1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ****************************************************************************** |
5 | * |
6 | * Copyright (C) 2000-2013, International Business Machines |
7 | * Corporation and others. All Rights Reserved. |
8 | * |
9 | ****************************************************************************** |
10 | * file name: ucnvmbcs.h |
11 | * encoding: UTF-8 |
12 | * tab size: 8 (not used) |
13 | * indentation:4 |
14 | * |
15 | * created on: 2000jul07 |
16 | * created by: Markus W. Scherer |
17 | */ |
18 | |
19 | #ifndef __UCNVMBCS_H__ |
20 | #define __UCNVMBCS_H__ |
21 | |
22 | #include "unicode/utypes.h" |
23 | |
24 | #if !UCONFIG_NO_CONVERSION |
25 | |
26 | #include "unicode/ucnv.h" |
27 | #include "ucnv_cnv.h" |
28 | #include "ucnv_ext.h" |
29 | |
30 | /** |
31 | * ICU conversion (.cnv) data file structure, following the usual UDataInfo |
32 | * header. |
33 | * |
34 | * Format version: 6.2 |
35 | * |
36 | * struct UConverterStaticData -- struct containing the converter name, IBM CCSID, |
37 | * min/max bytes per character, etc. |
38 | * see ucnv_bld.h |
39 | * |
40 | * -------------------- |
41 | * |
42 | * The static data is followed by conversionType-specific data structures. |
43 | * At the moment, there are only variations of MBCS converters. They all have |
44 | * the same toUnicode structures, while the fromUnicode structures for SBCS |
45 | * differ from those for other MBCS-style converters. |
46 | * |
47 | * _MBCSHeader.version 5 is optional and not backward-compatible |
48 | * (as usual for changes in the major version field). |
49 | * |
50 | * Versions 5.m work like versions 4.m except: |
51 | * - The _MBCSHeader has variable length (and is always longer than in version 4). |
52 | * See the struct _MBCSHeader further description below. |
53 | * - There is a set of flags which indicate further incompatible changes. |
54 | * (Reader code must reject the file if it does not recognize them all.) |
55 | * - In particular, one of these flags indicates that most of the fromUnicode |
56 | * data is missing and must be reconstituted from the toUnicode data |
57 | * and from the utf8Friendly mbcsIndex at load time. |
58 | * (This only works with a utf8Friendly table.) |
59 | * In this case, makeconv may increase maxFastUChar automatically to U+FFFF. |
60 | * |
61 | * The first of these versions is 5.3, which is like 4.3 except for the differences above. |
62 | * |
63 | * When possible, makeconv continues to generate version 4.m files. |
64 | * |
65 | * _MBCSHeader.version 5.4/4.4 supports "good one-way" mappings (|4) |
66 | * in the extension tables (fromUTableValues bit 30). See ucnv_ext.h for details. |
67 | * |
68 | * _MBCSHeader.version 4.3 optionally modifies the fromUnicode data structures |
69 | * slightly and optionally adds a table for conversion to MBCS (non-SBCS) |
70 | * charsets. |
71 | * |
72 | * The modifications are to make the data utf8Friendly. Not every 4.3 file |
73 | * file contains utf8Friendly data. |
74 | * It is utf8Friendly if _MBCSHeader.version[2]!=0. |
75 | * In this case, the data structures are utf8Friendly up to the code point |
76 | * maxFastUChar=((_MBCSHeader.version[2]<<8)|0xff) |
77 | * |
78 | * A utf8Friendly file has fromUnicode stage 3 entries for code points up to |
79 | * maxFastUChar allocated in blocks of 64 for indexing with the 6 bits from |
80 | * a UTF-8 trail byte. ASCII is allocated linearly with 128 contiguous entries. |
81 | * |
82 | * In addition, a utf8Friendly MBCS file contains an additional |
83 | * uint16_t mbcsIndex[(maxFastUChar+1)>>6]; |
84 | * which replaces the stage 1 and 2 tables for indexing with bits from the |
85 | * UTF-8 lead byte and middle trail byte. Unlike the older MBCS stage 2 table, |
86 | * the mbcsIndex does not contain roundtrip flags. Therefore, all fallbacks |
87 | * from code points up to maxFastUChar (and roundtrips to 0x00) are moved to |
88 | * the extension data structure. This also allows for faster roundtrip |
89 | * conversion from UTF-16. |
90 | * |
91 | * SBCS files do not contain an additional sbcsIndex[] array because the |
92 | * proportional size increase would be noticeable, but the runtime |
93 | * code builds one for the code point range for which the runtime conversion |
94 | * code is optimized. |
95 | * |
96 | * For SBCS, maxFastUChar should be at least U+0FFF. The initial makeconv |
97 | * implementation sets it to U+1FFF. Because the sbcsIndex is not stored in |
98 | * the file, a larger maxFastUChar only affects stage 3 block allocation size |
99 | * and is free in empty blocks. (Larger blocks with sparse contents cause larger |
100 | * files.) U+1FFF includes almost all of the small scripts. |
101 | * U+0FFF covers UTF-8 two-byte sequences and three-byte sequences starting with |
102 | * 0xe0. This includes most scripts with legacy SBCS charsets. |
103 | * The initial runtime implementation using 4.3 files only builds an sbcsIndex |
104 | * for code points up to U+0FFF. |
105 | * |
106 | * For MBCS, maxFastUChar should be at least U+D7FF (=initial value). |
107 | * This boundary is convenient because practically all of the commonly used |
108 | * characters are below it, and because it is the boundary to surrogate |
109 | * code points, above which special handling is necessary anyway. |
110 | * (Surrogate pair assembly for UTF-16, validity checking for UTF-8.) |
111 | * |
112 | * maxFastUChar could be up to U+FFFF to cover the whole BMP, which could be |
113 | * useful especially for conversion from UTF-8 when the input can be assumed |
114 | * to be valid, because the surrogate range would then not have to be |
115 | * checked. |
116 | * (With maxFastUChar=0xffff, makeconv would have to check for mbcsIndex value |
117 | * overflow because with the all-unassigned block 0 and nearly full mappings |
118 | * from the BMP it is theoretically possible that an index into stage 3 |
119 | * exceeds 16 bits.) |
120 | * |
121 | * _MBCSHeader.version 4.2 adds an optional conversion extension data structure. |
122 | * If it is present, then an ICU version reading header versions 4.0 or 4.1 |
123 | * will be able to use the base table and ignore the extension. |
124 | * |
125 | * The unicodeMask in the static data is part of the base table data structure. |
126 | * Especially, the UCNV_HAS_SUPPLEMENTARY flag determines the length of the |
127 | * fromUnicode stage 1 array. |
128 | * The static data unicodeMask refers only to the base table's properties if |
129 | * a base table is included. |
130 | * In an extension-only file, the static data unicodeMask is 0. |
131 | * The extension data indexes have a separate field with the unicodeMask flags. |
132 | * |
133 | * MBCS-style data structure following the static data. |
134 | * Offsets are counted in bytes from the beginning of the MBCS header structure. |
135 | * Details about usage in comments in ucnvmbcs.c. |
136 | * |
137 | * struct _MBCSHeader (see the definition in this header file below) |
138 | * contains 32-bit fields as follows: |
139 | * 8 values: |
140 | * 0 uint8_t[4] MBCS version in UVersionInfo format (currently 4.3.x.0) |
141 | * 1 uint32_t countStates |
142 | * 2 uint32_t countToUFallbacks |
143 | * 3 uint32_t offsetToUCodeUnits |
144 | * 4 uint32_t offsetFromUTable |
145 | * 5 uint32_t offsetFromUBytes |
146 | * 6 uint32_t flags, bits: |
147 | * 31.. 8 offsetExtension -- _MBCSHeader.version 4.2 (ICU 2.8) and higher |
148 | * 0 for older versions and if |
149 | * there is not extension structure |
150 | * 7.. 0 outputType |
151 | * 7 uint32_t fromUBytesLength -- _MBCSHeader.version 4.1 (ICU 2.4) and higher |
152 | * counts bytes in fromUBytes[] |
153 | * |
154 | * New and required in version 5: |
155 | * 8 uint32_t options, bits: |
156 | * 31..16 reserved for flags that can be added without breaking |
157 | * backward compatibility |
158 | * 15.. 6 reserved for flags whose addition will break |
159 | * backward compatibility |
160 | * 6 MBCS_OPT_FROM_U -- if set, |
161 | * then most of the fromUnicode data is omitted; |
162 | * fullStage2Length is present and the missing |
163 | * bottom part of stage 2 must be reconstituted from |
164 | * the toUnicode data; |
165 | * stage 3 is missing completely as well; |
166 | * not used for SBCS tables |
167 | * 5.. 0 length of the _MBCSHeader (number of uint32_t) |
168 | * |
169 | * New and optional in version 5: |
170 | * 9 uint32_t fullStage2Length: used if MBCS_OPT_FROM_U is set |
171 | * specifies the full length of stage 2 |
172 | * including the omitted part |
173 | * |
174 | * if(outputType==MBCS_OUTPUT_EXT_ONLY) { |
175 | * -- base table name for extension-only table |
176 | * char baseTableName[variable]; -- with NUL plus padding for 4-alignment |
177 | * |
178 | * -- all _MBCSHeader fields except for version and flags are 0 |
179 | * } else { |
180 | * -- normal base table with optional extension |
181 | * |
182 | * int32_t stateTable[countStates][256]; |
183 | * |
184 | * struct _MBCSToUFallback { (fallbacks are sorted by offset) |
185 | * uint32_t offset; |
186 | * UChar32 codePoint; |
187 | * } toUFallbacks[countToUFallbacks]; |
188 | * |
189 | * uint16_t unicodeCodeUnits[(offsetFromUTable-offsetToUCodeUnits)/2]; |
190 | * (padded to an even number of units) |
191 | * |
192 | * -- stage 1 tables |
193 | * if(staticData.unicodeMask&UCNV_HAS_SUPPLEMENTARY) { |
194 | * -- stage 1 table for all of Unicode |
195 | * uint16_t fromUTable[0x440]; (32-bit-aligned) |
196 | * } else { |
197 | * -- BMP-only tables have a smaller stage 1 table |
198 | * uint16_t fromUTable[0x40]; (32-bit-aligned) |
199 | * } |
200 | * |
201 | * -- stage 2 tables |
202 | * length determined by top of stage 1 and bottom of stage 3 tables |
203 | * if(outputType==MBCS_OUTPUT_1) { |
204 | * -- SBCS: pure indexes |
205 | * uint16_t stage 2 indexes[?]; |
206 | * } else { |
207 | * -- DBCS, MBCS, EBCDIC_STATEFUL, ...: roundtrip flags and indexes |
208 | * uint32_t stage 2 flags and indexes[?]; |
209 | * if(options&MBCS_OPT_NO_FROM_U) { |
210 | * stage 2 really has length fullStage2Length |
211 | * and the omitted lower part must be reconstituted from |
212 | * the toUnicode data |
213 | * } |
214 | * } |
215 | * |
216 | * -- stage 3 tables with byte results |
217 | * if(outputType==MBCS_OUTPUT_1) { |
218 | * -- SBCS: each 16-bit result contains flags and the result byte, see ucnvmbcs.c |
219 | * uint16_t fromUBytes[fromUBytesLength/2]; |
220 | * } else if(!(options&MBCS_OPT_NO_FROM_U)) { |
221 | * -- DBCS, MBCS, EBCDIC_STATEFUL, ... 2/3/4 bytes result, see ucnvmbcs.c |
222 | * uint8_t fromUBytes[fromUBytesLength]; or |
223 | * uint16_t fromUBytes[fromUBytesLength/2]; or |
224 | * uint32_t fromUBytes[fromUBytesLength/4]; |
225 | * } else { |
226 | * fromUBytes[] must be reconstituted from the toUnicode data |
227 | * } |
228 | * |
229 | * -- optional utf8Friendly mbcsIndex -- _MBCSHeader.version 4.3 (ICU 3.8) and higher |
230 | * if(outputType!=MBCS_OUTPUT_1 && |
231 | * _MBCSHeader.version[1]>=3 && |
232 | * (maxFastUChar=_MBCSHeader.version[2])!=0 |
233 | * ) { |
234 | * maxFastUChar=(maxFastUChar<<8)|0xff; |
235 | * uint16_t mbcsIndex[(maxFastUChar+1)>>6]; |
236 | * } |
237 | * } |
238 | * |
239 | * -- extension table, details see ucnv_ext.h |
240 | * int32_t indexes[>=32]; ... |
241 | */ |
242 | |
243 | /* MBCS converter data and state -------------------------------------------- */ |
244 | |
245 | enum { |
246 | MBCS_MAX_STATE_COUNT=128 |
247 | }; |
248 | |
249 | /** |
250 | * MBCS action codes for conversions to Unicode. |
251 | * These values are in bits 23..20 of the state table entries. |
252 | */ |
253 | enum { |
254 | MBCS_STATE_VALID_DIRECT_16, |
255 | MBCS_STATE_VALID_DIRECT_20, |
256 | |
257 | MBCS_STATE_FALLBACK_DIRECT_16, |
258 | MBCS_STATE_FALLBACK_DIRECT_20, |
259 | |
260 | MBCS_STATE_VALID_16, |
261 | MBCS_STATE_VALID_16_PAIR, |
262 | |
263 | MBCS_STATE_UNASSIGNED, |
264 | MBCS_STATE_ILLEGAL, |
265 | |
266 | MBCS_STATE_CHANGE_ONLY |
267 | }; |
268 | |
269 | /* Macros for state table entries */ |
270 | #define MBCS_ENTRY_TRANSITION(state, offset) (int32_t)(((int32_t)(state)<<24L)|(offset)) |
271 | #define MBCS_ENTRY_TRANSITION_SET_OFFSET(entry, offset) (int32_t)(((entry)&0xff000000)|(offset)) |
272 | #define MBCS_ENTRY_TRANSITION_ADD_OFFSET(entry, offset) (int32_t)((entry)+(offset)) |
273 | |
274 | #define MBCS_ENTRY_FINAL(state, action, value) (int32_t)(0x80000000|((int32_t)(state)<<24L)|((action)<<20L)|(value)) |
275 | #define MBCS_ENTRY_SET_FINAL(entry) (int32_t)((entry)|0x80000000) |
276 | #define MBCS_ENTRY_FINAL_SET_ACTION(entry, action) (int32_t)(((entry)&0xff0fffff)|((int32_t)(action)<<20L)) |
277 | #define MBCS_ENTRY_FINAL_SET_VALUE(entry, value) (int32_t)(((entry)&0xfff00000)|(value)) |
278 | #define MBCS_ENTRY_FINAL_SET_ACTION_VALUE(entry, action, value) (int32_t)(((entry)&0xff000000)|((int32_t)(action)<<20L)|(value)) |
279 | |
280 | #define MBCS_ENTRY_SET_STATE(entry, state) (int32_t)(((entry)&0x80ffffff)|((int32_t)(state)<<24L)) |
281 | |
282 | #define MBCS_ENTRY_STATE(entry) ((((uint32_t)entry)>>24)&0x7f) |
283 | |
284 | #define MBCS_ENTRY_IS_TRANSITION(entry) ((entry)>=0) |
285 | #define MBCS_ENTRY_IS_FINAL(entry) ((entry)<0) |
286 | |
287 | #define MBCS_ENTRY_TRANSITION_STATE(entry) (((uint32_t)entry)>>24) |
288 | #define MBCS_ENTRY_TRANSITION_OFFSET(entry) ((entry)&0xffffff) |
289 | |
290 | #define MBCS_ENTRY_FINAL_STATE(entry) ((((uint32_t)entry)>>24)&0x7f) |
291 | #define MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry) ((entry)<(int32_t)0x80100000) |
292 | #define MBCS_ENTRY_FINAL_ACTION(entry) ((((uint32_t)entry)>>20)&0xf) |
293 | #define MBCS_ENTRY_FINAL_VALUE(entry) ((entry)&0xfffff) |
294 | #define MBCS_ENTRY_FINAL_VALUE_16(entry) (uint16_t)(entry) |
295 | |
296 | #define IS_ASCII_ROUNDTRIP(b, asciiRoundtrips) (((asciiRoundtrips) & (1<<((b)>>2)))!=0) |
297 | |
298 | /* single-byte fromUnicode: get the 16-bit result word */ |
299 | #define MBCS_SINGLE_RESULT_FROM_U(table, results, c) (results)[ (table)[ (table)[(c)>>10] +(((c)>>4)&0x3f) ] +((c)&0xf) ] |
300 | |
301 | /* single-byte fromUnicode using the sbcsIndex */ |
302 | #define SBCS_RESULT_FROM_LOW_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ] |
303 | |
304 | /* single-byte fromUTF8 using the sbcsIndex; l and t must be masked externally; can be l=0 and t<=0x7f */ |
305 | #define SBCS_RESULT_FROM_UTF8(table, results, l, t) (results)[ (table)[l] +(t) ] |
306 | |
307 | /* multi-byte fromUnicode: get the 32-bit stage 2 entry */ |
308 | #define MBCS_STAGE_2_FROM_U(table, c) ((const uint32_t *)(table))[ (table)[(c)>>10] +(((c)>>4)&0x3f) ] |
309 | #define MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) ( ((stage2Entry) & ((uint32_t)1<< (16+((c)&0xf)) )) !=0) |
310 | |
311 | #define MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c) ((uint16_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)] |
312 | #define MBCS_VALUE_4_FROM_STAGE_2(bytes, stage2Entry, c) ((uint32_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)] |
313 | |
314 | #define MBCS_POINTER_3_FROM_STAGE_2(bytes, stage2Entry, c) ((bytes)+(16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf))*3) |
315 | |
316 | /* double-byte fromUnicode using the mbcsIndex */ |
317 | #define DBCS_RESULT_FROM_MOST_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ] |
318 | |
319 | /* double-byte fromUTF8 using the mbcsIndex; l and t1 combined into lt1; lt1 and t2 must be masked externally */ |
320 | #define DBCS_RESULT_FROM_UTF8(table, results, lt1, t2) (results)[ (table)[lt1] +(t2) ] |
321 | |
322 | |
323 | /** |
324 | * MBCS output types for conversions from Unicode. |
325 | * These per-converter types determine the storage method in stage 3 of the lookup table, |
326 | * mostly how many bytes are stored per entry. |
327 | */ |
328 | enum { |
329 | MBCS_OUTPUT_1, /* 0 */ |
330 | MBCS_OUTPUT_2, /* 1 */ |
331 | MBCS_OUTPUT_3, /* 2 */ |
332 | MBCS_OUTPUT_4, /* 3 */ |
333 | |
334 | MBCS_OUTPUT_3_EUC=8, /* 8 */ |
335 | MBCS_OUTPUT_4_EUC, /* 9 */ |
336 | |
337 | MBCS_OUTPUT_2_SISO=12, /* c */ |
338 | MBCS_OUTPUT_2_HZ, /* d */ |
339 | |
340 | MBCS_OUTPUT_EXT_ONLY, /* e */ |
341 | |
342 | MBCS_OUTPUT_COUNT, |
343 | |
344 | MBCS_OUTPUT_DBCS_ONLY=0xdb /* runtime-only type for DBCS-only handling of SISO tables */ |
345 | }; |
346 | |
347 | /** |
348 | * Fallbacks to Unicode are stored outside the normal state table and code point structures |
349 | * in a vector of items of this type. They are sorted by offset. |
350 | */ |
351 | typedef struct { |
352 | uint32_t offset; |
353 | UChar32 codePoint; |
354 | } _MBCSToUFallback; |
355 | |
356 | /** Constants for fast and UTF-8-friendly conversion. */ |
357 | enum { |
358 | SBCS_FAST_MAX=0x0fff, /* maximum code point with UTF-8-friendly SBCS runtime code, see makeconv SBCS_UTF8_MAX */ |
359 | SBCS_FAST_LIMIT=SBCS_FAST_MAX+1, /* =0x1000 */ |
360 | MBCS_FAST_MAX=0xd7ff, /* maximum code point with UTF-8-friendly MBCS runtime code, see makeconv MBCS_UTF8_MAX */ |
361 | MBCS_FAST_LIMIT=MBCS_FAST_MAX+1 /* =0xd800 */ |
362 | }; |
363 | |
364 | /** |
365 | * This is the MBCS part of the UConverterTable union (a runtime data structure). |
366 | * It keeps all the per-converter data and points into the loaded mapping tables. |
367 | * |
368 | * utf8Friendly data structures added with _MBCSHeader.version 4.3 |
369 | */ |
370 | typedef struct UConverterMBCSTable { |
371 | /* toUnicode */ |
372 | uint8_t countStates, dbcsOnlyState, stateTableOwned; |
373 | uint32_t countToUFallbacks; |
374 | |
375 | const int32_t (*stateTable)/*[countStates]*/[256]; |
376 | int32_t (*swapLFNLStateTable)/*[countStates]*/[256]; /* for swaplfnl */ |
377 | const uint16_t *unicodeCodeUnits/*[countUnicodeResults]*/; |
378 | const _MBCSToUFallback *toUFallbacks; |
379 | |
380 | /* fromUnicode */ |
381 | const uint16_t *fromUnicodeTable; |
382 | const uint16_t *mbcsIndex; /* for fast conversion from most of BMP to MBCS (utf8Friendly data) */ |
383 | uint16_t sbcsIndex[SBCS_FAST_LIMIT>>6]; /* for fast conversion from low BMP to SBCS (utf8Friendly data) */ |
384 | const uint8_t *fromUnicodeBytes; |
385 | uint8_t *swapLFNLFromUnicodeBytes; /* for swaplfnl */ |
386 | uint32_t fromUBytesLength; |
387 | uint8_t outputType, unicodeMask; |
388 | UBool utf8Friendly; /* for utf8Friendly data */ |
389 | UChar maxFastUChar; /* for utf8Friendly data */ |
390 | |
391 | /* roundtrips */ |
392 | uint32_t asciiRoundtrips; |
393 | |
394 | /* reconstituted data that was omitted from the .cnv file */ |
395 | uint8_t *reconstitutedData; |
396 | |
397 | /* converter name for swaplfnl */ |
398 | char *swapLFNLName; |
399 | |
400 | /* extension data */ |
401 | struct UConverterSharedData *baseSharedData; |
402 | const int32_t *extIndexes; |
403 | } UConverterMBCSTable; |
404 | |
405 | #define UCNV_MBCS_TABLE_INITIALIZER { \ |
406 | /* toUnicode */ \ |
407 | 0, 0, 0, \ |
408 | 0, \ |
409 | \ |
410 | NULL, \ |
411 | NULL, \ |
412 | NULL, \ |
413 | NULL, \ |
414 | \ |
415 | /* fromUnicode */ \ |
416 | NULL, \ |
417 | NULL, \ |
418 | { 0 }, \ |
419 | NULL, \ |
420 | NULL, \ |
421 | 0, \ |
422 | 0, 0, \ |
423 | FALSE, \ |
424 | 0, \ |
425 | \ |
426 | /* roundtrips */ \ |
427 | 0, \ |
428 | \ |
429 | /* reconstituted data that was omitted from the .cnv file */ \ |
430 | NULL, \ |
431 | \ |
432 | /* converter name for swaplfnl */ \ |
433 | NULL, \ |
434 | \ |
435 | /* extension data */ \ |
436 | NULL, \ |
437 | NULL \ |
438 | } |
439 | |
440 | enum { |
441 | MBCS_OPT_LENGTH_MASK=0x3f, |
442 | MBCS_OPT_NO_FROM_U=0x40, |
443 | /* |
444 | * If any of the following options bits are set, |
445 | * then the file must be rejected. |
446 | */ |
447 | MBCS_OPT_INCOMPATIBLE_MASK=0xffc0, |
448 | /* |
449 | * Remove bits from this mask as more options are recognized |
450 | * by all implementations that use this constant. |
451 | */ |
452 | MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK=0xff80 |
453 | }; |
454 | |
455 | enum { |
456 | =8, |
457 | =9 |
458 | }; |
459 | |
460 | /** |
461 | * MBCS data header. See data format description above. |
462 | */ |
463 | typedef struct { |
464 | UVersionInfo version; |
465 | uint32_t countStates, |
466 | countToUFallbacks, |
467 | offsetToUCodeUnits, |
468 | offsetFromUTable, |
469 | offsetFromUBytes, |
470 | flags, |
471 | fromUBytesLength; |
472 | |
473 | /* new and required in version 5 */ |
474 | uint32_t options; |
475 | |
476 | /* new and optional in version 5; used if options&MBCS_OPT_NO_FROM_U */ |
477 | uint32_t fullStage2Length; /* number of 32-bit units */ |
478 | } ; |
479 | |
480 | #define { { 0 }, 0, 0, 0, 0, 0, 0, 0, 0, 0 } |
481 | |
482 | /* |
483 | * This is a simple version of _MBCSGetNextUChar() that is used |
484 | * by other converter implementations. |
485 | * It only returns an "assigned" result if it consumes the entire input. |
486 | * It does not use state from the converter, nor error codes. |
487 | * It does not handle the EBCDIC swaplfnl option (set in UConverter). |
488 | * It handles conversion extensions but not GB 18030. |
489 | * |
490 | * Return value: |
491 | * U+fffe unassigned |
492 | * U+ffff illegal |
493 | * otherwise the Unicode code point |
494 | */ |
495 | U_CFUNC UChar32 |
496 | ucnv_MBCSSimpleGetNextUChar(UConverterSharedData *sharedData, |
497 | const char *source, int32_t length, |
498 | UBool useFallback); |
499 | |
500 | /** |
501 | * This version of _MBCSSimpleGetNextUChar() is optimized for single-byte, single-state codepages. |
502 | * It does not handle the EBCDIC swaplfnl option (set in UConverter). |
503 | * It does not handle conversion extensions (_extToU()). |
504 | */ |
505 | U_CFUNC UChar32 |
506 | ucnv_MBCSSingleSimpleGetNextUChar(UConverterSharedData *sharedData, |
507 | uint8_t b, UBool useFallback); |
508 | |
509 | /** |
510 | * This macro version of _MBCSSingleSimpleGetNextUChar() gets a code point from a byte. |
511 | * It works for single-byte, single-state codepages that only map |
512 | * to and from BMP code points, and it always |
513 | * returns fallback values. |
514 | */ |
515 | #define _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP(sharedData, b) \ |
516 | (UChar)MBCS_ENTRY_FINAL_VALUE_16((sharedData)->mbcs.stateTable[0][(uint8_t)(b)]) |
517 | |
518 | /** |
519 | * This is an internal function that allows other converter implementations |
520 | * to check whether a byte is a lead byte. |
521 | */ |
522 | U_CFUNC UBool |
523 | ucnv_MBCSIsLeadByte(UConverterSharedData *sharedData, char byte); |
524 | |
525 | /** This is a macro version of _MBCSIsLeadByte(). */ |
526 | #define _MBCS_IS_LEAD_BYTE(sharedData, byte) \ |
527 | (UBool)MBCS_ENTRY_IS_TRANSITION((sharedData)->mbcs.stateTable[0][(uint8_t)(byte)]) |
528 | |
529 | /* |
530 | * This is another simple conversion function for internal use by other |
531 | * conversion implementations. |
532 | * It does not use the converter state nor call callbacks. |
533 | * It does not handle the EBCDIC swaplfnl option (set in UConverter). |
534 | * It handles conversion extensions but not GB 18030. |
535 | * |
536 | * It converts one single Unicode code point into codepage bytes, encoded |
537 | * as one 32-bit value. The function returns the number of bytes in *pValue: |
538 | * 1..4 the number of bytes in *pValue |
539 | * 0 unassigned (*pValue undefined) |
540 | * -1 illegal (currently not used, *pValue undefined) |
541 | * |
542 | * *pValue will contain the resulting bytes with the last byte in bits 7..0, |
543 | * the second to last byte in bits 15..8, etc. |
544 | * Currently, the function assumes but does not check that 0<=c<=0x10ffff. |
545 | */ |
546 | U_CFUNC int32_t |
547 | ucnv_MBCSFromUChar32(UConverterSharedData *sharedData, |
548 | UChar32 c, uint32_t *pValue, |
549 | UBool useFallback); |
550 | |
551 | /** |
552 | * This version of _MBCSFromUChar32() is optimized for single-byte codepages. |
553 | * It does not handle the EBCDIC swaplfnl option (set in UConverter). |
554 | * |
555 | * It returns the codepage byte for the code point, or -1 if it is unassigned. |
556 | */ |
557 | U_CFUNC int32_t |
558 | ucnv_MBCSSingleFromUChar32(UConverterSharedData *sharedData, |
559 | UChar32 c, |
560 | UBool useFallback); |
561 | |
562 | /** |
563 | * SBCS, DBCS, and EBCDIC_STATEFUL are replaced by MBCS, but |
564 | * we cheat a little about the type, returning the old types if appropriate. |
565 | */ |
566 | U_CFUNC UConverterType |
567 | ucnv_MBCSGetType(const UConverter* converter); |
568 | |
569 | U_CFUNC void |
570 | ucnv_MBCSFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, |
571 | UErrorCode *pErrorCode); |
572 | U_CFUNC void |
573 | ucnv_MBCSToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
574 | UErrorCode *pErrorCode); |
575 | |
576 | /* |
577 | * Internal function returning a UnicodeSet for toUnicode() conversion. |
578 | * Currently only used for ISO-2022-CN, and only handles roundtrip mappings. |
579 | * In the future, if we add support for fallback sets, this function |
580 | * needs to be updated. |
581 | * Handles extensions. |
582 | * Does not empty the set first. |
583 | */ |
584 | U_CFUNC void |
585 | ucnv_MBCSGetUnicodeSetForUnicode(const UConverterSharedData *sharedData, |
586 | const USetAdder *sa, |
587 | UConverterUnicodeSet which, |
588 | UErrorCode *pErrorCode); |
589 | |
590 | /* |
591 | * Same as ucnv_MBCSGetUnicodeSetForUnicode() but |
592 | * the set can be filtered by encoding scheme. |
593 | * Used by stateful converters which share regular conversion tables |
594 | * but only use a subset of their mappings. |
595 | */ |
596 | U_CFUNC void |
597 | ucnv_MBCSGetFilteredUnicodeSetForUnicode(const UConverterSharedData *sharedData, |
598 | const USetAdder *sa, |
599 | UConverterUnicodeSet which, |
600 | UConverterSetFilter filter, |
601 | UErrorCode *pErrorCode); |
602 | |
603 | #endif |
604 | |
605 | #endif |
606 | |