icuSources/common/uprops.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2002-2003, 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
  26 /* indexes[] entries */
  27 enum {
  28     UPROPS_PROPS32_INDEX,
  29     UPROPS_EXCEPTIONS_INDEX,
  30     UPROPS_EXCEPTIONS_TOP_INDEX,
  31
  32     UPROPS_ADDITIONAL_TRIE_INDEX,
  33     UPROPS_ADDITIONAL_VECTORS_INDEX,
  34     UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
  35
  36     UPROPS_RESERVED_INDEX, /* 6 */
  37
  38     /* maximum values for code values in vector word 0 */
  39     UPROPS_MAX_VALUES_INDEX=10,
  40     /* maximum values for code values in vector word 2 */
  41     UPROPS_MAX_VALUES_2_INDEX,
  42
  43     UPROPS_INDEX_COUNT=16
  44 };
  45
  46 /* definitions for the main properties words */
  47 enum {
  48     /* general category shift==0                                0 (5 bits) */
  49     UPROPS_EXCEPTION_SHIFT=5,                               /*  5 (1 bit)  */
  50     UPROPS_BIDI_SHIFT,                                      /*  6 (5 bits) */
  51     UPROPS_MIRROR_SHIFT=UPROPS_BIDI_SHIFT+5,                /* 11 (1 bit)  */
  52     UPROPS_NUMERIC_TYPE_SHIFT,                              /* 12 (3 bits) */
  53     UPROPS_CASE_SENSITIVE_SHIFT=UPROPS_NUMERIC_TYPE_SHIFT+3,/* 15 (1 bit) format version 3.2 */
  54     UPROPS_RESERVED_SHIFT,                                  /* 16 (4 bits) */
  55     UPROPS_VALUE_SHIFT=20,                                  /* 20 */
  56
  57     UPROPS_EXCEPTION_BIT=1UL<<UPROPS_EXCEPTION_SHIFT,
  58     UPROPS_VALUE_BITS=32-UPROPS_VALUE_SHIFT,
  59
  60     UPROPS_MIN_VALUE=-(1L<<(UPROPS_VALUE_BITS-1)),
  61     UPROPS_MAX_VALUE=(1L<<(UPROPS_VALUE_BITS-1))-1,
  62     UPROPS_MAX_EXCEPTIONS_COUNT=1L<<UPROPS_VALUE_BITS
  63 };
  64
  65 #define PROPS_VALUE_IS_EXCEPTION(props) ((props)&UPROPS_EXCEPTION_BIT)
  66 #define GET_CATEGORY(props) ((props)&0x1f)
  67 #define GET_BIDI_CLASS(props) ((props>>UPROPS_BIDI_SHIFT)&0x1f)
  68 #define GET_NUMERIC_TYPE(props) (((props)>>UPROPS_NUMERIC_TYPE_SHIFT)&7)
  69 #define GET_UNSIGNED_VALUE(props) ((props)>>UPROPS_VALUE_SHIFT)
  70 #define GET_SIGNED_VALUE(props) ((int32_t)(props)>>UPROPS_VALUE_SHIFT)
  71 #define GET_EXCEPTIONS(props) (exceptionsTable+GET_UNSIGNED_VALUE(props))
  72
  73 #define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
  74
  75 enum {
  76     EXC_UPPERCASE,
  77     EXC_LOWERCASE,
  78     EXC_TITLECASE,
  79     EXC_UNUSED,
  80     EXC_NUMERIC_VALUE,
  81     EXC_DENOMINATOR_VALUE,
  82     EXC_MIRROR_MAPPING,
  83     EXC_SPECIAL_CASING,
  84     EXC_CASE_FOLDING
  85 };
  86
  87 /* number of properties vector words */
  88 #define UPROPS_VECTOR_WORDS     3
  89
  90 /*
  91  * Properties in vector word 0
  92  * Bits
  93  * 31..24   DerivedAge version major/minor one nibble each
  94  * 23       reserved
  95  * 22..18   Line Break
  96  * 17..15   East Asian Width
  97  * 14.. 7   UBlockCode
  98  *  6.. 0   UScriptCode
  99  */
 100
 101 /* derived age: one nibble each for major and minor version numbers */
 102 #define UPROPS_AGE_MASK         0xff000000
 103 #define UPROPS_AGE_SHIFT        24
 104
 105 #define UPROPS_LB_MASK          0x007C0000
 106 #define UPROPS_LB_SHIFT         18
 107
 108 #define UPROPS_EA_MASK          0x00038000
 109 #define UPROPS_EA_SHIFT         15
 110
 111 #define UPROPS_BLOCK_MASK       0x00007f80
 112 #define UPROPS_BLOCK_SHIFT      7
 113
 114 #define UPROPS_SCRIPT_MASK      0x0000007f
 115
 116 /*
 117  * Properties in vector word 1
 118  * Each bit encodes one binary property.
 119  * The following constants represent the bit number, use 1<<UPROPS_XYZ.
 120  * UPROPS_BINARY_1_TOP<=32!
 121  *
 122  * Keep this list of property enums in sync with
 123  * propListNames[] in icu/source/tools/genprops/props2.c!
 124  *
 125  * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
 126  */
 127 enum {
 128     UPROPS_WHITE_SPACE,
 129     UPROPS_BIDI_CONTROL,
 130     UPROPS_JOIN_CONTROL,
 131     UPROPS_DASH,
 132     UPROPS_HYPHEN,
 133     UPROPS_QUOTATION_MARK,
 134     UPROPS_TERMINAL_PUNCTUATION,
 135     UPROPS_MATH,
 136     UPROPS_HEX_DIGIT,
 137     UPROPS_ASCII_HEX_DIGIT,
 138     UPROPS_ALPHABETIC,
 139     UPROPS_IDEOGRAPHIC,
 140     UPROPS_DIACRITIC,
 141     UPROPS_EXTENDER,
 142     UPROPS_LOWERCASE,
 143     UPROPS_UPPERCASE,
 144     UPROPS_NONCHARACTER_CODE_POINT,
 145     UPROPS_GRAPHEME_EXTEND,
 146     UPROPS_GRAPHEME_LINK,
 147     UPROPS_IDS_BINARY_OPERATOR,
 148     UPROPS_IDS_TRINARY_OPERATOR,
 149     UPROPS_RADICAL,
 150     UPROPS_UNIFIED_IDEOGRAPH,
 151     UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
 152     UPROPS_DEPRECATED,
 153     UPROPS_SOFT_DOTTED,
 154     UPROPS_LOGICAL_ORDER_EXCEPTION,
 155     UPROPS_XID_START,
 156     UPROPS_XID_CONTINUE,
 157     UPROPS_ID_START,                            /* ICU 2.6, uprops format version 3.2 */
 158     UPROPS_ID_CONTINUE,
 159     UPROPS_GRAPHEME_BASE,
 160     UPROPS_BINARY_1_TOP                         /* ==32 - full! */
 161 };
 162
 163 /*
 164  * Properties in vector word 2
 165  * Bits
 166  * 13..11   Joining Type
 167  * 10.. 5   Joining Group
 168  *  4.. 0   Decomposition Type
 169  */
 170 #define UPROPS_JT_MASK          0x00003800
 171 #define UPROPS_JT_SHIFT         11
 172
 173 #define UPROPS_JG_MASK          0x000007e0
 174 #define UPROPS_JG_SHIFT         5
 175
 176 #define UPROPS_DT_MASK          0x0000001f
 177
 178 /**
 179  * Get a properties vector word for a code point.
 180  * Implemented in uchar.c for uprops.c.
 181  * column==-1 gets the 32-bit main properties word instead.
 182  * @return 0 if no data or illegal argument
 183  */
 184 U_CFUNC uint32_t
 185 u_getUnicodeProperties(UChar32 c, int32_t column);
 186
 187 /**
 188  * Get the the maximum values for some enum/int properties.
 189  * Use the same column numbers as for u_getUnicodeProperties().
 190  * The returned value will contain maximum values stored in the same bit fields
 191  * as where the enum values are stored in the u_getUnicodeProperties()
 192  * return values for the same columns.
 193  *
 194  * Valid columns are those for properties words that contain enumerated values.
 195  * (ICU 2.6: columns 0 and 2)
 196  * For other column numbers, this function will return 0.
 197  *
 198  * @internal
 199  */
 200 U_CFUNC int32_t
 201 uprv_getMaxValues(int32_t column);
 202
 203 /**
 204  * Unicode property names and property value names are compared
 205  * "loosely". Property[Value]Aliases.txt say:
 206  *   "With loose matching of property names, the case distinctions, whitespace,
 207  *    and '_' are ignored."
 208  *
 209  * This function does just that, for ASCII (char *) name strings.
 210  * It is almost identical to ucnv_compareNames() but also ignores
 211  * ASCII White_Space characters (U+0009..U+000d).
 212  *
 213  * @internal
 214  */
 215 U_CAPI int32_t U_EXPORT2
 216 uprv_comparePropertyNames(const char *name1, const char *name2);
 217
 218 /** Turn a bit index into a bit flag. @internal */
 219 #define FLAG(n) ((uint32_t)1<<(n))
 220
 221 /** Flags for general categories in the order of UCharCategory. @internal */
 222 #define _Cn     FLAG(U_GENERAL_OTHER_TYPES)
 223 #define _Lu     FLAG(U_UPPERCASE_LETTER)
 224 #define _Ll     FLAG(U_LOWERCASE_LETTER)
 225 #define _Lt     FLAG(U_TITLECASE_LETTER)
 226 #define _Lm     FLAG(U_MODIFIER_LETTER)
 227 #define _Lo     FLAG(U_OTHER_LETTER)
 228 #define _Mn     FLAG(U_NON_SPACING_MARK)
 229 #define _Me     FLAG(U_ENCLOSING_MARK)
 230 #define _Mc     FLAG(U_COMBINING_SPACING_MARK)
 231 #define _Nd     FLAG(U_DECIMAL_DIGIT_NUMBER)
 232 #define _Nl     FLAG(U_LETTER_NUMBER)
 233 #define _No     FLAG(U_OTHER_NUMBER)
 234 #define _Zs     FLAG(U_SPACE_SEPARATOR)
 235 #define _Zl     FLAG(U_LINE_SEPARATOR)
 236 #define _Zp     FLAG(U_PARAGRAPH_SEPARATOR)
 237 #define _Cc     FLAG(U_CONTROL_CHAR)
 238 #define _Cf     FLAG(U_FORMAT_CHAR)
 239 #define _Co     FLAG(U_PRIVATE_USE_CHAR)
 240 #define _Cs     FLAG(U_SURROGATE)
 241 #define _Pd     FLAG(U_DASH_PUNCTUATION)
 242 #define _Ps     FLAG(U_START_PUNCTUATION)
 243 #define _Pe     FLAG(U_END_PUNCTUATION)
 244 #define _Pc     FLAG(U_CONNECTOR_PUNCTUATION)
 245 #define _Po     FLAG(U_OTHER_PUNCTUATION)
 246 #define _Sm     FLAG(U_MATH_SYMBOL)
 247 #define _Sc     FLAG(U_CURRENCY_SYMBOL)
 248 #define _Sk     FLAG(U_MODIFIER_SYMBOL)
 249 #define _So     FLAG(U_OTHER_SYMBOL)
 250 #define _Pi     FLAG(U_INITIAL_PUNCTUATION)
 251 #define _Pf     FLAG(U_FINAL_PUNCTUATION)
 252
 253 /** Some code points. @internal */
 254 enum {
 255     TAB     =0x0009,
 256     LF      =0x000a,
 257     FF      =0x000c,
 258     CR      =0x000d,
 259     U_A     =0x0041,
 260     U_Z     =0x005a,
 261     U_a     =0x0061,
 262     U_z     =0x007a,
 263     DEL     =0x007f,
 264     NL      =0x0085,
 265     NBSP    =0x00a0,
 266     CGJ     =0x034f,
 267     FIGURESP=0x2007,
 268     HAIRSP  =0x200a,
 269     ZWNJ    =0x200c,
 270     ZWJ     =0x200d,
 271     RLM     =0x200f,
 272     NNBSP   =0x202f,
 273     WJ      =0x2060,
 274     INHSWAP =0x206a,
 275     NOMDIG  =0x206f,
 276     ZWNBSP  =0xfeff
 277 };
 278
 279 /**
 280  * Is this character a "white space" in the sense of ICU rule parsers?
 281  * @internal
 282  */
 283 U_CAPI UBool U_EXPORT2
 284 uprv_isRuleWhiteSpace(UChar32 c);
 285
 286 /**
 287  * Get the set of "white space" characters in the sense of ICU rule
 288  * parsers.  Caller must close/delete result.
 289  * @internal
 290  */
 291 U_CAPI USet* U_EXPORT2
 292 uprv_openRuleWhiteSpaceSet(UErrorCode* ec);
 293
 294 /**
 295  * Get the maximum length of a (regular/1.0/extended) character name.
 296  * @return 0 if no character names available.
 297  */
 298 U_CAPI int32_t U_EXPORT2
 299 uprv_getMaxCharNameLength(void);
 300
 301 #if 0
 302 /*
 303 Currently not used but left for future use. Probably by UnicodeSet.
 304 urename.h and unames.c changed accordingly.
 305 */
 306 /**
 307  * Get the maximum length of an ISO comment.
 308  * @return 0 if no ISO comments available.
 309  */
 310 U_CAPI int32_t U_EXPORT2
 311 uprv_getMaxISOCommentLength();
 312 #endif
 313
 314 /**
 315  * Fills set with characters that are used in Unicode character names.
 316  * Includes all characters that are used in regular/Unicode 1.0/extended names.
 317  * Just empties the set if no character names are available.
 318  * @param set USet to receive characters. Existing contents are deleted.
 319  */
 320 U_CAPI void U_EXPORT2
 321 uprv_getCharNameCharacters(USet* set);
 322
 323 #if 0
 324 /*
 325 Currently not used but left for future use. Probably by UnicodeSet.
 326 urename.h and unames.c changed accordingly.
 327 */
 328 /**
 329  * Fills set with characters that are used in Unicode character names.
 330  * Just empties the set if no ISO comments are available.
 331  * @param set USet to receive characters. Existing contents are deleted.
 332  */
 333 U_CAPI void U_EXPORT2
 334 uprv_getISOCommentCharacters(USet* set);
 335 */
 336 #endif
 337
 338 /**
 339  * Enumerate each core properties data trie and add the
 340  * start of each range of same properties to the set.
 341  * @internal
 342  */
 343 U_CAPI void U_EXPORT2
 344 uchar_addPropertyStarts(USet *set, UErrorCode *pErrorCode);
 345
 346 /**
 347  * Return a set of characters for property enumeration.
 348  * For each two consecutive characters (start, limit) in the set,
 349  * all of the properties for start..limit-1 are all the same.
 350  *
 351  * @param set USet to receive result. Existing contents are lost.
 352  * @internal
 353  */
 354 U_CAPI void U_EXPORT2
 355 uprv_getInclusions(USet* set, UErrorCode *pErrorCode);
 356
 357 #endif