lemur::file::SortMergeTextFiles Class Reference

#include <SortMergeTextFiles.hpp>

Inheritance diagram for lemur::file::SortMergeTextFiles:

List of all members.


Public Member Functions
	SortMergeTextFiles (std::string &outputFilePath, std::string &tempDirectory, int numMergeThreads=4, bool displayStatus=false)
	~SortMergeTextFiles ()
int	sort (std::vector< std::string > &inputFilePaths)
void	showStatus (bool displayStatus)
Static Public Member Functions
bool	_readLine (FILE _in, char &beginLine, size_t &lineLength, indri::utility::Buffer &_buffer)
Protected Member Functions
std::string	_flushChunks (std::string &basePathname, std::vector< std::string > *inMemRecords, int currentChunkNumber)
int	_mergeSortTwoFiles (std::string &firstFilePath, std::string &secondFilePath, std::string &outputFile, bool doCleanup=true)
std::vector< std::string >	_doMidFinalMerge (std::vector< std::string > &inputList, std::string &outputPathBase, int &recordCounter)
int	_doFinalMergesortFiles (std::vector< std::string > &inputFiles, std::string &outputFile)
virtual void	_doSingleFileMergesort (std::string &inputFile, std::string &outputFile, std::vector< std::string > &chunkList, int chunkRecordSize=16384 *10)
Protected Attributes
std::string	_outputFilePath
std::string	_tempDirectory
indri::utility::Buffer	_inputBuffer
indri::utility::Buffer	_inputBufferTwo
bool	_displayStatus
int	_numMergeThreads

Constructor & Destructor Documentation

lemur::file::SortMergeTextFiles::SortMergeTextFiles ( std::string & outputFilePath,

std::string & tempDirectory,

int numMergeThreads = 4,

bool displayStatus = false

)

lemur::file::SortMergeTextFiles::~SortMergeTextFiles ( )

Member Function Documentation

int lemur::file::SortMergeTextFiles::_doFinalMergesortFiles ( std::vector< std::string > & inputFiles,

std::string & outputFile

) [protected]

std::vector< std::string > lemur::file::SortMergeTextFiles::_doMidFinalMerge ( std::vector< std::string > & inputList,

std::string & outputPathBase,

int & recordCounter

) [protected]

void lemur::file::SortMergeTextFiles::_doSingleFileMergesort ( std::string & inputFile,

std::string & outputFile,

std::vector< std::string > & chunkList,

int chunkRecordSize = 16384 *10

) [protected, virtual]

Reimplemented in lemur::file::HarvestSortMerge.

std::string lemur::file::SortMergeTextFiles::_flushChunks ( std::string & basePathname,

std::vector< std::string > * inMemRecords,

int currentChunkNumber

) [protected]

int lemur::file::SortMergeTextFiles::_mergeSortTwoFiles ( std::string & firstFilePath,

std::string & secondFilePath,

std::string & outputFile,

bool doCleanup = true

) [protected]

bool lemur::file::SortMergeTextFiles::_readLine ( FILE * _in,

char *& beginLine,

size_t & lineLength,

indri::utility::Buffer & _buffer

) [static]