@Test
public void test_getOrcField_union() throws Exception {
final SchemaBuilder.FieldAssembler<Schema> builder =
SchemaBuilder.record("testRecord").namespace("any.data").fields();
builder.name("union").type().unionOf().intType().and().booleanType().endUnion().noDefault();
Schema testSchema = builder.endRecord();
TypeInfo orcType = NiFiOrcUtils.getOrcField(testSchema.getField("union").schema());
assertEquals(
TypeInfoFactory.getUnionTypeInfo(
Arrays.asList(TypeInfoCreator.createInt(), TypeInfoCreator.createBoolean())),
orcType);
}
@Test
public void test_getOrcField_map() throws Exception {
final SchemaBuilder.FieldAssembler<Schema> builder =
SchemaBuilder.record("testRecord").namespace("any.data").fields();
builder.name("map").type().map().values().doubleType().noDefault();
Schema testSchema = builder.endRecord();
TypeInfo orcType = NiFiOrcUtils.getOrcField(testSchema.getField("map").schema());
assertEquals(
TypeInfoFactory.getMapTypeInfo(
TypeInfoCreator.createString(), TypeInfoCreator.createDouble()),
orcType);
}
@Test
public void test_getOrcField_enum() throws Exception {
final SchemaBuilder.FieldAssembler<Schema> builder =
SchemaBuilder.record("testRecord").namespace("any.data").fields();
builder.name("enumField").type().enumeration("enum").symbols("a", "b", "c").enumDefault("a");
Schema testSchema = builder.endRecord();
TypeInfo orcType = NiFiOrcUtils.getOrcField(testSchema.getField("enumField").schema());
assertEquals(TypeInfoCreator.createString(), orcType);
}
@Test
public void test_getOrcField_record() throws Exception {
final SchemaBuilder.FieldAssembler<Schema> builder =
SchemaBuilder.record("testRecord").namespace("any.data").fields();
builder.name("int").type().intType().noDefault();
builder.name("long").type().longType().longDefault(1L);
builder.name("array").type().array().items().stringType().noDefault();
Schema testSchema = builder.endRecord();
TypeInfo orcType = NiFiOrcUtils.getOrcField(testSchema);
assertEquals(
TypeInfoFactory.getStructTypeInfo(
Arrays.asList("int", "long", "array"),
Arrays.asList(
TypeInfoCreator.createInt(),
TypeInfoCreator.createLong(),
TypeInfoFactory.getListTypeInfo(TypeInfoCreator.createString()))),
orcType);
}
@Test
public void test_getOrcField_array() throws Exception {
final SchemaBuilder.FieldAssembler<Schema> builder =
SchemaBuilder.record("testRecord").namespace("any.data").fields();
builder.name("array").type().array().items().longType().noDefault();
Schema testSchema = builder.endRecord();
TypeInfo orcType = NiFiOrcUtils.getOrcField(testSchema.getField("array").schema());
assertEquals(TypeInfoFactory.getListTypeInfo(TypeInfoCreator.createLong()), orcType);
}
public static TypeInfo buildPrimitiveOrcSchema() {
return TypeInfoFactory.getStructTypeInfo(
Arrays.asList("int", "long", "boolean", "float", "double", "bytes", "string"),
Arrays.asList(
TypeInfoCreator.createInt(),
TypeInfoCreator.createLong(),
TypeInfoCreator.createBoolean(),
TypeInfoCreator.createFloat(),
TypeInfoCreator.createDouble(),
TypeInfoCreator.createBinary(),
TypeInfoCreator.createString()));
}
@Test
public void test_getPrimitiveOrcTypeFromPrimitiveAvroType() throws Exception {
// Expected ORC types
TypeInfo[] expectedTypes = {
TypeInfoCreator.createInt(),
TypeInfoCreator.createLong(),
TypeInfoCreator.createBoolean(),
TypeInfoCreator.createFloat(),
TypeInfoCreator.createDouble(),
TypeInfoCreator.createBinary(),
TypeInfoCreator.createString(),
};
Schema testSchema = buildPrimitiveAvroSchema();
List<Schema.Field> fields = testSchema.getFields();
for (int i = 0; i < fields.size(); i++) {
assertEquals(
expectedTypes[i],
NiFiOrcUtils.getPrimitiveOrcTypeFromPrimitiveAvroType(fields.get(i).schema().getType()));
}
}
