#include <StringViewToStringDictEncoder.h>

Inheritance diagram for data_conversion::StringViewToStringDictEncoder< IdType >:

Collaboration diagram for data_conversion::StringViewToStringDictEncoder< IdType >:

Public Member Functions
	StringViewToStringDictEncoder (const Chunk_NS::Chunk &dst_chunk, const bool error_tracking_enabled)

void	encodeAndAppendData (const int8_t *data, const size_t num_elements) override

void	clear () override

const Chunk_NS::Chunk &	getDstChunk () const

std::optional< std::vector < bool > > &	getDeleteBuffer ()

Public Member Functions inherited from data_conversion::TypedBaseConvertEncoder< IdType >
	TypedBaseConvertEncoder (const bool error_tracking_enabled)

std::shared_ptr< ChunkMetadata >	getMetadata (const Chunk_NS::Chunk &chunk) const override

Public Member Functions inherited from data_conversion::BaseConvertEncoder
virtual	~BaseConvertEncoder ()

	BaseConvertEncoder (const bool error_tracking_enabled)

void	initializeDeleteBuffer (const size_t size_hint)

virtual void	finalize (const size_t rows_appended)

Private Member Functions
void	initialize ()

Private Attributes
SQLTypeInfo	dst_type_info_

const Chunk_NS::Chunk &	dst_chunk_

StringDictionary *	string_dict_

std::vector< IdType >	dict_encoding_output_buffer_

AbstractBuffer *	buffer_

Additional Inherited Members
Public Types inherited from data_conversion::TypedBaseConvertEncoder< IdType >
using	DataType = IdType

using	MetadataType = IdType

Protected Member Functions inherited from data_conversion::TypedBaseConvertEncoder< IdType >
DataType	getNull () const

void	updateMetadataStats (const DataType &typed_value, const bool is_date_in_days=false)

Protected Member Functions inherited from data_conversion::BaseConvertEncoder
template<typename DataType >
bool	isNull (const DataType &typed_value)

Protected Attributes inherited from data_conversion::TypedBaseConvertEncoder< IdType >
MetadataType	min_

MetadataType	max_

Protected Attributes inherited from data_conversion::BaseConvertEncoder
std::optional< std::vector < bool > >	delete_buffer_

const bool	error_tracking_enabled_

bool	has_nulls_

size_t	num_elements_

Detailed Description

template<typename IdType>
class data_conversion::StringViewToStringDictEncoder< IdType >

Definition at line 25 of file StringViewToStringDictEncoder.h.

Constructor & Destructor Documentation

template<typename IdType >

data_conversion::StringViewToStringDictEncoder< IdType >::StringViewToStringDictEncoder	(	const Chunk_NS::Chunk &	dst_chunk,
		const bool	error_tracking_enabled
	)

inline

Definition at line 27 of file StringViewToStringDictEncoder.h.

References data_conversion::StringViewToStringDictEncoder< IdType >::initialize().

       : TypedBaseConvertEncoder<IdType>(error_tracking_enabled), dst_chunk_(dst_chunk) {
     initialize();
   }

Here is the call graph for this function:

Member Function Documentation

template<typename IdType >

void data_conversion::StringViewToStringDictEncoder< IdType >::clear ( )

inlineoverridevirtual

Reimplemented from data_conversion::BaseConvertEncoder.

Definition at line 85 of file StringViewToStringDictEncoder.h.

References data_conversion::StringViewToStringDictEncoder< IdType >::buffer_, data_conversion::BaseConvertEncoder::clear(), and Data_Namespace::AbstractBuffer::resetToEmpty().

                         {
     buffer_->resetToEmpty();
     BaseConvertEncoder::clear();
   }

Here is the call graph for this function:

template<typename IdType >

void data_conversion::StringViewToStringDictEncoder< IdType >::encodeAndAppendData	(	const int8_t *	data,
		const size_t	num_elements
	)

inlineoverridevirtual

Implements data_conversion::BaseConvertEncoder.

Definition at line 33 of file StringViewToStringDictEncoder.h.

References Data_Namespace::AbstractBuffer::append(), data_conversion::StringViewToStringDictEncoder< IdType >::buffer_, CHECK, data_conversion::BaseConvertEncoder::delete_buffer_, data_conversion::StringViewToStringDictEncoder< IdType >::dict_encoding_output_buffer_, data_conversion::StringViewToStringDictEncoder< IdType >::dst_type_info_, data_conversion::BaseConvertEncoder::error_tracking_enabled_, SQLTypeInfo::get_notnull(), StringDictionary::getOrAddBulk(), data_conversion::BaseConvertEncoder::has_nulls_, data_conversion::BaseConvertEncoder::isNull(), StringDictionary::MAX_STRLEN, data_conversion::StringViewToStringDictEncoder< IdType >::string_dict_, to_string(), and data_conversion::TypedBaseConvertEncoder< DataType_, MetadataType_ >::updateMetadataStats().

                                                                                    {
     auto typed_data = reinterpret_cast<const std::string_view*>(data);
 
     CHECK(!BaseConvertEncoder::error_tracking_enabled_)
         << " unimplemented case for this encoder";
 
     std::vector<std::string_view> input_buffer(typed_data, typed_data + num_elements);
     CHECK(string_dict_);
     dict_encoding_output_buffer_.resize(num_elements);
     for (size_t i = 0; i < num_elements; ++i) {
       if (input_buffer[i].size() > StringDictionary::MAX_STRLEN) {
         if (!BaseConvertEncoder::error_tracking_enabled_) {
           throw std::runtime_error("String length of " +
                                    std::to_string(input_buffer[i].size()) +
                                    " exceeds allowed maximum string length of " +
                                    std::to_string(StringDictionary::MAX_STRLEN));
         } else {
           BaseConvertEncoder::delete_buffer_->push_back(true);
           input_buffer[i] = {};  // set to NULL/empty string to process
         }
       } else if (BaseConvertEncoder::isNull(input_buffer[i])) {
         if (!BaseConvertEncoder::error_tracking_enabled_) {
           if (dst_type_info_.get_notnull()) {
             throw std::runtime_error("NULL value not allowed in NOT NULL column");
           }
           BaseConvertEncoder::has_nulls_ = true;
         } else {
           if (dst_type_info_.get_notnull()) {
             BaseConvertEncoder::delete_buffer_->push_back(true);
           } else {
             BaseConvertEncoder::delete_buffer_->push_back(false);
           }
         }
       } else {
         if (BaseConvertEncoder::error_tracking_enabled_) {
           BaseConvertEncoder::delete_buffer_->push_back(false);
         }
       }
     }
     string_dict_->getOrAddBulk<IdType, std::string_view>(
         input_buffer, dict_encoding_output_buffer_.data());
     for (size_t i = 0; i < num_elements; ++i) {
       if (!BaseConvertEncoder::isNull(input_buffer[i])) {
         TypedBaseConvertEncoder<IdType>::updateMetadataStats(
             dict_encoding_output_buffer_[i]);
       }
     }
 
     buffer_->append(reinterpret_cast<int8_t*>(dict_encoding_output_buffer_.data()),
                     num_elements * sizeof(IdType));
   }

Here is the call graph for this function:

template<typename IdType >

std::optional<std::vector<bool> >& data_conversion::StringViewToStringDictEncoder< IdType >::getDeleteBuffer ( )

inline

Definition at line 92 of file StringViewToStringDictEncoder.h.

References data_conversion::BaseConvertEncoder::delete_buffer_.

                                                   {
     return BaseConvertEncoder::delete_buffer_;
   }

template<typename IdType >

const Chunk_NS::Chunk& data_conversion::StringViewToStringDictEncoder< IdType >::getDstChunk ( ) const

inline

Definition at line 90 of file StringViewToStringDictEncoder.h.

References data_conversion::StringViewToStringDictEncoder< IdType >::dst_chunk_.

90 { return dst_chunk_; }

data_conversion::StringViewToStringDictEncoder::dst_chunk_

const Chunk_NS::Chunk & dst_chunk_

Definition: StringViewToStringDictEncoder.h:108

template<typename IdType >

void data_conversion::StringViewToStringDictEncoder< IdType >::initialize ( )

inlineprivate

Definition at line 97 of file StringViewToStringDictEncoder.h.

References data_conversion::StringViewToStringDictEncoder< IdType >::buffer_, ColumnDescriptor::columnType, data_conversion::StringViewToStringDictEncoder< IdType >::dst_chunk_, data_conversion::StringViewToStringDictEncoder< IdType >::dst_type_info_, Chunk_NS::Chunk::getBuffer(), Catalog_Namespace::SysCatalog::getCatalog(), Chunk_NS::Chunk::getColumnDesc(), Catalog_Namespace::SysCatalog::instance(), and data_conversion::StringViewToStringDictEncoder< IdType >::string_dict_.

Referenced by data_conversion::StringViewToStringDictEncoder< IdType >::StringViewToStringDictEncoder().

                     {
     auto type_info = dst_chunk_.getColumnDesc()->columnType;
     auto catalog = Catalog_Namespace::SysCatalog::instance().getCatalog(
         type_info.getStringDictKey().db_id);
     string_dict_ = catalog->getMetadataForDict(type_info.getStringDictKey().dict_id, true)
                        ->stringDict.get();
     dst_type_info_ = type_info;
     buffer_ = dst_chunk_.getBuffer();
   }