THtmlLxChDef Class Reference

#include <html.h>

Public Member Functions
	THtmlLxChDef ()
	THtmlLxChDef (TSIn &SIn)
void	Save (TSOut &SOut)
THtmlLxChDef &	operator= (const THtmlLxChDef &)
int	GetChTy (const char &Ch) const
bool	IsEoln (const char &Ch) const
bool	IsWs (const char &Ch) const
bool	IsSpace (const char &Ch) const
bool	IsAlpha (const char &Ch) const
bool	IsNum (const char &Ch) const
bool	IsAlNum (const char &Ch) const
bool	IsSym (const char &Ch) const
bool	IsUrl (const char &Ch) const
bool	IsUc (const char &Ch) const
bool	IsLc (const char &Ch) const
char	GetUc (const char &Ch) const
char	GetLc (const char &Ch) const
void	GetUcChA (TChA &ChA) const
void	GetLcChA (TChA &ChA) const
TStr	GetUcStr (const TStr &Str) const
TStr	GetLcStr (const TStr &Str) const
TStr	GetEscStr (const TStr &Str) const

Static Public Member Functions
static PHtmlLxChDef	Load (TSIn &SIn)
static PHtmlLxChDef	GetChDef ()
static THtmlLxChDef &	GetChDefRef ()
static TStr	GetCSZFromYuascii (const TChA &ChA)
static TStr	GetCSZFromWin1250 (const TChA &ChA)
static TStr	GetWin1250FromYuascii (const TChA &ChA)
static TStr	GetIsoCeFromYuascii (const TChA &ChA)

Static Public Attributes
static PHtmlLxChDef	ChDef =PHtmlLxChDef(new THtmlLxChDef())

Private Member Functions
void	SetUcCh (const char &UcCh, const char &LcCh)
void	SetUcCh (const TStr &Str)
void	SetChTy (const THtmlLxChTy &ChTy, const TStr &Str)
void	SetEscStr (const TStr &SrcStr, const TStr &DstStr)

Private Attributes
TCRef	CRef
TIntV	ChTyV
TChV	UcChV
TChV	LcChV
TStrStrH	EscStrH

Friends
class	TPt< THtmlLxChDef >

Detailed Description

Definition at line 14 of file html.h.

Constructor & Destructor Documentation

THtmlLxChDef::THtmlLxChDef

(

)

Definition at line 48 of file html.cpp.

                          :
  ChTyV(TCh::Vals), UcChV(TCh::Vals), LcChV(TCh::Vals), EscStrH(100){

  // Character-Types
  ChTyV.PutAll(TInt(hlctSpace));
  SetChTy(hlctAlpha, "ABCDEFGHIJKLMNOPQRSTUVWXYZ");
  SetChTy(hlctAlpha, "abcdefghijklmnopqrstuvwxyz");
  SetChTy(hlctAlpha, "@_");
  SetChTy(hlctNum, "0123456789");
  SetChTy(hlctSym, "`~!#$%^&*()-=+[{]}\\|;:'\",<.>/?");
  SetChTy(hlctLTag, "<"); SetChTy(hlctRTag, ">");
  SetChTy(hlctEof, TStr(TCh::EofCh));
  for (int Ch=TCh::Mn; Ch<=TCh::Mx; Ch++){
    if ((Ch<0)||(127<Ch)){SetChTy(hlctAlpha, TStr(TCh(char(Ch))));}}
  //SetChTy(hlctSpace, TStr(TCh(char(160))));

  // Upper-Case
  {for (int Ch=TCh::Mn; Ch<=TCh::Mx; Ch++){
    SetUcCh(char(Ch), char(Ch));}}
  SetUcCh("Aa"); SetUcCh("\xc0\xe0"); SetUcCh("\xc1\xe1"); SetUcCh("\xc2\xe2");
  SetUcCh("\xc3\xe3"); SetUcCh("\xc4\xe4"); SetUcCh("\xc5\xe5"); SetUcCh("\xc6\xe6");
  SetUcCh("Bb"); SetUcCh("Cc"); SetUcCh("\xc7\xe7"); SetUcCh("Dd");
  SetUcCh("\xd0\xf0"); SetUcCh("Ee"); SetUcCh("\xc8\xe8"); SetUcCh("\xc9\xe9");
  SetUcCh("\xca\xea"); SetUcCh("\xcb\xeb"); SetUcCh("Ff"); SetUcCh("Gg");
  SetUcCh("Hh"); SetUcCh("Ii"); SetUcCh("\xcc\xec"); SetUcCh("\xcd\xed");
  SetUcCh("\xce\xee"); SetUcCh("\xcf\xef"); SetUcCh("Jj"); SetUcCh("Kk");
  SetUcCh("Ll"); SetUcCh("Mm"); SetUcCh("Nn"); SetUcCh("\xd1\xf1");
  SetUcCh("Oo"); SetUcCh("\xd2\xf2"); SetUcCh("\xd3\xf3"); SetUcCh("\xd4\xf4");
  SetUcCh("\xd5\xf5"); SetUcCh("\xd6\xf6"); SetUcCh("\xd8\xf8"); SetUcCh("Pp");
  SetUcCh("Qq"); SetUcCh("Rr"); SetUcCh("Ss"); SetUcCh("\x8a\x9a");
  SetUcCh("Tt"); SetUcCh("Uu"); SetUcCh("\xd9\xf9"); SetUcCh("\xda\xfa");
  SetUcCh("\xdb\xfb"); SetUcCh("\xdc\xfc"); SetUcCh("Vv"); SetUcCh("Ww");
  SetUcCh("Xx"); SetUcCh("Yy\xff"); SetUcCh("\xdd\xfd"); SetUcCh("Zz");
  SetUcCh("\x8e\x9e");
  // ISO-CE
  //SetUcCh(uchar(169), uchar(185)); /*Sh - \xa9\xb9*/
  //SetUcCh(uchar(174), uchar(190)); /*Zh - \xae\xbe*/
  //SetUcCh(uchar(200), uchar(232)); /*Ch - \xc8\xe8*/
  //SetUcCh(uchar(198), uchar(230)); /*Cs - \xc6\xe6*/
  //SetUcCh(uchar(208), uchar(240)); /*Dz - \xd0\xf0*/

  // Annoying Unicode-characters
  //SetChTy(hlctSpace, "\xc2\xef");

  // Escape-Sequences
  SetEscStr("&quot", "\""); SetEscStr("&amp", "&");
  SetEscStr("&lt", "<"); SetEscStr("&gt", ">");
  SetEscStr("&nbsp", " ");

  SetEscStr("&auml", "\xe4"); SetEscStr("&Auml", "\xc4");
  SetEscStr("&ouml", "\xf6"); SetEscStr("&Ouml", "\xd6");
  SetEscStr("&uuml", "\xfc"); SetEscStr("&Uuml", "\xdc");
  SetEscStr("&aring", "\xe5"); SetEscStr("&Aring", "\xc5");
  SetEscStr("&oslash", "\xf8"); SetEscStr("&Oslash", "\xd8");
  SetEscStr("&Aelig", "\xc6"); SetEscStr("&aelig", "\xe6");

  SetEscStr("&eacute", "e"); SetEscStr("&Eacute", "E");
  SetEscStr("&egrave", "e"); SetEscStr("&Egrave", "E");
  SetEscStr("&agrave", "a"); SetEscStr("&Agrave", "A");
}

THtmlLxChDef::THtmlLxChDef ( TSIn &  SIn )

inline

Definition at line 26 of file html.h.

: ChTyV(SIn), UcChV(SIn), LcChV(SIn), EscStrH(SIn){}

Member Function Documentation

static PHtmlLxChDef THtmlLxChDef::GetChDef ( )

inlinestatic

Definition at line 66 of file html.h.

{IAssert(!ChDef.Empty()); return ChDef;}

static THtmlLxChDef& THtmlLxChDef::GetChDefRef ( )

inlinestatic

Definition at line 67 of file html.h.

{IAssert(!ChDef.Empty()); return *ChDef;}

int THtmlLxChDef::GetChTy ( const char &  Ch ) const

inline

Definition at line 34 of file html.h.

{return ChTyV[Ch-TCh::Mn];}

TStr THtmlLxChDef::GetCSZFromWin1250 ( const TChA &  ChA )

static

Definition at line 132 of file html.cpp.

                                                   {
  TChA DstChA;
  for (int ChN=0; ChN<ChA.Len(); ChN++){
    const uchar Ch=ChA[ChN];
    switch (Ch){
      case 232: DstChA+='c'; break;
      case 200: DstChA+='C'; break;
      case 154: DstChA+='s'; break;
      case 138: DstChA+='S'; break;
      case 158: DstChA+='z'; break;
      case 142: DstChA+='Z'; break;
      default: DstChA+=Ch;
    }
  }
  return DstChA;
}

TStr THtmlLxChDef::GetCSZFromYuascii ( const TChA &  ChA )

static

Definition at line 111 of file html.cpp.

                                                   {
  TChA DstChA;
  for (int ChN=0; ChN<ChA.Len(); ChN++){
    char Ch=ChA[ChN];
    switch (Ch){
      case '~': DstChA+='c'; break;
      case '^': DstChA+='C'; break;
      case '}': DstChA+='c'; break;
      case ']': DstChA+='C'; break;
      case '|': DstChA+='d'; break;
      case '\\': DstChA+='D'; break;
      case '{': DstChA+='s'; break;
      case '[': DstChA+='S'; break;
      case '`': DstChA+='z'; break;
      case '@': DstChA+='Z'; break;
      default: DstChA+=Ch;
    }
  }
  return DstChA;
}

TStr THtmlLxChDef::GetEscStr

(

const TStr &

Str

)

const

Definition at line 33 of file html.cpp.

                                                  {
  int EscStrId;
  if ((EscStrId=EscStrH.GetKeyId(Str))!=-1){
    return EscStrH[EscStrId];
  } else
  if ((Str.Len()>=2)&&(Str[0]=='&')&&(Str[1]=='#')){
    int ChCd=0;
    for (int ChN=2; ChN<Str.Len(); ChN++){
      if (ChCd<=0xFFFF){ChCd=ChCd*10+Str[ChN]-'0';}}
    return TStr((char)ChCd);
  } else {
    return TStr(' ');
  }
}

TStr THtmlLxChDef::GetIsoCeFromYuascii ( const TChA &  ChA )

static

Definition at line 170 of file html.cpp.

                                                     {
  TChA DstChA;
  for (int ChN=0; ChN<ChA.Len(); ChN++){
    char Ch=ChA[ChN];
    switch (Ch){
      case '~': DstChA+=uchar(232); break;
      case '^': DstChA+=uchar(200); break;
      case '}': DstChA+=uchar(230); break;
      case ']': DstChA+=uchar(198); break;
      case '|': DstChA+=uchar(240); break;
      case '\\': DstChA+=uchar(208); break;
      case '{': DstChA+=uchar(185); break;
      case '[': DstChA+=uchar(169); break;
      case '`': DstChA+=uchar(190); break;
      case '@': DstChA+=uchar(174); break;
      default: DstChA+=Ch;
    }
  }
  return DstChA;
}

char THtmlLxChDef::GetLc ( const char &  Ch ) const

inline

Definition at line 53 of file html.h.

{return LcChV[Ch-TCh::Mn];}

void THtmlLxChDef::GetLcChA ( TChA &  ChA ) const

inline

Definition at line 56 of file html.h.

                                 {
    for (int ChN=0; ChN<ChA.Len(); ChN++){ChA.PutCh(ChN, GetLc(ChA[ChN]));}}

TStr THtmlLxChDef::GetLcStr ( const TStr &  Str ) const

inline

Definition at line 60 of file html.h.

                                       {
    TChA ChA(Str); GetLcChA(ChA); return ChA;}

char THtmlLxChDef::GetUc ( const char &  Ch ) const

inline

Definition at line 52 of file html.h.

{return UcChV[Ch-TCh::Mn];}

void THtmlLxChDef::GetUcChA ( TChA &  ChA ) const

inline

Definition at line 54 of file html.h.

                                 {
    for (int ChN=0; ChN<ChA.Len(); ChN++){ChA.PutCh(ChN, GetUc(ChA[ChN]));}}

TStr THtmlLxChDef::GetUcStr ( const TStr &  Str ) const

inline

Definition at line 58 of file html.h.

                                       {
    TChA ChA(Str); GetUcChA(ChA); return ChA;}

TStr THtmlLxChDef::GetWin1250FromYuascii ( const TChA &  ChA )

static

Definition at line 149 of file html.cpp.

                                                       {
  TChA DstChA;
  for (int ChN=0; ChN<ChA.Len(); ChN++){
    char Ch=ChA[ChN];
    switch (Ch){
      case '~': DstChA+=uchar(232); break;
      case '^': DstChA+=uchar(200); break;
      case '}': DstChA+='c'; break;
      case ']': DstChA+='C'; break;
      case '|': DstChA+='d'; break;
      case '\\': DstChA+='D'; break;
      case '{': DstChA+=uchar(154); break;
      case '[': DstChA+=uchar(138); break;
      case '`': DstChA+=uchar(158); break;
      case '@': DstChA+=uchar(142); break;
      default: DstChA+=Ch;
    }
  }
  return DstChA;
}

bool THtmlLxChDef::IsAlNum ( const char &  Ch ) const

inline

Definition at line 41 of file html.h.

                                     {
    return (int(ChTyV[Ch-TCh::Mn])==hlctAlpha)||(int(ChTyV[Ch-TCh::Mn])==hlctNum);}

bool THtmlLxChDef::IsAlpha ( const char &  Ch ) const

inline

Definition at line 39 of file html.h.

{return int(ChTyV[Ch-TCh::Mn])==hlctAlpha;}

bool THtmlLxChDef::IsEoln ( const char &  Ch ) const

inline

Definition at line 35 of file html.h.

{return (Ch==TCh::CrCh)||(Ch==TCh::LfCh);}

bool THtmlLxChDef::IsLc ( const char &  Ch ) const

inline

Definition at line 51 of file html.h.

{return Ch==LcChV[Ch-TCh::Mn];}

bool THtmlLxChDef::IsNum ( const char &  Ch ) const

inline

Definition at line 40 of file html.h.

{return int(ChTyV[Ch-TCh::Mn])==hlctNum;}

bool THtmlLxChDef::IsSpace ( const char &  Ch ) const

inline

Definition at line 38 of file html.h.

{return int(ChTyV[Ch-TCh::Mn])==hlctSpace;}

bool THtmlLxChDef::IsSym ( const char &  Ch ) const

inline

Definition at line 43 of file html.h.

{return int(ChTyV[Ch-TCh::Mn])==hlctSym;}

bool THtmlLxChDef::IsUc ( const char &  Ch ) const

inline

Definition at line 50 of file html.h.

{return Ch==UcChV[Ch-TCh::Mn];}

bool THtmlLxChDef::IsUrl ( const char &  Ch ) const

inline

Definition at line 44 of file html.h.

                                   {
    int ChTy=ChTyV[Ch-TCh::Mn];
    return (ChTy==hlctAlpha)||(ChTy==hlctNum)||
     (Ch=='.')||(Ch=='-')||(Ch==':')||(Ch=='/')||(Ch=='~');}

bool THtmlLxChDef::IsWs ( const char &  Ch ) const

inline

Definition at line 36 of file html.h.

                                  {
    return (Ch==' ')||(Ch==TCh::TabCh)||(Ch==TCh::CrCh)||(Ch==TCh::LfCh);}

static PHtmlLxChDef THtmlLxChDef::Load ( TSIn &  SIn )

inlinestatic

Definition at line 27 of file html.h.

{return new THtmlLxChDef(SIn);}

THtmlLxChDef& THtmlLxChDef::operator= ( const THtmlLxChDef &  )

inline

Definition at line 31 of file html.h.

{Fail; return *this;}

void THtmlLxChDef::Save ( TSOut &  SOut )

inline

Definition at line 28 of file html.h.

                        {
    ChTyV.Save(SOut); UcChV.Save(SOut); LcChV.Save(SOut); EscStrH.Save(SOut);}

void THtmlLxChDef::SetChTy ( const THtmlLxChTy &  ChTy, const TStr &  Str  )

private

Definition at line 24 of file html.cpp.

                                                                  {
  for (int ChN=0; ChN<Str.Len(); ChN++){
    ChTyV[Str[ChN]-TCh::Mn]=TInt(ChTy);}
}

void THtmlLxChDef::SetEscStr ( const TStr &  SrcStr, const TStr &  DstStr  )

private

Definition at line 29 of file html.cpp.

                                                                  {
  EscStrH.AddDat(SrcStr, DstStr);
}

void THtmlLxChDef::SetUcCh ( const char &  UcCh, const char &  LcCh  )

private

Definition at line 3 of file html.cpp.

                                                            {
  // update upper-case (more lower cases may have one upper case)
  IAssert(
   (UcChV[LcCh-TCh::Mn]==TCh(0))||
   (UcChV[LcCh-TCh::Mn]==TCh(LcCh)));
  UcChV[LcCh-TCh::Mn]=TCh(UcCh);
  // update lower-case (one upper case may have only one lower case)
  if ((LcChV[UcCh-TCh::Mn]==TCh(0))||(LcChV[UcCh-TCh::Mn]==TCh(UcCh))){
    LcChV[UcCh-TCh::Mn]=TCh(LcCh);
  }
}

void THtmlLxChDef::SetUcCh ( const TStr &  Str )

private

Definition at line 15 of file html.cpp.

                                         {
  // set type of characters as letters
  SetChTy(hlctAlpha, Str);
  // first char in string is upper-case, rest are lower-case
  for (int ChN=1; ChN<Str.Len(); ChN++){
    SetUcCh(Str[0], Str[ChN]);
  }
}

Friends And Related Function Documentation

friend class TPt< THtmlLxChDef >

friend

Definition at line 14 of file html.h.

Member Data Documentation

PHtmlLxChDef THtmlLxChDef::ChDef =PHtmlLxChDef(new THtmlLxChDef())

static

Definition at line 65 of file html.h.

TIntV THtmlLxChDef::ChTyV

private

Definition at line 16 of file html.h.

TCRef THtmlLxChDef::CRef

private

Definition at line 14 of file html.h.

TStrStrH THtmlLxChDef::EscStrH

private

Definition at line 19 of file html.h.

TChV THtmlLxChDef::LcChV

private

Definition at line 18 of file html.h.

TChV THtmlLxChDef::UcChV

private

Definition at line 17 of file html.h.

---

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

• glib-core/html.h
• glib-core/html.cpp