Java Group.append方法代码示例

import org.apache.parquet.example.data.Group; //导入方法依赖的package包/类
static File generateSparseParquetFile(File parentDir, String filename, int nrows) throws IOException {
  File f = new File(parentDir, filename);

  Configuration conf = new Configuration();
  MessageType schema = parseMessageType(
          "message test { optional int32 int32_field; optional binary string_field (UTF8); required int32 row; optional int32 int32_field2; } ");
  GroupWriteSupport.setSchema(schema, conf);
  SimpleGroupFactory fact = new SimpleGroupFactory(schema);
  ParquetWriter<Group> writer = new ParquetWriter<Group>(new Path(f.getPath()), new GroupWriteSupport(),
          UNCOMPRESSED, 1024, 1024, 512, true, false, ParquetProperties.WriterVersion.PARQUET_2_0, conf);
  try {
    for (int i = 0; i < nrows; i++) {
      Group g = fact.newGroup();
      if (i % 10 == 0) { g = g.append("int32_field", i); }
      if (i % 10 == 0) { g = g.append("string_field", "CAT_" + (i % 10)); }
      if (i % 10 == 0) { g = g.append("int32_field2", i); }
      writer.write(g.append("row", i));
    }
  } finally {
    writer.close();
  }
  return f;
}
 

import org.apache.parquet.example.data.Group; //导入方法依赖的package包/类
private static void writeFile(File out, Configuration conf, boolean useSchema2) throws IOException {
  if (!useSchema2) {
    GroupWriteSupport.setSchema(schema, conf);
  } else {
    GroupWriteSupport.setSchema(schema2, conf);
  }
  SimpleGroupFactory f = new SimpleGroupFactory(schema);

  Map<String, String> extraMetaData = new HashMap<String, String>();
  extraMetaData.put("schema_num", useSchema2 ? "2" : "1" );

  ParquetWriter<Group> writer = ExampleParquetWriter
      .builder(new Path(out.getAbsolutePath()))
      .withConf(conf)
      .withExtraMetaData(extraMetaData)
      .build();

    for (int i = 0; i < 1000; i++) {
      Group g = f.newGroup()
          .append("binary_field", "test" + i)
          .append("int32_field", i)
          .append("int64_field", (long) i)
          .append("boolean_field", i % 2 == 0)
          .append("float_field", (float) i)
          .append("double_field", (double)i)
          .append("flba_field", "foo");

      if (!useSchema2) {
        g = g.append("int96_field", Binary.fromConstantByteArray(new byte[12]));
      }

      writer.write(g);
    }
    writer.close();
}
 

import org.apache.parquet.example.data.Group; //导入方法依赖的package包/类
public static SimpleGroup groupFromUser(User user) {
  SimpleGroup root = new SimpleGroup(schema);
  root.append("id", user.getId());

  if (user.getName() != null) {
    root.append("name", user.getName());
  }

  if (user.getPhoneNumbers() != null) {
    Group phoneNumbers = root.addGroup("phoneNumbers");
    for (PhoneNumber number : user.getPhoneNumbers()) {
      Group phone = phoneNumbers.addGroup("phone");
      phone.append("number", number.getNumber());
      if (number.getKind() != null) {
        phone.append("kind", number.getKind());
      }
    }
  }

  if (user.getLocation() != null) {
    Group location = root.addGroup("location");
    if (user.getLocation().getLon() != null) {
      location.append("lon", user.getLocation().getLon());
    }
    if (user.getLocation().getLat() != null) {
      location.append("lat", user.getLocation().getLat());
    }
  }
  return root;
}
 

import org.apache.parquet.example.data.Group; //导入方法依赖的package包/类
@Override
public void write(ParquetWriter<Group> writer) throws IOException {
  for (int index = 0; index < recordCount; index++) {
    Group group = new SimpleGroup(super.schema);

    for (int column = 0, columnCnt = schema.getFieldCount(); column < columnCnt; ++column) {
      Type type = schema.getType(column);
      RandomValueGenerator<?> generator = randomGenerators.get(column);
      if (type.isRepetition(OPTIONAL) && generator.shouldGenerateNull()) {
        continue;
      }
      switch (type.asPrimitiveType().getPrimitiveTypeName()) {
      case BINARY:
      case FIXED_LEN_BYTE_ARRAY:
      case INT96:
        group.append(type.getName(), ((RandomBinaryBase<?>) generator).nextBinaryValue());
        break;
      case INT32:
        group.append(type.getName(), (Integer) generator.nextValue());
        break;
      case INT64:
        group.append(type.getName(), (Long) generator.nextValue());
        break;
      case FLOAT:
        group.append(type.getName(), (Float) generator.nextValue());
        break;
      case DOUBLE:
        group.append(type.getName(), (Double) generator.nextValue());
        break;
      case BOOLEAN:
        group.append(type.getName(), (Boolean) generator.nextValue());
        break;
      }
    }
    writer.write(group);
  }
}