReferenceable.set(PARAMETERS, storageDesc.getParameters());
    sdReferenceable.set("storedAsSubDirectories", storageDesc.isStoredAsSubDirectories());
    sdReferenceable.set(TABLE, tableId);

    return sdReferenceable;
}

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