icuSources/common/uprops.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2002-2004, 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.dat.
  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 "ucase.h"
  27 #include "udataswp.h"
  28
  29 /* indexes[] entries */
  30 enum {
  31     UPROPS_PROPS32_INDEX,
  32     UPROPS_EXCEPTIONS_INDEX,
  33     UPROPS_EXCEPTIONS_TOP_INDEX,
  34
  35     UPROPS_ADDITIONAL_TRIE_INDEX,
  36     UPROPS_ADDITIONAL_VECTORS_INDEX,
  37     UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
  38
  39     UPROPS_RESERVED_INDEX, /* 6 */
  40
  41     /* maximum values for code values in vector word 0 */
  42     UPROPS_MAX_VALUES_INDEX=10,
  43     /* maximum values for code values in vector word 2 */
  44     UPROPS_MAX_VALUES_2_INDEX,
  45
  46     UPROPS_INDEX_COUNT=16
  47 };
  48
  49 /* definitions for the main properties words */
  50 enum {
  51     /* general category shift==0                                0 (5 bits) */
  52     UPROPS_EXCEPTION_SHIFT=5,                               /*  5 (1 bit)  */
  53     UPROPS_BIDI_SHIFT,                                      /*  6 (5 bits) */
  54     UPROPS_MIRROR_SHIFT=UPROPS_BIDI_SHIFT+5,                /* 11 (1 bit)  */
  55     UPROPS_NUMERIC_TYPE_SHIFT,                              /* 12 (3 bits) */
  56     UPROPS_CASE_SENSITIVE_SHIFT=UPROPS_NUMERIC_TYPE_SHIFT+3,/* 15 (1 bit) format version 3.2 */
  57     UPROPS_RESERVED_SHIFT,                                  /* 16 (4 bits) */
  58     UPROPS_VALUE_SHIFT=20,                                  /* 20 */
  59
  60     UPROPS_EXCEPTION_BIT=1UL<<UPROPS_EXCEPTION_SHIFT,
  61     UPROPS_VALUE_BITS=32-UPROPS_VALUE_SHIFT,
  62
  63     UPROPS_MIN_VALUE=-(1L<<(UPROPS_VALUE_BITS-1)),
  64     UPROPS_MAX_VALUE=(1L<<(UPROPS_VALUE_BITS-1))-1,
  65     UPROPS_MAX_EXCEPTIONS_COUNT=1L<<UPROPS_VALUE_BITS
  66 };
  67
  68 #define PROPS_VALUE_IS_EXCEPTION(props) ((props)&UPROPS_EXCEPTION_BIT)
  69 #define GET_CATEGORY(props) ((props)&0x1f)
  70 #define GET_BIDI_CLASS(props) ((props>>UPROPS_BIDI_SHIFT)&0x1f)
  71 #define GET_NUMERIC_TYPE(props) (((props)>>UPROPS_NUMERIC_TYPE_SHIFT)&7)
  72 #define GET_UNSIGNED_VALUE(props) ((props)>>UPROPS_VALUE_SHIFT)
  73 #define GET_SIGNED_VALUE(props) ((int32_t)(props)>>UPROPS_VALUE_SHIFT)
  74 #define GET_EXCEPTIONS(props) (exceptionsTable+GET_UNSIGNED_VALUE(props))
  75
  76 #define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
  77
  78 enum {
  79     EXC_UPPERCASE,
  80     EXC_LOWERCASE,
  81     EXC_TITLECASE,
  82     EXC_UNUSED,
  83     EXC_NUMERIC_VALUE,
  84     EXC_DENOMINATOR_VALUE,
  85     EXC_MIRROR_MAPPING,
  86     EXC_SPECIAL_CASING,
  87     EXC_CASE_FOLDING
  88 };
  89
  90 /* number of properties vector words */
  91 #define UPROPS_VECTOR_WORDS     3
  92
  93 /*
  94  * Properties in vector word 0
  95  * Bits
  96  * 31..24   DerivedAge version major/minor one nibble each
  97  * 23       reserved
  98  * 22..18   Line Break
  99  * 17..15   East Asian Width
 100  * 14.. 7   UBlockCode
 101  *  6.. 0   UScriptCode
 102  */
 103
 104 /* derived age: one nibble each for major and minor version numbers */
 105 #define UPROPS_AGE_MASK         0xff000000
 106 #define UPROPS_AGE_SHIFT        24
 107
 108 #define UPROPS_LB_MASK          0x007C0000
 109 #define UPROPS_LB_SHIFT         18
 110
 111 #define UPROPS_EA_MASK          0x00038000
 112 #define UPROPS_EA_SHIFT         15
 113
 114 #define UPROPS_BLOCK_MASK       0x00007f80
 115 #define UPROPS_BLOCK_SHIFT      7
 116
 117 #define UPROPS_SCRIPT_MASK      0x0000007f
 118
 119 /*
 120  * Properties in vector word 1
 121  * Each bit encodes one binary property.
 122  * The following constants represent the bit number, use 1<<UPROPS_XYZ.
 123  * UPROPS_BINARY_1_TOP<=32!
 124  *
 125  * Keep this list of property enums in sync with
 126  * propListNames[] in icu/source/tools/genprops/props2.c!
 127  *
 128  * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
 129  */
 130 enum {
 131     UPROPS_WHITE_SPACE,
 132     UPROPS_BIDI_CONTROL,
 133     UPROPS_JOIN_CONTROL,
 134     UPROPS_DASH,
 135     UPROPS_HYPHEN,
 136     UPROPS_QUOTATION_MARK,
 137     UPROPS_TERMINAL_PUNCTUATION,
 138     UPROPS_MATH,
 139     UPROPS_HEX_DIGIT,
 140     UPROPS_ASCII_HEX_DIGIT,
 141     UPROPS_ALPHABETIC,
 142     UPROPS_IDEOGRAPHIC,
 143     UPROPS_DIACRITIC,
 144     UPROPS_EXTENDER,
 145     UPROPS_LOWERCASE,
 146     UPROPS_UPPERCASE,
 147     UPROPS_NONCHARACTER_CODE_POINT,
 148     UPROPS_GRAPHEME_EXTEND,
 149     UPROPS_GRAPHEME_LINK,
 150     UPROPS_IDS_BINARY_OPERATOR,
 151     UPROPS_IDS_TRINARY_OPERATOR,
 152     UPROPS_RADICAL,
 153     UPROPS_UNIFIED_IDEOGRAPH,
 154     UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
 155     UPROPS_DEPRECATED,
 156     UPROPS_SOFT_DOTTED,
 157     UPROPS_LOGICAL_ORDER_EXCEPTION,
 158     UPROPS_XID_START,
 159     UPROPS_XID_CONTINUE,
 160     UPROPS_ID_START,                            /* ICU 2.6, uprops format version 3.2 */
 161     UPROPS_ID_CONTINUE,
 162     UPROPS_GRAPHEME_BASE,
 163     UPROPS_BINARY_1_TOP                         /* ==32 - full! */
 164 };
 165
 166 /*
 167  * Properties in vector word 2
 168  * Bits
 169  * 31..24   More binary properties
 170  * 13..11   Joining Type
 171  * 10.. 5   Joining Group
 172  *  4.. 0   Decomposition Type
 173  */
 174 #define UPROPS_JT_MASK          0x00003800
 175 #define UPROPS_JT_SHIFT         11
 176
 177 #define UPROPS_JG_MASK          0x000007e0
 178 #define UPROPS_JG_SHIFT         5
 179
 180 #define UPROPS_DT_MASK          0x0000001f
 181
 182 enum {
 183     UPROPS_V2_S_TERM=24,                        /* new in ICU 3.0 and Unicode 4.0.1 */
 184     UPROPS_V2_VARIATION_SELECTOR,
 185     UPROPS_V2_TOP                               /* must be <=32 */
 186 };
 187
 188 /**
 189  * Get a properties vector word for a code point.
 190  * Implemented in uchar.c for uprops.c.
 191  * column==-1 gets the 32-bit main properties word instead.
 192  * @return 0 if no data or illegal argument
 193  */
 194 U_CFUNC uint32_t
 195 u_getUnicodeProperties(UChar32 c, int32_t column);
 196
 197 /**
 198  * Get the the maximum values for some enum/int properties.
 199  * Use the same column numbers as for u_getUnicodeProperties().
 200  * The returned value will contain maximum values stored in the same bit fields
 201  * as where the enum values are stored in the u_getUnicodeProperties()
 202  * return values for the same columns.
 203  *
 204  * Valid columns are those for properties words that contain enumerated values.
 205  * (ICU 2.6: columns 0 and 2)
 206  * For other column numbers, this function will return 0.
 207  *
 208  * @internal
 209  */
 210 U_CFUNC int32_t
 211 uprv_getMaxValues(int32_t column);
 212
 213 /**
 214  * Get the Hangul Syllable Type for c.
 215  * @internal
 216  */
 217 U_CFUNC UHangulSyllableType
 218 uchar_getHST(UChar32 c);
 219
 220 /** Turn a bit index into a bit flag. @internal */
 221 #define FLAG(n) ((uint32_t)1<<(n))
 222
 223 /** Flags for general categories in the order of UCharCategory. @internal */
 224 #define _Cn     FLAG(U_GENERAL_OTHER_TYPES)
 225 #define _Lu     FLAG(U_UPPERCASE_LETTER)
 226 #define _Ll     FLAG(U_LOWERCASE_LETTER)
 227 #define _Lt     FLAG(U_TITLECASE_LETTER)
 228 #define _Lm     FLAG(U_MODIFIER_LETTER)
 229 #define _Lo     FLAG(U_OTHER_LETTER)
 230 #define _Mn     FLAG(U_NON_SPACING_MARK)
 231 #define _Me     FLAG(U_ENCLOSING_MARK)
 232 #define _Mc     FLAG(U_COMBINING_SPACING_MARK)
 233 #define _Nd     FLAG(U_DECIMAL_DIGIT_NUMBER)
 234 #define _Nl     FLAG(U_LETTER_NUMBER)
 235 #define _No     FLAG(U_OTHER_NUMBER)
 236 #define _Zs     FLAG(U_SPACE_SEPARATOR)
 237 #define _Zl     FLAG(U_LINE_SEPARATOR)
 238 #define _Zp     FLAG(U_PARAGRAPH_SEPARATOR)
 239 #define _Cc     FLAG(U_CONTROL_CHAR)
 240 #define _Cf     FLAG(U_FORMAT_CHAR)
 241 #define _Co     FLAG(U_PRIVATE_USE_CHAR)
 242 #define _Cs     FLAG(U_SURROGATE)
 243 #define _Pd     FLAG(U_DASH_PUNCTUATION)
 244 #define _Ps     FLAG(U_START_PUNCTUATION)
 245 #define _Pe     FLAG(U_END_PUNCTUATION)
 246 #define _Pc     FLAG(U_CONNECTOR_PUNCTUATION)
 247 #define _Po     FLAG(U_OTHER_PUNCTUATION)
 248 #define _Sm     FLAG(U_MATH_SYMBOL)
 249 #define _Sc     FLAG(U_CURRENCY_SYMBOL)
 250 #define _Sk     FLAG(U_MODIFIER_SYMBOL)
 251 #define _So     FLAG(U_OTHER_SYMBOL)
 252 #define _Pi     FLAG(U_INITIAL_PUNCTUATION)
 253 #define _Pf     FLAG(U_FINAL_PUNCTUATION)
 254
 255 /** Some code points. @internal */
 256 enum {
 257     TAB     =0x0009,
 258     LF      =0x000a,
 259     FF      =0x000c,
 260     CR      =0x000d,
 261     U_A     =0x0041,
 262     U_Z     =0x005a,
 263     U_a     =0x0061,
 264     U_z     =0x007a,
 265     DEL     =0x007f,
 266     NL      =0x0085,
 267     NBSP    =0x00a0,
 268     CGJ     =0x034f,
 269     FIGURESP=0x2007,
 270     HAIRSP  =0x200a,
 271     ZWNJ    =0x200c,
 272     ZWJ     =0x200d,
 273     RLM     =0x200f,
 274     NNBSP   =0x202f,
 275     WJ      =0x2060,
 276     INHSWAP =0x206a,
 277     NOMDIG  =0x206f,
 278     ZWNBSP  =0xfeff
 279 };
 280
 281 /**
 282  * Get the maximum length of a (regular/1.0/extended) character name.
 283  * @return 0 if no character names available.
 284  */
 285 U_CAPI int32_t U_EXPORT2
 286 uprv_getMaxCharNameLength(void);
 287
 288 #if 0
 289 /*
 290 Currently not used but left for future use. Probably by UnicodeSet.
 291 urename.h and unames.c changed accordingly.
 292 */
 293 /**
 294  * Get the maximum length of an ISO comment.
 295  * @return 0 if no ISO comments available.
 296  */
 297 U_CAPI int32_t U_EXPORT2
 298 uprv_getMaxISOCommentLength();
 299 #endif
 300
 301 /**
 302  * Fills set with characters that are used in Unicode character names.
 303  * Includes all characters that are used in regular/Unicode 1.0/extended names.
 304  * Just empties the set if no character names are available.
 305  * @param sa USetAdder to receive characters.
 306  */
 307 U_CAPI void U_EXPORT2
 308 uprv_getCharNameCharacters(USetAdder *sa);
 309
 310 #if 0
 311 /*
 312 Currently not used but left for future use. Probably by UnicodeSet.
 313 urename.h and unames.c changed accordingly.
 314 */
 315 /**
 316  * Fills set with characters that are used in Unicode character names.
 317  * Just empties the set if no ISO comments are available.
 318  * @param sa USetAdder to receive characters.
 319  */
 320 U_CAPI void U_EXPORT2
 321 uprv_getISOCommentCharacters(USetAdder *sa);
 322 */
 323 #endif
 324
 325 /**
 326  * Constants for which data and implementation files provide which properties.
 327  * Used by UnicodeSet for service-specific property enumeration.
 328  * @internal
 329  */
 330 enum UPropertySource {
 331     /** No source, not a supported property. */
 332     UPROPS_SRC_NONE,
 333     /** From uchar.c/uprops.icu */
 334     UPROPS_SRC_CHAR,
 335     /** Hangul_Syllable_Type, from uchar.c/uprops.icu */
 336     UPROPS_SRC_HST,
 337     /** From unames.c/unames.icu */
 338     UPROPS_SRC_NAMES,
 339     /** From unorm.cpp/unorm.icu */
 340     UPROPS_SRC_NORM,
 341     /** From ucase.c/ucase.icu */
 342     UPROPS_SRC_CASE,
 343     /** From ubidi.c/ubidi.icu */
 344     UPROPS_SRC_BIDI,
 345     /** One more than the highes UPropertySource (UPROPS_SRC_) constant. */
 346     UPROPS_SRC_COUNT
 347 };
 348 typedef enum UPropertySource UPropertySource;
 349
 350 /**
 351  * @see UPropertySource
 352  * @internal
 353  */
 354 U_CAPI UPropertySource U_EXPORT2
 355 uprops_getSource(UProperty which);
 356
 357 /**
 358  * Enumerate each core properties data trie and add the
 359  * start of each range of same properties to the set.
 360  * @internal
 361  */
 362 U_CAPI void U_EXPORT2
 363 uchar_addPropertyStarts(USetAdder *sa, UErrorCode *pErrorCode);
 364
 365 /**
 366  * Same as uchar_addPropertyStarts() but only for Hangul_Syllable_Type.
 367  * @internal
 368  */
 369 U_CAPI void U_EXPORT2
 370 uhst_addPropertyStarts(USetAdder *sa, UErrorCode *pErrorCode);
 371
 372 /**
 373  * Return a set of characters for property enumeration.
 374  * For each two consecutive characters (start, limit) in the set,
 375  * all of the properties for start..limit-1 are all the same.
 376  *
 377  * @param sa USetAdder to receive result. Existing contents are lost.
 378  * @internal
 379  */
 380 U_CAPI void U_EXPORT2
 381 uprv_getInclusions(USetAdder *sa, UErrorCode *pErrorCode);
 382
 383 /**
 384  * Swap the ICU Unicode properties file. See uchar.c.
 385  * @internal
 386  */
 387 U_CAPI int32_t U_EXPORT2
 388 uprops_swap(const UDataSwapper *ds,
 389             const void *inData, int32_t length, void *outData,
 390             UErrorCode *pErrorCode);
 391
 392 /**
 393  * Swap the ICU Unicode character names file. See uchar.c.
 394  * @internal
 395  */
 396 U_CAPI int32_t U_EXPORT2
 397 uchar_swapNames(const UDataSwapper *ds,
 398                 const void *inData, int32_t length, void *outData,
 399                 UErrorCode *pErrorCode);
 400
 401 #endif