lxy-9602 commented on code in PR #269: URL: https://github.com/apache/paimon-cpp/pull/269#discussion_r3930165111
########## src/paimon/core/realtime/realtime_offset_batch_reader.cpp: ########## @@ -0,0 +1,110 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "paimon/core/realtime/realtime_offset_batch_reader.h" + +#include <cstdint> +#include <memory> +#include <utility> + +#include "arrow/array/array_base.h" +#include "arrow/array/array_nested.h" +#include "arrow/array/array_primitive.h" +#include "arrow/c/bridge.h" +#include "fmt/format.h" +#include "paimon/common/table/special_fields.h" +#include "paimon/common/utils/arrow/arrow_utils.h" +#include "paimon/common/utils/arrow/status_utils.h" +#include "paimon/common/utils/checked_cast.h" +#include "paimon/metrics.h" +#include "paimon/status.h" + +namespace paimon { +RealtimeOffsetBatchReader::RealtimeOffsetBatchReader(std::unique_ptr<BatchReader>&& reader, + const OffsetRange& visible_offsets) + : reader_(std::move(reader)), visible_offsets_(visible_offsets) {} + +Result<BatchReader::ReadBatch> RealtimeOffsetBatchReader::NextBatch() { + return Status::Invalid( + "paimon inner reader RealtimeOffsetBatchReader should use NextBatchWithBitmap"); +} + +Result<BatchReader::ReadBatchWithBitmap> RealtimeOffsetBatchReader::NextBatchWithBitmap() { + while (true) { + PAIMON_ASSIGN_OR_RAISE(ReadBatchWithBitmap batch_with_bitmap, + reader_->NextBatchWithBitmap()); + if (IsEofBatch(batch_with_bitmap)) { + return batch_with_bitmap; + } + auto& [batch, input_bitmap] = batch_with_bitmap; + auto& [c_array, c_schema] = batch; + PAIMON_ASSIGN_OR_RAISE_FROM_ARROW(std::shared_ptr<arrow::Array> arrow_array, + arrow::ImportArray(c_array.get(), c_schema.get())); + if (!arrow_array || arrow_array->type_id() != arrow::Type::STRUCT) { + return Status::Invalid("realtime query batch must be a StructArray"); + } + std::shared_ptr<arrow::StructArray> struct_array = + checked_pointer_cast<arrow::StructArray>(arrow_array); + std::shared_ptr<arrow::Array> offset_field = + struct_array->GetFieldByName(SpecialFields::RealtimeOffset().Name()); + if (!offset_field || offset_field->type_id() != arrow::Type::INT64) { + return Status::Invalid("realtime query batch must contain int64 _REALTIME_OFFSET"); + } + std::shared_ptr<arrow::Int64Array> offsets = + checked_pointer_cast<arrow::Int64Array>(offset_field); + if (offsets->null_count() != 0) { + return Status::Invalid("realtime query offset column contains null"); + } + + RoaringBitmap32 output_bitmap; + for (auto iter = input_bitmap.Begin(); iter != input_bitmap.End(); ++iter) { + const uint32_t row = *iter; + if (static_cast<int64_t>(row) >= offsets->length()) { + return Status::Invalid(fmt::format( + "selected row id {} is out of bounds for realtime query batch length {}", row, + offsets->length())); + } + const int64_t offset = offsets->Value(row); + if (offset >= visible_offsets_.begin && offset < visible_offsets_.end) { + output_bitmap.Add(row); + } + } + if (static_cast<int64_t>(input_bitmap.Cardinality()) != offsets->length()) { Review Comment: Thanks for the comments. This is not an issue in the current implementation, because predicate pushdown is currently applied at the batch level: an entire batch is either kept or skipped, and we do not go through a bitmap path. So this validation is fine for now. Once we support DV and finer-grained pushdown (like pk predicate pushdown), the `Cardinality` check will likely be removed for both PK and append mode. That is planned for the next PR or the one after that. -- 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]
