#include <unicode.h>

Collaboration diagram for TUniCaseFolding:

Public Member Functions
	TUniCaseFolding ()
	TUniCaseFolding (TSIn &SIn)
void	Load (TSIn &SIn)
void	Save (TSOut &SOut) const
void	Clr ()
void	LoadTxt (const TStr &fileName)
template<typename TSrcVec , typename TDestCh >
void	Fold (const TSrcVec &src, size_t srcIdx, const size_t srcCount, TVec< TDestCh > &dest, const bool clrDest, const bool full, const bool turkic) const
template<typename TSrcVec >
void	FoldInPlace (TSrcVec &src, size_t srcIdx, const size_t srcCount, const bool turkic) const
void	Test ()
Protected Types
typedef TUniVecIdx	TVecIdx
Protected Member Functions
void	Test (const TIntV &src, const TIntV &expectedDest, const bool full, const bool turkic, FILE *f)
Static Protected Member Functions
template<typename TSrcDat , typename TDestDat >
static void	AppendVector (const TVec< TSrcDat > &src, TVec< TDestDat > &dest)
Protected Attributes
TIntH	cfCommon
TIntH	cfSimple
TIntH	cfTurkic
TIntIntVH	cfFull
Friends
class	TUniChDb

Detailed Description

Definition at line 270 of file unicode.h.

Member Typedef Documentation

typedef TUniVecIdx TUniCaseFolding::TVecIdx [protected]

Definition at line 280 of file unicode.h.

Constructor & Destructor Documentation

TUniCaseFolding::TUniCaseFolding ( ) [inline]

Definition at line 283 of file unicode.h.

{ }

TUniCaseFolding::TUniCaseFolding ( TSIn & SIn ) [inline, explicit]

Definition at line 284 of file unicode.h.

References TSIn::LoadCs().

: cfCommon(SIn), cfSimple(SIn), cfTurkic(SIn), cfFull(SIn) { SIn.LoadCs(); }

Here is the call graph for this function:

Member Function Documentation

template<typename TSrcDat , typename TDestDat >

static void TUniCaseFolding::AppendVector	(	const TVec< TSrcDat > &	src,
		TVec< TDestDat > &	dest
	)		`[inline, static, protected]`

Definition at line 277 of file unicode.h.

References TVec< TVal >::Add(), and TVec< TVal >::Len().

Referenced by Fold(), and TUniChDb::GetCaseConverted().

                                                                                        {
                for (int i = 0; i < src.Len(); i++) dest.Add(src[i]); }

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::Clr ( ) [inline]

Definition at line 287 of file unicode.h.

References cfCommon, cfFull, cfSimple, cfTurkic, and THash< TKey, TDat, THashFunc >::Clr().

Referenced by TUniChDb::Clr(), and LoadTxt().

{ cfCommon.Clr(); cfSimple.Clr(); cfFull.Clr(); cfTurkic.Clr(); }

Here is the call graph for this function:

Here is the caller graph for this function:

template<typename TSrcVec , typename TDestCh >

void TUniCaseFolding::Fold	(	const TSrcVec &	src,
		size_t	srcIdx,
		const size_t	srcCount,
		TVec< TDestCh > &	dest,
		const bool	clrDest,
		const bool	full,
		const bool	turkic
	)		const `[inline]`

Definition at line 292 of file unicode.h.

References TVec< TVal >::Add(), AppendVector(), cfCommon, cfFull, cfSimple, cfTurkic, and THash< TKey, TDat, THashFunc >::GetKeyId().

Referenced by TUniChDb::GetCaseFolded(), and Test().

        {
                for (const size_t srcEnd = srcIdx + srcCount; srcIdx < srcEnd; )
                {
                        int c = src[TVecIdx(srcIdx)], i; srcIdx++;
                        if (turkic && ((i = cfTurkic.GetKeyId(c)) >= 0)) { dest.Add(cfTurkic[i]); continue; }
                        if (full && ((i = cfFull.GetKeyId(c)) >= 0)) { AppendVector(cfFull[i], dest); continue; }
                        if ((! full) && ((i = cfSimple.GetKeyId(c)) >= 0)) { dest.Add(cfSimple[i]); continue; }
                        i = cfCommon.GetKeyId(c); if (i >= 0) dest.Add(cfCommon[i]); else dest.Add(c);
                }
        }

Here is the call graph for this function:

Here is the caller graph for this function:

template<typename TSrcVec >

void TUniCaseFolding::FoldInPlace	(	TSrcVec &	src,
		size_t	srcIdx,
		const size_t	srcCount,
		const bool	turkic
	)		const `[inline]`

Definition at line 306 of file unicode.h.

References cfCommon, cfSimple, cfTurkic, and THash< TKey, TDat, THashFunc >::GetKeyId().

Referenced by TUniChDb::ToCaseFolded().

        {
                for (const size_t srcEnd = srcIdx + srcCount; srcIdx < srcEnd; srcIdx++)
                {
                        int c = src[TVecIdx(srcIdx)], i;
                        if (turkic && ((i = cfTurkic.GetKeyId(c)) >= 0)) { src[TVecIdx(srcIdx)] = cfTurkic[i]; continue; }
                        if ((i = cfSimple.GetKeyId(c)) >= 0) { src[TVecIdx(srcIdx)] = cfSimple[i]; continue; }
                        i = cfCommon.GetKeyId(c); if (i >= 0) src[TVecIdx(srcIdx)] = cfCommon[i];
                }
        }

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::Load ( TSIn & SIn ) [inline]

Definition at line 285 of file unicode.h.

References cfCommon, cfFull, cfSimple, cfTurkic, THash< TKey, TDat, THashFunc >::Load(), and TSIn::LoadCs().

Referenced by TUniChDb::Load().

{ cfCommon.Load(SIn); cfSimple.Load(SIn); cfFull.Load(SIn); cfTurkic.Load(SIn); SIn.LoadCs(); }

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::LoadTxt ( const TStr & fileName )

Definition at line 509 of file unicode.cpp.

References THash< TKey, TDat, THashFunc >::AddDat(), cfCommon, cfFull, cfSimple, cfTurkic, Clr(), TStr::CStr(), FailR, TUniChDb::TUcdFileReader::GetNextLine(), IAssert, THash< TKey, TDat, THashFunc >::IsKey(), THash< TKey, TDat, THashFunc >::Len(), TUniChDb::TUcdFileReader::Open(), TUniChDb::TUcdFileReader::ParseCodePoint(), and TUniChDb::TUcdFileReader::ParseCodePointList().

Referenced by TUniChDb::LoadTxt().

{
        Clr();
        TUniChDb::TUcdFileReader reader; reader.Open(fileName);
        TStrV fields;
        while (reader.GetNextLine(fields))
        {
                int cp = reader.ParseCodePoint(fields[0]);
                const TStr status = fields[1], mapsTo = fields[2];
                if (status == "C" || status == "S" || status == "T") {
                        TIntH &dest = (status == "C" ? cfCommon : status == "S" ? cfSimple : cfTurkic);
                        IAssert(! dest.IsKey(cp));
                        int cp2 = reader.ParseCodePoint(mapsTo);
                        dest.AddDat(cp, cp2); }
                else if (status == "F") {
                        TIntIntVH &dest = cfFull;
                        IAssert(! dest.IsKey(cp));
                        TIntV cps; reader.ParseCodePointList(mapsTo, cps); IAssert(cps.Len() > 0);
                        dest.AddDat(cp, cps); }
                else
                        FailR(status.CStr());
        }
        printf("TUniCaseFolding(\"%s\"): %d common, %d simple, %d full, %d Turkic.\n",
                fileName.CStr(), cfCommon.Len(), cfSimple.Len(), cfFull.Len(), cfTurkic.Len());
}

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::Save ( TSOut & SOut ) const [inline]

Definition at line 286 of file unicode.h.

References cfCommon, cfFull, cfSimple, cfTurkic, THash< TKey, TDat, THashFunc >::Save(), and TSOut::SaveCs().

Referenced by TUniChDb::Save().

{ cfCommon.Save(SOut); cfSimple.Save(SOut); cfFull.Save(SOut); cfTurkic.Save(SOut); SOut.SaveCs(); }

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::Test	(	const TIntV &	src,
		const TIntV &	expectedDest,
		const bool	full,
		const bool	turkic,
		FILE *	f
	)		`[protected]`

Definition at line 535 of file unicode.cpp.

References Fold(), IAssert, and TVec< TVal >::Len().

Referenced by TUniChDb::Test().

{
        fprintf(f, "TUniCaseFolding(%s%s): ", (full ? "full" : "simple"), (turkic ? ", turkic" : ""));
        for (int i = 0; i < src.Len(); i++) fprintf(f, " %04x", int(src[i]));
        TIntV dest; Fold(src, 0, src.Len(), dest, true, full, turkic);
        fprintf(f, "\n  -> ");
        for (int i = 0; i < dest.Len(); i++) fprintf(f, " %04x", int(dest[i]));
        fprintf(f, "\n");
        IAssert(dest.Len() == expectedDest.Len());
        for (int i = 0; i < dest.Len(); i++) IAssert(dest[i] == expectedDest[i]);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void TUniCaseFolding::Test ( )

Definition at line 553 of file unicode.cpp.

References anonymous_namespace{unicode.cpp}::VB.

{
        FILE *f = stderr;
        TVectorBuilder VB;
        // simple
        Test((VB, 0x41, 0x62, 0x49, 0x43, 0xdf), (VB, 0x61, 0x62, 0x69, 0x63, 0xdf), false, false, f);
        // simple + turkic
        Test((VB, 0x41, 0x62, 0x49, 0x43, 0xdf), (VB, 0x61, 0x62, 0x131, 0x63, 0xdf), false, true, f);
        // full
        Test((VB, 0x41, 0x62, 0x49, 0x43, 0xdf), (VB, 0x61, 0x62, 0x69, 0x63, 0x73, 0x73), true, false, f);
        // full + turkic
        Test((VB, 0x41, 0x62, 0x49, 0x43, 0xdf), (VB, 0x61, 0x62, 0x131, 0x63, 0x73, 0x73), true, true, f);
}