ConcIndexator.cpp File Reference

#include "StdConc.h"
#include "../common/util_classes.h"
#include "../common/DwdsThesaurus.h"
#include "ConcCommon.h"
#include "ConcIndexator.h"
#include "sys/stat.h"
#include "time.h"
#include "limits.h"
#include "ConcordAlgorithm.h"
#include "../tinyxml/tinyxml.h"
#include "../RmlTar/RmlTar.h"
#include "../GraphanLib/HTMLConv.h"

Include dependency graph for ConcIndexator.cpp:

Functions

• template<class T , class Y > void InitMorphologySystem (T *&Lemmatizer, Y *&Gramtab)
• const CLemmatizer * GetLemmatizerByLanguage (MorphLanguageEnum Langua)

  return a morphology dictionary by a language indentifier
  

• const CAgramtab * GetGramtabByLanguage (MorphLanguageEnum Langua)

  return a grammatical table by a language indentifier
  

• bool InitConcordDicts ()

  initializes morphology dictionaries
  

• void FreeConcordDicts ()

  deletes morphology dictionaries
  

• void concord_daemon_log (const string &t)

  write a dump message to a log file
  

• bool IsWord (const CGraphmatFile *piGraphmat, long GraLine)
• bool IsDigit (const CGraphmatFile *piGraphmat, long GraLine)
• bool IsSentenceEnd (const CGraphmatFile *piGraphmat, long GraLine)
• vector< string > GetGramInfosFromWord (const char *GraLine, MorphLanguageEnum Langua, bool bCapital)
• bool GetTextFromXMLRecursive (TiXmlNode *parent, string &Result)
• bool GetCWBFormattedStringRecursive (const TiXmlNode *parent, string &Result)
• bool IsXmlFile (const string &FileName)
• bool IsTarFile (const string &FileName)
• bool ReadSourceFile (const string &SourceFileName, string &CorpusFileName, TAR *pTar, vector< char > &Buffer, bool &bError)
• void newhandler ()

Variables

• bool bEnglishMorph = false
• bool bRussianMorph = false
• bool bGermanMorph = false
• CLemmatizerEnglish * engLemmatizer = 0
• CLemmatizerRussian * rusLemmatizer = 0
• CLemmatizerGerman * gerLemmatizer = 0
• CEngGramTab * engGramTab = 0
• CRusGramTab * rusGramTab = 0
• CGerGramTab * gerGramTab = 0

---

Function Documentation

template<class T , class Y >

void InitMorphologySystem	(	T *&	Lemmatizer,
		Y *&	Gramtab
	)

References ErrorMessage(), Format(), and GetStringByLanguage().

Referenced by GetGramtabByLanguage(), GetLemmatizerByLanguage(), and InitConcordDicts().

Here is the call graph for this function:

Here is the caller graph for this function:

const CLemmatizer* GetLemmatizerByLanguage

(

MorphLanguageEnum

Langua

)

return a morphology dictionary by a language indentifier

References bEnglishMorph, bGermanMorph, bRussianMorph, InitMorphologySystem(), morphEnglish, morphGerman, and morphRussian.

Referenced by GetGramInfosFromWord(), GetParadigmCollection(), and GetWordForms().

Here is the call graph for this function:

Here is the caller graph for this function:

const CAgramtab* GetGramtabByLanguage

(

MorphLanguageEnum

Langua

)

return a grammatical table by a language indentifier

References bEnglishMorph, bGermanMorph, bRussianMorph, InitMorphologySystem(), morphEnglish, morphGerman, and morphRussian.

Referenced by GetGramInfosFromWord(), GetGramInfoStr(), GetParadigmByGroups(), GetParadigmFromDictionary(), and GetStringByParadigm().

Here is the call graph for this function:

Here is the caller graph for this function:

bool InitConcordDicts

(

)

initializes morphology dictionaries

References bEnglishMorph, bGermanMorph, bRussianMorph, InitMorphologySystem(), and CExpc::m_strCause.

Referenced by main().

Here is the call graph for this function:

Here is the caller graph for this function:

void FreeConcordDicts

(

)

deletes morphology dictionaries

Referenced by UnloadData().

Here is the caller graph for this function:

void concord_daemon_log

(

const string &

t

)

write a dump message to a log file

References Format(), GetRegistryString(), and RmlGetCurrentTime().

Referenced by CConcHolder::GetFileSnippets(), CDDCServerListenHost::GetFirstHitsFromCorpora(), CConcHolder::GetHits(), CDDCServerListenHost::GetHitStringsFromOneCorpora(), CConcHolder::GetOccurrences(), CConcHolder::GetTokensFromStorageByBreak(), LoadDDCServer(), CDDCCorpusListenHost::LoadHolder(), LoadLocalCorpora(), CConcIndexator::LoadMaskedFiles(), CConcIndexator::LoadSourceFilesAndOptions(), main(), CDDCServerListenHost::ProcessSocketString(), CDDCCorpusListenHost::ProcessSocketString(), ReloadCorpus(), CDDCServerListenHost::ReloadCorpusProcessSignal(), CDDCServerListenHost::ReplaceHolderIfNeeded(), CConcHolder::SaveOccurrences(), SaveTrigger(), CConcHolder::SimpleQuery(), and termination_handler().

Here is the call graph for this function:

Here is the caller graph for this function:

bool IsWord	(	const CGraphmatFile *	piGraphmat,
		long	GraLine
	)			[inline]

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:

bool IsDigit	(	const CGraphmatFile *	piGraphmat,
		long	GraLine
	)			[inline]

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:

bool IsSentenceEnd	(	const CGraphmatFile *	piGraphmat,
		long	GraLine
	)			[inline]

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:

vector<string> GetGramInfosFromWord	(	const char *	GraLine,
		MorphLanguageEnum	Langua,
		bool	bCapital
	)

References _QM, gEIG, CLemmatizer::GetAllAncodesQuick(), GetGramtabByLanguage(), GetLemmatizerByLanguage(), Grammems, GrammemsCount, gSUB, MorphAnnotationsDelim, morphGerman, and UnknownPartOfSpeech.

Referenced by CConcIndexator::CreateMorphIndex().

Here is the call graph for this function:

Here is the caller graph for this function:

bool GetTextFromXMLRecursive	(	TiXmlNode *	parent,
		string &	Result
	)

References ErrorMessage(), TiXmlNode::FirstChild(), Format(), TiXmlNode::NextSibling(), 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:

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

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

Referenced by CConcIndexator::IndexTable().

Here is the call graph for this function:

Here is the caller graph for this function:

bool IsXmlFile

(

const string &

FileName

)

Referenced by CConcIndexator::LoadFileIntoGraphan().

Here is the caller graph for this function:

bool IsTarFile

(

const string &

FileName

)

Referenced by CConcIndexatorInvoker::BuildIndex().

Here is the caller graph for this function:

bool ReadSourceFile	(	const string &	SourceFileName,
		string &	CorpusFileName,
		TAR *	pTar,
		vector< char > &	Buffer,
		bool &	bError
	)

References HTML::GetTextFromHTMLBuffer(), IsHtmlFile(), ReadVector(), and tar_get_next_file().

Referenced by CConcIndexatorInvoker::BuildIndex().

Here is the call graph for this function:

Here is the caller graph for this function:

void newhandler

(

)

Referenced by CConcIndexatorInvoker::BuildIndex().

Here is the caller graph for this function:

---

Variable Documentation

bool bEnglishMorph = false

Referenced by GetGramtabByLanguage(), GetLemmatizerByLanguage(), and InitConcordDicts().

bool bRussianMorph = false

Referenced by GetGramtabByLanguage(), GetLemmatizerByLanguage(), and InitConcordDicts().

bool bGermanMorph = false

Referenced by GetGramtabByLanguage(), GetLemmatizerByLanguage(), and InitConcordDicts().

CLemmatizerEnglish* engLemmatizer = 0

CLemmatizerRussian* rusLemmatizer = 0

CLemmatizerGerman* gerLemmatizer = 0

CEngGramTab* engGramTab = 0

CRusGramTab* rusGramTab = 0

CGerGramTab* gerGramTab = 0