icuSources/tools/gensprep/store.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2009, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  store.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2003-02-06
  14 *   created by: Ram Viswanadha
  15 *
  16 */
  17
  18 #include <stdio.h>
  19 #include <stdlib.h>
  20 #include "unicode/utypes.h"
  21 #include "cmemory.h"
  22 #include "cstring.h"
  23 #include "filestrm.h"
  24 #include "unicode/udata.h"
  25 #include "utrie.h"
  26 #include "unewdata.h"
  27 #include "gensprep.h"
  28 #include "uhash.h"
  29
  30
  31 #define DO_DEBUG_OUT 0
  32
  33
  34 /*
  35  * StringPrep profile file format ------------------------------------
  36  *
  37  * The file format prepared and written here contains a 16-bit trie and a mapping table.
  38  *
  39  * Before the data contents described below, there are the headers required by
  40  * the udata API for loading ICU data. Especially, a UDataInfo structure
  41  * precedes the actual data. It contains platform properties values and the
  42  * file format version.
  43  *
  44  * The following is a description of format version 2.
  45  *
  46  * Data contents:
  47  *
  48  * The contents is a parsed, binary form of RFC3454 and possibly
  49  * NormalizationCorrections.txt depending on the options specified on the profile.
  50  *
  51  * Any Unicode code point from 0 to 0x10ffff can be looked up to get
  52  * the trie-word, if any, for that code point. This means that the input
  53  * to the lookup are 21-bit unsigned integers, with not all of the
  54  * 21-bit range used.
  55  *
  56  * *.spp files customarily begin with a UDataInfo structure, see udata.h and .c.
  57  * After that there are the following structures:
  58  *
  59  * int32_t indexes[_SPREP_INDEX_TOP];           -- _SPREP_INDEX_TOP=16, see enum in sprpimpl.h file
  60  *
  61  * UTrie stringPrepTrie;                        -- size in bytes=indexes[_SPREP_INDEX_TRIE_SIZE]
  62  *
  63  * uint16_t mappingTable[];                     -- Contains the sequecence of code units that the code point maps to
  64  *                                                 size in bytes = indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]
  65  *
  66  * The indexes array contains the following values:
  67  *  indexes[_SPREP_INDEX_TRIE_SIZE]                  -- The size of the StringPrep trie in bytes
  68  *  indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]          -- The size of the mappingTable in bytes
  69  *  indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION]  -- The index of Unicode version of last entry in NormalizationCorrections.txt
  70  *  indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START]    -- The starting index of 1 UChar  mapping index in the mapping table
  71  *  indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]   -- The starting index of 2 UChars mapping index in the mapping table
  72  *  indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] -- The starting index of 3 UChars mapping index in the mapping table
  73  *  indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]  -- The starting index of 4 UChars mapping index in the mapping table
  74  *  indexes[_SPREP_OPTIONS]                          -- Bit set of options to turn on in the profile, e.g: USPREP_NORMALIZATION_ON, USPREP_CHECK_BIDI_ON
  75  *
  76  *
  77  * StringPrep Trie :
  78  *
  79  * The StringPrep tries is a 16-bit trie that contains data for the profile.
  80  * Each code point is associated with a value (trie-word) in the trie.
  81  *
  82  * - structure of data words from the trie
  83  *
  84  *  i)  A value greater than or equal to _SPREP_TYPE_THRESHOLD (0xFFF0)
  85  *      represents the type associated with the code point
  86  *      if(trieWord >= _SPREP_TYPE_THRESHOLD){
  87  *          type = trieWord - 0xFFF0;
  88  *      }
  89  *      The type can be :
  90  *             USPREP_UNASSIGNED
  91  *             USPREP_PROHIBITED
  92  *             USPREP_DELETE
  93  *
  94  *  ii) A value less than _SPREP_TYPE_THRESHOLD means the type is USPREP_MAP and
  95  *      contains distribution described below
  96  *
  97  *      0       -  ON : The code point is prohibited (USPREP_PROHIBITED). This is to allow for codepoint that are both prohibited and mapped.
  98  *      1       -  ON : The value in the next 14 bits is an index into the mapping table
  99  *                 OFF: The value in the next 14 bits is an delta value from the code point
 100  *      2..15   -  Contains data as described by bit 1. If all bits are set
 101  *                 (value = _SPREP_MAX_INDEX_VALUE) then the type is USPREP_DELETE
 102  *
 103  *
 104  * Mapping Table:
 105  * The data in mapping table is sorted according to the length of the mapping sequence.
 106  * If the type of the code point is USPREP_MAP and value in trie word is an index, the index
 107  * is compared with start indexes of sequence length start to figure out the length according to
 108  * the following algorithm:
 109  *
 110  *              if(       index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
 111  *                        index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
 112  *                   length = 1;
 113  *               }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
 114  *                        index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
 115  *                   length = 2;
 116  *               }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
 117  *                        index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
 118  *                   length = 3;
 119  *               }else{
 120  *                   // The first position in the mapping table contains the length
 121  *                   // of the sequence
 122  *                   length = mappingTable[index++];
 123  *
 124  *               }
 125  *
 126  */
 127
 128 /* file data ---------------------------------------------------------------- */
 129 /* indexes[] value names */
 130
 131 #if UCONFIG_NO_IDNA
 132
 133 /* dummy UDataInfo cf. udata.h */
 134 static UDataInfo dataInfo = {
 135     sizeof(UDataInfo),
 136     0,
 137
 138     U_IS_BIG_ENDIAN,
 139     U_CHARSET_FAMILY,
 140     U_SIZEOF_UCHAR,
 141     0,
 142
 143     { 0, 0, 0, 0 },                 /* dummy dataFormat */
 144     { 0, 0, 0, 0 },                 /* dummy formatVersion */
 145     { 0, 0, 0, 0 }                  /* dummy dataVersion */
 146 };
 147
 148 #else
 149
 150 static int32_t indexes[_SPREP_INDEX_TOP]={ 0 };
 151
 152 static uint16_t* mappingData= NULL;
 153 static int32_t mappingDataCapacity = 0; /* we skip the first index in mapping data */
 154 static int16_t currentIndex = 0; /* the current index into the data trie */
 155 static int32_t maxLength = 0;  /* maximum length of mapping string */
 156
 157
 158 /* UDataInfo cf. udata.h */
 159 static UDataInfo dataInfo={
 160     sizeof(UDataInfo),
 161     0,
 162
 163     U_IS_BIG_ENDIAN,
 164     U_CHARSET_FAMILY,
 165     U_SIZEOF_UCHAR,
 166     0,
 167
 168     { 0x53, 0x50, 0x52, 0x50 },                 /* dataFormat="SPRP" */
 169     { 3, 2, UTRIE_SHIFT, UTRIE_INDEX_SHIFT },   /* formatVersion */
 170     { 3, 2, 0, 0 }                              /* dataVersion (Unicode version) */
 171 };
 172 void
 173 setUnicodeVersion(const char *v) {
 174     UVersionInfo version;
 175     u_versionFromString(version, v);
 176     uprv_memcpy(dataInfo.dataVersion, version, 4);
 177 }
 178
 179 void
 180 setUnicodeVersionNC(UVersionInfo version){
 181     uint32_t univer = version[0] << 24;
 182     univer += version[1] << 16;
 183     univer += version[2] << 8;
 184     univer += version[3];
 185     indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION] = univer;
 186 }
 187 static UNewTrie *sprepTrie;
 188
 189 #define MAX_DATA_LENGTH 11500
 190
 191
 192 #define SPREP_DELTA_RANGE_POSITIVE_LIMIT              8191
 193 #define SPREP_DELTA_RANGE_NEGATIVE_LIMIT              -8192
 194
 195
 196 extern void
 197 init() {
 198
 199     sprepTrie = (UNewTrie *)uprv_malloc(sizeof(UNewTrie));
 200     uprv_memset(sprepTrie, 0, sizeof(UNewTrie));
 201
 202     /* initialize the two tries */
 203     if(NULL==utrie_open(sprepTrie, NULL, MAX_DATA_LENGTH, 0, 0, FALSE)) {
 204         fprintf(stderr, "error: failed to initialize tries\n");
 205         exit(U_MEMORY_ALLOCATION_ERROR);
 206     }
 207 }
 208
 209 static UHashtable* hashTable = NULL;
 210
 211
 212 typedef struct ValueStruct {
 213     UChar* mapping;
 214     int16_t length;
 215     UStringPrepType type;
 216 } ValueStruct;
 217
 218 /* Callback for deleting the value from the hashtable */
 219 static void U_CALLCONV valueDeleter(void* obj){
 220     ValueStruct* value = (ValueStruct*) obj;
 221     uprv_free(value->mapping);
 222     uprv_free(value);
 223 }
 224
 225 /* Callback for hashing the entry */
 226 static int32_t U_CALLCONV hashEntry(const UHashTok parm) {
 227     return  parm.integer;
 228 }
 229
 230 /* Callback for comparing two entries */
 231 static UBool U_CALLCONV compareEntries(const UHashTok p1, const UHashTok p2) {
 232     return (UBool)(p1.integer != p2.integer);
 233 }
 234
 235
 236 static void
 237 storeMappingData(){
 238
 239     int32_t pos = -1;
 240     const UHashElement* element = NULL;
 241     ValueStruct* value  = NULL;
 242     int32_t codepoint = 0;
 243     int32_t elementCount = 0;
 244     int32_t writtenElementCount = 0;
 245     int32_t mappingLength = 1; /* minimum mapping length */
 246     int32_t oldMappingLength = 0;
 247     uint16_t trieWord =0;
 248     int32_t limitIndex = 0;
 249
 250     if (hashTable == NULL) {
 251         return;
 252     }
 253     elementCount = uhash_count(hashTable);
 254
 255         /*initialize the mapping data */
 256     mappingData = (uint16_t*) uprv_malloc(U_SIZEOF_UCHAR * (mappingDataCapacity));
 257
 258     uprv_memset(mappingData,0,U_SIZEOF_UCHAR * mappingDataCapacity);
 259
 260     while(writtenElementCount < elementCount){
 261
 262         while( (element = uhash_nextElement(hashTable, &pos))!=NULL){
 263
 264             codepoint = element->key.integer;
 265             value = (ValueStruct*)element->value.pointer;
 266
 267             /* store the start of indexes */
 268             if(oldMappingLength != mappingLength){
 269                 /* Assume that index[] is used according to the enums defined */
 270                 if(oldMappingLength <=_SPREP_MAX_INDEX_TOP_LENGTH){
 271                     indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION+mappingLength] = currentIndex;
 272                 }
 273                 if(oldMappingLength <= _SPREP_MAX_INDEX_TOP_LENGTH &&
 274                    mappingLength == _SPREP_MAX_INDEX_TOP_LENGTH +1){
 275
 276                     limitIndex = currentIndex;
 277
 278                 }
 279                 oldMappingLength = mappingLength;
 280             }
 281
 282             if(value->length == mappingLength){
 283                 uint32_t savedTrieWord = 0;
 284                 trieWord = currentIndex << 2;
 285                 /* turn on the 2nd bit to signal that the following bits contain an index */
 286                 trieWord += 0x02;
 287
 288                 if(trieWord > _SPREP_TYPE_THRESHOLD){
 289                     fprintf(stderr,"trieWord cannot contain value greater than 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
 290                     exit(U_ILLEGAL_CHAR_FOUND);
 291                 }
 292                 /* figure out if the code point has type already stored */
 293                 savedTrieWord= utrie_get32(sprepTrie,codepoint,NULL);
 294                 if(savedTrieWord!=0){
 295                     if((savedTrieWord- _SPREP_TYPE_THRESHOLD) == USPREP_PROHIBITED){
 296                         /* turn on the first bit in trie word */
 297                         trieWord += 0x01;
 298                     }else{
 299                         /*
 300                          * the codepoint has value something other than prohibited
 301                          * and a mapping .. error!
 302                          */
 303                         fprintf(stderr,"Type for codepoint \\U%08X already set!.\n", (int)codepoint);
 304                         exit(U_ILLEGAL_ARGUMENT_ERROR);
 305                     }
 306                 }
 307
 308                 /* now set the value in the trie */
 309                 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
 310                     fprintf(stderr,"Could not set the value for code point.\n");
 311                     exit(U_ILLEGAL_ARGUMENT_ERROR);
 312                 }
 313
 314                 /* written the trie word for the codepoint... increment the count*/
 315                 writtenElementCount++;
 316
 317                 /* sanity check are we exceeding the max number allowed */
 318                 if(currentIndex+value->length+1 > _SPREP_MAX_INDEX_VALUE){
 319                     fprintf(stderr, "Too many entries in the mapping table %i. Maximum allowed is %i\n", currentIndex+value->length, _SPREP_MAX_INDEX_VALUE);
 320                     exit(U_INDEX_OUTOFBOUNDS_ERROR);
 321                 }
 322
 323                 /* copy the mapping data */
 324                 if(currentIndex+value->length+1 <= mappingDataCapacity){
 325                     /* write the length */
 326                     if(mappingLength > _SPREP_MAX_INDEX_TOP_LENGTH ){
 327                          /* the cast here is safe since we donot expect the length to be > 65535 */
 328                          mappingData[currentIndex++] = (uint16_t) mappingLength;
 329                     }
 330                     /* copy the contents to mappindData array */
 331                     uprv_memmove(mappingData+currentIndex, value->mapping, value->length*U_SIZEOF_UCHAR);
 332                     currentIndex += value->length;
 333
 334                 }else{
 335                     /* realloc */
 336                     UChar* newMappingData = (uint16_t*) uprv_malloc(U_SIZEOF_UCHAR * mappingDataCapacity*2);
 337                     if(newMappingData == NULL){
 338                         fprintf(stderr, "Could not realloc the mapping data!\n");
 339                         exit(U_MEMORY_ALLOCATION_ERROR);
 340                     }
 341                     uprv_memmove(newMappingData, mappingData, U_SIZEOF_UCHAR * mappingDataCapacity);
 342                     mappingDataCapacity *= 2;
 343                     uprv_free(mappingData);
 344                     mappingData = newMappingData;
 345                     /* write the length */
 346                     if(mappingLength > _SPREP_MAX_INDEX_TOP_LENGTH ){
 347                          /* the cast here is safe since we donot expect the length to be > 65535 */
 348                          mappingData[currentIndex++] = (uint16_t) mappingLength;
 349                     }
 350                     /* continue copying */
 351                     uprv_memmove(mappingData+currentIndex, value->mapping, value->length*U_SIZEOF_UCHAR);
 352                     currentIndex += value->length;
 353                 }
 354
 355             }
 356         }
 357         mappingLength++;
 358         pos = -1;
 359     }
 360     /* set the last length for range check */
 361     if(mappingLength <= _SPREP_MAX_INDEX_TOP_LENGTH){
 362         indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION+mappingLength] = currentIndex+1;
 363     }else{
 364         indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START] = limitIndex;
 365     }
 366
 367 }
 368
 369 extern void setOptions(int32_t options){
 370     indexes[_SPREP_OPTIONS] = options;
 371 }
 372 extern void
 373 storeMapping(uint32_t codepoint, uint32_t* mapping,int32_t length,
 374              UStringPrepType type, UErrorCode* status){
 375
 376
 377     UChar* map = NULL;
 378     int16_t adjustedLen=0, i;
 379     uint16_t trieWord = 0;
 380     ValueStruct *value = NULL;
 381     uint32_t savedTrieWord = 0;
 382
 383     /* initialize the hashtable */
 384     if(hashTable==NULL){
 385         hashTable = uhash_open(hashEntry, compareEntries, NULL, status);
 386         uhash_setValueDeleter(hashTable, valueDeleter);
 387     }
 388
 389     /* figure out if the code point has type already stored */
 390     savedTrieWord= utrie_get32(sprepTrie,codepoint,NULL);
 391     if(savedTrieWord!=0){
 392         if((savedTrieWord- _SPREP_TYPE_THRESHOLD) == USPREP_PROHIBITED){
 393             /* turn on the first bit in trie word */
 394             trieWord += 0x01;
 395         }else{
 396             /*
 397              * the codepoint has value something other than prohibited
 398              * and a mapping .. error!
 399              */
 400             fprintf(stderr,"Type for codepoint \\U%08X already set!.\n", (int)codepoint);
 401             exit(U_ILLEGAL_ARGUMENT_ERROR);
 402         }
 403     }
 404
 405     /* figure out the real length */
 406     for(i=0; i<length; i++){
 407         if(mapping[i] > 0xFFFF){
 408             adjustedLen +=2;
 409         }else{
 410             adjustedLen++;
 411         }
 412     }
 413
 414     if(adjustedLen == 0){
 415         trieWord = (uint16_t)(_SPREP_MAX_INDEX_VALUE << 2);
 416         /* make sure that the value of trieWord is less than the threshold */
 417         if(trieWord < _SPREP_TYPE_THRESHOLD){
 418             /* now set the value in the trie */
 419             if(!utrie_set32(sprepTrie,codepoint,trieWord)){
 420                 fprintf(stderr,"Could not set the value for code point.\n");
 421                 exit(U_ILLEGAL_ARGUMENT_ERROR);
 422             }
 423             /* value is set so just return */
 424             return;
 425         }else{
 426             fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
 427             exit(U_ILLEGAL_CHAR_FOUND);
 428         }
 429     }
 430
 431     if(adjustedLen == 1){
 432         /* calculate the delta */
 433         int16_t delta = (int16_t)((int32_t)codepoint - (int16_t) mapping[0]);
 434         if(delta >= SPREP_DELTA_RANGE_NEGATIVE_LIMIT && delta <= SPREP_DELTA_RANGE_POSITIVE_LIMIT){
 435
 436             trieWord = delta << 2;
 437
 438
 439             /* make sure that the second bit is OFF */
 440             if((trieWord & 0x02) != 0 ){
 441                 fprintf(stderr,"The second bit in the trie word is not zero while storing a delta.\n");
 442                 exit(U_INTERNAL_PROGRAM_ERROR);
 443             }
 444             /* make sure that the value of trieWord is less than the threshold */
 445             if(trieWord < _SPREP_TYPE_THRESHOLD){
 446                 /* now set the value in the trie */
 447                 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
 448                     fprintf(stderr,"Could not set the value for code point.\n");
 449                     exit(U_ILLEGAL_ARGUMENT_ERROR);
 450                 }
 451                 /* value is set so just return */
 452                 return;
 453             }
 454         }
 455         /*
 456          * if the delta is not in the given range or if the trieWord is larger than the threshold
 457          * just fall through for storing the mapping in the mapping table
 458          */
 459     }
 460
 461     map = (UChar*) uprv_malloc(U_SIZEOF_UCHAR * (adjustedLen+1));
 462     uprv_memset(map,0,U_SIZEOF_UCHAR * (adjustedLen+1));
 463
 464     i=0;
 465
 466     while(i<length){
 467         if(mapping[i] <= 0xFFFF){
 468             map[i] = (uint16_t)mapping[i];
 469         }else{
 470             map[i]   = UTF16_LEAD(mapping[i]);
 471             map[i+1] = UTF16_TRAIL(mapping[i]);
 472         }
 473         i++;
 474     }
 475
 476     value = (ValueStruct*) uprv_malloc(sizeof(ValueStruct));
 477     value->mapping = map;
 478     value->type   = type;
 479     value->length  = adjustedLen;
 480     if(value->length > _SPREP_MAX_INDEX_TOP_LENGTH){
 481         mappingDataCapacity++;
 482     }
 483     if(maxLength < value->length){
 484         maxLength = value->length;
 485     }
 486     uhash_iput(hashTable,codepoint,value,status);
 487     mappingDataCapacity += adjustedLen;
 488
 489     if(U_FAILURE(*status)){
 490         fprintf(stderr, "Failed to put entries into the hastable. Error: %s\n", u_errorName(*status));
 491         exit(*status);
 492     }
 493 }
 494
 495
 496 extern void
 497 storeRange(uint32_t start, uint32_t end, UStringPrepType type,UErrorCode* status){
 498     uint16_t trieWord = 0;
 499
 500     if((int)(_SPREP_TYPE_THRESHOLD + type) > 0xFFFF){
 501         fprintf(stderr,"trieWord cannot contain value greater than 0xFFFF.\n");
 502         exit(U_ILLEGAL_CHAR_FOUND);
 503     }
 504     trieWord = (_SPREP_TYPE_THRESHOLD + type); /* the top 4 bits contain the value */
 505     if(start == end){
 506         uint32_t savedTrieWord = utrie_get32(sprepTrie, start, NULL);
 507         if(savedTrieWord>0){
 508             if(savedTrieWord < _SPREP_TYPE_THRESHOLD && type == USPREP_PROHIBITED){
 509                 /*
 510                  * A mapping is stored in the trie word
 511                  * and the only other possible type that a
 512                  * code point can have is USPREP_PROHIBITED
 513                  *
 514                  */
 515
 516                 /* turn on the 0th bit in the savedTrieWord */
 517                 savedTrieWord += 0x01;
 518
 519                 /* the downcast is safe since we only save 16 bit values */
 520                 trieWord = (uint16_t)savedTrieWord;
 521
 522                 /* make sure that the value of trieWord is less than the threshold */
 523                 if(trieWord < _SPREP_TYPE_THRESHOLD){
 524                     /* now set the value in the trie */
 525                     if(!utrie_set32(sprepTrie,start,trieWord)){
 526                         fprintf(stderr,"Could not set the value for code point.\n");
 527                         exit(U_ILLEGAL_ARGUMENT_ERROR);
 528                     }
 529                     /* value is set so just return */
 530                     return;
 531                 }else{
 532                     fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
 533                     exit(U_ILLEGAL_CHAR_FOUND);
 534                 }
 535
 536             }else if(savedTrieWord != trieWord){
 537                 fprintf(stderr,"Value for codepoint \\U%08X already set!.\n", (int)start);
 538                 exit(U_ILLEGAL_ARGUMENT_ERROR);
 539             }
 540             /* if savedTrieWord == trieWord .. fall through and set the value */
 541         }
 542         if(!utrie_set32(sprepTrie,start,trieWord)){
 543             fprintf(stderr,"Could not set the value for code point \\U%08X.\n", (int)start);
 544             exit(U_ILLEGAL_ARGUMENT_ERROR);
 545         }
 546     }else{
 547         if(!utrie_setRange32(sprepTrie, start, end+1, trieWord, FALSE)){
 548             fprintf(stderr,"Value for certain codepoint already set.\n");
 549             exit(U_ILLEGAL_CHAR_FOUND);
 550         }
 551     }
 552
 553 }
 554
 555 /* folding value: just store the offset (16 bits) if there is any non-0 entry */
 556 static uint32_t U_CALLCONV
 557 getFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) {
 558     uint32_t foldedValue, value;
 559     UChar32 limit=0;
 560     UBool inBlockZero;
 561
 562     foldedValue=0;
 563
 564     limit=start+0x400;
 565     while(start<limit) {
 566         value=utrie_get32(trie, start, &inBlockZero);
 567         if(inBlockZero) {
 568             start+=UTRIE_DATA_BLOCK_LENGTH;
 569         } else if(value!=0) {
 570             return (uint32_t)offset;
 571         } else {
 572             ++start;
 573         }
 574     }
 575     return 0;
 576
 577 }
 578
 579 #endif /* #if !UCONFIG_NO_IDNA */
 580
 581 extern void
 582 generateData(const char *dataDir, const char* bundleName) {
 583     static uint8_t sprepTrieBlock[100000];
 584
 585     UNewDataMemory *pData;
 586     UErrorCode errorCode=U_ZERO_ERROR;
 587     int32_t size, dataLength;
 588     char* fileName = (char*) uprv_malloc(uprv_strlen(bundleName) +100);
 589
 590 #if UCONFIG_NO_IDNA
 591
 592     size=0;
 593
 594 #else
 595
 596     int32_t sprepTrieSize;
 597
 598     /* sort and add mapping data */
 599     storeMappingData();
 600
 601     sprepTrieSize=utrie_serialize(sprepTrie, sprepTrieBlock, sizeof(sprepTrieBlock), getFoldedValue, TRUE, &errorCode);
 602     if(U_FAILURE(errorCode)) {
 603         fprintf(stderr, "error: utrie_serialize(sprep trie) failed, %s\n", u_errorName(errorCode));
 604         exit(errorCode);
 605     }
 606
 607     size = sprepTrieSize + mappingDataCapacity*U_SIZEOF_UCHAR + sizeof(indexes);
 608     if(beVerbose) {
 609         printf("size of sprep trie              %5u bytes\n", (int)sprepTrieSize);
 610         printf("size of " U_ICUDATA_NAME "_%s." DATA_TYPE " contents: %ld bytes\n", bundleName,(long)size);
 611         printf("size of mapping data array %5u bytes\n",(int)mappingDataCapacity * U_SIZEOF_UCHAR);
 612         printf("Number of code units in mappingData (currentIndex) are: %i \n", currentIndex);
 613         printf("Maximum length of the mapping string is : %i \n", (int)maxLength);
 614     }
 615
 616 #endif
 617
 618     fileName[0]=0;
 619     uprv_strcat(fileName,bundleName);
 620     /* write the data */
 621     pData=udata_create(dataDir, DATA_TYPE, fileName, &dataInfo,
 622                        haveCopyright ? U_COPYRIGHT_STRING : NULL, &errorCode);
 623     if(U_FAILURE(errorCode)) {
 624         fprintf(stderr, "gensprep: unable to create the output file, error %d\n", errorCode);
 625         exit(errorCode);
 626     }
 627
 628 #if !UCONFIG_NO_IDNA
 629
 630     indexes[_SPREP_INDEX_TRIE_SIZE]=sprepTrieSize;
 631     indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]=mappingDataCapacity*U_SIZEOF_UCHAR;
 632
 633     udata_writeBlock(pData, indexes, sizeof(indexes));
 634     udata_writeBlock(pData, sprepTrieBlock, sprepTrieSize);
 635     udata_writeBlock(pData, mappingData, indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]);
 636
 637
 638 #endif
 639
 640     /* finish up */
 641     dataLength=udata_finish(pData, &errorCode);
 642     if(U_FAILURE(errorCode)) {
 643         fprintf(stderr, "gensprep: error %d writing the output file\n", errorCode);
 644         exit(errorCode);
 645     }
 646
 647     if(dataLength!=size) {
 648         fprintf(stderr, "gensprep error: data length %ld != calculated size %ld\n",
 649             (long)dataLength, (long)size);
 650         exit(U_INTERNAL_PROGRAM_ERROR);
 651     }
 652
 653 #if !UCONFIG_NO_IDNA
 654     /* done with writing the data .. close the hashtable */
 655     if (hashTable != NULL) {
 656         uhash_close(hashTable);
 657     }
 658 #endif
 659 }
 660
 661 #if !UCONFIG_NO_IDNA
 662
 663 extern void
 664 cleanUpData(void) {
 665
 666     utrie_close(sprepTrie);
 667     uprv_free(sprepTrie);
 668 }
 669
 670 #endif /* #if !UCONFIG_NO_IDNA */
 671
 672 /*
 673  * Hey, Emacs, please set the following:
 674  *
 675  * Local Variables:
 676  * indent-tabs-mode: nil
 677  * End:
 678  *
 679  */