Examples with DataFileWriter org.apache.avro.file.DataFileWriter used on opensource projects

Example 91 with DataFileWriter

use of org.apache.avro.file.DataFileWriter in project flink by apache.

the class AvroBulkFormatTest method before.

@BeforeEach
public void before() throws IOException {
    tmpFile = Files.createTempFile("avro-bulk-format-test", ".avro").toFile();
    tmpFile.createNewFile();
    FileOutputStream out = new FileOutputStream(tmpFile);
    Schema schema = AvroSchemaConverter.convertToSchema(ROW_TYPE);
    RowDataToAvroConverters.RowDataToAvroConverter converter = RowDataToAvroConverters.createConverter(ROW_TYPE);
    DatumWriter<GenericRecord> datumWriter = new GenericDatumWriter<>(schema);
    DataFileWriter<GenericRecord> dataFileWriter = new DataFileWriter<>(datumWriter);
    dataFileWriter.create(schema, out);
    dataFileWriter.setSyncInterval(64);
    for (RowData rowData : TEST_DATA) {
        dataFileWriter.append((GenericRecord) converter.convert(schema, rowData));
    }
    dataFileWriter.close();
}

Example 92 with DataFileWriter

use of org.apache.avro.file.DataFileWriter in project flink by apache.

the class AvroRecordInputFormatTest method writeTestFile.

public static void writeTestFile(File testFile) throws IOException {
    ArrayList<CharSequence> stringArray = new ArrayList<>();
    stringArray.add(TEST_ARRAY_STRING_1);
    stringArray.add(TEST_ARRAY_STRING_2);
    ArrayList<Boolean> booleanArray = new ArrayList<>();
    booleanArray.add(TEST_ARRAY_BOOLEAN_1);
    booleanArray.add(TEST_ARRAY_BOOLEAN_2);
    HashMap<CharSequence, Long> longMap = new HashMap<>();
    longMap.put(TEST_MAP_KEY1, TEST_MAP_VALUE1);
    longMap.put(TEST_MAP_KEY2, TEST_MAP_VALUE2);
    Address addr = new Address();
    addr.setNum(TEST_NUM);
    addr.setStreet(TEST_STREET);
    addr.setCity(TEST_CITY);
    addr.setState(TEST_STATE);
    addr.setZip(TEST_ZIP);
    User user1 = new User();
    user1.setName(TEST_NAME);
    user1.setFavoriteNumber(256);
    user1.setTypeDoubleTest(123.45d);
    user1.setTypeBoolTest(true);
    user1.setTypeArrayString(stringArray);
    user1.setTypeArrayBoolean(booleanArray);
    user1.setTypeEnum(TEST_ENUM_COLOR);
    user1.setTypeMap(longMap);
    user1.setTypeNested(addr);
    user1.setTypeBytes(ByteBuffer.allocate(10));
    user1.setTypeDate(LocalDate.parse("2014-03-01"));
    user1.setTypeTimeMillis(LocalTime.parse("12:12:12"));
    user1.setTypeTimeMicros(LocalTime.ofSecondOfDay(0).plus(123456L, ChronoUnit.MICROS));
    user1.setTypeTimestampMillis(Instant.parse("2014-03-01T12:12:12.321Z"));
    user1.setTypeTimestampMicros(Instant.ofEpochSecond(0).plus(123456L, ChronoUnit.MICROS));
    // 20.00
    user1.setTypeDecimalBytes(ByteBuffer.wrap(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
    // 20.00
    user1.setTypeDecimalFixed(new Fixed2(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
    // Construct via builder
    User user2 = User.newBuilder().setName("Charlie").setFavoriteColor("blue").setFavoriteNumber(null).setTypeBoolTest(false).setTypeDoubleTest(1.337d).setTypeNullTest(null).setTypeLongTest(1337L).setTypeArrayString(new ArrayList<>()).setTypeArrayBoolean(new ArrayList<>()).setTypeNullableArray(null).setTypeEnum(Colors.RED).setTypeMap(new HashMap<>()).setTypeFixed(null).setTypeUnion(null).setTypeNested(Address.newBuilder().setNum(TEST_NUM).setStreet(TEST_STREET).setCity(TEST_CITY).setState(TEST_STATE).setZip(TEST_ZIP).build()).setTypeBytes(ByteBuffer.allocate(10)).setTypeDate(LocalDate.parse("2014-03-01")).setTypeTimeMillis(LocalTime.parse("12:12:12")).setTypeTimeMicros(LocalTime.ofSecondOfDay(0).plus(123456L, ChronoUnit.MICROS)).setTypeTimestampMillis(Instant.parse("2014-03-01T12:12:12.321Z")).setTypeTimestampMicros(Instant.ofEpochSecond(0).plus(123456L, ChronoUnit.MICROS)).setTypeDecimalBytes(ByteBuffer.wrap(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray())).setTypeDecimalFixed(new Fixed2(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray())).build();
    DatumWriter<User> userDatumWriter = new SpecificDatumWriter<>(User.class);
    DataFileWriter<User> dataFileWriter = new DataFileWriter<>(userDatumWriter);
    dataFileWriter.create(user1.getSchema(), testFile);
    dataFileWriter.append(user1);
    dataFileWriter.append(user2);
    dataFileWriter.close();
}

Example 93 with DataFileWriter

use of org.apache.avro.file.DataFileWriter in project flink by apache.

the class AvroSplittableInputFormatTest method createFiles.

@Before
public void createFiles() throws IOException {
    testFile = File.createTempFile("AvroSplittableInputFormatTest", null);
    ArrayList<CharSequence> stringArray = new ArrayList<>();
    stringArray.add(TEST_ARRAY_STRING_1);
    stringArray.add(TEST_ARRAY_STRING_2);
    ArrayList<Boolean> booleanArray = new ArrayList<>();
    booleanArray.add(TEST_ARRAY_BOOLEAN_1);
    booleanArray.add(TEST_ARRAY_BOOLEAN_2);
    HashMap<CharSequence, Long> longMap = new HashMap<>();
    longMap.put(TEST_MAP_KEY1, TEST_MAP_VALUE1);
    longMap.put(TEST_MAP_KEY2, TEST_MAP_VALUE2);
    Address addr = new Address();
    addr.setNum(TEST_NUM);
    addr.setStreet(TEST_STREET);
    addr.setCity(TEST_CITY);
    addr.setState(TEST_STATE);
    addr.setZip(TEST_ZIP);
    User user1 = new User();
    user1.setName(TEST_NAME);
    user1.setFavoriteNumber(256);
    user1.setTypeDoubleTest(123.45d);
    user1.setTypeBoolTest(true);
    user1.setTypeArrayString(stringArray);
    user1.setTypeArrayBoolean(booleanArray);
    user1.setTypeEnum(TEST_ENUM_COLOR);
    user1.setTypeMap(longMap);
    user1.setTypeNested(addr);
    user1.setTypeBytes(ByteBuffer.allocate(10));
    user1.setTypeDate(LocalDate.parse("2014-03-01"));
    user1.setTypeTimeMillis(LocalTime.parse("12:12:12"));
    user1.setTypeTimeMicros(LocalTime.ofSecondOfDay(0).plus(123456L, ChronoUnit.MICROS));
    user1.setTypeTimestampMillis(Instant.parse("2014-03-01T12:12:12.321Z"));
    user1.setTypeTimestampMicros(Instant.ofEpochSecond(0).plus(123456L, ChronoUnit.MICROS));
    // 20.00
    user1.setTypeDecimalBytes(ByteBuffer.wrap(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
    // 20.00
    user1.setTypeDecimalFixed(new Fixed2(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
    // Construct via builder
    User user2 = User.newBuilder().setName(TEST_NAME).setFavoriteColor("blue").setFavoriteNumber(null).setTypeBoolTest(false).setTypeDoubleTest(1.337d).setTypeNullTest(null).setTypeLongTest(1337L).setTypeArrayString(new ArrayList<>()).setTypeArrayBoolean(new ArrayList<>()).setTypeNullableArray(null).setTypeEnum(Colors.RED).setTypeMap(new HashMap<>()).setTypeFixed(new Fixed16()).setTypeUnion(123L).setTypeNested(Address.newBuilder().setNum(TEST_NUM).setStreet(TEST_STREET).setCity(TEST_CITY).setState(TEST_STATE).setZip(TEST_ZIP).build()).setTypeBytes(ByteBuffer.allocate(10)).setTypeDate(LocalDate.parse("2014-03-01")).setTypeTimeMillis(LocalTime.parse("12:12:12")).setTypeTimeMicros(LocalTime.ofSecondOfDay(0).plus(123456L, ChronoUnit.MICROS)).setTypeTimestampMillis(Instant.parse("2014-03-01T12:12:12.321Z")).setTypeTimestampMicros(Instant.ofEpochSecond(0).plus(123456L, ChronoUnit.MICROS)).setTypeDecimalBytes(ByteBuffer.wrap(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray())).setTypeDecimalFixed(new Fixed2(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray())).build();
    DatumWriter<User> userDatumWriter = new SpecificDatumWriter<>(User.class);
    DataFileWriter<User> dataFileWriter = new DataFileWriter<>(userDatumWriter);
    dataFileWriter.create(user1.getSchema(), testFile);
    dataFileWriter.append(user1);
    dataFileWriter.append(user2);
    Random rnd = new Random(1337);
    for (int i = 0; i < NUM_RECORDS - 2; i++) {
        User user = new User();
        user.setName(TEST_NAME + rnd.nextInt());
        user.setFavoriteNumber(rnd.nextInt());
        user.setTypeDoubleTest(rnd.nextDouble());
        user.setTypeBoolTest(true);
        user.setTypeArrayString(stringArray);
        user.setTypeArrayBoolean(booleanArray);
        user.setTypeEnum(TEST_ENUM_COLOR);
        user.setTypeMap(longMap);
        Address address = new Address();
        address.setNum(TEST_NUM);
        address.setStreet(TEST_STREET);
        address.setCity(TEST_CITY);
        address.setState(TEST_STATE);
        address.setZip(TEST_ZIP);
        user.setTypeNested(address);
        user.setTypeBytes(ByteBuffer.allocate(10));
        user.setTypeDate(LocalDate.parse("2014-03-01"));
        user.setTypeTimeMillis(LocalTime.parse("12:12:12"));
        user.setTypeTimeMicros(LocalTime.ofSecondOfDay(0).plus(123456L, ChronoUnit.MICROS));
        user.setTypeTimestampMillis(Instant.parse("2014-03-01T12:12:12.321Z"));
        user.setTypeTimestampMicros(Instant.ofEpochSecond(0).plus(123456L, ChronoUnit.MICROS));
        // 20.00
        user.setTypeDecimalBytes(ByteBuffer.wrap(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
        // 20.00
        user.setTypeDecimalFixed(new Fixed2(BigDecimal.valueOf(2000, 2).unscaledValue().toByteArray()));
        dataFileWriter.append(user);
    }
    dataFileWriter.close();
}

Example 94 with DataFileWriter

use of org.apache.avro.file.DataFileWriter in project druid by druid-io.

the class AvroHadoopInputRowParserTest method createAvroFile.

public static File createAvroFile(GenericRecord datum) throws IOException {
    final File tmpDir = FileUtils.createTempDir();
    File someAvroDatumFile = new File(tmpDir, "someAvroDatum.avro");
    try (DataFileWriter<GenericRecord> dataFileWriter = new DataFileWriter<>(new SpecificDatumWriter<>())) {
        dataFileWriter.create(SomeAvroDatum.getClassSchema(), someAvroDatumFile);
        dataFileWriter.append(datum);
    }
    return someAvroDatumFile;
}

Example 95 with DataFileWriter

use of org.apache.avro.file.DataFileWriter in project drill by apache.

the class AvroDataGenerator method generateDateTimeData.

public String generateDateTimeData(LocalDateTime dateTime) throws Exception {
    File file = File.createTempFile("avro-date-time-test", ".avro", dirTestWatcher.getRootDir());
    Schema timestampMillis = LogicalTypes.timestampMillis().addToSchema(SchemaBuilder.builder().longType());
    Schema timestampMicros = LogicalTypes.timestampMicros().addToSchema(SchemaBuilder.builder().longType());
    Schema date = LogicalTypes.date().addToSchema(SchemaBuilder.builder().intType());
    Schema timeMillis = LogicalTypes.timeMillis().addToSchema(SchemaBuilder.builder().intType());
    Schema timeMicros = LogicalTypes.timeMicros().addToSchema(SchemaBuilder.builder().longType());
    Schema schema = SchemaBuilder.record("rec").fields().name("col_timestamp_millis").type(timestampMillis).noDefault().name("col_timestamp_micros").type(timestampMicros).noDefault().name("col_date").type(date).noDefault().name("col_time_millis").type(timeMillis).noDefault().name("col_time_micros").type(timeMicros).noDefault().endRecord();
    try (DataFileWriter<GenericRecord> writer = new DataFileWriter<>(new GenericDatumWriter<>(schema))) {
        writer.create(schema, file);
        GenericRecord record = new GenericData.Record(schema);
        long timestampMillisValue = dateTime.toInstant(ZoneOffset.UTC).toEpochMilli();
        record.put("col_timestamp_millis", timestampMillisValue);
        record.put("col_timestamp_micros", timestampMillisValue * 1000);
        record.put("col_date", dateTime.toLocalDate().toEpochDay());
        long startOfDayMillis = dateTime.toLocalDate().atStartOfDay().toInstant(ZoneOffset.UTC).toEpochMilli();
        long timeMillisValue = timestampMillisValue - startOfDayMillis;
        record.put("col_time_millis", timeMillisValue);
        record.put("col_time_micros", timeMillisValue * 1000);
        writer.append(record);
    }
    return file.getName();
}