Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 1 | // © 2016 and later: Unicode, Inc. and others. |
| 2 | // License & terms of use: http://www.unicode.org/copyright.html |
| 3 | /* |
| 4 | ********************************************************************** |
| 5 | * Copyright (C) 1999-2015 International Business Machines |
| 6 | * Corporation and others. All Rights Reserved. |
| 7 | ********************************************************************** |
| 8 | * |
| 9 | * |
| 10 | * ucnv_bld.h: |
| 11 | * Contains internal data structure definitions |
| 12 | * Created by Bertrand A. Damiba |
| 13 | * |
| 14 | * Change history: |
| 15 | * |
| 16 | * 06/29/2000 helena Major rewrite of the callback APIs. |
| 17 | */ |
| 18 | |
| 19 | #ifndef UCNV_BLD_H |
| 20 | #define UCNV_BLD_H |
| 21 | |
| 22 | #include "unicode/utypes.h" |
| 23 | |
| 24 | #if !UCONFIG_NO_CONVERSION |
| 25 | |
| 26 | #include "unicode/ucnv.h" |
| 27 | #include "unicode/ucnv_err.h" |
| 28 | #include "unicode/utf16.h" |
| 29 | #include "ucnv_cnv.h" |
| 30 | #include "ucnvmbcs.h" |
| 31 | #include "ucnv_ext.h" |
| 32 | #include "udataswp.h" |
| 33 | |
| 34 | /* size of the overflow buffers in UConverter, enough for escaping callbacks */ |
| 35 | #define UCNV_ERROR_BUFFER_LENGTH 32 |
| 36 | |
| 37 | /* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */ |
| 38 | #define UCNV_MAX_SUBCHAR_LEN 4 |
| 39 | |
| 40 | /* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */ |
| 41 | #define UCNV_MAX_CHAR_LEN 8 |
| 42 | |
| 43 | /* converter options bits */ |
| 44 | #define UCNV_OPTION_VERSION 0xf |
| 45 | #define UCNV_OPTION_SWAP_LFNL 0x10 |
| 46 | |
| 47 | #define UCNV_GET_VERSION(cnv) ((cnv)->options&UCNV_OPTION_VERSION) |
| 48 | |
| 49 | U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv |
| 50 | itself is compiled under C++, the linkage of the funcptrs will |
| 51 | work. |
| 52 | */ |
| 53 | |
| 54 | union UConverterTable { |
| 55 | UConverterMBCSTable mbcs; |
| 56 | }; |
| 57 | |
| 58 | typedef union UConverterTable UConverterTable; |
| 59 | |
| 60 | struct UConverterImpl; |
| 61 | typedef struct UConverterImpl UConverterImpl; |
| 62 | |
| 63 | /** values for the unicodeMask */ |
| 64 | #define UCNV_HAS_SUPPLEMENTARY 1 |
| 65 | #define UCNV_HAS_SURROGATES 2 |
| 66 | |
| 67 | typedef struct UConverterStaticData { /* +offset: size */ |
| 68 | uint32_t structSize; /* +0: 4 Size of this structure */ |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 69 | |
| 70 | char name |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 71 | [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */ |
| 72 | |
| 73 | int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */ |
| 74 | |
| 75 | int8_t platform; /* +68: 1 platform of the converter (only IBM now) */ |
| 76 | int8_t conversionType; /* +69: 1 conversion type */ |
| 77 | |
| 78 | int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */ |
| 79 | int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */ |
| 80 | |
| 81 | uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */ |
| 82 | int8_t subCharLen; /* +76: 1 */ |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 83 | |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 84 | uint8_t hasToUnicodeFallback; /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */ |
| 85 | uint8_t hasFromUnicodeFallback; /* +78: 1 */ |
| 86 | uint8_t unicodeMask; /* +79: 1 bit 0: has supplementary bit 1: has single surrogates */ |
| 87 | uint8_t subChar1; /* +80: 1 single-byte substitution character for IBM MBCS (0 if none) */ |
| 88 | uint8_t reserved[19]; /* +81: 19 to round out the structure */ |
| 89 | /* total size: 100 */ |
| 90 | } UConverterStaticData; |
| 91 | |
| 92 | /* |
| 93 | * Defines the UConverterSharedData struct, |
| 94 | * the immutable, shared part of UConverter. |
| 95 | */ |
| 96 | struct UConverterSharedData { |
| 97 | uint32_t structSize; /* Size of this structure */ |
| 98 | uint32_t referenceCounter; /* used to count number of clients, unused for static/immutable SharedData */ |
| 99 | |
| 100 | const void *dataMemory; /* from udata_openChoice() - for cleanup */ |
| 101 | |
| 102 | const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */ |
| 103 | |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 104 | 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 */ |
| 105 | /** If false, then referenceCounter is not used. Must not change after initialization. */ |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 106 | UBool isReferenceCounted; |
| 107 | |
| 108 | const UConverterImpl *impl; /* vtable-style struct of mostly function pointers */ |
| 109 | |
| 110 | /*initial values of some members of the mutable part of object */ |
| 111 | uint32_t toUnicodeStatus; |
| 112 | |
| 113 | /* |
| 114 | * Shared data structures currently come in two flavors: |
| 115 | * - readonly for built-in algorithmic converters |
| 116 | * - allocated for MBCS, with a pointer to an allocated UConverterTable |
| 117 | * which always has a UConverterMBCSTable |
| 118 | * |
| 119 | * To eliminate one allocation, I am making the UConverterMBCSTable |
| 120 | * a member of the shared data. |
| 121 | * |
| 122 | * markus 2003-nov-07 |
| 123 | */ |
| 124 | UConverterMBCSTable mbcs; |
| 125 | }; |
| 126 | |
| 127 | /** UConverterSharedData initializer for static, non-reference-counted converters. */ |
| 128 | #define UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(pStaticData, pImpl) \ |
| 129 | { \ |
| 130 | sizeof(UConverterSharedData), ~((uint32_t)0), \ |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 131 | NULL, pStaticData, false, false, pImpl, \ |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 132 | 0, UCNV_MBCS_TABLE_INITIALIZER \ |
| 133 | } |
| 134 | |
| 135 | /* Defines a UConverter, the lightweight mutable part the user sees */ |
| 136 | |
| 137 | struct UConverter { |
| 138 | /* |
| 139 | * Error function pointer called when conversion issues |
| 140 | * occur during a ucnv_fromUnicode call |
| 141 | */ |
| 142 | void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context, |
| 143 | UConverterFromUnicodeArgs *args, |
| 144 | const UChar *codeUnits, |
| 145 | int32_t length, |
| 146 | UChar32 codePoint, |
| 147 | UConverterCallbackReason reason, |
| 148 | UErrorCode *); |
| 149 | /* |
| 150 | * Error function pointer called when conversion issues |
| 151 | * occur during a ucnv_toUnicode call |
| 152 | */ |
| 153 | void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context, |
| 154 | UConverterToUnicodeArgs *args, |
| 155 | const char *codeUnits, |
| 156 | int32_t length, |
| 157 | UConverterCallbackReason reason, |
| 158 | UErrorCode *); |
| 159 | |
| 160 | /* |
| 161 | * Pointer to additional data that depends on the converter type. |
| 162 | * Used by ISO 2022, SCSU, GB 18030 converters, possibly more. |
| 163 | */ |
| 164 | void *extraInfo; |
| 165 | |
| 166 | const void *fromUContext; |
| 167 | const void *toUContext; |
| 168 | |
| 169 | /* |
| 170 | * Pointer to charset bytes for substitution string if subCharLen>0, |
| 171 | * or pointer to Unicode string (UChar *) if subCharLen<0. |
| 172 | * subCharLen==0 is equivalent to using a skip callback. |
| 173 | * If the pointer is !=subUChars then it is allocated with |
| 174 | * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes. |
| 175 | * The subUChars field is declared as UChar[] not uint8_t[] to |
| 176 | * guarantee alignment for UChars. |
| 177 | */ |
| 178 | uint8_t *subChars; |
| 179 | |
| 180 | UConverterSharedData *sharedData; /* Pointer to the shared immutable part of the converter object */ |
| 181 | |
| 182 | uint32_t options; /* options flags from UConverterOpen, may contain additional bits */ |
| 183 | |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 184 | 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 */ |
| 185 | UBool isCopyLocal; /* true if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ |
| 186 | UBool isExtraLocal; /* true if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */ |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 187 | |
| 188 | UBool useFallback; |
| 189 | int8_t toULength; /* number of bytes in toUBytes */ |
| 190 | uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */ |
| 191 | uint32_t toUnicodeStatus; /* Used to internalize stream status information */ |
| 192 | int32_t mode; |
| 193 | uint32_t fromUnicodeStatus; |
| 194 | |
| 195 | /* |
| 196 | * More fromUnicode() status. Serves 3 purposes: |
| 197 | * - keeps a lead surrogate between buffers (similar to toUBytes[]) |
| 198 | * - keeps a lead surrogate at the end of the stream, |
| 199 | * which the framework handles as truncated input |
| 200 | * - if the fromUnicode() implementation returns to the framework |
| 201 | * (ucnv.c ucnv_fromUnicode()), then the framework calls the callback |
| 202 | * for this code point |
| 203 | */ |
| 204 | UChar32 fromUChar32; |
| 205 | |
| 206 | /* |
| 207 | * value for ucnv_getMaxCharSize() |
| 208 | * |
| 209 | * usually simply copied from the static data, but ucnvmbcs.c modifies |
| 210 | * the value depending on the converter type and options |
| 211 | */ |
| 212 | int8_t maxBytesPerUChar; |
| 213 | |
| 214 | int8_t subCharLen; /* length of the codepage specific character sequence */ |
| 215 | int8_t invalidCharLength; |
| 216 | int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */ |
| 217 | |
| 218 | int8_t invalidUCharLength; |
| 219 | int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */ |
| 220 | |
| 221 | uint8_t subChar1; /* single-byte substitution character if different from subChar */ |
| 222 | UBool useSubChar1; |
| 223 | char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */ |
| 224 | uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */ |
| 225 | UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */ |
| 226 | |
| 227 | UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */ |
| 228 | UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */ |
| 229 | |
| 230 | /* fields for conversion extension */ |
| 231 | |
| 232 | /* store previous UChars/chars to continue partial matches */ |
| 233 | UChar32 preFromUFirstCP; /* >=0: partial match */ |
| 234 | UChar preFromU[UCNV_EXT_MAX_UCHARS]; |
| 235 | char preToU[UCNV_EXT_MAX_BYTES]; |
| 236 | int8_t preFromULength, preToULength; /* negative: replay */ |
| 237 | int8_t preToUFirstLength; /* length of first character */ |
| 238 | |
| 239 | /* new fields for ICU 4.0 */ |
| 240 | UConverterCallbackReason toUCallbackReason; /* (*fromCharErrorBehaviour) reason, set when error is detected */ |
| 241 | }; |
| 242 | |
| 243 | U_CDECL_END /* end of UConverter */ |
| 244 | |
| 245 | #define CONVERTER_FILE_EXTENSION ".cnv" |
| 246 | |
| 247 | |
| 248 | /** |
| 249 | * Return the number of all converter names. |
| 250 | * @param pErrorCode The error code |
| 251 | * @return the number of all converter names |
| 252 | */ |
| 253 | U_CFUNC uint16_t |
| 254 | ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode); |
| 255 | |
| 256 | /** |
| 257 | * Return the (n)th converter name in mixed case, or NULL |
| 258 | * if there is none (typically, if the data cannot be loaded). |
| 259 | * 0<=index<ucnv_io_countAvailableConverters(). |
| 260 | * @param n The number specifies which converter name to get |
| 261 | * @param pErrorCode The error code |
| 262 | * @return the (n)th converter name in mixed case, or NULL if there is none. |
| 263 | */ |
| 264 | U_CFUNC const char * |
| 265 | ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode); |
| 266 | |
| 267 | /** |
| 268 | * Load a non-algorithmic converter. |
| 269 | * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex). |
| 270 | */ |
| 271 | U_CAPI UConverterSharedData * |
| 272 | ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err); |
| 273 | |
| 274 | /** |
| 275 | * Unload a non-algorithmic converter. |
| 276 | * It must be sharedData->isReferenceCounted |
| 277 | * and this function must be called inside umtx_lock(&cnvCacheMutex). |
| 278 | */ |
| 279 | U_CAPI void |
| 280 | ucnv_unload(UConverterSharedData *sharedData); |
| 281 | |
| 282 | /** |
| 283 | * Swap ICU .cnv conversion tables. See udataswp.h. |
| 284 | * @internal |
| 285 | */ |
| 286 | U_CAPI int32_t U_EXPORT2 |
| 287 | ucnv_swap(const UDataSwapper *ds, |
| 288 | const void *inData, int32_t length, void *outData, |
| 289 | UErrorCode *pErrorCode); |
| 290 | |
| 291 | U_CAPI void U_EXPORT2 |
Victor Chang | ce4bf3c | 2021-01-19 16:34:24 +0000 | [diff] [blame] | 292 | ucnv_enableCleanup(void); |
Victor Chang | 7322950 | 2020-09-17 13:39:19 +0100 | [diff] [blame] | 293 | |
| 294 | #endif |
| 295 | |
| 296 | #endif /* _UCNV_BLD */ |