org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat Java Examples

public static RecordWriter createRecordWriter(Path target, JobConf conf, Properties properties, String outputFormatName, ConnectorSession session)
{
    try {
        boolean compress = HiveConf.getBoolVar(conf, COMPRESSRESULT);
        if (outputFormatName.equals(MapredParquetOutputFormat.class.getName())) {
            return createParquetWriter(target, conf, properties, session);
        }
        if (outputFormatName.equals(HiveIgnoreKeyTextOutputFormat.class.getName())) {
            return new TextRecordWriter(target, conf, properties, compress);
        }
        if (outputFormatName.equals(HiveSequenceFileOutputFormat.class.getName())) {
            return new SequenceFileRecordWriter(target, conf, Text.class, compress);
        }
        if (outputFormatName.equals(AvroContainerOutputFormat.class.getName())) {
            return new AvroRecordWriter(target, conf, compress, properties);
        }
        Object writer = Class.forName(outputFormatName).getConstructor().newInstance();
        return ((HiveOutputFormat<?, ?>) writer).getHiveRecordWriter(conf, target, Text.class, compress, properties, Reporter.NULL);
    }
    catch (IOException | ReflectiveOperationException e) {
        throw new PrestoException(HIVE_WRITER_DATA_ERROR, e);
    }
}
 

public static String getStoreType(String fileFormat) {
  Preconditions.checkNotNull(fileFormat);

  String[] fileFormatArrary = fileFormat.split("\\.");
  if(fileFormatArrary.length < 1) {
    throw new CatalogException("Hive file output format is wrong. - file output format:" + fileFormat);
  }

  String outputFormatClass = fileFormatArrary[fileFormatArrary.length-1];
  if(outputFormatClass.equals(HiveIgnoreKeyTextOutputFormat.class.getSimpleName())) {
    return CatalogProtos.StoreType.CSV.name();
  } else if(outputFormatClass.equals(RCFileOutputFormat.class.getSimpleName())) {
      return CatalogProtos.StoreType.RCFILE.name();
  } else {
    throw new CatalogException("Not supported file output format. - file output format:" + fileFormat);
  }
}
 

@Test
public void testTableUsingRegex() throws Exception {
  TableMeta meta = new TableMeta(BuiltinStorages.REGEX, new KeyValueSet());
  meta.putProperty(StorageConstants.TEXT_REGEX, "([^ ]*)");
  meta.putProperty(StorageConstants.TEXT_REGEX_OUTPUT_FORMAT_STRING, "%1$s");

  org.apache.tajo.catalog.Schema schema = SchemaBuilder.builder()
      .add("c_custkey", TajoDataTypes.Type.TEXT)
      .build();

  TableDesc table = new TableDesc(IdentifierUtil.buildFQName(DB_NAME, CUSTOMER), schema, meta,
      new Path(warehousePath, new Path(DB_NAME, CUSTOMER)).toUri());
  store.createTable(table.getProto());
  assertTrue(store.existTable(DB_NAME, CUSTOMER));

  org.apache.hadoop.hive.ql.metadata.Table hiveTable = store.getHiveTable(DB_NAME, CUSTOMER);
  assertEquals(TextInputFormat.class.getName(), hiveTable.getSd().getInputFormat());
  assertEquals(HiveIgnoreKeyTextOutputFormat.class.getName(), hiveTable.getSd().getOutputFormat());
  assertEquals(RegexSerDe.class.getName(), hiveTable.getSerializationLib());

  TableDesc table1 = new TableDesc(store.getTable(DB_NAME, CUSTOMER));
  assertEquals(table.getName(), table1.getName());
  assertEquals(table.getUri(), table1.getUri());
  assertEquals(table.getSchema().size(), table1.getSchema().size());
  for (int i = 0; i < table.getSchema().size(); i++) {
    assertEquals(table.getSchema().getColumn(i).getSimpleName(), table1.getSchema().getColumn(i).getSimpleName());
  }

  store.dropTable(DB_NAME, CUSTOMER);
}
 

@Test
public void testTableUsingTextFile() throws Exception {
  TableMeta meta = new TableMeta(BuiltinStorages.TEXT, new KeyValueSet());

  org.apache.tajo.catalog.Schema schema = SchemaBuilder.builder()
      .add("c_custkey", TajoDataTypes.Type.INT4)
      .add("c_name", TajoDataTypes.Type.TEXT)
      .add("c_address", TajoDataTypes.Type.TEXT)
      .add("c_nationkey", TajoDataTypes.Type.INT4)
      .add("c_phone", TajoDataTypes.Type.TEXT)
      .add("c_acctbal", TajoDataTypes.Type.FLOAT8)
      .add("c_mktsegment", TajoDataTypes.Type.TEXT)
      .add("c_comment", TajoDataTypes.Type.TEXT)
      .build();

  TableDesc table = new TableDesc(IdentifierUtil.buildFQName(DB_NAME, CUSTOMER), schema, meta,
      new Path(warehousePath, new Path(DB_NAME, CUSTOMER)).toUri());
  store.createTable(table.getProto());
  assertTrue(store.existTable(DB_NAME, CUSTOMER));

  StorageFormatDescriptor descriptor = formatFactory.get(IOConstants.TEXTFILE);
  org.apache.hadoop.hive.ql.metadata.Table hiveTable = store.getHiveTable(DB_NAME, CUSTOMER);
  assertEquals(descriptor.getInputFormat(), hiveTable.getSd().getInputFormat());
  //IgnoreKeyTextOutputFormat was deprecated
  assertEquals(HiveIgnoreKeyTextOutputFormat.class.getName(), hiveTable.getSd().getOutputFormat());

  TableDesc table1 = new TableDesc(store.getTable(DB_NAME, CUSTOMER));
  assertEquals(table.getName(), table1.getName());
  assertEquals(table.getUri(), table1.getUri());
  assertEquals(table.getSchema().size(), table1.getSchema().size());
  for (int i = 0; i < table.getSchema().size(); i++) {
    assertEquals(table.getSchema().getColumn(i).getSimpleName(), table1.getSchema().getColumn(i).getSimpleName());
  }

  assertEquals(StringEscapeUtils.escapeJava(StorageConstants.DEFAULT_FIELD_DELIMITER),
      table1.getMeta().getProperty(StorageConstants.TEXT_DELIMITER));

  Map<String, String> expected = getProperties(DB_NAME, CUSTOMER);
  Map<String, String> toSet = new ImmutableMap.Builder<String, String>()
      .put("key1", "value1")
      .put("key2", "value2")
      .build();
  expected.putAll(toSet);

  setProperty(DB_NAME, CUSTOMER, toSet);
  Map<String, String> actual = getProperties(DB_NAME, CUSTOMER);
  assertEquals(actual.get(StorageConstants.TEXT_DELIMITER), expected.get(StorageConstants.TEXT_DELIMITER));
  assertEquals(actual.get("key1"), expected.get("key1"));
  assertEquals(actual.get("key2"), expected.get("key2"));

  Set<String> toUnset = Sets.newHashSet("key2", "key3");
  for (String key : toUnset) {
    expected.remove(key);
  }
  unSetProperty(DB_NAME, CUSTOMER, toUnset);
  actual = getProperties(DB_NAME, CUSTOMER);
  assertEquals(actual.get(StorageConstants.TEXT_DELIMITER), expected.get(StorageConstants.TEXT_DELIMITER));
  assertEquals(actual.get("key1"), expected.get("key1"));
  assertNull(actual.get("key2"));

  store.dropTable(DB_NAME, CUSTOMER);
}
 

@Test
public void testTableWithNullValue() throws Exception {
  KeyValueSet options = new KeyValueSet();
  options.set(StorageConstants.TEXT_DELIMITER, StringEscapeUtils.escapeJava("\u0002"));
  options.set(StorageConstants.TEXT_NULL, StringEscapeUtils.escapeJava("\u0003"));
  TableMeta meta = new TableMeta(BuiltinStorages.TEXT, options);

  org.apache.tajo.catalog.Schema schema = SchemaBuilder.builder()
      .add("s_suppkey", TajoDataTypes.Type.INT4)
      .add("s_name", TajoDataTypes.Type.TEXT)
      .add("s_address", TajoDataTypes.Type.TEXT)
      .add("s_nationkey", TajoDataTypes.Type.INT4)
      .add("s_phone", TajoDataTypes.Type.TEXT)
      .add("s_acctbal", TajoDataTypes.Type.FLOAT8)
      .add("s_comment", TajoDataTypes.Type.TEXT)
      .build();

  TableDesc table = new TableDesc(IdentifierUtil.buildFQName(DB_NAME, SUPPLIER), schema, meta,
      new Path(warehousePath, new Path(DB_NAME, SUPPLIER)).toUri());

  store.createTable(table.getProto());
  assertTrue(store.existTable(DB_NAME, SUPPLIER));

  StorageFormatDescriptor descriptor = formatFactory.get(IOConstants.TEXTFILE);
  org.apache.hadoop.hive.ql.metadata.Table hiveTable = store.getHiveTable(DB_NAME, SUPPLIER);
  assertEquals(descriptor.getInputFormat(), hiveTable.getSd().getInputFormat());
  //IgnoreKeyTextOutputFormat was deprecated
  assertEquals(HiveIgnoreKeyTextOutputFormat.class.getName(), hiveTable.getSd().getOutputFormat());

  TableDesc table1 = new TableDesc(store.getTable(DB_NAME, SUPPLIER));
  assertEquals(table.getName(), table1.getName());
  assertEquals(table.getUri(), table1.getUri());
  assertEquals(table.getSchema().size(), table1.getSchema().size());
  for (int i = 0; i < table.getSchema().size(); i++) {
    assertEquals(table.getSchema().getColumn(i).getSimpleName(), table1.getSchema().getColumn(i).getSimpleName());
  }

  assertEquals(table.getMeta().getProperty(StorageConstants.TEXT_DELIMITER),
      table1.getMeta().getProperty(StorageConstants.TEXT_DELIMITER));

  assertEquals(table.getMeta().getProperty(StorageConstants.TEXT_NULL),
      table1.getMeta().getProperty(StorageConstants.TEXT_NULL));

  assertEquals(table1.getMeta().getProperty(StorageConstants.TEXT_DELIMITER),
      StringEscapeUtils.escapeJava("\u0002"));

  assertEquals(table1.getMeta().getProperty(StorageConstants.TEXT_NULL),
      StringEscapeUtils.escapeJava("\u0003"));

  Map<String, String> expected = getProperties(DB_NAME, SUPPLIER);
  Map<String, String> toSet = new ImmutableMap.Builder<String, String>()
          .put("key1", "value1")
          .put("key2", "value2")
          .build();
  expected.putAll(toSet);

  setProperty(DB_NAME, SUPPLIER, toSet);
  Map<String, String> actual = getProperties(DB_NAME, SUPPLIER);
  assertEquals(actual.get(StorageConstants.TEXT_DELIMITER), expected.get(StorageConstants.TEXT_DELIMITER));
  assertEquals(actual.get("key1"), expected.get("key1"));
  assertEquals(actual.get("key2"), expected.get("key2"));

  Set<String> toUnset = Sets.newHashSet("key2", "key3");
  for (String key : toUnset) {
    expected.remove(key);
  }
  unSetProperty(DB_NAME, SUPPLIER, toUnset);
  actual = getProperties(DB_NAME, SUPPLIER);
  assertEquals(actual.get(StorageConstants.TEXT_DELIMITER), expected.get(StorageConstants.TEXT_DELIMITER));
  assertEquals(actual.get("key1"), expected.get("key1"));
  assertNull(actual.get("key2"));

  store.dropTable(DB_NAME, SUPPLIER);

}