1 /* 2 ******************************************************************************* 3 * 4 * Copyright (C) 2002-2010, International Business Machines 5 * Corporation and others. All Rights Reserved. 6 * 7 ******************************************************************************* 8 * file name: uprops.h 9 * encoding: US-ASCII 10 * tab size: 8 (not used) 11 * indentation:4 12 * 13 * created on: 2002feb24 14 * created by: Markus W. Scherer 15 * 16 * Constants for mostly non-core Unicode character properties 17 * stored in uprops.icu. 18 */ 19 20 #ifndef __UPROPS_H__ 21 #define __UPROPS_H__ 22 23 #include "unicode/utypes.h" 24 #include "unicode/uset.h" 25 #include "uset_imp.h" 26 #include "udataswp.h" 27 28 /* indexes[] entries */ 29 enum { 30 UPROPS_PROPS32_INDEX, 31 UPROPS_EXCEPTIONS_INDEX, 32 UPROPS_EXCEPTIONS_TOP_INDEX, 33 34 UPROPS_ADDITIONAL_TRIE_INDEX, 35 UPROPS_ADDITIONAL_VECTORS_INDEX, 36 UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX, 37 38 UPROPS_RESERVED_INDEX, /* 6 */ 39 40 /* maximum values for code values in vector word 0 */ 41 UPROPS_MAX_VALUES_INDEX=10, 42 /* maximum values for code values in vector word 2 */ 43 UPROPS_MAX_VALUES_2_INDEX, 44 45 UPROPS_INDEX_COUNT=16 46 }; 47 48 /* definitions for the main properties words */ 49 enum { 50 /* general category shift==0 0 (5 bits) */ 51 /* reserved 5 (1 bit) */ 52 UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6 /* 6 (10 bits) */ 53 }; 54 55 #define GET_CATEGORY(props) ((props)&0x1f) 56 #define CAT_MASK(props) U_MASK(GET_CATEGORY(props)) 57 58 #define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT) 59 60 /* constants for the storage form of numeric types and values */ 61 enum { 62 UPROPS_NTV_NONE=0, 63 UPROPS_NTV_DECIMAL_START=1, 64 UPROPS_NTV_DIGIT_START=11, 65 UPROPS_NTV_NUMERIC_START=21, 66 UPROPS_NTV_FRACTION_START=0xb0, 67 UPROPS_NTV_LARGE_START=0x1e0, 68 UPROPS_NTV_RESERVED_START=0x300, 69 70 UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1 71 }; 72 73 #define UPROPS_NTV_GET_TYPE(ntv) \ 74 ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \ 75 (ntv<UPROPS_NTV_DIGIT_START) ? U_NT_DECIMAL : \ 76 (ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \ 77 U_NT_NUMERIC) 78 79 /* number of properties vector words */ 80 #define UPROPS_VECTOR_WORDS 3 81 82 /* 83 * Properties in vector word 0 84 * Bits 85 * 31..24 DerivedAge version major/minor one nibble each 86 * 23..20 reserved 87 * 19..17 East Asian Width 88 * 16.. 8 UBlockCode 89 * 7.. 0 UScriptCode 90 */ 91 92 /* derived age: one nibble each for major and minor version numbers */ 93 #define UPROPS_AGE_MASK 0xff000000 94 #define UPROPS_AGE_SHIFT 24 95 96 #define UPROPS_EA_MASK 0x000e0000 97 #define UPROPS_EA_SHIFT 17 98 99 #define UPROPS_BLOCK_MASK 0x0001ff00 100 #define UPROPS_BLOCK_SHIFT 8 101 102 #define UPROPS_SCRIPT_MASK 0x000000ff 103 104 /* 105 * Properties in vector word 1 106 * Each bit encodes one binary property. 107 * The following constants represent the bit number, use 1<<UPROPS_XYZ. 108 * UPROPS_BINARY_1_TOP<=32! 109 * 110 * Keep this list of property enums in sync with 111 * propListNames[] in icu/source/tools/genprops/props2.c! 112 * 113 * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_". 114 */ 115 enum { 116 UPROPS_WHITE_SPACE, 117 UPROPS_DASH, 118 UPROPS_HYPHEN, 119 UPROPS_QUOTATION_MARK, 120 UPROPS_TERMINAL_PUNCTUATION, 121 UPROPS_MATH, 122 UPROPS_HEX_DIGIT, 123 UPROPS_ASCII_HEX_DIGIT, 124 UPROPS_ALPHABETIC, 125 UPROPS_IDEOGRAPHIC, 126 UPROPS_DIACRITIC, 127 UPROPS_EXTENDER, 128 UPROPS_NONCHARACTER_CODE_POINT, 129 UPROPS_GRAPHEME_EXTEND, 130 UPROPS_GRAPHEME_LINK, 131 UPROPS_IDS_BINARY_OPERATOR, 132 UPROPS_IDS_TRINARY_OPERATOR, 133 UPROPS_RADICAL, 134 UPROPS_UNIFIED_IDEOGRAPH, 135 UPROPS_DEFAULT_IGNORABLE_CODE_POINT, 136 UPROPS_DEPRECATED, 137 UPROPS_LOGICAL_ORDER_EXCEPTION, 138 UPROPS_XID_START, 139 UPROPS_XID_CONTINUE, 140 UPROPS_ID_START, /* ICU 2.6, uprops format version 3.2 */ 141 UPROPS_ID_CONTINUE, 142 UPROPS_GRAPHEME_BASE, 143 UPROPS_S_TERM, /* new in ICU 3.0 and Unicode 4.0.1 */ 144 UPROPS_VARIATION_SELECTOR, 145 UPROPS_PATTERN_SYNTAX, /* new in ICU 3.4 and Unicode 4.1 */ 146 UPROPS_PATTERN_WHITE_SPACE, 147 UPROPS_RESERVED, /* reserved & unused */ 148 UPROPS_BINARY_1_TOP /* ==32 - full! */ 149 }; 150 151 /* 152 * Properties in vector word 2 153 * Bits 154 * 31..26 reserved 155 * 25..20 Line Break 156 * 19..15 Sentence Break 157 * 14..10 Word Break 158 * 9.. 5 Grapheme Cluster Break 159 * 4.. 0 Decomposition Type 160 */ 161 #define UPROPS_LB_MASK 0x03f00000 162 #define UPROPS_LB_SHIFT 20 163 #define UPROPS_LB_VWORD 2 164 165 #define UPROPS_SB_MASK 0x000f8000 166 #define UPROPS_SB_SHIFT 15 167 168 #define UPROPS_WB_MASK 0x00007c00 169 #define UPROPS_WB_SHIFT 10 170 171 #define UPROPS_GCB_MASK 0x000003e0 172 #define UPROPS_GCB_SHIFT 5 173 174 #define UPROPS_DT_MASK 0x0000001f 175 176 /** 177 * Get a properties vector word for a code point. 178 * Implemented in uchar.c for uprops.c. 179 * column==-1 gets the 32-bit main properties word instead. 180 * @return 0 if no data or illegal argument 181 */ 182 U_CFUNC uint32_t 183 u_getUnicodeProperties(UChar32 c, int32_t column); 184 185 /** 186 * Get the the maximum values for some enum/int properties. 187 * Use the same column numbers as for u_getUnicodeProperties(). 188 * The returned value will contain maximum values stored in the same bit fields 189 * as where the enum values are stored in the u_getUnicodeProperties() 190 * return values for the same columns. 191 * 192 * Valid columns are those for properties words that contain enumerated values. 193 * (ICU 2.6: columns 0 and 2) 194 * For other column numbers, this function will return 0. 195 * 196 * @internal 197 */ 198 U_CFUNC int32_t 199 uprv_getMaxValues(int32_t column); 200 201 /** 202 * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM. 203 * @internal 204 */ 205 U_CFUNC UBool 206 u_isalnumPOSIX(UChar32 c); 207 208 /** 209 * Checks if c is in 210 * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}] 211 * with space=\p{Whitespace} and Control=Cc. 212 * Implements UCHAR_POSIX_GRAPH. 213 * @internal 214 */ 215 U_CFUNC UBool 216 u_isgraphPOSIX(UChar32 c); 217 218 /** 219 * Checks if c is in \p{graph}\p{blank} - \p{cntrl}. 220 * Implements UCHAR_POSIX_PRINT. 221 * @internal 222 */ 223 U_CFUNC UBool 224 u_isprintPOSIX(UChar32 c); 225 226 /** Turn a bit index into a bit flag. @internal */ 227 #define FLAG(n) ((uint32_t)1<<(n)) 228 229 /** Flags for general categories in the order of UCharCategory. @internal */ 230 #define _Cn FLAG(U_GENERAL_OTHER_TYPES) 231 #define _Lu FLAG(U_UPPERCASE_LETTER) 232 #define _Ll FLAG(U_LOWERCASE_LETTER) 233 #define _Lt FLAG(U_TITLECASE_LETTER) 234 #define _Lm FLAG(U_MODIFIER_LETTER) 235 /* #define _Lo FLAG(U_OTHER_LETTER) -- conflicts with MS Visual Studio 9.0 xiosbase */ 236 #define _Mn FLAG(U_NON_SPACING_MARK) 237 #define _Me FLAG(U_ENCLOSING_MARK) 238 #define _Mc FLAG(U_COMBINING_SPACING_MARK) 239 #define _Nd FLAG(U_DECIMAL_DIGIT_NUMBER) 240 #define _Nl FLAG(U_LETTER_NUMBER) 241 #define _No FLAG(U_OTHER_NUMBER) 242 #define _Zs FLAG(U_SPACE_SEPARATOR) 243 #define _Zl FLAG(U_LINE_SEPARATOR) 244 #define _Zp FLAG(U_PARAGRAPH_SEPARATOR) 245 #define _Cc FLAG(U_CONTROL_CHAR) 246 #define _Cf FLAG(U_FORMAT_CHAR) 247 #define _Co FLAG(U_PRIVATE_USE_CHAR) 248 #define _Cs FLAG(U_SURROGATE) 249 #define _Pd FLAG(U_DASH_PUNCTUATION) 250 #define _Ps FLAG(U_START_PUNCTUATION) 251 /* #define _Pe FLAG(U_END_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 xlocnum */ 252 /* #define _Pc FLAG(U_CONNECTOR_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */ 253 #define _Po FLAG(U_OTHER_PUNCTUATION) 254 #define _Sm FLAG(U_MATH_SYMBOL) 255 #define _Sc FLAG(U_CURRENCY_SYMBOL) 256 #define _Sk FLAG(U_MODIFIER_SYMBOL) 257 #define _So FLAG(U_OTHER_SYMBOL) 258 #define _Pi FLAG(U_INITIAL_PUNCTUATION) 259 /* #define _Pf FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */ 260 261 /** Some code points. @internal */ 262 enum { 263 TAB =0x0009, 264 LF =0x000a, 265 FF =0x000c, 266 CR =0x000d, 267 U_A =0x0041, 268 U_F =0x0046, 269 U_Z =0x005a, 270 U_a =0x0061, 271 U_f =0x0066, 272 U_z =0x007a, 273 DEL =0x007f, 274 NL =0x0085, 275 NBSP =0x00a0, 276 CGJ =0x034f, 277 FIGURESP=0x2007, 278 HAIRSP =0x200a, 279 ZWNJ =0x200c, 280 ZWJ =0x200d, 281 RLM =0x200f, 282 NNBSP =0x202f, 283 WJ =0x2060, 284 INHSWAP =0x206a, 285 NOMDIG =0x206f, 286 U_FW_A =0xff21, 287 U_FW_F =0xff26, 288 U_FW_Z =0xff3a, 289 U_FW_a =0xff41, 290 U_FW_f =0xff46, 291 U_FW_z =0xff5a, 292 ZWNBSP =0xfeff 293 }; 294 295 /** 296 * Get the maximum length of a (regular/1.0/extended) character name. 297 * @return 0 if no character names available. 298 */ 299 U_CAPI int32_t U_EXPORT2 300 uprv_getMaxCharNameLength(void); 301 302 /** 303 * Fills set with characters that are used in Unicode character names. 304 * Includes all characters that are used in regular/Unicode 1.0/extended names. 305 * Just empties the set if no character names are available. 306 * @param sa USetAdder to receive characters. 307 */ 308 U_CAPI void U_EXPORT2 309 uprv_getCharNameCharacters(const USetAdder *sa); 310 311 /** 312 * Constants for which data and implementation files provide which properties. 313 * Used by UnicodeSet for service-specific property enumeration. 314 * @internal 315 */ 316 enum UPropertySource { 317 /** No source, not a supported property. */ 318 UPROPS_SRC_NONE, 319 /** From uchar.c/uprops.icu main trie */ 320 UPROPS_SRC_CHAR, 321 /** From uchar.c/uprops.icu properties vectors trie */ 322 UPROPS_SRC_PROPSVEC, 323 /** From unames.c/unames.icu */ 324 UPROPS_SRC_NAMES, 325 /** From unorm.cpp/unorm.icu */ 326 UPROPS_SRC_NORM, 327 /** From ucase.c/ucase.icu */ 328 UPROPS_SRC_CASE, 329 /** From ubidi_props.c/ubidi.icu */ 330 UPROPS_SRC_BIDI, 331 /** From uchar.c/uprops.icu main trie as well as properties vectors trie */ 332 UPROPS_SRC_CHAR_AND_PROPSVEC, 333 /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */ 334 UPROPS_SRC_CASE_AND_NORM, 335 /** From normalizer2impl.cpp/nfc.nrm */ 336 UPROPS_SRC_NFC, 337 /** From normalizer2impl.cpp/nfkc.nrm */ 338 UPROPS_SRC_NFKC, 339 /** From normalizer2impl.cpp/nfkc_cf.nrm */ 340 UPROPS_SRC_NFKC_CF, 341 /** One more than the highest UPropertySource (UPROPS_SRC_) constant. */ 342 UPROPS_SRC_COUNT 343 }; 344 typedef enum UPropertySource UPropertySource; 345 346 /** 347 * @see UPropertySource 348 * @internal 349 */ 350 U_CFUNC UPropertySource U_EXPORT2 351 uprops_getSource(UProperty which); 352 353 /** 354 * Enumerate uprops.icu's main data trie and add the 355 * start of each range of same properties to the set. 356 * @internal 357 */ 358 U_CFUNC void U_EXPORT2 359 uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode); 360 361 /** 362 * Enumerate uprops.icu's properties vectors trie and add the 363 * start of each range of same properties to the set. 364 * @internal 365 */ 366 U_CFUNC void U_EXPORT2 367 upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode); 368 369 /** 370 * Return a set of characters for property enumeration. 371 * For each two consecutive characters (start, limit) in the set, 372 * all of the properties for start..limit-1 are all the same. 373 * 374 * @param sa USetAdder to receive result. Existing contents are lost. 375 * @internal 376 */ 377 /*U_CFUNC void U_EXPORT2 378 uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode); 379 */ 380 381 /** 382 * Swap the ICU Unicode properties file. See uchar.c. 383 * @internal 384 */ 385 U_CAPI int32_t U_EXPORT2 386 uprops_swap(const UDataSwapper *ds, 387 const void *inData, int32_t length, void *outData, 388 UErrorCode *pErrorCode); 389 390 /** 391 * Swap the ICU Unicode character names file. See uchar.c. 392 * @internal 393 */ 394 U_CAPI int32_t U_EXPORT2 395 uchar_swapNames(const UDataSwapper *ds, 396 const void *inData, int32_t length, void *outData, 397 UErrorCode *pErrorCode); 398 399 #ifdef XP_CPLUSPLUS 400 401 U_NAMESPACE_BEGIN 402 403 class UnicodeSet; 404 405 // implemented in uniset_props.cpp 406 U_CFUNC UnicodeSet * 407 uniset_getUnicode32Instance(UErrorCode &errorCode); 408 409 U_NAMESPACE_END 410 411 #endif 412 413 #endif 414