/**
* Returns the merger of schemas. The merged schema will include the union all columns. If there
* is a type conflict between columns with the same schemapath but different types, the merged
* schema will contain a Union type.
*
* @param schemas
* @return
*/
public static BatchSchema mergeSchemas(BatchSchema... schemas) {
Map<SchemaPath, Set<MinorType>> typeSetMap = Maps.newLinkedHashMap();
for (BatchSchema s : schemas) {
for (MaterializedField field : s) {
SchemaPath path = field.getPath();
Set<MinorType> currentTypes = typeSetMap.get(path);
if (currentTypes == null) {
currentTypes = Sets.newHashSet();
typeSetMap.put(path, currentTypes);
}
MinorType newType = field.getType().getMinorType();
if (newType == MinorType.MAP || newType == MinorType.LIST) {
throw new RuntimeException(
"Schema change not currently supported for schemas with complex types");
}
if (newType == MinorType.UNION) {
for (MinorType subType : field.getType().getSubTypeList()) {
currentTypes.add(subType);
}
} else {
currentTypes.add(newType);
}
}
}
List<MaterializedField> fields = Lists.newArrayList();
for (SchemaPath path : typeSetMap.keySet()) {
Set<MinorType> types = typeSetMap.get(path);
if (types.size() > 1) {
MajorType.Builder builder =
MajorType.newBuilder().setMinorType(MinorType.UNION).setMode(DataMode.OPTIONAL);
for (MinorType t : types) {
builder.addSubType(t);
}
MaterializedField field = MaterializedField.create(path, builder.build());
fields.add(field);
} else {
MaterializedField field =
MaterializedField.create(path, Types.optional(types.iterator().next()));
fields.add(field);
}
}
SchemaBuilder schemaBuilder = new SchemaBuilder();
BatchSchema s =
schemaBuilder
.addFields(fields)
.setSelectionVectorMode(schemas[0].getSelectionVectorMode())
.build();
return s;
}
