[apple/icu.git] / icuSources / common / ucnv_bld.h

/*
**********************************************************************
*   Copyright (C) 1999-2006, International Business Machines
*   Corporation and others.  All Rights Reserved.
**********************************************************************
*
*
*  ucnv_bld.h:
*  Contains internal data structure definitions
* Created by Bertrand A. Damiba
*
*   Change history:
*
*   06/29/2000  helena      Major rewrite of the callback APIs.
*/

#ifndef UCNV_BLD_H
#define UCNV_BLD_H

#include "unicode/utypes.h"

#if !UCONFIG_NO_CONVERSION

#include "unicode/ucnv.h"
#include "unicode/ucnv_err.h"
#include "ucnv_cnv.h"
#include "ucnvmbcs.h"
#include "ucnv_ext.h"
#include "udataswp.h"

/* size of the overflow buffers in UConverter, enough for escaping callbacks */
#define UCNV_ERROR_BUFFER_LENGTH 32

/* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */
#define UCNV_MAX_SUBCHAR_LEN 4

/* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */
#define UCNV_MAX_CHAR_LEN 8

/* converter options bits */
#define UCNV_OPTION_VERSION     0xf
#define UCNV_OPTION_SWAP_LFNL   0x10


U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv
                 itself is compiled under C++, the linkage of the funcptrs will
                 work.
              */

union UConverterTable {
    UConverterMBCSTable mbcs;
};

typedef union UConverterTable UConverterTable;

struct UConverterImpl;
typedef struct UConverterImpl UConverterImpl;

/** values for the unicodeMask */
#define UCNV_HAS_SUPPLEMENTARY 1
#define UCNV_HAS_SURROGATES    2

typedef struct UConverterStaticData {   /* +offset: size */
    uint32_t structSize;                /* +0: 4 Size of this structure */
    
    char name 
      [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60  internal name of the converter- invariant chars */

    int32_t codepage;               /* +64: 4 codepage # (now IBM-$codepage) */

    int8_t platform;                /* +68: 1 platform of the converter (only IBM now) */
    int8_t conversionType;          /* +69: 1 conversion type */

    int8_t minBytesPerChar;         /* +70: 1 Minimum # bytes per char in this codepage */
    int8_t maxBytesPerChar;         /* +71: 1 Maximum # bytes output per UChar in this codepage */

    uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4  [note:  4 and 8 byte boundary] */
    int8_t subCharLen;              /* +76: 1 */
    
    uint8_t hasToUnicodeFallback;   /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */
    uint8_t hasFromUnicodeFallback; /* +78: 1 */
    uint8_t unicodeMask;            /* +79: 1  bit 0: has supplementary  bit 1: has single surrogates */
    uint8_t subChar1;               /* +80: 1  single-byte substitution character for IBM MBCS (0 if none) */
    uint8_t reserved[19];           /* +81: 19 to round out the structure */
                                    /* total size: 100 */
} UConverterStaticData;

/*
 * Defines the UConverterSharedData struct,
 * the immutable, shared part of UConverter.
 */
struct UConverterSharedData {
    uint32_t structSize;            /* Size of this structure */
    uint32_t referenceCounter;      /* used to count number of clients, 0xffffffff for static SharedData */

    const void *dataMemory;         /* from udata_openChoice() - for cleanup */
    void *table;                    /* Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */

    const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */

    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 */
  /*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.   */

    const UConverterImpl *impl;     /* vtable-style struct of mostly function pointers */

    /*initial values of some members of the mutable part of object */
    uint32_t toUnicodeStatus;

    /*
     * Shared data structures currently come in two flavors:
     * - readonly for built-in algorithmic converters
     * - allocated for MBCS, with a pointer to an allocated UConverterTable
     *   which always has a UConverterMBCSTable
     *
     * To eliminate one allocation, I am making the UConverterMBCSTable
     * a member of the shared data. It is the last member so that static
     * definitions of UConverterSharedData work as before.
     * The table field above also remains to avoid updating all static
     * definitions, but is now unused.
     *
     * markus 2003-nov-07
     */
    UConverterMBCSTable mbcs;
};

/* Defines a UConverter, the lightweight mutable part the user sees */

struct UConverter {
    /*
     * Error function pointer called when conversion issues
     * occur during a ucnv_fromUnicode call
     */
    void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context,
                                     UConverterFromUnicodeArgs *args,
                                     const UChar *codeUnits,
                                     int32_t length,
                                     UChar32 codePoint,
                                     UConverterCallbackReason reason,
                                     UErrorCode *);
    /*
     * Error function pointer called when conversion issues
     * occur during a ucnv_toUnicode call
     */
    void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context,
                                    UConverterToUnicodeArgs *args,
                                    const char *codeUnits,
                                    int32_t length,
                                    UConverterCallbackReason reason,
                                    UErrorCode *);

    /*
     * Pointer to additional data that depends on the converter type.
     * Used by ISO 2022, SCSU, GB 18030 converters, possibly more.
     */
    void *extraInfo;

    const void *fromUContext;
    const void *toUContext;

    /*
     * Pointer to charset bytes for substitution string if subCharLen>0,
     * or pointer to Unicode string (UChar *) if subCharLen<0.
     * subCharLen==0 is equivalent to using a skip callback.
     * If the pointer is !=subUChars then it is allocated with
     * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes.
     * The subUChars field is declared as UChar[] not uint8_t[] to
     * guarantee alignment for UChars.
     */
    uint8_t *subChars;

    UConverterSharedData *sharedData;   /* Pointer to the shared immutable part of the converter object */

    uint32_t options; /* options flags from UConverterOpen, may contain additional bits */

    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 */
    UBool isCopyLocal;  /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
    UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */

    UBool  useFallback;
    int8_t toULength;                   /* number of bytes in toUBytes */
    uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */
    uint32_t toUnicodeStatus;           /* Used to internalize stream status information */
    int32_t mode;
    uint32_t fromUnicodeStatus;

    /*
     * More fromUnicode() status. Serves 3 purposes:
     * - keeps a lead surrogate between buffers (similar to toUBytes[])
     * - keeps a lead surrogate at the end of the stream,
     *   which the framework handles as truncated input
     * - if the fromUnicode() implementation returns to the framework
     *   (ucnv.c ucnv_fromUnicode()), then the framework calls the callback
     *   for this code point
     */
    UChar32 fromUChar32;

    /*
     * value for ucnv_getMaxCharSize()
     *
     * usually simply copied from the static data, but ucnvmbcs.c modifies
     * the value depending on the converter type and options
     */
    int8_t maxBytesPerUChar;

    int8_t subCharLen;                  /* length of the codepage specific character sequence */
    int8_t invalidCharLength;
    int8_t charErrorBufferLength;       /* number of valid bytes in charErrorBuffer */

    int8_t invalidUCharLength;
    int8_t UCharErrorBufferLength;      /* number of valid UChars in charErrorBuffer */

    uint8_t subChar1;                                   /* single-byte substitution character if different from subChar */
    UBool useSubChar1;
    char invalidCharBuffer[UCNV_MAX_CHAR_LEN];          /* bytes from last error/callback situation */
    uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH];  /* codepage output from Error functions */
    UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */

    UChar invalidUCharBuffer[U16_MAX_LENGTH];           /* UChars from last error/callback situation */
    UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH];   /* unicode output from Error functions */

    /* fields for conversion extension */

    /* store previous UChars/chars to continue partial matches */
    UChar32 preFromUFirstCP;                /* >=0: partial match */
    UChar preFromU[UCNV_EXT_MAX_UCHARS];
    char preToU[UCNV_EXT_MAX_BYTES];
    int8_t preFromULength, preToULength;    /* negative: replay */
    int8_t preToUFirstLength;               /* length of first character */
};

U_CDECL_END /* end of UConverter */

#define CONVERTER_FILE_EXTENSION ".cnv"


/**
 * Return the number of all converter names.
 * @param pErrorCode The error code
 * @return the number of all converter names
 */
U_CFUNC uint16_t
ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode);

/**
 * Return the (n)th converter name in mixed case, or NULL
 * if there is none (typically, if the data cannot be loaded).
 * 0<=index<ucnv_io_countAvailableConverters().
 * @param n The number specifies which converter name to get
 * @param pErrorCode The error code
 * @return the (n)th converter name in mixed case, or NULL if there is none.
 */
U_CFUNC const char *
ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode);

/**
 * Load a non-algorithmic converter.
 * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex).
 */
UConverterSharedData *
ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err);

/**
 * Unload a non-algorithmic converter.
 * It must be sharedData->referenceCounter != ~0
 * and this function must be called inside umtx_lock(&cnvCacheMutex).
 */
void
ucnv_unload(UConverterSharedData *sharedData);

/**
 * Swap ICU .cnv conversion tables. See udataswp.h.
 * @internal
 */
U_CAPI int32_t U_EXPORT2
ucnv_swap(const UDataSwapper *ds,
          const void *inData, int32_t length, void *outData,
          UErrorCode *pErrorCode);

#endif

#endif /* _UCNV_BLD */
Commit	Line	Data
b75a7d8f A	1	/*
b75a7d8f A	2	**********************************************************************
73c04bcf	3	* Copyright (C) 1999-2006, International Business Machines
b75a7d8f A	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"
374ca955 A	21
	22	#if !UCONFIG_NO_CONVERSION
	23
b75a7d8f A	24	#include "unicode/ucnv.h"
b75a7d8f A	25	#include "unicode/ucnv_err.h"
374ca955 A	26	#include "ucnv_cnv.h"
	27	#include "ucnvmbcs.h"
	28	#include "ucnv_ext.h"
	29	#include "udataswp.h"
b75a7d8f A	30
	31	/* size of the overflow buffers in UConverter, enough for escaping callbacks */
	32	#define UCNV_ERROR_BUFFER_LENGTH 32
	33
374ca955	34	/* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */
b75a7d8f A	35	#define UCNV_MAX_SUBCHAR_LEN 4
b75a7d8f A	36
374ca955 A	37	/* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */
	38	#define UCNV_MAX_CHAR_LEN 8
	39
b75a7d8f A	40	/* converter options bits */
	41	#define UCNV_OPTION_VERSION 0xf
	42	#define UCNV_OPTION_SWAP_LFNL 0x10
	43
	44
	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
	47	work.
	48	*/
	49
374ca955 A	50	union UConverterTable {
	51	UConverterMBCSTable mbcs;
	52	};
	53
b75a7d8f A	54	typedef union UConverterTable UConverterTable;
	55
	56	struct UConverterImpl;
	57	typedef struct UConverterImpl UConverterImpl;
	58
	59	/** values for the unicodeMask */
	60	#define UCNV_HAS_SUPPLEMENTARY 1
	61	#define UCNV_HAS_SURROGATES 2
	62
	63	typedef struct UConverterStaticData { /* +offset: size */
	64	uint32_t structSize; /* +0: 4 Size of this structure */
	65
	66	char name
	67	[UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60 internal name of the converter- invariant chars */
	68
	69	int32_t codepage; /* +64: 4 codepage # (now IBM-$codepage) */
	70
	71	int8_t platform; /* +68: 1 platform of the converter (only IBM now) */
	72	int8_t conversionType; /* +69: 1 conversion type */
	73
	74	int8_t minBytesPerChar; /* +70: 1 Minimum # bytes per char in this codepage */
374ca955	75	int8_t maxBytesPerChar; /* +71: 1 Maximum # bytes output per UChar in this codepage */
b75a7d8f A	76
	77	uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4 [note: 4 and 8 byte boundary] */
	78	int8_t subCharLen; /* +76: 1 */
	79
	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 */
	85	/* total size: 100 */
	86	} UConverterStaticData;
	87
	88	/*
	89	* Defines the UConverterSharedData struct,
	90	* the immutable, shared part of UConverter.
	91	*/
	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 */
	95
	96	const void dataMemory; / from udata_openChoice() - for cleanup */
374ca955	97	void table; / Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */
b75a7d8f A	98
	99	const UConverterStaticData staticData; / pointer to the static (non changing) data. */
	100
	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. /
	103
	104	const UConverterImpl impl; / vtable-style struct of mostly function pointers */
	105
	106	/initial values of some members of the mutable part of object /
	107	uint32_t toUnicodeStatus;
b75a7d8f	108
374ca955 A	109	/*
	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
	114	*
	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.
	120	*
	121	* markus 2003-nov-07
	122	*/
	123	UConverterMBCSTable mbcs;
	124	};
b75a7d8f A	125
	126	/* Defines a UConverter, the lightweight mutable part the user sees */
	127
	128	struct UConverter {
	129	/*
	130	* Error function pointer called when conversion issues
	131	* occur during a ucnv_fromUnicode call
	132	*/
	133	void (U_EXPORT2 fromUCharErrorBehaviour) (const void context,
	134	UConverterFromUnicodeArgs *args,
	135	const UChar *codeUnits,
	136	int32_t length,
	137	UChar32 codePoint,
	138	UConverterCallbackReason reason,
	139	UErrorCode *);
	140	/*
	141	* Error function pointer called when conversion issues
374ca955	142	* occur during a ucnv_toUnicode call
b75a7d8f A	143	*/
	144	void (U_EXPORT2 fromCharErrorBehaviour) (const void context,
	145	UConverterToUnicodeArgs *args,
	146	const char *codeUnits,
	147	int32_t length,
	148	UConverterCallbackReason reason,
	149	UErrorCode *);
	150
	151	/*
	152	* Pointer to additional data that depends on the converter type.
	153	* Used by ISO 2022, SCSU, GB 18030 converters, possibly more.
	154	*/
	155	void *extraInfo;
	156
	157	const void *fromUContext;
	158	const void *toUContext;
	159
73c04bcf A	160	/*
	161	* Pointer to charset bytes for substitution string if subCharLen>0,
	162	* or pointer to Unicode string (UChar *) if subCharLen<0.
	163	* subCharLen==0 is equivalent to using a skip callback.
	164	* If the pointer is !=subUChars then it is allocated with
	165	* UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes.
	166	* The subUChars field is declared as UChar[] not uint8_t[] to
	167	* guarantee alignment for UChars.
	168	*/
	169	uint8_t *subChars;
	170
b75a7d8f A	171	UConverterSharedData sharedData; / Pointer to the shared immutable part of the converter object */
	172
	173	uint32_t options; /* options flags from UConverterOpen, may contain additional bits */
	174
	175	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 */
	176	UBool isCopyLocal; /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
	177	UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
	178
	179	UBool useFallback;
	180	int8_t toULength; /* number of bytes in toUBytes */
374ca955	181	uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */
b75a7d8f A	182	uint32_t toUnicodeStatus; /* Used to internalize stream status information */
	183	int32_t mode;
	184	uint32_t fromUnicodeStatus;
374ca955 A	185
	186	/*
	187	* More fromUnicode() status. Serves 3 purposes:
	188	* - keeps a lead surrogate between buffers (similar to toUBytes[])
	189	* - keeps a lead surrogate at the end of the stream,
	190	* which the framework handles as truncated input
	191	* - if the fromUnicode() implementation returns to the framework
	192	* (ucnv.c ucnv_fromUnicode()), then the framework calls the callback
	193	* for this code point
	194	*/
	195	UChar32 fromUChar32;
	196
	197	/*
	198	* value for ucnv_getMaxCharSize()
	199	*
	200	* usually simply copied from the static data, but ucnvmbcs.c modifies
	201	* the value depending on the converter type and options
	202	*/
	203	int8_t maxBytesPerUChar;
b75a7d8f A	204
	205	int8_t subCharLen; /* length of the codepage specific character sequence */
	206	int8_t invalidCharLength;
	207	int8_t charErrorBufferLength; /* number of valid bytes in charErrorBuffer */
	208
	209	int8_t invalidUCharLength;
	210	int8_t UCharErrorBufferLength; /* number of valid UChars in charErrorBuffer */
	211
	212	uint8_t subChar1; /* single-byte substitution character if different from subChar */
374ca955	213	UBool useSubChar1;
374ca955	214	char invalidCharBuffer[UCNV_MAX_CHAR_LEN]; /* bytes from last error/callback situation */
b75a7d8f	215	uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* codepage output from Error functions */
73c04bcf	216	UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */
b75a7d8f	217
374ca955	218	UChar invalidUCharBuffer[U16_MAX_LENGTH]; /* UChars from last error/callback situation */
b75a7d8f A	219	UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH]; /* unicode output from Error functions */
b75a7d8f A	220
374ca955 A	221	/* fields for conversion extension */
	222
	223	/* store previous UChars/chars to continue partial matches */
	224	UChar32 preFromUFirstCP; /* >=0: partial match */
	225	UChar preFromU[UCNV_EXT_MAX_UCHARS];
	226	char preToU[UCNV_EXT_MAX_BYTES];
	227	int8_t preFromULength, preToULength; /* negative: replay */
	228	int8_t preToUFirstLength; /* length of first character */
b75a7d8f A	229	};
	230
	231	U_CDECL_END /* end of UConverter */
	232
374ca955 A	233	#define CONVERTER_FILE_EXTENSION ".cnv"
374ca955 A	234
73c04bcf A	235
	236	/**
	237	* Return the number of all converter names.
	238	* @param pErrorCode The error code
	239	* @return the number of all converter names
	240	*/
	241	U_CFUNC uint16_t
	242	ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode);
	243
	244	/**
	245	* Return the (n)th converter name in mixed case, or NULL
	246	* if there is none (typically, if the data cannot be loaded).
	247	* 0<=index<ucnv_io_countAvailableConverters().
	248	* @param n The number specifies which converter name to get
	249	* @param pErrorCode The error code
	250	* @return the (n)th converter name in mixed case, or NULL if there is none.
	251	*/
	252	U_CFUNC const char *
	253	ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode);
	254
374ca955 A	255	/**
	256	* Load a non-algorithmic converter.
	257	* If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex).
	258	*/
	259	UConverterSharedData *
	260	ucnv_load(UConverterLoadArgs pArgs, UErrorCode err);
	261
	262	/**
	263	* Unload a non-algorithmic converter.
	264	* It must be sharedData->referenceCounter != ~0
	265	* and this function must be called inside umtx_lock(&cnvCacheMutex).
	266	*/
	267	void
	268	ucnv_unload(UConverterSharedData *sharedData);
b75a7d8f	269
374ca955 A	270	/**
	271	* Swap ICU .cnv conversion tables. See udataswp.h.
	272	* @internal
	273	*/
	274	U_CAPI int32_t U_EXPORT2
	275	ucnv_swap(const UDataSwapper *ds,
	276	const void inData, int32_t length, void outData,
	277	UErrorCode *pErrorCode);
b75a7d8f	278
374ca955	279	#endif
b75a7d8f A	280
b75a7d8f A	281	#endif /* _UCNV_BLD */