1/*
2**********************************************************************
3*   Copyright (C) 1999-2009 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 "ucnv_cnv.h"
27#include "ucnvmbcs.h"
28#include "ucnv_ext.h"
29#include "udataswp.h"
30
31/* size of the overflow buffers in UConverter, enough for escaping callbacks */
32#define UCNV_ERROR_BUFFER_LENGTH 32
33
34/* at most 4 bytes per substitution character (part of .cnv file format! see UConverterStaticData) */
35#define UCNV_MAX_SUBCHAR_LEN 4
36
37/* at most 8 bytes per character in toUBytes[] (UTF-8 uses up to 6) */
38#define UCNV_MAX_CHAR_LEN 8
39
40/* converter options bits */
41#define UCNV_OPTION_VERSION     0xf
42#define UCNV_OPTION_SWAP_LFNL   0x10
43
44#define UCNV_GET_VERSION(cnv) ((cnv)->options&UCNV_OPTION_VERSION)
45
46U_CDECL_BEGIN /* We must declare the following as 'extern "C"' so that if ucnv
47                 itself is compiled under C++, the linkage of the funcptrs will
48                 work.
49              */
50
51union UConverterTable {
52    UConverterMBCSTable mbcs;
53};
54
55typedef union UConverterTable UConverterTable;
56
57struct UConverterImpl;
58typedef struct UConverterImpl UConverterImpl;
59
60/** values for the unicodeMask */
61#define UCNV_HAS_SUPPLEMENTARY 1
62#define UCNV_HAS_SURROGATES    2
63
64typedef struct UConverterStaticData {   /* +offset: size */
65    uint32_t structSize;                /* +0: 4 Size of this structure */
66
67    char name
68      [UCNV_MAX_CONVERTER_NAME_LENGTH]; /* +4: 60  internal name of the converter- invariant chars */
69
70    int32_t codepage;               /* +64: 4 codepage # (now IBM-$codepage) */
71
72    int8_t platform;                /* +68: 1 platform of the converter (only IBM now) */
73    int8_t conversionType;          /* +69: 1 conversion type */
74
75    int8_t minBytesPerChar;         /* +70: 1 Minimum # bytes per char in this codepage */
76    int8_t maxBytesPerChar;         /* +71: 1 Maximum # bytes output per UChar in this codepage */
77
78    uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4  [note:  4 and 8 byte boundary] */
79    int8_t subCharLen;              /* +76: 1 */
80
81    uint8_t hasToUnicodeFallback;   /* +77: 1 UBool needs to be changed to UBool to be consistent across platform */
82    uint8_t hasFromUnicodeFallback; /* +78: 1 */
83    uint8_t unicodeMask;            /* +79: 1  bit 0: has supplementary  bit 1: has single surrogates */
84    uint8_t subChar1;               /* +80: 1  single-byte substitution character for IBM MBCS (0 if none) */
85    uint8_t reserved[19];           /* +81: 19 to round out the structure */
86                                    /* total size: 100 */
87} UConverterStaticData;
88
89/*
90 * Defines the UConverterSharedData struct,
91 * the immutable, shared part of UConverter.
92 */
93struct UConverterSharedData {
94    uint32_t structSize;            /* Size of this structure */
95    uint32_t referenceCounter;      /* used to count number of clients, 0xffffffff for static SharedData */
96
97    const void *dataMemory;         /* from udata_openChoice() - for cleanup */
98    void *table;                    /* Unused. This used to be a UConverterTable - Pointer to conversion data - see mbcs below */
99
100    const UConverterStaticData *staticData; /* pointer to the static (non changing) data. */
101
102    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 */
103  /*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
105    const UConverterImpl *impl;     /* vtable-style struct of mostly function pointers */
106
107    /*initial values of some members of the mutable part of object */
108    uint32_t toUnicodeStatus;
109
110    /*
111     * Shared data structures currently come in two flavors:
112     * - readonly for built-in algorithmic converters
113     * - allocated for MBCS, with a pointer to an allocated UConverterTable
114     *   which always has a UConverterMBCSTable
115     *
116     * To eliminate one allocation, I am making the UConverterMBCSTable
117     * a member of the shared data. It is the last member so that static
118     * definitions of UConverterSharedData work as before.
119     * The table field above also remains to avoid updating all static
120     * definitions, but is now unused.
121     *
122     * markus 2003-nov-07
123     */
124    UConverterMBCSTable mbcs;
125};
126
127/* Defines a UConverter, the lightweight mutable part the user sees */
128
129struct UConverter {
130    /*
131     * Error function pointer called when conversion issues
132     * occur during a ucnv_fromUnicode call
133     */
134    void (U_EXPORT2 *fromUCharErrorBehaviour) (const void *context,
135                                     UConverterFromUnicodeArgs *args,
136                                     const UChar *codeUnits,
137                                     int32_t length,
138                                     UChar32 codePoint,
139                                     UConverterCallbackReason reason,
140                                     UErrorCode *);
141    /*
142     * Error function pointer called when conversion issues
143     * occur during a ucnv_toUnicode call
144     */
145    void (U_EXPORT2 *fromCharErrorBehaviour) (const void *context,
146                                    UConverterToUnicodeArgs *args,
147                                    const char *codeUnits,
148                                    int32_t length,
149                                    UConverterCallbackReason reason,
150                                    UErrorCode *);
151
152    /*
153     * Pointer to additional data that depends on the converter type.
154     * Used by ISO 2022, SCSU, GB 18030 converters, possibly more.
155     */
156    void *extraInfo;
157
158    const void *fromUContext;
159    const void *toUContext;
160
161    /*
162     * Pointer to charset bytes for substitution string if subCharLen>0,
163     * or pointer to Unicode string (UChar *) if subCharLen<0.
164     * subCharLen==0 is equivalent to using a skip callback.
165     * If the pointer is !=subUChars then it is allocated with
166     * UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR bytes.
167     * The subUChars field is declared as UChar[] not uint8_t[] to
168     * guarantee alignment for UChars.
169     */
170    uint8_t *subChars;
171
172    UConverterSharedData *sharedData;   /* Pointer to the shared immutable part of the converter object */
173
174    uint32_t options; /* options flags from UConverterOpen, may contain additional bits */
175
176    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 */
177    UBool isCopyLocal;  /* TRUE if UConverter is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
178    UBool isExtraLocal; /* TRUE if extraInfo is not owned and not released in ucnv_close() (stack-allocated, safeClone(), etc.) */
179
180    UBool  useFallback;
181    int8_t toULength;                   /* number of bytes in toUBytes */
182    uint8_t toUBytes[UCNV_MAX_CHAR_LEN-1];/* more "toU status"; keeps the bytes of the current character */
183    uint32_t toUnicodeStatus;           /* Used to internalize stream status information */
184    int32_t mode;
185    uint32_t fromUnicodeStatus;
186
187    /*
188     * More fromUnicode() status. Serves 3 purposes:
189     * - keeps a lead surrogate between buffers (similar to toUBytes[])
190     * - keeps a lead surrogate at the end of the stream,
191     *   which the framework handles as truncated input
192     * - if the fromUnicode() implementation returns to the framework
193     *   (ucnv.c ucnv_fromUnicode()), then the framework calls the callback
194     *   for this code point
195     */
196    UChar32 fromUChar32;
197
198    /*
199     * value for ucnv_getMaxCharSize()
200     *
201     * usually simply copied from the static data, but ucnvmbcs.c modifies
202     * the value depending on the converter type and options
203     */
204    int8_t maxBytesPerUChar;
205
206    int8_t subCharLen;                  /* length of the codepage specific character sequence */
207    int8_t invalidCharLength;
208    int8_t charErrorBufferLength;       /* number of valid bytes in charErrorBuffer */
209
210    int8_t invalidUCharLength;
211    int8_t UCharErrorBufferLength;      /* number of valid UChars in charErrorBuffer */
212
213    uint8_t subChar1;                                   /* single-byte substitution character if different from subChar */
214    UBool useSubChar1;
215    char invalidCharBuffer[UCNV_MAX_CHAR_LEN];          /* bytes from last error/callback situation */
216    uint8_t charErrorBuffer[UCNV_ERROR_BUFFER_LENGTH];  /* codepage output from Error functions */
217    UChar subUChars[UCNV_MAX_SUBCHAR_LEN/U_SIZEOF_UCHAR]; /* see subChars documentation */
218
219    UChar invalidUCharBuffer[U16_MAX_LENGTH];           /* UChars from last error/callback situation */
220    UChar UCharErrorBuffer[UCNV_ERROR_BUFFER_LENGTH];   /* unicode output from Error functions */
221
222    /* fields for conversion extension */
223
224    /* store previous UChars/chars to continue partial matches */
225    UChar32 preFromUFirstCP;                /* >=0: partial match */
226    UChar preFromU[UCNV_EXT_MAX_UCHARS];
227    char preToU[UCNV_EXT_MAX_BYTES];
228    int8_t preFromULength, preToULength;    /* negative: replay */
229    int8_t preToUFirstLength;               /* length of first character */
230
231    /* new fields for ICU 4.0 */
232    UConverterCallbackReason toUCallbackReason; /* (*fromCharErrorBehaviour) reason, set when error is detected */
233};
234
235U_CDECL_END /* end of UConverter */
236
237#define CONVERTER_FILE_EXTENSION ".cnv"
238
239
240/**
241 * Return the number of all converter names.
242 * @param pErrorCode The error code
243 * @return the number of all converter names
244 */
245U_CFUNC uint16_t
246ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode);
247
248/**
249 * Return the (n)th converter name in mixed case, or NULL
250 * if there is none (typically, if the data cannot be loaded).
251 * 0<=index<ucnv_io_countAvailableConverters().
252 * @param n The number specifies which converter name to get
253 * @param pErrorCode The error code
254 * @return the (n)th converter name in mixed case, or NULL if there is none.
255 */
256U_CFUNC const char *
257ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode);
258
259/**
260 * Load a non-algorithmic converter.
261 * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex).
262 */
263UConverterSharedData *
264ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err);
265
266/**
267 * Unload a non-algorithmic converter.
268 * It must be sharedData->referenceCounter != ~0
269 * and this function must be called inside umtx_lock(&cnvCacheMutex).
270 */
271void
272ucnv_unload(UConverterSharedData *sharedData);
273
274/**
275 * Swap ICU .cnv conversion tables. See udataswp.h.
276 * @internal
277 */
278U_CAPI int32_t U_EXPORT2
279ucnv_swap(const UDataSwapper *ds,
280          const void *inData, int32_t length, void *outData,
281          UErrorCode *pErrorCode);
282
283#endif
284
285#endif /* _UCNV_BLD */
286