ConcIndexator.cpp File Reference

#include "StdConc.h"
#include "DwdsThesaurus.h"
#include "ConcIndexator.h"
#include "DocumentIterator.h"

Include dependency graph for ConcIndexator.cpp:

Functions
bool	IsWord (const CGraphmatFile &piGraphmat, long GraLine)
bool	IsDigit (const CGraphmatFile &piGraphmat, long GraLine)
bool	IsSentenceEnd (const CGraphmatFile &piGraphmat, long GraLine)
void	GetTextFromXMLRecursive (TiXmlNode *parent, string &Result)
void	GetCWBFormattedStringRecursive (const TiXmlNode *parent, string &Result)
bool	IsXmlFile (const string &FileName)
bool	CreateMorphIndex (const CStringIndexator *indexator, string path, MorphLanguageEnum language, size_t maxTokenCountInOnePeriod)

Variables
const DWORD	DefaultMaxTokenCountInOnePeriod = 5000000
	global default value (5000000) for CConcIndexator::m_UserMaxTokenCountInOnePeriod More...
const DWORD	DefaultMaxInputLoadIndexSize = 400000
	global default value for (400000) CConcIndexator::m_UserMaxInputLoadIndexSize (must be <= DefaultMaxTokenCountInOnePeriod) More...

Function Documentation

IsWord()

bool IsWord	(	const CGraphmatFile &	piGraphmat,
		long	GraLine
	)

References CUnitHolder::HasDescr(), OLLE, and ORLE.

Referenced by CConcIndexator::IsDWDSToken().

Here is the call graph for this function:

Here is the caller graph for this function:

IsDigit()

bool IsDigit	(	const CGraphmatFile &	piGraphmat,
		long	GraLine
	)

References CUnitHolder::HasDescr(), ODigits, and ONumChar.

Referenced by CConcIndexator::IsDWDSToken().

Here is the call graph for this function:

Here is the caller graph for this function:

IsSentenceEnd()

bool IsSentenceEnd	(	const CGraphmatFile &	piGraphmat,
		long	GraLine
	)

References CUnitHolder::HasDescr(), and OSentEnd.

Referenced by CConcIndexator::IndexTextOrHtmlFile(), and CConcIndexator::IsDWDSToken().

Here is the call graph for this function:

Here is the caller graph for this function:

GetTextFromXMLRecursive()

void GetTextFromXMLRecursive	(	TiXmlNode *	parent,
		string &	Result
	)

References TiXmlNode::COMMENT, TiXmlNode::ELEMENT, TiXmlNode::FirstChild(), Format(), TiXmlNode::NextSibling(), TiXmlNode::TEXT, Trim(), TiXmlNode::Type(), UnknownPageNumber, and TiXmlNode::Value().

Referenced by CConcIndexator::LoadXmlFile().

Here is the call graph for this function:

Here is the caller graph for this function:

GetCWBFormattedStringRecursive()

void GetCWBFormattedStringRecursive	(	const TiXmlNode *	parent,
		string &	Result
	)

References TiXmlElement::Attribute(), TiXmlNode::COMMENT, TiXmlNode::ELEMENT, TiXmlNode::FirstChild(), Format(), TiXmlNode::NextSibling(), PredefinedTableLineTag, TiXmlNode::TEXT, TiXmlNode::ToElement(), Trim(), TiXmlNode::Type(), and TiXmlNode::Value().

Referenced by CConcIndexator::IndexFreeIndex().

Here is the call graph for this function:

Here is the caller graph for this function:

IsXmlFile()

bool IsXmlFile

(

const string &

FileName

)

Referenced by CConcIndexator::LoadFileIntoGraphan().

Here is the caller graph for this function:

CreateMorphIndex()

bool CreateMorphIndex	(	const CStringIndexator *	indexator,
		string	path,
		MorphLanguageEnum	language,
		size_t	maxTokenCountInOnePeriod
	)

References CIndexSetForLoadingStage::AddInputLoadIndexToMemoryLoadIndex(), CIndexSetForLoadingStage::AddMemoryLoadIndexToMainLoadIndex(), CheckLanguage(), CIndexSetForLoadingStage::CreateTempFiles(), CStringIndexSet::DestroyIndexSet(), ErrorMessage(), GetGramInfosFromWord(), CStringIndexator::GetIndexByAlias(), GetIndexItemSetByVectorString(), CStringIndexSet::GetIndexItemStr(), CIndexSetForLoadingStage::InsertToInputLoadIndex(), is_upper_alpha(), CIndexSetForQueryingStage::m_Index, CIndexSetForQueryingStage::ReadAllOccurrences(), CStringIndexSet::ReadFromTheDisk(), CIndexSetForLoadingStage::SaveMemoryLoadIndex(), ddcVecFile< T >::size(), CIndexSetForLoadingStage::SortInputAndMemoryIndices(), and CStringIndexSet::WriteToFile().

Referenced by CConcIndexator::CreateMorphIndexWrapper(), and CConcIndexator::IndexOneFile().

Here is the call graph for this function:

Here is the caller graph for this function:

Variable Documentation

DefaultMaxTokenCountInOnePeriod

const DWORD DefaultMaxTokenCountInOnePeriod = 5000000

global default value (5000000) for CConcIndexator::m_UserMaxTokenCountInOnePeriod

Referenced by CConcIndexator::GetMaxTokenCountInOnePeriod(), and CConcordance::InitDefaultOptions().

DefaultMaxInputLoadIndexSize

const DWORD DefaultMaxInputLoadIndexSize = 400000

global default value for (400000) CConcIndexator::m_UserMaxInputLoadIndexSize (must be <= DefaultMaxTokenCountInOnePeriod)

Referenced by CConcIndexator::GetMaxInputLoadIndexSize(), and CConcordance::InitDefaultOptions().