1 /* 2 ********************************************************************** 3 * Copyright (C) 1999-2011 International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ********************************************************************** 6 * 7 * 8 * ucnv_bld.h: 9 * Contains internal data structure definitions 10 * Created by Bertrand A. Damiba 11 * 12 * Change history: 13 * 14 * 06/29/2000 helena Major rewrite of the callback APIs. 15 */ 16 17 #ifndef UCNV_BLD_H 18 #define UCNV_BLD_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/utf16.h" 27 #include "ucnv_cnv.h" 28 #include "ucnvmbcs.h" 29 #include "ucnv_ext.h" 30 #include "udataswp.h" 31 32 /* size of the overflow buffers in UConverter, enough for escaping callbacks */ 33 #define UCNV_ERROR_BUFFER_LENGTH 32 34 35 /* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */ 36 #define UCNV_MAX_SUBCHAR_LEN 4 37 38 /* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */ 39 #define UCNV_MAX_CHAR_LEN 8 40 41 /* converter options bits */ 42 #define UCNV_OPTION_VERSION 0xf 43 #define UCNV_OPTION_SWAP_LFNL 0x10 44 45 #define UCNV_GET_VERSION(cnv) ((cnv)->options&UCNV_OPTION_VERSION) 46 47 U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv 48 itself is compiled under C++, the linkage of the funcptrs will 49 work. 50 */ 51 52 union UConverterTable { 53 UConverterMBCSTable mbcs; 54 }; 55 56 typedef union UConverterTable UConverterTable; 57 58 struct UConverterImpl; 59 typedef struct UConverterImpl UConverterImpl; 60 61 /** values for the unicodeMask */ 62 #define UCNV_HAS_SUPPLEMENTARY 1 63 #define UCNV_HAS_SURROGATES 2 64 65 typedef struct UConverterStaticData { /* +offset: size */ 66 uint32_t structSize; /* +0: 4 Size of this structure */ 67 68 char name 69 [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */ 70 71 int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */ 72 73 int8_t platform; /* +68: 1 platform of the converter (only IBM now) */ 74 int8_t conversionType; /* +69: 1 conversion type */ 75 76 int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */ 77 int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */ 78 79 uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */ 80 int8_t subCharLen; /* +76: 1 */ 81 82 uint8_t hasToUnicodeFallback; /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */ 83 uint8_t hasFromUnicodeFallback; /* +78: 1 */ 84 uint8_t unicodeMask; /* +79: 1 bit 0: has supplementary bit 1: has single surrogates */ 85 uint8_t subChar1; /* +80: 1 single-byte substitution character for IBM MBCS (0 if none) */ 86 uint8_t reserved[19]; /* +81: 19 to round out the structure */ 87 /* total size: 100 */ 88 } UConverterStaticData; 89 90 /* 91 * Defines the UConverterSharedData struct, 92 * the immutable, shared part of UConverter. 93 */ 94 struct UConverterSharedData { 95 uint32_t structSize; /* Size of this structure */ 96 uint32_t referenceCounter; /* used to count number of clients, 0xffffffff for static SharedData */ 97 98 const void *dataMemory; /* from udata_openChoice() - for cleanup */ 99 void *table; /* Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */ 100 101 const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */ 102 103 UBool sharedDataCached; /* TRUE: shared data is in cache, don't destroy on ucnv_close() if 0 ref. FALSE: shared data isn't in the cache, do attempt to clean it up if the ref is 0 */ 104 /*UBool staticDataOwned; TRUE if static data owned by shared data & should be freed with it, NEVER true for udata() loaded statics. This ignored variable was removed to make space for sharedDataCached. */ 105 106 const UConverterImpl *impl; /* vtable-style struct of mostly function pointers */ 107 108 /*initial values of some members of the mutable part of object */ 109 uint32_t toUnicodeStatus; 110 111 /* 112 * Shared data structures currently come in two flavors: 113 * - readonly for built-in algorithmic converters 114 * - allocated for MBCS, with a pointer to an allocated UConverterTable 115 * which always has a UConverterMBCSTable 116 * 117 * To eliminate one allocation, I am making the UConverterMBCSTable 118 * a member of the shared data. It is the last member so that static 119 * definitions of UConverterSharedData work as before. 120 * The table field above also remains to avoid updating all static 121 * definitions, but is now unused. 122 * 123 * markus 2003-nov-07 124 */ 125 UConverterMBCSTable mbcs; 126 }; 127 128 /* Defines a UConverter, the lightweight mutable part the user sees */ 129 130 struct UConverter { 131 /* 132 * Error function pointer called when conversion issues 133 * occur during a ucnv_fromUnicode call 134 */ 135 void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context, 136 UConverterFromUnicodeArgs *args, 137 const UChar *codeUnits, 138 int32_t length, 139 UChar32 codePoint, 140 UConverterCallbackReason reason, 141 UErrorCode *); 142 /* 143 * Error function pointer called when conversion issues 144 * occur during a ucnv_toUnicode call 145 */ 146 void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context, 147 UConverterToUnicodeArgs *args, 148 const char *codeUnits, 149 int32_t length, 150 UConverterCallbackReason reason, 151 UErrorCode *); 152 153 /* 154 * Pointer to additional data that depends on the converter type. 155 * Used by ISO 2022, SCSU, GB 18030 converters, possibly more. 156 */ 157 void *extraInfo; 158 159 const void *fromUContext; 160 const void *toUContext; 161 162 /* 163 * Pointer to charset bytes for substitution string if subCharLen>0, 164 * or pointer to Unicode string (UChar *) if subCharLen<0. 165 * subCharLen==0 is equivalent to using a skip callback. 166 * If the pointer is !=subUChars then it is allocated with 167 * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes. 168 * The subUChars field is declared as UChar[] not uint8_t[] to 169 * guarantee alignment for UChars. 170 */ 171 uint8_t *subChars; 172 173 UConverterSharedData *sharedData; /* Pointer to the shared immutable part of the converter object */ 174 175 uint32_t options; /* options flags from UConverterOpen, may contain additional bits */ 176 177 UBool sharedDataIsCached; /* TRUE: shared data is in cache, don't destroy on ucnv_close() if 0 ref. FALSE: shared data isn't in the cache, do attempt to clean it up if the ref is 0 */ 178 UBool isCopyLocal; /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ 179 UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ 180 181 UBool useFallback; 182 int8_t toULength; /* number of bytes in toUBytes */ 183 uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */ 184 uint32_t toUnicodeStatus; /* Used to internalize stream status information */ 185 int32_t mode; 186 uint32_t fromUnicodeStatus; 187 188 /* 189 * More fromUnicode() status. Serves 3 purposes: 190 * - keeps a lead surrogate between buffers (similar to toUBytes[]) 191 * - keeps a lead surrogate at the end of the stream, 192 * which the framework handles as truncated input 193 * - if the fromUnicode() implementation returns to the framework 194 * (ucnv.c ucnv_fromUnicode()), then the framework calls the callback 195 * for this code point 196 */ 197 UChar32 fromUChar32; 198 199 /* 200 * value for ucnv_getMaxCharSize() 201 * 202 * usually simply copied from the static data, but ucnvmbcs.c modifies 203 * the value depending on the converter type and options 204 */ 205 int8_t maxBytesPerUChar; 206 207 int8_t subCharLen; /* length of the codepage specific character sequence */ 208 int8_t invalidCharLength; 209 int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */ 210 211 int8_t invalidUCharLength; 212 int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */ 213 214 uint8_t subChar1; /* single-byte substitution character if different from subChar */ 215 UBool useSubChar1; 216 char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */ 217 uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */ 218 UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */ 219 220 UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */ 221 UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */ 222 223 /* fields for conversion extension */ 224 225 /* store previous UChars/chars to continue partial matches */ 226 UChar32 preFromUFirstCP; /* >=0: partial match */ 227 UChar preFromU[UCNV_EXT_MAX_UCHARS]; 228 char preToU[UCNV_EXT_MAX_BYTES]; 229 int8_t preFromULength, preToULength; /* negative: replay */ 230 int8_t preToUFirstLength; /* length of first character */ 231 232 /* new fields for ICU 4.0 */ 233 UConverterCallbackReason toUCallbackReason; /* (*fromCharErrorBehaviour) reason, set when error is detected */ 234 }; 235 236 U_CDECL_END /* end of UConverter */ 237 238 #define CONVERTER_FILE_EXTENSION ".cnv" 239 240 241 /** 242 * Return the number of all converter names. 243 * @param pErrorCode The error code 244 * @return the number of all converter names 245 */ 246 U_CFUNC uint16_t 247 ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode); 248 249 /** 250 * Return the (n)th converter name in mixed case, or NULL 251 * if there is none (typically, if the data cannot be loaded). 252 * 0<=index<ucnv_io_countAvailableConverters(). 253 * @param n The number specifies which converter name to get 254 * @param pErrorCode The error code 255 * @return the (n)th converter name in mixed case, or NULL if there is none. 256 */ 257 U_CFUNC const char * 258 ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode); 259 260 /** 261 * Load a non-algorithmic converter. 262 * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex). 263 */ 264 U_CAPI UConverterSharedData * 265 ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err); 266 267 /** 268 * Unload a non-algorithmic converter. 269 * It must be sharedData->referenceCounter != ~0 270 * and this function must be called inside umtx_lock(&cnvCacheMutex). 271 */ 272 U_CAPI void 273 ucnv_unload(UConverterSharedData *sharedData); 274 275 /** 276 * Swap ICU .cnv conversion tables. See udataswp.h. 277 * @internal 278 */ 279 U_CAPI int32_t U_EXPORT2 280 ucnv_swap(const UDataSwapper *ds, 281 const void *inData, int32_t length, void *outData, 282 UErrorCode *pErrorCode); 283 284 #endif 285 286 #endif /* _UCNV_BLD */ 287