emkornfield commented on code in PR #37400: URL: https://github.com/apache/arrow/pull/37400#discussion_r1648419112
########## cpp/src/parquet/bloom_filter_builder.cc: ########## @@ -0,0 +1,158 @@ +// 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. + +// This module defines an abstract interface for iterating through pages in a +// Parquet column chunk within a row group. It could be extended in the future +// to iterate through all data pages in all chunks in a file. + +#include "parquet/bloom_filter_builder.h" + +#include <map> +#include <utility> +#include <vector> + +#include "arrow/io/interfaces.h" + +#include "parquet/bloom_filter.h" +#include "parquet/exception.h" +#include "parquet/metadata.h" +#include "parquet/properties.h" + +namespace parquet::internal { + +namespace { +/// Column encryption for bloom filter is not implemented yet. +class BloomFilterBuilderImpl : public BloomFilterBuilder { + public: + explicit BloomFilterBuilderImpl(const SchemaDescriptor* schema, + const WriterProperties* properties) + : schema_(schema), properties_(properties) {} + /// Append a new row group to host all incoming bloom filters. + void AppendRowGroup() override; + + BloomFilter* GetOrCreateBloomFilter(int32_t column_ordinal) override; + + /// Serialize all bloom filters with header and bitset in the order of row group and + /// column id. The side effect is that it deletes all bloom filters after they have + /// been flushed. + void WriteTo(::arrow::io::OutputStream* sink, BloomFilterLocation* location) override; + + BloomFilterBuilderImpl(const BloomFilterBuilderImpl&) = delete; + BloomFilterBuilderImpl(BloomFilterBuilderImpl&&) = default; + + private: + /// Make sure column ordinal is not out of bound and the builder is in good state. + void CheckState(int32_t column_ordinal) const { + if (finished_) { + throw ParquetException("BloomFilterBuilder is already finished."); + } + if (column_ordinal < 0 || column_ordinal >= schema_->num_columns()) { + throw ParquetException("Invalid column ordinal: ", column_ordinal); + } + if (file_bloom_filters_.empty()) { + throw ParquetException("No row group appended to BloomFilterBuilder."); + } + if (schema_->Column(column_ordinal)->physical_type() == Type::BOOLEAN) { + throw ParquetException("BloomFilterBuilder does not support boolean type."); + } + } + + const SchemaDescriptor* schema_; + const WriterProperties* properties_; + bool finished_ = false; + + using RowGroupBloomFilters = std::map<int32_t, std::unique_ptr<BloomFilter>>; + // Using unique_ptr because the `std::unique_ptr<BloomFilter>` is not copyable. + // MSVC has the issue below: https://github.com/microsoft/STL/issues/1036 + // So we use `std::unique_ptr<std::map<>>` to avoid the issue. + std::vector<std::unique_ptr<RowGroupBloomFilters>> file_bloom_filters_; +}; + +void BloomFilterBuilderImpl::AppendRowGroup() { + if (finished_) { + throw ParquetException( + "Cannot call AppendRowGroup() to finished BloomFilterBuilder."); + } + file_bloom_filters_.emplace_back(std::make_unique<RowGroupBloomFilters>()); +} + +BloomFilter* BloomFilterBuilderImpl::GetOrCreateBloomFilter(int32_t column_ordinal) { + CheckState(column_ordinal); + const ColumnDescriptor* column_descr = schema_->Column(column_ordinal); + DCHECK_NE(column_descr->physical_type(), Type::BOOLEAN); + auto bloom_filter_options_opt = properties_->bloom_filter_options(column_descr->path()); + if (bloom_filter_options_opt == std::nullopt) { + return nullptr; + } + BloomFilterOptions bloom_filter_options = *bloom_filter_options_opt; + RowGroupBloomFilters& row_group_bloom_filter = *file_bloom_filters_.back(); + auto iter = row_group_bloom_filter.find(column_ordinal); + if (iter == row_group_bloom_filter.end()) { + auto block_split_bloom_filter = + std::make_unique<BlockSplitBloomFilter>(properties_->memory_pool()); + block_split_bloom_filter->Init(BlockSplitBloomFilter::OptimalNumOfBytes( + bloom_filter_options.ndv, bloom_filter_options.fpp)); + auto insert_result = row_group_bloom_filter.emplace( + column_ordinal, std::move(block_split_bloom_filter)); + DCHECK(insert_result.second); + iter = insert_result.first; + } + ARROW_CHECK(iter->second != nullptr); + return iter->second.get(); +} + +void BloomFilterBuilderImpl::WriteTo(::arrow::io::OutputStream* sink, + BloomFilterLocation* location) { + if (finished_) { + throw ParquetException("Cannot call WriteTo() multiple times."); + } + finished_ = true; + + for (size_t row_group_ordinal = 0; row_group_ordinal < file_bloom_filters_.size(); + ++row_group_ordinal) { + RowGroupBloomFilters& row_group_bloom_filters = + *file_bloom_filters_[row_group_ordinal]; + // the whole row group has no bloom filter + if (row_group_bloom_filters.empty()) { + continue; + } + bool has_valid_bloom_filter = false; + int num_columns = schema_->num_columns(); + std::vector<std::optional<IndexLocation>> locations(num_columns, std::nullopt); Review Comment: is it maybe better to make this a map. I expect the number of columns with a bloom filter to be relatively small compared to the number of overall columns? ########## cpp/src/parquet/bloom_filter_builder.cc: ########## @@ -0,0 +1,158 @@ +// 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. + +// This module defines an abstract interface for iterating through pages in a +// Parquet column chunk within a row group. It could be extended in the future +// to iterate through all data pages in all chunks in a file. + +#include "parquet/bloom_filter_builder.h" + +#include <map> +#include <utility> +#include <vector> + +#include "arrow/io/interfaces.h" + +#include "parquet/bloom_filter.h" +#include "parquet/exception.h" +#include "parquet/metadata.h" +#include "parquet/properties.h" + +namespace parquet::internal { + +namespace { +/// Column encryption for bloom filter is not implemented yet. +class BloomFilterBuilderImpl : public BloomFilterBuilder { + public: + explicit BloomFilterBuilderImpl(const SchemaDescriptor* schema, + const WriterProperties* properties) + : schema_(schema), properties_(properties) {} + /// Append a new row group to host all incoming bloom filters. + void AppendRowGroup() override; + + BloomFilter* GetOrCreateBloomFilter(int32_t column_ordinal) override; + + /// Serialize all bloom filters with header and bitset in the order of row group and + /// column id. The side effect is that it deletes all bloom filters after they have + /// been flushed. + void WriteTo(::arrow::io::OutputStream* sink, BloomFilterLocation* location) override; + + BloomFilterBuilderImpl(const BloomFilterBuilderImpl&) = delete; + BloomFilterBuilderImpl(BloomFilterBuilderImpl&&) = default; + + private: + /// Make sure column ordinal is not out of bound and the builder is in good state. + void CheckState(int32_t column_ordinal) const { + if (finished_) { + throw ParquetException("BloomFilterBuilder is already finished."); + } + if (column_ordinal < 0 || column_ordinal >= schema_->num_columns()) { + throw ParquetException("Invalid column ordinal: ", column_ordinal); + } + if (file_bloom_filters_.empty()) { + throw ParquetException("No row group appended to BloomFilterBuilder."); + } + if (schema_->Column(column_ordinal)->physical_type() == Type::BOOLEAN) { + throw ParquetException("BloomFilterBuilder does not support boolean type."); + } + } + + const SchemaDescriptor* schema_; + const WriterProperties* properties_; + bool finished_ = false; + + using RowGroupBloomFilters = std::map<int32_t, std::unique_ptr<BloomFilter>>; + // Using unique_ptr because the `std::unique_ptr<BloomFilter>` is not copyable. + // MSVC has the issue below: https://github.com/microsoft/STL/issues/1036 + // So we use `std::unique_ptr<std::map<>>` to avoid the issue. + std::vector<std::unique_ptr<RowGroupBloomFilters>> file_bloom_filters_; +}; + +void BloomFilterBuilderImpl::AppendRowGroup() { + if (finished_) { + throw ParquetException( + "Cannot call AppendRowGroup() to finished BloomFilterBuilder."); + } + file_bloom_filters_.emplace_back(std::make_unique<RowGroupBloomFilters>()); +} + +BloomFilter* BloomFilterBuilderImpl::GetOrCreateBloomFilter(int32_t column_ordinal) { + CheckState(column_ordinal); + const ColumnDescriptor* column_descr = schema_->Column(column_ordinal); + DCHECK_NE(column_descr->physical_type(), Type::BOOLEAN); + auto bloom_filter_options_opt = properties_->bloom_filter_options(column_descr->path()); + if (bloom_filter_options_opt == std::nullopt) { + return nullptr; + } + BloomFilterOptions bloom_filter_options = *bloom_filter_options_opt; + RowGroupBloomFilters& row_group_bloom_filter = *file_bloom_filters_.back(); + auto iter = row_group_bloom_filter.find(column_ordinal); + if (iter == row_group_bloom_filter.end()) { + auto block_split_bloom_filter = + std::make_unique<BlockSplitBloomFilter>(properties_->memory_pool()); + block_split_bloom_filter->Init(BlockSplitBloomFilter::OptimalNumOfBytes( + bloom_filter_options.ndv, bloom_filter_options.fpp)); + auto insert_result = row_group_bloom_filter.emplace( + column_ordinal, std::move(block_split_bloom_filter)); + DCHECK(insert_result.second); + iter = insert_result.first; + } + ARROW_CHECK(iter->second != nullptr); + return iter->second.get(); +} + +void BloomFilterBuilderImpl::WriteTo(::arrow::io::OutputStream* sink, + BloomFilterLocation* location) { + if (finished_) { + throw ParquetException("Cannot call WriteTo() multiple times."); + } + finished_ = true; + + for (size_t row_group_ordinal = 0; row_group_ordinal < file_bloom_filters_.size(); + ++row_group_ordinal) { + RowGroupBloomFilters& row_group_bloom_filters = + *file_bloom_filters_[row_group_ordinal]; + // the whole row group has no bloom filter + if (row_group_bloom_filters.empty()) { + continue; + } + bool has_valid_bloom_filter = false; + int num_columns = schema_->num_columns(); + std::vector<std::optional<IndexLocation>> locations(num_columns, std::nullopt); + + // serialize bloom filter in ascending order of column id + for (auto& [column_id, filter] : row_group_bloom_filters) { + ARROW_CHECK(filter != nullptr); + PARQUET_ASSIGN_OR_THROW(int64_t offset, sink->Tell()); + filter->WriteTo(sink); + PARQUET_ASSIGN_OR_THROW(int64_t pos, sink->Tell()); + has_valid_bloom_filter = true; + locations[column_id] = IndexLocation{offset, static_cast<int32_t>(pos - offset)}; Review Comment: Maybe we should check the bloom filter is less then 4GB (I know this is extremely unlikely ...) ########## cpp/src/parquet/bloom_filter_builder.h: ########## @@ -0,0 +1,67 @@ +// 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. + +#pragma once + +#include "arrow/io/type_fwd.h" +#include "parquet/types.h" + +namespace parquet { + +class BloomFilter; +class SchemaDescriptor; +struct BloomFilterOptions; +struct BloomFilterLocation; + +/// \brief Interface for collecting bloom filter of a parquet file. +class PARQUET_EXPORT BloomFilterBuilder { + public: + /// \brief API convenience to create a BloomFilterBuilder. + static std::unique_ptr<BloomFilterBuilder> Make(const SchemaDescriptor* schema, + const WriterProperties& properties); + + /// Append a new row group to host all incoming bloom filters. + /// + /// This method must be called before Finish. + virtual void AppendRowGroup() = 0; + + /// \brief Get the BloomFilter from column ordinal. + /// + /// \param column_ordinal Column ordinal in schema, which is only for leaf columns. + /// + /// \return BloomFilter for the column and its memory ownership belongs to the + /// BloomFilterBuilder. + virtual BloomFilter* GetOrCreateBloomFilter(int32_t column_ordinal) = 0; Review Comment: thanks. LGTM. ########## cpp/src/parquet/bloom_filter_reader_writer_test.cc: ########## @@ -0,0 +1,216 @@ +// Licensed to the Apache Software Foundation (ASF) under one Review Comment: I don't know if github can be smart about it but was git mv used? I thought that might limit the diff? ########## cpp/src/parquet/column_writer.cc: ########## @@ -2363,12 +2390,153 @@ Status TypedColumnWriterImpl<FLBAType>::WriteArrowDense( return Status::OK(); } +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilter(const T* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, static_cast<int>(current_hash_batch_size), Review Comment: it seems we should check num_values is within int32 range? (also be explicit about integer type)? ########## cpp/src/parquet/column_writer.cc: ########## @@ -2363,12 +2390,153 @@ Status TypedColumnWriterImpl<FLBAType>::WriteArrowDense( return Status::OK(); } +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilter(const T* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilter(const FLBA* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, descr_->type_length(), + static_cast<int>(current_hash_batch_size), hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilter(const bool*, int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilterSpaced(const T* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilterSpaced(const bool*, int64_t, + const uint8_t*, + int64_t) { + DCHECK(bloom_filter_ == nullptr); Review Comment: since this is a separate module it might be worth raising an exception/returning an error in case there is some weird case where it gets populated? At least add a comment on tthis. ########## cpp/src/parquet/column_writer.cc: ########## @@ -2363,12 +2390,153 @@ Status TypedColumnWriterImpl<FLBAType>::WriteArrowDense( return Status::OK(); } +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilter(const T* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilter(const FLBA* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, descr_->type_length(), + static_cast<int>(current_hash_batch_size), hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilter(const bool*, int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilterSpaced(const T* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilterSpaced(const bool*, int64_t, + const uint8_t*, + int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilterSpaced(const FLBA* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( Review Comment: I forget what version of C++ are we on. I wonder if we could consolidate these methods and use `if constexpr` instead? ########## cpp/src/parquet/column_writer.cc: ########## @@ -2363,12 +2390,151 @@ Status TypedColumnWriterImpl<FLBAType>::WriteArrowDense( return Status::OK(); } +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilter(const T* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilter(const FLBA* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, descr_->type_length(), + static_cast<int>(current_hash_batch_size), hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilter(const bool*, int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilterSpaced(const T* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilterSpaced(const bool*, int64_t, + const uint8_t*, + int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilterSpaced(const FLBA* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, descr_->type_length(), + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <typename ArrayType> +void UpdateBinaryBloomFilter(BloomFilter* bloom_filter, const ArrayType& array) { + // Using a smaller size because an extra `byte_arrays` are used. + constexpr int64_t kBinaryHashBatchSize = 64; + std::array<ByteArray, kBinaryHashBatchSize> byte_arrays; + std::array<uint64_t, kBinaryHashBatchSize> hashes; + int hashes_idx = 0; + auto flush_hashes = [&]() { + DCHECK_NE(0, hashes_idx); + bloom_filter->Hashes(byte_arrays.data(), static_cast<int>(hashes_idx), hashes.data()); + bloom_filter->InsertHashes(hashes.data(), static_cast<int>(hashes_idx)); + hashes_idx = 0; + }; + PARQUET_THROW_NOT_OK(::arrow::VisitArraySpanInline<typename ArrayType::TypeClass>( + *array.data(), + [&](const std::string_view& view) { + if (hashes_idx == kHashBatchSize) { + flush_hashes(); + } + byte_arrays[hashes_idx] = view; + ++hashes_idx; + return Status::OK(); + }, + []() { return Status::OK(); })); + if (hashes_idx != 0) { + flush_hashes(); + } +} + +template <> +void TypedColumnWriterImpl<ByteArrayType>::UpdateBloomFilterArray( + const ::arrow::Array& values) { + if (bloom_filter_) { + if (!::arrow::is_base_binary_like(values.type_id())) { + throw ParquetException("Only BaseBinaryArray and subclasses supported"); + } + + if (::arrow::is_binary_like(values.type_id())) { + UpdateBinaryBloomFilter(bloom_filter_, + checked_cast<const ::arrow::BinaryArray&>(values)); + } else { + DCHECK(::arrow::is_large_binary_like(values.type_id())); Review Comment: probably throwing is better. ########## cpp/src/parquet/column_writer.cc: ########## @@ -2363,12 +2390,153 @@ Status TypedColumnWriterImpl<FLBAType>::WriteArrowDense( return Status::OK(); } +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilter(const T* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilter(const FLBA* values, + int64_t num_values) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + for (int64_t i = 0; i < num_values; i += kHashBatchSize) { + int64_t current_hash_batch_size = std::min(kHashBatchSize, num_values - i); + bloom_filter_->Hashes(values, descr_->type_length(), + static_cast<int>(current_hash_batch_size), hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilter(const bool*, int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <typename DType> +void TypedColumnWriterImpl<DType>::UpdateBloomFilterSpaced(const T* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <> +void TypedColumnWriterImpl<BooleanType>::UpdateBloomFilterSpaced(const bool*, int64_t, + const uint8_t*, + int64_t) { + DCHECK(bloom_filter_ == nullptr); +} + +template <> +void TypedColumnWriterImpl<FLBAType>::UpdateBloomFilterSpaced(const FLBA* values, + int64_t num_values, + const uint8_t* valid_bits, + int64_t valid_bits_offset) { + if (bloom_filter_) { + std::array<uint64_t, kHashBatchSize> hashes; + ::arrow::internal::VisitSetBitRunsVoid( + valid_bits, valid_bits_offset, num_values, [&](int64_t position, int64_t length) { + for (int64_t i = 0; i < length; i += kHashBatchSize) { + auto current_hash_batch_size = std::min(kHashBatchSize, length - i); + bloom_filter_->Hashes(values + i + position, descr_->type_length(), + static_cast<int>(current_hash_batch_size), + hashes.data()); + bloom_filter_->InsertHashes(hashes.data(), + static_cast<int>(current_hash_batch_size)); + } + }); + } +} + +template <typename ArrayType> +void UpdateBinaryBloomFilter(BloomFilter* bloom_filter, const ArrayType& array) { + // Using a smaller size because an extra `byte_arrays` are used. + constexpr int64_t kBinaryHashBatchSize = 64; + std::array<ByteArray, kBinaryHashBatchSize> byte_arrays; + std::array<uint64_t, kBinaryHashBatchSize> hashes; + int hashes_idx = 0; + auto flush_hashes = [&]() { + DCHECK_NE(0, hashes_idx); + bloom_filter->Hashes(byte_arrays.data(), static_cast<int>(hashes_idx), hashes.data()); + bloom_filter->InsertHashes(hashes.data(), static_cast<int>(hashes_idx)); + hashes_idx = 0; + }; + PARQUET_THROW_NOT_OK(::arrow::VisitArraySpanInline<typename ArrayType::TypeClass>( + *array.data(), + [&](std::string_view view) { + if (hashes_idx == kHashBatchSize) { + flush_hashes(); + } + byte_arrays[hashes_idx] = view; + ++hashes_idx; + return Status::OK(); + }, + []() { return Status::OK(); })); + if (hashes_idx != 0) { + flush_hashes(); + } +} + +template <> +void TypedColumnWriterImpl<ByteArrayType>::UpdateBloomFilterArray( + const ::arrow::Array& values) { + if (bloom_filter_) { + // TODO(mwish): GH-37832 currently we don't support write StringView/BinaryView to + // parquet file. We can support + if (!::arrow::is_base_binary_like(values.type_id())) { Review Comment: so no integer types? what about dictionary types? Do we have checks at a higher level? ########## cpp/src/parquet/properties.h: ########## @@ -564,6 +591,43 @@ class PARQUET_EXPORT WriterProperties { return this->disable_statistics(path->ToDotString()); } + /// Disable bloom filter for the column specified by `path`. + /// Default disabled. + Builder* disable_bloom_filter(const std::string& path) { + bloom_filter_options_[path] = std::nullopt; + return this; + } + + /// Disable bloom filter for the column specified by `path`. + /// Default enabled. Review Comment: why is the default different then above? ########## cpp/src/parquet/properties.h: ########## @@ -789,6 +857,19 @@ class PARQUET_EXPORT WriterProperties { return false; } + bool bloom_filter_enabled() const { Review Comment: yeah, I'm not sure I would include this as an option it seems like it really should be path specific. -- 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
