pitrou commented on code in PR #37100:
URL: https://github.com/apache/arrow/pull/37100#discussion_r1299914912
##########
cpp/src/arrow/compute/kernels/aggregate_basic_internal.h:
##########
@@ -891,6 +891,105 @@ struct NullMinMaxImpl : public ScalarAggregator {
}
};
+template <SimdLevel::type SimdLevel>
+struct DictionaryMinMaxImpl : public ScalarAggregator {
+ using ThisType = DictionaryMinMaxImpl<SimdLevel>;
+
+ DictionaryMinMaxImpl(std::shared_ptr<DataType> out_type,
ScalarAggregateOptions options)
+ : options(std::move(options)),
+ out_type(std::move(out_type)),
+ has_nulls(false),
+ count(0),
+ min(nullptr),
+ max(nullptr) {
+ this->options.min_count = std::max<uint32_t>(1, this->options.min_count);
+ }
+
+ Status Consume(KernelContext*, const ExecSpan& batch) override {
+ if (batch[0].is_scalar()) {
+ return Status::NotImplemented("No min/max implemented for
DictionaryScalar");
+ }
+
+ DictionaryArray arr(batch[0].array.ToArrayData());
+ this->has_nulls = arr.null_count() > 0;
+ this->count += arr.length() - arr.null_count();
+
+ Datum dict_values(arr.dictionary());
+ ARROW_ASSIGN_OR_RAISE(Datum result, MinMax(std::move(dict_values)));
Review Comment:
Hmm, perhaps a more efficient routine can be written specifically for this
(find all dictionary indices actually used in a dictionary array)? Going
through "is_in" and "filter" sounds quite expensive.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
