org.apache.iceberg.types.Types.StringType Java Examples

@Test
public void convertPrimitive() {
  Schema iceberg = new Schema(
      Types.NestedField.optional(0, INTEGER_FIELD, IntegerType.get()),
      Types.NestedField.optional(1, BOOLEAN_FIELD, BooleanType.get()),
      Types.NestedField.required(2, DOUBLE_FIELD, DoubleType.get()),
      Types.NestedField.required(3, STRING_FIELD, StringType.get()),
      Types.NestedField.optional(4, DATE_FIELD, DateType.get()),
      Types.NestedField.optional(5, TIMESTAMP_FIELD, TimestampType.withZone()),
      Types.NestedField.optional(6, LONG_FIELD, LongType.get()),
      Types.NestedField.optional(7, FLOAT_FIELD, FloatType.get()),
      Types.NestedField.optional(8, TIME_FIELD, TimeType.get()),
      Types.NestedField.optional(9, BINARY_FIELD, Types.BinaryType.get()),
      Types.NestedField.optional(10, DECIMAL_FIELD, Types.DecimalType.of(1, 1)),
      Types.NestedField.optional(12, LIST_FIELD, Types.ListType.ofOptional(13, Types.IntegerType.get())),
      Types.NestedField.required(14, MAP_FIELD, Types.MapType.ofOptional(15, 16,
          StringType.get(), IntegerType.get())),
      Types.NestedField.optional(17, FIXED_WIDTH_BINARY_FIELD, Types.FixedType.ofLength(10)));

  org.apache.arrow.vector.types.pojo.Schema arrow = ArrowSchemaUtil.convert(iceberg);

  validate(iceberg, arrow);
}
 

@Test
public void testPrimitive() throws IOException {
  Schema icebergSchema = new Schema(
      optional(1, "b", BooleanType.get()),
      optional(2, "i", IntegerType.get()),
      optional(3, "l", LongType.get()),
      optional(4, "f", FloatType.get()),
      optional(5, "d", DoubleType.get()),
      optional(6, "dec", DecimalType.of(0, 2)),
      optional(7, "s", StringType.get()),
      optional(8, "bi", BinaryType.get())
  );

  ResourceSchema pigSchema = SchemaUtil.convert(icebergSchema);
  assertEquals(
      "b:boolean,i:int,l:long,f:float,d:double,dec:bigdecimal,s:chararray,bi:bytearray", pigSchema.toString());
}
 

@Test
public void complexNested() throws IOException {
  convertToPigSchema(new Schema(
      optional(1, "t", StructType.of(
          optional(2, "b", ListType.ofOptional(3, StructType.of(
              optional(4, "i", IntegerType.get()),
              optional(5, "s", StringType.get())
          )))
      )),
      optional(6, "m1", MapType.ofOptional(7, 8, StringType.get(), StructType.of(
          optional(9, "b", ListType.ofOptional(10, BinaryType.get())),
          optional(11, "m2", MapType.ofOptional(12, 13, StringType.get(), IntegerType.get()))
      ))),
      optional(14, "b1", ListType.ofOptional(15,
          MapType.ofOptional(16, 17, StringType.get(),
              ListType.ofOptional(18, FloatType.get()))))
  ), "t:(b:{(i:int,s:chararray)}),m1:[(b:{(bytearray)},m2:[int])],b1:{([{(float)}])}", "");
}
 

@Test
public void testTupleInMap() throws IOException {
  Schema icebergSchema = new Schema(
      optional(
          1, "nested_list",
          MapType.ofOptional(
              2, 3,
              StringType.get(),
              ListType.ofOptional(
                  4, StructType.of(
                      required(5, "id", LongType.get()),
                      optional(6, "data", StringType.get()))))));

  ResourceSchema pigSchema = SchemaUtil.convert(icebergSchema);
  // The output should contain a nested struct within a list within a map, I think.
  assertEquals("nested_list:[{(id:long,data:chararray)}]", pigSchema.toString());
}
 

@Test
public void doubleWrappingTuples() throws IOException {
  // struct<a:array<struct<b:string>>> -> (a:{(b:chararray)})
  convertToPigSchema(
      new Schema(
          StructType.of(
              required(1, "a",
                  ListType.ofRequired(2, StructType.of(required(3, "b", StringType.get()))))
          ).fields()),
      "a:{(b:chararray)}",
      "A tuple inside a bag should not be double wrapped");
  // struct<a:array<boolean>> -> "(a:{(boolean)})
  convertToPigSchema(
      new Schema(StructType.of(required(1, "a", ListType.ofRequired(2, BooleanType.get()))).fields()),
      "a:{(boolean)}",
      "boolean (or anything non-tuple) element inside a bag should be wrapped inside a tuple"
  );
}
 

@Test
public void testToStringMatchesSpecification()
{
    assertEquals(Transforms.identity(StringType.get()).toString(), "identity");
    assertEquals(Transforms.bucket(StringType.get(), 13).toString(), "bucket[13]");
    assertEquals(Transforms.truncate(StringType.get(), 19).toString(), "truncate[19]");
    assertEquals(Transforms.year(DateType.get()).toString(), "year");
    assertEquals(Transforms.month(DateType.get()).toString(), "month");
    assertEquals(Transforms.day(DateType.get()).toString(), "day");
    assertEquals(Transforms.hour(TimestampType.withoutZone()).toString(), "hour");
}
 

private static PartitionSpec partitionSpec(Consumer<PartitionSpec.Builder> consumer)
{
    Schema schema = new Schema(
            NestedField.required(1, "order_key", LongType.get()),
            NestedField.required(2, "ts", TimestampType.withoutZone()),
            NestedField.required(3, "price", DoubleType.get()),
            NestedField.optional(4, "comment", StringType.get()),
            NestedField.optional(5, "notes", ListType.ofRequired(6, StringType.get())));

    PartitionSpec.Builder builder = PartitionSpec.builderFor(schema);
    consumer.accept(builder);
    return builder.build();
}
 

@Test
public void convertComplex() {
  Schema iceberg = new Schema(
      Types.NestedField.optional(0, "m", MapType.ofOptional(
          1, 2, StringType.get(),
          LongType.get())
      ),
      Types.NestedField.required(3, "m2", MapType.ofOptional(
          4, 5, StringType.get(),
          ListType.ofOptional(6, TimestampType.withoutZone()))
      )
  );
  org.apache.arrow.vector.types.pojo.Schema arrow = ArrowSchemaUtil.convert(iceberg);
  Assert.assertEquals(iceberg.columns().size(), arrow.getFields().size());
}
 

@Test
public void testComplex() throws IOException {
  convertToPigSchema(
      new Schema(
          optional(1, "bag", ListType.ofOptional(2, BooleanType.get())),
          optional(3, "map", MapType.ofOptional(4, 5, StringType.get(), DoubleType.get())),
          optional(6, "tuple",
              StructType.of(optional(7, "i", IntegerType.get()), optional(8, "f", FloatType.get())))
      ), "bag:{(boolean)},map:[double],tuple:(i:int,f:float)", null
  );
}
 

@Test
public void nestedMaps() throws IOException {
  convertToPigSchema(new Schema(
      optional(1, "nested",
          MapType.ofOptional(2, 3, StringType.get(),
              MapType.ofOptional(4, 5, StringType.get(),
                  MapType.ofOptional(6, 7, StringType.get(), DecimalType.of(10, 2)))))
  ), "nested:[[[bigdecimal]]]", "");
}
 

@Test
public void nestedTuples() throws IOException {
  convertToPigSchema(new Schema(
      optional(1, "first", StructType.of(
          optional(2, "second", StructType.of(
              optional(3, "third", StructType.of(
                  optional(4, "val", StringType.get())
              ))
          ))
      ))
  ), "first:(second:(third:(val:chararray)))", "");
}
 

@Test
public void mapConversions() throws IOException {
  // consistent behavior for maps conversions. The below test case, correctly does not specify map key types
  convertToPigSchema(
      new Schema(
          required(
              1, "a",
              MapType.ofRequired(
                  2, 3,
                  StringType.get(),
                  ListType.ofRequired(
                      4, StructType.of(
                          required(5, "b", LongType.get()),
                          required(6, "c", StringType.get())))))),
      "a:[{(b:long,c:chararray)}]",
      "We do not specify the map key type here");
  // struct<a:map<string,map<string,double>>> -> (a:[[double]])
  // As per https://pig.apache.org/docs/latest/basic.html#map-schema. It seems that
  // we  only need to specify value type as keys are always of type chararray
  convertToPigSchema(
      new Schema(
          StructType.of(
              required(1, "a", MapType.ofRequired(
                  2, 3,
                  StringType.get(),
                  MapType.ofRequired(4, 5, StringType.get(), DoubleType.get())))
          ).fields()),
      "a:[[double]]",
      "A map key type does not need to be specified");
}
 

@Test
public void testLongInBag() throws IOException {
  Schema icebergSchema = new Schema(
      optional(
          1, "nested_list",
          MapType.ofOptional(
              2, 3,
              StringType.get(),
              ListType.ofRequired(5, LongType.get()))));
  SchemaUtil.convert(icebergSchema);
}
 

@Test
public void testMetricsForListAndMapElements() throws IOException {
  StructType structType = StructType.of(
      required(1, "leafIntCol", IntegerType.get()),
      optional(2, "leafStringCol", StringType.get())
  );
  Schema schema = new Schema(
      optional(3, "intListCol", ListType.ofRequired(4, IntegerType.get())),
      optional(5, "mapCol", MapType.ofRequired(6, 7, StringType.get(), structType))
  );

  Record record = GenericRecord.create(schema);
  record.setField("intListCol", Lists.newArrayList(10, 11, 12));
  Record struct = GenericRecord.create(structType);
  struct.setField("leafIntCol", 1);
  struct.setField("leafStringCol", "BBB");
  Map<String, Record> map = Maps.newHashMap();
  map.put("4", struct);
  record.set(1, map);

  InputFile recordsFile = writeRecords(schema, record);

  Metrics metrics = getMetrics(recordsFile);
  Assert.assertEquals(1L, (long) metrics.recordCount());
  if (fileFormat() != FileFormat.ORC) {
    assertCounts(1, 1L, 0L, metrics);
    assertCounts(2, 1L, 0L, metrics);
    assertCounts(4, 3L, 0L, metrics);
    assertCounts(6, 1L, 0L, metrics);
  } else {
    assertCounts(1, null, null, metrics);
    assertCounts(2, null, null, metrics);
    assertCounts(4, null, null, metrics);
    assertCounts(6, null, null, metrics);
  }
  assertBounds(1, IntegerType.get(), null, null, metrics);
  assertBounds(2, StringType.get(), null, null, metrics);
  assertBounds(4, IntegerType.get(), null, null, metrics);
  assertBounds(6, StringType.get(), null, null, metrics);
  assertBounds(7, structType, null, null, metrics);
}