sal/textenc/tenchelp.h

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: tenchelp.h,v $
  10  * $Revision: 1.11 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 #ifndef INCLUDED_RTL_TEXTENC_TENCHELP_H
  32 #define INCLUDED_RTL_TEXTENC_TENCHELP_H
  33
  34 #include "rtl/tencinfo.h"
  35 #include "rtl/textenc.h"
  36 #include "sal/types.h"
  37
  38 #if defined __cplusplus
  39 extern "C" {
  40 #endif /* __cplusplus */
  41
  42 #define RTL_TEXTCVT_BYTE_PRIVATE_START 0xF100
  43 #define RTL_TEXTCVT_BYTE_PRIVATE_END 0xF1FF
  44
  45 /* ----------------- */
  46 /* - TextConverter - */
  47 /* ----------------- */
  48
  49 typedef void ImplTextConverterData;
  50
  51 typedef
  52 sal_Size (* ImplConvertToUnicodeProc)(ImplTextConverterData const * pData,
  53                                       void * pContext,
  54                                       sal_Char const * pSrcBuf,
  55                                       sal_Size nSrcBytes,
  56                                       sal_Unicode * pDestBuf,
  57                                       sal_Size nDestChars,
  58                                       sal_uInt32 nFlags,
  59                                       sal_uInt32 * pInfo,
  60                                       sal_Size * pSrcCvtBytes);
  61
  62 typedef
  63 sal_Size (* ImplConvertToTextProc)(ImplTextConverterData const * pData,
  64                                    void * pContext,
  65                                    sal_Unicode const * pSrcBuf,
  66                                    sal_Size nSrcChars,
  67                                    sal_Char * pDestBuf,
  68                                    sal_Size nDestBytes,
  69                                    sal_uInt32 nFlags,
  70                                    sal_uInt32 * pInfo,
  71                                    sal_Size * pSrcCvtChars);
  72
  73 typedef void * (* ImplCreateTextContextProc)(void);
  74
  75 typedef void (* ImplDestroyTextContextProc)(void * pContext);
  76
  77 typedef void (* ImplResetTextContextProc)(void * pContext);
  78
  79 typedef void * (* ImplCreateUnicodeContextProc)(void);
  80
  81 typedef void (* ImplDestroyUnicodeContextProc)(void * pContext);
  82
  83 typedef void (* ImplResetUnicodeContextProc)(void * pContext);
  84
  85 typedef struct
  86 {
  87     ImplTextConverterData const * mpConvertData;
  88     ImplConvertToUnicodeProc mpConvertTextToUnicodeProc;
  89     ImplConvertToTextProc mpConvertUnicodeToTextProc;
  90     ImplCreateTextContextProc mpCreateTextToUnicodeContext;
  91     ImplDestroyTextContextProc mpDestroyTextToUnicodeContext;
  92     ImplResetTextContextProc mpResetTextToUnicodeContext;
  93     ImplCreateUnicodeContextProc mpCreateUnicodeToTextContext;
  94     ImplDestroyUnicodeContextProc mpDestroyUnicodeToTextContext;
  95     ImplResetUnicodeContextProc mpResetUnicodeToTextContext;
  96 } ImplTextConverter;
  97
  98 /* ----------------------------- */
  99 /* - TextEncoding - Structures - */
 100 /* ----------------------------- */
 101
 102 typedef struct
 103 {
 104     ImplTextConverter maConverter;
 105     sal_uInt8 mnMinCharSize;
 106     sal_uInt8 mnMaxCharSize;
 107     sal_uInt8 mnAveCharSize;
 108     sal_uInt8 mnBestWindowsCharset;
 109     char const * mpBestUnixCharset;
 110     char const * mpBestMimeCharset;
 111     sal_uInt32 mnInfoFlags;
 112 } ImplTextEncodingData;
 113
 114 /* ----------------------------------- */
 115 /* - TextConverter - Byte-Structures - */
 116 /* ----------------------------------- */
 117
 118 typedef struct
 119 {
 120     sal_uInt16                      mnUniChar;
 121     sal_uChar                       mnChar;
 122     sal_uChar                       mnChar2;
 123         // to cater for mappings like MS1258 with 1--2 bytes per Unicode char,
 124         // 0 if unused
 125 } ImplUniCharTabData;
 126
 127 typedef struct
 128 {
 129     const sal_uInt16*               mpToUniTab1;
 130     const sal_uInt16*               mpToUniTab2;
 131     sal_uChar                       mnToUniStart1;
 132     sal_uChar                       mnToUniEnd1;
 133     sal_uChar                       mnToUniStart2;
 134     sal_uChar                       mnToUniEnd2;
 135     const sal_uChar*                mpToCharTab1;
 136     const sal_uChar*                mpToCharTab2;
 137     const ImplUniCharTabData*       mpToCharExTab;
 138     sal_uInt16                      mnToCharStart1;
 139     sal_uInt16                      mnToCharEnd1;
 140     sal_uInt16                      mnToCharStart2;
 141     sal_uInt16                      mnToCharEnd2;
 142     sal_uInt16                      mnToCharExCount;
 143 } ImplByteConvertData;
 144
 145 /* ----------------------------------- */
 146 /* - TextConverter - DBCS-Structures - */
 147 /* ----------------------------------- */
 148
 149 typedef struct
 150 {
 151     sal_uChar                       mnLeadStart;
 152     sal_uChar                       mnLeadEnd;
 153     sal_uChar                       mnTrail1Start;
 154     sal_uChar                       mnTrail1End;
 155     sal_uChar                       mnTrail2Start;
 156     sal_uChar                       mnTrail2End;
 157     sal_uChar                       mnTrail3Start;
 158     sal_uChar                       mnTrail3End;
 159     sal_uChar                       mnTrailCount;
 160     sal_uInt16                      mnTrailRangeCount;
 161     sal_uInt16                      mnUniStart;
 162     sal_uInt16                      mnUniEnd;
 163 } ImplDBCSEUDCData;
 164
 165 typedef struct
 166 {
 167     sal_uInt16                      mnUniChar;
 168     sal_uInt8                       mnTrailStart;
 169     sal_uInt8                       mnTrailEnd;
 170     const sal_uInt16*               mpToUniTrailTab;
 171 } ImplDBCSToUniLeadTab;
 172
 173 typedef struct
 174 {
 175     sal_uInt8                       mnLowStart;
 176     sal_uInt8                       mnLowEnd;
 177     const sal_uInt16*               mpToUniTrailTab;
 178 } ImplUniToDBCSHighTab;
 179
 180 typedef struct
 181 {
 182     const ImplDBCSToUniLeadTab*     mpToUniLeadTab;
 183     const ImplUniToDBCSHighTab*     mpToDBCSHighTab;
 184     sal_uChar                       mnLeadStart;
 185     sal_uChar                       mnLeadEnd;
 186     sal_uChar                       mnTrailStart;
 187     sal_uChar                       mnTrailEnd;
 188     const ImplDBCSEUDCData*         mpEUDCTab;
 189     sal_uInt16                      mnEUDCCount;
 190 } ImplDBCSConvertData;
 191
 192 /* ---------------------------------- */
 193 /* - TextConverter - EUC-Structures - */
 194 /* ---------------------------------- */
 195
 196 typedef struct
 197 {
 198     const ImplDBCSToUniLeadTab*     mpJIS0208ToUniLeadTab;
 199     const ImplDBCSToUniLeadTab*     mpJIS0212ToUniLeadTab;
 200     const ImplUniToDBCSHighTab*     mpUniToJIS0208HighTab;
 201     const ImplUniToDBCSHighTab*     mpUniToJIS0212HighTab;
 202 } ImplEUCJPConvertData;
 203
 204 /* --------------------------------- */
 205 /* - TextConverter - HelpFunctions - */
 206 /* --------------------------------- */
 207
 208 sal_Unicode ImplGetUndefinedUnicodeChar(sal_uChar cChar, sal_uInt32 nFlags);
 209
 210 sal_Bool
 211 ImplHandleUndefinedUnicodeToTextChar(ImplTextConverterData const * pData,
 212                                      sal_Unicode const ** ppSrcBuf,
 213                                      sal_Unicode const * pEndSrcBuf,
 214                                      sal_Char ** ppDestBuf,
 215                                      sal_Char const * pEndDestBuf,
 216                                      sal_uInt32 nFlags,
 217                                      sal_uInt32 * pInfo);
 218     /* sal_True means 'continue,' sal_False means 'break' */
 219
 220 /* ----------------------------- */
 221 /* - TextConverter - Functions - */
 222 /* ----------------------------- */
 223
 224 sal_Size ImplSymbolToUnicode( const ImplTextConverterData* pData, void* pContext,
 225                               const sal_Char* pSrcBuf, sal_Size nSrcBytes,
 226                               sal_Unicode* pDestBuf, sal_Size nDestChars,
 227                               sal_uInt32 nFlags, sal_uInt32* pInfo, sal_Size* pSrcCvtBytes );
 228 sal_Size ImplUnicodeToSymbol( const ImplTextConverterData* pData, void* pContext,
 229                               const sal_Unicode* pSrcBuf, sal_Size nSrcChars,
 230                               sal_Char* pDestBuf, sal_Size nDestBytes,
 231                               sal_uInt32 nFlags, sal_uInt32* pInfo, sal_Size* pSrcCvtChars );
 232 sal_Size ImplCharToUnicode( const ImplTextConverterData* pData, void* pContext,
 233                             const sal_Char* pSrcBuf, sal_Size nSrcBytes,
 234                             sal_Unicode* pDestBuf, sal_Size nDestChars,
 235                             sal_uInt32 nFlags, sal_uInt32* pInfo, sal_Size* pSrcCvtBytes );
 236 sal_Size ImplUnicodeToChar( const ImplTextConverterData* pData, void* pContext,
 237                             const sal_Unicode* pSrcBuf, sal_Size nSrcChars,
 238                             sal_Char* pDestBuf, sal_Size nDestBytes,
 239                             sal_uInt32 nFlags, sal_uInt32* pInfo, sal_Size* pSrcCvtChars );
 240 sal_Size ImplDBCSToUnicode( const ImplTextConverterData* pData, void* pContext,
 241                             const sal_Char* pSrcBuf, sal_Size nSrcBytes,
 242                             sal_Unicode* pDestBuf, sal_Size nDestChars,
 243                             sal_uInt32 nFlags, sal_uInt32* pInfo,
 244                             sal_Size* pSrcCvtBytes );
 245 sal_Size ImplUnicodeToDBCS( const ImplTextConverterData* pData, void* pContext,
 246                             const sal_Unicode* pSrcBuf, sal_Size nSrcChars,
 247                             sal_Char* pDestBuf, sal_Size nDestBytes,
 248                             sal_uInt32 nFlags, sal_uInt32* pInfo,
 249                             sal_Size* pSrcCvtChars );
 250 sal_Size ImplEUCJPToUnicode( const ImplTextConverterData* pData,
 251                              void* pContext,
 252                              const sal_Char* pSrcBuf, sal_Size nSrcBytes,
 253                              sal_Unicode* pDestBuf, sal_Size nDestChars,
 254                              sal_uInt32 nFlags, sal_uInt32* pInfo,
 255                              sal_Size* pSrcCvtBytes );
 256 sal_Size ImplUnicodeToEUCJP( const ImplTextConverterData* pData,
 257                              void* pContext,
 258                              const sal_Unicode* pSrcBuf, sal_Size nSrcChars,
 259                              sal_Char* pDestBuf, sal_Size nDestBytes,
 260                              sal_uInt32 nFlags, sal_uInt32* pInfo,
 261                              sal_Size* pSrcCvtChars );
 262 void* ImplUTF7CreateUTF7TextToUnicodeContext( void );
 263 void ImplUTF7DestroyTextToUnicodeContext( void* pContext );
 264 void ImplUTF7ResetTextToUnicodeContext( void* pContext );
 265 sal_Size ImplUTF7ToUnicode( const ImplTextConverterData* pData, void* pContext,
 266                             const sal_Char* pSrcBuf, sal_Size nSrcBytes,
 267                             sal_Unicode* pDestBuf, sal_Size nDestChars,
 268                             sal_uInt32 nFlags, sal_uInt32* pInfo,
 269                             sal_Size* pSrcCvtBytes );
 270 void* ImplUTF7CreateUnicodeToTextContext( void );
 271 void ImplUTF7DestroyUnicodeToTextContext( void* pContext );
 272 void ImplUTF7ResetUnicodeToTextContext( void* pContext );
 273 sal_Size ImplUnicodeToUTF7( const ImplTextConverterData* pData, void* pContext,
 274                             const sal_Unicode* pSrcBuf, sal_Size nSrcChars,
 275                             sal_Char* pDestBuf, sal_Size nDestBytes,
 276                             sal_uInt32 nFlags, sal_uInt32* pInfo,
 277                             sal_Size* pSrcCvtChars );
 278
 279 void * ImplCreateUtf8ToUnicodeContext(void) SAL_THROW_EXTERN_C();
 280 void ImplResetUtf8ToUnicodeContext(void * pContext) SAL_THROW_EXTERN_C();
 281 sal_Size ImplConvertUtf8ToUnicode(ImplTextConverterData const * pData,
 282                                   void * pContext, sal_Char const * pSrcBuf,
 283                                   sal_Size nSrcBytes, sal_Unicode * pDestBuf,
 284                                   sal_Size nDestChars, sal_uInt32 nFlags,
 285                                   sal_uInt32 * pInfo, sal_Size * pSrcCvtBytes)
 286     SAL_THROW_EXTERN_C();
 287 void * ImplCreateUnicodeToUtf8Context(void) SAL_THROW_EXTERN_C();
 288 void ImplResetUnicodeToUtf8Context(void * pContext) SAL_THROW_EXTERN_C();
 289 sal_Size ImplConvertUnicodeToUtf8(ImplTextConverterData const * pData,
 290                                   void * pContext, sal_Unicode const * pSrcBuf,
 291                                   sal_Size nSrcChars, sal_Char * pDestBuf,
 292                                   sal_Size nDestBytes, sal_uInt32 nFlags,
 293                                   sal_uInt32 * pInfo, sal_Size* pSrcCvtChars)
 294     SAL_THROW_EXTERN_C();
 295
 296 #if defined __cplusplus
 297 }
 298 #endif /* __cplusplus */
 299
 300 #endif /* INCLUDED_RTL_TEXTENC_TENCHELP_H */