icuSources/test/intltest/idnaref.cpp

   1 /*
   2  *******************************************************************************
   3  *
   4  *   Copyright (C) 2003-2007, International Business Machines
   5  *   Corporation and others.  All Rights Reserved.
   6  *
   7  *******************************************************************************
   8  *   file name:  idnaref.cpp
   9  *   encoding:   US-ASCII
  10  *   tab size:   8 (not used)
  11  *   indentation:4
  12  *
  13  *   created on: 2003feb1
  14  *   created by: Ram Viswanadha
  15  */
  16
  17 #include "unicode/utypes.h"
  18
  19 #if !UCONFIG_NO_IDNA && !UCONFIG_NO_TRANSLITERATION
  20 #include "idnaref.h"
  21 #include "punyref.h"
  22 #include "ustr_imp.h"
  23 #include "cmemory.h"
  24 #include "sprpimpl.h"
  25 #include "nptrans.h"
  26 #include "testidna.h"
  27 #include "punycode.h"
  28 #include "unicode/ustring.h"
  29
  30 /* it is official IDNA ACE Prefix is "xn--" */
  31 static const UChar ACE_PREFIX[] ={ 0x0078,0x006E,0x002d,0x002d } ;
  32 #define ACE_PREFIX_LENGTH 4
  33
  34 #define MAX_LABEL_LENGTH 63
  35 #define HYPHEN      0x002D
  36 /* The Max length of the labels should not be more than 64 */
  37 #define MAX_LABEL_BUFFER_SIZE 100
  38 #define MAX_IDN_BUFFER_SIZE   300
  39
  40 #define CAPITAL_A        0x0041
  41 #define CAPITAL_Z        0x005A
  42 #define LOWER_CASE_DELTA 0x0020
  43 #define FULL_STOP        0x002E
  44
  45
  46 inline static UBool
  47 startsWithPrefix(const UChar* src , int32_t srcLength){
  48     UBool startsWithPrefix = TRUE;
  49
  50     if(srcLength < ACE_PREFIX_LENGTH){
  51         return FALSE;
  52     }
  53
  54     for(int8_t i=0; i< ACE_PREFIX_LENGTH; i++){
  55         if(u_tolower(src[i]) != ACE_PREFIX[i]){
  56             startsWithPrefix = FALSE;
  57         }
  58     }
  59     return startsWithPrefix;
  60 }
  61
  62 inline static UChar
  63 toASCIILower(UChar ch){
  64     if(CAPITAL_A <= ch && ch <= CAPITAL_Z){
  65         return ch + LOWER_CASE_DELTA;
  66     }
  67     return ch;
  68 }
  69
  70 inline static int32_t
  71 compareCaseInsensitiveASCII(const UChar* s1, int32_t s1Len,
  72                             const UChar* s2, int32_t s2Len){
  73     if(s1Len != s2Len){
  74         return (s1Len > s2Len) ? s1Len : s2Len;
  75     }
  76     UChar c1,c2;
  77     int32_t rc;
  78
  79     for(int32_t i =0;/* no condition */;i++) {
  80         /* If we reach the ends of both strings then they match */
  81         if(i == s1Len) {
  82             return 0;
  83         }
  84
  85         c1 = s1[i];
  86         c2 = s2[i];
  87
  88         /* Case-insensitive comparison */
  89         if(c1!=c2) {
  90             rc=(int32_t)toASCIILower(c1)-(int32_t)toASCIILower(c2);
  91             if(rc!=0) {
  92                 return rc;
  93             }
  94         }
  95     }
  96
  97 }
  98
  99 static UErrorCode getError(enum punycode_status status){
 100     switch(status){
 101     case punycode_success:
 102         return U_ZERO_ERROR;
 103     case punycode_bad_input:   /* Input is invalid.                         */
 104         return U_INVALID_CHAR_FOUND;
 105     case punycode_big_output:  /* Output would exceed the space provided.   */
 106         return U_BUFFER_OVERFLOW_ERROR;
 107     case punycode_overflow :    /* Input requires wider integers to process. */
 108         return U_INDEX_OUTOFBOUNDS_ERROR;
 109     default:
 110         return U_INTERNAL_PROGRAM_ERROR;
 111     }
 112 }
 113
 114 static inline int32_t convertASCIIToUChars(const char* src,UChar* dest, int32_t length){
 115     int i;
 116     for(i=0;i<length;i++){
 117         dest[i] = src[i];
 118     }
 119     return i;
 120 }
 121 static inline int32_t convertUCharsToASCII(const UChar* src,char* dest, int32_t length){
 122     int i;
 123     for(i=0;i<length;i++){
 124         dest[i] = (char)src[i];
 125     }
 126     return i;
 127 }
 128 // wrapper around the reference Punycode implementation
 129 static int32_t convertToPuny(const UChar* src, int32_t srcLength,
 130                              UChar* dest, int32_t destCapacity,
 131                              UErrorCode& status){
 132     uint32_t b1Stack[MAX_LABEL_BUFFER_SIZE];
 133     int32_t b1Len = 0, b1Capacity = MAX_LABEL_BUFFER_SIZE;
 134     uint32_t* b1 = b1Stack;
 135     char b2Stack[MAX_LABEL_BUFFER_SIZE];
 136     char* b2 = b2Stack;
 137     int32_t b2Len =MAX_LABEL_BUFFER_SIZE ;
 138     punycode_status error;
 139     unsigned char* caseFlags = NULL;
 140
 141     u_strToUTF32((UChar32*)b1,b1Capacity,&b1Len,src,srcLength,&status);
 142     if(status == U_BUFFER_OVERFLOW_ERROR){
 143         // redo processing of string
 144         /* we do not have enough room so grow the buffer*/
 145         b1 =  (uint32_t*) uprv_malloc(b1Len * sizeof(uint32_t));
 146         if(b1==NULL){
 147             status = U_MEMORY_ALLOCATION_ERROR;
 148             goto CLEANUP;
 149         }
 150
 151         status = U_ZERO_ERROR; // reset error
 152
 153         u_strToUTF32((UChar32*)b1,b1Len,&b1Len,src,srcLength,&status);
 154     }
 155     if(U_FAILURE(status)){
 156         goto CLEANUP;
 157     }
 158
 159     //caseFlags = (unsigned char*) uprv_malloc(b1Len *sizeof(unsigned char));
 160
 161     error = punycode_encode(b1Len,b1,caseFlags, (uint32_t*)&b2Len, b2);
 162     status = getError(error);
 163
 164     if(status == U_BUFFER_OVERFLOW_ERROR){
 165         /* we do not have enough room so grow the buffer*/
 166         b2 = (char*) uprv_malloc( b2Len * sizeof(char));
 167         if(b2==NULL){
 168             status = U_MEMORY_ALLOCATION_ERROR;
 169             goto CLEANUP;
 170         }
 171
 172         status = U_ZERO_ERROR; // reset error
 173
 174         punycode_status error = punycode_encode(b1Len,b1,caseFlags, (uint32_t*)&b2Len, b2);
 175         status = getError(error);
 176     }
 177     if(U_FAILURE(status)){
 178         goto CLEANUP;
 179     }
 180
 181     if(b2Len < destCapacity){
 182           convertASCIIToUChars(b2,dest,b2Len);
 183     }else{
 184         status =U_BUFFER_OVERFLOW_ERROR;
 185     }
 186
 187 CLEANUP:
 188     if(b1Stack != b1){
 189         uprv_free(b1);
 190     }
 191     if(b2Stack != b2){
 192         uprv_free(b2);
 193     }
 194     uprv_free(caseFlags);
 195
 196     return b2Len;
 197 }
 198
 199 static int32_t convertFromPuny(  const UChar* src, int32_t srcLength,
 200                                  UChar* dest, int32_t destCapacity,
 201                                  UErrorCode& status){
 202     char b1Stack[MAX_LABEL_BUFFER_SIZE];
 203     char* b1 = b1Stack;
 204     int32_t destLen =0;
 205
 206     convertUCharsToASCII(src, b1,srcLength);
 207
 208     uint32_t b2Stack[MAX_LABEL_BUFFER_SIZE];
 209     uint32_t* b2 = b2Stack;
 210     int32_t b2Len =MAX_LABEL_BUFFER_SIZE;
 211     unsigned char* caseFlags = NULL; //(unsigned char*) uprv_malloc(srcLength * sizeof(unsigned char*));
 212     punycode_status error = punycode_decode(srcLength,b1,(uint32_t*)&b2Len,b2,caseFlags);
 213     status = getError(error);
 214     if(status == U_BUFFER_OVERFLOW_ERROR){
 215         b2 =  (uint32_t*) uprv_malloc(b2Len * sizeof(uint32_t));
 216         if(b2 == NULL){
 217             status = U_MEMORY_ALLOCATION_ERROR;
 218             goto CLEANUP;
 219         }
 220         error = punycode_decode(srcLength,b1,(uint32_t*)&b2Len,b2,caseFlags);
 221         status = getError(error);
 222     }
 223
 224     if(U_FAILURE(status)){
 225         goto CLEANUP;
 226     }
 227
 228     u_strFromUTF32(dest,destCapacity,&destLen,(UChar32*)b2,b2Len,&status);
 229
 230 CLEANUP:
 231     if(b1Stack != b1){
 232         uprv_free(b1);
 233     }
 234     if(b2Stack != b2){
 235         uprv_free(b2);
 236     }
 237     uprv_free(caseFlags);
 238
 239     return destLen;
 240 }
 241
 242
 243 U_CFUNC int32_t U_EXPORT2
 244 idnaref_toASCII(const UChar* src, int32_t srcLength,
 245               UChar* dest, int32_t destCapacity,
 246               int32_t options,
 247               UParseError* parseError,
 248               UErrorCode* status){
 249
 250     if(status == NULL || U_FAILURE(*status)){
 251         return 0;
 252     }
 253     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 254         *status = U_ILLEGAL_ARGUMENT_ERROR;
 255         return 0;
 256     }
 257     UChar b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE];
 258     //initialize pointers to stack buffers
 259     UChar  *b1 = b1Stack, *b2 = b2Stack;
 260     int32_t b1Len=0, b2Len=0,
 261             b1Capacity = MAX_LABEL_BUFFER_SIZE,
 262             b2Capacity = MAX_LABEL_BUFFER_SIZE ,
 263             reqLength=0;
 264
 265     //get the options
 266     UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 267     UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 268
 269     UBool* caseFlags = NULL;
 270
 271     // assume the source contains all ascii codepoints
 272     UBool srcIsASCII  = TRUE;
 273     // assume the source contains all LDH codepoints
 274     UBool srcIsLDH = TRUE;
 275     int32_t j=0;
 276
 277     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 278
 279     if(U_FAILURE(*status)){
 280         goto CLEANUP;
 281     }
 282
 283     if(srcLength == -1){
 284         srcLength = u_strlen(src);
 285     }
 286     if(srcLength > b1Capacity){
 287         b1 = (UChar*) uprv_malloc(srcLength * U_SIZEOF_UCHAR);
 288         if(b1==NULL){
 289             *status = U_MEMORY_ALLOCATION_ERROR;
 290             goto CLEANUP;
 291         }
 292         b1Capacity = srcLength;
 293     }
 294     // step 1
 295     for( j=0;j<srcLength;j++){
 296         if(src[j] > 0x7F){
 297             srcIsASCII = FALSE;
 298         }
 299         b1[b1Len++] = src[j];
 300     }
 301
 302     // step 2
 303     b1Len = prep->process(src,srcLength,b1, b1Capacity,allowUnassigned,parseError,*status);
 304
 305     if(*status == U_BUFFER_OVERFLOW_ERROR){
 306         // redo processing of string
 307         /* we do not have enough room so grow the buffer*/
 308          if(b1 != b1Stack){
 309                 uprv_free(b1);
 310         }
 311         b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 312         if(b1==NULL){
 313             *status = U_MEMORY_ALLOCATION_ERROR;
 314             goto CLEANUP;
 315         }
 316
 317         *status = U_ZERO_ERROR; // reset error
 318
 319         b1Len = prep->process(src,srcLength,b1, b1Len,allowUnassigned, parseError, *status);
 320     }
 321     // error bail out
 322     if(U_FAILURE(*status)){
 323         goto CLEANUP;
 324     }
 325
 326     if(b1Len == 0){
 327         *status = U_IDNA_ZERO_LENGTH_LABEL_ERROR;
 328         goto CLEANUP;
 329     }
 330
 331     srcIsASCII = TRUE;
 332     // step 3 & 4
 333     for( j=0;j<b1Len;j++){
 334         if(b1[j] > 0x7F){// check if output of usprep_prepare is all ASCII
 335             srcIsASCII = FALSE;
 336         }else if(prep->isLDHChar(b1[j])==FALSE){  // if the char is in ASCII range verify that it is an LDH character{
 337             srcIsLDH = FALSE;
 338         }
 339     }
 340
 341     if(useSTD3ASCIIRules == TRUE){
 342         // verify 3a and 3b
 343         if( srcIsLDH == FALSE /* source contains some non-LDH characters */
 344             || b1[0] ==  HYPHEN || b1[b1Len-1] == HYPHEN){
 345             *status = U_IDNA_STD3_ASCII_RULES_ERROR;
 346             goto CLEANUP;
 347         }
 348     }
 349     if(srcIsASCII){
 350         if(b1Len <= destCapacity){
 351             uprv_memmove(dest, b1, b1Len * U_SIZEOF_UCHAR);
 352             reqLength = b1Len;
 353         }else{
 354             reqLength = b1Len;
 355             goto CLEANUP;
 356         }
 357     }else{
 358         // step 5 : verify the sequence does not begin with ACE prefix
 359         if(!startsWithPrefix(b1,b1Len)){
 360
 361             //step 6: encode the sequence with punycode
 362             //caseFlags = (UBool*) uprv_malloc(b1Len * sizeof(UBool));
 363
 364             b2Len = convertToPuny(b1,b1Len, b2,b2Capacity,*status);
 365             //b2Len = u_strToPunycode(b2,b2Capacity,b1,b1Len, caseFlags, status);
 366             if(*status == U_BUFFER_OVERFLOW_ERROR){
 367                 // redo processing of string
 368                 /* we do not have enough room so grow the buffer*/
 369                 b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
 370                 if(b2 == NULL){
 371                     *status = U_MEMORY_ALLOCATION_ERROR;
 372                     goto CLEANUP;
 373                 }
 374
 375                 *status = U_ZERO_ERROR; // reset error
 376
 377                 b2Len = convertToPuny(b1, b1Len, b2, b2Len, *status);
 378                 //b2Len = u_strToPunycode(b2,b2Len,b1,b1Len, caseFlags, status);
 379
 380             }
 381             //error bail out
 382             if(U_FAILURE(*status)){
 383                 goto CLEANUP;
 384             }
 385             reqLength = b2Len+ACE_PREFIX_LENGTH;
 386
 387             if(reqLength > destCapacity){
 388                 *status = U_BUFFER_OVERFLOW_ERROR;
 389                 goto CLEANUP;
 390             }
 391             //Step 7: prepend the ACE prefix
 392             uprv_memcpy(dest,ACE_PREFIX,ACE_PREFIX_LENGTH * U_SIZEOF_UCHAR);
 393             //Step 6: copy the contents in b2 into dest
 394             uprv_memcpy(dest+ACE_PREFIX_LENGTH, b2, b2Len * U_SIZEOF_UCHAR);
 395
 396         }else{
 397             *status = U_IDNA_ACE_PREFIX_ERROR;
 398             goto CLEANUP;
 399         }
 400     }
 401
 402     if(reqLength > MAX_LABEL_LENGTH){
 403         *status = U_IDNA_LABEL_TOO_LONG_ERROR;
 404     }
 405
 406 CLEANUP:
 407     if(b1 != b1Stack){
 408         uprv_free(b1);
 409     }
 410     if(b2 != b2Stack){
 411         uprv_free(b2);
 412     }
 413     uprv_free(caseFlags);
 414
 415 //    delete prep;
 416
 417     return u_terminateUChars(dest, destCapacity, reqLength, status);
 418 }
 419
 420
 421 U_CFUNC int32_t U_EXPORT2
 422 idnaref_toUnicode(const UChar* src, int32_t srcLength,
 423                 UChar* dest, int32_t destCapacity,
 424                 int32_t options,
 425                 UParseError* parseError,
 426                 UErrorCode* status){
 427
 428     if(status == NULL || U_FAILURE(*status)){
 429         return 0;
 430     }
 431     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 432         *status = U_ILLEGAL_ARGUMENT_ERROR;
 433         return 0;
 434     }
 435
 436
 437
 438     UChar b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE], b3Stack[MAX_LABEL_BUFFER_SIZE];
 439
 440     //initialize pointers to stack buffers
 441     UChar  *b1 = b1Stack, *b2 = b2Stack, *b1Prime=NULL, *b3=b3Stack;
 442     int32_t b1Len, b2Len, b1PrimeLen, b3Len,
 443             b1Capacity = MAX_LABEL_BUFFER_SIZE,
 444             b2Capacity = MAX_LABEL_BUFFER_SIZE,
 445             b3Capacity = MAX_LABEL_BUFFER_SIZE,
 446             reqLength=0;
 447 //    UParseError parseError;
 448
 449     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 450     b1Len = 0;
 451     UBool* caseFlags = NULL;
 452
 453     //get the options
 454     UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 455     UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 456
 457     UBool srcIsASCII = TRUE;
 458     UBool srcIsLDH = TRUE;
 459     int32_t failPos =0;
 460
 461     if(U_FAILURE(*status)){
 462         goto CLEANUP;
 463     }
 464     // step 1: find out if all the codepoints in src are ASCII
 465     if(srcLength==-1){
 466         srcLength = 0;
 467         for(;src[srcLength]!=0;){
 468             if(src[srcLength]> 0x7f){
 469                 srcIsASCII = FALSE;
 470             }if(prep->isLDHChar(src[srcLength])==FALSE){
 471                 // here we do not assemble surrogates
 472                 // since we know that LDH code points
 473                 // are in the ASCII range only
 474                 srcIsLDH = FALSE;
 475                 failPos = srcLength;
 476             }
 477             srcLength++;
 478         }
 479     }else{
 480         for(int32_t j=0; j<srcLength; j++){
 481             if(src[j]> 0x7f){
 482                 srcIsASCII = FALSE;
 483             }else if(prep->isLDHChar(src[j])==FALSE){
 484                 // here we do not assemble surrogates
 485                 // since we know that LDH code points
 486                 // are in the ASCII range only
 487                 srcIsLDH = FALSE;
 488                 failPos = j;
 489             }
 490         }
 491     }
 492
 493     if(srcIsASCII == FALSE){
 494         // step 2: process the string
 495         b1Len = prep->process(src,srcLength,b1,b1Capacity,allowUnassigned, parseError, *status);
 496         if(*status == U_BUFFER_OVERFLOW_ERROR){
 497             // redo processing of string
 498             /* we do not have enough room so grow the buffer*/
 499             b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 500             if(b1==NULL){
 501                 *status = U_MEMORY_ALLOCATION_ERROR;
 502                 goto CLEANUP;
 503             }
 504
 505             *status = U_ZERO_ERROR; // reset error
 506
 507             b1Len = prep->process(src,srcLength,b1, b1Len,allowUnassigned, parseError, *status);
 508         }
 509         //bail out on error
 510         if(U_FAILURE(*status)){
 511             goto CLEANUP;
 512         }
 513     }else{
 514
 515         // copy everything to b1
 516         if(srcLength < b1Capacity){
 517             uprv_memmove(b1,src, srcLength * U_SIZEOF_UCHAR);
 518         }else{
 519             /* we do not have enough room so grow the buffer*/
 520             b1 = (UChar*) uprv_malloc(srcLength * U_SIZEOF_UCHAR);
 521             if(b1==NULL){
 522                 *status = U_MEMORY_ALLOCATION_ERROR;
 523                 goto CLEANUP;
 524             }
 525             uprv_memmove(b1,src, srcLength * U_SIZEOF_UCHAR);
 526         }
 527         b1Len = srcLength;
 528     }
 529     //step 3: verify ACE Prefix
 530     if(startsWithPrefix(src,srcLength)){
 531
 532         //step 4: Remove the ACE Prefix
 533         b1Prime = b1 + ACE_PREFIX_LENGTH;
 534         b1PrimeLen  = b1Len - ACE_PREFIX_LENGTH;
 535
 536         //step 5: Decode using punycode
 537         b2Len = convertFromPuny(b1Prime,b1PrimeLen, b2, b2Capacity, *status);
 538         //b2Len = u_strFromPunycode(b2, b2Capacity,b1Prime,b1PrimeLen, caseFlags, status);
 539
 540         if(*status == U_BUFFER_OVERFLOW_ERROR){
 541             // redo processing of string
 542             /* we do not have enough room so grow the buffer*/
 543             b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
 544             if(b2==NULL){
 545                 *status = U_MEMORY_ALLOCATION_ERROR;
 546                 goto CLEANUP;
 547             }
 548
 549             *status = U_ZERO_ERROR; // reset error
 550
 551             b2Len =  convertFromPuny(b1Prime,b1PrimeLen, b2, b2Len, *status);
 552             //b2Len = u_strFromPunycode(b2, b2Len,b1Prime,b1PrimeLen,caseFlags, status);
 553         }
 554
 555
 556         //step 6:Apply toASCII
 557         b3Len = idnaref_toASCII(b2,b2Len,b3,b3Capacity,options,parseError, status);
 558
 559         if(*status == U_BUFFER_OVERFLOW_ERROR){
 560             // redo processing of string
 561             /* we do not have enough room so grow the buffer*/
 562             b3 = (UChar*) uprv_malloc(b3Len * U_SIZEOF_UCHAR);
 563             if(b3==NULL){
 564                 *status = U_MEMORY_ALLOCATION_ERROR;
 565                 goto CLEANUP;
 566             }
 567
 568             *status = U_ZERO_ERROR; // reset error
 569
 570             b3Len =  idnaref_toASCII(b2,b2Len,b3,b3Len, options, parseError, status);
 571
 572         }
 573         //bail out on error
 574         if(U_FAILURE(*status)){
 575             goto CLEANUP;
 576         }
 577
 578         //step 7: verify
 579         if(compareCaseInsensitiveASCII(b1, b1Len, b3, b3Len) !=0){
 580             *status = U_IDNA_VERIFICATION_ERROR;
 581             goto CLEANUP;
 582         }
 583
 584         //step 8: return output of step 5
 585         reqLength = b2Len;
 586         if(b2Len <= destCapacity) {
 587             uprv_memmove(dest, b2, b2Len * U_SIZEOF_UCHAR);
 588         }
 589     }else{
 590         // verify that STD3 ASCII rules are satisfied
 591         if(useSTD3ASCIIRules == TRUE){
 592             if( srcIsLDH == FALSE /* source contains some non-LDH characters */
 593                 || src[0] ==  HYPHEN || src[srcLength-1] == HYPHEN){
 594                 *status = U_IDNA_STD3_ASCII_RULES_ERROR;
 595
 596                 /* populate the parseError struct */
 597                 if(srcIsLDH==FALSE){
 598                     // failPos is always set the index of failure
 599                     uprv_syntaxError(src,failPos, srcLength,parseError);
 600                 }else if(src[0] == HYPHEN){
 601                     // fail position is 0
 602                     uprv_syntaxError(src,0,srcLength,parseError);
 603                 }else{
 604                     // the last index in the source is always length-1
 605                     uprv_syntaxError(src, (srcLength>0) ? srcLength-1 : srcLength, srcLength,parseError);
 606                 }
 607
 608                 goto CLEANUP;
 609             }
 610         }
 611         //copy the source to destination
 612         if(srcLength <= destCapacity){
 613             uprv_memmove(dest,src,srcLength * U_SIZEOF_UCHAR);
 614         }
 615         reqLength = srcLength;
 616     }
 617
 618 CLEANUP:
 619
 620     if(b1 != b1Stack){
 621         uprv_free(b1);
 622     }
 623     if(b2 != b2Stack){
 624         uprv_free(b2);
 625     }
 626     uprv_free(caseFlags);
 627
 628 //    delete prep;
 629
 630     return u_terminateUChars(dest, destCapacity, reqLength, status);
 631 }
 632
 633
 634 static int32_t
 635 getNextSeparator(UChar *src,int32_t srcLength,NamePrepTransform* prep,
 636                  UChar **limit,
 637                  UBool *done,
 638                  UErrorCode *status){
 639     if(srcLength == -1){
 640         int32_t i;
 641         for(i=0 ; ;i++){
 642             if(src[i] == 0){
 643                 *limit = src + i; // point to null
 644                 *done = TRUE;
 645                 return i;
 646             }
 647             if(prep->isLabelSeparator(src[i],*status)){
 648                 *limit = src + (i+1); // go past the delimiter
 649                 return i;
 650
 651             }
 652         }
 653     }else{
 654         int32_t i;
 655         for(i=0;i<srcLength;i++){
 656             if(prep->isLabelSeparator(src[i],*status)){
 657                 *limit = src + (i+1); // go past the delimiter
 658                 return i;
 659             }
 660         }
 661         // we have not found the delimiter
 662         if(i==srcLength){
 663             *limit = src+srcLength;
 664             *done = TRUE;
 665         }
 666         return i;
 667     }
 668 }
 669
 670 U_CFUNC int32_t U_EXPORT2
 671 idnaref_IDNToASCII(  const UChar* src, int32_t srcLength,
 672                    UChar* dest, int32_t destCapacity,
 673                    int32_t options,
 674                    UParseError* parseError,
 675                    UErrorCode* status){
 676
 677     if(status == NULL || U_FAILURE(*status)){
 678         return 0;
 679     }
 680     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 681         *status = U_ILLEGAL_ARGUMENT_ERROR;
 682         return 0;
 683     }
 684
 685     int32_t reqLength = 0;
 686 //    UParseError parseError;
 687
 688     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 689
 690     //initialize pointers to stack buffers
 691     UChar b1Stack[MAX_LABEL_BUFFER_SIZE];
 692     UChar  *b1 = b1Stack;
 693     int32_t b1Len, labelLen;
 694     UChar* delimiter = (UChar*)src;
 695     UChar* labelStart = (UChar*)src;
 696     int32_t remainingLen = srcLength;
 697     int32_t b1Capacity = MAX_LABEL_BUFFER_SIZE;
 698
 699     //get the options
 700 //    UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 701 //    UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 702     UBool done = FALSE;
 703
 704     if(U_FAILURE(*status)){
 705         goto CLEANUP;
 706     }
 707
 708
 709     if(srcLength == -1){
 710         for(;;){
 711
 712             if(*delimiter == 0){
 713                 break;
 714             }
 715
 716             labelLen = getNextSeparator(labelStart, -1, prep, &delimiter, &done, status);
 717             b1Len = 0;
 718             if(!(labelLen==0 && done)){// make sure this is not a root label separator.
 719
 720                 b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Capacity,
 721                                         options, parseError, status);
 722
 723                 if(*status == U_BUFFER_OVERFLOW_ERROR){
 724                     // redo processing of string
 725                     /* we do not have enough room so grow the buffer*/
 726                     b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 727                     if(b1==NULL){
 728                         *status = U_MEMORY_ALLOCATION_ERROR;
 729                         goto CLEANUP;
 730                     }
 731
 732                     *status = U_ZERO_ERROR; // reset error
 733
 734                     b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Len,
 735                                             options, parseError, status);
 736
 737                 }
 738             }
 739
 740             if(U_FAILURE(*status)){
 741                 goto CLEANUP;
 742             }
 743             int32_t tempLen = (reqLength + b1Len );
 744             // copy to dest
 745             if( tempLen< destCapacity){
 746                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 747             }
 748
 749             reqLength = tempLen;
 750
 751             // add the label separator
 752             if(done == FALSE){
 753                 if(reqLength < destCapacity){
 754                     dest[reqLength] = FULL_STOP;
 755                 }
 756                 reqLength++;
 757             }
 758
 759             labelStart = delimiter;
 760         }
 761     }else{
 762         for(;;){
 763
 764             if(delimiter == src+srcLength){
 765                 break;
 766             }
 767
 768             labelLen = getNextSeparator(labelStart, remainingLen, prep, &delimiter, &done, status);
 769
 770             b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Capacity,
 771                                     options,parseError, status);
 772
 773             if(*status == U_BUFFER_OVERFLOW_ERROR){
 774                 // redo processing of string
 775                 /* we do not have enough room so grow the buffer*/
 776                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 777                 if(b1==NULL){
 778                     *status = U_MEMORY_ALLOCATION_ERROR;
 779                     goto CLEANUP;
 780                 }
 781
 782                 *status = U_ZERO_ERROR; // reset error
 783
 784                 b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Len,
 785                                         options, parseError, status);
 786
 787             }
 788
 789             if(U_FAILURE(*status)){
 790                 goto CLEANUP;
 791             }
 792             int32_t tempLen = (reqLength + b1Len );
 793             // copy to dest
 794             if( tempLen< destCapacity){
 795                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 796             }
 797
 798             reqLength = tempLen;
 799
 800             // add the label separator
 801             if(done == FALSE){
 802                 if(reqLength < destCapacity){
 803                     dest[reqLength] = FULL_STOP;
 804                 }
 805                 reqLength++;
 806             }
 807
 808             labelStart = delimiter;
 809             remainingLen = srcLength - (delimiter - src);
 810         }
 811     }
 812
 813
 814 CLEANUP:
 815
 816     if(b1 != b1Stack){
 817         uprv_free(b1);
 818     }
 819
 820 //   delete prep;
 821
 822     return u_terminateUChars(dest, destCapacity, reqLength, status);
 823 }
 824
 825 U_CFUNC int32_t U_EXPORT2
 826 idnaref_IDNToUnicode(  const UChar* src, int32_t srcLength,
 827                      UChar* dest, int32_t destCapacity,
 828                      int32_t options,
 829                      UParseError* parseError,
 830                      UErrorCode* status){
 831
 832     if(status == NULL || U_FAILURE(*status)){
 833         return 0;
 834     }
 835     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 836         *status = U_ILLEGAL_ARGUMENT_ERROR;
 837         return 0;
 838     }
 839
 840     int32_t reqLength = 0;
 841
 842     UBool done = FALSE;
 843
 844     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 845
 846     //initialize pointers to stack buffers
 847     UChar b1Stack[MAX_LABEL_BUFFER_SIZE];
 848     UChar  *b1 = b1Stack;
 849     int32_t b1Len, labelLen;
 850     UChar* delimiter = (UChar*)src;
 851     UChar* labelStart = (UChar*)src;
 852     int32_t remainingLen = srcLength;
 853     int32_t b1Capacity = MAX_LABEL_BUFFER_SIZE;
 854
 855     //get the options
 856 //    UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 857 //    UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 858
 859     if(U_FAILURE(*status)){
 860         goto CLEANUP;
 861     }
 862
 863     if(srcLength == -1){
 864         for(;;){
 865
 866             if(*delimiter == 0){
 867                 break;
 868             }
 869
 870             labelLen = getNextSeparator(labelStart, -1, prep, &delimiter, &done, status);
 871
 872            if(labelLen==0 && done==FALSE){
 873                 *status = U_IDNA_ZERO_LENGTH_LABEL_ERROR;
 874             }
 875             b1Len = idnaref_toUnicode(labelStart, labelLen, b1, b1Capacity,
 876                                       options, parseError, status);
 877
 878             if(*status == U_BUFFER_OVERFLOW_ERROR){
 879                 // redo processing of string
 880                 /* we do not have enough room so grow the buffer*/
 881                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 882                 if(b1==NULL){
 883                     *status = U_MEMORY_ALLOCATION_ERROR;
 884                     goto CLEANUP;
 885                 }
 886
 887                 *status = U_ZERO_ERROR; // reset error
 888
 889                 b1Len = idnaref_toUnicode( labelStart, labelLen, b1, b1Len,
 890                                            options, parseError, status);
 891
 892             }
 893
 894             if(U_FAILURE(*status)){
 895                 goto CLEANUP;
 896             }
 897             int32_t tempLen = (reqLength + b1Len );
 898             // copy to dest
 899             if( tempLen< destCapacity){
 900                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 901             }
 902
 903             reqLength = tempLen;
 904             // add the label separator
 905             if(done == FALSE){
 906                 if(reqLength < destCapacity){
 907                     dest[reqLength] = FULL_STOP;
 908                 }
 909                 reqLength++;
 910             }
 911
 912             labelStart = delimiter;
 913         }
 914     }else{
 915         for(;;){
 916
 917             if(delimiter == src+srcLength){
 918                 break;
 919             }
 920
 921             labelLen = getNextSeparator(labelStart, remainingLen, prep, &delimiter, &done, status);
 922
 923             if(labelLen==0 && done==FALSE){
 924                 *status = U_IDNA_ZERO_LENGTH_LABEL_ERROR;
 925             }
 926
 927             b1Len = idnaref_toUnicode( labelStart,labelLen, b1, b1Capacity,
 928                                        options, parseError, status);
 929
 930             if(*status == U_BUFFER_OVERFLOW_ERROR){
 931                 // redo processing of string
 932                 /* we do not have enough room so grow the buffer*/
 933                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 934                 if(b1==NULL){
 935                     *status = U_MEMORY_ALLOCATION_ERROR;
 936                     goto CLEANUP;
 937                 }
 938
 939                 *status = U_ZERO_ERROR; // reset error
 940
 941                 b1Len = idnaref_toUnicode( labelStart, labelLen, b1, b1Len,
 942                                            options, parseError, status);
 943
 944             }
 945
 946             if(U_FAILURE(*status)){
 947                 goto CLEANUP;
 948             }
 949             int32_t tempLen = (reqLength + b1Len );
 950             // copy to dest
 951             if( tempLen< destCapacity){
 952                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 953             }
 954
 955             reqLength = tempLen;
 956
 957             // add the label separator
 958             if(done == FALSE){
 959                 if(reqLength < destCapacity){
 960                     dest[reqLength] = FULL_STOP;
 961                 }
 962                 reqLength++;
 963             }
 964
 965             labelStart = delimiter;
 966             remainingLen = srcLength - (delimiter - src);
 967         }
 968     }
 969
 970 CLEANUP:
 971
 972     if(b1 != b1Stack){
 973         uprv_free(b1);
 974     }
 975
 976 //    delete prep;
 977
 978     return u_terminateUChars(dest, destCapacity, reqLength, status);
 979 }
 980
 981 U_CFUNC int32_t U_EXPORT2
 982 idnaref_compare(  const UChar *s1, int32_t length1,
 983                 const UChar *s2, int32_t length2,
 984                 int32_t options,
 985                 UErrorCode* status){
 986
 987     if(status == NULL || U_FAILURE(*status)){
 988         return -1;
 989     }
 990
 991     UChar b1Stack[MAX_IDN_BUFFER_SIZE], b2Stack[MAX_IDN_BUFFER_SIZE];
 992     UChar *b1 = b1Stack, *b2 = b2Stack;
 993     int32_t b1Len, b2Len, b1Capacity = MAX_IDN_BUFFER_SIZE, b2Capacity = MAX_IDN_BUFFER_SIZE;
 994     int32_t result = -1;
 995
 996     UParseError parseError;
 997
 998     b1Len = idnaref_IDNToASCII(s1, length1, b1, b1Capacity, options, &parseError, status);
 999     if(*status == U_BUFFER_OVERFLOW_ERROR){
1000         // redo processing of string
1001         /* we do not have enough room so grow the buffer*/
1002         b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
1003         if(b1==NULL){
1004             *status = U_MEMORY_ALLOCATION_ERROR;
1005             goto CLEANUP;
1006         }
1007
1008         *status = U_ZERO_ERROR; // reset error
1009
1010         b1Len = idnaref_IDNToASCII(s1,length1,b1,b1Len, options, &parseError, status);
1011
1012     }
1013
1014     b2Len = idnaref_IDNToASCII(s2,length2,b2,b2Capacity,options, &parseError, status);
1015     if(*status == U_BUFFER_OVERFLOW_ERROR){
1016         // redo processing of string
1017         /* we do not have enough room so grow the buffer*/
1018         b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
1019         if(b2==NULL){
1020             *status = U_MEMORY_ALLOCATION_ERROR;
1021             goto CLEANUP;
1022         }
1023
1024         *status = U_ZERO_ERROR; // reset error
1025
1026         b2Len = idnaref_IDNToASCII(s2,length2,b2,b2Len,options, &parseError, status);
1027
1028     }
1029     // when toASCII is applied all label separators are replaced with FULL_STOP
1030     result = compareCaseInsensitiveASCII(b1,b1Len,b2,b2Len);
1031
1032 CLEANUP:
1033     if(b1 != b1Stack){
1034         uprv_free(b1);
1035     }
1036
1037     if(b2 != b2Stack){
1038         uprv_free(b2);
1039     }
1040
1041     return result;
1042 }
1043 #endif /* #if !UCONFIG_NO_IDNA */