icuSources/common/uprops.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2002-2011, 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_SCRIPT_EXTENSIONS_INDEX,
  39
  40     UPROPS_RESERVED_INDEX_7,
  41     UPROPS_RESERVED_INDEX_8,
  42
  43     /* size of the data file (number of 32-bit units after the header) */
  44     UPROPS_DATA_TOP_INDEX,
  45
  46     /* maximum values for code values in vector word 0 */
  47     UPROPS_MAX_VALUES_INDEX=10,
  48     /* maximum values for code values in vector word 2 */
  49     UPROPS_MAX_VALUES_2_INDEX,
  50
  51     UPROPS_INDEX_COUNT=16
  52 };
  53
  54 /* definitions for the main properties words */
  55 enum {
  56     /* general category shift==0                                0 (5 bits) */
  57     /* reserved                                                 5 (1 bit) */
  58     UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6                       /*  6 (10 bits) */
  59 };
  60
  61 #define GET_CATEGORY(props) ((props)&0x1f)
  62 #define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
  63
  64 #define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
  65
  66 /* constants for the storage form of numeric types and values */
  67 enum {
  68     UPROPS_NTV_NONE=0,
  69     UPROPS_NTV_DECIMAL_START=1,
  70     UPROPS_NTV_DIGIT_START=11,
  71     UPROPS_NTV_NUMERIC_START=21,
  72     UPROPS_NTV_FRACTION_START=0xb0,
  73     UPROPS_NTV_LARGE_START=0x1e0,
  74     UPROPS_NTV_RESERVED_START=0x300,
  75
  76     UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
  77 };
  78
  79 #define UPROPS_NTV_GET_TYPE(ntv) \
  80     ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
  81     (ntv<UPROPS_NTV_DIGIT_START) ?  U_NT_DECIMAL : \
  82     (ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \
  83     U_NT_NUMERIC)
  84
  85 /* number of properties vector words */
  86 #define UPROPS_VECTOR_WORDS     3
  87
  88 /*
  89  * Properties in vector word 0
  90  * Bits
  91  * 31..24   DerivedAge version major/minor one nibble each
  92  * 23..22   3..1: Bits 7..0 = Script_Extensions index
  93  *             3: Script value from Script_Extensions
  94  *             2: Script=Inherited
  95  *             1: Script=Common
  96  *             0: Script=bits 7..0
  97  * 21..20   reserved
  98  * 19..17   East Asian Width
  99  * 16.. 8   UBlockCode
 100  *  7.. 0   UScriptCode, or index to Script_Extensions
 101  */
 102
 103 /* derived age: one nibble each for major and minor version numbers */
 104 #define UPROPS_AGE_MASK         0xff000000
 105 #define UPROPS_AGE_SHIFT        24
 106
 107 /* Script_Extensions: mask includes Script */
 108 #define UPROPS_SCRIPT_X_MASK    0x00c000ff
 109 #define UPROPS_SCRIPT_X_SHIFT   22
 110
 111 #define UPROPS_EA_MASK          0x000e0000
 112 #define UPROPS_EA_SHIFT         17
 113
 114 #define UPROPS_BLOCK_MASK       0x0001ff00
 115 #define UPROPS_BLOCK_SHIFT      8
 116
 117 #define UPROPS_SCRIPT_MASK      0x000000ff
 118
 119 /* UPROPS_SCRIPT_X_WITH_COMMON must be the lowest value that involves Script_Extensions. */
 120 #define UPROPS_SCRIPT_X_WITH_COMMON     0x400000
 121 #define UPROPS_SCRIPT_X_WITH_INHERITED  0x800000
 122 #define UPROPS_SCRIPT_X_WITH_OTHER      0xc00000
 123
 124 /*
 125  * Properties in vector word 1
 126  * Each bit encodes one binary property.
 127  * The following constants represent the bit number, use 1<<UPROPS_XYZ.
 128  * UPROPS_BINARY_1_TOP<=32!
 129  *
 130  * Keep this list of property enums in sync with
 131  * propListNames[] in icu/source/tools/genprops/props2.c!
 132  *
 133  * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
 134  */
 135 enum {
 136     UPROPS_WHITE_SPACE,
 137     UPROPS_DASH,
 138     UPROPS_HYPHEN,
 139     UPROPS_QUOTATION_MARK,
 140     UPROPS_TERMINAL_PUNCTUATION,
 141     UPROPS_MATH,
 142     UPROPS_HEX_DIGIT,
 143     UPROPS_ASCII_HEX_DIGIT,
 144     UPROPS_ALPHABETIC,
 145     UPROPS_IDEOGRAPHIC,
 146     UPROPS_DIACRITIC,
 147     UPROPS_EXTENDER,
 148     UPROPS_NONCHARACTER_CODE_POINT,
 149     UPROPS_GRAPHEME_EXTEND,
 150     UPROPS_GRAPHEME_LINK,
 151     UPROPS_IDS_BINARY_OPERATOR,
 152     UPROPS_IDS_TRINARY_OPERATOR,
 153     UPROPS_RADICAL,
 154     UPROPS_UNIFIED_IDEOGRAPH,
 155     UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
 156     UPROPS_DEPRECATED,
 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_S_TERM,                              /* new in ICU 3.0 and Unicode 4.0.1 */
 164     UPROPS_VARIATION_SELECTOR,
 165     UPROPS_PATTERN_SYNTAX,                      /* new in ICU 3.4 and Unicode 4.1 */
 166     UPROPS_PATTERN_WHITE_SPACE,
 167     UPROPS_RESERVED,                            /* reserved & unused */
 168     UPROPS_BINARY_1_TOP                         /* ==32 - full! */
 169 };
 170
 171 /*
 172  * Properties in vector word 2
 173  * Bits
 174  * 31..26   reserved
 175  * 25..20   Line Break
 176  * 19..15   Sentence Break
 177  * 14..10   Word Break
 178  *  9.. 5   Grapheme Cluster Break
 179  *  4.. 0   Decomposition Type
 180  */
 181 #define UPROPS_LB_MASK          0x03f00000
 182 #define UPROPS_LB_SHIFT         20
 183
 184 #define UPROPS_SB_MASK          0x000f8000
 185 #define UPROPS_SB_SHIFT         15
 186
 187 #define UPROPS_WB_MASK          0x00007c00
 188 #define UPROPS_WB_SHIFT         10
 189
 190 #define UPROPS_GCB_MASK         0x000003e0
 191 #define UPROPS_GCB_SHIFT        5
 192
 193 #define UPROPS_DT_MASK          0x0000001f
 194
 195 /**
 196  * Gets the main properties value for a code point.
 197  * Implemented in uchar.c for uprops.cpp.
 198  */
 199 U_CFUNC uint32_t
 200 u_getMainProperties(UChar32 c);
 201
 202 /**
 203  * Get a properties vector word for a code point.
 204  * Implemented in uchar.c for uprops.cpp.
 205  * @return 0 if no data or illegal argument
 206  */
 207 U_CFUNC uint32_t
 208 u_getUnicodeProperties(UChar32 c, int32_t column);
 209
 210 /**
 211  * Get the the maximum values for some enum/int properties.
 212  * Use the same column numbers as for u_getUnicodeProperties().
 213  * The returned value will contain maximum values stored in the same bit fields
 214  * as where the enum values are stored in the u_getUnicodeProperties()
 215  * return values for the same columns.
 216  *
 217  * Valid columns are those for properties words that contain enumerated values.
 218  * (ICU 2.6: columns 0 and 2)
 219  * For other column numbers, this function will return 0.
 220  *
 221  * @internal
 222  */
 223 U_CFUNC int32_t
 224 uprv_getMaxValues(int32_t column);
 225
 226 /**
 227  * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
 228  * @internal
 229  */
 230 U_CFUNC UBool
 231 u_isalnumPOSIX(UChar32 c);
 232
 233 /**
 234  * Checks if c is in
 235  * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
 236  * with space=\p{Whitespace} and Control=Cc.
 237  * Implements UCHAR_POSIX_GRAPH.
 238  * @internal
 239  */
 240 U_CFUNC UBool
 241 u_isgraphPOSIX(UChar32 c);
 242
 243 /**
 244  * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
 245  * Implements UCHAR_POSIX_PRINT.
 246  * @internal
 247  */
 248 U_CFUNC UBool
 249 u_isprintPOSIX(UChar32 c);
 250
 251 /** Turn a bit index into a bit flag. @internal */
 252 #define FLAG(n) ((uint32_t)1<<(n))
 253
 254 /** Flags for general categories in the order of UCharCategory. @internal */
 255 #define _Cn     FLAG(U_GENERAL_OTHER_TYPES)
 256 #define _Lu     FLAG(U_UPPERCASE_LETTER)
 257 #define _Ll     FLAG(U_LOWERCASE_LETTER)
 258 #define _Lt     FLAG(U_TITLECASE_LETTER)
 259 #define _Lm     FLAG(U_MODIFIER_LETTER)
 260 /* #define _Lo     FLAG(U_OTHER_LETTER) -- conflicts with MS Visual Studio 9.0 xiosbase */
 261 #define _Mn     FLAG(U_NON_SPACING_MARK)
 262 #define _Me     FLAG(U_ENCLOSING_MARK)
 263 #define _Mc     FLAG(U_COMBINING_SPACING_MARK)
 264 #define _Nd     FLAG(U_DECIMAL_DIGIT_NUMBER)
 265 #define _Nl     FLAG(U_LETTER_NUMBER)
 266 #define _No     FLAG(U_OTHER_NUMBER)
 267 #define _Zs     FLAG(U_SPACE_SEPARATOR)
 268 #define _Zl     FLAG(U_LINE_SEPARATOR)
 269 #define _Zp     FLAG(U_PARAGRAPH_SEPARATOR)
 270 #define _Cc     FLAG(U_CONTROL_CHAR)
 271 #define _Cf     FLAG(U_FORMAT_CHAR)
 272 #define _Co     FLAG(U_PRIVATE_USE_CHAR)
 273 #define _Cs     FLAG(U_SURROGATE)
 274 #define _Pd     FLAG(U_DASH_PUNCTUATION)
 275 #define _Ps     FLAG(U_START_PUNCTUATION)
 276 /* #define _Pe     FLAG(U_END_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 xlocnum */
 277 /* #define _Pc     FLAG(U_CONNECTOR_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
 278 #define _Po     FLAG(U_OTHER_PUNCTUATION)
 279 #define _Sm     FLAG(U_MATH_SYMBOL)
 280 #define _Sc     FLAG(U_CURRENCY_SYMBOL)
 281 #define _Sk     FLAG(U_MODIFIER_SYMBOL)
 282 #define _So     FLAG(U_OTHER_SYMBOL)
 283 #define _Pi     FLAG(U_INITIAL_PUNCTUATION)
 284 /* #define _Pf     FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
 285
 286 /** Some code points. @internal */
 287 enum {
 288     TAB     =0x0009,
 289     LF      =0x000a,
 290     FF      =0x000c,
 291     CR      =0x000d,
 292     U_A     =0x0041,
 293     U_F     =0x0046,
 294     U_Z     =0x005a,
 295     U_a     =0x0061,
 296     U_f     =0x0066,
 297     U_z     =0x007a,
 298     DEL     =0x007f,
 299     NL      =0x0085,
 300     NBSP    =0x00a0,
 301     CGJ     =0x034f,
 302     FIGURESP=0x2007,
 303     HAIRSP  =0x200a,
 304     ZWNJ    =0x200c,
 305     ZWJ     =0x200d,
 306     RLM     =0x200f,
 307     NNBSP   =0x202f,
 308     WJ      =0x2060,
 309     INHSWAP =0x206a,
 310     NOMDIG  =0x206f,
 311     U_FW_A  =0xff21,
 312     U_FW_F  =0xff26,
 313     U_FW_Z  =0xff3a,
 314     U_FW_a  =0xff41,
 315     U_FW_f  =0xff46,
 316     U_FW_z  =0xff5a,
 317     ZWNBSP  =0xfeff
 318 };
 319
 320 /**
 321  * Get the maximum length of a (regular/1.0/extended) character name.
 322  * @return 0 if no character names available.
 323  */
 324 U_CAPI int32_t U_EXPORT2
 325 uprv_getMaxCharNameLength(void);
 326
 327 /**
 328  * Fills set with characters that are used in Unicode character names.
 329  * Includes all characters that are used in regular/Unicode 1.0/extended names.
 330  * Just empties the set if no character names are available.
 331  * @param sa USetAdder to receive characters.
 332  */
 333 U_CAPI void U_EXPORT2
 334 uprv_getCharNameCharacters(const USetAdder *sa);
 335
 336 /**
 337  * Constants for which data and implementation files provide which properties.
 338  * Used by UnicodeSet for service-specific property enumeration.
 339  * @internal
 340  */
 341 enum UPropertySource {
 342     /** No source, not a supported property. */
 343     UPROPS_SRC_NONE,
 344     /** From uchar.c/uprops.icu main trie */
 345     UPROPS_SRC_CHAR,
 346     /** From uchar.c/uprops.icu properties vectors trie */
 347     UPROPS_SRC_PROPSVEC,
 348     /** From unames.c/unames.icu */
 349     UPROPS_SRC_NAMES,
 350     /** From ucase.c/ucase.icu */
 351     UPROPS_SRC_CASE,
 352     /** From ubidi_props.c/ubidi.icu */
 353     UPROPS_SRC_BIDI,
 354     /** From uchar.c/uprops.icu main trie as well as properties vectors trie */
 355     UPROPS_SRC_CHAR_AND_PROPSVEC,
 356     /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
 357     UPROPS_SRC_CASE_AND_NORM,
 358     /** From normalizer2impl.cpp/nfc.nrm */
 359     UPROPS_SRC_NFC,
 360     /** From normalizer2impl.cpp/nfkc.nrm */
 361     UPROPS_SRC_NFKC,
 362     /** From normalizer2impl.cpp/nfkc_cf.nrm */
 363     UPROPS_SRC_NFKC_CF,
 364     /** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
 365     UPROPS_SRC_NFC_CANON_ITER,
 366     /** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
 367     UPROPS_SRC_COUNT
 368 };
 369 typedef enum UPropertySource UPropertySource;
 370
 371 /**
 372  * @see UPropertySource
 373  * @internal
 374  */
 375 U_CFUNC UPropertySource U_EXPORT2
 376 uprops_getSource(UProperty which);
 377
 378 /**
 379  * Enumerate uprops.icu's main data trie and add the
 380  * start of each range of same properties to the set.
 381  * @internal
 382  */
 383 U_CFUNC void U_EXPORT2
 384 uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
 385
 386 /**
 387  * Enumerate uprops.icu's properties vectors trie and add the
 388  * start of each range of same properties to the set.
 389  * @internal
 390  */
 391 U_CFUNC void U_EXPORT2
 392 upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
 393
 394 /**
 395  * Return a set of characters for property enumeration.
 396  * For each two consecutive characters (start, limit) in the set,
 397  * all of the properties for start..limit-1 are all the same.
 398  *
 399  * @param sa USetAdder to receive result. Existing contents are lost.
 400  * @internal
 401  */
 402 /*U_CFUNC void U_EXPORT2
 403 uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode);
 404 */
 405
 406 /**
 407  * Swap the ICU Unicode character names file. See uchar.c.
 408  * @internal
 409  */
 410 U_CAPI int32_t U_EXPORT2
 411 uchar_swapNames(const UDataSwapper *ds,
 412                 const void *inData, int32_t length, void *outData,
 413                 UErrorCode *pErrorCode);
 414
 415 #ifdef __cplusplus
 416
 417 U_NAMESPACE_BEGIN
 418
 419 class UnicodeSet;
 420
 421 // implemented in uniset_props.cpp
 422 U_CFUNC UnicodeSet *
 423 uniset_getUnicode32Instance(UErrorCode &errorCode);
 424
 425 U_NAMESPACE_END
 426
 427 #endif
 428
 429 #endif