1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/ucnv_bld.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,286 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (C) 1999-2011 International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +*
1.10 +*
1.11 +* ucnv_bld.h:
1.12 +* Contains internal data structure definitions
1.13 +* Created by Bertrand A. Damiba
1.14 +*
1.15 +* Change history:
1.16 +*
1.17 +* 06/29/2000 helena Major rewrite of the callback APIs.
1.18 +*/
1.19 +
1.20 +#ifndef UCNV_BLD_H
1.21 +#define UCNV_BLD_H
1.22 +
1.23 +#include "unicode/utypes.h"
1.24 +
1.25 +#if !UCONFIG_NO_CONVERSION
1.26 +
1.27 +#include "unicode/ucnv.h"
1.28 +#include "unicode/ucnv_err.h"
1.29 +#include "unicode/utf16.h"
1.30 +#include "ucnv_cnv.h"
1.31 +#include "ucnvmbcs.h"
1.32 +#include "ucnv_ext.h"
1.33 +#include "udataswp.h"
1.34 +
1.35 +/* size of the overflow buffers in UConverter, enough for escaping callbacks */
1.36 +#define UCNV_ERROR_BUFFER_LENGTH 32
1.37 +
1.38 +/* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */
1.39 +#define UCNV_MAX_SUBCHAR_LEN 4
1.40 +
1.41 +/* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */
1.42 +#define UCNV_MAX_CHAR_LEN 8
1.43 +
1.44 +/* converter options bits */
1.45 +#define UCNV_OPTION_VERSION 0xf
1.46 +#define UCNV_OPTION_SWAP_LFNL 0x10
1.47 +
1.48 +#define UCNV_GET_VERSION(cnv) ((cnv)->options&UCNV_OPTION_VERSION)
1.49 +
1.50 +U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv
1.51 + itself is compiled under C++, the linkage of the funcptrs will
1.52 + work.
1.53 + */
1.54 +
1.55 +union UConverterTable {
1.56 + UConverterMBCSTable mbcs;
1.57 +};
1.58 +
1.59 +typedef union UConverterTable UConverterTable;
1.60 +
1.61 +struct UConverterImpl;
1.62 +typedef struct UConverterImpl UConverterImpl;
1.63 +
1.64 +/** values for the unicodeMask */
1.65 +#define UCNV_HAS_SUPPLEMENTARY 1
1.66 +#define UCNV_HAS_SURROGATES 2
1.67 +
1.68 +typedef struct UConverterStaticData { /* +offset: size */
1.69 + uint32_t structSize; /* +0: 4 Size of this structure */
1.70 +
1.71 + char name
1.72 + [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */
1.73 +
1.74 + int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */
1.75 +
1.76 + int8_t platform; /* +68: 1 platform of the converter (only IBM now) */
1.77 + int8_t conversionType; /* +69: 1 conversion type */
1.78 +
1.79 + int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */
1.80 + int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */
1.81 +
1.82 + uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */
1.83 + int8_t subCharLen; /* +76: 1 */
1.84 +
1.85 + uint8_t hasToUnicodeFallback; /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */
1.86 + uint8_t hasFromUnicodeFallback; /* +78: 1 */
1.87 + uint8_t unicodeMask; /* +79: 1 bit 0: has supplementary bit 1: has single surrogates */
1.88 + uint8_t subChar1; /* +80: 1 single-byte substitution character for IBM MBCS (0 if none) */
1.89 + uint8_t reserved[19]; /* +81: 19 to round out the structure */
1.90 + /* total size: 100 */
1.91 +} UConverterStaticData;
1.92 +
1.93 +/*
1.94 + * Defines the UConverterSharedData struct,
1.95 + * the immutable, shared part of UConverter.
1.96 + */
1.97 +struct UConverterSharedData {
1.98 + uint32_t structSize; /* Size of this structure */
1.99 + uint32_t referenceCounter; /* used to count number of clients, 0xffffffff for static SharedData */
1.100 +
1.101 + const void *dataMemory; /* from udata_openChoice() - for cleanup */
1.102 + void *table; /* Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */
1.103 +
1.104 + const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */
1.105 +
1.106 + 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 */
1.107 + /*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. */
1.108 +
1.109 + const UConverterImpl *impl; /* vtable-style struct of mostly function pointers */
1.110 +
1.111 + /*initial values of some members of the mutable part of object */
1.112 + uint32_t toUnicodeStatus;
1.113 +
1.114 + /*
1.115 + * Shared data structures currently come in two flavors:
1.116 + * - readonly for built-in algorithmic converters
1.117 + * - allocated for MBCS, with a pointer to an allocated UConverterTable
1.118 + * which always has a UConverterMBCSTable
1.119 + *
1.120 + * To eliminate one allocation, I am making the UConverterMBCSTable
1.121 + * a member of the shared data. It is the last member so that static
1.122 + * definitions of UConverterSharedData work as before.
1.123 + * The table field above also remains to avoid updating all static
1.124 + * definitions, but is now unused.
1.125 + *
1.126 + * markus 2003-nov-07
1.127 + */
1.128 + UConverterMBCSTable mbcs;
1.129 +};
1.130 +
1.131 +/* Defines a UConverter, the lightweight mutable part the user sees */
1.132 +
1.133 +struct UConverter {
1.134 + /*
1.135 + * Error function pointer called when conversion issues
1.136 + * occur during a ucnv_fromUnicode call
1.137 + */
1.138 + void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context,
1.139 + UConverterFromUnicodeArgs *args,
1.140 + const UChar *codeUnits,
1.141 + int32_t length,
1.142 + UChar32 codePoint,
1.143 + UConverterCallbackReason reason,
1.144 + UErrorCode *);
1.145 + /*
1.146 + * Error function pointer called when conversion issues
1.147 + * occur during a ucnv_toUnicode call
1.148 + */
1.149 + void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context,
1.150 + UConverterToUnicodeArgs *args,
1.151 + const char *codeUnits,
1.152 + int32_t length,
1.153 + UConverterCallbackReason reason,
1.154 + UErrorCode *);
1.155 +
1.156 + /*
1.157 + * Pointer to additional data that depends on the converter type.
1.158 + * Used by ISO 2022, SCSU, GB 18030 converters, possibly more.
1.159 + */
1.160 + void *extraInfo;
1.161 +
1.162 + const void *fromUContext;
1.163 + const void *toUContext;
1.164 +
1.165 + /*
1.166 + * Pointer to charset bytes for substitution string if subCharLen>0,
1.167 + * or pointer to Unicode string (UChar *) if subCharLen<0.
1.168 + * subCharLen==0 is equivalent to using a skip callback.
1.169 + * If the pointer is !=subUChars then it is allocated with
1.170 + * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes.
1.171 + * The subUChars field is declared as UChar[] not uint8_t[] to
1.172 + * guarantee alignment for UChars.
1.173 + */
1.174 + uint8_t *subChars;
1.175 +
1.176 + UConverterSharedData *sharedData; /* Pointer to the shared immutable part of the converter object */
1.177 +
1.178 + uint32_t options; /* options flags from UConverterOpen, may contain additional bits */
1.179 +
1.180 + 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 */
1.181 + UBool isCopyLocal; /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
1.182 + UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
1.183 +
1.184 + UBool useFallback;
1.185 + int8_t toULength; /* number of bytes in toUBytes */
1.186 + uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */
1.187 + uint32_t toUnicodeStatus; /* Used to internalize stream status information */
1.188 + int32_t mode;
1.189 + uint32_t fromUnicodeStatus;
1.190 +
1.191 + /*
1.192 + * More fromUnicode() status. Serves 3 purposes:
1.193 + * - keeps a lead surrogate between buffers (similar to toUBytes[])
1.194 + * - keeps a lead surrogate at the end of the stream,
1.195 + * which the framework handles as truncated input
1.196 + * - if the fromUnicode() implementation returns to the framework
1.197 + * (ucnv.c ucnv_fromUnicode()), then the framework calls the callback
1.198 + * for this code point
1.199 + */
1.200 + UChar32 fromUChar32;
1.201 +
1.202 + /*
1.203 + * value for ucnv_getMaxCharSize()
1.204 + *
1.205 + * usually simply copied from the static data, but ucnvmbcs.c modifies
1.206 + * the value depending on the converter type and options
1.207 + */
1.208 + int8_t maxBytesPerUChar;
1.209 +
1.210 + int8_t subCharLen; /* length of the codepage specific character sequence */
1.211 + int8_t invalidCharLength;
1.212 + int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */
1.213 +
1.214 + int8_t invalidUCharLength;
1.215 + int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */
1.216 +
1.217 + uint8_t subChar1; /* single-byte substitution character if different from subChar */
1.218 + UBool useSubChar1;
1.219 + char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */
1.220 + uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */
1.221 + UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */
1.222 +
1.223 + UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */
1.224 + UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */
1.225 +
1.226 + /* fields for conversion extension */
1.227 +
1.228 + /* store previous UChars/chars to continue partial matches */
1.229 + UChar32 preFromUFirstCP; /* >=0: partial match */
1.230 + UChar preFromU[UCNV_EXT_MAX_UCHARS];
1.231 + char preToU[UCNV_EXT_MAX_BYTES];
1.232 + int8_t preFromULength, preToULength; /* negative: replay */
1.233 + int8_t preToUFirstLength; /* length of first character */
1.234 +
1.235 + /* new fields for ICU 4.0 */
1.236 + UConverterCallbackReason toUCallbackReason; /* (*fromCharErrorBehaviour) reason, set when error is detected */
1.237 +};
1.238 +
1.239 +U_CDECL_END /* end of UConverter */
1.240 +
1.241 +#define CONVERTER_FILE_EXTENSION ".cnv"
1.242 +
1.243 +
1.244 +/**
1.245 + * Return the number of all converter names.
1.246 + * @param pErrorCode The error code
1.247 + * @return the number of all converter names
1.248 + */
1.249 +U_CFUNC uint16_t
1.250 +ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode);
1.251 +
1.252 +/**
1.253 + * Return the (n)th converter name in mixed case, or NULL
1.254 + * if there is none (typically, if the data cannot be loaded).
1.255 + * 0<=index<ucnv_io_countAvailableConverters().
1.256 + * @param n The number specifies which converter name to get
1.257 + * @param pErrorCode The error code
1.258 + * @return the (n)th converter name in mixed case, or NULL if there is none.
1.259 + */
1.260 +U_CFUNC const char *
1.261 +ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode);
1.262 +
1.263 +/**
1.264 + * Load a non-algorithmic converter.
1.265 + * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex).
1.266 + */
1.267 +U_CAPI UConverterSharedData *
1.268 +ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err);
1.269 +
1.270 +/**
1.271 + * Unload a non-algorithmic converter.
1.272 + * It must be sharedData->referenceCounter != ~0
1.273 + * and this function must be called inside umtx_lock(&cnvCacheMutex).
1.274 + */
1.275 +U_CAPI void
1.276 +ucnv_unload(UConverterSharedData *sharedData);
1.277 +
1.278 +/**
1.279 + * Swap ICU .cnv conversion tables. See udataswp.h.
1.280 + * @internal
1.281 + */
1.282 +U_CAPI int32_t U_EXPORT2
1.283 +ucnv_swap(const UDataSwapper *ds,
1.284 + const void *inData, int32_t length, void *outData,
1.285 + UErrorCode *pErrorCode);
1.286 +
1.287 +#endif
1.288 +
1.289 +#endif /* _UCNV_BLD */
