SNAP Library 6.0, Developer Reference
2020-12-09 16:24:20
SNAP, a general purpose, high performance system for analysis and manipulation of large networks
|
#include <unicode.h>
Public Types | |
enum | { ccStarter = 0, ccOverlaysAndInterior = 1, ccNuktas = 7, ccHiraganaKatakanaVoicingMarks = 8, ccViramas = 9, ccFixedPositionStart = 10, ccFixedPositionEnd = 199, ccBelowLeftAttached = 200, ccBelowAttached = 202, ccBelowRightAttached = 204, ccLeftAttached = 208, ccRightAttached = 210, ccAboveLeftAttached = 212, ccAboveAttached = 214, ccAboveRightAttached = 216, ccBelowLeft = 218, ccBelow = 220, ccBelowRight = 222, ccLeft = 224, ccRight = 226, ccAboveLeft = 228, ccAbove = 230, ccAboveRight = 232, ccDoubleBelow = 233, ccDoubleAbove = 234, ccBelowIotaSubscript = 240, ccInvalid = 255 } |
Public Member Functions | |
void | InitAfterLoad () |
void | SetCatAndSubCat (const TUniChSubCategory catAndSubCat) |
void | Save (TSOut &SOut) const |
void | Load (TSIn &SIn) |
TUniChInfo (TSIn &SIn) | |
TUniChInfo () | |
bool | IsDcpFlag (const TUniChFlags flag) const |
void | ClrDcpFlags () |
void | SetDcpFlag (const TUniChFlags flag) |
bool | IsAlphabetic () const |
bool | IsUppercase () const |
bool | IsLowercase () const |
bool | IsMath () const |
bool | IsDefaultIgnorable () const |
bool | IsGraphemeBase () const |
bool | IsGraphemeExtend () const |
bool | IsIdStart () const |
bool | IsIdContinue () const |
bool | IsXidStart () const |
bool | IsXidContinue () const |
bool | IsProperty (const TUniChProperties flag) const |
void | SetProperty (const TUniChProperties flag) |
bool | IsAsciiHexDigit () const |
bool | IsBidiControl () const |
bool | IsDash () const |
bool | IsDeprecated () const |
bool | IsDiacritic () const |
bool | IsExtender () const |
bool | IsGraphemeLink () const |
bool | IsHexDigit () const |
bool | IsHyphen () const |
bool | IsIdeographic () const |
bool | IsJoinControl () const |
bool | IsLogicalOrderException () const |
bool | IsNoncharacter () const |
bool | IsQuotationMark () const |
bool | IsSoftDotted () const |
bool | IsSTerminal () const |
bool | IsTerminalPunctuation () const |
bool | IsVariationSelector () const |
bool | IsWhiteSpace () const |
bool | IsPropertyX (const TUniChPropertiesX flag) const |
void | SetPropertyX (const TUniChPropertiesX flag) |
bool | IsCompositionExclusion () const |
bool | IsCompatibilityDecomposition () const |
bool | IsWbFlag (const TUniChFlags flag) const |
void | ClrWbAndSbFlags () |
void | SetWbFlag (const TUniChFlags flag) |
int | GetWbFlags () const |
bool | IsWbFormat () const |
TStr | GetWbFlagsStr () const |
bool | IsSbFlag (const TUniChFlags flag) const |
void | SetSbFlag (const TUniChFlags flag) |
int | GetSbFlags () const |
bool | IsSbFormat () const |
TStr | GetSbFlagsStr () const |
bool | IsSbSep () const |
bool | IsGbExtend () const |
bool | IsCased () const |
TUniChCategory | GetCat () const |
TUniChSubCategory | GetSubCat () const |
bool | IsCurrency () const |
bool | IsPrivateUse () const |
bool | IsSurrogate () const |
Static Public Member Functions | |
static ushort | GetLineBreakCode (char c1, char c2) |
static void | LoadUShort (TSIn &SIn, ushort &u) |
static void | LoadSChar (TSIn &SIn, signed char &u) |
static void | SaveUShort (TSOut &SOut, ushort u) |
static void | SaveSChar (TSOut &SOut, signed char u) |
static TStr | GetWbFlagsStr (const int flags) |
static TStr | GetSbFlagsStr (const int flags) |
static bool | IsValidSubCat (const char chCat, const char chSubCat) |
Public Attributes | |
char | chCat |
char | chSubCat |
uchar | combClass |
TUniChCategory | cat |
TUniChSubCategory | subCat |
signed char | script |
int | simpleUpperCaseMapping |
int | simpleLowerCaseMapping |
int | simpleTitleCaseMapping |
int | decompOffset |
int | nameOffset |
int | flags |
int | properties |
int | propertiesX |
ushort | lineBreak |
Static Public Attributes | |
static const ushort | LineBreak_Unknown = TUniChInfo::GetLineBreakCode('X', 'X') |
static const ushort | LineBreak_ComplexContext = TUniChInfo::GetLineBreakCode('S', 'A') |
static const ushort | LineBreak_Numeric = TUniChInfo::GetLineBreakCode('N', 'U') |
static const ushort | LineBreak_InfixNumeric = TUniChInfo::GetLineBreakCode('I', 'S') |
static const ushort | LineBreak_Quotation = TUniChInfo::GetLineBreakCode('Q', 'U') |
Friends | |
class | TUniChDb |
anonymous enum |
Definition at line 988 of file unicode.h.
|
inlineexplicit |
|
inline |
Definition at line 1062 of file unicode.h.
References InitAfterLoad().
|
inline |
|
inline |
Definition at line 1116 of file unicode.h.
References ucfSbMask, and ucfWbMask.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
|
inlinestatic |
Definition at line 1031 of file unicode.h.
Referenced by TUniChDb::InitLineBreaks().
|
inline |
Definition at line 1128 of file unicode.h.
References ucfSbMask.
Referenced by GetSbFlagsStr(), and TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1130 of file unicode.h.
References GetSbFlags(), and GetSbFlagsStr().
Referenced by GetSbFlagsStr(), TUniChDb::InitWordAndSentenceBoundaryFlags(), and TUniChDb::TestFindNextWordOrSentenceBoundary().
|
inlinestatic |
Definition at line 1131 of file unicode.h.
References ucfSbATerm, ucfSbClose, ucfSbFormat, ucfSbLower, ucfSbNumeric, ucfSbOLetter, ucfSbSep, ucfSbSp, ucfSbSTerm, and ucfSbUpper.
|
inline |
|
inline |
Definition at line 1118 of file unicode.h.
References ucfWbMask.
Referenced by GetWbFlagsStr(), and TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1120 of file unicode.h.
References GetWbFlags(), and GetWbFlagsStr().
Referenced by GetWbFlagsStr(), and TUniChDb::TestFindNextWordOrSentenceBoundary().
|
inlinestatic |
Definition at line 1121 of file unicode.h.
References ucfWbALetter, ucfWbExtendNumLet, ucfWbFormat, ucfWbKatakana, ucfWbMidLetter, ucfWbMidNum, and ucfWbNumeric.
|
inline |
Definition at line 1035 of file unicode.h.
Referenced by Load(), TUniChDb::LoadTxt(), and TUniChInfo().
|
inline |
Definition at line 1071 of file unicode.h.
References IsDcpFlag(), and ucfDcpAlphabetic.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), and TUStr::IsAlphabetic().
|
inline |
Definition at line 1086 of file unicode.h.
References IsProperty(), and ucfPrAsciiHexDigit.
|
inline |
Definition at line 1087 of file unicode.h.
References IsProperty(), and ucfPrBidiControl.
|
inline |
Definition at line 1142 of file unicode.h.
References IsLowercase(), and IsUppercase().
Referenced by TUStr::IsCase().
|
inline |
Definition at line 1112 of file unicode.h.
References ucfCompatibilityDecomposition.
Referenced by TUniChDb::AddDecomposition(), and TUniChDb::LoadTxt().
|
inline |
Definition at line 1111 of file unicode.h.
References ucfCompositionExclusion.
Referenced by TUniChDb::LoadTxt().
|
inline |
Definition at line 1088 of file unicode.h.
References IsProperty(), and ucfPrDash.
|
inline |
Definition at line 1068 of file unicode.h.
References Assert, and ucfDcpMask.
Referenced by TUniChDb::InitDerivedCoreProperties(), IsAlphabetic(), IsDefaultIgnorable(), IsGraphemeBase(), IsGraphemeExtend(), IsIdContinue(), IsIdStart(), IsLowercase(), IsMath(), IsUppercase(), IsXidContinue(), and IsXidStart().
|
inline |
Definition at line 1075 of file unicode.h.
References IsDcpFlag(), and ucfDcpDefaultIgnorableCodePoint.
|
inline |
Definition at line 1089 of file unicode.h.
References IsProperty(), and ucfPrDeprecated.
|
inline |
Definition at line 1090 of file unicode.h.
References IsProperty(), and ucfPrDiacritic.
|
inline |
Definition at line 1091 of file unicode.h.
References IsProperty(), and ucfPrExtender.
|
inline |
Definition at line 1139 of file unicode.h.
References IsGraphemeExtend().
Referenced by TUniChDb::IsWbIgnored().
|
inline |
Definition at line 1076 of file unicode.h.
References IsDcpFlag(), and ucfDcpGraphemeBase.
|
inline |
Definition at line 1077 of file unicode.h.
References IsDcpFlag(), and ucfDcpGraphemeExtend.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), and IsGbExtend().
|
inline |
Definition at line 1092 of file unicode.h.
References IsProperty(), and ucfPrGraphemeLink.
|
inline |
Definition at line 1093 of file unicode.h.
References IsProperty(), and ucfPrHexDigit.
|
inline |
Definition at line 1094 of file unicode.h.
References IsProperty(), and ucfPrHyphen.
|
inline |
Definition at line 1079 of file unicode.h.
References IsDcpFlag(), and ucfDcpIdContinue.
|
inline |
Definition at line 1095 of file unicode.h.
References IsProperty(), and ucfPrIdeographic.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1078 of file unicode.h.
References IsDcpFlag(), and ucfDcpIdStart.
|
inline |
Definition at line 1096 of file unicode.h.
References IsProperty(), and ucfPrJoinControl.
|
inline |
Definition at line 1097 of file unicode.h.
References IsProperty(), and ucfPrLogicalOrderException.
|
inline |
Definition at line 1073 of file unicode.h.
References IsDcpFlag(), and ucfDcpLowercase.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), IsCased(), and TUStr::IsLowerCase().
|
inline |
Definition at line 1074 of file unicode.h.
References IsDcpFlag(), and ucfDcpMath.
Referenced by TUStr::IsMath().
|
inline |
Definition at line 1098 of file unicode.h.
References IsProperty(), and ucfPrNoncharacterCodePoint.
|
inline |
Definition at line 1084 of file unicode.h.
Referenced by TUniChDb::InitPropList(), IsAsciiHexDigit(), IsBidiControl(), IsDash(), IsDeprecated(), IsDiacritic(), IsExtender(), IsGraphemeLink(), IsHexDigit(), IsHyphen(), IsIdeographic(), IsJoinControl(), IsLogicalOrderException(), IsNoncharacter(), IsQuotationMark(), IsSoftDotted(), IsSTerminal(), IsTerminalPunctuation(), IsVariationSelector(), and IsWhiteSpace().
|
inline |
Definition at line 1107 of file unicode.h.
Referenced by TUniChDb::InitPropList().
|
inline |
Definition at line 1099 of file unicode.h.
References IsProperty(), and ucfPrQuotationMark.
|
inline |
Definition at line 1126 of file unicode.h.
References Assert, and ucfSbMask.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), and IsSbFormat().
|
inline |
Definition at line 1129 of file unicode.h.
References IsSbFlag(), and ucfSbFormat.
|
inline |
|
inline |
Definition at line 1100 of file unicode.h.
References IsProperty(), and ucfPrSoftDotted.
|
inline |
Definition at line 1101 of file unicode.h.
References IsProperty(), and ucfPrSTerm.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1102 of file unicode.h.
References IsProperty(), and ucfPrTerminalPunctuation.
|
inline |
Definition at line 1072 of file unicode.h.
References IsDcpFlag(), and ucfDcpUppercase.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), IsCased(), and TUStr::IsUpperCase().
|
inlinestatic |
Definition at line 1164 of file unicode.h.
Referenced by TUniChDb::TSubcatHelper::ProcessComment().
|
inline |
Definition at line 1103 of file unicode.h.
References IsProperty(), and ucfPrVariationSelector.
|
inline |
Definition at line 1115 of file unicode.h.
References Assert, and ucfWbMask.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags(), and IsWbFormat().
|
inline |
Definition at line 1119 of file unicode.h.
References IsWbFlag(), and ucfWbFormat.
Referenced by TUniChDb::IsWbIgnored().
|
inline |
Definition at line 1104 of file unicode.h.
References IsProperty(), and ucfPrWhiteSpace.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1081 of file unicode.h.
References IsDcpFlag(), and ucfDcpXidContinue.
|
inline |
Definition at line 1080 of file unicode.h.
References IsDcpFlag(), and ucfDcpXidStart.
|
inline |
Definition at line 1056 of file unicode.h.
References InitAfterLoad(), TSIn::Load(), LoadSChar(), and LoadUShort().
Referenced by TUniChInfo().
|
inlinestatic |
Definition at line 1046 of file unicode.h.
References TSIn::LoadBf().
Referenced by Load().
Definition at line 1045 of file unicode.h.
References TSIn::LoadBf().
Referenced by Load().
|
inline |
Definition at line 1051 of file unicode.h.
References TSOut::Save(), SaveSChar(), and SaveUShort().
|
inlinestatic |
Definition at line 1048 of file unicode.h.
References TSOut::SaveBf().
Referenced by Save().
Definition at line 1047 of file unicode.h.
References TSOut::SaveBf().
Referenced by Save().
|
inline |
|
inline |
Definition at line 1070 of file unicode.h.
References Assert, and ucfDcpMask.
Referenced by TUniChDb::InitDerivedCoreProperties().
|
inline |
Definition at line 1085 of file unicode.h.
Referenced by TUniChDb::InitPropList().
|
inline |
Definition at line 1108 of file unicode.h.
Referenced by TUniChDb::InitPropList().
|
inline |
Definition at line 1127 of file unicode.h.
References Assert, and ucfSbMask.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
inline |
Definition at line 1117 of file unicode.h.
References Assert, and ucfWbMask.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
TUniChCategory TUniChInfo::cat |
char TUniChInfo::chCat |
Definition at line 1017 of file unicode.h.
Referenced by TUniChDb::LoadTxt().
char TUniChInfo::chSubCat |
Definition at line 1017 of file unicode.h.
Referenced by TUniChDb::LoadTxt().
uchar TUniChInfo::combClass |
Definition at line 1018 of file unicode.h.
Referenced by TUniChDb::LoadTxt().
int TUniChInfo::decompOffset |
Definition at line 1023 of file unicode.h.
Referenced by TUniChDb::AddDecomposition(), TUniChDb::LoadTxt(), and TUniChDb::LoadTxt_ProcessDecomposition().
int TUniChInfo::flags |
Definition at line 1025 of file unicode.h.
Referenced by TUniChDb::LoadTxt(), TUniChDb::LoadTxt_ProcessDecomposition(), and TUniChDb::TestWbFindNonIgnored().
ushort TUniChInfo::lineBreak |
Definition at line 1028 of file unicode.h.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
static |
Definition at line 1032 of file unicode.h.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
static |
Definition at line 1032 of file unicode.h.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
static |
Definition at line 1032 of file unicode.h.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
static |
Definition at line 1032 of file unicode.h.
Referenced by TUniChDb::InitWordAndSentenceBoundaryFlags().
|
static |
Definition at line 1032 of file unicode.h.
Referenced by TUniChDb::InitLineBreaks().
int TUniChInfo::nameOffset |
Definition at line 1024 of file unicode.h.
Referenced by TUniChDb::LoadTxt().
int TUniChInfo::properties |
Definition at line 1026 of file unicode.h.
Referenced by TUniChDb::TestWbFindNonIgnored().
int TUniChInfo::propertiesX |
Definition at line 1027 of file unicode.h.
Referenced by TUniChDb::TestWbFindNonIgnored().
signed char TUniChInfo::script |
Definition at line 1021 of file unicode.h.
Referenced by TUniChDb::GetScript(), TUniChDb::InitScripts(), TUniChDb::InitWordAndSentenceBoundaryFlags(), TUniChDb::LoadTxt(), and TUniChDb::TestWbFindNonIgnored().
int TUniChInfo::simpleLowerCaseMapping |
Definition at line 1022 of file unicode.h.
Referenced by TUniChDb::GetCaseConverted(), TUniChDb::GetSimpleCaseConverted(), TUniChDb::LoadTxt(), and TUniChDb::ToSimpleCaseConverted().
int TUniChInfo::simpleTitleCaseMapping |
Definition at line 1022 of file unicode.h.
Referenced by TUniChDb::GetCaseConverted(), TUniChDb::GetSimpleCaseConverted(), TUniChDb::LoadTxt(), and TUniChDb::ToSimpleCaseConverted().
int TUniChInfo::simpleUpperCaseMapping |
Definition at line 1022 of file unicode.h.
Referenced by TUniChDb::GetCaseConverted(), TUniChDb::GetSimpleCaseConverted(), TUniChDb::LoadTxt(), and TUniChDb::ToSimpleCaseConverted().
TUniChSubCategory TUniChInfo::subCat |
Definition at line 1020 of file unicode.h.
Referenced by GetSubCat(), and TUniChDb::InitWordAndSentenceBoundaryFlags().