edgarRd commented on a change in pull request #1208:
URL: https://github.com/apache/iceberg/pull/1208#discussion_r459805494
##########
File path: orc/src/test/java/org/apache/iceberg/orc/TestORCSchemaUtil.java
##########
@@ -269,4 +278,68 @@ public void testSkipNonIcebergColumns() {
);
assertEquals("Schemas must match.", expectedSchema2.asStruct(),
icebergSchema2.asStruct());
}
+
+ @Test
+ public void testHasIds() {
+ Schema schema = new Schema(
+ optional(1, "data", Types.StructType.of(
+ optional(10, "entries", Types.MapType.ofOptional(11, 12,
Types.StringType.get(), Types.DateType.get()))
+ )),
+ optional(2, "intCol", Types.IntegerType.get()),
+ optional(3, "longCol", Types.LongType.get()),
+ optional(4, "listCol", Types.ListType.ofOptional(40,
Types.DoubleType.get()))
+ );
+
+ TypeDescription orcSchema =
ORCSchemaUtil.removeIds(ORCSchemaUtil.convert(schema));
+ assertFalse("Should not have Ids", ORCSchemaUtil.hasIds(orcSchema));
+
+ TypeDescription map2Col =
TypeDescription.createMap(TypeDescription.createString(),
TypeDescription.createBinary());
+ map2Col.setAttribute(ICEBERG_ID_ATTRIBUTE, "4");
+ orcSchema.addField("map2Col", map2Col);
+ assertTrue("Should have Ids after adding one type with Id",
ORCSchemaUtil.hasIds(orcSchema));
+ }
+
+ @Test
+ public void tetsAssignIdsByNameMapping() {
+ Types.StructType structType = Types.StructType.of(
+ required(0, "id", Types.LongType.get()),
+ optional(1, "list_of_maps",
+ Types.ListType.ofOptional(2, Types.MapType.ofOptional(3, 4,
+ Types.StringType.get(),
+ SUPPORTED_PRIMITIVES))),
+ optional(5, "map_of_lists",
+ Types.MapType.ofOptional(6, 7,
+ Types.StringType.get(),
+ Types.ListType.ofOptional(8, SUPPORTED_PRIMITIVES))),
+ required(9, "list_of_lists",
+ Types.ListType.ofOptional(10, Types.ListType.ofOptional(11,
SUPPORTED_PRIMITIVES))),
+ required(12, "map_of_maps",
+ Types.MapType.ofOptional(13, 14,
+ Types.StringType.get(),
+ Types.MapType.ofOptional(15, 16,
+ Types.StringType.get(),
+ SUPPORTED_PRIMITIVES))),
+ required(17, "list_of_struct_of_nested_types",
Types.ListType.ofOptional(19, Types.StructType.of(
+ Types.NestedField.required(20, "m1", Types.MapType.ofOptional(21,
22,
+ Types.StringType.get(),
+ SUPPORTED_PRIMITIVES)),
+ Types.NestedField.optional(23, "l1", Types.ListType.ofRequired(24,
SUPPORTED_PRIMITIVES)),
+ Types.NestedField.required(25, "l2", Types.ListType.ofRequired(26,
SUPPORTED_PRIMITIVES)),
+ Types.NestedField.optional(27, "m2", Types.MapType.ofOptional(28,
29,
+ Types.StringType.get(),
+ SUPPORTED_PRIMITIVES))
+ )))
+ );
+
+ Schema schema = new Schema(TypeUtil.assignFreshIds(structType, new
AtomicInteger(0)::incrementAndGet)
+ .asStructType().fields());
+
+ NameMapping nameMapping = MappingUtil.create(schema);
+ TypeDescription typeDescriptionWithIds = ORCSchemaUtil.convert(schema);
+ TypeDescription typeDescriptionWithIdsFromNameMapping = ORCSchemaUtil
+ .applyNameMapping(ORCSchemaUtil.removeIds(typeDescriptionWithIds),
nameMapping);
+
+ assertTrue("TypeDescription schemas should be equal, not comparing
Attributes",
+ typeDescriptionWithIds.equals(typeDescriptionWithIdsFromNameMapping,
false /* checkAttributes */));
Review comment:
Name mapping - using an Iceberg schema - will select the columns from an
ORC schema that match the column names from Iceberg, skipping anything that
does not match and adds the ids found to produce the reading schema. So, I
think the effect of name mapping is still being tested in this case since we're
checking the the `TypeDescription` generated with name mapping is the expected
one. That being said, I've added a check for the IDs as well.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
