#include <UnitHolder.h>
CUnitHolder::CUnitHolder | ( | ) |
References m_Language.
const vector<CGraLine>& CUnitHolder::GetUnits | ( | ) | const [inline] |
References m_Units.
Referenced by CalculateLMarg(), CanBeFileName(), CanBeRussianInitial(), CheckComma(), CGraphmatFile::CountEndL(), CGraphmatFile::CountSpaces(), CGraphmatFile::DealAbbrev(), CGraphmatFile::DealAsteriskBullet(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealFIO(), CGraphmatFile::DealGermanDividedCompounds(), DealIndention(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealModifierKey(), CGraphmatFile::DealNames(), CGraphmatFile::DealOborotto(), CGraphmatFile::DealReferences(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::DealShortFIO(), DealSimpleEnglishNames(), CGraphmatFile::FindKeySequence(), CGraphmatFile::FindOborotto(), FindSentEndAfterParagraph(), CGraphmatFile::GetGraphematicalLine(), CConSent::GetString(), GetToken(), GetTokenInputOffset(), GetTokenLength(), GetTokensCount(), CConSent::GetUnit(), GetUppercaseToken(), CGraphmatFile::GraphmatMain(), CConcIndexator::IndexTextOrHtmlFile(), CConSent::InitBasicInformation(), CGraphmatFile::InitContextDescriptors(), InitEnglishNameSlot(), CGraphmatFile::IsKey(), IsLastInGroupOrFree(), CGraphmatFile::MacSynHierarchy(), MapCorrectMinSpace(), ParagraphTagBeforeGraph(), CLemmatizer::ProcessHyphenWords(), RubiconText(), SetSentMarkers(), and CGraphmatFile::WriteGraphMat().
CGraLine & CUnitHolder::GetUnit | ( | size_t | UnitNo | ) |
References m_Units.
Referenced by CGraphmatFile::GraphmatMain(), CGraphmatFile::InitContextDescriptors(), InitEnglishNameSlot(), and MakeOneWord().
const vector< char > & CUnitHolder::GetUnitBuf | ( | ) | const |
References m_TokenBuf.
Referenced by GetUnitBufferSize(), and GetUnitBufferStart().
const vector< BYTE > & CUnitHolder::GetInputBuffer | ( | ) | const |
References m_InputBuffer.
Referenced by CGraphmatFile::GraphmatMain(), and CGraLine::ReadWord().
void CUnitHolder::SetState | ( | size_t | LB, | |
size_t | HB, | |||
WORD | state | |||
) |
References m_Units.
Referenced by CGraphmatFile::DealAbbrev(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealFIO(), CGraphmatFile::DealGermanDividedCompounds(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealModifierKey(), CGraphmatFile::DealOborotto(), CGraphmatFile::DealQuotedKeySequence(), CGraphmatFile::DealReferences(), CGraphmatFile::DealShortFIO(), DealSimpleEnglishNames(), and CGraphmatFile::DealSimpleKey().
void CUnitHolder::SetDes | ( | size_t | x, | |
Descriptors | des | |||
) |
References m_Units.
Referenced by CGraphmatFile::DealAbbrev(), CGraphmatFile::DealAsteriskBullet(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealFIO(), CGraphmatFile::DealGermanDividedCompounds(), DealIndention(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealModifierKey(), CGraphmatFile::DealNames(), CGraphmatFile::DealOborotto(), CGraphmatFile::DealQuotedKeySequence(), CGraphmatFile::DealReferences(), CGraphmatFile::DealShortFIO(), DealSimpleEnglishNames(), CGraphmatFile::DealSimpleKey(), InitTokenBuffer(), CGraphmatFile::MacSynHierarchy(), MakeOneWord(), and SetSentMarkers().
bool CUnitHolder::HasDescr | ( | size_t | i, | |
int | descr | |||
) | const [inline] |
Referenced by CanBeFileName(), CanBeRussianInitial(), CheckComma(), CheckGermanSentenceBreak(), ClearPairDescriptors(), CGraphmatFile::DealAbbrev(), CGraphmatFile::DealAsteriskBullet(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealFIO(), CGraphmatFile::DealGermanDividedCompounds(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealModifierKey(), CGraphmatFile::DealNames(), CGraphmatFile::DealQuotedKeySequence(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::DealShortFIO(), DealSimpleEnglishNames(), FindHeadings(), CGraphmatFile::FindKeySequence(), FindSentEndAfterParagraph(), FirstUpper(), CConSent::GetString(), CGraphmatFile::GetTokenLanguage(), CGraphmatFile::GraphmatMain(), CGraphmatFile::HasIndention(), CConcIndexator::IndexTextOrHtmlFile(), CConSent::InitBasicInformation(), CGraphmatFile::InitContextDescriptors(), InitEnglishNameSlot(), CConSent::IsBullet(), IsBulletWord(), IsDigit(), CConcIndexator::IsDWDSToken(), IsHyphen(), IsLastInGroupOrFree(), IsOneAlpha(), CConSent::IsParag(), IsSentenceEnd(), IsWord(), MakeOneWord(), PPunct(), PPunctOrSoft(), CLemmatizer::ProcessHyphenWords(), CPlmLineCollection::ProcessPlmLines(), RubiconText(), SetDepthOfBullet(), and SetSentMarkers().
bool CUnitHolder::AreGrouped | ( | size_t | LB, | |
size_t | HB | |||
) | const |
References m_Units.
bool CUnitHolder::HasGrouped | ( | size_t | LB, | |
size_t | HB | |||
) | const |
References m_Units.
Referenced by CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealModifierKey(), DealSimpleEnglishNames(), and CGraphmatFile::DealSimpleKey().
bool CUnitHolder::HasAbbreviation | ( | size_t | LB, | |
size_t | HB | |||
) | const |
References m_Units.
Referenced by CGraphmatFile::DealExtensionsAndLocalFileNames().
size_t CUnitHolder::PassSpace | ( | size_t | i, | |
size_t | HB | |||
) | const |
References m_Units.
Referenced by AdjustRubiconsInSpaceCompact(), CGraphmatFile::DealAsteriskBullet(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), CGraphmatFile::DealReferences(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::InitContextDescriptors(), and MapCorrectMinSpace().
size_t CUnitHolder::PPunctOrSoft | ( | size_t | i, | |
size_t | HB | |||
) | const |
References HasDescr(), m_Units, and OPun.
Referenced by CGraphmatFile::DealSentBreaker().
size_t CUnitHolder::PPunct | ( | size_t | i, | |
size_t | HB | |||
) | const |
References HasDescr(), and OPun.
Referenced by CGraphmatFile::DealSentBreaker().
size_t CUnitHolder::FindSpace | ( | size_t | i, | |
size_t | HB | |||
) | const |
References m_Units.
size_t CUnitHolder::BSpace | ( | size_t | i, | |
size_t | LB = 0 | |||
) | const |
References m_Units.
Referenced by AdjustRubiconsInSpaceCompact(), CGraphmatFile::DealAsteriskBullet(), CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), DealIndention(), EmptyLineBeforeGraph(), and CConSent::InitBasicInformation().
size_t CUnitHolder::PSoft | ( | size_t | i, | |
size_t | HB | |||
) | const |
References m_Units.
Referenced by CGraphmatFile::DealAbbrev(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealFIO(), CGraphmatFile::DealGermanDividedCompounds(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::DealShortFIO(), CGraphmatFile::FindKeySequence(), CGraphmatFile::FindOborotto(), CConSent::InitBasicInformation(), CGraphmatFile::MacSynHierarchy(), CLemmatizer::ProcessHyphenWords(), and RecognizeSimpleCS().
size_t CUnitHolder::BSoft | ( | size_t | i | ) | const |
References m_Units.
Referenced by CheckGermanSentenceBreak(), CGraphmatFile::DealFIO(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::DealShortFIO(), CGraphmatFile::FindKeySequence(), FindSentEndAfterParagraph(), and CConSent::InitBasicInformation().
bool CUnitHolder::IsHyphen | ( | size_t | x | ) | const |
bool CUnitHolder::is_latin_alpha | ( | int | ch | ) | const |
References is_english_alpha(), is_german_alpha(), m_Language, and morphGerman.
Referenced by CGraphmatFile::InitNonContextDescriptors(), and IsOneAlpha().
bool CUnitHolder::is_lowercase | ( | int | ch | ) | const |
References is_english_lower(), is_german_lower(), is_russian_lower(), m_Language, and morphGerman.
Referenced by CGraphmatFile::InitNonContextDescriptors(), and TestHeadingFirstLetter().
bool CUnitHolder::is_uppercase | ( | int | ch | ) | const |
References is_english_upper(), is_german_upper(), is_russian_upper(), m_Language, and morphGerman.
Referenced by CGraphmatFile::InitNonContextDescriptors().
bool CUnitHolder::StrSuperCompare | ( | int | UnitNo, | |
const char * | s | |||
) | const |
References GetToken(), GetTokenLength(), m_Language, m_Units, and strscmp().
Referenced by CConSent::IsFirstWord().
bool CUnitHolder::IsOneAlpha | ( | size_t | x | ) | const |
References GetTokenLength(), HasDescr(), is_latin_alpha(), m_Units, OLLE, and ORLE.
Referenced by CGraphmatFile::DealSimpleKey(), IsBulletWord(), and CGraphmatFile::IsKey().
bool CUnitHolder::IsOneChar | ( | size_t | x, | |
int | i | |||
) | const |
References m_Units.
Referenced by CheckComma(), and CGraphmatFile::DealGermanDividedCompounds().
bool CUnitHolder::IsOneULet | ( | size_t | x | ) | const |
bool CUnitHolder::FirstUpper | ( | size_t | x | ) | const |
References HasDescr(), OUp, and OUpLw.
Referenced by CGraphmatFile::DealFIO(), and CGraphmatFile::DealShortFIO().
bool CUnitHolder::IsBulletWord | ( | size_t | x | ) | const |
References HasDescr(), IsOneAlpha(), and ODigits.
Referenced by CGraphmatFile::DealBullet(), CGraphmatFile::DealBulletsWithTwoBrackets(), and CConSent::SetUsualLawDocumentDivision().
bool CUnitHolder::IsOneFullStop | ( | size_t | i | ) | const |
References GetTokenLength(), and m_Units.
Referenced by CheckGermanSentenceBreak(), CGraphmatFile::DealBullet(), CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealFIO(), CGraphmatFile::DealReferences(), CGraphmatFile::DealShortFIO(), IsSentenceEndMark(), and SetDepthOfBullet().
bool CUnitHolder::EmptyLineBeforeGraph | ( | size_t | i, | |
size_t | HB | |||
) | const |
References BSpace(), GetToken(), GetTokenLength(), and m_Units.
Referenced by RubiconText().
bool CUnitHolder::IsQuestionOrExclamationMarks | ( | size_t | i | ) | const |
References GetToken(), GetTokenLength(), and m_Units.
Referenced by IsSentenceEndMark().
bool CUnitHolder::IsSentenceEndMark | ( | size_t | i | ) | const |
References cEllipseChar, GetTokenLength(), IsOneFullStop(), IsQuestionOrExclamationMarks(), and m_Units.
Referenced by CGraphmatFile::DealSentBreaker(), FindSentEndAfterParagraph(), and CConSent::InitBasicInformation().
bool CUnitHolder::IsOneCloseQuotationMark | ( | size_t | i | ) | const |
References GetToken(), GetTokenLength(), and m_Units.
Referenced by CGraphmatFile::DealQuotedKeySequence(), and CGraphmatFile::DealSentBreaker().
bool CUnitHolder::IsOneOpenQuotationMark | ( | size_t | i | ) | const |
References GetToken(), GetTokenLength(), and m_Units.
Referenced by CGraphmatFile::DealQuotedKeySequence(), and SetSentMarkers().
void CUnitHolder::FreeTable | ( | ) |
References ClearVector(), m_FoundOborots, m_FoundPageBreaks, m_TokenBuf, and m_Units.
Referenced by CMorphologyHolder::GetMorphology(), CConcIndexator::IndexTextOrHtmlFile(), InitTokenBuffer(), and CGraphmatFile::~CGraphmatFile().
void CUnitHolder::BuildUnitBufferUpper | ( | ) |
References GerEngRusMakeUpperTemplate(), m_Language, m_UnitBufUpper, and m_Units.
Referenced by CGraphmatFile::GraphmatMain().
void CUnitHolder::InitTokenBuffer | ( | ) |
References FreeTable(), m_InputBuffer, m_TokenBuf, m_Units, OBeg, and SetDes().
Referenced by CGraphmatFile::GraphmatMain().
bool CUnitHolder::InitInputBuffer | ( | const string & | S | ) |
References m_InputBuffer.
Referenced by CGraphmatFile::LoadFileToGraphan(), and CGraphmatFile::LoadStringToGraphan().
void CUnitHolder::ClearInputBuffer | ( | ) |
References ClearVector(), and m_InputBuffer.
Referenced by CGraphmatFile::GraphmatMain().
void CUnitHolder::AddUnit | ( | const CGraLine & | NewLine | ) |
References m_Units.
Referenced by CGraphmatFile::GraphmatMain().
const char * CUnitHolder::GetUnitBufferStart | ( | ) | const |
References GetUnitBuf().
Referenced by GetUppercaseToken(), and CGraphmatFile::GraphmatMain().
const char * CUnitHolder::GetUnitUpperBufferStart | ( | ) | const |
References m_UnitBufUpper.
Referenced by GetUppercaseToken(), and MakeOneWord().
const char * CUnitHolder::GetUppercaseToken | ( | DWORD | LineNo | ) | const |
References GetUnitBufferStart(), GetUnits(), and GetUnitUpperBufferStart().
Referenced by CheckGermanSentenceBreak(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealGermanDividedCompounds(), CGraphmatFile::DealOborotto(), CGraphmatFile::DealReferences(), CConcIndexator::IndexTextOrHtmlFile(), InitEnglishNameSlot(), MakeOneWord(), and CPlmLineCollection::ProcessPlmLines().
string CUnitHolder::GetToken | ( | DWORD | LineNo | ) | const |
References CriticalTokenLength, GetTokenLength(), and GetUnits().
Referenced by EmptyLineBeforeGraph(), CGraphmatFile::GraphmatMain(), CGraphmatFile::IsKey(), IsOneCloseQuotationMark(), IsOneOpenQuotationMark(), IsQuestionOrExclamationMarks(), MakeOneWord(), CLemmatizer::ProcessHyphenWords(), CPlmLineCollection::ProcessPlmLines(), and StrSuperCompare().
size_t CUnitHolder::GetTokensCount | ( | ) | const |
References GetUnits().
Referenced by CMorphologyHolder::GetMorphology(), CConcIndexator::IndexTextOrHtmlFile(), CLemmatizer::ProcessHyphenWords(), and CPlmLineCollection::ProcessPlmLines().
References GetUnits().
Referenced by CGraphmatFile::DealSentBreaker(), and SetSentMarkers().
References GetUnits().
Referenced by CGraphmatFile::DealEnglishStyleFIO(), CGraphmatFile::DealExtensionsAndLocalFileNames(), CGraphmatFile::DealGermanDividedCompounds(), EmptyLineBeforeGraph(), CGraphmatFile::FindKeySequence(), GetToken(), CGraphmatFile::GraphmatMain(), IsOneAlpha(), IsOneCloseQuotationMark(), IsOneFullStop(), IsOneOpenQuotationMark(), IsQuestionOrExclamationMarks(), IsSentenceEndMark(), MakeOneWord(), and StrSuperCompare().
size_t CUnitHolder::GetUnitBufferSize | ( | ) | const |
void CUnitHolder::DeleteDescr | ( | size_t | LineNo, | |
Descriptors | d | |||
) |
References m_Units.
Referenced by ClearPairDescriptors(), CGraphmatFile::DealKeySequence(), CGraphmatFile::DealQuotedKeySequence(), and CGraphmatFile::InitContextDescriptors().
void CUnitHolder::MakeOneWord | ( | size_t | StartLineNo, | |
size_t | EndLineNo | |||
) |
References CGraLine::AddLength(), ClearPairDescriptors(), GetToken(), GetTokenLength(), GetUnit(), GetUnitUpperBufferStart(), GetUppercaseToken(), HasDescr(), IsEndTextPeriodDescriptor(), m_Language, m_UnitBufUpper, m_Units, RmlMakeUpper(), SetDes(), and SetOborotNo().
Referenced by CLemmatizer::ProcessHyphenWords().
void CUnitHolder::ClearPairDescriptors | ( | size_t | StartLineNo, | |
size_t | EndLineNo | |||
) |
References DeleteDescr(), GetFirstMemberByTheSecond(), GetSecondMemberByTheFirst(), HasDescr(), IsFirstMemberOfPairDesciptor(), IsSecondMemberOfPairDesciptor(), and OSentEnd.
Referenced by MakeOneWord().
void CUnitHolder::SetOborotNo | ( | size_t | LineNo, | |
short | OborotNo | |||
) |
References m_FoundOborots, and m_Units.
Referenced by CGraphmatFile::DealOborotto(), and MakeOneWord().
short CUnitHolder::GetOborotNo | ( | size_t | LineNo | ) | const |
References m_FoundOborots, and m_Units.
Referenced by CGraphmatFile::GetGraphematicalLine(), and CGraphmatFile::StartsFixedOborot().
void CUnitHolder::SetPageNumber | ( | size_t | LineNo, | |
DWORD | PageNumber | |||
) |
References m_FoundPageBreaks, m_Units, and UnknownPageNumber.
Referenced by CGraphmatFile::GraphmatMain().
DWORD CUnitHolder::GetPageNumber | ( | size_t | LineNo | ) | const |
References m_FoundPageBreaks, m_Units, and UnknownPageNumber.
Referenced by CGraphmatFile::GetGraphematicalLine(), and CConcIndexator::IndexTextOrHtmlFile().
vector<CGraLine> CUnitHolder::m_Units [private] |
Referenced by AddUnit(), AreGrouped(), BSoft(), BSpace(), BuildUnitBufferUpper(), DeleteDescr(), EmptyLineBeforeGraph(), FindSpace(), FreeTable(), GetOborotNo(), GetPageNumber(), GetUnit(), GetUnits(), HasAbbreviation(), HasDescr(), HasGrouped(), InitTokenBuffer(), IsHyphen(), IsOneAlpha(), IsOneChar(), IsOneCloseQuotationMark(), IsOneFullStop(), IsOneOpenQuotationMark(), IsQuestionOrExclamationMarks(), IsSentenceEndMark(), MakeOneWord(), PassSpace(), PPunctOrSoft(), PSoft(), SetDes(), SetOborotNo(), SetPageNumber(), SetState(), and StrSuperCompare().
vector<char> CUnitHolder::m_TokenBuf [private] |
Referenced by FreeTable(), GetUnitBuf(), and InitTokenBuffer().
vector<char> CUnitHolder::m_UnitBufUpper [private] |
Referenced by BuildUnitBufferUpper(), GetUnitUpperBufferStart(), and MakeOneWord().
vector<BYTE> CUnitHolder::m_InputBuffer [private] |
Referenced by ClearInputBuffer(), GetInputBuffer(), InitInputBuffer(), and InitTokenBuffer().
map<size_t, short> CUnitHolder::m_FoundOborots [private] |
Referenced by FreeTable(), GetOborotNo(), and SetOborotNo().
map<size_t, DWORD> CUnitHolder::m_FoundPageBreaks [private] |
Referenced by FreeTable(), GetPageNumber(), and SetPageNumber().
Referenced by BuildUnitBufferUpper(), CUnitHolder(), CGraphmatFile::DealSentBreaker(), CGraphmatFile::FindKeySequence(), CGraphmatFile::GetTokenLanguage(), CGraphmatFile::InitContextDescriptors(), CConcIndexator::InitGraphanProperties(), CGraphmatFile::InitNonContextDescriptors(), is_latin_alpha(), is_lowercase(), is_uppercase(), CGraphmatFile::IsKey(), CGraLine::LengthUntilDelimiters(), CGraphmatFile::LoadDicts(), CMorphologyHolder::LoadGraphanAndLemmatizer(), MakeOneWord(), and StrSuperCompare().