jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1 | /* |
| 2 | ****************************************************************************** |
| 3 | * |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 4 | * Copyright (C) 1999-2014, International Business Machines |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 5 | * Corporation and others. All Rights Reserved. |
| 6 | * |
| 7 | ****************************************************************************** |
| 8 | * file name: unames.c |
| 9 | * encoding: US-ASCII |
| 10 | * tab size: 8 (not used) |
| 11 | * indentation:4 |
| 12 | * |
| 13 | * created on: 1999oct04 |
| 14 | * created by: Markus W. Scherer |
| 15 | */ |
| 16 | |
| 17 | #include "unicode/utypes.h" |
| 18 | #include "unicode/putil.h" |
| 19 | #include "unicode/uchar.h" |
| 20 | #include "unicode/udata.h" |
| 21 | #include "unicode/utf.h" |
| 22 | #include "unicode/utf16.h" |
| 23 | #include "uassert.h" |
| 24 | #include "ustr_imp.h" |
| 25 | #include "umutex.h" |
| 26 | #include "cmemory.h" |
| 27 | #include "cstring.h" |
| 28 | #include "ucln_cmn.h" |
| 29 | #include "udataswp.h" |
| 30 | #include "uprops.h" |
| 31 | |
| 32 | U_NAMESPACE_BEGIN |
| 33 | |
| 34 | /* prototypes ------------------------------------------------------------- */ |
| 35 | |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 36 | static const char DATA_NAME[] = "unames"; |
| 37 | static const char DATA_TYPE[] = "icu"; |
| 38 | |
| 39 | #define GROUP_SHIFT 5 |
| 40 | #define LINES_PER_GROUP (1L<<GROUP_SHIFT) |
| 41 | #define GROUP_MASK (LINES_PER_GROUP-1) |
| 42 | |
| 43 | /* |
| 44 | * This struct was replaced by explicitly accessing equivalent |
| 45 | * fields from triples of uint16_t. |
| 46 | * The Group struct was padded to 8 bytes on compilers for early ARM CPUs, |
| 47 | * which broke the assumption that sizeof(Group)==6 and that the ++ operator |
| 48 | * would advance by 6 bytes (3 uint16_t). |
| 49 | * |
| 50 | * We can't just change the data structure because it's loaded from a data file, |
| 51 | * and we don't want to make it less compact, so we changed the access code. |
| 52 | * |
| 53 | * For details see ICU tickets 6331 and 6008. |
| 54 | typedef struct { |
| 55 | uint16_t groupMSB, |
| 56 | offsetHigh, offsetLow; / * avoid padding * / |
| 57 | } Group; |
| 58 | */ |
| 59 | enum { |
| 60 | GROUP_MSB, |
| 61 | GROUP_OFFSET_HIGH, |
| 62 | GROUP_OFFSET_LOW, |
| 63 | GROUP_LENGTH |
| 64 | }; |
| 65 | |
| 66 | /* |
| 67 | * Get the 32-bit group offset. |
| 68 | * @param group (const uint16_t *) pointer to a Group triple of uint16_t |
| 69 | * @return group offset (int32_t) |
| 70 | */ |
| 71 | #define GET_GROUP_OFFSET(group) ((int32_t)(group)[GROUP_OFFSET_HIGH]<<16|(group)[GROUP_OFFSET_LOW]) |
| 72 | |
| 73 | #define NEXT_GROUP(group) ((group)+GROUP_LENGTH) |
| 74 | #define PREV_GROUP(group) ((group)-GROUP_LENGTH) |
| 75 | |
| 76 | typedef struct { |
| 77 | uint32_t start, end; |
| 78 | uint8_t type, variant; |
| 79 | uint16_t size; |
| 80 | } AlgorithmicRange; |
| 81 | |
| 82 | typedef struct { |
| 83 | uint32_t tokenStringOffset, groupsOffset, groupStringOffset, algNamesOffset; |
| 84 | } UCharNames; |
| 85 | |
| 86 | /* |
| 87 | * Get the groups table from a UCharNames struct. |
| 88 | * The groups table consists of one uint16_t groupCount followed by |
| 89 | * groupCount groups. Each group is a triple of uint16_t, see GROUP_LENGTH |
| 90 | * and the comment for the old struct Group above. |
| 91 | * |
| 92 | * @param names (const UCharNames *) pointer to the UCharNames indexes |
| 93 | * @return (const uint16_t *) pointer to the groups table |
| 94 | */ |
| 95 | #define GET_GROUPS(names) (const uint16_t *)((const char *)names+names->groupsOffset) |
| 96 | |
| 97 | typedef struct { |
| 98 | const char *otherName; |
| 99 | UChar32 code; |
| 100 | } FindName; |
| 101 | |
| 102 | #define DO_FIND_NAME NULL |
| 103 | |
| 104 | static UDataMemory *uCharNamesData=NULL; |
| 105 | static UCharNames *uCharNames=NULL; |
| 106 | static icu::UInitOnce gCharNamesInitOnce = U_INITONCE_INITIALIZER; |
| 107 | |
| 108 | /* |
| 109 | * Maximum length of character names (regular & 1.0). |
| 110 | */ |
| 111 | static int32_t gMaxNameLength=0; |
| 112 | |
| 113 | /* |
| 114 | * Set of chars used in character names (regular & 1.0). |
| 115 | * Chars are platform-dependent (can be EBCDIC). |
| 116 | */ |
| 117 | static uint32_t gNameSet[8]={ 0 }; |
| 118 | |
| 119 | #define U_NONCHARACTER_CODE_POINT U_CHAR_CATEGORY_COUNT |
| 120 | #define U_LEAD_SURROGATE U_CHAR_CATEGORY_COUNT + 1 |
| 121 | #define U_TRAIL_SURROGATE U_CHAR_CATEGORY_COUNT + 2 |
| 122 | |
| 123 | #define U_CHAR_EXTENDED_CATEGORY_COUNT (U_CHAR_CATEGORY_COUNT + 3) |
| 124 | |
| 125 | static const char * const charCatNames[U_CHAR_EXTENDED_CATEGORY_COUNT] = { |
| 126 | "unassigned", |
| 127 | "uppercase letter", |
| 128 | "lowercase letter", |
| 129 | "titlecase letter", |
| 130 | "modifier letter", |
| 131 | "other letter", |
| 132 | "non spacing mark", |
| 133 | "enclosing mark", |
| 134 | "combining spacing mark", |
| 135 | "decimal digit number", |
| 136 | "letter number", |
| 137 | "other number", |
| 138 | "space separator", |
| 139 | "line separator", |
| 140 | "paragraph separator", |
| 141 | "control", |
| 142 | "format", |
| 143 | "private use area", |
| 144 | "surrogate", |
| 145 | "dash punctuation", |
| 146 | "start punctuation", |
| 147 | "end punctuation", |
| 148 | "connector punctuation", |
| 149 | "other punctuation", |
| 150 | "math symbol", |
| 151 | "currency symbol", |
| 152 | "modifier symbol", |
| 153 | "other symbol", |
| 154 | "initial punctuation", |
| 155 | "final punctuation", |
| 156 | "noncharacter", |
| 157 | "lead surrogate", |
| 158 | "trail surrogate" |
| 159 | }; |
| 160 | |
| 161 | /* implementation ----------------------------------------------------------- */ |
| 162 | |
| 163 | static UBool U_CALLCONV unames_cleanup(void) |
| 164 | { |
| 165 | if(uCharNamesData) { |
| 166 | udata_close(uCharNamesData); |
| 167 | uCharNamesData = NULL; |
| 168 | } |
| 169 | if(uCharNames) { |
| 170 | uCharNames = NULL; |
| 171 | } |
| 172 | gCharNamesInitOnce.reset(); |
| 173 | gMaxNameLength=0; |
| 174 | return TRUE; |
| 175 | } |
| 176 | |
| 177 | static UBool U_CALLCONV |
| 178 | isAcceptable(void * /*context*/, |
| 179 | const char * /*type*/, const char * /*name*/, |
| 180 | const UDataInfo *pInfo) { |
| 181 | return (UBool)( |
| 182 | pInfo->size>=20 && |
| 183 | pInfo->isBigEndian==U_IS_BIG_ENDIAN && |
| 184 | pInfo->charsetFamily==U_CHARSET_FAMILY && |
| 185 | pInfo->dataFormat[0]==0x75 && /* dataFormat="unam" */ |
| 186 | pInfo->dataFormat[1]==0x6e && |
| 187 | pInfo->dataFormat[2]==0x61 && |
| 188 | pInfo->dataFormat[3]==0x6d && |
| 189 | pInfo->formatVersion[0]==1); |
| 190 | } |
| 191 | |
| 192 | static void U_CALLCONV |
| 193 | loadCharNames(UErrorCode &status) { |
| 194 | U_ASSERT(uCharNamesData == NULL); |
| 195 | U_ASSERT(uCharNames == NULL); |
| 196 | |
| 197 | uCharNamesData = udata_openChoice(NULL, DATA_TYPE, DATA_NAME, isAcceptable, NULL, &status); |
| 198 | if(U_FAILURE(status)) { |
| 199 | uCharNamesData = NULL; |
| 200 | } else { |
| 201 | uCharNames = (UCharNames *)udata_getMemory(uCharNamesData); |
| 202 | } |
| 203 | ucln_common_registerCleanup(UCLN_COMMON_UNAMES, unames_cleanup); |
| 204 | } |
| 205 | |
| 206 | |
| 207 | static UBool |
| 208 | isDataLoaded(UErrorCode *pErrorCode) { |
| 209 | umtx_initOnce(gCharNamesInitOnce, &loadCharNames, *pErrorCode); |
| 210 | return U_SUCCESS(*pErrorCode); |
| 211 | } |
| 212 | |
| 213 | #define WRITE_CHAR(buffer, bufferLength, bufferPos, c) { \ |
| 214 | if((bufferLength)>0) { \ |
| 215 | *(buffer)++=c; \ |
| 216 | --(bufferLength); \ |
| 217 | } \ |
| 218 | ++(bufferPos); \ |
| 219 | } |
| 220 | |
| 221 | #define U_ISO_COMMENT U_CHAR_NAME_CHOICE_COUNT |
| 222 | |
| 223 | /* |
| 224 | * Important: expandName() and compareName() are almost the same - |
| 225 | * apply fixes to both. |
| 226 | * |
| 227 | * UnicodeData.txt uses ';' as a field separator, so no |
| 228 | * field can contain ';' as part of its contents. |
| 229 | * In unames.dat, it is marked as token[';']==-1 only if the |
| 230 | * semicolon is used in the data file - which is iff we |
| 231 | * have Unicode 1.0 names or ISO comments or aliases. |
| 232 | * So, it will be token[';']==-1 if we store U1.0 names/ISO comments/aliases |
| 233 | * although we know that it will never be part of a name. |
| 234 | */ |
| 235 | static uint16_t |
| 236 | expandName(UCharNames *names, |
| 237 | const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice, |
| 238 | char *buffer, uint16_t bufferLength) { |
| 239 | uint16_t *tokens=(uint16_t *)names+8; |
| 240 | uint16_t token, tokenCount=*tokens++, bufferPos=0; |
| 241 | uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset; |
| 242 | uint8_t c; |
| 243 | |
| 244 | if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { |
| 245 | /* |
| 246 | * skip the modern name if it is not requested _and_ |
| 247 | * if the semicolon byte value is a character, not a token number |
| 248 | */ |
| 249 | if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { |
| 250 | int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice; |
| 251 | do { |
| 252 | while(nameLength>0) { |
| 253 | --nameLength; |
| 254 | if(*name++==';') { |
| 255 | break; |
| 256 | } |
| 257 | } |
| 258 | } while(--fieldIndex>0); |
| 259 | } else { |
| 260 | /* |
| 261 | * the semicolon byte value is a token number, therefore |
| 262 | * only modern names are stored in unames.dat and there is no |
| 263 | * such requested alternate name here |
| 264 | */ |
| 265 | nameLength=0; |
| 266 | } |
| 267 | } |
| 268 | |
| 269 | /* write each letter directly, and write a token word per token */ |
| 270 | while(nameLength>0) { |
| 271 | --nameLength; |
| 272 | c=*name++; |
| 273 | |
| 274 | if(c>=tokenCount) { |
| 275 | if(c!=';') { |
| 276 | /* implicit letter */ |
| 277 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 278 | } else { |
| 279 | /* finished */ |
| 280 | break; |
| 281 | } |
| 282 | } else { |
| 283 | token=tokens[c]; |
| 284 | if(token==(uint16_t)(-2)) { |
| 285 | /* this is a lead byte for a double-byte token */ |
| 286 | token=tokens[c<<8|*name++]; |
| 287 | --nameLength; |
| 288 | } |
| 289 | if(token==(uint16_t)(-1)) { |
| 290 | if(c!=';') { |
| 291 | /* explicit letter */ |
| 292 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 293 | } else { |
| 294 | /* stop, but skip the semicolon if we are seeking |
| 295 | extended names and there was no 2.0 name but there |
| 296 | is a 1.0 name. */ |
| 297 | if(!bufferPos && nameChoice == U_EXTENDED_CHAR_NAME) { |
| 298 | if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { |
| 299 | continue; |
| 300 | } |
| 301 | } |
| 302 | /* finished */ |
| 303 | break; |
| 304 | } |
| 305 | } else { |
| 306 | /* write token word */ |
| 307 | uint8_t *tokenString=tokenStrings+token; |
| 308 | while((c=*tokenString++)!=0) { |
| 309 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 310 | } |
| 311 | } |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | /* zero-terminate */ |
| 316 | if(bufferLength>0) { |
| 317 | *buffer=0; |
| 318 | } |
| 319 | |
| 320 | return bufferPos; |
| 321 | } |
| 322 | |
| 323 | /* |
| 324 | * compareName() is almost the same as expandName() except that it compares |
| 325 | * the currently expanded name to an input name. |
| 326 | * It returns the match/no match result as soon as possible. |
| 327 | */ |
| 328 | static UBool |
| 329 | compareName(UCharNames *names, |
| 330 | const uint8_t *name, uint16_t nameLength, UCharNameChoice nameChoice, |
| 331 | const char *otherName) { |
| 332 | uint16_t *tokens=(uint16_t *)names+8; |
| 333 | uint16_t token, tokenCount=*tokens++; |
| 334 | uint8_t *tokenStrings=(uint8_t *)names+names->tokenStringOffset; |
| 335 | uint8_t c; |
| 336 | const char *origOtherName = otherName; |
| 337 | |
| 338 | if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { |
| 339 | /* |
| 340 | * skip the modern name if it is not requested _and_ |
| 341 | * if the semicolon byte value is a character, not a token number |
| 342 | */ |
| 343 | if((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { |
| 344 | int fieldIndex= nameChoice==U_ISO_COMMENT ? 2 : nameChoice; |
| 345 | do { |
| 346 | while(nameLength>0) { |
| 347 | --nameLength; |
| 348 | if(*name++==';') { |
| 349 | break; |
| 350 | } |
| 351 | } |
| 352 | } while(--fieldIndex>0); |
| 353 | } else { |
| 354 | /* |
| 355 | * the semicolon byte value is a token number, therefore |
| 356 | * only modern names are stored in unames.dat and there is no |
| 357 | * such requested alternate name here |
| 358 | */ |
| 359 | nameLength=0; |
| 360 | } |
| 361 | } |
| 362 | |
| 363 | /* compare each letter directly, and compare a token word per token */ |
| 364 | while(nameLength>0) { |
| 365 | --nameLength; |
| 366 | c=*name++; |
| 367 | |
| 368 | if(c>=tokenCount) { |
| 369 | if(c!=';') { |
| 370 | /* implicit letter */ |
| 371 | if((char)c!=*otherName++) { |
| 372 | return FALSE; |
| 373 | } |
| 374 | } else { |
| 375 | /* finished */ |
| 376 | break; |
| 377 | } |
| 378 | } else { |
| 379 | token=tokens[c]; |
| 380 | if(token==(uint16_t)(-2)) { |
| 381 | /* this is a lead byte for a double-byte token */ |
| 382 | token=tokens[c<<8|*name++]; |
| 383 | --nameLength; |
| 384 | } |
| 385 | if(token==(uint16_t)(-1)) { |
| 386 | if(c!=';') { |
| 387 | /* explicit letter */ |
| 388 | if((char)c!=*otherName++) { |
| 389 | return FALSE; |
| 390 | } |
| 391 | } else { |
| 392 | /* stop, but skip the semicolon if we are seeking |
| 393 | extended names and there was no 2.0 name but there |
| 394 | is a 1.0 name. */ |
| 395 | if(otherName == origOtherName && nameChoice == U_EXTENDED_CHAR_NAME) { |
| 396 | if ((uint8_t)';'>=tokenCount || tokens[(uint8_t)';']==(uint16_t)(-1)) { |
| 397 | continue; |
| 398 | } |
| 399 | } |
| 400 | /* finished */ |
| 401 | break; |
| 402 | } |
| 403 | } else { |
| 404 | /* write token word */ |
| 405 | uint8_t *tokenString=tokenStrings+token; |
| 406 | while((c=*tokenString++)!=0) { |
| 407 | if((char)c!=*otherName++) { |
| 408 | return FALSE; |
| 409 | } |
| 410 | } |
| 411 | } |
| 412 | } |
| 413 | } |
| 414 | |
| 415 | /* complete match? */ |
| 416 | return (UBool)(*otherName==0); |
| 417 | } |
| 418 | |
| 419 | static uint8_t getCharCat(UChar32 cp) { |
| 420 | uint8_t cat; |
| 421 | |
| 422 | if (U_IS_UNICODE_NONCHAR(cp)) { |
| 423 | return U_NONCHARACTER_CODE_POINT; |
| 424 | } |
| 425 | |
| 426 | if ((cat = u_charType(cp)) == U_SURROGATE) { |
| 427 | cat = U_IS_LEAD(cp) ? U_LEAD_SURROGATE : U_TRAIL_SURROGATE; |
| 428 | } |
| 429 | |
| 430 | return cat; |
| 431 | } |
| 432 | |
| 433 | static const char *getCharCatName(UChar32 cp) { |
| 434 | uint8_t cat = getCharCat(cp); |
| 435 | |
| 436 | /* Return unknown if the table of names above is not up to |
| 437 | date. */ |
| 438 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 439 | if (cat >= UPRV_LENGTHOF(charCatNames)) { |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 440 | return "unknown"; |
| 441 | } else { |
| 442 | return charCatNames[cat]; |
| 443 | } |
| 444 | } |
| 445 | |
| 446 | static uint16_t getExtName(uint32_t code, char *buffer, uint16_t bufferLength) { |
| 447 | const char *catname = getCharCatName(code); |
| 448 | uint16_t length = 0; |
| 449 | |
| 450 | UChar32 cp; |
| 451 | int ndigits, i; |
| 452 | |
| 453 | WRITE_CHAR(buffer, bufferLength, length, '<'); |
| 454 | while (catname[length - 1]) { |
| 455 | WRITE_CHAR(buffer, bufferLength, length, catname[length - 1]); |
| 456 | } |
| 457 | WRITE_CHAR(buffer, bufferLength, length, '-'); |
| 458 | for (cp = code, ndigits = 0; cp; ++ndigits, cp >>= 4) |
| 459 | ; |
| 460 | if (ndigits < 4) |
| 461 | ndigits = 4; |
| 462 | for (cp = code, i = ndigits; (cp || i > 0) && bufferLength; cp >>= 4, bufferLength--) { |
| 463 | uint8_t v = (uint8_t)(cp & 0xf); |
| 464 | buffer[--i] = (v < 10 ? '0' + v : 'A' + v - 10); |
| 465 | } |
| 466 | buffer += ndigits; |
| 467 | length += ndigits; |
| 468 | WRITE_CHAR(buffer, bufferLength, length, '>'); |
| 469 | |
| 470 | return length; |
| 471 | } |
| 472 | |
| 473 | /* |
| 474 | * getGroup() does a binary search for the group that contains the |
| 475 | * Unicode code point "code". |
| 476 | * The return value is always a valid Group* that may contain "code" |
| 477 | * or else is the highest group before "code". |
| 478 | * If the lowest group is after "code", then that one is returned. |
| 479 | */ |
| 480 | static const uint16_t * |
| 481 | getGroup(UCharNames *names, uint32_t code) { |
| 482 | const uint16_t *groups=GET_GROUPS(names); |
| 483 | uint16_t groupMSB=(uint16_t)(code>>GROUP_SHIFT), |
| 484 | start=0, |
| 485 | limit=*groups++, |
| 486 | number; |
| 487 | |
| 488 | /* binary search for the group of names that contains the one for code */ |
| 489 | while(start<limit-1) { |
| 490 | number=(uint16_t)((start+limit)/2); |
| 491 | if(groupMSB<groups[number*GROUP_LENGTH+GROUP_MSB]) { |
| 492 | limit=number; |
| 493 | } else { |
| 494 | start=number; |
| 495 | } |
| 496 | } |
| 497 | |
| 498 | /* return this regardless of whether it is an exact match */ |
| 499 | return groups+start*GROUP_LENGTH; |
| 500 | } |
| 501 | |
| 502 | /* |
| 503 | * expandGroupLengths() reads a block of compressed lengths of 32 strings and |
| 504 | * expands them into offsets and lengths for each string. |
| 505 | * Lengths are stored with a variable-width encoding in consecutive nibbles: |
| 506 | * If a nibble<0xc, then it is the length itself (0=empty string). |
| 507 | * If a nibble>=0xc, then it forms a length value with the following nibble. |
| 508 | * Calculation see below. |
| 509 | * The offsets and lengths arrays must be at least 33 (one more) long because |
| 510 | * there is no check here at the end if the last nibble is still used. |
| 511 | */ |
| 512 | static const uint8_t * |
| 513 | expandGroupLengths(const uint8_t *s, |
| 514 | uint16_t offsets[LINES_PER_GROUP+1], uint16_t lengths[LINES_PER_GROUP+1]) { |
| 515 | /* read the lengths of the 32 strings in this group and get each string's offset */ |
| 516 | uint16_t i=0, offset=0, length=0; |
| 517 | uint8_t lengthByte; |
| 518 | |
| 519 | /* all 32 lengths must be read to get the offset of the first group string */ |
| 520 | while(i<LINES_PER_GROUP) { |
| 521 | lengthByte=*s++; |
| 522 | |
| 523 | /* read even nibble - MSBs of lengthByte */ |
| 524 | if(length>=12) { |
| 525 | /* double-nibble length spread across two bytes */ |
| 526 | length=(uint16_t)(((length&0x3)<<4|lengthByte>>4)+12); |
| 527 | lengthByte&=0xf; |
| 528 | } else if((lengthByte /* &0xf0 */)>=0xc0) { |
| 529 | /* double-nibble length spread across this one byte */ |
| 530 | length=(uint16_t)((lengthByte&0x3f)+12); |
| 531 | } else { |
| 532 | /* single-nibble length in MSBs */ |
| 533 | length=(uint16_t)(lengthByte>>4); |
| 534 | lengthByte&=0xf; |
| 535 | } |
| 536 | |
| 537 | *offsets++=offset; |
| 538 | *lengths++=length; |
| 539 | |
| 540 | offset+=length; |
| 541 | ++i; |
| 542 | |
| 543 | /* read odd nibble - LSBs of lengthByte */ |
| 544 | if((lengthByte&0xf0)==0) { |
| 545 | /* this nibble was not consumed for a double-nibble length above */ |
| 546 | length=lengthByte; |
| 547 | if(length<12) { |
| 548 | /* single-nibble length in LSBs */ |
| 549 | *offsets++=offset; |
| 550 | *lengths++=length; |
| 551 | |
| 552 | offset+=length; |
| 553 | ++i; |
| 554 | } |
| 555 | } else { |
| 556 | length=0; /* prevent double-nibble detection in the next iteration */ |
| 557 | } |
| 558 | } |
| 559 | |
| 560 | /* now, s is at the first group string */ |
| 561 | return s; |
| 562 | } |
| 563 | |
| 564 | static uint16_t |
| 565 | expandGroupName(UCharNames *names, const uint16_t *group, |
| 566 | uint16_t lineNumber, UCharNameChoice nameChoice, |
| 567 | char *buffer, uint16_t bufferLength) { |
| 568 | uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; |
| 569 | const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group); |
| 570 | s=expandGroupLengths(s, offsets, lengths); |
| 571 | return expandName(names, s+offsets[lineNumber], lengths[lineNumber], nameChoice, |
| 572 | buffer, bufferLength); |
| 573 | } |
| 574 | |
| 575 | static uint16_t |
| 576 | getName(UCharNames *names, uint32_t code, UCharNameChoice nameChoice, |
| 577 | char *buffer, uint16_t bufferLength) { |
| 578 | const uint16_t *group=getGroup(names, code); |
| 579 | if((uint16_t)(code>>GROUP_SHIFT)==group[GROUP_MSB]) { |
| 580 | return expandGroupName(names, group, (uint16_t)(code&GROUP_MASK), nameChoice, |
| 581 | buffer, bufferLength); |
| 582 | } else { |
| 583 | /* group not found */ |
| 584 | /* zero-terminate */ |
| 585 | if(bufferLength>0) { |
| 586 | *buffer=0; |
| 587 | } |
| 588 | return 0; |
| 589 | } |
| 590 | } |
| 591 | |
| 592 | /* |
| 593 | * enumGroupNames() enumerates all the names in a 32-group |
| 594 | * and either calls the enumerator function or finds a given input name. |
| 595 | */ |
| 596 | static UBool |
| 597 | enumGroupNames(UCharNames *names, const uint16_t *group, |
| 598 | UChar32 start, UChar32 end, |
| 599 | UEnumCharNamesFn *fn, void *context, |
| 600 | UCharNameChoice nameChoice) { |
| 601 | uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; |
| 602 | const uint8_t *s=(uint8_t *)names+names->groupStringOffset+GET_GROUP_OFFSET(group); |
| 603 | |
| 604 | s=expandGroupLengths(s, offsets, lengths); |
| 605 | if(fn!=DO_FIND_NAME) { |
| 606 | char buffer[200]; |
| 607 | uint16_t length; |
| 608 | |
| 609 | while(start<=end) { |
| 610 | length=expandName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, buffer, sizeof(buffer)); |
| 611 | if (!length && nameChoice == U_EXTENDED_CHAR_NAME) { |
| 612 | buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0; |
| 613 | } |
| 614 | /* here, we assume that the buffer is large enough */ |
| 615 | if(length>0) { |
| 616 | if(!fn(context, start, nameChoice, buffer, length)) { |
| 617 | return FALSE; |
| 618 | } |
| 619 | } |
| 620 | ++start; |
| 621 | } |
| 622 | } else { |
| 623 | const char *otherName=((FindName *)context)->otherName; |
| 624 | while(start<=end) { |
| 625 | if(compareName(names, s+offsets[start&GROUP_MASK], lengths[start&GROUP_MASK], nameChoice, otherName)) { |
| 626 | ((FindName *)context)->code=start; |
| 627 | return FALSE; |
| 628 | } |
| 629 | ++start; |
| 630 | } |
| 631 | } |
| 632 | return TRUE; |
| 633 | } |
| 634 | |
| 635 | /* |
| 636 | * enumExtNames enumerate extended names. |
| 637 | * It only needs to do it if it is called with a real function and not |
| 638 | * with the dummy DO_FIND_NAME, because u_charFromName() does a check |
| 639 | * for extended names by itself. |
| 640 | */ |
| 641 | static UBool |
| 642 | enumExtNames(UChar32 start, UChar32 end, |
| 643 | UEnumCharNamesFn *fn, void *context) |
| 644 | { |
| 645 | if(fn!=DO_FIND_NAME) { |
| 646 | char buffer[200]; |
| 647 | uint16_t length; |
| 648 | |
| 649 | while(start<=end) { |
| 650 | buffer[length = getExtName(start, buffer, sizeof(buffer))] = 0; |
| 651 | /* here, we assume that the buffer is large enough */ |
| 652 | if(length>0) { |
| 653 | if(!fn(context, start, U_EXTENDED_CHAR_NAME, buffer, length)) { |
| 654 | return FALSE; |
| 655 | } |
| 656 | } |
| 657 | ++start; |
| 658 | } |
| 659 | } |
| 660 | |
| 661 | return TRUE; |
| 662 | } |
| 663 | |
| 664 | static UBool |
| 665 | enumNames(UCharNames *names, |
| 666 | UChar32 start, UChar32 limit, |
| 667 | UEnumCharNamesFn *fn, void *context, |
| 668 | UCharNameChoice nameChoice) { |
| 669 | uint16_t startGroupMSB, endGroupMSB, groupCount; |
| 670 | const uint16_t *group, *groupLimit; |
| 671 | |
| 672 | startGroupMSB=(uint16_t)(start>>GROUP_SHIFT); |
| 673 | endGroupMSB=(uint16_t)((limit-1)>>GROUP_SHIFT); |
| 674 | |
| 675 | /* find the group that contains start, or the highest before it */ |
| 676 | group=getGroup(names, start); |
| 677 | |
| 678 | if(startGroupMSB<group[GROUP_MSB] && nameChoice==U_EXTENDED_CHAR_NAME) { |
| 679 | /* enumerate synthetic names between start and the group start */ |
| 680 | UChar32 extLimit=((UChar32)group[GROUP_MSB]<<GROUP_SHIFT); |
| 681 | if(extLimit>limit) { |
| 682 | extLimit=limit; |
| 683 | } |
| 684 | if(!enumExtNames(start, extLimit-1, fn, context)) { |
| 685 | return FALSE; |
| 686 | } |
| 687 | start=extLimit; |
| 688 | } |
| 689 | |
| 690 | if(startGroupMSB==endGroupMSB) { |
| 691 | if(startGroupMSB==group[GROUP_MSB]) { |
| 692 | /* if start and limit-1 are in the same group, then enumerate only in that one */ |
| 693 | return enumGroupNames(names, group, start, limit-1, fn, context, nameChoice); |
| 694 | } |
| 695 | } else { |
| 696 | const uint16_t *groups=GET_GROUPS(names); |
| 697 | groupCount=*groups++; |
| 698 | groupLimit=groups+groupCount*GROUP_LENGTH; |
| 699 | |
| 700 | if(startGroupMSB==group[GROUP_MSB]) { |
| 701 | /* enumerate characters in the partial start group */ |
| 702 | if((start&GROUP_MASK)!=0) { |
| 703 | if(!enumGroupNames(names, group, |
| 704 | start, ((UChar32)startGroupMSB<<GROUP_SHIFT)+LINES_PER_GROUP-1, |
| 705 | fn, context, nameChoice)) { |
| 706 | return FALSE; |
| 707 | } |
| 708 | group=NEXT_GROUP(group); /* continue with the next group */ |
| 709 | } |
| 710 | } else if(startGroupMSB>group[GROUP_MSB]) { |
| 711 | /* make sure that we start enumerating with the first group after start */ |
| 712 | const uint16_t *nextGroup=NEXT_GROUP(group); |
| 713 | if (nextGroup < groupLimit && nextGroup[GROUP_MSB] > startGroupMSB && nameChoice == U_EXTENDED_CHAR_NAME) { |
| 714 | UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT; |
| 715 | if (end > limit) { |
| 716 | end = limit; |
| 717 | } |
| 718 | if (!enumExtNames(start, end - 1, fn, context)) { |
| 719 | return FALSE; |
| 720 | } |
| 721 | } |
| 722 | group=nextGroup; |
| 723 | } |
| 724 | |
| 725 | /* enumerate entire groups between the start- and end-groups */ |
| 726 | while(group<groupLimit && group[GROUP_MSB]<endGroupMSB) { |
| 727 | const uint16_t *nextGroup; |
| 728 | start=(UChar32)group[GROUP_MSB]<<GROUP_SHIFT; |
| 729 | if(!enumGroupNames(names, group, start, start+LINES_PER_GROUP-1, fn, context, nameChoice)) { |
| 730 | return FALSE; |
| 731 | } |
| 732 | nextGroup=NEXT_GROUP(group); |
| 733 | if (nextGroup < groupLimit && nextGroup[GROUP_MSB] > group[GROUP_MSB] + 1 && nameChoice == U_EXTENDED_CHAR_NAME) { |
| 734 | UChar32 end = nextGroup[GROUP_MSB] << GROUP_SHIFT; |
| 735 | if (end > limit) { |
| 736 | end = limit; |
| 737 | } |
| 738 | if (!enumExtNames((group[GROUP_MSB] + 1) << GROUP_SHIFT, end - 1, fn, context)) { |
| 739 | return FALSE; |
| 740 | } |
| 741 | } |
| 742 | group=nextGroup; |
| 743 | } |
| 744 | |
| 745 | /* enumerate within the end group (group[GROUP_MSB]==endGroupMSB) */ |
| 746 | if(group<groupLimit && group[GROUP_MSB]==endGroupMSB) { |
| 747 | return enumGroupNames(names, group, (limit-1)&~GROUP_MASK, limit-1, fn, context, nameChoice); |
| 748 | } else if (nameChoice == U_EXTENDED_CHAR_NAME && group == groupLimit) { |
| 749 | UChar32 next = (PREV_GROUP(group)[GROUP_MSB] + 1) << GROUP_SHIFT; |
| 750 | if (next > start) { |
| 751 | start = next; |
| 752 | } |
| 753 | } else { |
| 754 | return TRUE; |
| 755 | } |
| 756 | } |
| 757 | |
| 758 | /* we have not found a group, which means everything is made of |
| 759 | extended names. */ |
| 760 | if (nameChoice == U_EXTENDED_CHAR_NAME) { |
| 761 | if (limit > UCHAR_MAX_VALUE + 1) { |
| 762 | limit = UCHAR_MAX_VALUE + 1; |
| 763 | } |
| 764 | return enumExtNames(start, limit - 1, fn, context); |
| 765 | } |
| 766 | |
| 767 | return TRUE; |
| 768 | } |
| 769 | |
| 770 | static uint16_t |
| 771 | writeFactorSuffix(const uint16_t *factors, uint16_t count, |
| 772 | const char *s, /* suffix elements */ |
| 773 | uint32_t code, |
| 774 | uint16_t indexes[8], /* output fields from here */ |
| 775 | const char *elementBases[8], const char *elements[8], |
| 776 | char *buffer, uint16_t bufferLength) { |
| 777 | uint16_t i, factor, bufferPos=0; |
| 778 | char c; |
| 779 | |
| 780 | /* write elements according to the factors */ |
| 781 | |
| 782 | /* |
| 783 | * the factorized elements are determined by modulo arithmetic |
| 784 | * with the factors of this algorithm |
| 785 | * |
| 786 | * note that for fewer operations, count is decremented here |
| 787 | */ |
| 788 | --count; |
| 789 | for(i=count; i>0; --i) { |
| 790 | factor=factors[i]; |
| 791 | indexes[i]=(uint16_t)(code%factor); |
| 792 | code/=factor; |
| 793 | } |
| 794 | /* |
| 795 | * we don't need to calculate the last modulus because start<=code<=end |
| 796 | * guarantees here that code<=factors[0] |
| 797 | */ |
| 798 | indexes[0]=(uint16_t)code; |
| 799 | |
| 800 | /* write each element */ |
| 801 | for(;;) { |
| 802 | if(elementBases!=NULL) { |
| 803 | *elementBases++=s; |
| 804 | } |
| 805 | |
| 806 | /* skip indexes[i] strings */ |
| 807 | factor=indexes[i]; |
| 808 | while(factor>0) { |
| 809 | while(*s++!=0) {} |
| 810 | --factor; |
| 811 | } |
| 812 | if(elements!=NULL) { |
| 813 | *elements++=s; |
| 814 | } |
| 815 | |
| 816 | /* write element */ |
| 817 | while((c=*s++)!=0) { |
| 818 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 819 | } |
| 820 | |
| 821 | /* we do not need to perform the rest of this loop for i==count - break here */ |
| 822 | if(i>=count) { |
| 823 | break; |
| 824 | } |
| 825 | |
| 826 | /* skip the rest of the strings for this factors[i] */ |
| 827 | factor=(uint16_t)(factors[i]-indexes[i]-1); |
| 828 | while(factor>0) { |
| 829 | while(*s++!=0) {} |
| 830 | --factor; |
| 831 | } |
| 832 | |
| 833 | ++i; |
| 834 | } |
| 835 | |
| 836 | /* zero-terminate */ |
| 837 | if(bufferLength>0) { |
| 838 | *buffer=0; |
| 839 | } |
| 840 | |
| 841 | return bufferPos; |
| 842 | } |
| 843 | |
| 844 | /* |
| 845 | * Important: |
| 846 | * Parts of findAlgName() are almost the same as some of getAlgName(). |
| 847 | * Fixes must be applied to both. |
| 848 | */ |
| 849 | static uint16_t |
| 850 | getAlgName(AlgorithmicRange *range, uint32_t code, UCharNameChoice nameChoice, |
| 851 | char *buffer, uint16_t bufferLength) { |
| 852 | uint16_t bufferPos=0; |
| 853 | |
| 854 | /* Only the normative character name can be algorithmic. */ |
| 855 | if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { |
| 856 | /* zero-terminate */ |
| 857 | if(bufferLength>0) { |
| 858 | *buffer=0; |
| 859 | } |
| 860 | return 0; |
| 861 | } |
| 862 | |
| 863 | switch(range->type) { |
| 864 | case 0: { |
| 865 | /* name = prefix hex-digits */ |
| 866 | const char *s=(const char *)(range+1); |
| 867 | char c; |
| 868 | |
| 869 | uint16_t i, count; |
| 870 | |
| 871 | /* copy prefix */ |
| 872 | while((c=*s++)!=0) { |
| 873 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 874 | } |
| 875 | |
| 876 | /* write hexadecimal code point value */ |
| 877 | count=range->variant; |
| 878 | |
| 879 | /* zero-terminate */ |
| 880 | if(count<bufferLength) { |
| 881 | buffer[count]=0; |
| 882 | } |
| 883 | |
| 884 | for(i=count; i>0;) { |
| 885 | if(--i<bufferLength) { |
| 886 | c=(char)(code&0xf); |
| 887 | if(c<10) { |
| 888 | c+='0'; |
| 889 | } else { |
| 890 | c+='A'-10; |
| 891 | } |
| 892 | buffer[i]=c; |
| 893 | } |
| 894 | code>>=4; |
| 895 | } |
| 896 | |
| 897 | bufferPos+=count; |
| 898 | break; |
| 899 | } |
| 900 | case 1: { |
| 901 | /* name = prefix factorized-elements */ |
| 902 | uint16_t indexes[8]; |
| 903 | const uint16_t *factors=(const uint16_t *)(range+1); |
| 904 | uint16_t count=range->variant; |
| 905 | const char *s=(const char *)(factors+count); |
| 906 | char c; |
| 907 | |
| 908 | /* copy prefix */ |
| 909 | while((c=*s++)!=0) { |
| 910 | WRITE_CHAR(buffer, bufferLength, bufferPos, c); |
| 911 | } |
| 912 | |
| 913 | bufferPos+=writeFactorSuffix(factors, count, |
| 914 | s, code-range->start, indexes, NULL, NULL, buffer, bufferLength); |
| 915 | break; |
| 916 | } |
| 917 | default: |
| 918 | /* undefined type */ |
| 919 | /* zero-terminate */ |
| 920 | if(bufferLength>0) { |
| 921 | *buffer=0; |
| 922 | } |
| 923 | break; |
| 924 | } |
| 925 | |
| 926 | return bufferPos; |
| 927 | } |
| 928 | |
| 929 | /* |
| 930 | * Important: enumAlgNames() and findAlgName() are almost the same. |
| 931 | * Any fix must be applied to both. |
| 932 | */ |
| 933 | static UBool |
| 934 | enumAlgNames(AlgorithmicRange *range, |
| 935 | UChar32 start, UChar32 limit, |
| 936 | UEnumCharNamesFn *fn, void *context, |
| 937 | UCharNameChoice nameChoice) { |
| 938 | char buffer[200]; |
| 939 | uint16_t length; |
| 940 | |
| 941 | if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { |
| 942 | return TRUE; |
| 943 | } |
| 944 | |
| 945 | switch(range->type) { |
| 946 | case 0: { |
| 947 | char *s, *end; |
| 948 | char c; |
| 949 | |
| 950 | /* get the full name of the start character */ |
| 951 | length=getAlgName(range, (uint32_t)start, nameChoice, buffer, sizeof(buffer)); |
| 952 | if(length<=0) { |
| 953 | return TRUE; |
| 954 | } |
| 955 | |
| 956 | /* call the enumerator function with this first character */ |
| 957 | if(!fn(context, start, nameChoice, buffer, length)) { |
| 958 | return FALSE; |
| 959 | } |
| 960 | |
| 961 | /* go to the end of the name; all these names have the same length */ |
| 962 | end=buffer; |
| 963 | while(*end!=0) { |
| 964 | ++end; |
| 965 | } |
| 966 | |
| 967 | /* enumerate the rest of the names */ |
| 968 | while(++start<limit) { |
| 969 | /* increment the hexadecimal number on a character-basis */ |
| 970 | s=end; |
| 971 | for (;;) { |
| 972 | c=*--s; |
| 973 | if(('0'<=c && c<'9') || ('A'<=c && c<'F')) { |
| 974 | *s=(char)(c+1); |
| 975 | break; |
| 976 | } else if(c=='9') { |
| 977 | *s='A'; |
| 978 | break; |
| 979 | } else if(c=='F') { |
| 980 | *s='0'; |
| 981 | } |
| 982 | } |
| 983 | |
| 984 | if(!fn(context, start, nameChoice, buffer, length)) { |
| 985 | return FALSE; |
| 986 | } |
| 987 | } |
| 988 | break; |
| 989 | } |
| 990 | case 1: { |
| 991 | uint16_t indexes[8]; |
| 992 | const char *elementBases[8], *elements[8]; |
| 993 | const uint16_t *factors=(const uint16_t *)(range+1); |
| 994 | uint16_t count=range->variant; |
| 995 | const char *s=(const char *)(factors+count); |
| 996 | char *suffix, *t; |
| 997 | uint16_t prefixLength, i, idx; |
| 998 | |
| 999 | char c; |
| 1000 | |
| 1001 | /* name = prefix factorized-elements */ |
| 1002 | |
| 1003 | /* copy prefix */ |
| 1004 | suffix=buffer; |
| 1005 | prefixLength=0; |
| 1006 | while((c=*s++)!=0) { |
| 1007 | *suffix++=c; |
| 1008 | ++prefixLength; |
| 1009 | } |
| 1010 | |
| 1011 | /* append the suffix of the start character */ |
| 1012 | length=(uint16_t)(prefixLength+writeFactorSuffix(factors, count, |
| 1013 | s, (uint32_t)start-range->start, |
| 1014 | indexes, elementBases, elements, |
| 1015 | suffix, (uint16_t)(sizeof(buffer)-prefixLength))); |
| 1016 | |
| 1017 | /* call the enumerator function with this first character */ |
| 1018 | if(!fn(context, start, nameChoice, buffer, length)) { |
| 1019 | return FALSE; |
| 1020 | } |
| 1021 | |
| 1022 | /* enumerate the rest of the names */ |
| 1023 | while(++start<limit) { |
| 1024 | /* increment the indexes in lexical order bound by the factors */ |
| 1025 | i=count; |
| 1026 | for (;;) { |
| 1027 | idx=(uint16_t)(indexes[--i]+1); |
| 1028 | if(idx<factors[i]) { |
| 1029 | /* skip one index and its element string */ |
| 1030 | indexes[i]=idx; |
| 1031 | s=elements[i]; |
| 1032 | while(*s++!=0) { |
| 1033 | } |
| 1034 | elements[i]=s; |
| 1035 | break; |
| 1036 | } else { |
| 1037 | /* reset this index to 0 and its element string to the first one */ |
| 1038 | indexes[i]=0; |
| 1039 | elements[i]=elementBases[i]; |
| 1040 | } |
| 1041 | } |
| 1042 | |
| 1043 | /* to make matters a little easier, just append all elements to the suffix */ |
| 1044 | t=suffix; |
| 1045 | length=prefixLength; |
| 1046 | for(i=0; i<count; ++i) { |
| 1047 | s=elements[i]; |
| 1048 | while((c=*s++)!=0) { |
| 1049 | *t++=c; |
| 1050 | ++length; |
| 1051 | } |
| 1052 | } |
| 1053 | /* zero-terminate */ |
| 1054 | *t=0; |
| 1055 | |
| 1056 | if(!fn(context, start, nameChoice, buffer, length)) { |
| 1057 | return FALSE; |
| 1058 | } |
| 1059 | } |
| 1060 | break; |
| 1061 | } |
| 1062 | default: |
| 1063 | /* undefined type */ |
| 1064 | break; |
| 1065 | } |
| 1066 | |
| 1067 | return TRUE; |
| 1068 | } |
| 1069 | |
| 1070 | /* |
| 1071 | * findAlgName() is almost the same as enumAlgNames() except that it |
| 1072 | * returns the code point for a name if it fits into the range. |
| 1073 | * It returns 0xffff otherwise. |
| 1074 | */ |
| 1075 | static UChar32 |
| 1076 | findAlgName(AlgorithmicRange *range, UCharNameChoice nameChoice, const char *otherName) { |
| 1077 | UChar32 code; |
| 1078 | |
| 1079 | if(nameChoice!=U_UNICODE_CHAR_NAME && nameChoice!=U_EXTENDED_CHAR_NAME) { |
| 1080 | return 0xffff; |
| 1081 | } |
| 1082 | |
| 1083 | switch(range->type) { |
| 1084 | case 0: { |
| 1085 | /* name = prefix hex-digits */ |
| 1086 | const char *s=(const char *)(range+1); |
| 1087 | char c; |
| 1088 | |
| 1089 | uint16_t i, count; |
| 1090 | |
| 1091 | /* compare prefix */ |
| 1092 | while((c=*s++)!=0) { |
| 1093 | if((char)c!=*otherName++) { |
| 1094 | return 0xffff; |
| 1095 | } |
| 1096 | } |
| 1097 | |
| 1098 | /* read hexadecimal code point value */ |
| 1099 | count=range->variant; |
| 1100 | code=0; |
| 1101 | for(i=0; i<count; ++i) { |
| 1102 | c=*otherName++; |
| 1103 | if('0'<=c && c<='9') { |
| 1104 | code=(code<<4)|(c-'0'); |
| 1105 | } else if('A'<=c && c<='F') { |
| 1106 | code=(code<<4)|(c-'A'+10); |
| 1107 | } else { |
| 1108 | return 0xffff; |
| 1109 | } |
| 1110 | } |
| 1111 | |
| 1112 | /* does it fit into the range? */ |
| 1113 | if(*otherName==0 && range->start<=(uint32_t)code && (uint32_t)code<=range->end) { |
| 1114 | return code; |
| 1115 | } |
| 1116 | break; |
| 1117 | } |
| 1118 | case 1: { |
| 1119 | char buffer[64]; |
| 1120 | uint16_t indexes[8]; |
| 1121 | const char *elementBases[8], *elements[8]; |
| 1122 | const uint16_t *factors=(const uint16_t *)(range+1); |
| 1123 | uint16_t count=range->variant; |
| 1124 | const char *s=(const char *)(factors+count), *t; |
| 1125 | UChar32 start, limit; |
| 1126 | uint16_t i, idx; |
| 1127 | |
| 1128 | char c; |
| 1129 | |
| 1130 | /* name = prefix factorized-elements */ |
| 1131 | |
| 1132 | /* compare prefix */ |
| 1133 | while((c=*s++)!=0) { |
| 1134 | if((char)c!=*otherName++) { |
| 1135 | return 0xffff; |
| 1136 | } |
| 1137 | } |
| 1138 | |
| 1139 | start=(UChar32)range->start; |
| 1140 | limit=(UChar32)(range->end+1); |
| 1141 | |
| 1142 | /* initialize the suffix elements for enumeration; indexes should all be set to 0 */ |
| 1143 | writeFactorSuffix(factors, count, s, 0, |
| 1144 | indexes, elementBases, elements, buffer, sizeof(buffer)); |
| 1145 | |
| 1146 | /* compare the first suffix */ |
| 1147 | if(0==uprv_strcmp(otherName, buffer)) { |
| 1148 | return start; |
| 1149 | } |
| 1150 | |
| 1151 | /* enumerate and compare the rest of the suffixes */ |
| 1152 | while(++start<limit) { |
| 1153 | /* increment the indexes in lexical order bound by the factors */ |
| 1154 | i=count; |
| 1155 | for (;;) { |
| 1156 | idx=(uint16_t)(indexes[--i]+1); |
| 1157 | if(idx<factors[i]) { |
| 1158 | /* skip one index and its element string */ |
| 1159 | indexes[i]=idx; |
| 1160 | s=elements[i]; |
| 1161 | while(*s++!=0) {} |
| 1162 | elements[i]=s; |
| 1163 | break; |
| 1164 | } else { |
| 1165 | /* reset this index to 0 and its element string to the first one */ |
| 1166 | indexes[i]=0; |
| 1167 | elements[i]=elementBases[i]; |
| 1168 | } |
| 1169 | } |
| 1170 | |
| 1171 | /* to make matters a little easier, just compare all elements of the suffix */ |
| 1172 | t=otherName; |
| 1173 | for(i=0; i<count; ++i) { |
| 1174 | s=elements[i]; |
| 1175 | while((c=*s++)!=0) { |
| 1176 | if(c!=*t++) { |
| 1177 | s=""; /* does not match */ |
| 1178 | i=99; |
| 1179 | } |
| 1180 | } |
| 1181 | } |
| 1182 | if(i<99 && *t==0) { |
| 1183 | return start; |
| 1184 | } |
| 1185 | } |
| 1186 | break; |
| 1187 | } |
| 1188 | default: |
| 1189 | /* undefined type */ |
| 1190 | break; |
| 1191 | } |
| 1192 | |
| 1193 | return 0xffff; |
| 1194 | } |
| 1195 | |
| 1196 | /* sets of name characters, maximum name lengths ---------------------------- */ |
| 1197 | |
| 1198 | #define SET_ADD(set, c) ((set)[(uint8_t)c>>5]|=((uint32_t)1<<((uint8_t)c&0x1f))) |
| 1199 | #define SET_CONTAINS(set, c) (((set)[(uint8_t)c>>5]&((uint32_t)1<<((uint8_t)c&0x1f)))!=0) |
| 1200 | |
| 1201 | static int32_t |
| 1202 | calcStringSetLength(uint32_t set[8], const char *s) { |
| 1203 | int32_t length=0; |
| 1204 | char c; |
| 1205 | |
| 1206 | while((c=*s++)!=0) { |
| 1207 | SET_ADD(set, c); |
| 1208 | ++length; |
| 1209 | } |
| 1210 | return length; |
| 1211 | } |
| 1212 | |
| 1213 | static int32_t |
| 1214 | calcAlgNameSetsLengths(int32_t maxNameLength) { |
| 1215 | AlgorithmicRange *range; |
| 1216 | uint32_t *p; |
| 1217 | uint32_t rangeCount; |
| 1218 | int32_t length; |
| 1219 | |
| 1220 | /* enumerate algorithmic ranges */ |
| 1221 | p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); |
| 1222 | rangeCount=*p; |
| 1223 | range=(AlgorithmicRange *)(p+1); |
| 1224 | while(rangeCount>0) { |
| 1225 | switch(range->type) { |
| 1226 | case 0: |
| 1227 | /* name = prefix + (range->variant times) hex-digits */ |
| 1228 | /* prefix */ |
| 1229 | length=calcStringSetLength(gNameSet, (const char *)(range+1))+range->variant; |
| 1230 | if(length>maxNameLength) { |
| 1231 | maxNameLength=length; |
| 1232 | } |
| 1233 | break; |
| 1234 | case 1: { |
| 1235 | /* name = prefix factorized-elements */ |
| 1236 | const uint16_t *factors=(const uint16_t *)(range+1); |
| 1237 | const char *s; |
| 1238 | int32_t i, count=range->variant, factor, factorLength, maxFactorLength; |
| 1239 | |
| 1240 | /* prefix length */ |
| 1241 | s=(const char *)(factors+count); |
| 1242 | length=calcStringSetLength(gNameSet, s); |
| 1243 | s+=length+1; /* start of factor suffixes */ |
| 1244 | |
| 1245 | /* get the set and maximum factor suffix length for each factor */ |
| 1246 | for(i=0; i<count; ++i) { |
| 1247 | maxFactorLength=0; |
| 1248 | for(factor=factors[i]; factor>0; --factor) { |
| 1249 | factorLength=calcStringSetLength(gNameSet, s); |
| 1250 | s+=factorLength+1; |
| 1251 | if(factorLength>maxFactorLength) { |
| 1252 | maxFactorLength=factorLength; |
| 1253 | } |
| 1254 | } |
| 1255 | length+=maxFactorLength; |
| 1256 | } |
| 1257 | |
| 1258 | if(length>maxNameLength) { |
| 1259 | maxNameLength=length; |
| 1260 | } |
| 1261 | break; |
| 1262 | } |
| 1263 | default: |
| 1264 | /* unknown type */ |
| 1265 | break; |
| 1266 | } |
| 1267 | |
| 1268 | range=(AlgorithmicRange *)((uint8_t *)range+range->size); |
| 1269 | --rangeCount; |
| 1270 | } |
| 1271 | return maxNameLength; |
| 1272 | } |
| 1273 | |
| 1274 | static int32_t |
| 1275 | calcExtNameSetsLengths(int32_t maxNameLength) { |
| 1276 | int32_t i, length; |
| 1277 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1278 | for(i=0; i<UPRV_LENGTHOF(charCatNames); ++i) { |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1279 | /* |
| 1280 | * for each category, count the length of the category name |
| 1281 | * plus 9= |
| 1282 | * 2 for <> |
| 1283 | * 1 for - |
| 1284 | * 6 for most hex digits per code point |
| 1285 | */ |
| 1286 | length=9+calcStringSetLength(gNameSet, charCatNames[i]); |
| 1287 | if(length>maxNameLength) { |
| 1288 | maxNameLength=length; |
| 1289 | } |
| 1290 | } |
| 1291 | return maxNameLength; |
| 1292 | } |
| 1293 | |
| 1294 | static int32_t |
| 1295 | calcNameSetLength(const uint16_t *tokens, uint16_t tokenCount, const uint8_t *tokenStrings, int8_t *tokenLengths, |
| 1296 | uint32_t set[8], |
| 1297 | const uint8_t **pLine, const uint8_t *lineLimit) { |
| 1298 | const uint8_t *line=*pLine; |
| 1299 | int32_t length=0, tokenLength; |
| 1300 | uint16_t c, token; |
| 1301 | |
| 1302 | while(line!=lineLimit && (c=*line++)!=(uint8_t)';') { |
| 1303 | if(c>=tokenCount) { |
| 1304 | /* implicit letter */ |
| 1305 | SET_ADD(set, c); |
| 1306 | ++length; |
| 1307 | } else { |
| 1308 | token=tokens[c]; |
| 1309 | if(token==(uint16_t)(-2)) { |
| 1310 | /* this is a lead byte for a double-byte token */ |
| 1311 | c=c<<8|*line++; |
| 1312 | token=tokens[c]; |
| 1313 | } |
| 1314 | if(token==(uint16_t)(-1)) { |
| 1315 | /* explicit letter */ |
| 1316 | SET_ADD(set, c); |
| 1317 | ++length; |
| 1318 | } else { |
| 1319 | /* count token word */ |
| 1320 | if(tokenLengths!=NULL) { |
| 1321 | /* use cached token length */ |
| 1322 | tokenLength=tokenLengths[c]; |
| 1323 | if(tokenLength==0) { |
| 1324 | tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token); |
| 1325 | tokenLengths[c]=(int8_t)tokenLength; |
| 1326 | } |
| 1327 | } else { |
| 1328 | tokenLength=calcStringSetLength(set, (const char *)tokenStrings+token); |
| 1329 | } |
| 1330 | length+=tokenLength; |
| 1331 | } |
| 1332 | } |
| 1333 | } |
| 1334 | |
| 1335 | *pLine=line; |
| 1336 | return length; |
| 1337 | } |
| 1338 | |
| 1339 | static void |
| 1340 | calcGroupNameSetsLengths(int32_t maxNameLength) { |
| 1341 | uint16_t offsets[LINES_PER_GROUP+2], lengths[LINES_PER_GROUP+2]; |
| 1342 | |
| 1343 | uint16_t *tokens=(uint16_t *)uCharNames+8; |
| 1344 | uint16_t tokenCount=*tokens++; |
| 1345 | uint8_t *tokenStrings=(uint8_t *)uCharNames+uCharNames->tokenStringOffset; |
| 1346 | |
| 1347 | int8_t *tokenLengths; |
| 1348 | |
| 1349 | const uint16_t *group; |
| 1350 | const uint8_t *s, *line, *lineLimit; |
| 1351 | |
| 1352 | int32_t groupCount, lineNumber, length; |
| 1353 | |
| 1354 | tokenLengths=(int8_t *)uprv_malloc(tokenCount); |
| 1355 | if(tokenLengths!=NULL) { |
| 1356 | uprv_memset(tokenLengths, 0, tokenCount); |
| 1357 | } |
| 1358 | |
| 1359 | group=GET_GROUPS(uCharNames); |
| 1360 | groupCount=*group++; |
| 1361 | |
| 1362 | /* enumerate all groups */ |
| 1363 | while(groupCount>0) { |
| 1364 | s=(uint8_t *)uCharNames+uCharNames->groupStringOffset+GET_GROUP_OFFSET(group); |
| 1365 | s=expandGroupLengths(s, offsets, lengths); |
| 1366 | |
| 1367 | /* enumerate all lines in each group */ |
| 1368 | for(lineNumber=0; lineNumber<LINES_PER_GROUP; ++lineNumber) { |
| 1369 | line=s+offsets[lineNumber]; |
| 1370 | length=lengths[lineNumber]; |
| 1371 | if(length==0) { |
| 1372 | continue; |
| 1373 | } |
| 1374 | |
| 1375 | lineLimit=line+length; |
| 1376 | |
| 1377 | /* read regular name */ |
| 1378 | length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gNameSet, &line, lineLimit); |
| 1379 | if(length>maxNameLength) { |
| 1380 | maxNameLength=length; |
| 1381 | } |
| 1382 | if(line==lineLimit) { |
| 1383 | continue; |
| 1384 | } |
| 1385 | |
| 1386 | /* read Unicode 1.0 name */ |
| 1387 | length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gNameSet, &line, lineLimit); |
| 1388 | if(length>maxNameLength) { |
| 1389 | maxNameLength=length; |
| 1390 | } |
| 1391 | if(line==lineLimit) { |
| 1392 | continue; |
| 1393 | } |
| 1394 | |
| 1395 | /* read ISO comment */ |
| 1396 | /*length=calcNameSetLength(tokens, tokenCount, tokenStrings, tokenLengths, gISOCommentSet, &line, lineLimit);*/ |
| 1397 | } |
| 1398 | |
| 1399 | group=NEXT_GROUP(group); |
| 1400 | --groupCount; |
| 1401 | } |
| 1402 | |
| 1403 | if(tokenLengths!=NULL) { |
| 1404 | uprv_free(tokenLengths); |
| 1405 | } |
| 1406 | |
| 1407 | /* set gMax... - name length last for threading */ |
| 1408 | gMaxNameLength=maxNameLength; |
| 1409 | } |
| 1410 | |
| 1411 | static UBool |
| 1412 | calcNameSetsLengths(UErrorCode *pErrorCode) { |
| 1413 | static const char extChars[]="0123456789ABCDEF<>-"; |
| 1414 | int32_t i, maxNameLength; |
| 1415 | |
| 1416 | if(gMaxNameLength!=0) { |
| 1417 | return TRUE; |
| 1418 | } |
| 1419 | |
| 1420 | if(!isDataLoaded(pErrorCode)) { |
| 1421 | return FALSE; |
| 1422 | } |
| 1423 | |
| 1424 | /* set hex digits, used in various names, and <>-, used in extended names */ |
| 1425 | for(i=0; i<(int32_t)sizeof(extChars)-1; ++i) { |
| 1426 | SET_ADD(gNameSet, extChars[i]); |
| 1427 | } |
| 1428 | |
| 1429 | /* set sets and lengths from algorithmic names */ |
| 1430 | maxNameLength=calcAlgNameSetsLengths(0); |
| 1431 | |
| 1432 | /* set sets and lengths from extended names */ |
| 1433 | maxNameLength=calcExtNameSetsLengths(maxNameLength); |
| 1434 | |
| 1435 | /* set sets and lengths from group names, set global maximum values */ |
| 1436 | calcGroupNameSetsLengths(maxNameLength); |
| 1437 | |
| 1438 | return TRUE; |
| 1439 | } |
| 1440 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1441 | U_NAMESPACE_END |
| 1442 | |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1443 | /* public API --------------------------------------------------------------- */ |
| 1444 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1445 | U_NAMESPACE_USE |
| 1446 | |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1447 | U_CAPI int32_t U_EXPORT2 |
| 1448 | u_charName(UChar32 code, UCharNameChoice nameChoice, |
| 1449 | char *buffer, int32_t bufferLength, |
| 1450 | UErrorCode *pErrorCode) { |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1451 | AlgorithmicRange *algRange; |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1452 | uint32_t *p; |
| 1453 | uint32_t i; |
| 1454 | int32_t length; |
| 1455 | |
| 1456 | /* check the argument values */ |
| 1457 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 1458 | return 0; |
| 1459 | } else if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || |
| 1460 | bufferLength<0 || (bufferLength>0 && buffer==NULL) |
| 1461 | ) { |
| 1462 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 1463 | return 0; |
| 1464 | } |
| 1465 | |
| 1466 | if((uint32_t)code>UCHAR_MAX_VALUE || !isDataLoaded(pErrorCode)) { |
| 1467 | return u_terminateChars(buffer, bufferLength, 0, pErrorCode); |
| 1468 | } |
| 1469 | |
| 1470 | length=0; |
| 1471 | |
| 1472 | /* try algorithmic names first */ |
| 1473 | p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); |
| 1474 | i=*p; |
| 1475 | algRange=(AlgorithmicRange *)(p+1); |
| 1476 | while(i>0) { |
| 1477 | if(algRange->start<=(uint32_t)code && (uint32_t)code<=algRange->end) { |
| 1478 | length=getAlgName(algRange, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength); |
| 1479 | break; |
| 1480 | } |
| 1481 | algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size); |
| 1482 | --i; |
| 1483 | } |
| 1484 | |
| 1485 | if(i==0) { |
| 1486 | if (nameChoice == U_EXTENDED_CHAR_NAME) { |
| 1487 | length = getName(uCharNames, (uint32_t )code, U_EXTENDED_CHAR_NAME, buffer, (uint16_t) bufferLength); |
| 1488 | if (!length) { |
| 1489 | /* extended character name */ |
| 1490 | length = getExtName((uint32_t) code, buffer, (uint16_t) bufferLength); |
| 1491 | } |
| 1492 | } else { |
| 1493 | /* normal character name */ |
| 1494 | length=getName(uCharNames, (uint32_t)code, nameChoice, buffer, (uint16_t)bufferLength); |
| 1495 | } |
| 1496 | } |
| 1497 | |
| 1498 | return u_terminateChars(buffer, bufferLength, length, pErrorCode); |
| 1499 | } |
| 1500 | |
| 1501 | U_CAPI int32_t U_EXPORT2 |
| 1502 | u_getISOComment(UChar32 /*c*/, |
| 1503 | char *dest, int32_t destCapacity, |
| 1504 | UErrorCode *pErrorCode) { |
| 1505 | /* check the argument values */ |
| 1506 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 1507 | return 0; |
| 1508 | } else if(destCapacity<0 || (destCapacity>0 && dest==NULL)) { |
| 1509 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 1510 | return 0; |
| 1511 | } |
| 1512 | |
| 1513 | return u_terminateChars(dest, destCapacity, 0, pErrorCode); |
| 1514 | } |
| 1515 | |
| 1516 | U_CAPI UChar32 U_EXPORT2 |
| 1517 | u_charFromName(UCharNameChoice nameChoice, |
| 1518 | const char *name, |
| 1519 | UErrorCode *pErrorCode) { |
| 1520 | char upper[120], lower[120]; |
| 1521 | FindName findName; |
| 1522 | AlgorithmicRange *algRange; |
| 1523 | uint32_t *p; |
| 1524 | uint32_t i; |
| 1525 | UChar32 cp = 0; |
| 1526 | char c0; |
| 1527 | UChar32 error = 0xffff; /* Undefined, but use this for backwards compatibility. */ |
| 1528 | |
| 1529 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 1530 | return error; |
| 1531 | } |
| 1532 | |
| 1533 | if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || name==NULL || *name==0) { |
| 1534 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 1535 | return error; |
| 1536 | } |
| 1537 | |
| 1538 | if(!isDataLoaded(pErrorCode)) { |
| 1539 | return error; |
| 1540 | } |
| 1541 | |
| 1542 | /* construct the uppercase and lowercase of the name first */ |
| 1543 | for(i=0; i<sizeof(upper); ++i) { |
| 1544 | if((c0=*name++)!=0) { |
| 1545 | upper[i]=uprv_toupper(c0); |
| 1546 | lower[i]=uprv_tolower(c0); |
| 1547 | } else { |
| 1548 | upper[i]=lower[i]=0; |
| 1549 | break; |
| 1550 | } |
| 1551 | } |
| 1552 | if(i==sizeof(upper)) { |
| 1553 | /* name too long, there is no such character */ |
| 1554 | *pErrorCode = U_ILLEGAL_CHAR_FOUND; |
| 1555 | return error; |
| 1556 | } |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1557 | // i==strlen(name)==strlen(lower)==strlen(upper) |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1558 | |
| 1559 | /* try extended names first */ |
| 1560 | if (lower[0] == '<') { |
| 1561 | if (nameChoice == U_EXTENDED_CHAR_NAME) { |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1562 | // Parse a string like "<category-HHHH>" where HHHH is a hex code point. |
| 1563 | if (lower[--i] == '>' && i >= 3 && lower[--i] != '-') { |
| 1564 | while (i >= 3 && lower[--i] != '-') {} |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1565 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1566 | if (i >= 2 && lower[i] == '-') { |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1567 | uint32_t cIdx; |
| 1568 | |
| 1569 | lower[i] = 0; |
| 1570 | |
| 1571 | for (++i; lower[i] != '>'; ++i) { |
| 1572 | if (lower[i] >= '0' && lower[i] <= '9') { |
| 1573 | cp = (cp << 4) + lower[i] - '0'; |
| 1574 | } else if (lower[i] >= 'a' && lower[i] <= 'f') { |
| 1575 | cp = (cp << 4) + lower[i] - 'a' + 10; |
| 1576 | } else { |
| 1577 | *pErrorCode = U_ILLEGAL_CHAR_FOUND; |
| 1578 | return error; |
| 1579 | } |
| 1580 | } |
| 1581 | |
| 1582 | /* Now validate the category name. |
| 1583 | We could use a binary search, or a trie, if |
| 1584 | we really wanted to. */ |
| 1585 | |
Jungshik Shin (jungshik at google) | 0f8746a | 2015-01-08 15:46:45 -0800 | [diff] [blame^] | 1586 | for (lower[i] = 0, cIdx = 0; cIdx < UPRV_LENGTHOF(charCatNames); ++cIdx) { |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 1587 | |
| 1588 | if (!uprv_strcmp(lower + 1, charCatNames[cIdx])) { |
| 1589 | if (getCharCat(cp) == cIdx) { |
| 1590 | return cp; |
| 1591 | } |
| 1592 | break; |
| 1593 | } |
| 1594 | } |
| 1595 | } |
| 1596 | } |
| 1597 | } |
| 1598 | |
| 1599 | *pErrorCode = U_ILLEGAL_CHAR_FOUND; |
| 1600 | return error; |
| 1601 | } |
| 1602 | |
| 1603 | /* try algorithmic names now */ |
| 1604 | p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); |
| 1605 | i=*p; |
| 1606 | algRange=(AlgorithmicRange *)(p+1); |
| 1607 | while(i>0) { |
| 1608 | if((cp=findAlgName(algRange, nameChoice, upper))!=0xffff) { |
| 1609 | return cp; |
| 1610 | } |
| 1611 | algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size); |
| 1612 | --i; |
| 1613 | } |
| 1614 | |
| 1615 | /* normal character name */ |
| 1616 | findName.otherName=upper; |
| 1617 | findName.code=error; |
| 1618 | enumNames(uCharNames, 0, UCHAR_MAX_VALUE + 1, DO_FIND_NAME, &findName, nameChoice); |
| 1619 | if (findName.code == error) { |
| 1620 | *pErrorCode = U_ILLEGAL_CHAR_FOUND; |
| 1621 | } |
| 1622 | return findName.code; |
| 1623 | } |
| 1624 | |
| 1625 | U_CAPI void U_EXPORT2 |
| 1626 | u_enumCharNames(UChar32 start, UChar32 limit, |
| 1627 | UEnumCharNamesFn *fn, |
| 1628 | void *context, |
| 1629 | UCharNameChoice nameChoice, |
| 1630 | UErrorCode *pErrorCode) { |
| 1631 | AlgorithmicRange *algRange; |
| 1632 | uint32_t *p; |
| 1633 | uint32_t i; |
| 1634 | |
| 1635 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 1636 | return; |
| 1637 | } |
| 1638 | |
| 1639 | if(nameChoice>=U_CHAR_NAME_CHOICE_COUNT || fn==NULL) { |
| 1640 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 1641 | return; |
| 1642 | } |
| 1643 | |
| 1644 | if((uint32_t) limit > UCHAR_MAX_VALUE + 1) { |
| 1645 | limit = UCHAR_MAX_VALUE + 1; |
| 1646 | } |
| 1647 | if((uint32_t)start>=(uint32_t)limit) { |
| 1648 | return; |
| 1649 | } |
| 1650 | |
| 1651 | if(!isDataLoaded(pErrorCode)) { |
| 1652 | return; |
| 1653 | } |
| 1654 | |
| 1655 | /* interleave the data-driven ones with the algorithmic ones */ |
| 1656 | /* iterate over all algorithmic ranges; assume that they are in ascending order */ |
| 1657 | p=(uint32_t *)((uint8_t *)uCharNames+uCharNames->algNamesOffset); |
| 1658 | i=*p; |
| 1659 | algRange=(AlgorithmicRange *)(p+1); |
| 1660 | while(i>0) { |
| 1661 | /* enumerate the character names before the current algorithmic range */ |
| 1662 | /* here: start<limit */ |
| 1663 | if((uint32_t)start<algRange->start) { |
| 1664 | if((uint32_t)limit<=algRange->start) { |
| 1665 | enumNames(uCharNames, start, limit, fn, context, nameChoice); |
| 1666 | return; |
| 1667 | } |
| 1668 | if(!enumNames(uCharNames, start, (UChar32)algRange->start, fn, context, nameChoice)) { |
| 1669 | return; |
| 1670 | } |
| 1671 | start=(UChar32)algRange->start; |
| 1672 | } |
| 1673 | /* enumerate the character names in the current algorithmic range */ |
| 1674 | /* here: algRange->start<=start<limit */ |
| 1675 | if((uint32_t)start<=algRange->end) { |
| 1676 | if((uint32_t)limit<=(algRange->end+1)) { |
| 1677 | enumAlgNames(algRange, start, limit, fn, context, nameChoice); |
| 1678 | return; |
| 1679 | } |
| 1680 | if(!enumAlgNames(algRange, start, (UChar32)algRange->end+1, fn, context, nameChoice)) { |
| 1681 | return; |
| 1682 | } |
| 1683 | start=(UChar32)algRange->end+1; |
| 1684 | } |
| 1685 | /* continue to the next algorithmic range (here: start<limit) */ |
| 1686 | algRange=(AlgorithmicRange *)((uint8_t *)algRange+algRange->size); |
| 1687 | --i; |
| 1688 | } |
| 1689 | /* enumerate the character names after the last algorithmic range */ |
| 1690 | enumNames(uCharNames, start, limit, fn, context, nameChoice); |
| 1691 | } |
| 1692 | |
| 1693 | U_CAPI int32_t U_EXPORT2 |
| 1694 | uprv_getMaxCharNameLength() { |
| 1695 | UErrorCode errorCode=U_ZERO_ERROR; |
| 1696 | if(calcNameSetsLengths(&errorCode)) { |
| 1697 | return gMaxNameLength; |
| 1698 | } else { |
| 1699 | return 0; |
| 1700 | } |
| 1701 | } |
| 1702 | |
| 1703 | /** |
| 1704 | * Converts the char set cset into a Unicode set uset. |
| 1705 | * @param cset Set of 256 bit flags corresponding to a set of chars. |
| 1706 | * @param uset USet to receive characters. Existing contents are deleted. |
| 1707 | */ |
| 1708 | static void |
| 1709 | charSetToUSet(uint32_t cset[8], const USetAdder *sa) { |
| 1710 | UChar us[256]; |
| 1711 | char cs[256]; |
| 1712 | |
| 1713 | int32_t i, length; |
| 1714 | UErrorCode errorCode; |
| 1715 | |
| 1716 | errorCode=U_ZERO_ERROR; |
| 1717 | |
| 1718 | if(!calcNameSetsLengths(&errorCode)) { |
| 1719 | return; |
| 1720 | } |
| 1721 | |
| 1722 | /* build a char string with all chars that are used in character names */ |
| 1723 | length=0; |
| 1724 | for(i=0; i<256; ++i) { |
| 1725 | if(SET_CONTAINS(cset, i)) { |
| 1726 | cs[length++]=(char)i; |
| 1727 | } |
| 1728 | } |
| 1729 | |
| 1730 | /* convert the char string to a UChar string */ |
| 1731 | u_charsToUChars(cs, us, length); |
| 1732 | |
| 1733 | /* add each UChar to the USet */ |
| 1734 | for(i=0; i<length; ++i) { |
| 1735 | if(us[i]!=0 || cs[i]==0) { /* non-invariant chars become (UChar)0 */ |
| 1736 | sa->add(sa->set, us[i]); |
| 1737 | } |
| 1738 | } |
| 1739 | } |
| 1740 | |
| 1741 | /** |
| 1742 | * Fills set with characters that are used in Unicode character names. |
| 1743 | * @param set USet to receive characters. |
| 1744 | */ |
| 1745 | U_CAPI void U_EXPORT2 |
| 1746 | uprv_getCharNameCharacters(const USetAdder *sa) { |
| 1747 | charSetToUSet(gNameSet, sa); |
| 1748 | } |
| 1749 | |
| 1750 | /* data swapping ------------------------------------------------------------ */ |
| 1751 | |
| 1752 | /* |
| 1753 | * The token table contains non-negative entries for token bytes, |
| 1754 | * and -1 for bytes that represent themselves in the data file's charset. |
| 1755 | * -2 entries are used for lead bytes. |
| 1756 | * |
| 1757 | * Direct bytes (-1 entries) must be translated from the input charset family |
| 1758 | * to the output charset family. |
| 1759 | * makeTokenMap() writes a permutation mapping for this. |
| 1760 | * Use it once for single-/lead-byte tokens and once more for all trail byte |
| 1761 | * tokens. (';' is an unused trail byte marked with -1.) |
| 1762 | */ |
| 1763 | static void |
| 1764 | makeTokenMap(const UDataSwapper *ds, |
| 1765 | int16_t tokens[], uint16_t tokenCount, |
| 1766 | uint8_t map[256], |
| 1767 | UErrorCode *pErrorCode) { |
| 1768 | UBool usedOutChar[256]; |
| 1769 | uint16_t i, j; |
| 1770 | uint8_t c1, c2; |
| 1771 | |
| 1772 | if(U_FAILURE(*pErrorCode)) { |
| 1773 | return; |
| 1774 | } |
| 1775 | |
| 1776 | if(ds->inCharset==ds->outCharset) { |
| 1777 | /* Same charset family: identity permutation */ |
| 1778 | for(i=0; i<256; ++i) { |
| 1779 | map[i]=(uint8_t)i; |
| 1780 | } |
| 1781 | } else { |
| 1782 | uprv_memset(map, 0, 256); |
| 1783 | uprv_memset(usedOutChar, 0, 256); |
| 1784 | |
| 1785 | if(tokenCount>256) { |
| 1786 | tokenCount=256; |
| 1787 | } |
| 1788 | |
| 1789 | /* set the direct bytes (byte 0 always maps to itself) */ |
| 1790 | for(i=1; i<tokenCount; ++i) { |
| 1791 | if(tokens[i]==-1) { |
| 1792 | /* convert the direct byte character */ |
| 1793 | c1=(uint8_t)i; |
| 1794 | ds->swapInvChars(ds, &c1, 1, &c2, pErrorCode); |
| 1795 | if(U_FAILURE(*pErrorCode)) { |
| 1796 | udata_printError(ds, "unames/makeTokenMap() finds variant character 0x%02x used (input charset family %d)\n", |
| 1797 | i, ds->inCharset); |
| 1798 | return; |
| 1799 | } |
| 1800 | |
| 1801 | /* enter the converted character into the map and mark it used */ |
| 1802 | map[c1]=c2; |
| 1803 | usedOutChar[c2]=TRUE; |
| 1804 | } |
| 1805 | } |
| 1806 | |
| 1807 | /* set the mappings for the rest of the permutation */ |
| 1808 | for(i=j=1; i<tokenCount; ++i) { |
| 1809 | /* set mappings that were not set for direct bytes */ |
| 1810 | if(map[i]==0) { |
| 1811 | /* set an output byte value that was not used as an output byte above */ |
| 1812 | while(usedOutChar[j]) { |
| 1813 | ++j; |
| 1814 | } |
| 1815 | map[i]=(uint8_t)j++; |
| 1816 | } |
| 1817 | } |
| 1818 | |
| 1819 | /* |
| 1820 | * leave mappings at tokenCount and above unset if tokenCount<256 |
| 1821 | * because they won't be used |
| 1822 | */ |
| 1823 | } |
| 1824 | } |
| 1825 | |
| 1826 | U_CAPI int32_t U_EXPORT2 |
| 1827 | uchar_swapNames(const UDataSwapper *ds, |
| 1828 | const void *inData, int32_t length, void *outData, |
| 1829 | UErrorCode *pErrorCode) { |
| 1830 | const UDataInfo *pInfo; |
| 1831 | int32_t headerSize; |
| 1832 | |
| 1833 | const uint8_t *inBytes; |
| 1834 | uint8_t *outBytes; |
| 1835 | |
| 1836 | uint32_t tokenStringOffset, groupsOffset, groupStringOffset, algNamesOffset, |
| 1837 | offset, i, count, stringsCount; |
| 1838 | |
| 1839 | const AlgorithmicRange *inRange; |
| 1840 | AlgorithmicRange *outRange; |
| 1841 | |
| 1842 | /* udata_swapDataHeader checks the arguments */ |
| 1843 | headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); |
| 1844 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 1845 | return 0; |
| 1846 | } |
| 1847 | |
| 1848 | /* check data format and format version */ |
| 1849 | pInfo=(const UDataInfo *)((const char *)inData+4); |
| 1850 | if(!( |
| 1851 | pInfo->dataFormat[0]==0x75 && /* dataFormat="unam" */ |
| 1852 | pInfo->dataFormat[1]==0x6e && |
| 1853 | pInfo->dataFormat[2]==0x61 && |
| 1854 | pInfo->dataFormat[3]==0x6d && |
| 1855 | pInfo->formatVersion[0]==1 |
| 1856 | )) { |
| 1857 | udata_printError(ds, "uchar_swapNames(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as unames.icu\n", |
| 1858 | pInfo->dataFormat[0], pInfo->dataFormat[1], |
| 1859 | pInfo->dataFormat[2], pInfo->dataFormat[3], |
| 1860 | pInfo->formatVersion[0]); |
| 1861 | *pErrorCode=U_UNSUPPORTED_ERROR; |
| 1862 | return 0; |
| 1863 | } |
| 1864 | |
| 1865 | inBytes=(const uint8_t *)inData+headerSize; |
| 1866 | outBytes=(uint8_t *)outData+headerSize; |
| 1867 | if(length<0) { |
| 1868 | algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3]); |
| 1869 | } else { |
| 1870 | length-=headerSize; |
| 1871 | if( length<20 || |
| 1872 | (uint32_t)length<(algNamesOffset=ds->readUInt32(((const uint32_t *)inBytes)[3])) |
| 1873 | ) { |
| 1874 | udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu\n", |
| 1875 | length); |
| 1876 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 1877 | return 0; |
| 1878 | } |
| 1879 | } |
| 1880 | |
| 1881 | if(length<0) { |
| 1882 | /* preflighting: iterate through algorithmic ranges */ |
| 1883 | offset=algNamesOffset; |
| 1884 | count=ds->readUInt32(*((const uint32_t *)(inBytes+offset))); |
| 1885 | offset+=4; |
| 1886 | |
| 1887 | for(i=0; i<count; ++i) { |
| 1888 | inRange=(const AlgorithmicRange *)(inBytes+offset); |
| 1889 | offset+=ds->readUInt16(inRange->size); |
| 1890 | } |
| 1891 | } else { |
| 1892 | /* swap data */ |
| 1893 | const uint16_t *p; |
| 1894 | uint16_t *q, *temp; |
| 1895 | |
| 1896 | int16_t tokens[512]; |
| 1897 | uint16_t tokenCount; |
| 1898 | |
| 1899 | uint8_t map[256], trailMap[256]; |
| 1900 | |
| 1901 | /* copy the data for inaccessible bytes */ |
| 1902 | if(inBytes!=outBytes) { |
| 1903 | uprv_memcpy(outBytes, inBytes, length); |
| 1904 | } |
| 1905 | |
| 1906 | /* the initial 4 offsets first */ |
| 1907 | tokenStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[0]); |
| 1908 | groupsOffset=ds->readUInt32(((const uint32_t *)inBytes)[1]); |
| 1909 | groupStringOffset=ds->readUInt32(((const uint32_t *)inBytes)[2]); |
| 1910 | ds->swapArray32(ds, inBytes, 16, outBytes, pErrorCode); |
| 1911 | |
| 1912 | /* |
| 1913 | * now the tokens table |
| 1914 | * it needs to be permutated along with the compressed name strings |
| 1915 | */ |
| 1916 | p=(const uint16_t *)(inBytes+16); |
| 1917 | q=(uint16_t *)(outBytes+16); |
| 1918 | |
| 1919 | /* read and swap the tokenCount */ |
| 1920 | tokenCount=ds->readUInt16(*p); |
| 1921 | ds->swapArray16(ds, p, 2, q, pErrorCode); |
| 1922 | ++p; |
| 1923 | ++q; |
| 1924 | |
| 1925 | /* read the first 512 tokens and make the token maps */ |
| 1926 | if(tokenCount<=512) { |
| 1927 | count=tokenCount; |
| 1928 | } else { |
| 1929 | count=512; |
| 1930 | } |
| 1931 | for(i=0; i<count; ++i) { |
| 1932 | tokens[i]=udata_readInt16(ds, p[i]); |
| 1933 | } |
| 1934 | for(; i<512; ++i) { |
| 1935 | tokens[i]=0; /* fill the rest of the tokens array if tokenCount<512 */ |
| 1936 | } |
| 1937 | makeTokenMap(ds, tokens, tokenCount, map, pErrorCode); |
| 1938 | makeTokenMap(ds, tokens+256, (uint16_t)(tokenCount>256 ? tokenCount-256 : 0), trailMap, pErrorCode); |
| 1939 | if(U_FAILURE(*pErrorCode)) { |
| 1940 | return 0; |
| 1941 | } |
| 1942 | |
| 1943 | /* |
| 1944 | * swap and permutate the tokens |
| 1945 | * go through a temporary array to support in-place swapping |
| 1946 | */ |
| 1947 | temp=(uint16_t *)uprv_malloc(tokenCount*2); |
| 1948 | if(temp==NULL) { |
| 1949 | udata_printError(ds, "out of memory swapping %u unames.icu tokens\n", |
| 1950 | tokenCount); |
| 1951 | *pErrorCode=U_MEMORY_ALLOCATION_ERROR; |
| 1952 | return 0; |
| 1953 | } |
| 1954 | |
| 1955 | /* swap and permutate single-/lead-byte tokens */ |
| 1956 | for(i=0; i<tokenCount && i<256; ++i) { |
| 1957 | ds->swapArray16(ds, p+i, 2, temp+map[i], pErrorCode); |
| 1958 | } |
| 1959 | |
| 1960 | /* swap and permutate trail-byte tokens */ |
| 1961 | for(; i<tokenCount; ++i) { |
| 1962 | ds->swapArray16(ds, p+i, 2, temp+(i&0xffffff00)+trailMap[i&0xff], pErrorCode); |
| 1963 | } |
| 1964 | |
| 1965 | /* copy the result into the output and free the temporary array */ |
| 1966 | uprv_memcpy(q, temp, tokenCount*2); |
| 1967 | uprv_free(temp); |
| 1968 | |
| 1969 | /* |
| 1970 | * swap the token strings but not a possible padding byte after |
| 1971 | * the terminating NUL of the last string |
| 1972 | */ |
| 1973 | udata_swapInvStringBlock(ds, inBytes+tokenStringOffset, (int32_t)(groupsOffset-tokenStringOffset), |
| 1974 | outBytes+tokenStringOffset, pErrorCode); |
| 1975 | if(U_FAILURE(*pErrorCode)) { |
| 1976 | udata_printError(ds, "uchar_swapNames(token strings) failed\n"); |
| 1977 | return 0; |
| 1978 | } |
| 1979 | |
| 1980 | /* swap the group table */ |
| 1981 | count=ds->readUInt16(*((const uint16_t *)(inBytes+groupsOffset))); |
| 1982 | ds->swapArray16(ds, inBytes+groupsOffset, (int32_t)((1+count*3)*2), |
| 1983 | outBytes+groupsOffset, pErrorCode); |
| 1984 | |
| 1985 | /* |
| 1986 | * swap the group strings |
| 1987 | * swap the string bytes but not the nibble-encoded string lengths |
| 1988 | */ |
| 1989 | if(ds->inCharset!=ds->outCharset) { |
| 1990 | uint16_t offsets[LINES_PER_GROUP+1], lengths[LINES_PER_GROUP+1]; |
| 1991 | |
| 1992 | const uint8_t *inStrings, *nextInStrings; |
| 1993 | uint8_t *outStrings; |
| 1994 | |
| 1995 | uint8_t c; |
| 1996 | |
| 1997 | inStrings=inBytes+groupStringOffset; |
| 1998 | outStrings=outBytes+groupStringOffset; |
| 1999 | |
| 2000 | stringsCount=algNamesOffset-groupStringOffset; |
| 2001 | |
| 2002 | /* iterate through string groups until only a few padding bytes are left */ |
| 2003 | while(stringsCount>32) { |
| 2004 | nextInStrings=expandGroupLengths(inStrings, offsets, lengths); |
| 2005 | |
| 2006 | /* move past the length bytes */ |
| 2007 | stringsCount-=(uint32_t)(nextInStrings-inStrings); |
| 2008 | outStrings+=nextInStrings-inStrings; |
| 2009 | inStrings=nextInStrings; |
| 2010 | |
| 2011 | count=offsets[31]+lengths[31]; /* total number of string bytes in this group */ |
| 2012 | stringsCount-=count; |
| 2013 | |
| 2014 | /* swap the string bytes using map[] and trailMap[] */ |
| 2015 | while(count>0) { |
| 2016 | c=*inStrings++; |
| 2017 | *outStrings++=map[c]; |
| 2018 | if(tokens[c]!=-2) { |
| 2019 | --count; |
| 2020 | } else { |
| 2021 | /* token lead byte: swap the trail byte, too */ |
| 2022 | *outStrings++=trailMap[*inStrings++]; |
| 2023 | count-=2; |
| 2024 | } |
| 2025 | } |
| 2026 | } |
| 2027 | } |
| 2028 | |
| 2029 | /* swap the algorithmic ranges */ |
| 2030 | offset=algNamesOffset; |
| 2031 | count=ds->readUInt32(*((const uint32_t *)(inBytes+offset))); |
| 2032 | ds->swapArray32(ds, inBytes+offset, 4, outBytes+offset, pErrorCode); |
| 2033 | offset+=4; |
| 2034 | |
| 2035 | for(i=0; i<count; ++i) { |
| 2036 | if(offset>(uint32_t)length) { |
| 2037 | udata_printError(ds, "uchar_swapNames(): too few bytes (%d after header) for unames.icu algorithmic range %u\n", |
| 2038 | length, i); |
| 2039 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 2040 | return 0; |
| 2041 | } |
| 2042 | |
| 2043 | inRange=(const AlgorithmicRange *)(inBytes+offset); |
| 2044 | outRange=(AlgorithmicRange *)(outBytes+offset); |
| 2045 | offset+=ds->readUInt16(inRange->size); |
| 2046 | |
| 2047 | ds->swapArray32(ds, inRange, 8, outRange, pErrorCode); |
| 2048 | ds->swapArray16(ds, &inRange->size, 2, &outRange->size, pErrorCode); |
| 2049 | switch(inRange->type) { |
| 2050 | case 0: |
| 2051 | /* swap prefix string */ |
| 2052 | ds->swapInvChars(ds, inRange+1, (int32_t)uprv_strlen((const char *)(inRange+1)), |
| 2053 | outRange+1, pErrorCode); |
| 2054 | if(U_FAILURE(*pErrorCode)) { |
| 2055 | udata_printError(ds, "uchar_swapNames(prefix string of algorithmic range %u) failed\n", |
| 2056 | i); |
| 2057 | return 0; |
| 2058 | } |
| 2059 | break; |
| 2060 | case 1: |
| 2061 | { |
| 2062 | /* swap factors and the prefix and factor strings */ |
| 2063 | uint32_t factorsCount; |
| 2064 | |
| 2065 | factorsCount=inRange->variant; |
| 2066 | p=(const uint16_t *)(inRange+1); |
| 2067 | q=(uint16_t *)(outRange+1); |
| 2068 | ds->swapArray16(ds, p, (int32_t)(factorsCount*2), q, pErrorCode); |
| 2069 | |
| 2070 | /* swap the strings, up to the last terminating NUL */ |
| 2071 | p+=factorsCount; |
| 2072 | q+=factorsCount; |
| 2073 | stringsCount=(uint32_t)((inBytes+offset)-(const uint8_t *)p); |
| 2074 | while(stringsCount>0 && ((const uint8_t *)p)[stringsCount-1]!=0) { |
| 2075 | --stringsCount; |
| 2076 | } |
| 2077 | ds->swapInvChars(ds, p, (int32_t)stringsCount, q, pErrorCode); |
| 2078 | } |
| 2079 | break; |
| 2080 | default: |
| 2081 | udata_printError(ds, "uchar_swapNames(): unknown type %u of algorithmic range %u\n", |
| 2082 | inRange->type, i); |
| 2083 | *pErrorCode=U_UNSUPPORTED_ERROR; |
| 2084 | return 0; |
| 2085 | } |
| 2086 | } |
| 2087 | } |
| 2088 | |
| 2089 | return headerSize+(int32_t)offset; |
| 2090 | } |
| 2091 | |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 2092 | /* |
| 2093 | * Hey, Emacs, please set the following: |
| 2094 | * |
| 2095 | * Local Variables: |
| 2096 | * indent-tabs-mode: nil |
| 2097 | * End: |
| 2098 | * |
| 2099 | */ |