2 **********************************************************************
3 * Copyright (C) 1999-2005, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 **********************************************************************
9 * Contains internal data structure definitions
10 * Created by Bertrand A. Damiba
14 * 06/29/2000 helena Major rewrite of the callback APIs.
20 #include "unicode/utypes.h"
22 #if !UCONFIG_NO_CONVERSION
24 #include "unicode/ucnv.h"
25 #include "unicode/ucnv_err.h"
31 /* size of the overflow buffers in UConverter, enough for escaping callbacks */
32 #define UCNV_ERROR_BUFFER_LENGTH 32
34 /* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */
35 #define UCNV_MAX_SUBCHAR_LEN 4
37 /* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */
38 #define UCNV_MAX_CHAR_LEN 8
40 /* converter options bits */
41 #define UCNV_OPTION_VERSION 0xf
42 #define UCNV_OPTION_SWAP_LFNL 0x10
45 U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv
46 itself is compiled under C++, the linkage of the funcptrs will
50 union UConverterTable {
51 UConverterMBCSTable mbcs;
54 typedef union UConverterTable UConverterTable;
56 struct UConverterImpl;
57 typedef struct UConverterImpl UConverterImpl;
59 /** values for the unicodeMask */
60 #define UCNV_HAS_SUPPLEMENTARY 1
61 #define UCNV_HAS_SURROGATES 2
63 typedef struct UConverterStaticData { /* +offset: size */
64 uint32_t structSize; /* +0: 4 Size of this structure */
67 [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */
69 int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */
71 int8_t platform; /* +68: 1 platform of the converter (only IBM now) */
72 int8_t conversionType; /* +69: 1 conversion type */
74 int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */
75 int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */
77 uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */
78 int8_t subCharLen; /* +76: 1 */
80 uint8_t hasToUnicodeFallback; /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */
81 uint8_t hasFromUnicodeFallback; /* +78: 1 */
82 uint8_t unicodeMask; /* +79: 1 bit 0: has supplementary bit 1: has single surrogates */
83 uint8_t subChar1; /* +80: 1 single-byte substitution character for IBM MBCS (0 if none) */
84 uint8_t reserved[19]; /* +81: 19 to round out the structure */
86 } UConverterStaticData;
89 * Defines the UConverterSharedData struct,
90 * the immutable, shared part of UConverter.
92 struct UConverterSharedData {
93 uint32_t structSize; /* Size of this structure */
94 uint32_t referenceCounter; /* used to count number of clients, 0xffffffff for static SharedData */
96 const void *dataMemory; /* from udata_openChoice() - for cleanup */
97 void *table; /* Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */
99 const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */
101 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 */
102 /*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. */
104 const UConverterImpl *impl; /* vtable-style struct of mostly function pointers */
106 /*initial values of some members of the mutable part of object */
107 uint32_t toUnicodeStatus;
110 * Shared data structures currently come in two flavors:
111 * - readonly for built-in algorithmic converters
112 * - allocated for MBCS, with a pointer to an allocated UConverterTable
113 * which always has a UConverterMBCSTable
115 * To eliminate one allocation, I am making the UConverterMBCSTable
116 * a member of the shared data. It is the last member so that static
117 * definitions of UConverterSharedData work as before.
118 * The table field above also remains to avoid updating all static
119 * definitions, but is now unused.
123 UConverterMBCSTable mbcs;
126 /* Defines a UConverter, the lightweight mutable part the user sees */
130 * Error function pointer called when conversion issues
131 * occur during a ucnv_fromUnicode call
133 void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context,
134 UConverterFromUnicodeArgs *args,
135 const UChar *codeUnits,
138 UConverterCallbackReason reason,
141 * Error function pointer called when conversion issues
142 * occur during a ucnv_toUnicode call
144 void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context,
145 UConverterToUnicodeArgs *args,
146 const char *codeUnits,
148 UConverterCallbackReason reason,
152 * Pointer to additional data that depends on the converter type.
153 * Used by ISO 2022, SCSU, GB 18030 converters, possibly more.
157 const void *fromUContext;
158 const void *toUContext;
160 UConverterSharedData *sharedData; /* Pointer to the shared immutable part of the converter object */
162 uint32_t options; /* options flags from UConverterOpen, may contain additional bits */
164 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 */
165 UBool isCopyLocal; /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
166 UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
169 int8_t toULength; /* number of bytes in toUBytes */
170 uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */
171 uint32_t toUnicodeStatus; /* Used to internalize stream status information */
173 uint32_t fromUnicodeStatus;
176 * More fromUnicode() status. Serves 3 purposes:
177 * - keeps a lead surrogate between buffers (similar to toUBytes[])
178 * - keeps a lead surrogate at the end of the stream,
179 * which the framework handles as truncated input
180 * - if the fromUnicode() implementation returns to the framework
181 * (ucnv.c ucnv_fromUnicode()), then the framework calls the callback
182 * for this code point
187 * value for ucnv_getMaxCharSize()
189 * usually simply copied from the static data, but ucnvmbcs.c modifies
190 * the value depending on the converter type and options
192 int8_t maxBytesPerUChar;
194 int8_t subCharLen; /* length of the codepage specific character sequence */
195 int8_t invalidCharLength;
196 int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */
198 int8_t invalidUCharLength;
199 int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */
201 uint8_t subChar1; /* single-byte substitution character if different from subChar */
203 uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* codepage specific character sequence */
204 char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */
205 uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */
207 UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */
208 UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */
210 /* fields for conversion extension */
212 /* store previous UChars/chars to continue partial matches */
213 UChar32 preFromUFirstCP; /* >=0: partial match */
214 UChar preFromU[UCNV_EXT_MAX_UCHARS];
215 char preToU[UCNV_EXT_MAX_BYTES];
216 int8_t preFromULength, preToULength; /* negative: replay */
217 int8_t preToUFirstLength; /* length of first character */
220 U_CDECL_END /* end of UConverter */
222 #define CONVERTER_FILE_EXTENSION ".cnv"
226 * Return the number of all converter names.
227 * @param pErrorCode The error code
228 * @return the number of all converter names
231 ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode);
234 * Return the (n)th converter name in mixed case, or NULL
235 * if there is none (typically, if the data cannot be loaded).
236 * 0<=index<ucnv_io_countAvailableConverters().
237 * @param n The number specifies which converter name to get
238 * @param pErrorCode The error code
239 * @return the (n)th converter name in mixed case, or NULL if there is none.
242 ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode);
245 * Load a non-algorithmic converter.
246 * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex).
248 UConverterSharedData *
249 ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err);
252 * Unload a non-algorithmic converter.
253 * It must be sharedData->referenceCounter != ~0
254 * and this function must be called inside umtx_lock(&cnvCacheMutex).
257 ucnv_unload(UConverterSharedData *sharedData);
260 * Swap ICU .cnv conversion tables. See udataswp.h.
263 U_CAPI int32_t U_EXPORT2
264 ucnv_swap(const UDataSwapper *ds,
265 const void *inData, int32_t length, void *outData,
266 UErrorCode *pErrorCode);
270 #endif /* _UCNV_BLD */