pitrou commented on code in PR #46229:
URL: https://github.com/apache/arrow/pull/46229#discussion_r2207419059
##########
cpp/src/arrow/array/array_binary.cc:
##########
@@ -105,6 +111,392 @@
BinaryViewArray::BinaryViewArray(std::shared_ptr<DataType> type, int64_t length,
ArrayData::Make(std::move(type), length, std::move(buffers), null_count,
offset));
}
+namespace {
+
+// TODO Should We move this to bitmap_ops.h and Remove from
compute/kernels/util.s
+Result<std::shared_ptr<Buffer>> GetOrCopyNullBitmapBuffer(const ArrayData&
in_array,
+ MemoryPool* pool) {
+ if (in_array.buffers[0]->data() == nullptr) {
+ return nullptr;
+ } else if (in_array.offset == 0) {
+ return in_array.buffers[0];
+ } else if (in_array.offset % 8 == 0) {
+ return SliceBuffer(in_array.buffers[0], /*offset=*/in_array.offset / 8);
+ } else {
+ // If a non-zero offset, we need to shift the bitmap
+ return internal::CopyBitmap(pool, in_array.buffers[0]->data(),
in_array.offset,
+ in_array.length);
+ }
+}
+
+struct Interval {
+ int64_t start;
+ int64_t end;
+ int32_t offset = -1;
+};
+
+struct IntervalComparator {
+ bool operator()(const Interval& left, const Interval& right) const {
+ return left.start < right.start;
+ }
+};
+
+// inspired from boost::icl::interval_set
+class IntervalMerger {
+ public:
+ using IntervalSet = std::set<Interval, IntervalComparator>;
+ using Iterator = std::set<Interval, IntervalComparator>::iterator;
+
+ void AddInterval(const Interval& interval) {
+ auto [it, is_inserted] = interval_set.insert(interval);
+ if (is_inserted) {
+ JointLeft(it);
+ JoinRight(it);
+ } else {
+ if (it->end < interval.end) {
+ const_cast<int64_t&>(it->end) = interval.end;
+ JoinRight(it);
+ }
+ }
+ }
+
+ int64_t CalculateOffsetAndTotalSize() {
+ int64_t total_size = 0;
+ for (auto& it : interval_set) {
+ const_cast<int32_t&>(it.offset) = static_cast<int32_t>(total_size);
Review Comment:
Why this? It's not obvious to me why this is necessary nor correct.
--
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: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
