icuSources/common/ucnv_cnv.h

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 1999-2011, 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 #define UCNV_LOAD_ARGS_INITIALIZER \
  60     { (int32_t)sizeof(UConverterLoadArgs), 0, FALSE, FALSE, 0, 0, NULL, NULL, NULL }
  61
  62 typedef void (*UConverterLoad) (UConverterSharedData *sharedData,
  63                                 UConverterLoadArgs *pArgs,
  64                                 const uint8_t *raw, UErrorCode *pErrorCode);
  65 typedef void (*UConverterUnload) (UConverterSharedData *sharedData);
  66
  67 typedef void (*UConverterOpen) (UConverter *cnv, UConverterLoadArgs *pArgs, UErrorCode *pErrorCode);
  68 typedef void (*UConverterClose) (UConverter *cnv);
  69
  70 typedef enum UConverterResetChoice {
  71     UCNV_RESET_BOTH,
  72     UCNV_RESET_TO_UNICODE,
  73     UCNV_RESET_FROM_UNICODE
  74 } UConverterResetChoice;
  75
  76 typedef void (*UConverterReset) (UConverter *cnv, UConverterResetChoice choice);
  77
  78 /*
  79  * Converter implementation function(s) for ucnv_toUnicode().
  80  * If the toUnicodeWithOffsets function pointer is NULL,
  81  * then the toUnicode function will be used and the offsets will be set to -1.
  82  *
  83  * Must maintain state across buffers. Use toUBytes[toULength] for partial input
  84  * sequences; it will be checked in ucnv.c at the end of the input stream
  85  * to detect truncated input.
  86  * Some converters may need additional detection and may then set U_TRUNCATED_CHAR_FOUND.
  87  *
  88  * The toUnicodeWithOffsets must write exactly as many offset values as target
  89  * units. Write offset values of -1 for when the source index corresponding to
  90  * the output unit is not known (e.g., the character started in an earlier buffer).
  91  * The pArgs->offsets pointer need not be moved forward.
  92  *
  93  * At function return, either one of the following conditions must be true:
  94  * - U_BUFFER_OVERFLOW_ERROR and the target is full: target==targetLimit
  95  * - another error code with toUBytes[toULength] set to the offending input
  96  * - no error, and the source is consumed: source==sourceLimit
  97  *
  98  * The ucnv.c code will handle the end of the input (reset)
  99  * (reset, and truncation detection) and callbacks.
 100  */
 101 typedef void (*UConverterToUnicode) (UConverterToUnicodeArgs *, UErrorCode *);
 102
 103 /*
 104  * Same rules as for UConverterToUnicode.
 105  * A lead surrogate is kept in fromUChar32 across buffers, and if an error
 106  * occurs, then the offending input code point must be put into fromUChar32
 107  * as well.
 108  */
 109 typedef void (*UConverterFromUnicode) (UConverterFromUnicodeArgs *, UErrorCode *);
 110
 111 /*
 112  * Converter implementation function for ucnv_convertEx(), for direct conversion
 113  * between two charsets without pivoting through UTF-16.
 114  * The rules are the same as for UConverterToUnicode and UConverterFromUnicode.
 115  * In addition,
 116  * - The toUnicode side must behave and keep state exactly like the
 117  *   UConverterToUnicode implementation for the same source charset.
 118  * - A U_USING_DEFAULT_WARNING can be set to request to temporarily fall back
 119  *   to pivoting. When this function is called, the conversion framework makes
 120  *   sure that this warning is not set on input.
 121  * - Continuing a partial match and flushing the toUnicode replay buffer
 122  *   are handled by pivoting, using the toUnicode and fromUnicode functions.
 123  */
 124 typedef void (*UConverterConvert) (UConverterFromUnicodeArgs *pFromUArgs,
 125                                    UConverterToUnicodeArgs *pToUArgs,
 126                                    UErrorCode *pErrorCode);
 127
 128 /*
 129  * Converter implementation function for ucnv_getNextUChar().
 130  * If the function pointer is NULL, then the toUnicode function will be used.
 131  *
 132  * Will be called at a character boundary (toULength==0).
 133  * May return with
 134  * - U_INDEX_OUTOFBOUNDS_ERROR if there was no output for the input
 135  *   (the return value will be ignored)
 136  * - U_TRUNCATED_CHAR_FOUND or another error code (never U_BUFFER_OVERFLOW_ERROR!)
 137  *   with toUBytes[toULength] set to the offending input
 138  *   (the return value will be ignored)
 139  * - return UCNV_GET_NEXT_UCHAR_USE_TO_U, without moving the source pointer,
 140  *   to indicate that the ucnv.c code shall call the toUnicode function instead
 141  * - return a real code point result
 142  *
 143  * Unless UCNV_GET_NEXT_UCHAR_USE_TO_U is returned, the source bytes must be consumed.
 144  *
 145  * The ucnv.c code will handle the end of the input (reset)
 146  * (except for truncation detection!) and callbacks.
 147  */
 148 typedef UChar32 (*UConverterGetNextUChar) (UConverterToUnicodeArgs *, UErrorCode *);
 149
 150 typedef void (*UConverterGetStarters)(const UConverter* converter,
 151                                       UBool starters[256],
 152                                       UErrorCode *pErrorCode);
 153
 154 /* If this function pointer is null or if the function returns null
 155  * the name field in static data struct should be returned by
 156  * ucnv_getName() API function
 157  */
 158 typedef const char * (*UConverterGetName) (const UConverter *cnv);
 159
 160 /**
 161  * Write the codepage substitution character.
 162  * If this function is not set, then ucnv_cbFromUWriteSub() writes
 163  * the substitution character from UConverter.
 164  * For stateful converters, it is typically necessary to handle this
 165  * specificially for the converter in order to properly maintain the state.
 166  */
 167 typedef void (*UConverterWriteSub) (UConverterFromUnicodeArgs *pArgs, int32_t offsetIndex, UErrorCode *pErrorCode);
 168
 169 /**
 170  * For converter-specific safeClone processing
 171  * If this function is not set, then ucnv_safeClone assumes that the converter has no private data that changes
 172  * after the converter is done opening.
 173  * If this function is set, then it is called just after a memcpy() of
 174  * converter data to the new, empty converter, and is expected to set up
 175  * the initial state of the converter.  It is not expected to increment the
 176  * reference counts of the standard data types such as the shared data.
 177  */
 178 typedef UConverter * (*UConverterSafeClone) (const UConverter   *cnv,
 179                                              void               *stackBuffer,
 180                                              int32_t            *pBufferSize,
 181                                              UErrorCode         *status);
 182
 183 /**
 184  * Filters for some ucnv_getUnicodeSet() implementation code.
 185  */
 186 typedef enum UConverterSetFilter {
 187     UCNV_SET_FILTER_NONE,
 188     UCNV_SET_FILTER_DBCS_ONLY,
 189     UCNV_SET_FILTER_2022_CN,
 190     UCNV_SET_FILTER_SJIS,
 191     UCNV_SET_FILTER_GR94DBCS,
 192     UCNV_SET_FILTER_HZ,
 193     UCNV_SET_FILTER_COUNT
 194 } UConverterSetFilter;
 195
 196 /**
 197  * Fills the set of Unicode code points that can be converted by an ICU converter.
 198  * The API function ucnv_getUnicodeSet() clears the USet before calling
 199  * the converter's getUnicodeSet() implementation; the converter should only
 200  * add the appropriate code points to allow recursive use.
 201  * For example, the ISO-2022-JP converter will call each subconverter's
 202  * getUnicodeSet() implementation to consecutively add code points to
 203  * the same USet, which will result in a union of the sets of all subconverters.
 204  *
 205  * For more documentation, see ucnv_getUnicodeSet() in ucnv.h.
 206  */
 207 typedef void (*UConverterGetUnicodeSet) (const UConverter *cnv,
 208                                          const USetAdder *sa,
 209                                          UConverterUnicodeSet which,
 210                                          UErrorCode *pErrorCode);
 211
 212 UBool CONVERSION_U_SUCCESS (UErrorCode err);
 213
 214 /**
 215  * UConverterImpl contains all the data and functions for a converter type.
 216  * Its function pointers work much like a C++ vtable.
 217  * Many converter types need to define only a subset of the functions;
 218  * when a function pointer is NULL, then a default action will be performed.
 219  *
 220  * Every converter type must implement toUnicode, fromUnicode, and getNextUChar,
 221  * otherwise the converter may crash.
 222  * Every converter type that has variable-length codepage sequences should
 223  * also implement toUnicodeWithOffsets and fromUnicodeWithOffsets for
 224  * correct offset handling.
 225  * All other functions may or may not be implemented - it depends only on
 226  * whether the converter type needs them.
 227  *
 228  * When open() fails, then close() will be called, if present.
 229  */
 230 struct UConverterImpl {
 231     UConverterType type;
 232
 233     UConverterLoad load;
 234     UConverterUnload unload;
 235
 236     UConverterOpen open;
 237     UConverterClose close;
 238     UConverterReset reset;
 239
 240     UConverterToUnicode toUnicode;
 241     UConverterToUnicode toUnicodeWithOffsets;
 242     UConverterFromUnicode fromUnicode;
 243     UConverterFromUnicode fromUnicodeWithOffsets;
 244     UConverterGetNextUChar getNextUChar;
 245
 246     UConverterGetStarters getStarters;
 247     UConverterGetName getName;
 248     UConverterWriteSub writeSub;
 249     UConverterSafeClone safeClone;
 250     UConverterGetUnicodeSet getUnicodeSet;
 251
 252     UConverterConvert toUTF8;
 253     UConverterConvert fromUTF8;
 254 };
 255
 256 extern const UConverterSharedData
 257     _MBCSData, _Latin1Data,
 258     _UTF8Data, _UTF16BEData, _UTF16LEData, _UTF32BEData, _UTF32LEData,
 259     _ISO2022Data,
 260     _LMBCSData1,_LMBCSData2, _LMBCSData3, _LMBCSData4, _LMBCSData5, _LMBCSData6,
 261     _LMBCSData8,_LMBCSData11,_LMBCSData16,_LMBCSData17,_LMBCSData18,_LMBCSData19,
 262     _HZData,_ISCIIData, _SCSUData, _ASCIIData,
 263     _UTF7Data, _Bocu1Data, _UTF16Data, _UTF32Data, _CESU8Data, _IMAPData, _CompoundTextData;
 264
 265 U_CDECL_END
 266
 267 /** Always use fallbacks from codepage to Unicode */
 268 #define TO_U_USE_FALLBACK(useFallback) TRUE
 269 #define UCNV_TO_U_USE_FALLBACK(cnv) TRUE
 270
 271 /** Use fallbacks from Unicode to codepage when cnv->useFallback or for private-use code points */
 272 #define IS_PRIVATE_USE(c) ((uint32_t)((c)-0xe000)<0x1900 || (uint32_t)((c)-0xf0000)<0x20000)
 273 #define FROM_U_USE_FALLBACK(useFallback, c) ((useFallback) || IS_PRIVATE_USE(c))
 274 #define UCNV_FROM_U_USE_FALLBACK(cnv, c) FROM_U_USE_FALLBACK((cnv)->useFallback, c)
 275
 276 /**
 277  * Magic number for ucnv_getNextUChar(), returned by a
 278  * getNextUChar() implementation to indicate to use the converter's toUnicode()
 279  * instead of the native function.
 280  * @internal
 281  */
 282 #define UCNV_GET_NEXT_UCHAR_USE_TO_U -9
 283
 284 U_CFUNC void
 285 ucnv_getCompleteUnicodeSet(const UConverter *cnv,
 286                    const USetAdder *sa,
 287                    UConverterUnicodeSet which,
 288                    UErrorCode *pErrorCode);
 289
 290 U_CFUNC void
 291 ucnv_getNonSurrogateUnicodeSet(const UConverter *cnv,
 292                                const USetAdder *sa,
 293                                UConverterUnicodeSet which,
 294                                UErrorCode *pErrorCode);
 295
 296 U_CFUNC void
 297 ucnv_fromUWriteBytes(UConverter *cnv,
 298                      const char *bytes, int32_t length,
 299                      char **target, const char *targetLimit,
 300                      int32_t **offsets,
 301                      int32_t sourceIndex,
 302                      UErrorCode *pErrorCode);
 303 U_CFUNC void
 304 ucnv_toUWriteUChars(UConverter *cnv,
 305                     const UChar *uchars, int32_t length,
 306                     UChar **target, const UChar *targetLimit,
 307                     int32_t **offsets,
 308                     int32_t sourceIndex,
 309                     UErrorCode *pErrorCode);
 310
 311 U_CFUNC void
 312 ucnv_toUWriteCodePoint(UConverter *cnv,
 313                        UChar32 c,
 314                        UChar **target, const UChar *targetLimit,
 315                        int32_t **offsets,
 316                        int32_t sourceIndex,
 317                        UErrorCode *pErrorCode);
 318
 319 #endif
 320
 321 #endif /* UCNV_CNV */