#include <ParquetVariableLengthArrayEncoder.h>

Inheritance diagram for foreign_storage::ParquetVariableLengthArrayEncoder:

Collaboration diagram for foreign_storage::ParquetVariableLengthArrayEncoder:

Public Member Functions
	ParquetVariableLengthArrayEncoder (Data_Namespace::AbstractBuffer data_buffer, Data_Namespace::AbstractBuffer index_buffer, std::shared_ptr< ParquetScalarEncoder > scalar_encoder, const ColumnDescriptor *column_desciptor)

void	appendData (const int16_t def_levels, const int16_t rep_levels, const int64_t values_read, const int64_t levels_read, int8_t *values) override

Public Member Functions inherited from foreign_storage::ParquetArrayEncoder
	ParquetArrayEncoder (Data_Namespace::AbstractBuffer data_buffer, std::shared_ptr< ParquetScalarEncoder > scalar_encoder, const ColumnDescriptor column_desciptor)

void	appendDataTrackErrors (const int16_t def_levels, const int16_t rep_levels, const int64_t values_read, const int64_t levels_read, int8_t *values) override

void	finalizeRowGroup ()

std::shared_ptr< ChunkMetadata >	getRowGroupMetadata (const parquet::RowGroupMetaData *group_metadata, const int parquet_column_index, const SQLTypeInfo &column_type) override

virtual void	disableMetadataStatsValidation () override

virtual void	initializeErrorTracking (const SQLTypeInfo &column_type) override

Public Member Functions inherited from foreign_storage::ParquetEncoder
	ParquetEncoder (Data_Namespace::AbstractBuffer *buffer)

virtual	~ParquetEncoder ()=default

RejectedRowIndices	getRejectedRowIndices () const

Protected Member Functions
void	appendArraysToBuffer () override

void	processLastArray () override

Protected Member Functions inherited from foreign_storage::ParquetArrayEncoder
bool	isLastArrayNull () const

bool	isLastArrayEmpty () const

size_t	sizeOfLastArray () const

int8_t *	resizeArrayDataBytes (const size_t additional_num_elements)

virtual void	resetLastArrayMetadata ()

bool	isNewArray (const int16_t rep_level) const

int8_t *	encodedDataAtIndex (const size_t index)

void	updateMetadataForAppendedArrayItem (const int64_t encoded_index)

virtual void	appendArrayItem (const int64_t encoded_index)

virtual void	encodeAllValues (const int8_t *values, const int64_t values_read)

Private Member Functions
void	setFirstOffsetForBuffer (const int16_t def_level)

void	appendLastArrayOffset ()

Private Attributes
Data_Namespace::AbstractBuffer *	index_buffer_

std::vector< ArrayOffsetT >	offsets_

Additional Inherited Members
Static Protected Member Functions inherited from foreign_storage::ParquetEncoder
static std::shared_ptr < ChunkMetadata >	createMetadata (const SQLTypeInfo &column_type)

static void	throwNotNullViolation (const std::string &parquet_column_name)

static void	validateNullCount (const std::string &parquet_column_name, int64_t null_count, const SQLTypeInfo &column_type)

Protected Attributes inherited from foreign_storage::ParquetArrayEncoder
size_t	omnisci_data_type_byte_size_

std::shared_ptr < ParquetScalarEncoder >	scalar_encoder_

std::vector< int8_t >	data_buffer_bytes_

Protected Attributes inherited from foreign_storage::ParquetEncoder
Data_Namespace::AbstractBuffer *	buffer_

bool	is_error_tracking_enabled_

RejectedRowIndices	invalid_indices_

size_t	current_chunk_offset_

SQLTypeInfo	column_type_

bool	validate_metadata_stats_

Static Protected Attributes inherited from foreign_storage::ParquetArrayEncoder
static const int16_t	non_null_def_level = 3

static const int16_t	item_null_def_level = 2

static const int16_t	empty_list_def_level = 1

static const int16_t	list_null_def_level = 0

Detailed Description

Definition at line 25 of file ParquetVariableLengthArrayEncoder.h.

Constructor & Destructor Documentation

foreign_storage::ParquetVariableLengthArrayEncoder::ParquetVariableLengthArrayEncoder	(	Data_Namespace::AbstractBuffer *	data_buffer,
		Data_Namespace::AbstractBuffer *	index_buffer,
		std::shared_ptr< ParquetScalarEncoder >	scalar_encoder,
		const ColumnDescriptor *	column_desciptor
	)

inline

Definition at line 27 of file ParquetVariableLengthArrayEncoder.h.

31 : ParquetArrayEncoder(data_buffer, scalar_encoder, column_desciptor)

32 , index_buffer_(index_buffer) {}

foreign_storage::ParquetVariableLengthArrayEncoder::index_buffer_

Data_Namespace::AbstractBuffer * index_buffer_

Definition: ParquetVariableLengthArrayEncoder.h:91

foreign_storage::ParquetArrayEncoder::ParquetArrayEncoder

ParquetArrayEncoder(Data_Namespace::AbstractBuffer *data_buffer, std::shared_ptr< ParquetScalarEncoder > scalar_encoder, const ColumnDescriptor *column_desciptor)

Definition: ParquetArrayEncoder.h:25

Member Function Documentation

void foreign_storage::ParquetVariableLengthArrayEncoder::appendArraysToBuffer ( )

inlineoverrideprotectedvirtual

Reimplemented from foreign_storage::ParquetArrayEncoder.

Definition at line 46 of file ParquetVariableLengthArrayEncoder.h.

References Data_Namespace::AbstractBuffer::append(), foreign_storage::ParquetArrayEncoder::appendArraysToBuffer(), index_buffer_, and offsets_.

                                        {
     index_buffer_->append(reinterpret_cast<int8_t*>(offsets_.data()),
                           offsets_.size() * sizeof(ArrayOffsetT));
     offsets_.clear();
     ParquetArrayEncoder::appendArraysToBuffer();
   }

Here is the call graph for this function:

void foreign_storage::ParquetVariableLengthArrayEncoder::appendData	(	const int16_t *	def_levels,
		const int16_t *	rep_levels,
		const int64_t	values_read,
		const int64_t	levels_read,
		int8_t *	values
	)

inlineoverridevirtual

Reimplemented from foreign_storage::ParquetArrayEncoder.

Definition at line 34 of file ParquetVariableLengthArrayEncoder.h.

References foreign_storage::ParquetArrayEncoder::appendData(), CHECK, and setFirstOffsetForBuffer().

                                            {
     CHECK(levels_read > 0);
     setFirstOffsetForBuffer(def_levels[0]);
     ParquetArrayEncoder::appendData(
         def_levels, rep_levels, values_read, levels_read, values);
   }

Here is the call graph for this function:

void foreign_storage::ParquetVariableLengthArrayEncoder::appendLastArrayOffset ( )

inlineprivate

Definition at line 78 of file ParquetVariableLengthArrayEncoder.h.

References foreign_storage::ParquetEncoder::buffer_, foreign_storage::ParquetArrayEncoder::data_buffer_bytes_, foreign_storage::ParquetArrayEncoder::isLastArrayEmpty(), foreign_storage::ParquetArrayEncoder::isLastArrayNull(), offsets_, and Data_Namespace::AbstractBuffer::size().

Referenced by processLastArray().

                                {
     int64_t last_offset = buffer_->size() + data_buffer_bytes_.size();
     if (isLastArrayNull()) {
       // append a null array offset
       offsets_.push_back(-last_offset);
     } else if (isLastArrayEmpty()) {
       offsets_.push_back(last_offset);
     } else {
       // append array data offset
       offsets_.push_back(last_offset);
     }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

void foreign_storage::ParquetVariableLengthArrayEncoder::processLastArray ( )

inlineoverrideprotectedvirtual

Reimplemented from foreign_storage::ParquetArrayEncoder.

Definition at line 53 of file ParquetVariableLengthArrayEncoder.h.

References appendLastArrayOffset(), and foreign_storage::ParquetArrayEncoder::processLastArray().

                                    {
     ParquetArrayEncoder::processLastArray();
     appendLastArrayOffset();
   }

Here is the call graph for this function:

void foreign_storage::ParquetVariableLengthArrayEncoder::setFirstOffsetForBuffer ( const int16_t def_level )

inlineprivate

Definition at line 59 of file ParquetVariableLengthArrayEncoder.h.

References foreign_storage::ParquetEncoder::buffer_, foreign_storage::ParquetArrayEncoder::data_buffer_bytes_, ArrayNoneEncoder::DEFAULT_NULL_PADDING_SIZE, foreign_storage::ParquetArrayEncoder::empty_list_def_level, foreign_storage::ParquetArrayEncoder::list_null_def_level, offsets_, and Data_Namespace::AbstractBuffer::size().

Referenced by appendData().

                                                         {
     if (data_buffer_bytes_.size() == 0 && buffer_->size() == 0) {  // first  element
       if (def_level == ParquetArrayEncoder::list_null_def_level ||
           def_level == ParquetArrayEncoder::empty_list_def_level) {
         // OmniSci variable array types have a special encoding for chunks in
         // which the first array is null: the first `DEFAULT_NULL_PADDING_SIZE`
         // bytes of the chunk are filled and the offset is set appropriately.
         // Ostensibly, this is done to allow marking a null array by negating
         // a non-zero value.
         offsets_.push_back(ArrayNoneEncoder::DEFAULT_NULL_PADDING_SIZE);
         std::vector<int8_t> zero_bytes(ArrayNoneEncoder::DEFAULT_NULL_PADDING_SIZE, 0);
         data_buffer_bytes_.insert(
             data_buffer_bytes_.end(), zero_bytes.begin(), zero_bytes.end());
       } else {
         offsets_.push_back(0);
       }
     }
   }