icuSources/test/intltest/idnaref.cpp

   1 /*
   2  *******************************************************************************
   3  *
   4  *   Copyright (C) 2003-2004, 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
 244 U_CFUNC int32_t U_EXPORT2
 245 idnaref_toASCII(const UChar* src, int32_t srcLength,
 246               UChar* dest, int32_t destCapacity,
 247               int32_t options,
 248               UParseError* parseError,
 249               UErrorCode* status){
 250
 251     if(status == NULL || U_FAILURE(*status)){
 252         return 0;
 253     }
 254     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 255         *status = U_ILLEGAL_ARGUMENT_ERROR;
 256         return 0;
 257     }
 258     UChar b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE];
 259     //initialize pointers to stack buffers
 260     UChar  *b1 = b1Stack, *b2 = b2Stack;
 261     int32_t b1Len, b2Len,
 262             b1Capacity = MAX_LABEL_BUFFER_SIZE,
 263             b2Capacity = MAX_LABEL_BUFFER_SIZE ,
 264             reqLength=0;
 265
 266     //get the options
 267     UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 268     UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 269
 270     UBool* caseFlags = NULL;
 271
 272     // assume the source contains all ascii codepoints
 273     UBool srcIsASCII  = TRUE;
 274     // assume the source contains all LDH codepoints
 275     UBool srcIsLDH = TRUE;
 276     int32_t j=0;
 277 //    UParseError parseError;
 278     // step 2
 279     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 280
 281     if(U_FAILURE(*status)){
 282         goto CLEANUP;
 283     }
 284
 285     b1Len = prep->process(src,srcLength,b1, b1Capacity,allowUnassigned,parseError,*status);
 286
 287     if(*status == U_BUFFER_OVERFLOW_ERROR){
 288         // redo processing of string
 289         /* we do not have enough room so grow the buffer*/
 290         b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 291         if(b1==NULL){
 292             *status = U_MEMORY_ALLOCATION_ERROR;
 293             goto CLEANUP;
 294         }
 295
 296         *status = U_ZERO_ERROR; // reset error
 297
 298         b1Len = prep->process(src,srcLength,b1, b1Len,allowUnassigned, parseError, *status);
 299     }
 300     // error bail out
 301     if(U_FAILURE(*status)){
 302         goto CLEANUP;
 303     }
 304
 305     // step 3 & 4
 306     for( j=0;j<b1Len;j++){
 307         if(b1[j] > 0x7F){
 308             srcIsASCII = FALSE;
 309         }else if(prep->isLDHChar(b1[j])==FALSE){  // if the char is in ASCII range verify that it is an LDH character{
 310             srcIsLDH = FALSE;
 311         }
 312     }
 313
 314     if(useSTD3ASCIIRules == TRUE){
 315         // verify 3a and 3b
 316         if( srcIsLDH == FALSE /* source contains some non-LDH characters */
 317             || b1[0] ==  HYPHEN || b1[b1Len-1] == HYPHEN){
 318             *status = U_IDNA_STD3_ASCII_RULES_ERROR;
 319             goto CLEANUP;
 320         }
 321     }
 322     if(srcIsASCII){
 323         if(b1Len <= destCapacity){
 324             uprv_memmove(dest, b1, b1Len * U_SIZEOF_UCHAR);
 325             reqLength = b1Len;
 326         }else{
 327             reqLength = b1Len;
 328             goto CLEANUP;
 329         }
 330     }else{
 331         // step 5 : verify the sequence does not begin with ACE prefix
 332         if(!startsWithPrefix(b1,b1Len)){
 333
 334             //step 6: encode the sequence with punycode
 335             //caseFlags = (UBool*) uprv_malloc(b1Len * sizeof(UBool));
 336
 337             b2Len = convertToPuny(b1,b1Len, b2,b2Capacity,*status);
 338             //b2Len = u_strToPunycode(b2,b2Capacity,b1,b1Len, caseFlags, status);
 339             if(*status == U_BUFFER_OVERFLOW_ERROR){
 340                 // redo processing of string
 341                 /* we do not have enough room so grow the buffer*/
 342                 b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
 343                 if(b2 == NULL){
 344                     *status = U_MEMORY_ALLOCATION_ERROR;
 345                     goto CLEANUP;
 346                 }
 347
 348                 *status = U_ZERO_ERROR; // reset error
 349
 350                 b2Len = convertToPuny(b1, b1Len, b2, b2Len, *status);
 351                 //b2Len = u_strToPunycode(b2,b2Len,b1,b1Len, caseFlags, status);
 352
 353             }
 354             //error bail out
 355             if(U_FAILURE(*status)){
 356                 goto CLEANUP;
 357             }
 358             reqLength = b2Len+ACE_PREFIX_LENGTH;
 359
 360             if(reqLength > destCapacity){
 361                 *status = U_BUFFER_OVERFLOW_ERROR;
 362                 goto CLEANUP;
 363             }
 364             //Step 7: prepend the ACE prefix
 365             uprv_memcpy(dest,ACE_PREFIX,ACE_PREFIX_LENGTH * U_SIZEOF_UCHAR);
 366             //Step 6: copy the contents in b2 into dest
 367             uprv_memcpy(dest+ACE_PREFIX_LENGTH, b2, b2Len * U_SIZEOF_UCHAR);
 368
 369         }else{
 370             *status = U_IDNA_ACE_PREFIX_ERROR;
 371             goto CLEANUP;
 372         }
 373     }
 374
 375     if(reqLength > MAX_LABEL_LENGTH){
 376         *status = U_IDNA_LABEL_TOO_LONG_ERROR;
 377     }
 378
 379 CLEANUP:
 380     if(b1 != b1Stack){
 381         uprv_free(b1);
 382     }
 383     if(b2 != b2Stack){
 384         uprv_free(b2);
 385     }
 386     uprv_free(caseFlags);
 387
 388 //    delete prep;
 389
 390     return u_terminateUChars(dest, destCapacity, reqLength, status);
 391 }
 392
 393
 394 U_CFUNC int32_t U_EXPORT2
 395 idnaref_toUnicode(const UChar* src, int32_t srcLength,
 396                 UChar* dest, int32_t destCapacity,
 397                 int32_t options,
 398                 UParseError* parseError,
 399                 UErrorCode* status){
 400
 401     if(status == NULL || U_FAILURE(*status)){
 402         return 0;
 403     }
 404     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 405         *status = U_ILLEGAL_ARGUMENT_ERROR;
 406         return 0;
 407     }
 408
 409
 410
 411     UChar b1Stack[MAX_LABEL_BUFFER_SIZE], b2Stack[MAX_LABEL_BUFFER_SIZE], b3Stack[MAX_LABEL_BUFFER_SIZE];
 412
 413     //initialize pointers to stack buffers
 414     UChar  *b1 = b1Stack, *b2 = b2Stack, *b1Prime=NULL, *b3=b3Stack;
 415     int32_t b1Len, b2Len, b1PrimeLen, b3Len,
 416             b1Capacity = MAX_LABEL_BUFFER_SIZE,
 417             b2Capacity = MAX_LABEL_BUFFER_SIZE,
 418             b3Capacity = MAX_LABEL_BUFFER_SIZE,
 419             reqLength=0;
 420 //    UParseError parseError;
 421
 422     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 423     b1Len = 0;
 424     UBool* caseFlags = NULL;
 425
 426     //get the options
 427     UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 428     UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 429
 430     UBool srcIsASCII = TRUE;
 431     UBool srcIsLDH = TRUE;
 432     int32_t failPos =0;
 433
 434     if(U_FAILURE(*status)){
 435         goto CLEANUP;
 436     }
 437     // step 1: find out if all the codepoints in src are ASCII
 438     if(srcLength==-1){
 439         srcLength = 0;
 440         for(;src[srcLength]!=0;){
 441             if(src[srcLength]> 0x7f){
 442                 srcIsASCII = FALSE;
 443             }if(prep->isLDHChar(src[srcLength])==FALSE){
 444                 // here we do not assemble surrogates
 445                 // since we know that LDH code points
 446                 // are in the ASCII range only
 447                 srcIsLDH = FALSE;
 448                 failPos = srcLength;
 449             }
 450             srcLength++;
 451         }
 452     }else{
 453         for(int32_t j=0; j<srcLength; j++){
 454             if(src[j]> 0x7f){
 455                 srcIsASCII = FALSE;
 456             }else if(prep->isLDHChar(src[j])==FALSE){
 457                 // here we do not assemble surrogates
 458                 // since we know that LDH code points
 459                 // are in the ASCII range only
 460                 srcIsLDH = FALSE;
 461                 failPos = j;
 462             }
 463         }
 464     }
 465
 466     if(srcIsASCII == FALSE){
 467         // step 2: process the string
 468         b1Len = prep->process(src,srcLength,b1,b1Capacity,allowUnassigned, parseError, *status);
 469         if(*status == U_BUFFER_OVERFLOW_ERROR){
 470             // redo processing of string
 471             /* we do not have enough room so grow the buffer*/
 472             b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 473             if(b1==NULL){
 474                 *status = U_MEMORY_ALLOCATION_ERROR;
 475                 goto CLEANUP;
 476             }
 477
 478             *status = U_ZERO_ERROR; // reset error
 479
 480             b1Len = prep->process(src,srcLength,b1, b1Len,allowUnassigned, parseError, *status);
 481         }
 482         //bail out on error
 483         if(U_FAILURE(*status)){
 484             goto CLEANUP;
 485         }
 486     }else{
 487
 488         // copy everything to b1
 489         if(srcLength < b1Capacity){
 490             uprv_memmove(b1,src, srcLength * U_SIZEOF_UCHAR);
 491         }else{
 492             /* we do not have enough room so grow the buffer*/
 493             b1 = (UChar*) uprv_malloc(srcLength * U_SIZEOF_UCHAR);
 494             if(b1==NULL){
 495                 *status = U_MEMORY_ALLOCATION_ERROR;
 496                 goto CLEANUP;
 497             }
 498             uprv_memmove(b1,src, srcLength * U_SIZEOF_UCHAR);
 499         }
 500         b1Len = srcLength;
 501     }
 502     //step 3: verify ACE Prefix
 503     if(startsWithPrefix(src,srcLength)){
 504
 505         //step 4: Remove the ACE Prefix
 506         b1Prime = b1 + ACE_PREFIX_LENGTH;
 507         b1PrimeLen  = b1Len - ACE_PREFIX_LENGTH;
 508
 509         //step 5: Decode using punycode
 510         b2Len = convertFromPuny(b1Prime,b1PrimeLen, b2, b2Capacity, *status);
 511         //b2Len = u_strFromPunycode(b2, b2Capacity,b1Prime,b1PrimeLen, caseFlags, status);
 512
 513         if(*status == U_BUFFER_OVERFLOW_ERROR){
 514             // redo processing of string
 515             /* we do not have enough room so grow the buffer*/
 516             b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
 517             if(b2==NULL){
 518                 *status = U_MEMORY_ALLOCATION_ERROR;
 519                 goto CLEANUP;
 520             }
 521
 522             *status = U_ZERO_ERROR; // reset error
 523
 524             b2Len =  convertFromPuny(b1Prime,b1PrimeLen, b2, b2Len, *status);
 525             //b2Len = u_strFromPunycode(b2, b2Len,b1Prime,b1PrimeLen,caseFlags, status);
 526         }
 527
 528
 529         //step 6:Apply toASCII
 530         b3Len = idnaref_toASCII(b2,b2Len,b3,b3Capacity,options,parseError, status);
 531
 532         if(*status == U_BUFFER_OVERFLOW_ERROR){
 533             // redo processing of string
 534             /* we do not have enough room so grow the buffer*/
 535             b3 = (UChar*) uprv_malloc(b3Len * U_SIZEOF_UCHAR);
 536             if(b3==NULL){
 537                 *status = U_MEMORY_ALLOCATION_ERROR;
 538                 goto CLEANUP;
 539             }
 540
 541             *status = U_ZERO_ERROR; // reset error
 542
 543             b3Len =  idnaref_toASCII(b2,b2Len,b3,b3Len, options, parseError, status);
 544
 545         }
 546         //bail out on error
 547         if(U_FAILURE(*status)){
 548             goto CLEANUP;
 549         }
 550
 551         //step 7: verify
 552         if(compareCaseInsensitiveASCII(b1, b1Len, b3, b3Len) !=0){
 553             *status = U_IDNA_VERIFICATION_ERROR;
 554             goto CLEANUP;
 555         }
 556
 557         //step 8: return output of step 5
 558         reqLength = b2Len;
 559         if(b2Len <= destCapacity) {
 560             uprv_memmove(dest, b2, b2Len * U_SIZEOF_UCHAR);
 561         }
 562     }else{
 563         // verify that STD3 ASCII rules are satisfied
 564         if(useSTD3ASCIIRules == TRUE){
 565             if( srcIsLDH == FALSE /* source contains some non-LDH characters */
 566                 || src[0] ==  HYPHEN || src[srcLength-1] == HYPHEN){
 567                 *status = U_IDNA_STD3_ASCII_RULES_ERROR;
 568
 569                 /* populate the parseError struct */
 570                 if(srcIsLDH==FALSE){
 571                     // failPos is always set the index of failure
 572                     uprv_syntaxError(src,failPos, srcLength,parseError);
 573                 }else if(src[0] == HYPHEN){
 574                     // fail position is 0
 575                     uprv_syntaxError(src,0,srcLength,parseError);
 576                 }else{
 577                     // the last index in the source is always length-1
 578                     uprv_syntaxError(src, (srcLength>0) ? srcLength-1 : srcLength, srcLength,parseError);
 579                 }
 580
 581                 goto CLEANUP;
 582             }
 583         }
 584         //copy the source to destination
 585         if(srcLength <= destCapacity){
 586             uprv_memmove(dest,src,srcLength * U_SIZEOF_UCHAR);
 587         }
 588         reqLength = srcLength;
 589     }
 590
 591 CLEANUP:
 592
 593     if(b1 != b1Stack){
 594         uprv_free(b1);
 595     }
 596     if(b2 != b2Stack){
 597         uprv_free(b2);
 598     }
 599     uprv_free(caseFlags);
 600
 601 //    delete prep;
 602
 603     return u_terminateUChars(dest, destCapacity, reqLength, status);
 604 }
 605
 606
 607 static int32_t
 608 getNextSeparator(UChar *src,int32_t srcLength,NamePrepTransform* prep,
 609                  UChar **limit,
 610                  UBool *done,
 611                  UErrorCode *status){
 612     if(srcLength == -1){
 613         int32_t i;
 614         for(i=0 ; ;i++){
 615             if(src[i] == 0){
 616                 *limit = src + i; // point to null
 617                 *done = TRUE;
 618                 return i;
 619             }
 620             if(prep->isLabelSeparator(src[i],*status)){
 621                 *limit = src + (i+1); // go past the delimiter
 622                 return i;
 623
 624             }
 625         }
 626     }else{
 627         int32_t i;
 628         for(i=0;i<srcLength;i++){
 629             if(prep->isLabelSeparator(src[i],*status)){
 630                 *limit = src + (i+1); // go past the delimiter
 631                 return i;
 632             }
 633         }
 634         // we have not found the delimiter
 635         if(i==srcLength){
 636             *limit = src+srcLength;
 637             *done = TRUE;
 638         }
 639         return i;
 640     }
 641 }
 642
 643 U_CFUNC int32_t U_EXPORT2
 644 idnaref_IDNToASCII(  const UChar* src, int32_t srcLength,
 645                    UChar* dest, int32_t destCapacity,
 646                    int32_t options,
 647                    UParseError* parseError,
 648                    UErrorCode* status){
 649
 650     if(status == NULL || U_FAILURE(*status)){
 651         return 0;
 652     }
 653     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 654         *status = U_ILLEGAL_ARGUMENT_ERROR;
 655         return 0;
 656     }
 657
 658     int32_t reqLength = 0;
 659 //    UParseError parseError;
 660
 661     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 662
 663     //initialize pointers to stack buffers
 664     UChar b1Stack[MAX_LABEL_BUFFER_SIZE];
 665     UChar  *b1 = b1Stack;
 666     int32_t b1Len, labelLen;
 667     UChar* delimiter = (UChar*)src;
 668     UChar* labelStart = (UChar*)src;
 669     int32_t remainingLen = srcLength;
 670     int32_t b1Capacity = MAX_LABEL_BUFFER_SIZE;
 671
 672     //get the options
 673 //    UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 674 //    UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 675     UBool done = FALSE;
 676
 677     if(U_FAILURE(*status)){
 678         goto CLEANUP;
 679     }
 680
 681
 682     if(srcLength == -1){
 683         for(;;){
 684
 685             if(*delimiter == 0){
 686                 break;
 687             }
 688
 689             labelLen = getNextSeparator(labelStart, -1, prep, &delimiter, &done, status);
 690
 691             b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Capacity,
 692                                     options, parseError, status);
 693
 694             if(*status == U_BUFFER_OVERFLOW_ERROR){
 695                 // redo processing of string
 696                 /* we do not have enough room so grow the buffer*/
 697                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 698                 if(b1==NULL){
 699                     *status = U_MEMORY_ALLOCATION_ERROR;
 700                     goto CLEANUP;
 701                 }
 702
 703                 *status = U_ZERO_ERROR; // reset error
 704
 705                 b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Len,
 706                                         options, parseError, status);
 707
 708             }
 709
 710             if(U_FAILURE(*status)){
 711                 goto CLEANUP;
 712             }
 713             int32_t tempLen = (reqLength + b1Len );
 714             // copy to dest
 715             if( tempLen< destCapacity){
 716                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 717             }
 718
 719             reqLength = tempLen;
 720
 721             // add the label separator
 722             if(done == FALSE){
 723                 if(reqLength < destCapacity){
 724                     dest[reqLength] = FULL_STOP;
 725                 }
 726                 reqLength++;
 727             }
 728
 729             labelStart = delimiter;
 730         }
 731     }else{
 732         for(;;){
 733
 734             if(delimiter == src+srcLength){
 735                 break;
 736             }
 737
 738             labelLen = getNextSeparator(labelStart, remainingLen, prep, &delimiter, &done, status);
 739
 740             b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Capacity,
 741                                     options,parseError, status);
 742
 743             if(*status == U_BUFFER_OVERFLOW_ERROR){
 744                 // redo processing of string
 745                 /* we do not have enough room so grow the buffer*/
 746                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 747                 if(b1==NULL){
 748                     *status = U_MEMORY_ALLOCATION_ERROR;
 749                     goto CLEANUP;
 750                 }
 751
 752                 *status = U_ZERO_ERROR; // reset error
 753
 754                 b1Len = idnaref_toASCII(labelStart, labelLen, b1, b1Len,
 755                                         options, parseError, status);
 756
 757             }
 758
 759             if(U_FAILURE(*status)){
 760                 goto CLEANUP;
 761             }
 762             int32_t tempLen = (reqLength + b1Len );
 763             // copy to dest
 764             if( tempLen< destCapacity){
 765                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 766             }
 767
 768             reqLength = tempLen;
 769
 770             // add the label separator
 771             if(done == FALSE){
 772                 if(reqLength < destCapacity){
 773                     dest[reqLength] = FULL_STOP;
 774                 }
 775                 reqLength++;
 776             }
 777
 778             labelStart = delimiter;
 779             remainingLen = srcLength - (delimiter - src);
 780         }
 781     }
 782
 783
 784 CLEANUP:
 785
 786     if(b1 != b1Stack){
 787         uprv_free(b1);
 788     }
 789
 790 //   delete prep;
 791
 792     return u_terminateUChars(dest, destCapacity, reqLength, status);
 793 }
 794
 795 U_CFUNC int32_t U_EXPORT2
 796 idnaref_IDNToUnicode(  const UChar* src, int32_t srcLength,
 797                      UChar* dest, int32_t destCapacity,
 798                      int32_t options,
 799                      UParseError* parseError,
 800                      UErrorCode* status){
 801
 802     if(status == NULL || U_FAILURE(*status)){
 803         return 0;
 804     }
 805     if((src == NULL) || (srcLength < -1) || (destCapacity<0) || (!dest && destCapacity > 0)){
 806         *status = U_ILLEGAL_ARGUMENT_ERROR;
 807         return 0;
 808     }
 809
 810     int32_t reqLength = 0;
 811
 812     UBool done = FALSE;
 813
 814     NamePrepTransform* prep = TestIDNA::getInstance(*status);
 815
 816     //initialize pointers to stack buffers
 817     UChar b1Stack[MAX_LABEL_BUFFER_SIZE];
 818     UChar  *b1 = b1Stack;
 819     int32_t b1Len, labelLen;
 820     UChar* delimiter = (UChar*)src;
 821     UChar* labelStart = (UChar*)src;
 822     int32_t remainingLen = srcLength;
 823     int32_t b1Capacity = MAX_LABEL_BUFFER_SIZE;
 824
 825     //get the options
 826 //    UBool allowUnassigned   = (UBool)((options & IDNAREF_ALLOW_UNASSIGNED) != 0);
 827 //    UBool useSTD3ASCIIRules = (UBool)((options & IDNAREF_USE_STD3_RULES) != 0);
 828
 829     if(U_FAILURE(*status)){
 830         goto CLEANUP;
 831     }
 832
 833     if(srcLength == -1){
 834         for(;;){
 835
 836             if(*delimiter == 0){
 837                 break;
 838             }
 839
 840             labelLen = getNextSeparator(labelStart, -1, prep, &delimiter, &done, status);
 841
 842             b1Len = idnaref_toUnicode(labelStart, labelLen, b1, b1Capacity,
 843                                       options, parseError, status);
 844
 845             if(*status == U_BUFFER_OVERFLOW_ERROR){
 846                 // redo processing of string
 847                 /* we do not have enough room so grow the buffer*/
 848                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 849                 if(b1==NULL){
 850                     *status = U_MEMORY_ALLOCATION_ERROR;
 851                     goto CLEANUP;
 852                 }
 853
 854                 *status = U_ZERO_ERROR; // reset error
 855
 856                 b1Len = idnaref_toUnicode( labelStart, labelLen, b1, b1Len,
 857                                            options, parseError, status);
 858
 859             }
 860
 861             if(U_FAILURE(*status)){
 862                 goto CLEANUP;
 863             }
 864             int32_t tempLen = (reqLength + b1Len );
 865             // copy to dest
 866             if( tempLen< destCapacity){
 867                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 868             }
 869
 870             reqLength = tempLen;
 871             // add the label separator
 872             if(done == FALSE){
 873                 if(reqLength < destCapacity){
 874                     dest[reqLength] = FULL_STOP;
 875                 }
 876                 reqLength++;
 877             }
 878
 879             labelStart = delimiter;
 880         }
 881     }else{
 882         for(;;){
 883
 884             if(delimiter == src+srcLength){
 885                 break;
 886             }
 887
 888             labelLen = getNextSeparator(labelStart, remainingLen, prep, &delimiter, &done, status);
 889
 890             b1Len = idnaref_toUnicode( labelStart,labelLen, b1, b1Capacity,
 891                                        options, parseError, status);
 892
 893             if(*status == U_BUFFER_OVERFLOW_ERROR){
 894                 // redo processing of string
 895                 /* we do not have enough room so grow the buffer*/
 896                 b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 897                 if(b1==NULL){
 898                     *status = U_MEMORY_ALLOCATION_ERROR;
 899                     goto CLEANUP;
 900                 }
 901
 902                 *status = U_ZERO_ERROR; // reset error
 903
 904                 b1Len = idnaref_toUnicode( labelStart, labelLen, b1, b1Len,
 905                                            options, parseError, status);
 906
 907             }
 908
 909             if(U_FAILURE(*status)){
 910                 goto CLEANUP;
 911             }
 912             int32_t tempLen = (reqLength + b1Len );
 913             // copy to dest
 914             if( tempLen< destCapacity){
 915                 uprv_memmove(dest+reqLength, b1, b1Len * U_SIZEOF_UCHAR);
 916             }
 917
 918             reqLength = tempLen;
 919
 920             // add the label separator
 921             if(done == FALSE){
 922                 if(reqLength < destCapacity){
 923                     dest[reqLength] = FULL_STOP;
 924                 }
 925                 reqLength++;
 926             }
 927
 928             labelStart = delimiter;
 929             remainingLen = srcLength - (delimiter - src);
 930         }
 931     }
 932
 933 CLEANUP:
 934
 935     if(b1 != b1Stack){
 936         uprv_free(b1);
 937     }
 938
 939 //    delete prep;
 940
 941     return u_terminateUChars(dest, destCapacity, reqLength, status);
 942 }
 943
 944 U_CFUNC int32_t U_EXPORT2
 945 idnaref_compare(  const UChar *s1, int32_t length1,
 946                 const UChar *s2, int32_t length2,
 947                 int32_t options,
 948                 UErrorCode* status){
 949
 950     if(status == NULL || U_FAILURE(*status)){
 951         return -1;
 952     }
 953
 954     UChar b1Stack[MAX_IDN_BUFFER_SIZE], b2Stack[MAX_IDN_BUFFER_SIZE];
 955     UChar *b1 = b1Stack, *b2 = b2Stack;
 956     int32_t b1Len, b2Len, b1Capacity = MAX_IDN_BUFFER_SIZE, b2Capacity = MAX_IDN_BUFFER_SIZE;
 957     int32_t result = -1;
 958
 959     UParseError parseError;
 960
 961     b1Len = idnaref_IDNToASCII(s1, length1, b1, b1Capacity, options, &parseError, status);
 962     if(*status == U_BUFFER_OVERFLOW_ERROR){
 963         // redo processing of string
 964         /* we do not have enough room so grow the buffer*/
 965         b1 = (UChar*) uprv_malloc(b1Len * U_SIZEOF_UCHAR);
 966         if(b1==NULL){
 967             *status = U_MEMORY_ALLOCATION_ERROR;
 968             goto CLEANUP;
 969         }
 970
 971         *status = U_ZERO_ERROR; // reset error
 972
 973         b1Len = idnaref_IDNToASCII(s1,length1,b1,b1Len, options, &parseError, status);
 974
 975     }
 976
 977     b2Len = idnaref_IDNToASCII(s2,length2,b2,b2Capacity,options, &parseError, status);
 978     if(*status == U_BUFFER_OVERFLOW_ERROR){
 979         // redo processing of string
 980         /* we do not have enough room so grow the buffer*/
 981         b2 = (UChar*) uprv_malloc(b2Len * U_SIZEOF_UCHAR);
 982         if(b2==NULL){
 983             *status = U_MEMORY_ALLOCATION_ERROR;
 984             goto CLEANUP;
 985         }
 986
 987         *status = U_ZERO_ERROR; // reset error
 988
 989         b2Len = idnaref_IDNToASCII(s2,length2,b2,b2Len,options, &parseError, status);
 990
 991     }
 992     // when toASCII is applied all label separators are replaced with FULL_STOP
 993     result = compareCaseInsensitiveASCII(b1,b1Len,b2,b2Len);
 994
 995 CLEANUP:
 996     if(b1 != b1Stack){
 997         uprv_free(b1);
 998     }
 999
1000     if(b2 != b2Stack){
1001         uprv_free(b2);
1002     }
1003
1004     return result;
1005 }
1006 #endif /* #if !UCONFIG_NO_IDNA */