| 1 | // © 2016 and later: Unicode, Inc. and others. | 
|---|
| 2 | // License & terms of use: http://www.unicode.org/copyright.html | 
|---|
| 3 | /* | 
|---|
| 4 | ******************************************************************************* | 
|---|
| 5 | *   Copyright (C) 2004-2016, International Business Machines | 
|---|
| 6 | *   Corporation and others.  All Rights Reserved. | 
|---|
| 7 | ******************************************************************************* | 
|---|
| 8 | *   file name:  ucol_sit.cpp | 
|---|
| 9 | *   encoding:   UTF-8 | 
|---|
| 10 | *   tab size:   8 (not used) | 
|---|
| 11 | *   indentation:4 | 
|---|
| 12 | * | 
|---|
| 13 | * Modification history | 
|---|
| 14 | * Date        Name      Comments | 
|---|
| 15 | * 03/12/2004  weiv      Creation | 
|---|
| 16 | */ | 
|---|
| 17 |  | 
|---|
| 18 | #include "unicode/ustring.h" | 
|---|
| 19 | #include "unicode/udata.h" | 
|---|
| 20 | #include "unicode/utf16.h" | 
|---|
| 21 | #include "utracimp.h" | 
|---|
| 22 | #include "ucol_imp.h" | 
|---|
| 23 | #include "cmemory.h" | 
|---|
| 24 | #include "cstring.h" | 
|---|
| 25 | #include "uresimp.h" | 
|---|
| 26 | #include "unicode/coll.h" | 
|---|
| 27 |  | 
|---|
| 28 | #ifdef UCOL_TRACE_SIT | 
|---|
| 29 | # include <stdio.h> | 
|---|
| 30 | #endif | 
|---|
| 31 |  | 
|---|
| 32 | #if !UCONFIG_NO_COLLATION | 
|---|
| 33 |  | 
|---|
| 34 | #include "unicode/tblcoll.h" | 
|---|
| 35 |  | 
|---|
| 36 | enum OptionsList { | 
|---|
| 37 | UCOL_SIT_LANGUAGE = 0, | 
|---|
| 38 | UCOL_SIT_SCRIPT   = 1, | 
|---|
| 39 | UCOL_SIT_REGION   = 2, | 
|---|
| 40 | UCOL_SIT_VARIANT  = 3, | 
|---|
| 41 | UCOL_SIT_KEYWORD  = 4, | 
|---|
| 42 | UCOL_SIT_PROVIDER = 5, | 
|---|
| 43 | UCOL_SIT_LOCELEMENT_MAX = UCOL_SIT_PROVIDER, /* the last element that's part of LocElements */ | 
|---|
| 44 |  | 
|---|
| 45 | UCOL_SIT_BCP47, | 
|---|
| 46 | UCOL_SIT_STRENGTH, | 
|---|
| 47 | UCOL_SIT_CASE_LEVEL, | 
|---|
| 48 | UCOL_SIT_CASE_FIRST, | 
|---|
| 49 | UCOL_SIT_NUMERIC_COLLATION, | 
|---|
| 50 | UCOL_SIT_ALTERNATE_HANDLING, | 
|---|
| 51 | UCOL_SIT_NORMALIZATION_MODE, | 
|---|
| 52 | UCOL_SIT_FRENCH_COLLATION, | 
|---|
| 53 | UCOL_SIT_HIRAGANA_QUATERNARY, | 
|---|
| 54 | UCOL_SIT_VARIABLE_TOP, | 
|---|
| 55 | UCOL_SIT_VARIABLE_TOP_VALUE, | 
|---|
| 56 | UCOL_SIT_ITEMS_COUNT | 
|---|
| 57 | }; | 
|---|
| 58 |  | 
|---|
| 59 | /* option starters chars. */ | 
|---|
| 60 | static const char alternateHArg     = 'A'; | 
|---|
| 61 | static const char variableTopValArg = 'B'; | 
|---|
| 62 | static const char caseFirstArg      = 'C'; | 
|---|
| 63 | static const char numericCollArg    = 'D'; | 
|---|
| 64 | static const char caseLevelArg      = 'E'; | 
|---|
| 65 | static const char frenchCollArg     = 'F'; | 
|---|
| 66 | static const char hiraganaQArg      = 'H'; | 
|---|
| 67 | static const char keywordArg        = 'K'; | 
|---|
| 68 | static const char languageArg       = 'L'; | 
|---|
| 69 | static const char normArg           = 'N'; | 
|---|
| 70 | static const char providerArg       = 'P'; | 
|---|
| 71 | static const char regionArg         = 'R'; | 
|---|
| 72 | static const char strengthArg       = 'S'; | 
|---|
| 73 | static const char variableTopArg    = 'T'; | 
|---|
| 74 | static const char variantArg        = 'V'; | 
|---|
| 75 | static const char RFC3066Arg        = 'X'; | 
|---|
| 76 | static const char scriptArg         = 'Z'; | 
|---|
| 77 |  | 
|---|
| 78 | static const char collationKeyword[]  = "@collation="; | 
|---|
| 79 | static const char providerKeyword[]  = "@sp="; | 
|---|
| 80 |  | 
|---|
| 81 |  | 
|---|
| 82 | static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1; | 
|---|
| 83 | static const int32_t locElementCapacity = 32; | 
|---|
| 84 | static const int32_t loc3066Capacity = 256; | 
|---|
| 85 | static const int32_t locProviderCapacity = 10; | 
|---|
| 86 | static const int32_t internalBufferSize = 512; | 
|---|
| 87 |  | 
|---|
| 88 | /* structure containing specification of a collator. Initialized | 
|---|
| 89 | * from a short string. Also used to construct a short string from a | 
|---|
| 90 | * collator instance | 
|---|
| 91 | */ | 
|---|
| 92 | struct CollatorSpec { | 
|---|
| 93 | char locElements[locElementCount][locElementCapacity]; | 
|---|
| 94 | char locale[loc3066Capacity]; | 
|---|
| 95 | char provider[locProviderCapacity]; | 
|---|
| 96 | UColAttributeValue options[UCOL_ATTRIBUTE_COUNT]; | 
|---|
| 97 | uint32_t variableTopValue; | 
|---|
| 98 | UChar variableTopString[locElementCapacity]; | 
|---|
| 99 | int32_t variableTopStringLen; | 
|---|
| 100 | UBool variableTopSet; | 
|---|
| 101 | struct { | 
|---|
| 102 | const char *start; | 
|---|
| 103 | int32_t len; | 
|---|
| 104 | } entries[UCOL_SIT_ITEMS_COUNT]; | 
|---|
| 105 | }; | 
|---|
| 106 |  | 
|---|
| 107 |  | 
|---|
| 108 | /* structure for converting between character attribute | 
|---|
| 109 | * representation and real collation attribute value. | 
|---|
| 110 | */ | 
|---|
| 111 | struct AttributeConversion { | 
|---|
| 112 | char letter; | 
|---|
| 113 | UColAttributeValue value; | 
|---|
| 114 | }; | 
|---|
| 115 |  | 
|---|
| 116 | static const AttributeConversion conversions[12] = { | 
|---|
| 117 | { '1', UCOL_PRIMARY }, | 
|---|
| 118 | { '2', UCOL_SECONDARY }, | 
|---|
| 119 | { '3', UCOL_TERTIARY }, | 
|---|
| 120 | { '4', UCOL_QUATERNARY }, | 
|---|
| 121 | { 'D', UCOL_DEFAULT }, | 
|---|
| 122 | { 'I', UCOL_IDENTICAL }, | 
|---|
| 123 | { 'L', UCOL_LOWER_FIRST }, | 
|---|
| 124 | { 'N', UCOL_NON_IGNORABLE }, | 
|---|
| 125 | { 'O', UCOL_ON }, | 
|---|
| 126 | { 'S', UCOL_SHIFTED }, | 
|---|
| 127 | { 'U', UCOL_UPPER_FIRST }, | 
|---|
| 128 | { 'X', UCOL_OFF } | 
|---|
| 129 | }; | 
|---|
| 130 |  | 
|---|
| 131 |  | 
|---|
| 132 | static UColAttributeValue | 
|---|
| 133 | ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) { | 
|---|
| 134 | uint32_t i = 0; | 
|---|
| 135 | for(i = 0; i < UPRV_LENGTHOF(conversions); i++) { | 
|---|
| 136 | if(conversions[i].letter == letter) { | 
|---|
| 137 | return conversions[i].value; | 
|---|
| 138 | } | 
|---|
| 139 | } | 
|---|
| 140 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 141 | #ifdef UCOL_TRACE_SIT | 
|---|
| 142 | fprintf(stderr, "%s:%d: unknown letter %c: %s\n", __FILE__, __LINE__, letter, u_errorName(*status)); | 
|---|
| 143 | #endif | 
|---|
| 144 | return UCOL_DEFAULT; | 
|---|
| 145 | } | 
|---|
| 146 |  | 
|---|
| 147 | /* function prototype for functions used to parse a short string */ | 
|---|
| 148 | U_CDECL_BEGIN | 
|---|
| 149 | typedef const char* U_CALLCONV | 
|---|
| 150 | ActionFunction(CollatorSpec *spec, uint32_t value1, const char* string, | 
|---|
| 151 | UErrorCode *status); | 
|---|
| 152 | U_CDECL_END | 
|---|
| 153 |  | 
|---|
| 154 | U_CDECL_BEGIN | 
|---|
| 155 | static const char* U_CALLCONV | 
|---|
| 156 | _processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string, | 
|---|
| 157 | UErrorCode *status) | 
|---|
| 158 | { | 
|---|
| 159 | int32_t len = 0; | 
|---|
| 160 | do { | 
|---|
| 161 | if(value == UCOL_SIT_LANGUAGE || value == UCOL_SIT_KEYWORD || value == UCOL_SIT_PROVIDER) { | 
|---|
| 162 | spec->locElements[value][len++] = uprv_tolower(*string); | 
|---|
| 163 | } else { | 
|---|
| 164 | spec->locElements[value][len++] = *string; | 
|---|
| 165 | } | 
|---|
| 166 | } while(*(++string) != '_' && *string && len < locElementCapacity); | 
|---|
| 167 | if(len >= locElementCapacity) { | 
|---|
| 168 | *status = U_BUFFER_OVERFLOW_ERROR; | 
|---|
| 169 | return string; | 
|---|
| 170 | } | 
|---|
| 171 | // don't skip the underscore at the end | 
|---|
| 172 | return string; | 
|---|
| 173 | } | 
|---|
| 174 | U_CDECL_END | 
|---|
| 175 |  | 
|---|
| 176 | U_CDECL_BEGIN | 
|---|
| 177 | static const char* U_CALLCONV | 
|---|
| 178 | _processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string, | 
|---|
| 179 | UErrorCode *status) | 
|---|
| 180 | { | 
|---|
| 181 | char terminator = *string; | 
|---|
| 182 | string++; | 
|---|
| 183 | const char *end = uprv_strchr(string+1, terminator); | 
|---|
| 184 | if(end == NULL || end - string >= loc3066Capacity) { | 
|---|
| 185 | *status = U_BUFFER_OVERFLOW_ERROR; | 
|---|
| 186 | return string; | 
|---|
| 187 | } else { | 
|---|
| 188 | uprv_strncpy(spec->locale, string, end-string); | 
|---|
| 189 | return end+1; | 
|---|
| 190 | } | 
|---|
| 191 | } | 
|---|
| 192 |  | 
|---|
| 193 | U_CDECL_END | 
|---|
| 194 |  | 
|---|
| 195 | U_CDECL_BEGIN | 
|---|
| 196 | static const char* U_CALLCONV | 
|---|
| 197 | _processCollatorOption(CollatorSpec *spec, uint32_t option, const char* string, | 
|---|
| 198 | UErrorCode *status) | 
|---|
| 199 | { | 
|---|
| 200 | spec->options[option] = ucol_sit_letterToAttributeValue(*string, status); | 
|---|
| 201 | if((*(++string) != '_' && *string) || U_FAILURE(*status)) { | 
|---|
| 202 | #ifdef UCOL_TRACE_SIT | 
|---|
| 203 | fprintf(stderr, "%s:%d: unknown collator option at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(*status)); | 
|---|
| 204 | #endif | 
|---|
| 205 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 206 | } | 
|---|
| 207 | return string; | 
|---|
| 208 | } | 
|---|
| 209 | U_CDECL_END | 
|---|
| 210 |  | 
|---|
| 211 |  | 
|---|
| 212 | static UChar | 
|---|
| 213 | readHexCodeUnit(const char **string, UErrorCode *status) | 
|---|
| 214 | { | 
|---|
| 215 | UChar result = 0; | 
|---|
| 216 | int32_t value = 0; | 
|---|
| 217 | char c; | 
|---|
| 218 | int32_t noDigits = 0; | 
|---|
| 219 | while((c = **string) != 0 && noDigits < 4) { | 
|---|
| 220 | if( c >= '0' && c <= '9') { | 
|---|
| 221 | value = c - '0'; | 
|---|
| 222 | } else if ( c >= 'a' && c <= 'f') { | 
|---|
| 223 | value = c - 'a' + 10; | 
|---|
| 224 | } else if ( c >= 'A' && c <= 'F') { | 
|---|
| 225 | value = c - 'A' + 10; | 
|---|
| 226 | } else { | 
|---|
| 227 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 228 | #ifdef UCOL_TRACE_SIT | 
|---|
| 229 | fprintf(stderr, "%s:%d: Bad hex char at '%s': %s\n", __FILE__, __LINE__, *string, u_errorName(*status)); | 
|---|
| 230 | #endif | 
|---|
| 231 | return 0; | 
|---|
| 232 | } | 
|---|
| 233 | result = (result << 4) | (UChar)value; | 
|---|
| 234 | noDigits++; | 
|---|
| 235 | (*string)++; | 
|---|
| 236 | } | 
|---|
| 237 | // if the string was terminated before we read 4 digits, set an error | 
|---|
| 238 | if(noDigits < 4) { | 
|---|
| 239 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 240 | #ifdef UCOL_TRACE_SIT | 
|---|
| 241 | fprintf(stderr, "%s:%d: Short (only %d digits, wanted 4) at '%s': %s\n", __FILE__, __LINE__, noDigits,*string, u_errorName(*status)); | 
|---|
| 242 | #endif | 
|---|
| 243 | } | 
|---|
| 244 | return result; | 
|---|
| 245 | } | 
|---|
| 246 |  | 
|---|
| 247 | U_CDECL_BEGIN | 
|---|
| 248 | static const char* U_CALLCONV | 
|---|
| 249 | _processVariableTop(CollatorSpec *spec, uint32_t value1, const char* string, UErrorCode *status) | 
|---|
| 250 | { | 
|---|
| 251 | // get four digits | 
|---|
| 252 | int32_t i = 0; | 
|---|
| 253 | if(!value1) { | 
|---|
| 254 | while(U_SUCCESS(*status) && i < locElementCapacity && *string != 0 && *string != '_') { | 
|---|
| 255 | spec->variableTopString[i++] = readHexCodeUnit(&string, status); | 
|---|
| 256 | } | 
|---|
| 257 | spec->variableTopStringLen = i; | 
|---|
| 258 | if(i == locElementCapacity && *string != 0 && *string != '_') { | 
|---|
| 259 | *status = U_BUFFER_OVERFLOW_ERROR; | 
|---|
| 260 | } | 
|---|
| 261 | } else { | 
|---|
| 262 | spec->variableTopValue = readHexCodeUnit(&string, status); | 
|---|
| 263 | } | 
|---|
| 264 | if(U_SUCCESS(*status)) { | 
|---|
| 265 | spec->variableTopSet = TRUE; | 
|---|
| 266 | } | 
|---|
| 267 | return string; | 
|---|
| 268 | } | 
|---|
| 269 | U_CDECL_END | 
|---|
| 270 |  | 
|---|
| 271 |  | 
|---|
| 272 | /* Table for parsing short strings */ | 
|---|
| 273 | struct ShortStringOptions { | 
|---|
| 274 | char optionStart; | 
|---|
| 275 | ActionFunction *action; | 
|---|
| 276 | uint32_t attr; | 
|---|
| 277 | }; | 
|---|
| 278 |  | 
|---|
| 279 | static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] = | 
|---|
| 280 | { | 
|---|
| 281 | /* 10 ALTERNATE_HANDLING */   {alternateHArg,     _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate  N, S, D | 
|---|
| 282 | /* 15 VARIABLE_TOP_VALUE */   {variableTopValArg, _processVariableTop,    1 }, | 
|---|
| 283 | /* 08 CASE_FIRST */           {caseFirstArg,      _processCollatorOption, UCOL_CASE_FIRST }, // case first L, U, X, D | 
|---|
| 284 | /* 09 NUMERIC_COLLATION */    {numericCollArg,    _processCollatorOption, UCOL_NUMERIC_COLLATION }, // codan      O, X, D | 
|---|
| 285 | /* 07 CASE_LEVEL */           {caseLevelArg,      _processCollatorOption, UCOL_CASE_LEVEL }, // case level O, X, D | 
|---|
| 286 | /* 12 FRENCH_COLLATION */     {frenchCollArg,     _processCollatorOption, UCOL_FRENCH_COLLATION }, // french     O, X, D | 
|---|
| 287 | /* 13 HIRAGANA_QUATERNARY] */ {hiraganaQArg,      _processCollatorOption, UCOL_HIRAGANA_QUATERNARY_MODE }, // hiragana   O, X, D | 
|---|
| 288 | /* 04 KEYWORD */              {keywordArg,        _processLocaleElement,  UCOL_SIT_KEYWORD }, // keyword | 
|---|
| 289 | /* 00 LANGUAGE */             {languageArg,       _processLocaleElement,  UCOL_SIT_LANGUAGE }, // language | 
|---|
| 290 | /* 11 NORMALIZATION_MODE */   {normArg,           _processCollatorOption, UCOL_NORMALIZATION_MODE }, // norm       O, X, D | 
|---|
| 291 | /* 02 REGION */               {regionArg,         _processLocaleElement,  UCOL_SIT_REGION }, // region | 
|---|
| 292 | /* 06 STRENGTH */             {strengthArg,       _processCollatorOption, UCOL_STRENGTH }, // strength   1, 2, 3, 4, I, D | 
|---|
| 293 | /* 14 VARIABLE_TOP */         {variableTopArg,    _processVariableTop,    0 }, | 
|---|
| 294 | /* 03 VARIANT */              {variantArg,        _processLocaleElement,  UCOL_SIT_VARIANT }, // variant | 
|---|
| 295 | /* 05 RFC3066BIS */           {RFC3066Arg,        _processRFC3066Locale,  0 }, // rfc3066bis locale name | 
|---|
| 296 | /* 01 SCRIPT */               {scriptArg,         _processLocaleElement,  UCOL_SIT_SCRIPT },  // script | 
|---|
| 297 | /*    PROVIDER */             {providerArg,       _processLocaleElement, UCOL_SIT_PROVIDER } | 
|---|
| 298 | }; | 
|---|
| 299 |  | 
|---|
| 300 |  | 
|---|
| 301 | static | 
|---|
| 302 | const char* ucol_sit_readOption(const char *start, CollatorSpec *spec, | 
|---|
| 303 | UErrorCode *status) | 
|---|
| 304 | { | 
|---|
| 305 | int32_t i = 0; | 
|---|
| 306 |  | 
|---|
| 307 | for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) { | 
|---|
| 308 | if(*start == options[i].optionStart) { | 
|---|
| 309 | spec->entries[i].start = start; | 
|---|
| 310 | const char* end = options[i].action(spec, options[i].attr, start+1, status); | 
|---|
| 311 | spec->entries[i].len = (int32_t)(end - start); | 
|---|
| 312 | return end; | 
|---|
| 313 | } | 
|---|
| 314 | } | 
|---|
| 315 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 316 | #ifdef UCOL_TRACE_SIT | 
|---|
| 317 | fprintf(stderr, "%s:%d: Unknown option at '%s': %s\n", __FILE__, __LINE__, start, u_errorName(*status)); | 
|---|
| 318 | #endif | 
|---|
| 319 | return start; | 
|---|
| 320 | } | 
|---|
| 321 |  | 
|---|
| 322 | static | 
|---|
| 323 | void ucol_sit_initCollatorSpecs(CollatorSpec *spec) | 
|---|
| 324 | { | 
|---|
| 325 | // reset everything | 
|---|
| 326 | uprv_memset(spec, 0, sizeof(CollatorSpec)); | 
|---|
| 327 | // set collation options to default | 
|---|
| 328 | int32_t i = 0; | 
|---|
| 329 | for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) { | 
|---|
| 330 | spec->options[i] = UCOL_DEFAULT; | 
|---|
| 331 | } | 
|---|
| 332 | } | 
|---|
| 333 |  | 
|---|
| 334 | static const char* | 
|---|
| 335 | ucol_sit_readSpecs(CollatorSpec *s, const char *string, | 
|---|
| 336 | UParseError *parseError, UErrorCode *status) | 
|---|
| 337 | { | 
|---|
| 338 | const char *definition = string; | 
|---|
| 339 | while(U_SUCCESS(*status) && *string) { | 
|---|
| 340 | string = ucol_sit_readOption(string, s, status); | 
|---|
| 341 | // advance over '_' | 
|---|
| 342 | while(*string && *string == '_') { | 
|---|
| 343 | string++; | 
|---|
| 344 | } | 
|---|
| 345 | } | 
|---|
| 346 | if(U_FAILURE(*status)) { | 
|---|
| 347 | parseError->offset = (int32_t)(string - definition); | 
|---|
| 348 | } | 
|---|
| 349 | return string; | 
|---|
| 350 | } | 
|---|
| 351 |  | 
|---|
| 352 | static | 
|---|
| 353 | int32_t ucol_sit_dumpSpecs(CollatorSpec *s, char *destination, int32_t capacity, UErrorCode *status) | 
|---|
| 354 | { | 
|---|
| 355 | int32_t i = 0, j = 0; | 
|---|
| 356 | int32_t len = 0; | 
|---|
| 357 | char optName; | 
|---|
| 358 | if(U_SUCCESS(*status)) { | 
|---|
| 359 | for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) { | 
|---|
| 360 | if(s->entries[i].start) { | 
|---|
| 361 | if(len) { | 
|---|
| 362 | if(len < capacity) { | 
|---|
| 363 | uprv_strcat(destination, "_"); | 
|---|
| 364 | } | 
|---|
| 365 | len++; | 
|---|
| 366 | } | 
|---|
| 367 | optName = *(s->entries[i].start); | 
|---|
| 368 | if(optName == languageArg || optName == regionArg || optName == variantArg || optName == keywordArg) { | 
|---|
| 369 | for(j = 0; j < s->entries[i].len; j++) { | 
|---|
| 370 | if(len + j < capacity) { | 
|---|
| 371 | destination[len+j] = uprv_toupper(*(s->entries[i].start+j)); | 
|---|
| 372 | } | 
|---|
| 373 | } | 
|---|
| 374 | len += s->entries[i].len; | 
|---|
| 375 | } else { | 
|---|
| 376 | len += s->entries[i].len; | 
|---|
| 377 | if(len < capacity) { | 
|---|
| 378 | uprv_strncat(destination,s->entries[i].start, s->entries[i].len); | 
|---|
| 379 | } | 
|---|
| 380 | } | 
|---|
| 381 | } | 
|---|
| 382 | } | 
|---|
| 383 | return len; | 
|---|
| 384 | } else { | 
|---|
| 385 | return 0; | 
|---|
| 386 | } | 
|---|
| 387 | } | 
|---|
| 388 |  | 
|---|
| 389 | static void | 
|---|
| 390 | ucol_sit_calculateWholeLocale(CollatorSpec *s) { | 
|---|
| 391 | // put the locale together, unless we have a done | 
|---|
| 392 | // locale | 
|---|
| 393 | if(s->locale[0] == 0) { | 
|---|
| 394 | // first the language | 
|---|
| 395 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_LANGUAGE]); | 
|---|
| 396 | // then the script, if present | 
|---|
| 397 | if(*(s->locElements[UCOL_SIT_SCRIPT])) { | 
|---|
| 398 | uprv_strcat(s->locale, "_"); | 
|---|
| 399 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_SCRIPT]); | 
|---|
| 400 | } | 
|---|
| 401 | // then the region, if present | 
|---|
| 402 | if(*(s->locElements[UCOL_SIT_REGION])) { | 
|---|
| 403 | uprv_strcat(s->locale, "_"); | 
|---|
| 404 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_REGION]); | 
|---|
| 405 | } else if(*(s->locElements[UCOL_SIT_VARIANT])) { // if there is a variant, we need an underscore | 
|---|
| 406 | uprv_strcat(s->locale, "_"); | 
|---|
| 407 | } | 
|---|
| 408 | // add variant, if there | 
|---|
| 409 | if(*(s->locElements[UCOL_SIT_VARIANT])) { | 
|---|
| 410 | uprv_strcat(s->locale, "_"); | 
|---|
| 411 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_VARIANT]); | 
|---|
| 412 | } | 
|---|
| 413 |  | 
|---|
| 414 | // if there is a collation keyword, add that too | 
|---|
| 415 | if(*(s->locElements[UCOL_SIT_KEYWORD])) { | 
|---|
| 416 | uprv_strcat(s->locale, collationKeyword); | 
|---|
| 417 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_KEYWORD]); | 
|---|
| 418 | } | 
|---|
| 419 |  | 
|---|
| 420 | // if there is a provider keyword, add that too | 
|---|
| 421 | if(*(s->locElements[UCOL_SIT_PROVIDER])) { | 
|---|
| 422 | uprv_strcat(s->locale, providerKeyword); | 
|---|
| 423 | uprv_strcat(s->locale, s->locElements[UCOL_SIT_PROVIDER]); | 
|---|
| 424 | } | 
|---|
| 425 | } | 
|---|
| 426 | } | 
|---|
| 427 |  | 
|---|
| 428 |  | 
|---|
| 429 | U_CAPI void U_EXPORT2 | 
|---|
| 430 | ucol_prepareShortStringOpen( const char *definition, | 
|---|
| 431 | UBool, | 
|---|
| 432 | UParseError *parseError, | 
|---|
| 433 | UErrorCode *status) | 
|---|
| 434 | { | 
|---|
| 435 | if(U_FAILURE(*status)) return; | 
|---|
| 436 |  | 
|---|
| 437 | UParseError internalParseError; | 
|---|
| 438 |  | 
|---|
| 439 | if(!parseError) { | 
|---|
| 440 | parseError = &internalParseError; | 
|---|
| 441 | } | 
|---|
| 442 | parseError->line = 0; | 
|---|
| 443 | parseError->offset = 0; | 
|---|
| 444 | parseError->preContext[0] = 0; | 
|---|
| 445 | parseError->postContext[0] = 0; | 
|---|
| 446 |  | 
|---|
| 447 |  | 
|---|
| 448 | // first we want to pick stuff out of short string. | 
|---|
| 449 | // we'll end up with an UCA version, locale and a bunch of | 
|---|
| 450 | // settings | 
|---|
| 451 |  | 
|---|
| 452 | // analyse the string in order to get everything we need. | 
|---|
| 453 | CollatorSpec s; | 
|---|
| 454 | ucol_sit_initCollatorSpecs(&s); | 
|---|
| 455 | ucol_sit_readSpecs(&s, definition, parseError, status); | 
|---|
| 456 | ucol_sit_calculateWholeLocale(&s); | 
|---|
| 457 |  | 
|---|
| 458 | char buffer[internalBufferSize]; | 
|---|
| 459 | uprv_memset(buffer, 0, internalBufferSize); | 
|---|
| 460 | uloc_canonicalize(s.locale, buffer, internalBufferSize, status); | 
|---|
| 461 |  | 
|---|
| 462 | UResourceBundle *b = ures_open(U_ICUDATA_COLL, buffer, status); | 
|---|
| 463 | /* we try to find stuff from keyword */ | 
|---|
| 464 | UResourceBundle *collations = ures_getByKey(b, "collations", NULL, status); | 
|---|
| 465 | UResourceBundle *collElem = NULL; | 
|---|
| 466 | char keyBuffer[256]; | 
|---|
| 467 | // if there is a keyword, we pick it up and try to get elements | 
|---|
| 468 | int32_t keyLen = uloc_getKeywordValue(buffer, "collation", keyBuffer, sizeof(keyBuffer), status); | 
|---|
| 469 | // Treat too long a value as no keyword. | 
|---|
| 470 | if(keyLen >= (int32_t)sizeof(keyBuffer)) { | 
|---|
| 471 | keyLen = 0; | 
|---|
| 472 | *status = U_ZERO_ERROR; | 
|---|
| 473 | } | 
|---|
| 474 | if(keyLen == 0) { | 
|---|
| 475 | // no keyword | 
|---|
| 476 | // we try to find the default setting, which will give us the keyword value | 
|---|
| 477 | UResourceBundle *defaultColl = ures_getByKeyWithFallback(collations, "default", NULL, status); | 
|---|
| 478 | if(U_SUCCESS(*status)) { | 
|---|
| 479 | int32_t defaultKeyLen = 0; | 
|---|
| 480 | const UChar *defaultKey = ures_getString(defaultColl, &defaultKeyLen, status); | 
|---|
| 481 | u_UCharsToChars(defaultKey, keyBuffer, defaultKeyLen); | 
|---|
| 482 | keyBuffer[defaultKeyLen] = 0; | 
|---|
| 483 | } else { | 
|---|
| 484 | *status = U_INTERNAL_PROGRAM_ERROR; | 
|---|
| 485 | return; | 
|---|
| 486 | } | 
|---|
| 487 | ures_close(defaultColl); | 
|---|
| 488 | } | 
|---|
| 489 | collElem = ures_getByKeyWithFallback(collations, keyBuffer, collElem, status); | 
|---|
| 490 | ures_close(collElem); | 
|---|
| 491 | ures_close(collations); | 
|---|
| 492 | ures_close(b); | 
|---|
| 493 | } | 
|---|
| 494 |  | 
|---|
| 495 |  | 
|---|
| 496 | U_CAPI UCollator* U_EXPORT2 | 
|---|
| 497 | ucol_openFromShortString( const char *definition, | 
|---|
| 498 | UBool forceDefaults, | 
|---|
| 499 | UParseError *parseError, | 
|---|
| 500 | UErrorCode *status) | 
|---|
| 501 | { | 
|---|
| 502 | UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING); | 
|---|
| 503 | UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition); | 
|---|
| 504 |  | 
|---|
| 505 | if(U_FAILURE(*status)) return 0; | 
|---|
| 506 |  | 
|---|
| 507 | UParseError internalParseError; | 
|---|
| 508 |  | 
|---|
| 509 | if(!parseError) { | 
|---|
| 510 | parseError = &internalParseError; | 
|---|
| 511 | } | 
|---|
| 512 | parseError->line = 0; | 
|---|
| 513 | parseError->offset = 0; | 
|---|
| 514 | parseError->preContext[0] = 0; | 
|---|
| 515 | parseError->postContext[0] = 0; | 
|---|
| 516 |  | 
|---|
| 517 |  | 
|---|
| 518 | // first we want to pick stuff out of short string. | 
|---|
| 519 | // we'll end up with an UCA version, locale and a bunch of | 
|---|
| 520 | // settings | 
|---|
| 521 |  | 
|---|
| 522 | // analyse the string in order to get everything we need. | 
|---|
| 523 | const char *string = definition; | 
|---|
| 524 | CollatorSpec s; | 
|---|
| 525 | ucol_sit_initCollatorSpecs(&s); | 
|---|
| 526 | string = ucol_sit_readSpecs(&s, definition, parseError, status); | 
|---|
| 527 | ucol_sit_calculateWholeLocale(&s); | 
|---|
| 528 |  | 
|---|
| 529 | char buffer[internalBufferSize]; | 
|---|
| 530 | uprv_memset(buffer, 0, internalBufferSize); | 
|---|
| 531 | uloc_canonicalize(s.locale, buffer, internalBufferSize, status); | 
|---|
| 532 |  | 
|---|
| 533 | UCollator *result = ucol_open(buffer, status); | 
|---|
| 534 | int32_t i = 0; | 
|---|
| 535 |  | 
|---|
| 536 | for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) { | 
|---|
| 537 | if(s.options[i] != UCOL_DEFAULT) { | 
|---|
| 538 | if(forceDefaults || ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) { | 
|---|
| 539 | ucol_setAttribute(result, (UColAttribute)i, s.options[i], status); | 
|---|
| 540 | } | 
|---|
| 541 |  | 
|---|
| 542 | if(U_FAILURE(*status)) { | 
|---|
| 543 | parseError->offset = (int32_t)(string - definition); | 
|---|
| 544 | ucol_close(result); | 
|---|
| 545 | return NULL; | 
|---|
| 546 | } | 
|---|
| 547 |  | 
|---|
| 548 | } | 
|---|
| 549 | } | 
|---|
| 550 | if(s.variableTopSet) { | 
|---|
| 551 | if(s.variableTopString[0]) { | 
|---|
| 552 | ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status); | 
|---|
| 553 | } else { // we set by value, using 'B' | 
|---|
| 554 | ucol_restoreVariableTop(result, s.variableTopValue, status); | 
|---|
| 555 | } | 
|---|
| 556 | } | 
|---|
| 557 |  | 
|---|
| 558 |  | 
|---|
| 559 | if(U_FAILURE(*status)) { // here it can only be a bogus value | 
|---|
| 560 | ucol_close(result); | 
|---|
| 561 | result = NULL; | 
|---|
| 562 | } | 
|---|
| 563 |  | 
|---|
| 564 | UTRACE_EXIT_PTR_STATUS(result, *status); | 
|---|
| 565 | return result; | 
|---|
| 566 | } | 
|---|
| 567 |  | 
|---|
| 568 |  | 
|---|
| 569 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 570 | ucol_getShortDefinitionString(const UCollator *coll, | 
|---|
| 571 | const char *locale, | 
|---|
| 572 | char *dst, | 
|---|
| 573 | int32_t capacity, | 
|---|
| 574 | UErrorCode *status) | 
|---|
| 575 | { | 
|---|
| 576 | if(U_FAILURE(*status)) return 0; | 
|---|
| 577 | if(coll == NULL) { | 
|---|
| 578 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 579 | return 0; | 
|---|
| 580 | } | 
|---|
| 581 | return ((icu::Collator*)coll)->internalGetShortDefinitionString(locale,dst,capacity,*status); | 
|---|
| 582 | } | 
|---|
| 583 |  | 
|---|
| 584 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 585 | ucol_normalizeShortDefinitionString(const char *definition, | 
|---|
| 586 | char *destination, | 
|---|
| 587 | int32_t capacity, | 
|---|
| 588 | UParseError *parseError, | 
|---|
| 589 | UErrorCode *status) | 
|---|
| 590 | { | 
|---|
| 591 |  | 
|---|
| 592 | if(U_FAILURE(*status)) { | 
|---|
| 593 | return 0; | 
|---|
| 594 | } | 
|---|
| 595 |  | 
|---|
| 596 | if(destination) { | 
|---|
| 597 | uprv_memset(destination, 0, capacity*sizeof(char)); | 
|---|
| 598 | } | 
|---|
| 599 |  | 
|---|
| 600 | UParseError pe; | 
|---|
| 601 | if(!parseError) { | 
|---|
| 602 | parseError = &pe; | 
|---|
| 603 | } | 
|---|
| 604 |  | 
|---|
| 605 | // validate | 
|---|
| 606 | CollatorSpec s; | 
|---|
| 607 | ucol_sit_initCollatorSpecs(&s); | 
|---|
| 608 | ucol_sit_readSpecs(&s, definition, parseError, status); | 
|---|
| 609 | return ucol_sit_dumpSpecs(&s, destination, capacity, status); | 
|---|
| 610 | } | 
|---|
| 611 |  | 
|---|
| 612 | /** | 
|---|
| 613 | * Get a set containing the contractions defined by the collator. The set includes | 
|---|
| 614 | * both the UCA contractions and the contractions defined by the collator | 
|---|
| 615 | * @param coll collator | 
|---|
| 616 | * @param conts the set to hold the result | 
|---|
| 617 | * @param status to hold the error code | 
|---|
| 618 | * @return the size of the contraction set | 
|---|
| 619 | */ | 
|---|
| 620 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 621 | ucol_getContractions( const UCollator *coll, | 
|---|
| 622 | USet *contractions, | 
|---|
| 623 | UErrorCode *status) | 
|---|
| 624 | { | 
|---|
| 625 | ucol_getContractionsAndExpansions(coll, contractions, NULL, FALSE, status); | 
|---|
| 626 | return uset_getItemCount(contractions); | 
|---|
| 627 | } | 
|---|
| 628 |  | 
|---|
| 629 | /** | 
|---|
| 630 | * Get a set containing the expansions defined by the collator. The set includes | 
|---|
| 631 | * both the UCA expansions and the expansions defined by the tailoring | 
|---|
| 632 | * @param coll collator | 
|---|
| 633 | * @param conts the set to hold the result | 
|---|
| 634 | * @param addPrefixes add the prefix contextual elements to contractions | 
|---|
| 635 | * @param status to hold the error code | 
|---|
| 636 | * | 
|---|
| 637 | * @draft ICU 3.4 | 
|---|
| 638 | */ | 
|---|
| 639 | U_CAPI void U_EXPORT2 | 
|---|
| 640 | ucol_getContractionsAndExpansions( const UCollator *coll, | 
|---|
| 641 | USet *contractions, | 
|---|
| 642 | USet *expansions, | 
|---|
| 643 | UBool addPrefixes, | 
|---|
| 644 | UErrorCode *status) | 
|---|
| 645 | { | 
|---|
| 646 | if(U_FAILURE(*status)) { | 
|---|
| 647 | return; | 
|---|
| 648 | } | 
|---|
| 649 | if(coll == NULL) { | 
|---|
| 650 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 651 | return; | 
|---|
| 652 | } | 
|---|
| 653 | const icu::RuleBasedCollator *rbc = icu::RuleBasedCollator::rbcFromUCollator(coll); | 
|---|
| 654 | if(rbc == NULL) { | 
|---|
| 655 | *status = U_UNSUPPORTED_ERROR; | 
|---|
| 656 | return; | 
|---|
| 657 | } | 
|---|
| 658 | rbc->internalGetContractionsAndExpansions( | 
|---|
| 659 | icu::UnicodeSet::fromUSet(contractions), | 
|---|
| 660 | icu::UnicodeSet::fromUSet(expansions), | 
|---|
| 661 | addPrefixes, *status); | 
|---|
| 662 | } | 
|---|
| 663 | #endif | 
|---|
| 664 |  | 
|---|