@Test
public void testAliases1() throws TupleMRException {
TupleMRConfigBuilder b = new TupleMRConfigBuilder();
b.addIntermediateSchema(
new Schema("schema1", Fields.parse("ax:int,bx:string,cx:string,blablax:string,p2:string")));
b.addIntermediateSchema(
new Schema("schema2", Fields.parse("ay:int,cy:string,by:string,blobloy:string,p:string")));
{
Aliases aliases1 = new Aliases();
aliases1.add("a", "ax");
aliases1.add("b", "bx");
aliases1.add("c", "cx");
aliases1.add("blabla", "blablax");
aliases1.add("p", "p2");
b.setFieldAliases("schema1", aliases1);
}
{
Aliases aliases2 = new Aliases();
aliases2.add("a", "ay");
aliases2.add("b", "by");
aliases2.add("c", "cy");
aliases2.add("bloblo", "blobloy");
b.setFieldAliases("schema2", aliases2);
}
b.setGroupByFields("c", "b");
b.setOrderBy(
new OrderBy()
.add("b", Order.ASC)
.add("c", Order.DESC)
.addSchemaOrder(Order.DESC)
.add("a", Order.DESC));
b.setSpecificOrderBy("schema1", new OrderBy().add("blabla", Order.DESC));
b.setCustomPartitionFields("p");
TupleMRConfig config = b.buildConf();
SerializationInfo serInfo = config.getSerializationInfo();
System.out.println(serInfo.getCommonSchema());
System.out.println(serInfo.getPartitionFieldsIndexes());
{
List<SortElement> expectedCommon = new ArrayList<SortElement>();
expectedCommon.add(new SortElement("b", Order.ASC, Criteria.NullOrder.NULL_SMALLEST));
expectedCommon.add(new SortElement("c", Order.DESC, Criteria.NullOrder.NULL_SMALLEST));
Assert.assertEquals(new Criteria(expectedCommon), config.getCommonCriteria());
}
{
List<SortElement> expectedSchema1 = new ArrayList<SortElement>();
expectedSchema1.add(new SortElement("a", Order.DESC, Criteria.NullOrder.NULL_SMALLEST));
expectedSchema1.add(new SortElement("blabla", Order.DESC, Criteria.NullOrder.NULL_SMALLEST));
Assert.assertEquals(new Criteria(expectedSchema1), config.getSpecificOrderBys().get(0));
}
{
List<SortElement> expectedSchema2 = new ArrayList<SortElement>();
expectedSchema2.add(new SortElement("a", Order.DESC, Criteria.NullOrder.NULL_SMALLEST));
Assert.assertEquals(new Criteria(expectedSchema2), config.getSpecificOrderBys().get(1));
}
}
@Test(expected = TupleMRException.class)
public void testAliasesUnknownSchema() throws TupleMRException {
TupleMRConfigBuilder b = new TupleMRConfigBuilder();
b.addIntermediateSchema(
new Schema("schema1", Fields.parse("a:int,b:string,c:string,blabla:string")));
b.setFieldAliases("schemaX", new Aliases().add("bx", "b"));
}
@Test
public void testWithFieldAliases() throws TupleMRException, IOException {
TupleMRConfigBuilder b = new TupleMRConfigBuilder();
Schema schema1 =
new Schema(
"schema1",
Fields.parse("user_id:int,operation:string,age:long,timestamp:int,country:string"));
Schema schema2 =
new Schema("schema2", Fields.parse("id:int,op:string,another_id:int,time:int"));
b.addIntermediateSchema(schema1);
b.addIntermediateSchema(schema2);
b.setFieldAliases("schema1", new Aliases().add("id", "user_id").add("op", "operation"));
b.setFieldAliases("schema2", new Aliases().add("timestamp", "time"));
b.setGroupByFields("id", "op");
b.setOrderBy(
new OrderBy()
.add("op", Order.ASC)
.add("id", Order.DESC)
.addSchemaOrder(Order.DESC)
.add("timestamp", Order.DESC));
b.setSpecificOrderBy("schema1", new OrderBy().add("country", Order.DESC));
TupleMRConfig conf = b.buildConf();
Configuration hconf = new Configuration();
TupleMRConfig.set(conf, hconf);
TupleMRConfig deserConf = TupleMRConfig.get(hconf);
System.out.println(conf);
System.out.println("------------");
System.out.println(deserConf);
Assert.assertEquals(conf, deserConf);
hconf = new Configuration();
TupleMRConfig.set(deserConf, hconf);
TupleMRConfig deserConf2 = TupleMRConfig.get(hconf);
Assert.assertEquals(conf, deserConf2);
}
