icuSources/common/ucnv_cnv.h

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 1999-2009, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 **********************************************************************
   6 *
   7 *   ucnv_cnv.h:
   8 *   Definitions for converter implementations.
   9 *
  10 * Modification History:
  11 *
  12 *   Date        Name        Description
  13 *   05/09/00    helena      Added implementation to handle fallback mappings.
  14 *   06/29/2000  helena      Major rewrite of the callback APIs.
  15 */
  16
  17 #ifndef UCNV_CNV_H
  18 #define UCNV_CNV_H
  19
  20 #include "unicode/utypes.h"
  21
  22 #if !UCONFIG_NO_CONVERSION
  23
  24 #include "unicode/ucnv.h"
  25 #include "unicode/ucnv_err.h"
  26 #include "unicode/uset.h"
  27 #include "uset_imp.h"
  28
  29 U_CDECL_BEGIN
  30
  31 /* this is used in fromUnicode DBCS tables as an "unassigned" marker */
  32 #define missingCharMarker 0xFFFF
  33
  34 /*
  35  * #define missingUCharMarker 0xfffe
  36  *
  37  * commented out because there are actually two values used in toUnicode tables:
  38  * U+fffe "unassigned"
  39  * U+ffff "illegal"
  40  */
  41
  42 /** Forward declaration, see ucnv_bld.h */
  43 struct UConverterSharedData;
  44 typedef struct UConverterSharedData UConverterSharedData;
  45
  46 /* function types for UConverterImpl ---------------------------------------- */
  47
  48 /* struct with arguments for UConverterLoad and ucnv_load() */
  49 typedef struct {
  50     int32_t size;               /* sizeof(UConverterLoadArgs) */
  51     int32_t nestedLoads;        /* count nested ucnv_load() calls */
  52     UBool onlyTestIsLoadable;   /* input: don't actually load */
  53     UBool reserved0;            /* reserved - for good alignment of the pointers */
  54     int16_t reserved;           /* reserved - for good alignment of the pointers */
  55     uint32_t options;
  56     const char *pkg, *name, *locale;
  57 } UConverterLoadArgs;
  58
  59 typedef void (*UConverterLoad) (UConverterSharedData *sharedData,
  60                                 UConverterLoadArgs *pArgs,
  61                                 const uint8_t *raw, UErrorCode *pErrorCode);
  62 typedef void (*UConverterUnload) (UConverterSharedData *sharedData);
  63
  64 typedef void (*UConverterOpen) (UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *pErrorCode);
  65 typedef void (*UConverterClose) (UConverter *cnv);
  66
  67 typedef enum UConverterResetChoice {
  68     UCNV_RESET_BOTH,
  69     UCNV_RESET_TO_UNICODE,
  70     UCNV_RESET_FROM_UNICODE
  71 } UConverterResetChoice;
  72
  73 typedef void (*UConverterReset) (UConverter *cnv, UConverterResetChoice choice);
  74
  75 /*
  76  * Converter implementation function(s) for ucnv_toUnicode().
  77  * If the toUnicodeWithOffsets function pointer is NULL,
  78  * then the toUnicode function will be used and the offsets will be set to -1.
  79  *
  80  * Must maintain state across buffers. Use toUBytes[toULength] for partial input
  81  * sequences; it will be checked in ucnv.c at the end of the input stream
  82  * to detect truncated input.
  83  * Some converters may need additional detection and may then set U_TRUNCATED_CHAR_FOUND.
  84  *
  85  * The toUnicodeWithOffsets must write exactly as many offset values as target
  86  * units. Write offset values of -1 for when the source index corresponding to
  87  * the output unit is not known (e.g., the character started in an earlier buffer).
  88  * The pArgs->offsets pointer need not be moved forward.
  89  *
  90  * At function return, either one of the following conditions must be true:
  91  * - U_BUFFER_OVERFLOW_ERROR and the target is full: target==targetLimit
  92  * - another error code with toUBytes[toULength] set to the offending input
  93  * - no error, and the source is consumed: source==sourceLimit
  94  *
  95  * The ucnv.c code will handle the end of the input (reset)
  96  * (reset, and truncation detection) and callbacks.
  97  */
  98 typedef void (*UConverterToUnicode) (UConverterToUnicodeArgs *, UErrorCode *);
  99
 100 /*
 101  * Same rules as for UConverterToUnicode.
 102  * A lead surrogate is kept in fromUChar32 across buffers, and if an error
 103  * occurs, then the offending input code point must be put into fromUChar32
 104  * as well.
 105  */
 106 typedef void (*UConverterFromUnicode) (UConverterFromUnicodeArgs *, UErrorCode *);
 107
 108 /*
 109  * Converter implementation function for ucnv_convertEx(), for direct conversion
 110  * between two charsets without pivoting through UTF-16.
 111  * The rules are the same as for UConverterToUnicode and UConverterFromUnicode.
 112  * In addition,
 113  * - The toUnicode side must behave and keep state exactly like the
 114  *   UConverterToUnicode implementation for the same source charset.
 115  * - A U_USING_DEFAULT_WARNING can be set to request to temporarily fall back
 116  *   to pivoting. When this function is called, the conversion framework makes
 117  *   sure that this warning is not set on input.
 118  * - Continuing a partial match and flushing the toUnicode replay buffer
 119  *   are handled by pivoting, using the toUnicode and fromUnicode functions.
 120  */
 121 typedef void (*UConverterConvert) (UConverterFromUnicodeArgs *pFromUArgs,
 122                                    UConverterToUnicodeArgs *pToUArgs,
 123                                    UErrorCode *pErrorCode);
 124
 125 /*
 126  * Converter implementation function for ucnv_getNextUChar().
 127  * If the function pointer is NULL, then the toUnicode function will be used.
 128  *
 129  * Will be called at a character boundary (toULength==0).
 130  * May return with
 131  * - U_INDEX_OUTOFBOUNDS_ERROR if there was no output for the input
 132  *   (the return value will be ignored)
 133  * - U_TRUNCATED_CHAR_FOUND or another error code (never U_BUFFER_OVERFLOW_ERROR!)
 134  *   with toUBytes[toULength] set to the offending input
 135  *   (the return value will be ignored)
 136  * - return UCNV_GET_NEXT_UCHAR_USE_TO_U, without moving the source pointer,
 137  *   to indicate that the ucnv.c code shall call the toUnicode function instead
 138  * - return a real code point result
 139  *
 140  * Unless UCNV_GET_NEXT_UCHAR_USE_TO_U is returned, the source bytes must be consumed.
 141  *
 142  * The ucnv.c code will handle the end of the input (reset)
 143  * (except for truncation detection!) and callbacks.
 144  */
 145 typedef UChar32 (*UConverterGetNextUChar) (UConverterToUnicodeArgs *, UErrorCode *);
 146
 147 typedef void (*UConverterGetStarters)(const UConverter* converter,
 148                                       UBool starters[256],
 149                                       UErrorCode *pErrorCode);
 150
 151 /* If this function pointer is null or if the function returns null
 152  * the name field in static data struct should be returned by
 153  * ucnv_getName() API function
 154  */
 155 typedef const char * (*UConverterGetName) (const UConverter *cnv);
 156
 157 /**
 158  * Write the codepage substitution character.
 159  * If this function is not set, then ucnv_cbFromUWriteSub() writes
 160  * the substitution character from UConverter.
 161  * For stateful converters, it is typically necessary to handle this
 162  * specificially for the converter in order to properly maintain the state.
 163  */
 164 typedef void (*UConverterWriteSub) (UConverterFromUnicodeArgs *pArgs, int32_t offsetIndex, UErrorCode *pErrorCode);
 165
 166 /**
 167  * For converter-specific safeClone processing
 168  * If this function is not set, then ucnv_safeClone assumes that the converter has no private data that changes
 169  * after the converter is done opening.
 170  * If this function is set, then it is called just after a memcpy() of
 171  * converter data to the new, empty converter, and is expected to set up
 172  * the initial state of the converter.  It is not expected to increment the
 173  * reference counts of the standard data types such as the shared data.
 174  */
 175 typedef UConverter * (*UConverterSafeClone) (const UConverter   *cnv,
 176                                              void               *stackBuffer,
 177                                              int32_t            *pBufferSize,
 178                                              UErrorCode         *status);
 179
 180 /**
 181  * Filters for some ucnv_getUnicodeSet() implementation code.
 182  */
 183 typedef enum UConverterSetFilter {
 184     UCNV_SET_FILTER_NONE,
 185     UCNV_SET_FILTER_DBCS_ONLY,
 186     UCNV_SET_FILTER_2022_CN,
 187     UCNV_SET_FILTER_SJIS,
 188     UCNV_SET_FILTER_GR94DBCS,
 189     UCNV_SET_FILTER_HZ,
 190     UCNV_SET_FILTER_COUNT
 191 } UConverterSetFilter;
 192
 193 /**
 194  * Fills the set of Unicode code points that can be converted by an ICU converter.
 195  * The API function ucnv_getUnicodeSet() clears the USet before calling
 196  * the converter's getUnicodeSet() implementation; the converter should only
 197  * add the appropriate code points to allow recursive use.
 198  * For example, the ISO-2022-JP converter will call each subconverter's
 199  * getUnicodeSet() implementation to consecutively add code points to
 200  * the same USet, which will result in a union of the sets of all subconverters.
 201  *
 202  * For more documentation, see ucnv_getUnicodeSet() in ucnv.h.
 203  */
 204 typedef void (*UConverterGetUnicodeSet) (const UConverter *cnv,
 205                                          const USetAdder *sa,
 206                                          UConverterUnicodeSet which,
 207                                          UErrorCode *pErrorCode);
 208
 209 UBool CONVERSION_U_SUCCESS (UErrorCode err);
 210
 211 /**
 212  * UConverterImpl contains all the data and functions for a converter type.
 213  * Its function pointers work much like a C++ vtable.
 214  * Many converter types need to define only a subset of the functions;
 215  * when a function pointer is NULL, then a default action will be performed.
 216  *
 217  * Every converter type must implement toUnicode, fromUnicode, and getNextUChar,
 218  * otherwise the converter may crash.
 219  * Every converter type that has variable-length codepage sequences should
 220  * also implement toUnicodeWithOffsets and fromUnicodeWithOffsets for
 221  * correct offset handling.
 222  * All other functions may or may not be implemented - it depends only on
 223  * whether the converter type needs them.
 224  *
 225  * When open() fails, then close() will be called, if present.
 226  */
 227 struct UConverterImpl {
 228     UConverterType type;
 229
 230     UConverterLoad load;
 231     UConverterUnload unload;
 232
 233     UConverterOpen open;
 234     UConverterClose close;
 235     UConverterReset reset;
 236
 237     UConverterToUnicode toUnicode;
 238     UConverterToUnicode toUnicodeWithOffsets;
 239     UConverterFromUnicode fromUnicode;
 240     UConverterFromUnicode fromUnicodeWithOffsets;
 241     UConverterGetNextUChar getNextUChar;
 242
 243     UConverterGetStarters getStarters;
 244     UConverterGetName getName;
 245     UConverterWriteSub writeSub;
 246     UConverterSafeClone safeClone;
 247     UConverterGetUnicodeSet getUnicodeSet;
 248
 249     UConverterConvert toUTF8;
 250     UConverterConvert fromUTF8;
 251 };
 252
 253 extern const UConverterSharedData
 254     _MBCSData, _Latin1Data,
 255     _UTF8Data, _UTF16BEData, _UTF16LEData, _UTF32BEData, _UTF32LEData,
 256     _ISO2022Data,
 257     _LMBCSData1,_LMBCSData2, _LMBCSData3, _LMBCSData4, _LMBCSData5, _LMBCSData6,
 258     _LMBCSData8,_LMBCSData11,_LMBCSData16,_LMBCSData17,_LMBCSData18,_LMBCSData19,
 259     _HZData,_ISCIIData, _SCSUData, _ASCIIData,
 260     _UTF7Data, _Bocu1Data, _UTF16Data, _UTF32Data, _CESU8Data, _IMAPData;
 261
 262 U_CDECL_END
 263
 264 /** Always use fallbacks from codepage to Unicode */
 265 #define TO_U_USE_FALLBACK(useFallback) TRUE
 266 #define UCNV_TO_U_USE_FALLBACK(cnv) TRUE
 267
 268 /** Use fallbacks from Unicode to codepage when cnv->useFallback or for private-use code points */
 269 #define IS_PRIVATE_USE(c) ((uint32_t)((c)-0xe000)<0x1900 || (uint32_t)((c)-0xf0000)<0x20000)
 270 #define FROM_U_USE_FALLBACK(useFallback, c) ((useFallback) || IS_PRIVATE_USE(c))
 271 #define UCNV_FROM_U_USE_FALLBACK(cnv, c) FROM_U_USE_FALLBACK((cnv)->useFallback, c)
 272
 273 /**
 274  * Magic number for ucnv_getNextUChar(), returned by a
 275  * getNextUChar() implementation to indicate to use the converter's toUnicode()
 276  * instead of the native function.
 277  * @internal
 278  */
 279 #define UCNV_GET_NEXT_UCHAR_USE_TO_U -9
 280
 281 U_CFUNC void
 282 ucnv_getCompleteUnicodeSet(const UConverter *cnv,
 283                    const USetAdder *sa,
 284                    UConverterUnicodeSet which,
 285                    UErrorCode *pErrorCode);
 286
 287 U_CFUNC void
 288 ucnv_getNonSurrogateUnicodeSet(const UConverter *cnv,
 289                                const USetAdder *sa,
 290                                UConverterUnicodeSet which,
 291                                UErrorCode *pErrorCode);
 292
 293 U_CFUNC void
 294 ucnv_fromUWriteBytes(UConverter *cnv,
 295                      const char *bytes, int32_t length,
 296                      char **target, const char *targetLimit,
 297                      int32_t **offsets,
 298                      int32_t sourceIndex,
 299                      UErrorCode *pErrorCode);
 300 U_CFUNC void
 301 ucnv_toUWriteUChars(UConverter *cnv,
 302                     const UChar *uchars, int32_t length,
 303                     UChar **target, const UChar *targetLimit,
 304                     int32_t **offsets,
 305                     int32_t sourceIndex,
 306                     UErrorCode *pErrorCode);
 307
 308 U_CFUNC void
 309 ucnv_toUWriteCodePoint(UConverter *cnv,
 310                        UChar32 c,
 311                        UChar **target, const UChar *targetLimit,
 312                        int32_t **offsets,
 313                        int32_t sourceIndex,
 314                        UErrorCode *pErrorCode);
 315
 316 #endif
 317
 318 #endif /* UCNV_CNV */