mapleFU commented on code in PR #14341:
URL: https://github.com/apache/arrow/pull/14341#discussion_r1149515227
##########
cpp/src/parquet/encoding.cc:
##########
@@ -3037,11 +3073,180 @@ class RleBooleanDecoder : public DecoderImpl, virtual
public BooleanDecoder {
// ----------------------------------------------------------------------
// DELTA_BYTE_ARRAY
-class DeltaByteArrayDecoder : public DecoderImpl,
- virtual public TypedDecoder<ByteArrayType> {
+/// Delta Byte Array encoding also known as incremental encoding or front
compression:
+/// for each element in a sequence of strings, store the prefix length of the
previous
+/// entry plus the suffix.
+///
+/// This is stored as a sequence of delta-encoded prefix lengths
(DELTA_BINARY_PACKED),
+/// followed by the suffixes encoded as delta length byte arrays
+/// (DELTA_LENGTH_BYTE_ARRAY).
+
+// ----------------------------------------------------------------------
+// DeltaByteArrayEncoder
+
+template <typename DType>
+class DeltaByteArrayEncoder : public EncoderImpl, virtual public
TypedEncoder<DType> {
public:
- explicit DeltaByteArrayDecoder(const ColumnDescriptor* descr,
+ using T = typename DType::c_type;
+
+ explicit DeltaByteArrayEncoder(const ColumnDescriptor* descr,
MemoryPool* pool =
::arrow::default_memory_pool())
+ : EncoderImpl(descr, Encoding::DELTA_BYTE_ARRAY, pool),
+ sink_(pool),
+ prefix_length_encoder_(nullptr, pool),
+ suffix_encoder_(nullptr, pool),
+ last_value_("") {}
+
+ std::shared_ptr<Buffer> FlushValues() override;
+
+ int64_t EstimatedDataEncodedSize() override {
+ return prefix_length_encoder_.EstimatedDataEncodedSize() +
+ suffix_encoder_.EstimatedDataEncodedSize();
+ }
+
+ using TypedEncoder<DType>::Put;
+
+ void Put(const ::arrow::Array& values) override;
+
+ void Put(const T* buffer, int num_values) override;
+
+ void PutSpaced(const T* src, int num_values, const uint8_t* valid_bits,
+ int64_t valid_bits_offset) override {
+ if (valid_bits != NULLPTR) {
+ PARQUET_ASSIGN_OR_THROW(auto buffer, ::arrow::AllocateBuffer(num_values
* sizeof(T),
+
this->memory_pool()));
+ T* data = reinterpret_cast<T*>(buffer->mutable_data());
+ int num_valid_values = ::arrow::util::internal::SpacedCompress<T>(
+ src, num_values, valid_bits, valid_bits_offset, data);
+ Put(data, num_valid_values);
+ } else {
+ Put(src, num_values);
+ }
+ }
+
+ protected:
+ template <typename ArrayType>
+ void PutBinaryArray(const ArrayType& array) {
+ uint32_t previous_len = 0;
+ std::string_view last_value_view = last_value_;
+
+ PARQUET_THROW_NOT_OK(::arrow::VisitArraySpanInline<typename
ArrayType::TypeClass>(
+ *array.data(),
+ [&](::std::string_view view) {
+ if (ARROW_PREDICT_FALSE(view.size() >= kMaxByteArraySize)) {
+ return Status::Invalid("Parquet cannot store strings with size 2GB
or more");
+ }
+ // Convert view to ByteArray so it can be passed to the
suffix_encoder_.
+ const ByteArray src{view};
+
+ uint32_t j = 0;
+ while (j < std::min(previous_len, src.len)) {
+ if (last_value_view[j] != view[j]) {
+ break;
+ }
+ j++;
+ }
+ previous_len = j;
Review Comment:
```suggestion
previous_len = src.len;
```
##########
cpp/src/parquet/encoding.cc:
##########
@@ -3037,11 +3073,180 @@ class RleBooleanDecoder : public DecoderImpl, virtual
public BooleanDecoder {
// ----------------------------------------------------------------------
// DELTA_BYTE_ARRAY
-class DeltaByteArrayDecoder : public DecoderImpl,
- virtual public TypedDecoder<ByteArrayType> {
+/// Delta Byte Array encoding also known as incremental encoding or front
compression:
+/// for each element in a sequence of strings, store the prefix length of the
previous
+/// entry plus the suffix.
+///
+/// This is stored as a sequence of delta-encoded prefix lengths
(DELTA_BINARY_PACKED),
+/// followed by the suffixes encoded as delta length byte arrays
+/// (DELTA_LENGTH_BYTE_ARRAY).
+
+// ----------------------------------------------------------------------
+// DeltaByteArrayEncoder
+
+template <typename DType>
+class DeltaByteArrayEncoder : public EncoderImpl, virtual public
TypedEncoder<DType> {
public:
- explicit DeltaByteArrayDecoder(const ColumnDescriptor* descr,
+ using T = typename DType::c_type;
+
+ explicit DeltaByteArrayEncoder(const ColumnDescriptor* descr,
MemoryPool* pool =
::arrow::default_memory_pool())
+ : EncoderImpl(descr, Encoding::DELTA_BYTE_ARRAY, pool),
+ sink_(pool),
+ prefix_length_encoder_(nullptr, pool),
+ suffix_encoder_(nullptr, pool),
+ last_value_("") {}
+
+ std::shared_ptr<Buffer> FlushValues() override;
+
+ int64_t EstimatedDataEncodedSize() override {
+ return prefix_length_encoder_.EstimatedDataEncodedSize() +
+ suffix_encoder_.EstimatedDataEncodedSize();
+ }
+
+ using TypedEncoder<DType>::Put;
+
+ void Put(const ::arrow::Array& values) override;
+
+ void Put(const T* buffer, int num_values) override;
+
+ void PutSpaced(const T* src, int num_values, const uint8_t* valid_bits,
+ int64_t valid_bits_offset) override {
+ if (valid_bits != NULLPTR) {
+ PARQUET_ASSIGN_OR_THROW(auto buffer, ::arrow::AllocateBuffer(num_values
* sizeof(T),
+
this->memory_pool()));
+ T* data = reinterpret_cast<T*>(buffer->mutable_data());
+ int num_valid_values = ::arrow::util::internal::SpacedCompress<T>(
+ src, num_values, valid_bits, valid_bits_offset, data);
+ Put(data, num_valid_values);
+ } else {
+ Put(src, num_values);
+ }
+ }
+
+ protected:
+ template <typename ArrayType>
+ void PutBinaryArray(const ArrayType& array) {
+ uint32_t previous_len = 0;
Review Comment:
```suggestion
uint32_t previous_len = static_cast<uint32_t>(last_value_.size());
```
##########
cpp/src/parquet/encoding.cc:
##########
@@ -3037,11 +3073,180 @@ class RleBooleanDecoder : public DecoderImpl, virtual
public BooleanDecoder {
// ----------------------------------------------------------------------
// DELTA_BYTE_ARRAY
-class DeltaByteArrayDecoder : public DecoderImpl,
- virtual public TypedDecoder<ByteArrayType> {
+/// Delta Byte Array encoding also known as incremental encoding or front
compression:
+/// for each element in a sequence of strings, store the prefix length of the
previous
+/// entry plus the suffix.
+///
+/// This is stored as a sequence of delta-encoded prefix lengths
(DELTA_BINARY_PACKED),
+/// followed by the suffixes encoded as delta length byte arrays
+/// (DELTA_LENGTH_BYTE_ARRAY).
+
+// ----------------------------------------------------------------------
+// DeltaByteArrayEncoder
+
+template <typename DType>
+class DeltaByteArrayEncoder : public EncoderImpl, virtual public
TypedEncoder<DType> {
public:
- explicit DeltaByteArrayDecoder(const ColumnDescriptor* descr,
+ using T = typename DType::c_type;
+
+ explicit DeltaByteArrayEncoder(const ColumnDescriptor* descr,
MemoryPool* pool =
::arrow::default_memory_pool())
+ : EncoderImpl(descr, Encoding::DELTA_BYTE_ARRAY, pool),
+ sink_(pool),
+ prefix_length_encoder_(nullptr, pool),
+ suffix_encoder_(nullptr, pool),
+ last_value_("") {}
+
+ std::shared_ptr<Buffer> FlushValues() override;
+
+ int64_t EstimatedDataEncodedSize() override {
+ return prefix_length_encoder_.EstimatedDataEncodedSize() +
+ suffix_encoder_.EstimatedDataEncodedSize();
+ }
+
+ using TypedEncoder<DType>::Put;
+
+ void Put(const ::arrow::Array& values) override;
+
+ void Put(const T* buffer, int num_values) override;
+
+ void PutSpaced(const T* src, int num_values, const uint8_t* valid_bits,
+ int64_t valid_bits_offset) override {
+ if (valid_bits != NULLPTR) {
+ PARQUET_ASSIGN_OR_THROW(auto buffer, ::arrow::AllocateBuffer(num_values
* sizeof(T),
+
this->memory_pool()));
+ T* data = reinterpret_cast<T*>(buffer->mutable_data());
+ int num_valid_values = ::arrow::util::internal::SpacedCompress<T>(
+ src, num_values, valid_bits, valid_bits_offset, data);
+ Put(data, num_valid_values);
+ } else {
+ Put(src, num_values);
+ }
+ }
+
+ protected:
+ template <typename ArrayType>
+ void PutBinaryArray(const ArrayType& array) {
+ uint32_t previous_len = 0;
+ std::string_view last_value_view = last_value_;
+
+ PARQUET_THROW_NOT_OK(::arrow::VisitArraySpanInline<typename
ArrayType::TypeClass>(
+ *array.data(),
+ [&](::std::string_view view) {
+ if (ARROW_PREDICT_FALSE(view.size() >= kMaxByteArraySize)) {
+ return Status::Invalid("Parquet cannot store strings with size 2GB
or more");
+ }
+ // Convert view to ByteArray so it can be passed to the
suffix_encoder_.
+ const ByteArray src{view};
+
+ uint32_t j = 0;
+ while (j < std::min(previous_len, src.len)) {
+ if (last_value_view[j] != view[j]) {
+ break;
+ }
+ j++;
+ }
+ previous_len = j;
+ prefix_length_encoder_.Put({static_cast<int32_t>(j)}, 1);
+
+ const uint8_t* suffix_ptr = src.ptr + j;
+ const uint32_t suffix_length = static_cast<uint32_t>(src.len - j);
+ last_value_view =
+ string_view{reinterpret_cast<const char*>(suffix_ptr),
suffix_length};
+ // Convert suffix to ByteArray so it can be passed to the
suffix_encoder_.
+ const ByteArray suffix(suffix_length, suffix_ptr);
+ suffix_encoder_.Put(&suffix, 1);
+
+ return Status::OK();
+ },
+ []() { return Status::OK(); }));
+ last_value_ = last_value_view;
+ }
+
+ ::arrow::BufferBuilder sink_;
+ DeltaBitPackEncoder<Int32Type> prefix_length_encoder_;
+ DeltaLengthByteArrayEncoder<ByteArrayType> suffix_encoder_;
+ std::string last_value_;
+};
+
+template <typename DType>
+void DeltaByteArrayEncoder<DType>::Put(const T* src, int num_values) {
+ if (num_values == 0) {
+ return;
+ }
+ ArrowPoolVector<int32_t> prefix_lengths(num_values,
+
::arrow::stl::allocator<int32_t>(pool_));
+ std::string_view last_value_view = last_value_;
+
+ int i = 0;
+ while (i < num_values) {
+ // Convert to ByteArray so we can pass to the suffix_encoder_.
+ auto value = reinterpret_cast<const ByteArray*>(&src[i]);
+ if (ARROW_PREDICT_FALSE(value->len >= kMaxByteArraySize)) {
+ throw Status::Invalid("Parquet cannot store strings with size 2GB or
more");
+ }
+
+ auto view = string_view{reinterpret_cast<const char*>(value->ptr),
value->len};
+ uint32_t j = 0;
+ while (j < std::min(value->len,
static_cast<uint32_t>(last_value_view.length()))) {
+ if (last_value_view[j] != view[j]) {
+ break;
+ }
+ j++;
+ }
+
+ prefix_lengths[i] = j;
+ const uint8_t* suffix_ptr = value->ptr + j;
+ const uint32_t suffix_length = static_cast<uint32_t>(value->len - j);
+ last_value_view =
+ string_view{reinterpret_cast<const char*>(suffix_ptr), suffix_length};
+ // Convert suffix to ByteArray so it can be passed to the suffix_encoder_.
+ const ByteArray suffix(suffix_length, suffix_ptr);
+ suffix_encoder_.Put(&suffix, 1);
+ i++;
+ }
+ prefix_length_encoder_.Put(prefix_lengths.data(), num_values);
+ last_value_ = last_value_view;
+}
+
+template <typename DType>
+void DeltaByteArrayEncoder<DType>::Put(const ::arrow::Array& values) {
+ if (::arrow::is_binary_like(values.type_id())) {
+ PutBinaryArray(checked_cast<const ::arrow::BinaryArray&>(values));
+ } else if (::arrow::is_large_binary_like(values.type_id())) {
+ PutBinaryArray(checked_cast<const ::arrow::LargeBinaryArray&>(values));
+ } else if (::arrow::is_fixed_size_binary(values.type_id())) {
+ PutBinaryArray(checked_cast<const ::arrow::FixedSizeBinaryArray&>(values));
+ } else {
+ throw ParquetException("Only BaseBinaryArray and subclasses supported");
+ }
+}
+
+template <typename DType>
+std::shared_ptr<Buffer> DeltaByteArrayEncoder<DType>::FlushValues() {
+ PARQUET_THROW_NOT_OK(sink_.Resize(EstimatedDataEncodedSize(), false));
+
+ std::shared_ptr<Buffer> prefix_lengths =
prefix_length_encoder_.FlushValues();
+ PARQUET_THROW_NOT_OK(sink_.Append(prefix_lengths->data(),
prefix_lengths->size()));
+
+ std::shared_ptr<Buffer> suffixes = suffix_encoder_.FlushValues();
+ PARQUET_THROW_NOT_OK(sink_.Append(suffixes->data(), suffixes->size()));
+
+ std::shared_ptr<Buffer> buffer;
+ PARQUET_THROW_NOT_OK(sink_.Finish(&buffer, true));
+ return buffer;
Review Comment:
Should clear context here.
```suggestion
last_value_.clear();
return buffer;
```
##########
cpp/src/parquet/encoding.cc:
##########
@@ -3037,11 +3073,180 @@ class RleBooleanDecoder : public DecoderImpl, virtual
public BooleanDecoder {
// ----------------------------------------------------------------------
// DELTA_BYTE_ARRAY
-class DeltaByteArrayDecoder : public DecoderImpl,
- virtual public TypedDecoder<ByteArrayType> {
+/// Delta Byte Array encoding also known as incremental encoding or front
compression:
+/// for each element in a sequence of strings, store the prefix length of the
previous
+/// entry plus the suffix.
+///
+/// This is stored as a sequence of delta-encoded prefix lengths
(DELTA_BINARY_PACKED),
+/// followed by the suffixes encoded as delta length byte arrays
+/// (DELTA_LENGTH_BYTE_ARRAY).
+
+// ----------------------------------------------------------------------
+// DeltaByteArrayEncoder
+
+template <typename DType>
+class DeltaByteArrayEncoder : public EncoderImpl, virtual public
TypedEncoder<DType> {
public:
- explicit DeltaByteArrayDecoder(const ColumnDescriptor* descr,
+ using T = typename DType::c_type;
+
+ explicit DeltaByteArrayEncoder(const ColumnDescriptor* descr,
MemoryPool* pool =
::arrow::default_memory_pool())
+ : EncoderImpl(descr, Encoding::DELTA_BYTE_ARRAY, pool),
+ sink_(pool),
+ prefix_length_encoder_(nullptr, pool),
+ suffix_encoder_(nullptr, pool),
+ last_value_("") {}
+
+ std::shared_ptr<Buffer> FlushValues() override;
+
+ int64_t EstimatedDataEncodedSize() override {
+ return prefix_length_encoder_.EstimatedDataEncodedSize() +
+ suffix_encoder_.EstimatedDataEncodedSize();
+ }
+
+ using TypedEncoder<DType>::Put;
+
+ void Put(const ::arrow::Array& values) override;
+
+ void Put(const T* buffer, int num_values) override;
+
+ void PutSpaced(const T* src, int num_values, const uint8_t* valid_bits,
+ int64_t valid_bits_offset) override {
+ if (valid_bits != NULLPTR) {
+ PARQUET_ASSIGN_OR_THROW(auto buffer, ::arrow::AllocateBuffer(num_values
* sizeof(T),
+
this->memory_pool()));
+ T* data = reinterpret_cast<T*>(buffer->mutable_data());
+ int num_valid_values = ::arrow::util::internal::SpacedCompress<T>(
+ src, num_values, valid_bits, valid_bits_offset, data);
+ Put(data, num_valid_values);
+ } else {
+ Put(src, num_values);
+ }
+ }
+
+ protected:
+ template <typename ArrayType>
+ void PutBinaryArray(const ArrayType& array) {
+ uint32_t previous_len = 0;
+ std::string_view last_value_view = last_value_;
+
+ PARQUET_THROW_NOT_OK(::arrow::VisitArraySpanInline<typename
ArrayType::TypeClass>(
+ *array.data(),
+ [&](::std::string_view view) {
+ if (ARROW_PREDICT_FALSE(view.size() >= kMaxByteArraySize)) {
+ return Status::Invalid("Parquet cannot store strings with size 2GB
or more");
+ }
+ // Convert view to ByteArray so it can be passed to the
suffix_encoder_.
+ const ByteArray src{view};
+
+ uint32_t j = 0;
+ while (j < std::min(previous_len, src.len)) {
+ if (last_value_view[j] != view[j]) {
+ break;
+ }
+ j++;
+ }
+ previous_len = j;
Review Comment:
Or we can discard `previous_len`, just use `last_value_view`
##########
cpp/src/parquet/encoding_test.cc:
##########
@@ -1955,5 +1955,159 @@ TEST(DeltaLengthByteArrayEncodingAdHoc, ArrowDirectPut)
{
CheckDecode(encoded, ::arrow::ArrayFromJSON(::arrow::large_binary(),
values));
}
+// ----------------------------------------------------------------------
+// DELTA_BYTE_ARRAY encode/decode tests.
+
+template <typename Type>
+class TestDeltaByteArrayEncoding : public TestEncodingBase<Type> {
+ public:
+ using c_type = typename Type::c_type;
+ static constexpr int TYPE = Type::type_num;
+
+ virtual void CheckRoundtrip() {
+ auto encoder =
+ MakeTypedEncoder<Type>(Encoding::DELTA_BYTE_ARRAY, false,
descr_.get());
+ auto decoder = MakeTypedDecoder<Type>(Encoding::DELTA_BYTE_ARRAY,
descr_.get());
+
+ encoder->Put(draws_, num_values_);
+ encode_buffer_ = encoder->FlushValues();
+
+ decoder->SetData(num_values_, encode_buffer_->data(),
+ static_cast<int>(encode_buffer_->size()));
+ int values_decoded = decoder->Decode(decode_buf_, num_values_);
+ ASSERT_EQ(num_values_, values_decoded);
+ ASSERT_NO_FATAL_FAILURE(VerifyResults<c_type>(decode_buf_, draws_,
num_values_));
+ }
+
+ void CheckRoundtripSpaced(const uint8_t* valid_bits, int64_t
valid_bits_offset) {
+ auto encoder =
+ MakeTypedEncoder<Type>(Encoding::DELTA_BYTE_ARRAY, false,
descr_.get());
+ auto decoder = MakeTypedDecoder<Type>(Encoding::DELTA_BYTE_ARRAY,
descr_.get());
+ int null_count = 0;
+ for (auto i = 0; i < num_values_; i++) {
+ if (!bit_util::GetBit(valid_bits, valid_bits_offset + i)) {
+ null_count++;
+ }
+ }
+
+ encoder->PutSpaced(draws_, num_values_, valid_bits, valid_bits_offset);
+ encode_buffer_ = encoder->FlushValues();
+ decoder->SetData(num_values_ - null_count, encode_buffer_->data(),
+ static_cast<int>(encode_buffer_->size()));
+ auto values_decoded = decoder->DecodeSpaced(decode_buf_, num_values_,
null_count,
+ valid_bits, valid_bits_offset);
+ ASSERT_EQ(num_values_, values_decoded);
+ ASSERT_NO_FATAL_FAILURE(VerifyResultsSpaced<c_type>(decode_buf_, draws_,
num_values_,
+ valid_bits,
valid_bits_offset));
+ }
+
+ protected:
+ USING_BASE_MEMBERS();
+};
+
+typedef ::testing::Types<ByteArrayType> TestDeltaByteArrayEncodingTypes;
+TYPED_TEST_SUITE(TestDeltaByteArrayEncoding, TestDeltaByteArrayEncodingTypes);
+
+// TODO: add FLBAType and Decimal type tests
+
+TYPED_TEST(TestDeltaByteArrayEncoding, BasicRoundTrip) {
+ ASSERT_NO_FATAL_FAILURE(this->Execute(0, 0));
+ ASSERT_NO_FATAL_FAILURE(this->Execute(250, 2));
+ ASSERT_NO_FATAL_FAILURE(this->ExecuteSpaced(
+ /*nvalues*/ 1234, /*repeats*/ 1, /*valid_bits_offset*/ 64, /*null_prob*/
0));
+
+ ASSERT_NO_FATAL_FAILURE(this->Execute(2000, 200));
+ ASSERT_NO_FATAL_FAILURE(this->ExecuteSpaced(
+ /*nvalues*/ 1234, /*repeats*/ 10, /*valid_bits_offset*/ 64,
+ /*null_probability*/ 0.1));
+}
+
+TEST(DeltaByteArrayEncodingAdHoc, ArrowBinaryDirectPut) {
+ const int64_t size = 50;
+ const int32_t min_length = 0;
+ const int32_t max_length = 10;
+ const int32_t num_unique = 10;
+ const double null_probability = 0.25;
+ auto encoder = MakeTypedEncoder<ByteArrayType>(Encoding::DELTA_BYTE_ARRAY);
+ auto decoder = MakeTypedDecoder<ByteArrayType>(Encoding::DELTA_BYTE_ARRAY);
+
+ auto CheckSeed = [&](std::shared_ptr<::arrow::Array> values) {
+ ASSERT_NO_THROW(encoder->Put(*values));
+ auto buf = encoder->FlushValues();
+
+ int num_values = static_cast<int>(values->length() - values->null_count());
+ decoder->SetData(num_values, buf->data(), static_cast<int>(buf->size()));
+
+ typename EncodingTraits<ByteArrayType>::Accumulator acc;
+ if (::arrow::is_string(values->type()->id())) {
+ acc.builder = std::make_unique<::arrow::StringBuilder>();
+ } else {
+ acc.builder = std::make_unique<::arrow::BinaryBuilder>();
+ }
+ ASSERT_EQ(num_values,
+ decoder->DecodeArrow(static_cast<int>(values->length()),
+ static_cast<int>(values->null_count()),
+ values->null_bitmap_data(),
values->offset(), &acc));
+
+ std::shared_ptr<::arrow::Array> result;
+ ASSERT_OK(acc.builder->Finish(&result));
+ ASSERT_EQ(values->length(), result->length());
+ ASSERT_OK(result->ValidateFull());
+
+ auto upcast_result = CastBinaryTypesHelper(result, values->type());
+ ::arrow::AssertArraysEqual(*values, *result);
+ };
+
+ ::arrow::random::RandomArrayGenerator rag(42);
+ auto values = rag.String(0, min_length, max_length, null_probability);
+ CheckSeed(values);
+ for (auto seed : {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}) {
+ rag = ::arrow::random::RandomArrayGenerator(seed);
+
+ values = rag.String(size, min_length, max_length, null_probability);
+ CheckSeed(values);
+
+ values =
+ rag.BinaryWithRepeats(size, num_unique, min_length, max_length,
null_probability);
+ CheckSeed(values);
+ }
+}
+
+TEST(DeltaByteArrayEncodingAdHoc, ArrowBinaryDirectPutFixedLength) {
+ const int64_t size = 50;
+ const double null_probability = 0.25;
+ ::arrow::random::RandomArrayGenerator rag(0);
+ auto encoder = MakeTypedEncoder<FLBAType>(Encoding::DELTA_BYTE_ARRAY);
+ auto decoder = MakeTypedDecoder<FLBAType>(Encoding::DELTA_BYTE_ARRAY);
+
+ auto CheckSeed = [&](std::shared_ptr<::arrow::Array> values) {
+ ASSERT_NO_THROW(encoder->Put(*values));
+ auto buf = encoder->FlushValues();
+
+ int num_values = static_cast<int>(values->length() - values->null_count());
+ decoder->SetData(num_values, buf->data(), static_cast<int>(buf->size()));
+
+ typename EncodingTraits<FLBAType>::Accumulator acc(values->type());
+ ASSERT_EQ(num_values,
+ decoder->DecodeArrow(static_cast<int>(values->length()),
+ static_cast<int>(values->null_count()),
+ values->null_bitmap_data(),
values->offset(), &acc));
+
+ std::shared_ptr<::arrow::Array> result;
+ ASSERT_OK(acc.Finish(&result));
+ ASSERT_EQ(values->length(), result->length());
+ ASSERT_OK(result->ValidateFull());
+ ::arrow::AssertArraysEqual(*values, *result);
+ };
+
+ for (auto seed : {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}) {
+ for (auto length : {0, 10, 100, 1000}) {
+ rag = ::arrow::random::RandomArrayGenerator(seed);
+ auto values = rag.FixedSizeBinary(size, length, null_probability);
+ CheckSeed(values);
+ }
+ }
+}
+
Review Comment:
Should add a test for common prefix, seems that arrow generate didn't tests
that:
```suggestion
TEST(DeltaByteArrayEncodingAdHoc, ArrowDirectPut) {
auto CheckEncode = [](std::shared_ptr<::arrow::Array> values,
std::shared_ptr<::arrow::Array> prefix_lengths,
std::shared_ptr<::arrow::Array> suffix_lengths,
std::string_view value) {
auto encoder =
MakeTypedEncoder<ByteArrayType>(Encoding::DELTA_BYTE_ARRAY);
ASSERT_NO_THROW(encoder->Put(*values));
auto buf = encoder->FlushValues();
auto prefix_lengths_encoder =
MakeTypedEncoder<Int32Type>(Encoding::DELTA_BINARY_PACKED);
ASSERT_NO_THROW(prefix_lengths_encoder->Put(*prefix_lengths));
auto prefix_lengths_buf = prefix_lengths_encoder->FlushValues();
auto encoded_prefix_lengths_buf = SliceBuffer(buf, 0,
prefix_lengths_buf->size());
auto suffix_lengths_encoder =
MakeTypedEncoder<Int32Type>(Encoding::DELTA_BINARY_PACKED);
ASSERT_NO_THROW(suffix_lengths_encoder->Put(*suffix_lengths));
auto suffix_lengths_buf = suffix_lengths_encoder->FlushValues();
auto encoded_values_buf = SliceBuffer(buf, prefix_lengths_buf->size() +
suffix_lengths_buf->size());
auto encoded_prefix_length_buf = SliceBuffer(buf, 0,
prefix_lengths_buf->size());
EXPECT_TRUE(prefix_lengths_buf->Equals(*encoded_prefix_length_buf));
auto encoded_suffix_length_buf = SliceBuffer(buf,
prefix_lengths_buf->size(), suffix_lengths_buf->size());
EXPECT_TRUE(suffix_lengths_buf->Equals(*encoded_suffix_length_buf));
EXPECT_EQ(value, encoded_values_buf->ToString());
};
auto values = R"(["axis", "axle", "babble", "babyhood"])";
auto prefix_lengths = ::arrow::ArrayFromJSON(::arrow::int32(), R"([0, 2,
0, 3])");
auto suffix_lengths = ::arrow::ArrayFromJSON(::arrow::int32(), R"([4, 2,
6, 5])");
CheckEncode(::arrow::ArrayFromJSON(::arrow::utf8(), values),
prefix_lengths, suffix_lengths, "axislebabbleyhood");
CheckEncode(::arrow::ArrayFromJSON(::arrow::large_utf8(), values),
prefix_lengths, suffix_lengths, "axislebabbleyhood");
CheckEncode(::arrow::ArrayFromJSON(::arrow::binary(), values),
prefix_lengths, suffix_lengths, "axislebabbleyhood");
CheckEncode(::arrow::ArrayFromJSON(::arrow::large_binary(), values),
prefix_lengths, suffix_lengths, "axislebabbleyhood");
}
```
This code is modified from
`DeltaLengthByteArrayEncodingAdHoc.ArrowDirectPut`, you can just go ahead and
add `CheckDecode`, and add more cases
--
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]
