icuSources/tools/gensprep/store.c

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 1999-2014, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  store.c
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2003-02-06
  16 *   created by: Ram Viswanadha
  17 *
  18 */
  19
  20 #include <stdio.h>
  21 #include <stdlib.h>
  22 #include "unicode/utypes.h"
  23 #include "cmemory.h"
  24 #include "cstring.h"
  25 #include "filestrm.h"
  26 #include "unicode/udata.h"
  27 #include "unicode/utf16.h"
  28 #include "utrie.h"
  29 #include "unewdata.h"
  30 #include "gensprep.h"
  31 #include "uhash.h"
  32
  33
  34 #define DO_DEBUG_OUT 0
  35
  36
  37 /*
  38  * StringPrep profile file format ------------------------------------
  39  *
  40  * The file format prepared and written here contains a 16-bit trie and a mapping table.
  41  *
  42  * Before the data contents described below, there are the headers required by
  43  * the udata API for loading ICU data. Especially, a UDataInfo structure
  44  * precedes the actual data. It contains platform properties values and the
  45  * file format version.
  46  *
  47  * The following is a description of format version 2.
  48  *
  49  * Data contents:
  50  *
  51  * The contents is a parsed, binary form of RFC3454 and possibly
  52  * NormalizationCorrections.txt depending on the options specified on the profile.
  53  *
  54  * Any Unicode code point from 0 to 0x10ffff can be looked up to get
  55  * the trie-word, if any, for that code point. This means that the input
  56  * to the lookup are 21-bit unsigned integers, with not all of the
  57  * 21-bit range used.
  58  *
  59  * *.spp files customarily begin with a UDataInfo structure, see udata.h and .c.
  60  * After that there are the following structures:
  61  *
  62  * int32_t indexes[_SPREP_INDEX_TOP];           -- _SPREP_INDEX_TOP=16, see enum in sprpimpl.h file
  63  *
  64  * UTrie stringPrepTrie;                        -- size in bytes=indexes[_SPREP_INDEX_TRIE_SIZE]
  65  *
  66  * uint16_t mappingTable[];                     -- Contains the sequecence of code units that the code point maps to
  67  *                                                 size in bytes = indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]
  68  *
  69  * The indexes array contains the following values:
  70  *  indexes[_SPREP_INDEX_TRIE_SIZE]                  -- The size of the StringPrep trie in bytes
  71  *  indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]          -- The size of the mappingTable in bytes
  72  *  indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION]  -- The index of Unicode version of last entry in NormalizationCorrections.txt
  73  *  indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START]    -- The starting index of 1 UChar  mapping index in the mapping table
  74  *  indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]   -- The starting index of 2 UChars mapping index in the mapping table
  75  *  indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] -- The starting index of 3 UChars mapping index in the mapping table
  76  *  indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]  -- The starting index of 4 UChars mapping index in the mapping table
  77  *  indexes[_SPREP_OPTIONS]                          -- Bit set of options to turn on in the profile, e.g: USPREP_NORMALIZATION_ON, USPREP_CHECK_BIDI_ON
  78  *
  79  *
  80  * StringPrep Trie :
  81  *
  82  * The StringPrep tries is a 16-bit trie that contains data for the profile.
  83  * Each code point is associated with a value (trie-word) in the trie.
  84  *
  85  * - structure of data words from the trie
  86  *
  87  *  i)  A value greater than or equal to _SPREP_TYPE_THRESHOLD (0xFFF0)
  88  *      represents the type associated with the code point
  89  *      if(trieWord >= _SPREP_TYPE_THRESHOLD){
  90  *          type = trieWord - 0xFFF0;
  91  *      }
  92  *      The type can be :
  93  *             USPREP_UNASSIGNED
  94  *             USPREP_PROHIBITED
  95  *             USPREP_DELETE
  96  *
  97  *  ii) A value less than _SPREP_TYPE_THRESHOLD means the type is USPREP_MAP and
  98  *      contains distribution described below
  99  *
 100  *      0       -  ON : The code point is prohibited (USPREP_PROHIBITED). This is to allow for codepoint that are both prohibited and mapped.
 101  *      1       -  ON : The value in the next 14 bits is an index into the mapping table
 102  *                 OFF: The value in the next 14 bits is an delta value from the code point
 103  *      2..15   -  Contains data as described by bit 1. If all bits are set
 104  *                 (value = _SPREP_MAX_INDEX_VALUE) then the type is USPREP_DELETE
 105  *
 106  *
 107  * Mapping Table:
 108  * The data in mapping table is sorted according to the length of the mapping sequence.
 109  * If the type of the code point is USPREP_MAP and value in trie word is an index, the index
 110  * is compared with start indexes of sequence length start to figure out the length according to
 111  * the following algorithm:
 112  *
 113  *              if(       index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
 114  *                        index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
 115  *                   length = 1;
 116  *               }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
 117  *                        index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
 118  *                   length = 2;
 119  *               }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
 120  *                        index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
 121  *                   length = 3;
 122  *               }else{
 123  *                   // The first position in the mapping table contains the length
 124  *                   // of the sequence
 125  *                   length = mappingTable[index++];
 126  *
 127  *               }
 128  *
 129  */
 130
 131 /* file data ---------------------------------------------------------------- */
 132 /* indexes[] value names */
 133
 134 #if UCONFIG_NO_IDNA
 135
 136 /* dummy UDataInfo cf. udata.h */
 137 static UDataInfo dataInfo = {
 138     sizeof(UDataInfo),
 139     0,
 140
 141     U_IS_BIG_ENDIAN,
 142     U_CHARSET_FAMILY,
 143     U_SIZEOF_UCHAR,
 144     0,
 145
 146     { 0, 0, 0, 0 },                 /* dummy dataFormat */
 147     { 0, 0, 0, 0 },                 /* dummy formatVersion */
 148     { 0, 0, 0, 0 }                  /* dummy dataVersion */
 149 };
 150
 151 #else
 152
 153 static int32_t indexes[_SPREP_INDEX_TOP]={ 0 };
 154
 155 static uint16_t* mappingData= NULL;
 156 static int32_t mappingDataCapacity = 0; /* we skip the first index in mapping data */
 157 static int16_t currentIndex = 0; /* the current index into the data trie */
 158 static int32_t maxLength = 0;  /* maximum length of mapping string */
 159
 160
 161 /* UDataInfo cf. udata.h */
 162 static UDataInfo dataInfo={
 163     sizeof(UDataInfo),
 164     0,
 165
 166     U_IS_BIG_ENDIAN,
 167     U_CHARSET_FAMILY,
 168     U_SIZEOF_UCHAR,
 169     0,
 170
 171     { 0x53, 0x50, 0x52, 0x50 },                 /* dataFormat="SPRP" */
 172     { 3, 2, UTRIE_SHIFT, UTRIE_INDEX_SHIFT },   /* formatVersion */
 173     { 3, 2, 0, 0 }                              /* dataVersion (Unicode version) */
 174 };
 175 void
 176 setUnicodeVersion(const char *v) {
 177     UVersionInfo version;
 178     u_versionFromString(version, v);
 179     uprv_memcpy(dataInfo.dataVersion, version, 4);
 180 }
 181
 182 void
 183 setUnicodeVersionNC(UVersionInfo version){
 184     uint32_t univer = version[0] << 24;
 185     univer += version[1] << 16;
 186     univer += version[2] << 8;
 187     univer += version[3];
 188     indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION] = univer;
 189 }
 190 static UNewTrie *sprepTrie;
 191
 192 #define MAX_DATA_LENGTH 11500
 193
 194
 195 #define SPREP_DELTA_RANGE_POSITIVE_LIMIT              8191
 196 #define SPREP_DELTA_RANGE_NEGATIVE_LIMIT              -8192
 197
 198
 199 extern void
 200 init() {
 201
 202     sprepTrie = (UNewTrie *)uprv_calloc(1, sizeof(UNewTrie));
 203
 204     /* initialize the two tries */
 205     if(NULL==utrie_open(sprepTrie, NULL, MAX_DATA_LENGTH, 0, 0, FALSE)) {
 206         fprintf(stderr, "error: failed to initialize tries\n");
 207         exit(U_MEMORY_ALLOCATION_ERROR);
 208     }
 209 }
 210
 211 static UHashtable* hashTable = NULL;
 212
 213
 214 typedef struct ValueStruct {
 215     UChar* mapping;
 216     int16_t length;
 217     UStringPrepType type;
 218 } ValueStruct;
 219
 220 /* Callback for deleting the value from the hashtable */
 221 static void U_CALLCONV valueDeleter(void* obj){
 222     ValueStruct* value = (ValueStruct*) obj;
 223     uprv_free(value->mapping);
 224     uprv_free(value);
 225 }
 226
 227 /* Callback for hashing the entry */
 228 static int32_t U_CALLCONV hashEntry(const UHashTok parm) {
 229     return  parm.integer;
 230 }
 231
 232 /* Callback for comparing two entries */
 233 static UBool U_CALLCONV compareEntries(const UHashTok p1, const UHashTok p2) {
 234     return (UBool)(p1.integer != p2.integer);
 235 }
 236
 237
 238 static void
 239 storeMappingData(){
 240
 241     int32_t pos = UHASH_FIRST;
 242     const UHashElement* element = NULL;
 243     ValueStruct* value  = NULL;
 244     int32_t codepoint = 0;
 245     int32_t elementCount = 0;
 246     int32_t writtenElementCount = 0;
 247     int32_t mappingLength = 1; /* minimum mapping length */
 248     int32_t oldMappingLength = 0;
 249     uint16_t trieWord =0;
 250     int32_t limitIndex = 0;
 251
 252     if (hashTable == NULL) {
 253         return;
 254     }
 255     elementCount = uhash_count(hashTable);
 256
 257         /*initialize the mapping data */
 258     mappingData = (uint16_t*) uprv_calloc(mappingDataCapacity, U_SIZEOF_UCHAR);
 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",
 320                         currentIndex+value->length, _SPREP_MAX_INDEX_VALUE);
 321                     exit(U_INDEX_OUTOFBOUNDS_ERROR);
 322                 }
 323
 324                 /* copy the mapping data */
 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                 u_memmove(mappingData+currentIndex, value->mapping, value->length);
 332                 currentIndex += value->length;
 333                 if (currentIndex > mappingDataCapacity) {
 334                     /* If this happens there is a bug in the computation of the mapping data size in storeMapping() */
 335                     fprintf(stderr, "gensprep, fatal error at %s, %d.  Aborting.\n", __FILE__, __LINE__);
 336                     exit(U_INTERNAL_PROGRAM_ERROR);
 337                 }
 338             }
 339         }
 340         mappingLength++;
 341         pos = -1;
 342     }
 343     /* set the last length for range check */
 344     if(mappingLength <= _SPREP_MAX_INDEX_TOP_LENGTH){
 345         indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION+mappingLength] = currentIndex+1;
 346     }else{
 347         indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START] = limitIndex;
 348     }
 349
 350 }
 351
 352 extern void setOptions(int32_t options){
 353     indexes[_SPREP_OPTIONS] = options;
 354 }
 355 extern void
 356 storeMapping(uint32_t codepoint, uint32_t* mapping,int32_t length,
 357              UStringPrepType type, UErrorCode* status){
 358
 359
 360     UChar* map = NULL;
 361     int16_t adjustedLen=0, i, j;
 362     uint16_t trieWord = 0;
 363     ValueStruct *value = NULL;
 364     uint32_t savedTrieWord = 0;
 365
 366     /* initialize the hashtable */
 367     if(hashTable==NULL){
 368         hashTable = uhash_open(hashEntry, compareEntries, NULL, status);
 369         uhash_setValueDeleter(hashTable, valueDeleter);
 370     }
 371
 372     /* figure out if the code point has type already stored */
 373     savedTrieWord= utrie_get32(sprepTrie,codepoint,NULL);
 374     if(savedTrieWord!=0){
 375         if((savedTrieWord- _SPREP_TYPE_THRESHOLD) == USPREP_PROHIBITED){
 376             /* turn on the first bit in trie word */
 377             trieWord += 0x01;
 378         }else{
 379             /*
 380              * the codepoint has value something other than prohibited
 381              * and a mapping .. error!
 382              */
 383             fprintf(stderr,"Type for codepoint \\U%08X already set!.\n", (int)codepoint);
 384             exit(U_ILLEGAL_ARGUMENT_ERROR);
 385         }
 386     }
 387
 388     /* figure out the real length */
 389     for(i=0; i<length; i++){
 390         adjustedLen += U16_LENGTH(mapping[i]);
 391     }
 392
 393     if(adjustedLen == 0){
 394         trieWord = (uint16_t)(_SPREP_MAX_INDEX_VALUE << 2);
 395         /* make sure that the value of trieWord is less than the threshold */
 396         if(trieWord < _SPREP_TYPE_THRESHOLD){
 397             /* now set the value in the trie */
 398             if(!utrie_set32(sprepTrie,codepoint,trieWord)){
 399                 fprintf(stderr,"Could not set the value for code point.\n");
 400                 exit(U_ILLEGAL_ARGUMENT_ERROR);
 401             }
 402             /* value is set so just return */
 403             return;
 404         }else{
 405             fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
 406             exit(U_ILLEGAL_CHAR_FOUND);
 407         }
 408     }
 409
 410     if(adjustedLen == 1){
 411         /* calculate the delta */
 412         int16_t delta = (int16_t)((int32_t)codepoint - (int16_t) mapping[0]);
 413         if(delta >= SPREP_DELTA_RANGE_NEGATIVE_LIMIT && delta <= SPREP_DELTA_RANGE_POSITIVE_LIMIT){
 414
 415             trieWord = delta << 2;
 416
 417
 418             /* make sure that the second bit is OFF */
 419             if((trieWord & 0x02) != 0 ){
 420                 fprintf(stderr,"The second bit in the trie word is not zero while storing a delta.\n");
 421                 exit(U_INTERNAL_PROGRAM_ERROR);
 422             }
 423             /* make sure that the value of trieWord is less than the threshold */
 424             if(trieWord < _SPREP_TYPE_THRESHOLD){
 425                 /* now set the value in the trie */
 426                 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
 427                     fprintf(stderr,"Could not set the value for code point.\n");
 428                     exit(U_ILLEGAL_ARGUMENT_ERROR);
 429                 }
 430                 /* value is set so just return */
 431                 return;
 432             }
 433         }
 434         /*
 435          * if the delta is not in the given range or if the trieWord is larger than the threshold
 436          * just fall through for storing the mapping in the mapping table
 437          */
 438     }
 439
 440     map = (UChar*) uprv_calloc(adjustedLen + 1, U_SIZEOF_UCHAR);
 441
 442     for (i=0, j=0; i<length; i++) {
 443         U16_APPEND_UNSAFE(map, j, mapping[i]);
 444     }
 445
 446     value = (ValueStruct*) uprv_malloc(sizeof(ValueStruct));
 447     value->mapping = map;
 448     value->type    = type;
 449     value->length  = adjustedLen;
 450     if(value->length > _SPREP_MAX_INDEX_TOP_LENGTH){
 451         mappingDataCapacity++;
 452     }
 453     if(maxLength < value->length){
 454         maxLength = value->length;
 455     }
 456     uhash_iput(hashTable,codepoint,value,status);
 457     mappingDataCapacity += adjustedLen;
 458
 459     if(U_FAILURE(*status)){
 460         fprintf(stderr, "Failed to put entries into the hastable. Error: %s\n", u_errorName(*status));
 461         exit(*status);
 462     }
 463 }
 464
 465
 466 extern void
 467 storeRange(uint32_t start, uint32_t end, UStringPrepType type,UErrorCode* status){
 468     uint16_t trieWord = 0;
 469
 470     if((int)(_SPREP_TYPE_THRESHOLD + type) > 0xFFFF){
 471         fprintf(stderr,"trieWord cannot contain value greater than 0xFFFF.\n");
 472         exit(U_ILLEGAL_CHAR_FOUND);
 473     }
 474     trieWord = (_SPREP_TYPE_THRESHOLD + type); /* the top 4 bits contain the value */
 475     if(start == end){
 476         uint32_t savedTrieWord = utrie_get32(sprepTrie, start, NULL);
 477         if(savedTrieWord>0){
 478             if(savedTrieWord < _SPREP_TYPE_THRESHOLD && type == USPREP_PROHIBITED){
 479                 /*
 480                  * A mapping is stored in the trie word
 481                  * and the only other possible type that a
 482                  * code point can have is USPREP_PROHIBITED
 483                  *
 484                  */
 485
 486                 /* turn on the 0th bit in the savedTrieWord */
 487                 savedTrieWord += 0x01;
 488
 489                 /* the downcast is safe since we only save 16 bit values */
 490                 trieWord = (uint16_t)savedTrieWord;
 491
 492                 /* make sure that the value of trieWord is less than the threshold */
 493                 if(trieWord < _SPREP_TYPE_THRESHOLD){
 494                     /* now set the value in the trie */
 495                     if(!utrie_set32(sprepTrie,start,trieWord)){
 496                         fprintf(stderr,"Could not set the value for code point.\n");
 497                         exit(U_ILLEGAL_ARGUMENT_ERROR);
 498                     }
 499                     /* value is set so just return */
 500                     return;
 501                 }else{
 502                     fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
 503                     exit(U_ILLEGAL_CHAR_FOUND);
 504                 }
 505
 506             }else if(savedTrieWord != trieWord){
 507                 fprintf(stderr,"Value for codepoint \\U%08X already set!.\n", (int)start);
 508                 exit(U_ILLEGAL_ARGUMENT_ERROR);
 509             }
 510             /* if savedTrieWord == trieWord .. fall through and set the value */
 511         }
 512         if(!utrie_set32(sprepTrie,start,trieWord)){
 513             fprintf(stderr,"Could not set the value for code point \\U%08X.\n", (int)start);
 514             exit(U_ILLEGAL_ARGUMENT_ERROR);
 515         }
 516     }else{
 517         if(!utrie_setRange32(sprepTrie, start, end+1, trieWord, FALSE)){
 518             fprintf(stderr,"Value for certain codepoint already set.\n");
 519             exit(U_ILLEGAL_CHAR_FOUND);
 520         }
 521     }
 522
 523 }
 524
 525 /* folding value: just store the offset (16 bits) if there is any non-0 entry */
 526 static uint32_t U_CALLCONV
 527 getFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) {
 528     uint32_t value;
 529     UChar32 limit=0;
 530     UBool inBlockZero;
 531
 532     limit=start+0x400;
 533     while(start<limit) {
 534         value=utrie_get32(trie, start, &inBlockZero);
 535         if(inBlockZero) {
 536             start+=UTRIE_DATA_BLOCK_LENGTH;
 537         } else if(value!=0) {
 538             return (uint32_t)offset;
 539         } else {
 540             ++start;
 541         }
 542     }
 543     return 0;
 544
 545 }
 546
 547 #endif /* #if !UCONFIG_NO_IDNA */
 548
 549 extern void
 550 generateData(const char *dataDir, const char* bundleName) {
 551     static uint8_t sprepTrieBlock[100000];
 552
 553     UNewDataMemory *pData;
 554     UErrorCode errorCode=U_ZERO_ERROR;
 555     int32_t size, dataLength;
 556     char* fileName = (char*) uprv_malloc(uprv_strlen(bundleName) +100);
 557
 558 #if UCONFIG_NO_IDNA
 559
 560     size=0;
 561
 562 #else
 563
 564     int32_t sprepTrieSize;
 565
 566     /* sort and add mapping data */
 567     storeMappingData();
 568
 569     sprepTrieSize=utrie_serialize(sprepTrie, sprepTrieBlock, sizeof(sprepTrieBlock), getFoldedValue, TRUE, &errorCode);
 570     if(U_FAILURE(errorCode)) {
 571         fprintf(stderr, "error: utrie_serialize(sprep trie) failed, %s\n", u_errorName(errorCode));
 572         exit(errorCode);
 573     }
 574
 575     size = sprepTrieSize + mappingDataCapacity*U_SIZEOF_UCHAR + sizeof(indexes);
 576     if(beVerbose) {
 577         printf("size of sprep trie              %5u bytes\n", (int)sprepTrieSize);
 578         printf("size of " U_ICUDATA_NAME "_%s." DATA_TYPE " contents: %ld bytes\n", bundleName,(long)size);
 579         printf("size of mapping data array %5u bytes\n",(int)mappingDataCapacity * U_SIZEOF_UCHAR);
 580         printf("Number of code units in mappingData (currentIndex) are: %i \n", currentIndex);
 581         printf("Maximum length of the mapping string is : %i \n", (int)maxLength);
 582     }
 583
 584 #endif
 585
 586     fileName[0]=0;
 587     uprv_strcat(fileName,bundleName);
 588     /* write the data */
 589     pData=udata_create(dataDir, DATA_TYPE, fileName, &dataInfo,
 590                        haveCopyright ? U_COPYRIGHT_STRING : NULL, &errorCode);
 591     if(U_FAILURE(errorCode)) {
 592         fprintf(stderr, "gensprep: unable to create the output file, error %d\n", errorCode);
 593         exit(errorCode);
 594     }
 595
 596 #if !UCONFIG_NO_IDNA
 597
 598     indexes[_SPREP_INDEX_TRIE_SIZE]=sprepTrieSize;
 599     indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]=mappingDataCapacity*U_SIZEOF_UCHAR;
 600
 601     udata_writeBlock(pData, indexes, sizeof(indexes));
 602     udata_writeBlock(pData, sprepTrieBlock, sprepTrieSize);
 603     udata_writeBlock(pData, mappingData, indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]);
 604
 605
 606 #endif
 607
 608     /* finish up */
 609     dataLength=udata_finish(pData, &errorCode);
 610     if(U_FAILURE(errorCode)) {
 611         fprintf(stderr, "gensprep: error %d writing the output file\n", errorCode);
 612         exit(errorCode);
 613     }
 614
 615     if(dataLength!=size) {
 616         fprintf(stderr, "gensprep error: data length %ld != calculated size %ld\n",
 617             (long)dataLength, (long)size);
 618         exit(U_INTERNAL_PROGRAM_ERROR);
 619     }
 620
 621 #if !UCONFIG_NO_IDNA
 622     /* done with writing the data .. close the hashtable */
 623     if (hashTable != NULL) {
 624         uhash_close(hashTable);
 625     }
 626 #endif
 627
 628     uprv_free(fileName);
 629 }
 630
 631 #if !UCONFIG_NO_IDNA
 632
 633 extern void
 634 cleanUpData(void) {
 635     uprv_free(mappingData);
 636     utrie_close(sprepTrie);
 637     uprv_free(sprepTrie);
 638 }
 639
 640 #endif /* #if !UCONFIG_NO_IDNA */
 641
 642 /*
 643  * Hey, Emacs, please set the following:
 644  *
 645  * Local Variables:
 646  * indent-tabs-mode: nil
 647  * End:
 648  *
 649  */