icuSources/common/udataswp.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2003-2014, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  udataswp.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2003jun05
  14 *   created by: Markus W. Scherer
  15 *
  16 *   Definitions for ICU data transformations for different platforms,
  17 *   changing between big- and little-endian data and/or between
  18 *   charset families (ASCII<->EBCDIC).
  19 */
  20
  21 #include <stdarg.h>
  22 #include "unicode/utypes.h"
  23 #include "unicode/udata.h" /* UDataInfo */
  24 #include "ucmndata.h" /* DataHeader */
  25 #include "cmemory.h"
  26 #include "udataswp.h"
  27
  28 /* swapping primitives ------------------------------------------------------ */
  29
  30 static int32_t U_CALLCONV
  31 uprv_swapArray16(const UDataSwapper *ds,
  32                  const void *inData, int32_t length, void *outData,
  33                  UErrorCode *pErrorCode) {
  34     const uint16_t *p;
  35     uint16_t *q;
  36     int32_t count;
  37     uint16_t x;
  38
  39     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
  40         return 0;
  41     }
  42     if(ds==NULL || inData==NULL || length<0 || (length&1)!=0 || outData==NULL) {
  43         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
  44         return 0;
  45     }
  46
  47     /* setup and swapping */
  48     p=(const uint16_t *)inData;
  49     q=(uint16_t *)outData;
  50     count=length/2;
  51     while(count>0) {
  52         x=*p++;
  53         *q++=(uint16_t)((x<<8)|(x>>8));
  54         --count;
  55     }
  56
  57     return length;
  58 }
  59
  60 static int32_t U_CALLCONV
  61 uprv_copyArray16(const UDataSwapper *ds,
  62                  const void *inData, int32_t length, void *outData,
  63                  UErrorCode *pErrorCode) {
  64     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
  65         return 0;
  66     }
  67     if(ds==NULL || inData==NULL || length<0 || (length&1)!=0 || outData==NULL) {
  68         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
  69         return 0;
  70     }
  71
  72     if(length>0 && inData!=outData) {
  73         uprv_memcpy(outData, inData, length);
  74     }
  75     return length;
  76 }
  77
  78 static int32_t U_CALLCONV
  79 uprv_swapArray32(const UDataSwapper *ds,
  80                  const void *inData, int32_t length, void *outData,
  81                  UErrorCode *pErrorCode) {
  82     const uint32_t *p;
  83     uint32_t *q;
  84     int32_t count;
  85     uint32_t x;
  86
  87     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
  88         return 0;
  89     }
  90     if(ds==NULL || inData==NULL || length<0 || (length&3)!=0 || outData==NULL) {
  91         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
  92         return 0;
  93     }
  94
  95     /* setup and swapping */
  96     p=(const uint32_t *)inData;
  97     q=(uint32_t *)outData;
  98     count=length/4;
  99     while(count>0) {
 100         x=*p++;
 101         *q++=(uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24));
 102         --count;
 103     }
 104
 105     return length;
 106 }
 107
 108 static int32_t U_CALLCONV
 109 uprv_copyArray32(const UDataSwapper *ds,
 110                  const void *inData, int32_t length, void *outData,
 111                  UErrorCode *pErrorCode) {
 112     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 113         return 0;
 114     }
 115     if(ds==NULL || inData==NULL || length<0 || (length&3)!=0 || outData==NULL) {
 116         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 117         return 0;
 118     }
 119
 120     if(length>0 && inData!=outData) {
 121         uprv_memcpy(outData, inData, length);
 122     }
 123     return length;
 124 }
 125
 126 static int32_t U_CALLCONV
 127 uprv_swapArray64(const UDataSwapper *ds,
 128                  const void *inData, int32_t length, void *outData,
 129                  UErrorCode *pErrorCode) {
 130     const uint64_t *p;
 131     uint64_t *q;
 132     int32_t count;
 133
 134     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 135         return 0;
 136     }
 137     if(ds==NULL || inData==NULL || length<0 || (length&7)!=0 || outData==NULL) {
 138         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 139         return 0;
 140     }
 141
 142     /* setup and swapping */
 143     p=(const uint64_t *)inData;
 144     q=(uint64_t *)outData;
 145     count=length/8;
 146     while(count>0) {
 147         uint64_t x=*p++;
 148         x=(x<<56)|((x&0xff00)<<40)|((x&0xff0000)<<24)|((x&0xff000000)<<8)|
 149             ((x>>8)&0xff000000)|((x>>24)&0xff0000)|((x>>40)&0xff00)|(x>>56);
 150         *q++=x;
 151         --count;
 152     }
 153
 154     return length;
 155 }
 156
 157 static int32_t U_CALLCONV
 158 uprv_copyArray64(const UDataSwapper *ds,
 159                  const void *inData, int32_t length, void *outData,
 160                  UErrorCode *pErrorCode) {
 161     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 162         return 0;
 163     }
 164     if(ds==NULL || inData==NULL || length<0 || (length&7)!=0 || outData==NULL) {
 165         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 166         return 0;
 167     }
 168
 169     if(length>0 && inData!=outData) {
 170         uprv_memcpy(outData, inData, length);
 171     }
 172     return length;
 173 }
 174
 175 static uint16_t U_CALLCONV
 176 uprv_readSwapUInt16(uint16_t x) {
 177     return (uint16_t)((x<<8)|(x>>8));
 178 }
 179
 180 static uint16_t U_CALLCONV
 181 uprv_readDirectUInt16(uint16_t x) {
 182     return x;
 183 }
 184
 185 static uint32_t U_CALLCONV
 186 uprv_readSwapUInt32(uint32_t x) {
 187     return (uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24));
 188 }
 189
 190 static uint32_t U_CALLCONV
 191 uprv_readDirectUInt32(uint32_t x) {
 192     return x;
 193 }
 194
 195 static void U_CALLCONV
 196 uprv_writeSwapUInt16(uint16_t *p, uint16_t x) {
 197     *p=(uint16_t)((x<<8)|(x>>8));
 198 }
 199
 200 static void U_CALLCONV
 201 uprv_writeDirectUInt16(uint16_t *p, uint16_t x) {
 202     *p=x;
 203 }
 204
 205 static void U_CALLCONV
 206 uprv_writeSwapUInt32(uint32_t *p, uint32_t x) {
 207     *p=(uint32_t)((x<<24)|((x<<8)&0xff0000)|((x>>8)&0xff00)|(x>>24));
 208 }
 209
 210 static void U_CALLCONV
 211 uprv_writeDirectUInt32(uint32_t *p, uint32_t x) {
 212     *p=x;
 213 }
 214
 215 U_CAPI int16_t U_EXPORT2
 216 udata_readInt16(const UDataSwapper *ds, int16_t x) {
 217     return (int16_t)ds->readUInt16((uint16_t)x);
 218 }
 219
 220 U_CAPI int32_t U_EXPORT2
 221 udata_readInt32(const UDataSwapper *ds, int32_t x) {
 222     return (int32_t)ds->readUInt32((uint32_t)x);
 223 }
 224
 225 /**
 226  * Swap a block of invariant, NUL-terminated strings, but not padding
 227  * bytes after the last string.
 228  * @internal
 229  */
 230 U_CAPI int32_t U_EXPORT2
 231 udata_swapInvStringBlock(const UDataSwapper *ds,
 232                          const void *inData, int32_t length, void *outData,
 233                          UErrorCode *pErrorCode) {
 234     const char *inChars;
 235     int32_t stringsLength;
 236
 237     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 238         return 0;
 239     }
 240     if(ds==NULL || inData==NULL || length<0 || (length>0 && outData==NULL)) {
 241         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 242         return 0;
 243     }
 244
 245     /* reduce the strings length to not include bytes after the last NUL */
 246     inChars=(const char *)inData;
 247     stringsLength=length;
 248     while(stringsLength>0 && inChars[stringsLength-1]!=0) {
 249         --stringsLength;
 250     }
 251
 252     /* swap up to the last NUL */
 253     ds->swapInvChars(ds, inData, stringsLength, outData, pErrorCode);
 254
 255     /* copy the bytes after the last NUL */
 256     if(inData!=outData && length>stringsLength) {
 257         uprv_memcpy((char *)outData+stringsLength, inChars+stringsLength, length-stringsLength);
 258     }
 259
 260     /* return the length including padding bytes */
 261     if(U_SUCCESS(*pErrorCode)) {
 262         return length;
 263     } else {
 264         return 0;
 265     }
 266 }
 267
 268 U_CAPI void U_EXPORT2
 269 udata_printError(const UDataSwapper *ds,
 270                  const char *fmt,
 271                  ...) {
 272     va_list args;
 273
 274     if(ds->printError!=NULL) {
 275         va_start(args, fmt);
 276         ds->printError(ds->printErrorContext, fmt, args);
 277         va_end(args);
 278     }
 279 }
 280
 281 /* swap a data header ------------------------------------------------------- */
 282
 283 U_CAPI int32_t U_EXPORT2
 284 udata_swapDataHeader(const UDataSwapper *ds,
 285                      const void *inData, int32_t length, void *outData,
 286                      UErrorCode *pErrorCode) {
 287     const DataHeader *pHeader;
 288     uint16_t headerSize, infoSize;
 289
 290     /* argument checking */
 291     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 292         return 0;
 293     }
 294     if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) {
 295         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 296         return 0;
 297     }
 298
 299     /* check minimum length and magic bytes */
 300     pHeader=(const DataHeader *)inData;
 301     if( (length>=0 && length<sizeof(DataHeader)) ||
 302         pHeader->dataHeader.magic1!=0xda ||
 303         pHeader->dataHeader.magic2!=0x27 ||
 304         pHeader->info.sizeofUChar!=2
 305     ) {
 306         udata_printError(ds, "udata_swapDataHeader(): initial bytes do not look like ICU data\n");
 307         *pErrorCode=U_UNSUPPORTED_ERROR;
 308         return 0;
 309     }
 310
 311     headerSize=ds->readUInt16(pHeader->dataHeader.headerSize);
 312     infoSize=ds->readUInt16(pHeader->info.size);
 313
 314     if( headerSize<sizeof(DataHeader) ||
 315         infoSize<sizeof(UDataInfo) ||
 316         headerSize<(sizeof(pHeader->dataHeader)+infoSize) ||
 317         (length>=0 && length<headerSize)
 318     ) {
 319         udata_printError(ds, "udata_swapDataHeader(): header size mismatch - headerSize %d infoSize %d length %d\n",
 320                          headerSize, infoSize, length);
 321         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
 322         return 0;
 323     }
 324
 325     if(length>0) {
 326         DataHeader *outHeader;
 327         const char *s;
 328         int32_t maxLength;
 329
 330         /* Most of the fields are just bytes and need no swapping. */
 331         if(inData!=outData) {
 332             uprv_memcpy(outData, inData, headerSize);
 333         }
 334         outHeader=(DataHeader *)outData;
 335
 336         outHeader->info.isBigEndian = ds->outIsBigEndian;
 337         outHeader->info.charsetFamily = ds->outCharset;
 338
 339         /* swap headerSize */
 340         ds->swapArray16(ds, &pHeader->dataHeader.headerSize, 2, &outHeader->dataHeader.headerSize, pErrorCode);
 341
 342         /* swap UDataInfo size and reservedWord */
 343         ds->swapArray16(ds, &pHeader->info.size, 4, &outHeader->info.size, pErrorCode);
 344
 345         /* swap copyright statement after the UDataInfo */
 346         infoSize+=sizeof(pHeader->dataHeader);
 347         s=(const char *)inData+infoSize;
 348         maxLength=headerSize-infoSize;
 349         /* get the length of the string */
 350         for(length=0; length<maxLength && s[length]!=0; ++length) {}
 351         /* swap the string contents */
 352         ds->swapInvChars(ds, s, length, (char *)outData+infoSize, pErrorCode);
 353     }
 354
 355     return headerSize;
 356 }
 357
 358 /* API functions ------------------------------------------------------------ */
 359
 360 U_CAPI UDataSwapper * U_EXPORT2
 361 udata_openSwapper(UBool inIsBigEndian, uint8_t inCharset,
 362                   UBool outIsBigEndian, uint8_t outCharset,
 363                   UErrorCode *pErrorCode) {
 364     UDataSwapper *swapper;
 365
 366     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 367         return NULL;
 368     }
 369     if(inCharset>U_EBCDIC_FAMILY || outCharset>U_EBCDIC_FAMILY) {
 370         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 371         return NULL;
 372     }
 373
 374     /* allocate the swapper */
 375     swapper=uprv_malloc(sizeof(UDataSwapper));
 376     if(swapper==NULL) {
 377         *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
 378         return NULL;
 379     }
 380     uprv_memset(swapper, 0, sizeof(UDataSwapper));
 381
 382     /* set values and functions pointers according to in/out parameters */
 383     swapper->inIsBigEndian=inIsBigEndian;
 384     swapper->inCharset=inCharset;
 385     swapper->outIsBigEndian=outIsBigEndian;
 386     swapper->outCharset=outCharset;
 387
 388     swapper->readUInt16= inIsBigEndian==U_IS_BIG_ENDIAN ? uprv_readDirectUInt16 : uprv_readSwapUInt16;
 389     swapper->readUInt32= inIsBigEndian==U_IS_BIG_ENDIAN ? uprv_readDirectUInt32 : uprv_readSwapUInt32;
 390
 391     swapper->writeUInt16= outIsBigEndian==U_IS_BIG_ENDIAN ? uprv_writeDirectUInt16 : uprv_writeSwapUInt16;
 392     swapper->writeUInt32= outIsBigEndian==U_IS_BIG_ENDIAN ? uprv_writeDirectUInt32 : uprv_writeSwapUInt32;
 393
 394     swapper->compareInvChars= outCharset==U_ASCII_FAMILY ? uprv_compareInvAscii : uprv_compareInvEbcdic;
 395
 396     if(inIsBigEndian==outIsBigEndian) {
 397         swapper->swapArray16=uprv_copyArray16;
 398         swapper->swapArray32=uprv_copyArray32;
 399         swapper->swapArray64=uprv_copyArray64;
 400     } else {
 401         swapper->swapArray16=uprv_swapArray16;
 402         swapper->swapArray32=uprv_swapArray32;
 403         swapper->swapArray64=uprv_swapArray64;
 404     }
 405
 406     if(inCharset==U_ASCII_FAMILY) {
 407         swapper->swapInvChars= outCharset==U_ASCII_FAMILY ? uprv_copyAscii : uprv_ebcdicFromAscii;
 408     } else /* U_EBCDIC_FAMILY */ {
 409         swapper->swapInvChars= outCharset==U_EBCDIC_FAMILY ? uprv_copyEbcdic : uprv_asciiFromEbcdic;
 410     }
 411
 412     return swapper;
 413 }
 414
 415 U_CAPI UDataSwapper * U_EXPORT2
 416 udata_openSwapperForInputData(const void *data, int32_t length,
 417                               UBool outIsBigEndian, uint8_t outCharset,
 418                               UErrorCode *pErrorCode) {
 419     const DataHeader *pHeader;
 420     uint16_t headerSize, infoSize;
 421     UBool inIsBigEndian;
 422     int8_t inCharset;
 423
 424     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 425         return NULL;
 426     }
 427     if( data==NULL ||
 428         (length>=0 && length<sizeof(DataHeader)) ||
 429         outCharset>U_EBCDIC_FAMILY
 430     ) {
 431         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 432         return NULL;
 433     }
 434
 435     pHeader=(const DataHeader *)data;
 436     if( (length>=0 && length<sizeof(DataHeader)) ||
 437         pHeader->dataHeader.magic1!=0xda ||
 438         pHeader->dataHeader.magic2!=0x27 ||
 439         pHeader->info.sizeofUChar!=2
 440     ) {
 441         *pErrorCode=U_UNSUPPORTED_ERROR;
 442         return 0;
 443     }
 444
 445     inIsBigEndian=(UBool)pHeader->info.isBigEndian;
 446     inCharset=pHeader->info.charsetFamily;
 447
 448     if(inIsBigEndian==U_IS_BIG_ENDIAN) {
 449         headerSize=pHeader->dataHeader.headerSize;
 450         infoSize=pHeader->info.size;
 451     } else {
 452         headerSize=uprv_readSwapUInt16(pHeader->dataHeader.headerSize);
 453         infoSize=uprv_readSwapUInt16(pHeader->info.size);
 454     }
 455
 456     if( headerSize<sizeof(DataHeader) ||
 457         infoSize<sizeof(UDataInfo) ||
 458         headerSize<(sizeof(pHeader->dataHeader)+infoSize) ||
 459         (length>=0 && length<headerSize)
 460     ) {
 461         *pErrorCode=U_UNSUPPORTED_ERROR;
 462         return 0;
 463     }
 464
 465     return udata_openSwapper(inIsBigEndian, inCharset, outIsBigEndian, outCharset, pErrorCode);
 466 }
 467
 468 U_CAPI void U_EXPORT2
 469 udata_closeSwapper(UDataSwapper *ds) {
 470     uprv_free(ds);
 471 }