SNAP Library 2.0, User Reference  2013-05-13 16:33:57
SNAP, a general purpose, high performance system for analysis and manipulation of large networks
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines
Public Types | Public Member Functions | Static Public Member Functions | Public Attributes | Static Protected Member Functions | Protected Attributes
TUnicode Class Reference

#include <unicode.h>

List of all members.

Public Types

typedef TUniChDb::TCaseConversion TCaseConversion

Public Member Functions

 TUnicode ()
 TUnicode (const TStr &fnBinUcd)
void Init ()
int DecodeUtf8 (const TIntV &src, TIntV &dest) const
int DecodeUtf8 (const TStr &src, TIntV &dest) const
int EncodeUtf8 (const TIntV &src, TIntV &dest) const
TStr EncodeUtf8Str (const TIntV &src) const
int DecodeUtf16FromBytes (const TIntV &src, TIntV &dest, const TUtf16BomHandling bomHandling=bomAllowed, const TUniByteOrder defaultByteOrder=boMachineEndian) const
int DecodeUtf16FromWords (const TIntV &src, TIntV &dest, const TUtf16BomHandling bomHandling=bomAllowed, const TUniByteOrder defaultByteOrder=boMachineEndian) const
int EncodeUtf16ToWords (const TIntV &src, TIntV &dest, const bool insertBom, const TUniByteOrder destByteOrder=boMachineEndian) const
int EncodeUtf16ToBytes (const TIntV &src, TIntV &dest, const bool insertBom, const TUniByteOrder destByteOrder=boMachineEndian) const
void RegisterCodec (const TStr &nameList, const PCodecBase &codec)
void UnregisterCodec (const TStr &nameList)
void ClrCodecs ()
void InitCodecs ()
PCodecBase GetCodec (const TStr &name) const
void GetAllCodecs (TCodecBaseV &dest) const
bool FindNextWordBoundary (const TIntV &src, int &position) const
void FindWordBoundaries (const TIntV &src, TBoolV &dest) const
bool FindNextSentenceBoundary (const TIntV &src, int &position) const
void FindSentenceBoundaries (const TIntV &src, TBoolV &dest) const
void ClrSentenceBoundaryExceptions ()
void UseEnglishSentenceBoundaryExceptions ()
void Decompose (const TIntV &src, TIntV &dest, bool compatibility) const
void Compose (const TIntV &src, TIntV &dest) const
void DecomposeAndCompose (const TIntV &src, TIntV &dest, bool compatibility) const
int ExtractStarters (const TIntV &src, TIntV &dest) const
int ExtractStarters (TIntV &src) const
void GetLowerCase (const TIntV &src, TIntV &dest) const
void GetUpperCase (const TIntV &src, TIntV &dest) const
void GetTitleCase (const TIntV &src, TIntV &dest) const
void GetSimpleLowerCase (const TIntV &src, TIntV &dest) const
void GetSimpleUpperCase (const TIntV &src, TIntV &dest) const
void GetSimpleTitleCase (const TIntV &src, TIntV &dest) const
void ToSimpleUpperCase (TIntV &src) const
void ToSimpleLowerCase (TIntV &src) const
void ToSimpleTitleCase (TIntV &src) const
void GetCaseFolded (const TIntV &src, TIntV &dest, const bool full=true) const
void ToCaseFolded (TIntV &src) const
TStr GetUtf8CaseFolded (const TStr &s) const
DECLARE_FORWARDED_PROPERTY_METHODS ___UniFwd2 (IsPrivateUse, IsSurrogate) TUniChCategory GetCat(const int cp) const
TUniChSubCategory GetSubCat (const int cp) const
const char * GetCharName (const int cp) const
TStr GetCharNameS (const int cp) const

Static Public Member Functions

static void EncodeUtf8 (const uint &Ch, TChA &Dest)
static TStr EncodeUtf8 (const uint &Ch)

Public Attributes

TUniCodec codec
TUniChDb ucd
T8BitCodec< TEncoding_ISO8859_1 > iso8859_1
T8BitCodec< TEncoding_ISO8859_2 > iso8859_2
T8BitCodec< TEncoding_ISO8859_3 > iso8859_3
T8BitCodec< TEncoding_ISO8859_4 > iso8859_4
T8BitCodec< TEncoding_YuAscii > yuAscii
T8BitCodec< TEncoding_CP1250 > cp1250
T8BitCodec< TEncoding_CP852 > cp852
T8BitCodec< TEncoding_CP437 > cp437

Static Protected Member Functions

static TStr NormalizeCodecName (const TStr &name)

Protected Attributes

THash< TStr, PCodecBase > codecs

Detailed Description

Definition at line 1771 of file unicode.h.

Member Typedef Documentation

typedef TUniChDb::TCaseConversion TUnicode::TCaseConversion

Definition at line 1963 of file unicode.h.

Constructor & Destructor Documentation

TUnicode::TUnicode ( ) [inline]

Definition at line 1777 of file unicode.h.

{ Init(); }
TUnicode::TUnicode ( const TStr &  fnBinUcd) [inline, explicit]

Definition at line 1778 of file unicode.h.

{ ucd.LoadBin(fnBinUcd); Init(); }

Member Function Documentation

DECLARE_FORWARDED_PROPERTY_METHODS TUnicode::___UniFwd2 ( IsPrivateUse  ,
IsSurrogate   
) const [inline]

Definition at line 2018 of file unicode.h.

                                                  { return ucd.GetCat(cp); }
void TUnicode::ClrCodecs ( ) [inline]

Definition at line 1881 of file unicode.h.

{ codecs.Clr(); }
void TUnicode::ClrSentenceBoundaryExceptions ( ) [inline]

Definition at line 1924 of file unicode.h.

{ ucd.SbEx_Clr(); }
void TUnicode::Compose ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1941 of file unicode.h.

{ return ucd.Compose(src, dest, true); }
int TUnicode::DecodeUtf16FromBytes ( const TIntV &  src,
TIntV &  dest,
const TUtf16BomHandling  bomHandling = bomAllowed,
const TUniByteOrder  defaultByteOrder = boMachineEndian 
) const [inline]

Definition at line 1810 of file unicode.h.

                                                                              {
                        return (int) codec.DecodeUtf16FromBytes(src, 0, src.Len(), dest, true, bomHandling, defaultByteOrder); }
int TUnicode::DecodeUtf16FromWords ( const TIntV &  src,
TIntV &  dest,
const TUtf16BomHandling  bomHandling = bomAllowed,
const TUniByteOrder  defaultByteOrder = boMachineEndian 
) const [inline]

Definition at line 1823 of file unicode.h.

                                                                              {
                        return (int) codec.DecodeUtf16FromWords(src, 0, src.Len(), dest, true, bomHandling, defaultByteOrder); }
int TUnicode::DecodeUtf8 ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1787 of file unicode.h.

{ return (int) codec.DecodeUtf8(src, dest); }
int TUnicode::DecodeUtf8 ( const TStr &  src,
TIntV &  dest 
) const [inline]

Definition at line 1788 of file unicode.h.

{ return (int) codec.DecodeUtf8(src, dest); }
void TUnicode::Decompose ( const TIntV &  src,
TIntV &  dest,
bool  compatibility 
) const [inline]

Definition at line 1934 of file unicode.h.

{ ucd.Decompose(src, dest, compatibility, true); }
void TUnicode::DecomposeAndCompose ( const TIntV &  src,
TIntV &  dest,
bool  compatibility 
) const [inline]

Definition at line 1946 of file unicode.h.

{ return ucd.DecomposeAndCompose(src, dest, compatibility); }
int TUnicode::EncodeUtf16ToBytes ( const TIntV &  src,
TIntV &  dest,
const bool  insertBom,
const TUniByteOrder  destByteOrder = boMachineEndian 
) const [inline]

Definition at line 1838 of file unicode.h.

                                                                           {
                        return (int) codec.EncodeUtf16ToBytes(src, 0, src.Len(), dest, true, insertBom, destByteOrder); }
int TUnicode::EncodeUtf16ToWords ( const TIntV &  src,
TIntV &  dest,
const bool  insertBom,
const TUniByteOrder  destByteOrder = boMachineEndian 
) const [inline]

Definition at line 1834 of file unicode.h.

                                                                           {
                        return (int) codec.EncodeUtf16ToWords(src, 0, src.Len(), dest, true, insertBom, destByteOrder); }
int TUnicode::EncodeUtf8 ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1792 of file unicode.h.

{ return (int) codec.EncodeUtf8(src, dest); }
void TUnicode::EncodeUtf8 ( const uint &  Ch,
TChA &  Dest 
) [static]

Definition at line 1700 of file unicode.cpp.

                                                   {
        if (c > 0x10ffff) {
                throw TExcept::New(TStr::Fmt("Unkown Unicode character %u", c)); }
        if (c < 0x80u)
                dest.AddCh(char(c & 0xffu));
        else if (c < 0x800u) {
                dest.AddCh(char(TUniCodec::_1100_0000 | ((c >> 6) & TUniCodec::_0001_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | (c & TUniCodec::_0011_1111))); }
        else if (c < 0x10000u) {
                dest.AddCh(char(TUniCodec::_1110_0000 | ((c >> 12) & TUniCodec::_0000_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 6) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | (c & TUniCodec::_0011_1111))); }
        else if (c < 0x200000u) {
                dest.AddCh(char(TUniCodec::_1111_0000 | ((c >> 18) & TUniCodec::_0000_0111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 12) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 6) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | (c & TUniCodec::_0011_1111))); }
        else if (c < 0x4000000u) {
                dest.AddCh(char(TUniCodec::_1111_1000 | ((c >> 24) & TUniCodec::_0000_0011)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 18) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 12) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 6) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | (c & TUniCodec::_0011_1111))); }
        else {
                dest.AddCh(char(TUniCodec::_1111_1100 | ((c >> 30) & TUniCodec::_0000_0011)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 24) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 18) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 12) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | ((c >> 6) & TUniCodec::_0011_1111)));
                dest.AddCh(char(TUniCodec::_1000_0000 | (c & TUniCodec::_0011_1111))); }
}
TStr TUnicode::EncodeUtf8 ( const uint &  Ch) [static]

Definition at line 1732 of file unicode.cpp.

                                        {
        TChA ChA; EncodeUtf8(Ch, ChA); return ChA;
}
TStr TUnicode::EncodeUtf8Str ( const TIntV &  src) const [inline]

Definition at line 1796 of file unicode.h.

{ return codec.EncodeUtf8Str(src); }
int TUnicode::ExtractStarters ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1951 of file unicode.h.

{ return (int) ucd.ExtractStarters(src, dest); }
int TUnicode::ExtractStarters ( TIntV &  src) const [inline]

Definition at line 1953 of file unicode.h.

{ return (int) ucd.ExtractStarters(src); }
bool TUnicode::FindNextSentenceBoundary ( const TIntV &  src,
int &  position 
) const [inline]

Definition at line 1916 of file unicode.h.

                                                                             {
                if (position < 0) { position = 0; return true; }
                size_t position_; bool retVal = ucd.FindNextSentenceBoundary(src, 0, src.Len(), position_); position = int(position_); return retVal; }
bool TUnicode::FindNextWordBoundary ( const TIntV &  src,
int &  position 
) const [inline]

Definition at line 1901 of file unicode.h.

                                                                         {
                if (position < 0) { position = 0; return true; }
                size_t position_; bool retVal = ucd.FindNextWordBoundary(src, 0, src.Len(), position_); position = int(position_); return retVal; }
void TUnicode::FindSentenceBoundaries ( const TIntV &  src,
TBoolV &  dest 
) const [inline]

Definition at line 1922 of file unicode.h.

{ ucd.FindSentenceBoundaries(src, 0, src.Len(), dest); }
void TUnicode::FindWordBoundaries ( const TIntV &  src,
TBoolV &  dest 
) const [inline]

Definition at line 1907 of file unicode.h.

{ ucd.FindWordBoundaries(src, 0, src.Len(), dest); }
void TUnicode::GetAllCodecs ( TCodecBaseV &  dest) const [inline]

Definition at line 1887 of file unicode.h.

                                                   {
                dest.Clr();
                for (int i = codecs.FFirstKeyId(); codecs.FNextKeyId(i); ) {
                        PCodecBase codec = codecs[i]; bool found = false;
                        for (int j = 0; j < dest.Len(); j++) if (dest[j]() == codec()) { found = true; break; }
                        if (! found) dest.Add(codec); }}
void TUnicode::GetCaseFolded ( const TIntV &  src,
TIntV &  dest,
const bool  full = true 
) const [inline]

Definition at line 1989 of file unicode.h.

{ return ucd.GetCaseFolded(src, dest, true, full, false); }
const char* TUnicode::GetCharName ( const int  cp) const [inline]

Definition at line 2024 of file unicode.h.

{ return ucd.GetCharName(cp); }
TStr TUnicode::GetCharNameS ( const int  cp) const [inline]

Definition at line 2025 of file unicode.h.

{ return ucd.GetCharNameS(cp); }
PCodecBase TUnicode::GetCodec ( const TStr &  name) const [inline]

Definition at line 1883 of file unicode.h.

                                                    {
                TStr s = NormalizeCodecName(name);
                PCodecBase p; if (! codecs.IsKeyGetDat(s, p)) p.Clr();
                return p; }
void TUnicode::GetLowerCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1965 of file unicode.h.

{ ucd.GetLowerCase(src, dest, true, false, false); }
void TUnicode::GetSimpleLowerCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1972 of file unicode.h.

{ ucd.GetSimpleLowerCase(src, dest, true); }
void TUnicode::GetSimpleTitleCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1974 of file unicode.h.

{ ucd.GetSimpleTitleCase(src, dest, true); }
void TUnicode::GetSimpleUpperCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1973 of file unicode.h.

{ ucd.GetSimpleUpperCase(src, dest, true); }
TUniChSubCategory TUnicode::GetSubCat ( const int  cp) const [inline]

Definition at line 2021 of file unicode.h.

{ return ucd.GetSubCat(cp); }
void TUnicode::GetTitleCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1967 of file unicode.h.

{ ucd.GetTitleCase(src, dest, true, false, false); }
void TUnicode::GetUpperCase ( const TIntV &  src,
TIntV &  dest 
) const [inline]

Definition at line 1966 of file unicode.h.

{ ucd.GetUpperCase(src, dest, true, false, false); }
TStr TUnicode::GetUtf8CaseFolded ( const TStr &  s) const [inline]

Definition at line 1994 of file unicode.h.

                                                    {
                bool isAscii = true;
                for (int i = 0, n = s.Len(); i < n; i++) if (uchar(s[i]) >= 128) { isAscii = false; break; }
                if (isAscii) return s.GetLc();
                TIntV src; DecodeUtf8(s, src);
                TIntV dest; GetCaseFolded(src, dest);
                return EncodeUtf8Str(dest); }
void TUnicode::Init ( ) [inline]

Definition at line 1779 of file unicode.h.

{ InitCodecs(); }
void TUnicode::InitCodecs ( )

Definition at line 1687 of file unicode.cpp.

{
        ClrCodecs();
        RegisterCodec("ISO-8859-1 ISO_8859-1 ISO_8859-1:1987 ISO-IR-100 CP819 IBM819 LATIN1 L1 csISOLatin1 ISO8859-1 ISO8859_1 CP28591", TCodecBase::New<TCodec_ISO8859_1>());
        RegisterCodec("ISO-8859-2 ISO_8859-2 ISO_8859-2:1987 ISO-IR-101 LATIN2 L2 csISOLatin2 ISO8859-2 ISO8859_2 CP28592", TCodecBase::New<TCodec_ISO8859_2>());
        RegisterCodec("ISO-8859-3 ISO_8859-3 ISO_8859-3:1988 ISO-IR-109 LATIN3 L3 csISOLatin3 ISO8859-3 ISO8859_3 CP28593", TCodecBase::New<TCodec_ISO8859_3>());
        RegisterCodec("ISO-8859-4 ISO_8859-4 ISO_8859-4:1988 ISO-IR-110 LATIN4 L4 csISOLatin4 ISO8859-4 ISO8859_4 CP28594", TCodecBase::New<TCodec_ISO8859_4>());
        RegisterCodec("YUASCII YU-ASCII YU_ASCII", TCodecBase::New<TCodec_YuAscii>());
        RegisterCodec("CP1250 Windows-1250 MS-EE", TCodecBase::New<TCodec_CP1250>());
        RegisterCodec("CP852 cp852_DOSLatin2 DOSLatin2", TCodecBase::New<TCodec_CP852>());
        RegisterCodec("CP437 cp437_DOSLatinUS DOSLatinUS", TCodecBase::New<TCodec_CP437>());
}
static TStr TUnicode::NormalizeCodecName ( const TStr &  name) [inline, static, protected]

Definition at line 1870 of file unicode.h.

                                                                {
                TStr s = name.GetLc(); s.ChangeStrAll("_", ""); s.ChangeStrAll("-", ""); return s; }
void TUnicode::RegisterCodec ( const TStr &  nameList,
const PCodecBase &  codec 
) [inline]

Definition at line 1873 of file unicode.h.

                                                                          {
                TStrV names; nameList.SplitOnWs(names);
                for (int i = 0; i < names.Len(); i++)
                        codecs.AddDat(NormalizeCodecName(names[i]), codec); }
void TUnicode::ToCaseFolded ( TIntV &  src) const [inline]

Definition at line 1992 of file unicode.h.

{ return ucd.ToCaseFolded(src, false); }
void TUnicode::ToSimpleLowerCase ( TIntV &  src) const [inline]

Definition at line 1978 of file unicode.h.

{ ucd.ToSimpleLowerCase(src); }
void TUnicode::ToSimpleTitleCase ( TIntV &  src) const [inline]

Definition at line 1979 of file unicode.h.

{ ucd.ToSimpleTitleCase(src); }
void TUnicode::ToSimpleUpperCase ( TIntV &  src) const [inline]

Definition at line 1977 of file unicode.h.

{ ucd.ToSimpleUpperCase(src); }
void TUnicode::UnregisterCodec ( const TStr &  nameList) [inline]

Definition at line 1877 of file unicode.h.

                                                   {
                TStrV names; nameList.SplitOnWs(names);
                for (int i = 0; i < names.Len(); i++)
                        codecs.DelKey(NormalizeCodecName(names[i])); }
void TUnicode::UseEnglishSentenceBoundaryExceptions ( ) [inline]

Definition at line 1925 of file unicode.h.

{ ucd.SbEx_SetStdEnglish(); }

Member Data Documentation

TUniCodec TUnicode::codec

Definition at line 1774 of file unicode.h.

THash<TStr, PCodecBase> TUnicode::codecs [protected]

Definition at line 1869 of file unicode.h.

T8BitCodec<TEncoding_CP1250> TUnicode::cp1250

Definition at line 1851 of file unicode.h.

T8BitCodec<TEncoding_CP437> TUnicode::cp437

Definition at line 1853 of file unicode.h.

T8BitCodec<TEncoding_CP852> TUnicode::cp852

Definition at line 1852 of file unicode.h.

T8BitCodec<TEncoding_ISO8859_1> TUnicode::iso8859_1

Definition at line 1846 of file unicode.h.

T8BitCodec<TEncoding_ISO8859_2> TUnicode::iso8859_2

Definition at line 1847 of file unicode.h.

T8BitCodec<TEncoding_ISO8859_3> TUnicode::iso8859_3

Definition at line 1848 of file unicode.h.

T8BitCodec<TEncoding_ISO8859_4> TUnicode::iso8859_4

Definition at line 1849 of file unicode.h.

TUniChDb TUnicode::ucd

Definition at line 1775 of file unicode.h.

T8BitCodec<TEncoding_YuAscii> TUnicode::yuAscii

Definition at line 1850 of file unicode.h.

The documentation for this class was generated from the following files: