icuSources/common/ucnvmbcs.h

   1 /*
   2 ******************************************************************************
   3 *
   4 *   Copyright (C) 2000-2013, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 ******************************************************************************
   8 *   file name:  ucnvmbcs.h
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2000jul07
  14 *   created by: Markus W. Scherer
  15 */
  16
  17 #ifndef __UCNVMBCS_H__
  18 #define __UCNVMBCS_H__
  19
  20 #include "unicode/utypes.h"
  21
  22 #if !UCONFIG_NO_CONVERSION
  23
  24 #include "unicode/ucnv.h"
  25 #include "ucnv_cnv.h"
  26 #include "ucnv_ext.h"
  27
  28 /**
  29  * ICU conversion (.cnv) data file structure, following the usual UDataInfo
  30  * header.
  31  *
  32  * Format version: 6.2
  33  *
  34  * struct UConverterStaticData -- struct containing the converter name, IBM CCSID,
  35  *                                min/max bytes per character, etc.
  36  *                                see ucnv_bld.h
  37  *
  38  * --------------------
  39  *
  40  * The static data is followed by conversionType-specific data structures.
  41  * At the moment, there are only variations of MBCS converters. They all have
  42  * the same toUnicode structures, while the fromUnicode structures for SBCS
  43  * differ from those for other MBCS-style converters.
  44  *
  45  * _MBCSHeader.version 5 is optional and not backward-compatible
  46  * (as usual for changes in the major version field).
  47  *
  48  * Versions 5.m work like versions 4.m except:
  49  * - The _MBCSHeader has variable length (and is always longer than in version 4).
  50  *   See the struct _MBCSHeader further description below.
  51  * - There is a set of flags which indicate further incompatible changes.
  52  *   (Reader code must reject the file if it does not recognize them all.)
  53  * - In particular, one of these flags indicates that most of the fromUnicode
  54  *   data is missing and must be reconstituted from the toUnicode data
  55  *   and from the utf8Friendly mbcsIndex at load time.
  56  *   (This only works with a utf8Friendly table.)
  57  *   In this case, makeconv may increase maxFastUChar automatically to U+FFFF.
  58  *
  59  * The first of these versions is 5.3, which is like 4.3 except for the differences above.
  60  *
  61  * When possible, makeconv continues to generate version 4.m files.
  62  *
  63  * _MBCSHeader.version 5.4/4.4 supports "good one-way" mappings (|4)
  64  * in the extension tables (fromUTableValues bit 30). See ucnv_ext.h for details.
  65  *
  66  * _MBCSHeader.version 4.3 optionally modifies the fromUnicode data structures
  67  * slightly and optionally adds a table for conversion to MBCS (non-SBCS)
  68  * charsets.
  69  *
  70  * The modifications are to make the data utf8Friendly. Not every 4.3 file
  71  * file contains utf8Friendly data.
  72  * It is utf8Friendly if _MBCSHeader.version[2]!=0.
  73  * In this case, the data structures are utf8Friendly up to the code point
  74  *   maxFastUChar=((_MBCSHeader.version[2]<<8)|0xff)
  75  *
  76  * A utf8Friendly file has fromUnicode stage 3 entries for code points up to
  77  * maxFastUChar allocated in blocks of 64 for indexing with the 6 bits from
  78  * a UTF-8 trail byte. ASCII is allocated linearly with 128 contiguous entries.
  79  *
  80  * In addition, a utf8Friendly MBCS file contains an additional
  81  *   uint16_t mbcsIndex[(maxFastUChar+1)>>6];
  82  * which replaces the stage 1 and 2 tables for indexing with bits from the
  83  * UTF-8 lead byte and middle trail byte. Unlike the older MBCS stage 2 table,
  84  * the mbcsIndex does not contain roundtrip flags. Therefore, all fallbacks
  85  * from code points up to maxFastUChar (and roundtrips to 0x00) are moved to
  86  * the extension data structure. This also allows for faster roundtrip
  87  * conversion from UTF-16.
  88  *
  89  * SBCS files do not contain an additional sbcsIndex[] array because the
  90  * proportional size increase would be noticeable, but the runtime
  91  * code builds one for the code point range for which the runtime conversion
  92  * code is optimized.
  93  *
  94  * For SBCS, maxFastUChar should be at least U+0FFF. The initial makeconv
  95  * implementation sets it to U+1FFF. Because the sbcsIndex is not stored in
  96  * the file, a larger maxFastUChar only affects stage 3 block allocation size
  97  * and is free in empty blocks. (Larger blocks with sparse contents cause larger
  98  * files.) U+1FFF includes almost all of the small scripts.
  99  * U+0FFF covers UTF-8 two-byte sequences and three-byte sequences starting with
 100  * 0xe0. This includes most scripts with legacy SBCS charsets.
 101  * The initial runtime implementation using 4.3 files only builds an sbcsIndex
 102  * for code points up to U+0FFF.
 103  *
 104  * For MBCS, maxFastUChar should be at least U+D7FF (=initial value).
 105  * This boundary is convenient because practically all of the commonly used
 106  * characters are below it, and because it is the boundary to surrogate
 107  * code points, above which special handling is necessary anyway.
 108  * (Surrogate pair assembly for UTF-16, validity checking for UTF-8.)
 109  *
 110  * maxFastUChar could be up to U+FFFF to cover the whole BMP, which could be
 111  * useful especially for conversion from UTF-8 when the input can be assumed
 112  * to be valid, because the surrogate range would then not have to be
 113  * checked.
 114  * (With maxFastUChar=0xffff, makeconv would have to check for mbcsIndex value
 115  * overflow because with the all-unassigned block 0 and nearly full mappings
 116  * from the BMP it is theoretically possible that an index into stage 3
 117  * exceeds 16 bits.)
 118  *
 119  * _MBCSHeader.version 4.2 adds an optional conversion extension data structure.
 120  * If it is present, then an ICU version reading header versions 4.0 or 4.1
 121  * will be able to use the base table and ignore the extension.
 122  *
 123  * The unicodeMask in the static data is part of the base table data structure.
 124  * Especially, the UCNV_HAS_SUPPLEMENTARY flag determines the length of the
 125  * fromUnicode stage 1 array.
 126  * The static data unicodeMask refers only to the base table's properties if
 127  * a base table is included.
 128  * In an extension-only file, the static data unicodeMask is 0.
 129  * The extension data indexes have a separate field with the unicodeMask flags.
 130  *
 131  * MBCS-style data structure following the static data.
 132  * Offsets are counted in bytes from the beginning of the MBCS header structure.
 133  * Details about usage in comments in ucnvmbcs.c.
 134  *
 135  * struct _MBCSHeader (see the definition in this header file below)
 136  * contains 32-bit fields as follows:
 137  * 8 values:
 138  *  0   uint8_t[4]  MBCS version in UVersionInfo format (currently 4.3.x.0)
 139  *  1   uint32_t    countStates
 140  *  2   uint32_t    countToUFallbacks
 141  *  3   uint32_t    offsetToUCodeUnits
 142  *  4   uint32_t    offsetFromUTable
 143  *  5   uint32_t    offsetFromUBytes
 144  *  6   uint32_t    flags, bits:
 145  *                      31.. 8 offsetExtension -- _MBCSHeader.version 4.2 (ICU 2.8) and higher
 146  *                                                0 for older versions and if
 147  *                                                there is not extension structure
 148  *                       7.. 0 outputType
 149  *  7   uint32_t    fromUBytesLength -- _MBCSHeader.version 4.1 (ICU 2.4) and higher
 150  *                  counts bytes in fromUBytes[]
 151  *
 152  * New and required in version 5:
 153  *  8   uint32_t    options, bits:
 154  *                      31..16 reserved for flags that can be added without breaking
 155  *                                 backward compatibility
 156  *                      15.. 6 reserved for flags whose addition will break
 157  *                                 backward compatibility
 158  *                           6 MBCS_OPT_FROM_U -- if set,
 159  *                                 then most of the fromUnicode data is omitted;
 160  *                                 fullStage2Length is present and the missing
 161  *                                 bottom part of stage 2 must be reconstituted from
 162  *                                 the toUnicode data;
 163  *                                 stage 3 is missing completely as well;
 164  *                                 not used for SBCS tables
 165  *                       5.. 0 length of the _MBCSHeader (number of uint32_t)
 166  *
 167  * New and optional in version 5:
 168  *  9   uint32_t    fullStage2Length: used if MBCS_OPT_FROM_U is set
 169  *                                 specifies the full length of stage 2
 170  *                                 including the omitted part
 171  *
 172  * if(outputType==MBCS_OUTPUT_EXT_ONLY) {
 173  *     -- base table name for extension-only table
 174  *     char baseTableName[variable]; -- with NUL plus padding for 4-alignment
 175  *
 176  *     -- all _MBCSHeader fields except for version and flags are 0
 177  * } else {
 178  *     -- normal base table with optional extension
 179  *
 180  *     int32_t stateTable[countStates][256];
 181  *
 182  *     struct _MBCSToUFallback { (fallbacks are sorted by offset)
 183  *         uint32_t offset;
 184  *         UChar32 codePoint;
 185  *     } toUFallbacks[countToUFallbacks];
 186  *
 187  *     uint16_t unicodeCodeUnits[(offsetFromUTable-offsetToUCodeUnits)/2];
 188  *                  (padded to an even number of units)
 189  *
 190  *     -- stage 1 tables
 191  *     if(staticData.unicodeMask&UCNV_HAS_SUPPLEMENTARY) {
 192  *         -- stage 1 table for all of Unicode
 193  *         uint16_t fromUTable[0x440]; (32-bit-aligned)
 194  *     } else {
 195  *         -- BMP-only tables have a smaller stage 1 table
 196  *         uint16_t fromUTable[0x40]; (32-bit-aligned)
 197  *     }
 198  *
 199  *     -- stage 2 tables
 200  *        length determined by top of stage 1 and bottom of stage 3 tables
 201  *     if(outputType==MBCS_OUTPUT_1) {
 202  *         -- SBCS: pure indexes
 203  *         uint16_t stage 2 indexes[?];
 204  *     } else {
 205  *         -- DBCS, MBCS, EBCDIC_STATEFUL, ...: roundtrip flags and indexes
 206  *         uint32_t stage 2 flags and indexes[?];
 207  *         if(options&MBCS_OPT_NO_FROM_U) {
 208  *             stage 2 really has length fullStage2Length
 209  *             and the omitted lower part must be reconstituted from
 210  *             the toUnicode data
 211  *         }
 212  *     }
 213  *
 214  *     -- stage 3 tables with byte results
 215  *     if(outputType==MBCS_OUTPUT_1) {
 216  *         -- SBCS: each 16-bit result contains flags and the result byte, see ucnvmbcs.c
 217  *         uint16_t fromUBytes[fromUBytesLength/2];
 218  *     } else if(!(options&MBCS_OPT_NO_FROM_U)) {
 219  *         -- DBCS, MBCS, EBCDIC_STATEFUL, ... 2/3/4 bytes result, see ucnvmbcs.c
 220  *         uint8_t fromUBytes[fromUBytesLength]; or
 221  *         uint16_t fromUBytes[fromUBytesLength/2]; or
 222  *         uint32_t fromUBytes[fromUBytesLength/4];
 223  *     } else {
 224  *         fromUBytes[] must be reconstituted from the toUnicode data
 225  *     }
 226  *
 227  *     -- optional utf8Friendly mbcsIndex -- _MBCSHeader.version 4.3 (ICU 3.8) and higher
 228  *     if(outputType!=MBCS_OUTPUT_1 &&
 229  *        _MBCSHeader.version[1]>=3 &&
 230  *        (maxFastUChar=_MBCSHeader.version[2])!=0
 231  *     ) {
 232  *         maxFastUChar=(maxFastUChar<<8)|0xff;
 233  *         uint16_t mbcsIndex[(maxFastUChar+1)>>6];
 234  *     }
 235  * }
 236  *
 237  * -- extension table, details see ucnv_ext.h
 238  * int32_t indexes[>=32]; ...
 239  */
 240
 241 /* MBCS converter data and state -------------------------------------------- */
 242
 243 enum {
 244     MBCS_MAX_STATE_COUNT=128
 245 };
 246
 247 /**
 248  * MBCS action codes for conversions to Unicode.
 249  * These values are in bits 23..20 of the state table entries.
 250  */
 251 enum {
 252     MBCS_STATE_VALID_DIRECT_16,
 253     MBCS_STATE_VALID_DIRECT_20,
 254
 255     MBCS_STATE_FALLBACK_DIRECT_16,
 256     MBCS_STATE_FALLBACK_DIRECT_20,
 257
 258     MBCS_STATE_VALID_16,
 259     MBCS_STATE_VALID_16_PAIR,
 260
 261     MBCS_STATE_UNASSIGNED,
 262     MBCS_STATE_ILLEGAL,
 263
 264     MBCS_STATE_CHANGE_ONLY
 265 };
 266
 267 /* Macros for state table entries */
 268 #define MBCS_ENTRY_TRANSITION(state, offset) (int32_t)(((int32_t)(state)<<24L)|(offset))
 269 #define MBCS_ENTRY_TRANSITION_SET_OFFSET(entry, offset) (int32_t)(((entry)&0xff000000)|(offset))
 270 #define MBCS_ENTRY_TRANSITION_ADD_OFFSET(entry, offset) (int32_t)((entry)+(offset))
 271
 272 #define MBCS_ENTRY_FINAL(state, action, value) (int32_t)(0x80000000|((int32_t)(state)<<24L)|((action)<<20L)|(value))
 273 #define MBCS_ENTRY_SET_FINAL(entry) (int32_t)((entry)|0x80000000)
 274 #define MBCS_ENTRY_FINAL_SET_ACTION(entry, action) (int32_t)(((entry)&0xff0fffff)|((int32_t)(action)<<20L))
 275 #define MBCS_ENTRY_FINAL_SET_VALUE(entry, value) (int32_t)(((entry)&0xfff00000)|(value))
 276 #define MBCS_ENTRY_FINAL_SET_ACTION_VALUE(entry, action, value) (int32_t)(((entry)&0xff000000)|((int32_t)(action)<<20L)|(value))
 277
 278 #define MBCS_ENTRY_SET_STATE(entry, state) (int32_t)(((entry)&0x80ffffff)|((int32_t)(state)<<24L))
 279
 280 #define MBCS_ENTRY_STATE(entry) ((((uint32_t)entry)>>24)&0x7f)
 281
 282 #define MBCS_ENTRY_IS_TRANSITION(entry) ((entry)>=0)
 283 #define MBCS_ENTRY_IS_FINAL(entry) ((entry)<0)
 284
 285 #define MBCS_ENTRY_TRANSITION_STATE(entry) (((uint32_t)entry)>>24)
 286 #define MBCS_ENTRY_TRANSITION_OFFSET(entry) ((entry)&0xffffff)
 287
 288 #define MBCS_ENTRY_FINAL_STATE(entry) ((((uint32_t)entry)>>24)&0x7f)
 289 #define MBCS_ENTRY_FINAL_IS_VALID_DIRECT_16(entry) ((entry)<(int32_t)0x80100000)
 290 #define MBCS_ENTRY_FINAL_ACTION(entry) ((((uint32_t)entry)>>20)&0xf)
 291 #define MBCS_ENTRY_FINAL_VALUE(entry) ((entry)&0xfffff)
 292 #define MBCS_ENTRY_FINAL_VALUE_16(entry) (uint16_t)(entry)
 293
 294 #define IS_ASCII_ROUNDTRIP(b, asciiRoundtrips) (((asciiRoundtrips) & (1<<((b)>>2)))!=0)
 295
 296 /* single-byte fromUnicode: get the 16-bit result word */
 297 #define MBCS_SINGLE_RESULT_FROM_U(table, results, c) (results)[ (table)[ (table)[(c)>>10] +(((c)>>4)&0x3f) ] +((c)&0xf) ]
 298
 299 /* single-byte fromUnicode using the sbcsIndex */
 300 #define SBCS_RESULT_FROM_LOW_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ]
 301
 302 /* single-byte fromUTF8 using the sbcsIndex; l and t must be masked externally; can be l=0 and t<=0x7f */
 303 #define SBCS_RESULT_FROM_UTF8(table, results, l, t) (results)[ (table)[l] +(t) ]
 304
 305 /* multi-byte fromUnicode: get the 32-bit stage 2 entry */
 306 #define MBCS_STAGE_2_FROM_U(table, c) ((const uint32_t *)(table))[ (table)[(c)>>10] +(((c)>>4)&0x3f) ]
 307 #define MBCS_FROM_U_IS_ROUNDTRIP(stage2Entry, c) ( ((stage2Entry) & ((uint32_t)1<< (16+((c)&0xf)) )) !=0)
 308
 309 #define MBCS_VALUE_2_FROM_STAGE_2(bytes, stage2Entry, c) ((uint16_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)]
 310 #define MBCS_VALUE_4_FROM_STAGE_2(bytes, stage2Entry, c) ((uint32_t *)(bytes))[16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf)]
 311
 312 #define MBCS_POINTER_3_FROM_STAGE_2(bytes, stage2Entry, c) ((bytes)+(16*(uint32_t)(uint16_t)(stage2Entry)+((c)&0xf))*3)
 313
 314 /* double-byte fromUnicode using the mbcsIndex */
 315 #define DBCS_RESULT_FROM_MOST_BMP(table, results, c) (results)[ (table)[(c)>>6] +((c)&0x3f) ]
 316
 317 /* double-byte fromUTF8 using the mbcsIndex; l and t1 combined into lt1; lt1 and t2 must be masked externally */
 318 #define DBCS_RESULT_FROM_UTF8(table, results, lt1, t2) (results)[ (table)[lt1] +(t2) ]
 319
 320
 321 /**
 322  * MBCS output types for conversions from Unicode.
 323  * These per-converter types determine the storage method in stage 3 of the lookup table,
 324  * mostly how many bytes are stored per entry.
 325  */
 326 enum {
 327     MBCS_OUTPUT_1,          /* 0 */
 328     MBCS_OUTPUT_2,          /* 1 */
 329     MBCS_OUTPUT_3,          /* 2 */
 330     MBCS_OUTPUT_4,          /* 3 */
 331
 332     MBCS_OUTPUT_3_EUC=8,    /* 8 */
 333     MBCS_OUTPUT_4_EUC,      /* 9 */
 334
 335     MBCS_OUTPUT_2_SISO=12,  /* c */
 336     MBCS_OUTPUT_2_HZ,       /* d */
 337
 338     MBCS_OUTPUT_EXT_ONLY,   /* e */
 339
 340     MBCS_OUTPUT_COUNT,
 341
 342     MBCS_OUTPUT_DBCS_ONLY=0xdb  /* runtime-only type for DBCS-only handling of SISO tables */
 343 };
 344
 345 /**
 346  * Fallbacks to Unicode are stored outside the normal state table and code point structures
 347  * in a vector of items of this type. They are sorted by offset.
 348  */
 349 typedef struct {
 350     uint32_t offset;
 351     UChar32 codePoint;
 352 } _MBCSToUFallback;
 353
 354 /** Constants for fast and UTF-8-friendly conversion. */
 355 enum {
 356     SBCS_FAST_MAX=0x0fff,               /* maximum code point with UTF-8-friendly SBCS runtime code, see makeconv SBCS_UTF8_MAX */
 357     SBCS_FAST_LIMIT=SBCS_FAST_MAX+1,    /* =0x1000 */
 358     MBCS_FAST_MAX=0xd7ff,               /* maximum code point with UTF-8-friendly MBCS runtime code, see makeconv MBCS_UTF8_MAX */
 359     MBCS_FAST_LIMIT=MBCS_FAST_MAX+1     /* =0xd800 */
 360 };
 361
 362 /**
 363  * This is the MBCS part of the UConverterTable union (a runtime data structure).
 364  * It keeps all the per-converter data and points into the loaded mapping tables.
 365  *
 366  * utf8Friendly data structures added with _MBCSHeader.version 4.3
 367  */
 368 typedef struct UConverterMBCSTable {
 369     /* toUnicode */
 370     uint8_t countStates, dbcsOnlyState, stateTableOwned;
 371     uint32_t countToUFallbacks;
 372
 373     const int32_t (*stateTable)/*[countStates]*/[256];
 374     int32_t (*swapLFNLStateTable)/*[countStates]*/[256]; /* for swaplfnl */
 375     const uint16_t *unicodeCodeUnits/*[countUnicodeResults]*/;
 376     const _MBCSToUFallback *toUFallbacks;
 377
 378     /* fromUnicode */
 379     const uint16_t *fromUnicodeTable;
 380     const uint16_t *mbcsIndex;              /* for fast conversion from most of BMP to MBCS (utf8Friendly data) */
 381     uint16_t sbcsIndex[SBCS_FAST_LIMIT>>6]; /* for fast conversion from low BMP to SBCS (utf8Friendly data) */
 382     const uint8_t *fromUnicodeBytes;
 383     uint8_t *swapLFNLFromUnicodeBytes;      /* for swaplfnl */
 384     uint32_t fromUBytesLength;
 385     uint8_t outputType, unicodeMask;
 386     UBool utf8Friendly;                     /* for utf8Friendly data */
 387     UChar maxFastUChar;                     /* for utf8Friendly data */
 388
 389     /* roundtrips */
 390     uint32_t asciiRoundtrips;
 391
 392     /* reconstituted data that was omitted from the .cnv file */
 393     uint8_t *reconstitutedData;
 394
 395     /* converter name for swaplfnl */
 396     char *swapLFNLName;
 397
 398     /* extension data */
 399     struct UConverterSharedData *baseSharedData;
 400     const int32_t *extIndexes;
 401 } UConverterMBCSTable;
 402
 403 #define UCNV_MBCS_TABLE_INITIALIZER { \
 404     /* toUnicode */ \
 405     0, 0, 0, \
 406     0, \
 407      \
 408     NULL, \
 409     NULL, \
 410     NULL, \
 411     NULL, \
 412      \
 413     /* fromUnicode */ \
 414     NULL, \
 415     NULL, \
 416     { 0 }, \
 417     NULL, \
 418     NULL, \
 419     0, \
 420     0, 0, \
 421     FALSE, \
 422     0, \
 423      \
 424     /* roundtrips */ \
 425     0, \
 426      \
 427     /* reconstituted data that was omitted from the .cnv file */ \
 428     NULL, \
 429      \
 430     /* converter name for swaplfnl */ \
 431     NULL, \
 432      \
 433     /* extension data */ \
 434     NULL, \
 435     NULL \
 436 }
 437
 438 enum {
 439     MBCS_OPT_LENGTH_MASK=0x3f,
 440     MBCS_OPT_NO_FROM_U=0x40,
 441     /*
 442      * If any of the following options bits are set,
 443      * then the file must be rejected.
 444      */
 445     MBCS_OPT_INCOMPATIBLE_MASK=0xffc0,
 446     /*
 447      * Remove bits from this mask as more options are recognized
 448      * by all implementations that use this constant.
 449      */
 450     MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK=0xff80
 451 };
 452
 453 enum {
 454     MBCS_HEADER_V4_LENGTH=8,
 455     MBCS_HEADER_V5_MIN_LENGTH=9
 456 };
 457
 458 /**
 459  * MBCS data header. See data format description above.
 460  */
 461 typedef struct {
 462     UVersionInfo version;
 463     uint32_t countStates,
 464              countToUFallbacks,
 465              offsetToUCodeUnits,
 466              offsetFromUTable,
 467              offsetFromUBytes,
 468              flags,
 469              fromUBytesLength;
 470
 471     /* new and required in version 5 */
 472     uint32_t options;
 473
 474     /* new and optional in version 5; used if options&MBCS_OPT_NO_FROM_U */
 475     uint32_t fullStage2Length;  /* number of 32-bit units */
 476 } _MBCSHeader;
 477
 478 #define UCNV_MBCS_HEADER_INITIALIZER { { 0 },  0, 0, 0, 0, 0, 0, 0,  0,  0 }
 479
 480 /*
 481  * This is a simple version of _MBCSGetNextUChar() that is used
 482  * by other converter implementations.
 483  * It only returns an "assigned" result if it consumes the entire input.
 484  * It does not use state from the converter, nor error codes.
 485  * It does not handle the EBCDIC swaplfnl option (set in UConverter).
 486  * It handles conversion extensions but not GB 18030.
 487  *
 488  * Return value:
 489  * U+fffe   unassigned
 490  * U+ffff   illegal
 491  * otherwise the Unicode code point
 492  */
 493 U_CFUNC UChar32
 494 ucnv_MBCSSimpleGetNextUChar(UConverterSharedData *sharedData,
 495                         const char *source, int32_t length,
 496                         UBool useFallback);
 497
 498 /**
 499  * This version of _MBCSSimpleGetNextUChar() is optimized for single-byte, single-state codepages.
 500  * It does not handle the EBCDIC swaplfnl option (set in UConverter).
 501  * It does not handle conversion extensions (_extToU()).
 502  */
 503 U_CFUNC UChar32
 504 ucnv_MBCSSingleSimpleGetNextUChar(UConverterSharedData *sharedData,
 505                               uint8_t b, UBool useFallback);
 506
 507 /**
 508  * This macro version of _MBCSSingleSimpleGetNextUChar() gets a code point from a byte.
 509  * It works for single-byte, single-state codepages that only map
 510  * to and from BMP code points, and it always
 511  * returns fallback values.
 512  */
 513 #define _MBCS_SINGLE_SIMPLE_GET_NEXT_BMP(sharedData, b) \
 514     (UChar)MBCS_ENTRY_FINAL_VALUE_16((sharedData)->mbcs.stateTable[0][(uint8_t)(b)])
 515
 516 /**
 517  * This is an internal function that allows other converter implementations
 518  * to check whether a byte is a lead byte.
 519  */
 520 U_CFUNC UBool
 521 ucnv_MBCSIsLeadByte(UConverterSharedData *sharedData, char byte);
 522
 523 /** This is a macro version of _MBCSIsLeadByte(). */
 524 #define _MBCS_IS_LEAD_BYTE(sharedData, byte) \
 525     (UBool)MBCS_ENTRY_IS_TRANSITION((sharedData)->mbcs.stateTable[0][(uint8_t)(byte)])
 526
 527 /*
 528  * This is another simple conversion function for internal use by other
 529  * conversion implementations.
 530  * It does not use the converter state nor call callbacks.
 531  * It does not handle the EBCDIC swaplfnl option (set in UConverter).
 532  * It handles conversion extensions but not GB 18030.
 533  *
 534  * It converts one single Unicode code point into codepage bytes, encoded
 535  * as one 32-bit value. The function returns the number of bytes in *pValue:
 536  * 1..4 the number of bytes in *pValue
 537  * 0    unassigned (*pValue undefined)
 538  * -1   illegal (currently not used, *pValue undefined)
 539  *
 540  * *pValue will contain the resulting bytes with the last byte in bits 7..0,
 541  * the second to last byte in bits 15..8, etc.
 542  * Currently, the function assumes but does not check that 0<=c<=0x10ffff.
 543  */
 544 U_CFUNC int32_t
 545 ucnv_MBCSFromUChar32(UConverterSharedData *sharedData,
 546                  UChar32 c, uint32_t *pValue,
 547                  UBool useFallback);
 548
 549 /**
 550  * This version of _MBCSFromUChar32() is optimized for single-byte codepages.
 551  * It does not handle the EBCDIC swaplfnl option (set in UConverter).
 552  *
 553  * It returns the codepage byte for the code point, or -1 if it is unassigned.
 554  */
 555 U_CFUNC int32_t
 556 ucnv_MBCSSingleFromUChar32(UConverterSharedData *sharedData,
 557                        UChar32 c,
 558                        UBool useFallback);
 559
 560 /**
 561  * SBCS, DBCS, and EBCDIC_STATEFUL are replaced by MBCS, but
 562  * we cheat a little about the type, returning the old types if appropriate.
 563  */
 564 U_CFUNC UConverterType
 565 ucnv_MBCSGetType(const UConverter* converter);
 566
 567 U_CFUNC void
 568 ucnv_MBCSFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs,
 569                             UErrorCode *pErrorCode);
 570 U_CFUNC void
 571 ucnv_MBCSToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs,
 572                           UErrorCode *pErrorCode);
 573
 574 /*
 575  * Internal function returning a UnicodeSet for toUnicode() conversion.
 576  * Currently only used for ISO-2022-CN, and only handles roundtrip mappings.
 577  * In the future, if we add support for fallback sets, this function
 578  * needs to be updated.
 579  * Handles extensions.
 580  * Does not empty the set first.
 581  */
 582 U_CFUNC void
 583 ucnv_MBCSGetUnicodeSetForUnicode(const UConverterSharedData *sharedData,
 584                                  const USetAdder *sa,
 585                                  UConverterUnicodeSet which,
 586                                  UErrorCode *pErrorCode);
 587
 588 /*
 589  * Same as ucnv_MBCSGetUnicodeSetForUnicode() but
 590  * the set can be filtered by encoding scheme.
 591  * Used by stateful converters which share regular conversion tables
 592  * but only use a subset of their mappings.
 593  */
 594 U_CFUNC void
 595 ucnv_MBCSGetFilteredUnicodeSetForUnicode(const UConverterSharedData *sharedData,
 596                                          const USetAdder *sa,
 597                                          UConverterUnicodeSet which,
 598                                          UConverterSetFilter filter,
 599                                          UErrorCode *pErrorCode);
 600
 601 #endif
 602
 603 #endif