intl/icu/source/tools/gensprep/store.c

Wed, 31 Dec 2014 07:22:50 +0100

author
Michael Schloh von Bennewitz <michael@schloh.com>
date
Wed, 31 Dec 2014 07:22:50 +0100
branch
TOR_BUG_3246
changeset 4
fc2d59ddac77
permissions
-rw-r--r--

Correct previous dual key logic pending first delivery installment.

michael@0 1 /*
michael@0 2 *******************************************************************************
michael@0 3 *
michael@0 4 * Copyright (C) 1999-2012, International Business Machines
michael@0 5 * Corporation and others. All Rights Reserved.
michael@0 6 *
michael@0 7 *******************************************************************************
michael@0 8 * file name: store.c
michael@0 9 * encoding: US-ASCII
michael@0 10 * tab size: 8 (not used)
michael@0 11 * indentation:4
michael@0 12 *
michael@0 13 * created on: 2003-02-06
michael@0 14 * created by: Ram Viswanadha
michael@0 15 *
michael@0 16 */
michael@0 17
michael@0 18 #include <stdio.h>
michael@0 19 #include <stdlib.h>
michael@0 20 #include "unicode/utypes.h"
michael@0 21 #include "cmemory.h"
michael@0 22 #include "cstring.h"
michael@0 23 #include "filestrm.h"
michael@0 24 #include "unicode/udata.h"
michael@0 25 #include "unicode/utf16.h"
michael@0 26 #include "utrie.h"
michael@0 27 #include "unewdata.h"
michael@0 28 #include "gensprep.h"
michael@0 29 #include "uhash.h"
michael@0 30
michael@0 31
michael@0 32 #define DO_DEBUG_OUT 0
michael@0 33
michael@0 34
michael@0 35 /*
michael@0 36 * StringPrep profile file format ------------------------------------
michael@0 37 *
michael@0 38 * The file format prepared and written here contains a 16-bit trie and a mapping table.
michael@0 39 *
michael@0 40 * Before the data contents described below, there are the headers required by
michael@0 41 * the udata API for loading ICU data. Especially, a UDataInfo structure
michael@0 42 * precedes the actual data. It contains platform properties values and the
michael@0 43 * file format version.
michael@0 44 *
michael@0 45 * The following is a description of format version 2.
michael@0 46 *
michael@0 47 * Data contents:
michael@0 48 *
michael@0 49 * The contents is a parsed, binary form of RFC3454 and possibly
michael@0 50 * NormalizationCorrections.txt depending on the options specified on the profile.
michael@0 51 *
michael@0 52 * Any Unicode code point from 0 to 0x10ffff can be looked up to get
michael@0 53 * the trie-word, if any, for that code point. This means that the input
michael@0 54 * to the lookup are 21-bit unsigned integers, with not all of the
michael@0 55 * 21-bit range used.
michael@0 56 *
michael@0 57 * *.spp files customarily begin with a UDataInfo structure, see udata.h and .c.
michael@0 58 * After that there are the following structures:
michael@0 59 *
michael@0 60 * int32_t indexes[_SPREP_INDEX_TOP]; -- _SPREP_INDEX_TOP=16, see enum in sprpimpl.h file
michael@0 61 *
michael@0 62 * UTrie stringPrepTrie; -- size in bytes=indexes[_SPREP_INDEX_TRIE_SIZE]
michael@0 63 *
michael@0 64 * uint16_t mappingTable[]; -- Contains the sequecence of code units that the code point maps to
michael@0 65 * size in bytes = indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]
michael@0 66 *
michael@0 67 * The indexes array contains the following values:
michael@0 68 * indexes[_SPREP_INDEX_TRIE_SIZE] -- The size of the StringPrep trie in bytes
michael@0 69 * indexes[_SPREP_INDEX_MAPPING_DATA_SIZE] -- The size of the mappingTable in bytes
michael@0 70 * indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION] -- The index of Unicode version of last entry in NormalizationCorrections.txt
michael@0 71 * indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] -- The starting index of 1 UChar mapping index in the mapping table
michael@0 72 * indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] -- The starting index of 2 UChars mapping index in the mapping table
michael@0 73 * indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] -- The starting index of 3 UChars mapping index in the mapping table
michael@0 74 * indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START] -- The starting index of 4 UChars mapping index in the mapping table
michael@0 75 * indexes[_SPREP_OPTIONS] -- Bit set of options to turn on in the profile, e.g: USPREP_NORMALIZATION_ON, USPREP_CHECK_BIDI_ON
michael@0 76 *
michael@0 77 *
michael@0 78 * StringPrep Trie :
michael@0 79 *
michael@0 80 * The StringPrep tries is a 16-bit trie that contains data for the profile.
michael@0 81 * Each code point is associated with a value (trie-word) in the trie.
michael@0 82 *
michael@0 83 * - structure of data words from the trie
michael@0 84 *
michael@0 85 * i) A value greater than or equal to _SPREP_TYPE_THRESHOLD (0xFFF0)
michael@0 86 * represents the type associated with the code point
michael@0 87 * if(trieWord >= _SPREP_TYPE_THRESHOLD){
michael@0 88 * type = trieWord - 0xFFF0;
michael@0 89 * }
michael@0 90 * The type can be :
michael@0 91 * USPREP_UNASSIGNED
michael@0 92 * USPREP_PROHIBITED
michael@0 93 * USPREP_DELETE
michael@0 94 *
michael@0 95 * ii) A value less than _SPREP_TYPE_THRESHOLD means the type is USPREP_MAP and
michael@0 96 * contains distribution described below
michael@0 97 *
michael@0 98 * 0 - ON : The code point is prohibited (USPREP_PROHIBITED). This is to allow for codepoint that are both prohibited and mapped.
michael@0 99 * 1 - ON : The value in the next 14 bits is an index into the mapping table
michael@0 100 * OFF: The value in the next 14 bits is an delta value from the code point
michael@0 101 * 2..15 - Contains data as described by bit 1. If all bits are set
michael@0 102 * (value = _SPREP_MAX_INDEX_VALUE) then the type is USPREP_DELETE
michael@0 103 *
michael@0 104 *
michael@0 105 * Mapping Table:
michael@0 106 * The data in mapping table is sorted according to the length of the mapping sequence.
michael@0 107 * If the type of the code point is USPREP_MAP and value in trie word is an index, the index
michael@0 108 * is compared with start indexes of sequence length start to figure out the length according to
michael@0 109 * the following algorithm:
michael@0 110 *
michael@0 111 * if( index >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
michael@0 112 * index < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
michael@0 113 * length = 1;
michael@0 114 * }else if(index >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
michael@0 115 * index < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
michael@0 116 * length = 2;
michael@0 117 * }else if(index >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
michael@0 118 * index < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
michael@0 119 * length = 3;
michael@0 120 * }else{
michael@0 121 * // The first position in the mapping table contains the length
michael@0 122 * // of the sequence
michael@0 123 * length = mappingTable[index++];
michael@0 124 *
michael@0 125 * }
michael@0 126 *
michael@0 127 */
michael@0 128
michael@0 129 /* file data ---------------------------------------------------------------- */
michael@0 130 /* indexes[] value names */
michael@0 131
michael@0 132 #if UCONFIG_NO_IDNA
michael@0 133
michael@0 134 /* dummy UDataInfo cf. udata.h */
michael@0 135 static UDataInfo dataInfo = {
michael@0 136 sizeof(UDataInfo),
michael@0 137 0,
michael@0 138
michael@0 139 U_IS_BIG_ENDIAN,
michael@0 140 U_CHARSET_FAMILY,
michael@0 141 U_SIZEOF_UCHAR,
michael@0 142 0,
michael@0 143
michael@0 144 { 0, 0, 0, 0 }, /* dummy dataFormat */
michael@0 145 { 0, 0, 0, 0 }, /* dummy formatVersion */
michael@0 146 { 0, 0, 0, 0 } /* dummy dataVersion */
michael@0 147 };
michael@0 148
michael@0 149 #else
michael@0 150
michael@0 151 static int32_t indexes[_SPREP_INDEX_TOP]={ 0 };
michael@0 152
michael@0 153 static uint16_t* mappingData= NULL;
michael@0 154 static int32_t mappingDataCapacity = 0; /* we skip the first index in mapping data */
michael@0 155 static int16_t currentIndex = 0; /* the current index into the data trie */
michael@0 156 static int32_t maxLength = 0; /* maximum length of mapping string */
michael@0 157
michael@0 158
michael@0 159 /* UDataInfo cf. udata.h */
michael@0 160 static UDataInfo dataInfo={
michael@0 161 sizeof(UDataInfo),
michael@0 162 0,
michael@0 163
michael@0 164 U_IS_BIG_ENDIAN,
michael@0 165 U_CHARSET_FAMILY,
michael@0 166 U_SIZEOF_UCHAR,
michael@0 167 0,
michael@0 168
michael@0 169 { 0x53, 0x50, 0x52, 0x50 }, /* dataFormat="SPRP" */
michael@0 170 { 3, 2, UTRIE_SHIFT, UTRIE_INDEX_SHIFT }, /* formatVersion */
michael@0 171 { 3, 2, 0, 0 } /* dataVersion (Unicode version) */
michael@0 172 };
michael@0 173 void
michael@0 174 setUnicodeVersion(const char *v) {
michael@0 175 UVersionInfo version;
michael@0 176 u_versionFromString(version, v);
michael@0 177 uprv_memcpy(dataInfo.dataVersion, version, 4);
michael@0 178 }
michael@0 179
michael@0 180 void
michael@0 181 setUnicodeVersionNC(UVersionInfo version){
michael@0 182 uint32_t univer = version[0] << 24;
michael@0 183 univer += version[1] << 16;
michael@0 184 univer += version[2] << 8;
michael@0 185 univer += version[3];
michael@0 186 indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION] = univer;
michael@0 187 }
michael@0 188 static UNewTrie *sprepTrie;
michael@0 189
michael@0 190 #define MAX_DATA_LENGTH 11500
michael@0 191
michael@0 192
michael@0 193 #define SPREP_DELTA_RANGE_POSITIVE_LIMIT 8191
michael@0 194 #define SPREP_DELTA_RANGE_NEGATIVE_LIMIT -8192
michael@0 195
michael@0 196
michael@0 197 extern void
michael@0 198 init() {
michael@0 199
michael@0 200 sprepTrie = (UNewTrie *)uprv_calloc(1, sizeof(UNewTrie));
michael@0 201
michael@0 202 /* initialize the two tries */
michael@0 203 if(NULL==utrie_open(sprepTrie, NULL, MAX_DATA_LENGTH, 0, 0, FALSE)) {
michael@0 204 fprintf(stderr, "error: failed to initialize tries\n");
michael@0 205 exit(U_MEMORY_ALLOCATION_ERROR);
michael@0 206 }
michael@0 207 }
michael@0 208
michael@0 209 static UHashtable* hashTable = NULL;
michael@0 210
michael@0 211
michael@0 212 typedef struct ValueStruct {
michael@0 213 UChar* mapping;
michael@0 214 int16_t length;
michael@0 215 UStringPrepType type;
michael@0 216 } ValueStruct;
michael@0 217
michael@0 218 /* Callback for deleting the value from the hashtable */
michael@0 219 static void U_CALLCONV valueDeleter(void* obj){
michael@0 220 ValueStruct* value = (ValueStruct*) obj;
michael@0 221 uprv_free(value->mapping);
michael@0 222 uprv_free(value);
michael@0 223 }
michael@0 224
michael@0 225 /* Callback for hashing the entry */
michael@0 226 static int32_t U_CALLCONV hashEntry(const UHashTok parm) {
michael@0 227 return parm.integer;
michael@0 228 }
michael@0 229
michael@0 230 /* Callback for comparing two entries */
michael@0 231 static UBool U_CALLCONV compareEntries(const UHashTok p1, const UHashTok p2) {
michael@0 232 return (UBool)(p1.integer != p2.integer);
michael@0 233 }
michael@0 234
michael@0 235
michael@0 236 static void
michael@0 237 storeMappingData(){
michael@0 238
michael@0 239 int32_t pos = -1;
michael@0 240 const UHashElement* element = NULL;
michael@0 241 ValueStruct* value = NULL;
michael@0 242 int32_t codepoint = 0;
michael@0 243 int32_t elementCount = 0;
michael@0 244 int32_t writtenElementCount = 0;
michael@0 245 int32_t mappingLength = 1; /* minimum mapping length */
michael@0 246 int32_t oldMappingLength = 0;
michael@0 247 uint16_t trieWord =0;
michael@0 248 int32_t limitIndex = 0;
michael@0 249
michael@0 250 if (hashTable == NULL) {
michael@0 251 return;
michael@0 252 }
michael@0 253 elementCount = uhash_count(hashTable);
michael@0 254
michael@0 255 /*initialize the mapping data */
michael@0 256 mappingData = (uint16_t*) uprv_calloc(mappingDataCapacity, U_SIZEOF_UCHAR);
michael@0 257
michael@0 258 while(writtenElementCount < elementCount){
michael@0 259
michael@0 260 while( (element = uhash_nextElement(hashTable, &pos))!=NULL){
michael@0 261
michael@0 262 codepoint = element->key.integer;
michael@0 263 value = (ValueStruct*)element->value.pointer;
michael@0 264
michael@0 265 /* store the start of indexes */
michael@0 266 if(oldMappingLength != mappingLength){
michael@0 267 /* Assume that index[] is used according to the enums defined */
michael@0 268 if(oldMappingLength <=_SPREP_MAX_INDEX_TOP_LENGTH){
michael@0 269 indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION+mappingLength] = currentIndex;
michael@0 270 }
michael@0 271 if(oldMappingLength <= _SPREP_MAX_INDEX_TOP_LENGTH &&
michael@0 272 mappingLength == _SPREP_MAX_INDEX_TOP_LENGTH +1){
michael@0 273
michael@0 274 limitIndex = currentIndex;
michael@0 275
michael@0 276 }
michael@0 277 oldMappingLength = mappingLength;
michael@0 278 }
michael@0 279
michael@0 280 if(value->length == mappingLength){
michael@0 281 uint32_t savedTrieWord = 0;
michael@0 282 trieWord = currentIndex << 2;
michael@0 283 /* turn on the 2nd bit to signal that the following bits contain an index */
michael@0 284 trieWord += 0x02;
michael@0 285
michael@0 286 if(trieWord > _SPREP_TYPE_THRESHOLD){
michael@0 287 fprintf(stderr,"trieWord cannot contain value greater than 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
michael@0 288 exit(U_ILLEGAL_CHAR_FOUND);
michael@0 289 }
michael@0 290 /* figure out if the code point has type already stored */
michael@0 291 savedTrieWord= utrie_get32(sprepTrie,codepoint,NULL);
michael@0 292 if(savedTrieWord!=0){
michael@0 293 if((savedTrieWord- _SPREP_TYPE_THRESHOLD) == USPREP_PROHIBITED){
michael@0 294 /* turn on the first bit in trie word */
michael@0 295 trieWord += 0x01;
michael@0 296 }else{
michael@0 297 /*
michael@0 298 * the codepoint has value something other than prohibited
michael@0 299 * and a mapping .. error!
michael@0 300 */
michael@0 301 fprintf(stderr,"Type for codepoint \\U%08X already set!.\n", (int)codepoint);
michael@0 302 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 303 }
michael@0 304 }
michael@0 305
michael@0 306 /* now set the value in the trie */
michael@0 307 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
michael@0 308 fprintf(stderr,"Could not set the value for code point.\n");
michael@0 309 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 310 }
michael@0 311
michael@0 312 /* written the trie word for the codepoint... increment the count*/
michael@0 313 writtenElementCount++;
michael@0 314
michael@0 315 /* sanity check are we exceeding the max number allowed */
michael@0 316 if(currentIndex+value->length+1 > _SPREP_MAX_INDEX_VALUE){
michael@0 317 fprintf(stderr, "Too many entries in the mapping table %i. Maximum allowed is %i\n",
michael@0 318 currentIndex+value->length, _SPREP_MAX_INDEX_VALUE);
michael@0 319 exit(U_INDEX_OUTOFBOUNDS_ERROR);
michael@0 320 }
michael@0 321
michael@0 322 /* copy the mapping data */
michael@0 323 /* write the length */
michael@0 324 if(mappingLength > _SPREP_MAX_INDEX_TOP_LENGTH ){
michael@0 325 /* the cast here is safe since we donot expect the length to be > 65535 */
michael@0 326 mappingData[currentIndex++] = (uint16_t) mappingLength;
michael@0 327 }
michael@0 328 /* copy the contents to mappindData array */
michael@0 329 uprv_memmove(mappingData+currentIndex, value->mapping, value->length*U_SIZEOF_UCHAR);
michael@0 330 currentIndex += value->length;
michael@0 331 if (currentIndex > mappingDataCapacity) {
michael@0 332 /* If this happens there is a bug in the computation of the mapping data size in storeMapping() */
michael@0 333 fprintf(stderr, "gensprep, fatal error at %s, %d. Aborting.\n", __FILE__, __LINE__);
michael@0 334 exit(U_INTERNAL_PROGRAM_ERROR);
michael@0 335 }
michael@0 336 }
michael@0 337 }
michael@0 338 mappingLength++;
michael@0 339 pos = -1;
michael@0 340 }
michael@0 341 /* set the last length for range check */
michael@0 342 if(mappingLength <= _SPREP_MAX_INDEX_TOP_LENGTH){
michael@0 343 indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION+mappingLength] = currentIndex+1;
michael@0 344 }else{
michael@0 345 indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START] = limitIndex;
michael@0 346 }
michael@0 347
michael@0 348 }
michael@0 349
michael@0 350 extern void setOptions(int32_t options){
michael@0 351 indexes[_SPREP_OPTIONS] = options;
michael@0 352 }
michael@0 353 extern void
michael@0 354 storeMapping(uint32_t codepoint, uint32_t* mapping,int32_t length,
michael@0 355 UStringPrepType type, UErrorCode* status){
michael@0 356
michael@0 357
michael@0 358 UChar* map = NULL;
michael@0 359 int16_t adjustedLen=0, i, j;
michael@0 360 uint16_t trieWord = 0;
michael@0 361 ValueStruct *value = NULL;
michael@0 362 uint32_t savedTrieWord = 0;
michael@0 363
michael@0 364 /* initialize the hashtable */
michael@0 365 if(hashTable==NULL){
michael@0 366 hashTable = uhash_open(hashEntry, compareEntries, NULL, status);
michael@0 367 uhash_setValueDeleter(hashTable, valueDeleter);
michael@0 368 }
michael@0 369
michael@0 370 /* figure out if the code point has type already stored */
michael@0 371 savedTrieWord= utrie_get32(sprepTrie,codepoint,NULL);
michael@0 372 if(savedTrieWord!=0){
michael@0 373 if((savedTrieWord- _SPREP_TYPE_THRESHOLD) == USPREP_PROHIBITED){
michael@0 374 /* turn on the first bit in trie word */
michael@0 375 trieWord += 0x01;
michael@0 376 }else{
michael@0 377 /*
michael@0 378 * the codepoint has value something other than prohibited
michael@0 379 * and a mapping .. error!
michael@0 380 */
michael@0 381 fprintf(stderr,"Type for codepoint \\U%08X already set!.\n", (int)codepoint);
michael@0 382 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 383 }
michael@0 384 }
michael@0 385
michael@0 386 /* figure out the real length */
michael@0 387 for(i=0; i<length; i++){
michael@0 388 adjustedLen += U16_LENGTH(mapping[i]);
michael@0 389 }
michael@0 390
michael@0 391 if(adjustedLen == 0){
michael@0 392 trieWord = (uint16_t)(_SPREP_MAX_INDEX_VALUE << 2);
michael@0 393 /* make sure that the value of trieWord is less than the threshold */
michael@0 394 if(trieWord < _SPREP_TYPE_THRESHOLD){
michael@0 395 /* now set the value in the trie */
michael@0 396 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
michael@0 397 fprintf(stderr,"Could not set the value for code point.\n");
michael@0 398 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 399 }
michael@0 400 /* value is set so just return */
michael@0 401 return;
michael@0 402 }else{
michael@0 403 fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
michael@0 404 exit(U_ILLEGAL_CHAR_FOUND);
michael@0 405 }
michael@0 406 }
michael@0 407
michael@0 408 if(adjustedLen == 1){
michael@0 409 /* calculate the delta */
michael@0 410 int16_t delta = (int16_t)((int32_t)codepoint - (int16_t) mapping[0]);
michael@0 411 if(delta >= SPREP_DELTA_RANGE_NEGATIVE_LIMIT && delta <= SPREP_DELTA_RANGE_POSITIVE_LIMIT){
michael@0 412
michael@0 413 trieWord = delta << 2;
michael@0 414
michael@0 415
michael@0 416 /* make sure that the second bit is OFF */
michael@0 417 if((trieWord & 0x02) != 0 ){
michael@0 418 fprintf(stderr,"The second bit in the trie word is not zero while storing a delta.\n");
michael@0 419 exit(U_INTERNAL_PROGRAM_ERROR);
michael@0 420 }
michael@0 421 /* make sure that the value of trieWord is less than the threshold */
michael@0 422 if(trieWord < _SPREP_TYPE_THRESHOLD){
michael@0 423 /* now set the value in the trie */
michael@0 424 if(!utrie_set32(sprepTrie,codepoint,trieWord)){
michael@0 425 fprintf(stderr,"Could not set the value for code point.\n");
michael@0 426 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 427 }
michael@0 428 /* value is set so just return */
michael@0 429 return;
michael@0 430 }
michael@0 431 }
michael@0 432 /*
michael@0 433 * if the delta is not in the given range or if the trieWord is larger than the threshold
michael@0 434 * just fall through for storing the mapping in the mapping table
michael@0 435 */
michael@0 436 }
michael@0 437
michael@0 438 map = (UChar*) uprv_calloc(adjustedLen + 1, U_SIZEOF_UCHAR);
michael@0 439
michael@0 440 for (i=0, j=0; i<length; i++) {
michael@0 441 U16_APPEND_UNSAFE(map, j, mapping[i]);
michael@0 442 }
michael@0 443
michael@0 444 value = (ValueStruct*) uprv_malloc(sizeof(ValueStruct));
michael@0 445 value->mapping = map;
michael@0 446 value->type = type;
michael@0 447 value->length = adjustedLen;
michael@0 448 if(value->length > _SPREP_MAX_INDEX_TOP_LENGTH){
michael@0 449 mappingDataCapacity++;
michael@0 450 }
michael@0 451 if(maxLength < value->length){
michael@0 452 maxLength = value->length;
michael@0 453 }
michael@0 454 uhash_iput(hashTable,codepoint,value,status);
michael@0 455 mappingDataCapacity += adjustedLen;
michael@0 456
michael@0 457 if(U_FAILURE(*status)){
michael@0 458 fprintf(stderr, "Failed to put entries into the hastable. Error: %s\n", u_errorName(*status));
michael@0 459 exit(*status);
michael@0 460 }
michael@0 461 }
michael@0 462
michael@0 463
michael@0 464 extern void
michael@0 465 storeRange(uint32_t start, uint32_t end, UStringPrepType type,UErrorCode* status){
michael@0 466 uint16_t trieWord = 0;
michael@0 467
michael@0 468 if((int)(_SPREP_TYPE_THRESHOLD + type) > 0xFFFF){
michael@0 469 fprintf(stderr,"trieWord cannot contain value greater than 0xFFFF.\n");
michael@0 470 exit(U_ILLEGAL_CHAR_FOUND);
michael@0 471 }
michael@0 472 trieWord = (_SPREP_TYPE_THRESHOLD + type); /* the top 4 bits contain the value */
michael@0 473 if(start == end){
michael@0 474 uint32_t savedTrieWord = utrie_get32(sprepTrie, start, NULL);
michael@0 475 if(savedTrieWord>0){
michael@0 476 if(savedTrieWord < _SPREP_TYPE_THRESHOLD && type == USPREP_PROHIBITED){
michael@0 477 /*
michael@0 478 * A mapping is stored in the trie word
michael@0 479 * and the only other possible type that a
michael@0 480 * code point can have is USPREP_PROHIBITED
michael@0 481 *
michael@0 482 */
michael@0 483
michael@0 484 /* turn on the 0th bit in the savedTrieWord */
michael@0 485 savedTrieWord += 0x01;
michael@0 486
michael@0 487 /* the downcast is safe since we only save 16 bit values */
michael@0 488 trieWord = (uint16_t)savedTrieWord;
michael@0 489
michael@0 490 /* make sure that the value of trieWord is less than the threshold */
michael@0 491 if(trieWord < _SPREP_TYPE_THRESHOLD){
michael@0 492 /* now set the value in the trie */
michael@0 493 if(!utrie_set32(sprepTrie,start,trieWord)){
michael@0 494 fprintf(stderr,"Could not set the value for code point.\n");
michael@0 495 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 496 }
michael@0 497 /* value is set so just return */
michael@0 498 return;
michael@0 499 }else{
michael@0 500 fprintf(stderr,"trieWord cannot contain value greater than threshold 0x%04X.\n",_SPREP_TYPE_THRESHOLD);
michael@0 501 exit(U_ILLEGAL_CHAR_FOUND);
michael@0 502 }
michael@0 503
michael@0 504 }else if(savedTrieWord != trieWord){
michael@0 505 fprintf(stderr,"Value for codepoint \\U%08X already set!.\n", (int)start);
michael@0 506 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 507 }
michael@0 508 /* if savedTrieWord == trieWord .. fall through and set the value */
michael@0 509 }
michael@0 510 if(!utrie_set32(sprepTrie,start,trieWord)){
michael@0 511 fprintf(stderr,"Could not set the value for code point \\U%08X.\n", (int)start);
michael@0 512 exit(U_ILLEGAL_ARGUMENT_ERROR);
michael@0 513 }
michael@0 514 }else{
michael@0 515 if(!utrie_setRange32(sprepTrie, start, end+1, trieWord, FALSE)){
michael@0 516 fprintf(stderr,"Value for certain codepoint already set.\n");
michael@0 517 exit(U_ILLEGAL_CHAR_FOUND);
michael@0 518 }
michael@0 519 }
michael@0 520
michael@0 521 }
michael@0 522
michael@0 523 /* folding value: just store the offset (16 bits) if there is any non-0 entry */
michael@0 524 static uint32_t U_CALLCONV
michael@0 525 getFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) {
michael@0 526 uint32_t value;
michael@0 527 UChar32 limit=0;
michael@0 528 UBool inBlockZero;
michael@0 529
michael@0 530 limit=start+0x400;
michael@0 531 while(start<limit) {
michael@0 532 value=utrie_get32(trie, start, &inBlockZero);
michael@0 533 if(inBlockZero) {
michael@0 534 start+=UTRIE_DATA_BLOCK_LENGTH;
michael@0 535 } else if(value!=0) {
michael@0 536 return (uint32_t)offset;
michael@0 537 } else {
michael@0 538 ++start;
michael@0 539 }
michael@0 540 }
michael@0 541 return 0;
michael@0 542
michael@0 543 }
michael@0 544
michael@0 545 #endif /* #if !UCONFIG_NO_IDNA */
michael@0 546
michael@0 547 extern void
michael@0 548 generateData(const char *dataDir, const char* bundleName) {
michael@0 549 static uint8_t sprepTrieBlock[100000];
michael@0 550
michael@0 551 UNewDataMemory *pData;
michael@0 552 UErrorCode errorCode=U_ZERO_ERROR;
michael@0 553 int32_t size, dataLength;
michael@0 554 char* fileName = (char*) uprv_malloc(uprv_strlen(bundleName) +100);
michael@0 555
michael@0 556 #if UCONFIG_NO_IDNA
michael@0 557
michael@0 558 size=0;
michael@0 559
michael@0 560 #else
michael@0 561
michael@0 562 int32_t sprepTrieSize;
michael@0 563
michael@0 564 /* sort and add mapping data */
michael@0 565 storeMappingData();
michael@0 566
michael@0 567 sprepTrieSize=utrie_serialize(sprepTrie, sprepTrieBlock, sizeof(sprepTrieBlock), getFoldedValue, TRUE, &errorCode);
michael@0 568 if(U_FAILURE(errorCode)) {
michael@0 569 fprintf(stderr, "error: utrie_serialize(sprep trie) failed, %s\n", u_errorName(errorCode));
michael@0 570 exit(errorCode);
michael@0 571 }
michael@0 572
michael@0 573 size = sprepTrieSize + mappingDataCapacity*U_SIZEOF_UCHAR + sizeof(indexes);
michael@0 574 if(beVerbose) {
michael@0 575 printf("size of sprep trie %5u bytes\n", (int)sprepTrieSize);
michael@0 576 printf("size of " U_ICUDATA_NAME "_%s." DATA_TYPE " contents: %ld bytes\n", bundleName,(long)size);
michael@0 577 printf("size of mapping data array %5u bytes\n",(int)mappingDataCapacity * U_SIZEOF_UCHAR);
michael@0 578 printf("Number of code units in mappingData (currentIndex) are: %i \n", currentIndex);
michael@0 579 printf("Maximum length of the mapping string is : %i \n", (int)maxLength);
michael@0 580 }
michael@0 581
michael@0 582 #endif
michael@0 583
michael@0 584 fileName[0]=0;
michael@0 585 uprv_strcat(fileName,bundleName);
michael@0 586 /* write the data */
michael@0 587 pData=udata_create(dataDir, DATA_TYPE, fileName, &dataInfo,
michael@0 588 haveCopyright ? U_COPYRIGHT_STRING : NULL, &errorCode);
michael@0 589 if(U_FAILURE(errorCode)) {
michael@0 590 fprintf(stderr, "gensprep: unable to create the output file, error %d\n", errorCode);
michael@0 591 exit(errorCode);
michael@0 592 }
michael@0 593
michael@0 594 #if !UCONFIG_NO_IDNA
michael@0 595
michael@0 596 indexes[_SPREP_INDEX_TRIE_SIZE]=sprepTrieSize;
michael@0 597 indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]=mappingDataCapacity*U_SIZEOF_UCHAR;
michael@0 598
michael@0 599 udata_writeBlock(pData, indexes, sizeof(indexes));
michael@0 600 udata_writeBlock(pData, sprepTrieBlock, sprepTrieSize);
michael@0 601 udata_writeBlock(pData, mappingData, indexes[_SPREP_INDEX_MAPPING_DATA_SIZE]);
michael@0 602
michael@0 603
michael@0 604 #endif
michael@0 605
michael@0 606 /* finish up */
michael@0 607 dataLength=udata_finish(pData, &errorCode);
michael@0 608 if(U_FAILURE(errorCode)) {
michael@0 609 fprintf(stderr, "gensprep: error %d writing the output file\n", errorCode);
michael@0 610 exit(errorCode);
michael@0 611 }
michael@0 612
michael@0 613 if(dataLength!=size) {
michael@0 614 fprintf(stderr, "gensprep error: data length %ld != calculated size %ld\n",
michael@0 615 (long)dataLength, (long)size);
michael@0 616 exit(U_INTERNAL_PROGRAM_ERROR);
michael@0 617 }
michael@0 618
michael@0 619 #if !UCONFIG_NO_IDNA
michael@0 620 /* done with writing the data .. close the hashtable */
michael@0 621 if (hashTable != NULL) {
michael@0 622 uhash_close(hashTable);
michael@0 623 }
michael@0 624 #endif
michael@0 625
michael@0 626 uprv_free(fileName);
michael@0 627 }
michael@0 628
michael@0 629 #if !UCONFIG_NO_IDNA
michael@0 630
michael@0 631 extern void
michael@0 632 cleanUpData(void) {
michael@0 633 uprv_free(mappingData);
michael@0 634 utrie_close(sprepTrie);
michael@0 635 uprv_free(sprepTrie);
michael@0 636 }
michael@0 637
michael@0 638 #endif /* #if !UCONFIG_NO_IDNA */
michael@0 639
michael@0 640 /*
michael@0 641 * Hey, Emacs, please set the following:
michael@0 642 *
michael@0 643 * Local Variables:
michael@0 644 * indent-tabs-mode: nil
michael@0 645 * End:
michael@0 646 *
michael@0 647 */

mercurial