mathyingzhou commented on a change in pull request #8648:
URL: https://github.com/apache/arrow/pull/8648#discussion_r614551452
##########
File path: cpp/src/arrow/adapters/orc/adapter_util.cc
##########
@@ -315,13 +344,662 @@ Status AppendBatch(const liborc::Type* type,
liborc::ColumnVectorBatch* batch,
return Status::NotImplemented("Not implemented type kind: ", kind);
}
}
+} // namespace orc
+} // namespace adapters
+} // namespace arrow
+
+namespace {
+
+using arrow::internal::checked_cast;
+
+arrow::Status WriteBatch(const arrow::Array& parray, int64_t orc_offset,
+ liborc::ColumnVectorBatch* column_vector_batch,
+ bool normalized = false);
+
+// Make sure children of StructArray have appropriate null.
+std::shared_ptr<arrow::Array> NormalizeArray(const
std::shared_ptr<arrow::Array>& array) {
+ arrow::Type::type kind = array->type_id();
+ switch (kind) {
+ case arrow::Type::type::BOOL:
+ case arrow::Type::type::INT8:
+ case arrow::Type::type::INT16:
+ case arrow::Type::type::INT32:
+ case arrow::Type::type::INT64:
+ case arrow::Type::type::FLOAT:
+ case arrow::Type::type::DOUBLE:
+ case arrow::Type::type::STRING:
+ case arrow::Type::type::LARGE_STRING:
+ case arrow::Type::type::BINARY:
+ case arrow::Type::type::LARGE_BINARY:
+ case arrow::Type::type::FIXED_SIZE_BINARY:
+ case arrow::Type::type::DATE32:
+ case arrow::Type::type::DATE64:
+ case arrow::Type::type::TIMESTAMP:
+ case arrow::Type::type::DECIMAL128: {
+ return array;
+ }
+ case arrow::Type::type::STRUCT: {
+ if (array->null_count() == 0) {
+ return array;
+ } else {
+ auto struct_array =
std::static_pointer_cast<arrow::StructArray>(array);
+ const std::shared_ptr<arrow::Buffer> bitmap =
struct_array->null_bitmap();
+ std::shared_ptr<arrow::DataType> struct_type = struct_array->type();
+ std::size_t size = struct_type->fields().size();
+ std::vector<std::shared_ptr<arrow::Array>> new_children(size, nullptr);
+ for (std::size_t i = 0; i < size; i++) {
+ std::shared_ptr<arrow::Array> child = struct_array->field(i);
+ const std::shared_ptr<arrow::Buffer> child_bitmap =
child->null_bitmap();
Review comment:
Ah actually we don't need it for this release since we don't support NA
or union types. I need to add it when implementing the ORC reader & writer for
union types though.
--
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]
