00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023 #ifndef UCHAR_H
00024 #define UCHAR_H
00025
00026 #include "unicode/utypes.h"
00027
00028 U_CDECL_BEGIN
00029
00030
00031
00032
00042 #define U_UNICODE_VERSION "5.0"
00043
00124 #define UCHAR_MIN_VALUE 0
00125
00134 #define UCHAR_MAX_VALUE 0x10ffff
00135
00140 #define U_MASK(x) ((uint32_t)1<<(x))
00141
00142
00143
00144
00145
00146
00147
00148
00149
00150
00151
00152
00153
00154
00174 typedef enum UProperty {
00175
00176
00177
00178
00179
00180
00181
00182
00183
00186 UCHAR_ALPHABETIC=0,
00188 UCHAR_BINARY_START=UCHAR_ALPHABETIC,
00190 UCHAR_ASCII_HEX_DIGIT=1,
00194 UCHAR_BIDI_CONTROL=2,
00199 UCHAR_BIDI_MIRRORED=3,
00201 UCHAR_DASH=4,
00205 UCHAR_DEFAULT_IGNORABLE_CODE_POINT=5,
00208 UCHAR_DEPRECATED=6,
00211 UCHAR_DIACRITIC=7,
00215 UCHAR_EXTENDER=8,
00219 UCHAR_FULL_COMPOSITION_EXCLUSION=9,
00223 UCHAR_GRAPHEME_BASE=10,
00227 UCHAR_GRAPHEME_EXTEND=11,
00230 UCHAR_GRAPHEME_LINK=12,
00233 UCHAR_HEX_DIGIT=13,
00236 UCHAR_HYPHEN=14,
00241 UCHAR_ID_CONTINUE=15,
00245 UCHAR_ID_START=16,
00248 UCHAR_IDEOGRAPHIC=17,
00252 UCHAR_IDS_BINARY_OPERATOR=18,
00256 UCHAR_IDS_TRINARY_OPERATOR=19,
00259 UCHAR_JOIN_CONTROL=20,
00263 UCHAR_LOGICAL_ORDER_EXCEPTION=21,
00266 UCHAR_LOWERCASE=22,
00268 UCHAR_MATH=23,
00272 UCHAR_NONCHARACTER_CODE_POINT=24,
00274 UCHAR_QUOTATION_MARK=25,
00278 UCHAR_RADICAL=26,
00283 UCHAR_SOFT_DOTTED=27,
00287 UCHAR_TERMINAL_PUNCTUATION=28,
00291 UCHAR_UNIFIED_IDEOGRAPH=29,
00294 UCHAR_UPPERCASE=30,
00298 UCHAR_WHITE_SPACE=31,
00302 UCHAR_XID_CONTINUE=32,
00305 UCHAR_XID_START=33,
00309 UCHAR_CASE_SENSITIVE=34,
00314 UCHAR_S_TERM=35,
00320 UCHAR_VARIATION_SELECTOR=36,
00343 UCHAR_NFD_INERT=37,
00352 UCHAR_NFKD_INERT=38,
00361 UCHAR_NFC_INERT=39,
00370 UCHAR_NFKC_INERT=40,
00381 UCHAR_SEGMENT_STARTER=41,
00386 UCHAR_PATTERN_SYNTAX=42,
00391 UCHAR_PATTERN_WHITE_SPACE=43,
00396 UCHAR_POSIX_ALNUM=44,
00401 UCHAR_POSIX_BLANK=45,
00406 UCHAR_POSIX_GRAPH=46,
00411 UCHAR_POSIX_PRINT=47,
00416 UCHAR_POSIX_XDIGIT=48,
00418 UCHAR_BINARY_LIMIT=49,
00419
00422 UCHAR_BIDI_CLASS=0x1000,
00424 UCHAR_INT_START=UCHAR_BIDI_CLASS,
00427 UCHAR_BLOCK=0x1001,
00430 UCHAR_CANONICAL_COMBINING_CLASS=0x1002,
00433 UCHAR_DECOMPOSITION_TYPE=0x1003,
00437 UCHAR_EAST_ASIAN_WIDTH=0x1004,
00440 UCHAR_GENERAL_CATEGORY=0x1005,
00443 UCHAR_JOINING_GROUP=0x1006,
00446 UCHAR_JOINING_TYPE=0x1007,
00449 UCHAR_LINE_BREAK=0x1008,
00452 UCHAR_NUMERIC_TYPE=0x1009,
00455 UCHAR_SCRIPT=0x100A,
00458 UCHAR_HANGUL_SYLLABLE_TYPE=0x100B,
00461 UCHAR_NFD_QUICK_CHECK=0x100C,
00464 UCHAR_NFKD_QUICK_CHECK=0x100D,
00467 UCHAR_NFC_QUICK_CHECK=0x100E,
00470 UCHAR_NFKC_QUICK_CHECK=0x100F,
00477 UCHAR_LEAD_CANONICAL_COMBINING_CLASS=0x1010,
00484 UCHAR_TRAIL_CANONICAL_COMBINING_CLASS=0x1011,
00489 UCHAR_GRAPHEME_CLUSTER_BREAK=0x1012,
00494 UCHAR_SENTENCE_BREAK=0x1013,
00499 UCHAR_WORD_BREAK=0x1014,
00501 UCHAR_INT_LIMIT=0x1015,
00502
00511 UCHAR_GENERAL_CATEGORY_MASK=0x2000,
00513 UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK,
00515 UCHAR_MASK_LIMIT=0x2001,
00516
00519 UCHAR_NUMERIC_VALUE=0x3000,
00521 UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE,
00523 UCHAR_DOUBLE_LIMIT=0x3001,
00524
00527 UCHAR_AGE=0x4000,
00529 UCHAR_STRING_START=UCHAR_AGE,
00532 UCHAR_BIDI_MIRRORING_GLYPH=0x4001,
00535 UCHAR_CASE_FOLDING=0x4002,
00538 UCHAR_ISO_COMMENT=0x4003,
00541 UCHAR_LOWERCASE_MAPPING=0x4004,
00544 UCHAR_NAME=0x4005,
00547 UCHAR_SIMPLE_CASE_FOLDING=0x4006,
00550 UCHAR_SIMPLE_LOWERCASE_MAPPING=0x4007,
00553 UCHAR_SIMPLE_TITLECASE_MAPPING=0x4008,
00556 UCHAR_SIMPLE_UPPERCASE_MAPPING=0x4009,
00559 UCHAR_TITLECASE_MAPPING=0x400A,
00562 UCHAR_UNICODE_1_NAME=0x400B,
00565 UCHAR_UPPERCASE_MAPPING=0x400C,
00567 UCHAR_STRING_LIMIT=0x400D,
00568
00570 UCHAR_INVALID_CODE = -1
00571 } UProperty;
00572
00578 typedef enum UCharCategory
00579 {
00583 U_UNASSIGNED = 0,
00585 U_GENERAL_OTHER_TYPES = 0,
00587 U_UPPERCASE_LETTER = 1,
00589 U_LOWERCASE_LETTER = 2,
00591 U_TITLECASE_LETTER = 3,
00593 U_MODIFIER_LETTER = 4,
00595 U_OTHER_LETTER = 5,
00597 U_NON_SPACING_MARK = 6,
00599 U_ENCLOSING_MARK = 7,
00601 U_COMBINING_SPACING_MARK = 8,
00603 U_DECIMAL_DIGIT_NUMBER = 9,
00605 U_LETTER_NUMBER = 10,
00607 U_OTHER_NUMBER = 11,
00609 U_SPACE_SEPARATOR = 12,
00611 U_LINE_SEPARATOR = 13,
00613 U_PARAGRAPH_SEPARATOR = 14,
00615 U_CONTROL_CHAR = 15,
00617 U_FORMAT_CHAR = 16,
00619 U_PRIVATE_USE_CHAR = 17,
00621 U_SURROGATE = 18,
00623 U_DASH_PUNCTUATION = 19,
00625 U_START_PUNCTUATION = 20,
00627 U_END_PUNCTUATION = 21,
00629 U_CONNECTOR_PUNCTUATION = 22,
00631 U_OTHER_PUNCTUATION = 23,
00633 U_MATH_SYMBOL = 24,
00635 U_CURRENCY_SYMBOL = 25,
00637 U_MODIFIER_SYMBOL = 26,
00639 U_OTHER_SYMBOL = 27,
00641 U_INITIAL_PUNCTUATION = 28,
00643 U_FINAL_PUNCTUATION = 29,
00645 U_CHAR_CATEGORY_COUNT
00646 } UCharCategory;
00647
00662 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
00663
00665 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
00666
00667 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
00668
00669 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
00670
00671 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
00672
00673 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
00674
00676 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
00677
00678 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
00679
00680 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
00681
00683 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
00684
00685 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
00686
00687 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
00688
00690 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
00691
00692 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
00693
00694 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
00695
00697 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
00698
00699 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
00700
00701 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
00702
00703 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
00704
00706 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
00707
00708 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
00709
00710 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
00711
00712 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
00713
00714 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
00715
00717 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
00718
00719 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
00720
00721 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
00722
00723 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
00724
00726 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
00727
00728 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
00729
00730
00732 #define U_GC_L_MASK \
00733 (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
00734
00736 #define U_GC_LC_MASK \
00737 (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
00738
00740 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
00741
00743 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
00744
00746 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
00747
00749 #define U_GC_C_MASK \
00750 (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
00751
00753 #define U_GC_P_MASK \
00754 (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
00755 U_GC_PI_MASK|U_GC_PF_MASK)
00756
00758 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
00759
00764 typedef enum UCharDirection {
00768 U_LEFT_TO_RIGHT = 0,
00770 U_RIGHT_TO_LEFT = 1,
00772 U_EUROPEAN_NUMBER = 2,
00774 U_EUROPEAN_NUMBER_SEPARATOR = 3,
00776 U_EUROPEAN_NUMBER_TERMINATOR = 4,
00778 U_ARABIC_NUMBER = 5,
00780 U_COMMON_NUMBER_SEPARATOR = 6,
00782 U_BLOCK_SEPARATOR = 7,
00784 U_SEGMENT_SEPARATOR = 8,
00786 U_WHITE_SPACE_NEUTRAL = 9,
00788 U_OTHER_NEUTRAL = 10,
00790 U_LEFT_TO_RIGHT_EMBEDDING = 11,
00792 U_LEFT_TO_RIGHT_OVERRIDE = 12,
00794 U_RIGHT_TO_LEFT_ARABIC = 13,
00796 U_RIGHT_TO_LEFT_EMBEDDING = 14,
00798 U_RIGHT_TO_LEFT_OVERRIDE = 15,
00800 U_POP_DIRECTIONAL_FORMAT = 16,
00802 U_DIR_NON_SPACING_MARK = 17,
00804 U_BOUNDARY_NEUTRAL = 18,
00806 U_CHAR_DIRECTION_COUNT
00807 } UCharDirection;
00808
00813 enum UBlockCode {
00814
00816 UBLOCK_NO_BLOCK = 0,
00817
00819 UBLOCK_BASIC_LATIN = 1,
00820
00822 UBLOCK_LATIN_1_SUPPLEMENT=2,
00823
00825 UBLOCK_LATIN_EXTENDED_A =3,
00826
00828 UBLOCK_LATIN_EXTENDED_B =4,
00829
00831 UBLOCK_IPA_EXTENSIONS =5,
00832
00834 UBLOCK_SPACING_MODIFIER_LETTERS =6,
00835
00837 UBLOCK_COMBINING_DIACRITICAL_MARKS =7,
00838
00843 UBLOCK_GREEK =8,
00844
00846 UBLOCK_CYRILLIC =9,
00847
00849 UBLOCK_ARMENIAN =10,
00850
00852 UBLOCK_HEBREW =11,
00853
00855 UBLOCK_ARABIC =12,
00856
00858 UBLOCK_SYRIAC =13,
00859
00861 UBLOCK_THAANA =14,
00862
00864 UBLOCK_DEVANAGARI =15,
00865
00867 UBLOCK_BENGALI =16,
00868
00870 UBLOCK_GURMUKHI =17,
00871
00873 UBLOCK_GUJARATI =18,
00874
00876 UBLOCK_ORIYA =19,
00877
00879 UBLOCK_TAMIL =20,
00880
00882 UBLOCK_TELUGU =21,
00883
00885 UBLOCK_KANNADA =22,
00886
00888 UBLOCK_MALAYALAM =23,
00889
00891 UBLOCK_SINHALA =24,
00892
00894 UBLOCK_THAI =25,
00895
00897 UBLOCK_LAO =26,
00898
00900 UBLOCK_TIBETAN =27,
00901
00903 UBLOCK_MYANMAR =28,
00904
00906 UBLOCK_GEORGIAN =29,
00907
00909 UBLOCK_HANGUL_JAMO =30,
00910
00912 UBLOCK_ETHIOPIC =31,
00913
00915 UBLOCK_CHEROKEE =32,
00916
00918 UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS =33,
00919
00921 UBLOCK_OGHAM =34,
00922
00924 UBLOCK_RUNIC =35,
00925
00927 UBLOCK_KHMER =36,
00928
00930 UBLOCK_MONGOLIAN =37,
00931
00933 UBLOCK_LATIN_EXTENDED_ADDITIONAL =38,
00934
00936 UBLOCK_GREEK_EXTENDED =39,
00937
00939 UBLOCK_GENERAL_PUNCTUATION =40,
00940
00942 UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS =41,
00943
00945 UBLOCK_CURRENCY_SYMBOLS =42,
00946
00951 UBLOCK_COMBINING_MARKS_FOR_SYMBOLS =43,
00952
00954 UBLOCK_LETTERLIKE_SYMBOLS =44,
00955
00957 UBLOCK_NUMBER_FORMS =45,
00958
00960 UBLOCK_ARROWS =46,
00961
00963 UBLOCK_MATHEMATICAL_OPERATORS =47,
00964
00966 UBLOCK_MISCELLANEOUS_TECHNICAL =48,
00967
00969 UBLOCK_CONTROL_PICTURES =49,
00970
00972 UBLOCK_OPTICAL_CHARACTER_RECOGNITION =50,
00973
00975 UBLOCK_ENCLOSED_ALPHANUMERICS =51,
00976
00978 UBLOCK_BOX_DRAWING =52,
00979
00981 UBLOCK_BLOCK_ELEMENTS =53,
00982
00984 UBLOCK_GEOMETRIC_SHAPES =54,
00985
00987 UBLOCK_MISCELLANEOUS_SYMBOLS =55,
00988
00990 UBLOCK_DINGBATS =56,
00991
00993 UBLOCK_BRAILLE_PATTERNS =57,
00994
00996 UBLOCK_CJK_RADICALS_SUPPLEMENT =58,
00997
00999 UBLOCK_KANGXI_RADICALS =59,
01000
01002 UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS =60,
01003
01005 UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION =61,
01006
01008 UBLOCK_HIRAGANA =62,
01009
01011 UBLOCK_KATAKANA =63,
01012
01014 UBLOCK_BOPOMOFO =64,
01015
01017 UBLOCK_HANGUL_COMPATIBILITY_JAMO =65,
01018
01020 UBLOCK_KANBUN =66,
01021
01023 UBLOCK_BOPOMOFO_EXTENDED =67,
01024
01026 UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS =68,
01027
01029 UBLOCK_CJK_COMPATIBILITY =69,
01030
01032 UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A =70,
01033
01035 UBLOCK_CJK_UNIFIED_IDEOGRAPHS =71,
01036
01038 UBLOCK_YI_SYLLABLES =72,
01039
01041 UBLOCK_YI_RADICALS =73,
01042
01044 UBLOCK_HANGUL_SYLLABLES =74,
01045
01047 UBLOCK_HIGH_SURROGATES =75,
01048
01050 UBLOCK_HIGH_PRIVATE_USE_SURROGATES =76,
01051
01053 UBLOCK_LOW_SURROGATES =77,
01054
01064 UBLOCK_PRIVATE_USE = 78,
01074 UBLOCK_PRIVATE_USE_AREA =UBLOCK_PRIVATE_USE,
01075
01077 UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS =79,
01078
01080 UBLOCK_ALPHABETIC_PRESENTATION_FORMS =80,
01081
01083 UBLOCK_ARABIC_PRESENTATION_FORMS_A =81,
01084
01086 UBLOCK_COMBINING_HALF_MARKS =82,
01087
01089 UBLOCK_CJK_COMPATIBILITY_FORMS =83,
01090
01092 UBLOCK_SMALL_FORM_VARIANTS =84,
01093
01095 UBLOCK_ARABIC_PRESENTATION_FORMS_B =85,
01096
01098 UBLOCK_SPECIALS =86,
01099
01101 UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS =87,
01102
01103
01104
01106 UBLOCK_OLD_ITALIC = 88 ,
01108 UBLOCK_GOTHIC = 89 ,
01110 UBLOCK_DESERET = 90 ,
01112 UBLOCK_BYZANTINE_MUSICAL_SYMBOLS = 91 ,
01114 UBLOCK_MUSICAL_SYMBOLS = 92 ,
01116 UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 93 ,
01118 UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B = 94 ,
01120 UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 95 ,
01122 UBLOCK_TAGS = 96,
01123
01124
01125
01130 UBLOCK_CYRILLIC_SUPPLEMENTARY = 97,
01132 UBLOCK_CYRILLIC_SUPPLEMENT = UBLOCK_CYRILLIC_SUPPLEMENTARY,
01134 UBLOCK_TAGALOG = 98,
01136 UBLOCK_HANUNOO = 99,
01138 UBLOCK_BUHID = 100,
01140 UBLOCK_TAGBANWA = 101,
01142 UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 102,
01144 UBLOCK_SUPPLEMENTAL_ARROWS_A = 103,
01146 UBLOCK_SUPPLEMENTAL_ARROWS_B = 104,
01148 UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 105,
01150 UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 106,
01152 UBLOCK_KATAKANA_PHONETIC_EXTENSIONS = 107,
01154 UBLOCK_VARIATION_SELECTORS = 108,
01156 UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 109,
01158 UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 110,
01159
01160
01161
01163 UBLOCK_LIMBU = 111,
01165 UBLOCK_TAI_LE = 112,
01167 UBLOCK_KHMER_SYMBOLS = 113,
01169 UBLOCK_PHONETIC_EXTENSIONS = 114,
01171 UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 115,
01173 UBLOCK_YIJING_HEXAGRAM_SYMBOLS = 116,
01175 UBLOCK_LINEAR_B_SYLLABARY = 117,
01177 UBLOCK_LINEAR_B_IDEOGRAMS = 118,
01179 UBLOCK_AEGEAN_NUMBERS = 119,
01181 UBLOCK_UGARITIC = 120,
01183 UBLOCK_SHAVIAN = 121,
01185 UBLOCK_OSMANYA = 122,
01187 UBLOCK_CYPRIOT_SYLLABARY = 123,
01189 UBLOCK_TAI_XUAN_JING_SYMBOLS = 124,
01191 UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125,
01192
01193
01194
01196 UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION = 126,
01198 UBLOCK_ANCIENT_GREEK_NUMBERS = 127,
01200 UBLOCK_ARABIC_SUPPLEMENT = 128,
01202 UBLOCK_BUGINESE = 129,
01204 UBLOCK_CJK_STROKES = 130,
01206 UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 131,
01208 UBLOCK_COPTIC = 132,
01210 UBLOCK_ETHIOPIC_EXTENDED = 133,
01212 UBLOCK_ETHIOPIC_SUPPLEMENT = 134,
01214 UBLOCK_GEORGIAN_SUPPLEMENT = 135,
01216 UBLOCK_GLAGOLITIC = 136,
01218 UBLOCK_KHAROSHTHI = 137,
01220 UBLOCK_MODIFIER_TONE_LETTERS = 138,
01222 UBLOCK_NEW_TAI_LUE = 139,
01224 UBLOCK_OLD_PERSIAN = 140,
01226 UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT = 141,
01228 UBLOCK_SUPPLEMENTAL_PUNCTUATION = 142,
01230 UBLOCK_SYLOTI_NAGRI = 143,
01232 UBLOCK_TIFINAGH = 144,
01234 UBLOCK_VERTICAL_FORMS = 145,
01235
01236
01237
01239 UBLOCK_NKO = 146,
01241 UBLOCK_BALINESE = 147,
01243 UBLOCK_LATIN_EXTENDED_C = 148,
01245 UBLOCK_LATIN_EXTENDED_D = 149,
01247 UBLOCK_PHAGS_PA = 150,
01249 UBLOCK_PHOENICIAN = 151,
01251 UBLOCK_CUNEIFORM = 152,
01253 UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 153,
01255 UBLOCK_COUNTING_ROD_NUMERALS = 154,
01256
01258 UBLOCK_COUNT = 155,
01259
01261 UBLOCK_INVALID_CODE=-1
01262 };
01263
01265 typedef enum UBlockCode UBlockCode;
01266
01274 typedef enum UEastAsianWidth {
01275 U_EA_NEUTRAL,
01276 U_EA_AMBIGUOUS,
01277 U_EA_HALFWIDTH,
01278 U_EA_FULLWIDTH,
01279 U_EA_NARROW,
01280 U_EA_WIDE,
01281 U_EA_COUNT
01282 } UEastAsianWidth;
01283
01284
01285
01286
01287
01299 typedef enum UCharNameChoice {
01300 U_UNICODE_CHAR_NAME,
01301 U_UNICODE_10_CHAR_NAME,
01302 U_EXTENDED_CHAR_NAME,
01303 U_CHAR_NAME_CHOICE_COUNT
01304 } UCharNameChoice;
01305
01319 typedef enum UPropertyNameChoice {
01320 U_SHORT_PROPERTY_NAME,
01321 U_LONG_PROPERTY_NAME,
01322 U_PROPERTY_NAME_CHOICE_COUNT
01323 } UPropertyNameChoice;
01324
01331 typedef enum UDecompositionType {
01332 U_DT_NONE,
01333 U_DT_CANONICAL,
01334 U_DT_COMPAT,
01335 U_DT_CIRCLE,
01336 U_DT_FINAL,
01337 U_DT_FONT,
01338 U_DT_FRACTION,
01339 U_DT_INITIAL,
01340 U_DT_ISOLATED,
01341 U_DT_MEDIAL,
01342 U_DT_NARROW,
01343 U_DT_NOBREAK,
01344 U_DT_SMALL,
01345 U_DT_SQUARE,
01346 U_DT_SUB,
01347 U_DT_SUPER,
01348 U_DT_VERTICAL,
01349 U_DT_WIDE,
01350 U_DT_COUNT
01351 } UDecompositionType;
01352
01359 typedef enum UJoiningType {
01360 U_JT_NON_JOINING,
01361 U_JT_JOIN_CAUSING,
01362 U_JT_DUAL_JOINING,
01363 U_JT_LEFT_JOINING,
01364 U_JT_RIGHT_JOINING,
01365 U_JT_TRANSPARENT,
01366 U_JT_COUNT
01367 } UJoiningType;
01368
01375 typedef enum UJoiningGroup {
01376 U_JG_NO_JOINING_GROUP,
01377 U_JG_AIN,
01378 U_JG_ALAPH,
01379 U_JG_ALEF,
01380 U_JG_BEH,
01381 U_JG_BETH,
01382 U_JG_DAL,
01383 U_JG_DALATH_RISH,
01384 U_JG_E,
01385 U_JG_FEH,
01386 U_JG_FINAL_SEMKATH,
01387 U_JG_GAF,
01388 U_JG_GAMAL,
01389 U_JG_HAH,
01390 U_JG_HAMZA_ON_HEH_GOAL,
01391 U_JG_HE,
01392 U_JG_HEH,
01393 U_JG_HEH_GOAL,
01394 U_JG_HETH,
01395 U_JG_KAF,
01396 U_JG_KAPH,
01397 U_JG_KNOTTED_HEH,
01398 U_JG_LAM,
01399 U_JG_LAMADH,
01400 U_JG_MEEM,
01401 U_JG_MIM,
01402 U_JG_NOON,
01403 U_JG_NUN,
01404 U_JG_PE,
01405 U_JG_QAF,
01406 U_JG_QAPH,
01407 U_JG_REH,
01408 U_JG_REVERSED_PE,
01409 U_JG_SAD,
01410 U_JG_SADHE,
01411 U_JG_SEEN,
01412 U_JG_SEMKATH,
01413 U_JG_SHIN,
01414 U_JG_SWASH_KAF,
01415 U_JG_SYRIAC_WAW,
01416 U_JG_TAH,
01417 U_JG_TAW,
01418 U_JG_TEH_MARBUTA,
01419 U_JG_TETH,
01420 U_JG_WAW,
01421 U_JG_YEH,
01422 U_JG_YEH_BARREE,
01423 U_JG_YEH_WITH_TAIL,
01424 U_JG_YUDH,
01425 U_JG_YUDH_HE,
01426 U_JG_ZAIN,
01427 U_JG_FE,
01428 U_JG_KHAPH,
01429 U_JG_ZHAIN,
01430 U_JG_COUNT
01431 } UJoiningGroup;
01432
01439 typedef enum UGraphemeClusterBreak {
01440 U_GCB_OTHER = 0,
01441 U_GCB_CONTROL = 1,
01442 U_GCB_CR = 2,
01443 U_GCB_EXTEND = 3,
01444 U_GCB_L = 4,
01445 U_GCB_LF = 5,
01446 U_GCB_LV = 6,
01447 U_GCB_LVT = 7,
01448 U_GCB_T = 8,
01449 U_GCB_V = 9,
01450 U_GCB_COUNT = 10
01451 } UGraphemeClusterBreak;
01452
01460 typedef enum UWordBreakValues {
01461 U_WB_OTHER = 0,
01462 U_WB_ALETTER = 1,
01463 U_WB_FORMAT = 2,
01464 U_WB_KATAKANA = 3,
01465 U_WB_MIDLETTER = 4,
01466 U_WB_MIDNUM = 5,
01467 U_WB_NUMERIC = 6,
01468 U_WB_EXTENDNUMLET = 7,
01469 U_WB_COUNT = 8
01470 } UWordBreakValues;
01471
01478 typedef enum USentenceBreak {
01479 U_SB_OTHER = 0,
01480 U_SB_ATERM = 1,
01481 U_SB_CLOSE = 2,
01482 U_SB_FORMAT = 3,
01483 U_SB_LOWER = 4,
01484 U_SB_NUMERIC = 5,
01485 U_SB_OLETTER = 6,
01486 U_SB_SEP = 7,
01487 U_SB_SP = 8,
01488 U_SB_STERM = 9,
01489 U_SB_UPPER = 10,
01490 U_SB_COUNT = 11
01491 } USentenceBreak;
01492
01499 typedef enum ULineBreak {
01500 U_LB_UNKNOWN = 0,
01501 U_LB_AMBIGUOUS = 1,
01502 U_LB_ALPHABETIC = 2,
01503 U_LB_BREAK_BOTH = 3,
01504 U_LB_BREAK_AFTER = 4,
01505 U_LB_BREAK_BEFORE = 5,
01506 U_LB_MANDATORY_BREAK = 6,
01507 U_LB_CONTINGENT_BREAK = 7,
01508 U_LB_CLOSE_PUNCTUATION = 8,
01509 U_LB_COMBINING_MARK = 9,
01510 U_LB_CARRIAGE_RETURN = 10,
01511 U_LB_EXCLAMATION = 11,
01512 U_LB_GLUE = 12,
01513 U_LB_HYPHEN = 13,
01514 U_LB_IDEOGRAPHIC = 14,
01515 U_LB_INSEPERABLE = 15,
01517 U_LB_INSEPARABLE=U_LB_INSEPERABLE,
01518 U_LB_INFIX_NUMERIC = 16,
01519 U_LB_LINE_FEED = 17,
01520 U_LB_NONSTARTER = 18,
01521 U_LB_NUMERIC = 19,
01522 U_LB_OPEN_PUNCTUATION = 20,
01523 U_LB_POSTFIX_NUMERIC = 21,
01524 U_LB_PREFIX_NUMERIC = 22,
01525 U_LB_QUOTATION = 23,
01526 U_LB_COMPLEX_CONTEXT = 24,
01527 U_LB_SURROGATE = 25,
01528 U_LB_SPACE = 26,
01529 U_LB_BREAK_SYMBOLS = 27,
01530 U_LB_ZWSPACE = 28,
01531 U_LB_NEXT_LINE = 29,
01532 U_LB_WORD_JOINER = 30,
01533 U_LB_H2 = 31,
01534 U_LB_H3 = 32,
01535 U_LB_JL = 33,
01536 U_LB_JT = 34,
01537 U_LB_JV = 35,
01538 U_LB_COUNT = 36
01539 } ULineBreak;
01540
01547 typedef enum UNumericType {
01548 U_NT_NONE,
01549 U_NT_DECIMAL,
01550 U_NT_DIGIT,
01551 U_NT_NUMERIC,
01552 U_NT_COUNT
01553 } UNumericType;
01554
01561 typedef enum UHangulSyllableType {
01562 U_HST_NOT_APPLICABLE,
01563 U_HST_LEADING_JAMO,
01564 U_HST_VOWEL_JAMO,
01565 U_HST_TRAILING_JAMO,
01566 U_HST_LV_SYLLABLE,
01567 U_HST_LVT_SYLLABLE,
01568 U_HST_COUNT
01569 } UHangulSyllableType;
01570
01597 U_STABLE UBool U_EXPORT2
01598 u_hasBinaryProperty(UChar32 c, UProperty which);
01599
01612 U_STABLE UBool U_EXPORT2
01613 u_isUAlphabetic(UChar32 c);
01614
01627 U_STABLE UBool U_EXPORT2
01628 u_isULowercase(UChar32 c);
01629
01642 U_STABLE UBool U_EXPORT2
01643 u_isUUppercase(UChar32 c);
01644
01663 U_STABLE UBool U_EXPORT2
01664 u_isUWhiteSpace(UChar32 c);
01665
01703 U_STABLE int32_t U_EXPORT2
01704 u_getIntPropertyValue(UChar32 c, UProperty which);
01705
01724 U_STABLE int32_t U_EXPORT2
01725 u_getIntPropertyMinValue(UProperty which);
01726
01753 U_STABLE int32_t U_EXPORT2
01754 u_getIntPropertyMaxValue(UProperty which);
01755
01776 U_STABLE double U_EXPORT2
01777 u_getNumericValue(UChar32 c);
01778
01786 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
01787
01811 U_STABLE UBool U_EXPORT2
01812 u_islower(UChar32 c);
01813
01838 U_STABLE UBool U_EXPORT2
01839 u_isupper(UChar32 c);
01840
01855 U_STABLE UBool U_EXPORT2
01856 u_istitle(UChar32 c);
01857
01876 U_STABLE UBool U_EXPORT2
01877 u_isdigit(UChar32 c);
01878
01897 U_STABLE UBool U_EXPORT2
01898 u_isalpha(UChar32 c);
01899
01918 U_STABLE UBool U_EXPORT2
01919 u_isalnum(UChar32 c);
01920
01941 U_STABLE UBool U_EXPORT2
01942 u_isxdigit(UChar32 c);
01943
01957 U_STABLE UBool U_EXPORT2
01958 u_ispunct(UChar32 c);
01959
01976 U_STABLE UBool U_EXPORT2
01977 u_isgraph(UChar32 c);
01978
02005 U_STABLE UBool U_EXPORT2
02006 u_isblank(UChar32 c);
02007
02030 U_STABLE UBool U_EXPORT2
02031 u_isdefined(UChar32 c);
02032
02051 U_STABLE UBool U_EXPORT2
02052 u_isspace(UChar32 c);
02053
02072 U_STABLE UBool U_EXPORT2
02073 u_isJavaSpaceChar(UChar32 c);
02074
02106 U_STABLE UBool U_EXPORT2
02107 u_isWhitespace(UChar32 c);
02108
02130 U_STABLE UBool U_EXPORT2
02131 u_iscntrl(UChar32 c);
02132
02145 U_STABLE UBool U_EXPORT2
02146 u_isISOControl(UChar32 c);
02147
02163 U_STABLE UBool U_EXPORT2
02164 u_isprint(UChar32 c);
02165
02184 U_STABLE UBool U_EXPORT2
02185 u_isbase(UChar32 c);
02186
02203 U_STABLE UCharDirection U_EXPORT2
02204 u_charDirection(UChar32 c);
02205
02221 U_STABLE UBool U_EXPORT2
02222 u_isMirrored(UChar32 c);
02223
02243 U_STABLE UChar32 U_EXPORT2
02244 u_charMirror(UChar32 c);
02245
02257 U_STABLE int8_t U_EXPORT2
02258 u_charType(UChar32 c);
02259
02273 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
02274
02292 typedef UBool U_CALLCONV
02293 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
02294
02314 U_STABLE void U_EXPORT2
02315 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
02316
02317 #if !UCONFIG_NO_NORMALIZATION
02318
02326 U_STABLE uint8_t U_EXPORT2
02327 u_getCombiningClass(UChar32 c);
02328
02329 #endif
02330
02354 U_STABLE int32_t U_EXPORT2
02355 u_charDigitValue(UChar32 c);
02356
02366 U_STABLE UBlockCode U_EXPORT2
02367 ublock_getCode(UChar32 c);
02368
02401 U_STABLE int32_t U_EXPORT2
02402 u_charName(UChar32 code, UCharNameChoice nameChoice,
02403 char *buffer, int32_t bufferLength,
02404 UErrorCode *pErrorCode);
02405
02428 U_STABLE int32_t U_EXPORT2
02429 u_getISOComment(UChar32 c,
02430 char *dest, int32_t destCapacity,
02431 UErrorCode *pErrorCode);
02432
02453 U_STABLE UChar32 U_EXPORT2
02454 u_charFromName(UCharNameChoice nameChoice,
02455 const char *name,
02456 UErrorCode *pErrorCode);
02457
02475 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
02476 UChar32 code,
02477 UCharNameChoice nameChoice,
02478 const char *name,
02479 int32_t length);
02480
02502 U_STABLE void U_EXPORT2
02503 u_enumCharNames(UChar32 start, UChar32 limit,
02504 UEnumCharNamesFn *fn,
02505 void *context,
02506 UCharNameChoice nameChoice,
02507 UErrorCode *pErrorCode);
02508
02540 U_STABLE const char* U_EXPORT2
02541 u_getPropertyName(UProperty property,
02542 UPropertyNameChoice nameChoice);
02543
02563 U_STABLE UProperty U_EXPORT2
02564 u_getPropertyEnum(const char* alias);
02565
02613 U_STABLE const char* U_EXPORT2
02614 u_getPropertyValueName(UProperty property,
02615 int32_t value,
02616 UPropertyNameChoice nameChoice);
02617
02649 U_STABLE int32_t U_EXPORT2
02650 u_getPropertyValueEnum(UProperty property,
02651 const char* alias);
02652
02670 U_STABLE UBool U_EXPORT2
02671 u_isIDStart(UChar32 c);
02672
02694 U_STABLE UBool U_EXPORT2
02695 u_isIDPart(UChar32 c);
02696
02719 U_STABLE UBool U_EXPORT2
02720 u_isIDIgnorable(UChar32 c);
02721
02738 U_STABLE UBool U_EXPORT2
02739 u_isJavaIDStart(UChar32 c);
02740
02759 U_STABLE UBool U_EXPORT2
02760 u_isJavaIDPart(UChar32 c);
02761
02784 U_STABLE UChar32 U_EXPORT2
02785 u_tolower(UChar32 c);
02786
02809 U_STABLE UChar32 U_EXPORT2
02810 u_toupper(UChar32 c);
02811
02834 U_STABLE UChar32 U_EXPORT2
02835 u_totitle(UChar32 c);
02836
02838 #define U_FOLD_CASE_DEFAULT 0
02839
02856 #define U_FOLD_CASE_EXCLUDE_SPECIAL_I 1
02857
02880 U_STABLE UChar32 U_EXPORT2
02881 u_foldCase(UChar32 c, uint32_t options);
02882
02921 U_STABLE int32_t U_EXPORT2
02922 u_digit(UChar32 ch, int8_t radix);
02923
02952 U_STABLE UChar32 U_EXPORT2
02953 u_forDigit(int32_t digit, int8_t radix);
02954
02969 U_STABLE void U_EXPORT2
02970 u_charAge(UChar32 c, UVersionInfo versionArray);
02971
02983 U_STABLE void U_EXPORT2
02984 u_getUnicodeVersion(UVersionInfo versionArray);
02985
03007 U_STABLE int32_t U_EXPORT2
03008 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
03009
03010 U_CDECL_END
03011
03012 #endif
03013