mirror of
https://github.com/unicode-org/icu.git
synced 2025-04-04 13:05:31 +00:00
1375 lines
68 KiB
C
1375 lines
68 KiB
C
// © 2016 and later: Unicode, Inc. and others.
|
|
// License & terms of use: http://www.unicode.org/copyright.html
|
|
/**
|
|
* Copyright (C) 2002-2016, International Business Machines Corporation and
|
|
* others. All Rights Reserved.
|
|
*
|
|
* machine-generated by: icu/tools/unicode/py/preparseucd.py
|
|
*/
|
|
|
|
#define UNICODE_VERSION { 16, 0, 0, 0 }
|
|
|
|
static const Value VALUES_binprop[2] = {
|
|
Value(0, "N No F False"),
|
|
Value(1, "Y Yes T True"),
|
|
};
|
|
|
|
static const Value VALUES_bc[23] = {
|
|
Value(U_LEFT_TO_RIGHT, "L Left_To_Right"),
|
|
Value(U_RIGHT_TO_LEFT, "R Right_To_Left"),
|
|
Value(U_EUROPEAN_NUMBER, "EN European_Number"),
|
|
Value(U_EUROPEAN_NUMBER_SEPARATOR, "ES European_Separator"),
|
|
Value(U_EUROPEAN_NUMBER_TERMINATOR, "ET European_Terminator"),
|
|
Value(U_ARABIC_NUMBER, "AN Arabic_Number"),
|
|
Value(U_COMMON_NUMBER_SEPARATOR, "CS Common_Separator"),
|
|
Value(U_BLOCK_SEPARATOR, "B Paragraph_Separator"),
|
|
Value(U_SEGMENT_SEPARATOR, "S Segment_Separator"),
|
|
Value(U_WHITE_SPACE_NEUTRAL, "WS White_Space"),
|
|
Value(U_OTHER_NEUTRAL, "ON Other_Neutral"),
|
|
Value(U_LEFT_TO_RIGHT_EMBEDDING, "LRE Left_To_Right_Embedding"),
|
|
Value(U_LEFT_TO_RIGHT_OVERRIDE, "LRO Left_To_Right_Override"),
|
|
Value(U_RIGHT_TO_LEFT_ARABIC, "AL Arabic_Letter"),
|
|
Value(U_RIGHT_TO_LEFT_EMBEDDING, "RLE Right_To_Left_Embedding"),
|
|
Value(U_RIGHT_TO_LEFT_OVERRIDE, "RLO Right_To_Left_Override"),
|
|
Value(U_POP_DIRECTIONAL_FORMAT, "PDF Pop_Directional_Format"),
|
|
Value(U_DIR_NON_SPACING_MARK, "NSM Nonspacing_Mark"),
|
|
Value(U_BOUNDARY_NEUTRAL, "BN Boundary_Neutral"),
|
|
Value(U_FIRST_STRONG_ISOLATE, "FSI First_Strong_Isolate"),
|
|
Value(U_LEFT_TO_RIGHT_ISOLATE, "LRI Left_To_Right_Isolate"),
|
|
Value(U_RIGHT_TO_LEFT_ISOLATE, "RLI Right_To_Left_Isolate"),
|
|
Value(U_POP_DIRECTIONAL_ISOLATE, "PDI Pop_Directional_Isolate"),
|
|
};
|
|
|
|
static const Value VALUES_blk[339] = {
|
|
Value(UBLOCK_NO_BLOCK, "NB No_Block"),
|
|
Value(UBLOCK_BASIC_LATIN, "ASCII Basic_Latin"),
|
|
Value(UBLOCK_LATIN_1_SUPPLEMENT, "Latin_1_Sup Latin_1_Supplement Latin_1"),
|
|
Value(UBLOCK_LATIN_EXTENDED_A, "Latin_Ext_A Latin_Extended_A"),
|
|
Value(UBLOCK_LATIN_EXTENDED_B, "Latin_Ext_B Latin_Extended_B"),
|
|
Value(UBLOCK_IPA_EXTENSIONS, "IPA_Ext IPA_Extensions"),
|
|
Value(UBLOCK_SPACING_MODIFIER_LETTERS, "Modifier_Letters Spacing_Modifier_Letters"),
|
|
Value(UBLOCK_COMBINING_DIACRITICAL_MARKS, "Diacriticals Combining_Diacritical_Marks"),
|
|
Value(UBLOCK_GREEK, "Greek Greek_And_Coptic"),
|
|
Value(UBLOCK_CYRILLIC, "Cyrillic Cyrillic"),
|
|
Value(UBLOCK_ARMENIAN, "Armenian Armenian"),
|
|
Value(UBLOCK_HEBREW, "Hebrew Hebrew"),
|
|
Value(UBLOCK_ARABIC, "Arabic Arabic"),
|
|
Value(UBLOCK_SYRIAC, "Syriac Syriac"),
|
|
Value(UBLOCK_THAANA, "Thaana Thaana"),
|
|
Value(UBLOCK_DEVANAGARI, "Devanagari Devanagari"),
|
|
Value(UBLOCK_BENGALI, "Bengali Bengali"),
|
|
Value(UBLOCK_GURMUKHI, "Gurmukhi Gurmukhi"),
|
|
Value(UBLOCK_GUJARATI, "Gujarati Gujarati"),
|
|
Value(UBLOCK_ORIYA, "Oriya Oriya"),
|
|
Value(UBLOCK_TAMIL, "Tamil Tamil"),
|
|
Value(UBLOCK_TELUGU, "Telugu Telugu"),
|
|
Value(UBLOCK_KANNADA, "Kannada Kannada"),
|
|
Value(UBLOCK_MALAYALAM, "Malayalam Malayalam"),
|
|
Value(UBLOCK_SINHALA, "Sinhala Sinhala"),
|
|
Value(UBLOCK_THAI, "Thai Thai"),
|
|
Value(UBLOCK_LAO, "Lao Lao"),
|
|
Value(UBLOCK_TIBETAN, "Tibetan Tibetan"),
|
|
Value(UBLOCK_MYANMAR, "Myanmar Myanmar"),
|
|
Value(UBLOCK_GEORGIAN, "Georgian Georgian"),
|
|
Value(UBLOCK_HANGUL_JAMO, "Jamo Hangul_Jamo"),
|
|
Value(UBLOCK_ETHIOPIC, "Ethiopic Ethiopic"),
|
|
Value(UBLOCK_CHEROKEE, "Cherokee Cherokee"),
|
|
Value(UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS, "UCAS Unified_Canadian_Aboriginal_Syllabics Canadian_Syllabics"),
|
|
Value(UBLOCK_OGHAM, "Ogham Ogham"),
|
|
Value(UBLOCK_RUNIC, "Runic Runic"),
|
|
Value(UBLOCK_KHMER, "Khmer Khmer"),
|
|
Value(UBLOCK_MONGOLIAN, "Mongolian Mongolian"),
|
|
Value(UBLOCK_LATIN_EXTENDED_ADDITIONAL, "Latin_Ext_Additional Latin_Extended_Additional"),
|
|
Value(UBLOCK_GREEK_EXTENDED, "Greek_Ext Greek_Extended"),
|
|
Value(UBLOCK_GENERAL_PUNCTUATION, "Punctuation General_Punctuation"),
|
|
Value(UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS, "Super_And_Sub Superscripts_And_Subscripts"),
|
|
Value(UBLOCK_CURRENCY_SYMBOLS, "Currency_Symbols Currency_Symbols"),
|
|
Value(UBLOCK_COMBINING_MARKS_FOR_SYMBOLS, "Diacriticals_For_Symbols Combining_Diacritical_Marks_For_Symbols Combining_Marks_For_Symbols"),
|
|
Value(UBLOCK_LETTERLIKE_SYMBOLS, "Letterlike_Symbols Letterlike_Symbols"),
|
|
Value(UBLOCK_NUMBER_FORMS, "Number_Forms Number_Forms"),
|
|
Value(UBLOCK_ARROWS, "Arrows Arrows"),
|
|
Value(UBLOCK_MATHEMATICAL_OPERATORS, "Math_Operators Mathematical_Operators"),
|
|
Value(UBLOCK_MISCELLANEOUS_TECHNICAL, "Misc_Technical Miscellaneous_Technical"),
|
|
Value(UBLOCK_CONTROL_PICTURES, "Control_Pictures Control_Pictures"),
|
|
Value(UBLOCK_OPTICAL_CHARACTER_RECOGNITION, "OCR Optical_Character_Recognition"),
|
|
Value(UBLOCK_ENCLOSED_ALPHANUMERICS, "Enclosed_Alphanum Enclosed_Alphanumerics"),
|
|
Value(UBLOCK_BOX_DRAWING, "Box_Drawing Box_Drawing"),
|
|
Value(UBLOCK_BLOCK_ELEMENTS, "Block_Elements Block_Elements"),
|
|
Value(UBLOCK_GEOMETRIC_SHAPES, "Geometric_Shapes Geometric_Shapes"),
|
|
Value(UBLOCK_MISCELLANEOUS_SYMBOLS, "Misc_Symbols Miscellaneous_Symbols"),
|
|
Value(UBLOCK_DINGBATS, "Dingbats Dingbats"),
|
|
Value(UBLOCK_BRAILLE_PATTERNS, "Braille Braille_Patterns"),
|
|
Value(UBLOCK_CJK_RADICALS_SUPPLEMENT, "CJK_Radicals_Sup CJK_Radicals_Supplement"),
|
|
Value(UBLOCK_KANGXI_RADICALS, "Kangxi Kangxi_Radicals"),
|
|
Value(UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS, "IDC Ideographic_Description_Characters"),
|
|
Value(UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION, "CJK_Symbols CJK_Symbols_And_Punctuation"),
|
|
Value(UBLOCK_HIRAGANA, "Hiragana Hiragana"),
|
|
Value(UBLOCK_KATAKANA, "Katakana Katakana"),
|
|
Value(UBLOCK_BOPOMOFO, "Bopomofo Bopomofo"),
|
|
Value(UBLOCK_HANGUL_COMPATIBILITY_JAMO, "Compat_Jamo Hangul_Compatibility_Jamo"),
|
|
Value(UBLOCK_KANBUN, "Kanbun Kanbun"),
|
|
Value(UBLOCK_BOPOMOFO_EXTENDED, "Bopomofo_Ext Bopomofo_Extended"),
|
|
Value(UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS, "Enclosed_CJK Enclosed_CJK_Letters_And_Months"),
|
|
Value(UBLOCK_CJK_COMPATIBILITY, "CJK_Compat CJK_Compatibility"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A, "CJK_Ext_A CJK_Unified_Ideographs_Extension_A"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS, "CJK CJK_Unified_Ideographs"),
|
|
Value(UBLOCK_YI_SYLLABLES, "Yi_Syllables Yi_Syllables"),
|
|
Value(UBLOCK_YI_RADICALS, "Yi_Radicals Yi_Radicals"),
|
|
Value(UBLOCK_HANGUL_SYLLABLES, "Hangul Hangul_Syllables"),
|
|
Value(UBLOCK_HIGH_SURROGATES, "High_Surrogates High_Surrogates"),
|
|
Value(UBLOCK_HIGH_PRIVATE_USE_SURROGATES, "High_PU_Surrogates High_Private_Use_Surrogates"),
|
|
Value(UBLOCK_LOW_SURROGATES, "Low_Surrogates Low_Surrogates"),
|
|
Value(UBLOCK_PRIVATE_USE_AREA, "PUA Private_Use_Area Private_Use"),
|
|
Value(UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS, "CJK_Compat_Ideographs CJK_Compatibility_Ideographs"),
|
|
Value(UBLOCK_ALPHABETIC_PRESENTATION_FORMS, "Alphabetic_PF Alphabetic_Presentation_Forms"),
|
|
Value(UBLOCK_ARABIC_PRESENTATION_FORMS_A, "Arabic_PF_A Arabic_Presentation_Forms_A Arabic_Presentation_Forms-A"),
|
|
Value(UBLOCK_COMBINING_HALF_MARKS, "Half_Marks Combining_Half_Marks"),
|
|
Value(UBLOCK_CJK_COMPATIBILITY_FORMS, "CJK_Compat_Forms CJK_Compatibility_Forms"),
|
|
Value(UBLOCK_SMALL_FORM_VARIANTS, "Small_Forms Small_Form_Variants"),
|
|
Value(UBLOCK_ARABIC_PRESENTATION_FORMS_B, "Arabic_PF_B Arabic_Presentation_Forms_B"),
|
|
Value(UBLOCK_SPECIALS, "Specials Specials"),
|
|
Value(UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS, "Half_And_Full_Forms Halfwidth_And_Fullwidth_Forms"),
|
|
Value(UBLOCK_OLD_ITALIC, "Old_Italic Old_Italic"),
|
|
Value(UBLOCK_GOTHIC, "Gothic Gothic"),
|
|
Value(UBLOCK_DESERET, "Deseret Deseret"),
|
|
Value(UBLOCK_BYZANTINE_MUSICAL_SYMBOLS, "Byzantine_Music Byzantine_Musical_Symbols"),
|
|
Value(UBLOCK_MUSICAL_SYMBOLS, "Music Musical_Symbols"),
|
|
Value(UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS, "Math_Alphanum Mathematical_Alphanumeric_Symbols"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B, "CJK_Ext_B CJK_Unified_Ideographs_Extension_B"),
|
|
Value(UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT, "CJK_Compat_Ideographs_Sup CJK_Compatibility_Ideographs_Supplement"),
|
|
Value(UBLOCK_TAGS, "Tags Tags"),
|
|
Value(UBLOCK_CYRILLIC_SUPPLEMENT, "Cyrillic_Sup Cyrillic_Supplement Cyrillic_Supplementary"),
|
|
Value(UBLOCK_TAGALOG, "Tagalog Tagalog"),
|
|
Value(UBLOCK_HANUNOO, "Hanunoo Hanunoo"),
|
|
Value(UBLOCK_BUHID, "Buhid Buhid"),
|
|
Value(UBLOCK_TAGBANWA, "Tagbanwa Tagbanwa"),
|
|
Value(UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A, "Misc_Math_Symbols_A Miscellaneous_Mathematical_Symbols_A"),
|
|
Value(UBLOCK_SUPPLEMENTAL_ARROWS_A, "Sup_Arrows_A Supplemental_Arrows_A"),
|
|
Value(UBLOCK_SUPPLEMENTAL_ARROWS_B, "Sup_Arrows_B Supplemental_Arrows_B"),
|
|
Value(UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B, "Misc_Math_Symbols_B Miscellaneous_Mathematical_Symbols_B"),
|
|
Value(UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS, "Sup_Math_Operators Supplemental_Mathematical_Operators"),
|
|
Value(UBLOCK_KATAKANA_PHONETIC_EXTENSIONS, "Katakana_Ext Katakana_Phonetic_Extensions"),
|
|
Value(UBLOCK_VARIATION_SELECTORS, "VS Variation_Selectors"),
|
|
Value(UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A, "Sup_PUA_A Supplementary_Private_Use_Area_A"),
|
|
Value(UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B, "Sup_PUA_B Supplementary_Private_Use_Area_B"),
|
|
Value(UBLOCK_LIMBU, "Limbu Limbu"),
|
|
Value(UBLOCK_TAI_LE, "Tai_Le Tai_Le"),
|
|
Value(UBLOCK_KHMER_SYMBOLS, "Khmer_Symbols Khmer_Symbols"),
|
|
Value(UBLOCK_PHONETIC_EXTENSIONS, "Phonetic_Ext Phonetic_Extensions"),
|
|
Value(UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS, "Misc_Arrows Miscellaneous_Symbols_And_Arrows"),
|
|
Value(UBLOCK_YIJING_HEXAGRAM_SYMBOLS, "Yijing Yijing_Hexagram_Symbols"),
|
|
Value(UBLOCK_LINEAR_B_SYLLABARY, "Linear_B_Syllabary Linear_B_Syllabary"),
|
|
Value(UBLOCK_LINEAR_B_IDEOGRAMS, "Linear_B_Ideograms Linear_B_Ideograms"),
|
|
Value(UBLOCK_AEGEAN_NUMBERS, "Aegean_Numbers Aegean_Numbers"),
|
|
Value(UBLOCK_UGARITIC, "Ugaritic Ugaritic"),
|
|
Value(UBLOCK_SHAVIAN, "Shavian Shavian"),
|
|
Value(UBLOCK_OSMANYA, "Osmanya Osmanya"),
|
|
Value(UBLOCK_CYPRIOT_SYLLABARY, "Cypriot_Syllabary Cypriot_Syllabary"),
|
|
Value(UBLOCK_TAI_XUAN_JING_SYMBOLS, "Tai_Xuan_Jing Tai_Xuan_Jing_Symbols"),
|
|
Value(UBLOCK_VARIATION_SELECTORS_SUPPLEMENT, "VS_Sup Variation_Selectors_Supplement"),
|
|
Value(UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION, "Ancient_Greek_Music Ancient_Greek_Musical_Notation"),
|
|
Value(UBLOCK_ANCIENT_GREEK_NUMBERS, "Ancient_Greek_Numbers Ancient_Greek_Numbers"),
|
|
Value(UBLOCK_ARABIC_SUPPLEMENT, "Arabic_Sup Arabic_Supplement"),
|
|
Value(UBLOCK_BUGINESE, "Buginese Buginese"),
|
|
Value(UBLOCK_CJK_STROKES, "CJK_Strokes CJK_Strokes"),
|
|
Value(UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT, "Diacriticals_Sup Combining_Diacritical_Marks_Supplement"),
|
|
Value(UBLOCK_COPTIC, "Coptic Coptic"),
|
|
Value(UBLOCK_ETHIOPIC_EXTENDED, "Ethiopic_Ext Ethiopic_Extended"),
|
|
Value(UBLOCK_ETHIOPIC_SUPPLEMENT, "Ethiopic_Sup Ethiopic_Supplement"),
|
|
Value(UBLOCK_GEORGIAN_SUPPLEMENT, "Georgian_Sup Georgian_Supplement"),
|
|
Value(UBLOCK_GLAGOLITIC, "Glagolitic Glagolitic"),
|
|
Value(UBLOCK_KHAROSHTHI, "Kharoshthi Kharoshthi"),
|
|
Value(UBLOCK_MODIFIER_TONE_LETTERS, "Modifier_Tone_Letters Modifier_Tone_Letters"),
|
|
Value(UBLOCK_NEW_TAI_LUE, "New_Tai_Lue New_Tai_Lue"),
|
|
Value(UBLOCK_OLD_PERSIAN, "Old_Persian Old_Persian"),
|
|
Value(UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT, "Phonetic_Ext_Sup Phonetic_Extensions_Supplement"),
|
|
Value(UBLOCK_SUPPLEMENTAL_PUNCTUATION, "Sup_Punctuation Supplemental_Punctuation"),
|
|
Value(UBLOCK_SYLOTI_NAGRI, "Syloti_Nagri Syloti_Nagri"),
|
|
Value(UBLOCK_TIFINAGH, "Tifinagh Tifinagh"),
|
|
Value(UBLOCK_VERTICAL_FORMS, "Vertical_Forms Vertical_Forms"),
|
|
Value(UBLOCK_NKO, "NKo NKo"),
|
|
Value(UBLOCK_BALINESE, "Balinese Balinese"),
|
|
Value(UBLOCK_LATIN_EXTENDED_C, "Latin_Ext_C Latin_Extended_C"),
|
|
Value(UBLOCK_LATIN_EXTENDED_D, "Latin_Ext_D Latin_Extended_D"),
|
|
Value(UBLOCK_PHAGS_PA, "Phags_Pa Phags_Pa"),
|
|
Value(UBLOCK_PHOENICIAN, "Phoenician Phoenician"),
|
|
Value(UBLOCK_CUNEIFORM, "Cuneiform Cuneiform"),
|
|
Value(UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION, "Cuneiform_Numbers Cuneiform_Numbers_And_Punctuation"),
|
|
Value(UBLOCK_COUNTING_ROD_NUMERALS, "Counting_Rod Counting_Rod_Numerals"),
|
|
Value(UBLOCK_SUNDANESE, "Sundanese Sundanese"),
|
|
Value(UBLOCK_LEPCHA, "Lepcha Lepcha"),
|
|
Value(UBLOCK_OL_CHIKI, "Ol_Chiki Ol_Chiki"),
|
|
Value(UBLOCK_CYRILLIC_EXTENDED_A, "Cyrillic_Ext_A Cyrillic_Extended_A"),
|
|
Value(UBLOCK_VAI, "Vai Vai"),
|
|
Value(UBLOCK_CYRILLIC_EXTENDED_B, "Cyrillic_Ext_B Cyrillic_Extended_B"),
|
|
Value(UBLOCK_SAURASHTRA, "Saurashtra Saurashtra"),
|
|
Value(UBLOCK_KAYAH_LI, "Kayah_Li Kayah_Li"),
|
|
Value(UBLOCK_REJANG, "Rejang Rejang"),
|
|
Value(UBLOCK_CHAM, "Cham Cham"),
|
|
Value(UBLOCK_ANCIENT_SYMBOLS, "Ancient_Symbols Ancient_Symbols"),
|
|
Value(UBLOCK_PHAISTOS_DISC, "Phaistos Phaistos_Disc"),
|
|
Value(UBLOCK_LYCIAN, "Lycian Lycian"),
|
|
Value(UBLOCK_CARIAN, "Carian Carian"),
|
|
Value(UBLOCK_LYDIAN, "Lydian Lydian"),
|
|
Value(UBLOCK_MAHJONG_TILES, "Mahjong Mahjong_Tiles"),
|
|
Value(UBLOCK_DOMINO_TILES, "Domino Domino_Tiles"),
|
|
Value(UBLOCK_SAMARITAN, "Samaritan Samaritan"),
|
|
Value(UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED, "UCAS_Ext Unified_Canadian_Aboriginal_Syllabics_Extended"),
|
|
Value(UBLOCK_TAI_THAM, "Tai_Tham Tai_Tham"),
|
|
Value(UBLOCK_VEDIC_EXTENSIONS, "Vedic_Ext Vedic_Extensions"),
|
|
Value(UBLOCK_LISU, "Lisu Lisu"),
|
|
Value(UBLOCK_BAMUM, "Bamum Bamum"),
|
|
Value(UBLOCK_COMMON_INDIC_NUMBER_FORMS, "Indic_Number_Forms Common_Indic_Number_Forms"),
|
|
Value(UBLOCK_DEVANAGARI_EXTENDED, "Devanagari_Ext Devanagari_Extended"),
|
|
Value(UBLOCK_HANGUL_JAMO_EXTENDED_A, "Jamo_Ext_A Hangul_Jamo_Extended_A"),
|
|
Value(UBLOCK_JAVANESE, "Javanese Javanese"),
|
|
Value(UBLOCK_MYANMAR_EXTENDED_A, "Myanmar_Ext_A Myanmar_Extended_A"),
|
|
Value(UBLOCK_TAI_VIET, "Tai_Viet Tai_Viet"),
|
|
Value(UBLOCK_MEETEI_MAYEK, "Meetei_Mayek Meetei_Mayek"),
|
|
Value(UBLOCK_HANGUL_JAMO_EXTENDED_B, "Jamo_Ext_B Hangul_Jamo_Extended_B"),
|
|
Value(UBLOCK_IMPERIAL_ARAMAIC, "Imperial_Aramaic Imperial_Aramaic"),
|
|
Value(UBLOCK_OLD_SOUTH_ARABIAN, "Old_South_Arabian Old_South_Arabian"),
|
|
Value(UBLOCK_AVESTAN, "Avestan Avestan"),
|
|
Value(UBLOCK_INSCRIPTIONAL_PARTHIAN, "Inscriptional_Parthian Inscriptional_Parthian"),
|
|
Value(UBLOCK_INSCRIPTIONAL_PAHLAVI, "Inscriptional_Pahlavi Inscriptional_Pahlavi"),
|
|
Value(UBLOCK_OLD_TURKIC, "Old_Turkic Old_Turkic"),
|
|
Value(UBLOCK_RUMI_NUMERAL_SYMBOLS, "Rumi Rumi_Numeral_Symbols"),
|
|
Value(UBLOCK_KAITHI, "Kaithi Kaithi"),
|
|
Value(UBLOCK_EGYPTIAN_HIEROGLYPHS, "Egyptian_Hieroglyphs Egyptian_Hieroglyphs"),
|
|
Value(UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT, "Enclosed_Alphanum_Sup Enclosed_Alphanumeric_Supplement"),
|
|
Value(UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT, "Enclosed_Ideographic_Sup Enclosed_Ideographic_Supplement"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C, "CJK_Ext_C CJK_Unified_Ideographs_Extension_C"),
|
|
Value(UBLOCK_MANDAIC, "Mandaic Mandaic"),
|
|
Value(UBLOCK_BATAK, "Batak Batak"),
|
|
Value(UBLOCK_ETHIOPIC_EXTENDED_A, "Ethiopic_Ext_A Ethiopic_Extended_A"),
|
|
Value(UBLOCK_BRAHMI, "Brahmi Brahmi"),
|
|
Value(UBLOCK_BAMUM_SUPPLEMENT, "Bamum_Sup Bamum_Supplement"),
|
|
Value(UBLOCK_KANA_SUPPLEMENT, "Kana_Sup Kana_Supplement"),
|
|
Value(UBLOCK_PLAYING_CARDS, "Playing_Cards Playing_Cards"),
|
|
Value(UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS, "Misc_Pictographs Miscellaneous_Symbols_And_Pictographs"),
|
|
Value(UBLOCK_EMOTICONS, "Emoticons Emoticons"),
|
|
Value(UBLOCK_TRANSPORT_AND_MAP_SYMBOLS, "Transport_And_Map Transport_And_Map_Symbols"),
|
|
Value(UBLOCK_ALCHEMICAL_SYMBOLS, "Alchemical Alchemical_Symbols"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D, "CJK_Ext_D CJK_Unified_Ideographs_Extension_D"),
|
|
Value(UBLOCK_ARABIC_EXTENDED_A, "Arabic_Ext_A Arabic_Extended_A"),
|
|
Value(UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS, "Arabic_Math Arabic_Mathematical_Alphabetic_Symbols"),
|
|
Value(UBLOCK_CHAKMA, "Chakma Chakma"),
|
|
Value(UBLOCK_MEETEI_MAYEK_EXTENSIONS, "Meetei_Mayek_Ext Meetei_Mayek_Extensions"),
|
|
Value(UBLOCK_MEROITIC_CURSIVE, "Meroitic_Cursive Meroitic_Cursive"),
|
|
Value(UBLOCK_MEROITIC_HIEROGLYPHS, "Meroitic_Hieroglyphs Meroitic_Hieroglyphs"),
|
|
Value(UBLOCK_MIAO, "Miao Miao"),
|
|
Value(UBLOCK_SHARADA, "Sharada Sharada"),
|
|
Value(UBLOCK_SORA_SOMPENG, "Sora_Sompeng Sora_Sompeng"),
|
|
Value(UBLOCK_SUNDANESE_SUPPLEMENT, "Sundanese_Sup Sundanese_Supplement"),
|
|
Value(UBLOCK_TAKRI, "Takri Takri"),
|
|
Value(UBLOCK_BASSA_VAH, "Bassa_Vah Bassa_Vah"),
|
|
Value(UBLOCK_CAUCASIAN_ALBANIAN, "Caucasian_Albanian Caucasian_Albanian"),
|
|
Value(UBLOCK_COPTIC_EPACT_NUMBERS, "Coptic_Epact_Numbers Coptic_Epact_Numbers"),
|
|
Value(UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED, "Diacriticals_Ext Combining_Diacritical_Marks_Extended"),
|
|
Value(UBLOCK_DUPLOYAN, "Duployan Duployan"),
|
|
Value(UBLOCK_ELBASAN, "Elbasan Elbasan"),
|
|
Value(UBLOCK_GEOMETRIC_SHAPES_EXTENDED, "Geometric_Shapes_Ext Geometric_Shapes_Extended"),
|
|
Value(UBLOCK_GRANTHA, "Grantha Grantha"),
|
|
Value(UBLOCK_KHOJKI, "Khojki Khojki"),
|
|
Value(UBLOCK_KHUDAWADI, "Khudawadi Khudawadi"),
|
|
Value(UBLOCK_LATIN_EXTENDED_E, "Latin_Ext_E Latin_Extended_E"),
|
|
Value(UBLOCK_LINEAR_A, "Linear_A Linear_A"),
|
|
Value(UBLOCK_MAHAJANI, "Mahajani Mahajani"),
|
|
Value(UBLOCK_MANICHAEAN, "Manichaean Manichaean"),
|
|
Value(UBLOCK_MENDE_KIKAKUI, "Mende_Kikakui Mende_Kikakui"),
|
|
Value(UBLOCK_MODI, "Modi Modi"),
|
|
Value(UBLOCK_MRO, "Mro Mro"),
|
|
Value(UBLOCK_MYANMAR_EXTENDED_B, "Myanmar_Ext_B Myanmar_Extended_B"),
|
|
Value(UBLOCK_NABATAEAN, "Nabataean Nabataean"),
|
|
Value(UBLOCK_OLD_NORTH_ARABIAN, "Old_North_Arabian Old_North_Arabian"),
|
|
Value(UBLOCK_OLD_PERMIC, "Old_Permic Old_Permic"),
|
|
Value(UBLOCK_ORNAMENTAL_DINGBATS, "Ornamental_Dingbats Ornamental_Dingbats"),
|
|
Value(UBLOCK_PAHAWH_HMONG, "Pahawh_Hmong Pahawh_Hmong"),
|
|
Value(UBLOCK_PALMYRENE, "Palmyrene Palmyrene"),
|
|
Value(UBLOCK_PAU_CIN_HAU, "Pau_Cin_Hau Pau_Cin_Hau"),
|
|
Value(UBLOCK_PSALTER_PAHLAVI, "Psalter_Pahlavi Psalter_Pahlavi"),
|
|
Value(UBLOCK_SHORTHAND_FORMAT_CONTROLS, "Shorthand_Format_Controls Shorthand_Format_Controls"),
|
|
Value(UBLOCK_SIDDHAM, "Siddham Siddham"),
|
|
Value(UBLOCK_SINHALA_ARCHAIC_NUMBERS, "Sinhala_Archaic_Numbers Sinhala_Archaic_Numbers"),
|
|
Value(UBLOCK_SUPPLEMENTAL_ARROWS_C, "Sup_Arrows_C Supplemental_Arrows_C"),
|
|
Value(UBLOCK_TIRHUTA, "Tirhuta Tirhuta"),
|
|
Value(UBLOCK_WARANG_CITI, "Warang_Citi Warang_Citi"),
|
|
Value(UBLOCK_AHOM, "Ahom Ahom"),
|
|
Value(UBLOCK_ANATOLIAN_HIEROGLYPHS, "Anatolian_Hieroglyphs Anatolian_Hieroglyphs"),
|
|
Value(UBLOCK_CHEROKEE_SUPPLEMENT, "Cherokee_Sup Cherokee_Supplement"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E, "CJK_Ext_E CJK_Unified_Ideographs_Extension_E"),
|
|
Value(UBLOCK_EARLY_DYNASTIC_CUNEIFORM, "Early_Dynastic_Cuneiform Early_Dynastic_Cuneiform"),
|
|
Value(UBLOCK_HATRAN, "Hatran Hatran"),
|
|
Value(UBLOCK_MULTANI, "Multani Multani"),
|
|
Value(UBLOCK_OLD_HUNGARIAN, "Old_Hungarian Old_Hungarian"),
|
|
Value(UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS, "Sup_Symbols_And_Pictographs Supplemental_Symbols_And_Pictographs"),
|
|
Value(UBLOCK_SUTTON_SIGNWRITING, "Sutton_SignWriting Sutton_SignWriting"),
|
|
Value(UBLOCK_ADLAM, "Adlam Adlam"),
|
|
Value(UBLOCK_BHAIKSUKI, "Bhaiksuki Bhaiksuki"),
|
|
Value(UBLOCK_CYRILLIC_EXTENDED_C, "Cyrillic_Ext_C Cyrillic_Extended_C"),
|
|
Value(UBLOCK_GLAGOLITIC_SUPPLEMENT, "Glagolitic_Sup Glagolitic_Supplement"),
|
|
Value(UBLOCK_IDEOGRAPHIC_SYMBOLS_AND_PUNCTUATION, "Ideographic_Symbols Ideographic_Symbols_And_Punctuation"),
|
|
Value(UBLOCK_MARCHEN, "Marchen Marchen"),
|
|
Value(UBLOCK_MONGOLIAN_SUPPLEMENT, "Mongolian_Sup Mongolian_Supplement"),
|
|
Value(UBLOCK_NEWA, "Newa Newa"),
|
|
Value(UBLOCK_OSAGE, "Osage Osage"),
|
|
Value(UBLOCK_TANGUT, "Tangut Tangut"),
|
|
Value(UBLOCK_TANGUT_COMPONENTS, "Tangut_Components Tangut_Components"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F, "CJK_Ext_F CJK_Unified_Ideographs_Extension_F"),
|
|
Value(UBLOCK_KANA_EXTENDED_A, "Kana_Ext_A Kana_Extended_A"),
|
|
Value(UBLOCK_MASARAM_GONDI, "Masaram_Gondi Masaram_Gondi"),
|
|
Value(UBLOCK_NUSHU, "Nushu Nushu"),
|
|
Value(UBLOCK_SOYOMBO, "Soyombo Soyombo"),
|
|
Value(UBLOCK_SYRIAC_SUPPLEMENT, "Syriac_Sup Syriac_Supplement"),
|
|
Value(UBLOCK_ZANABAZAR_SQUARE, "Zanabazar_Square Zanabazar_Square"),
|
|
Value(UBLOCK_CHESS_SYMBOLS, "Chess_Symbols Chess_Symbols"),
|
|
Value(UBLOCK_DOGRA, "Dogra Dogra"),
|
|
Value(UBLOCK_GEORGIAN_EXTENDED, "Georgian_Ext Georgian_Extended"),
|
|
Value(UBLOCK_GUNJALA_GONDI, "Gunjala_Gondi Gunjala_Gondi"),
|
|
Value(UBLOCK_HANIFI_ROHINGYA, "Hanifi_Rohingya Hanifi_Rohingya"),
|
|
Value(UBLOCK_INDIC_SIYAQ_NUMBERS, "Indic_Siyaq_Numbers Indic_Siyaq_Numbers"),
|
|
Value(UBLOCK_MAKASAR, "Makasar Makasar"),
|
|
Value(UBLOCK_MAYAN_NUMERALS, "Mayan_Numerals Mayan_Numerals"),
|
|
Value(UBLOCK_MEDEFAIDRIN, "Medefaidrin Medefaidrin"),
|
|
Value(UBLOCK_OLD_SOGDIAN, "Old_Sogdian Old_Sogdian"),
|
|
Value(UBLOCK_SOGDIAN, "Sogdian Sogdian"),
|
|
Value(UBLOCK_EGYPTIAN_HIEROGLYPH_FORMAT_CONTROLS, "Egyptian_Hieroglyph_Format_Controls Egyptian_Hieroglyph_Format_Controls"),
|
|
Value(UBLOCK_ELYMAIC, "Elymaic Elymaic"),
|
|
Value(UBLOCK_NANDINAGARI, "Nandinagari Nandinagari"),
|
|
Value(UBLOCK_NYIAKENG_PUACHUE_HMONG, "Nyiakeng_Puachue_Hmong Nyiakeng_Puachue_Hmong"),
|
|
Value(UBLOCK_OTTOMAN_SIYAQ_NUMBERS, "Ottoman_Siyaq_Numbers Ottoman_Siyaq_Numbers"),
|
|
Value(UBLOCK_SMALL_KANA_EXTENSION, "Small_Kana_Ext Small_Kana_Extension"),
|
|
Value(UBLOCK_SYMBOLS_AND_PICTOGRAPHS_EXTENDED_A, "Symbols_And_Pictographs_Ext_A Symbols_And_Pictographs_Extended_A"),
|
|
Value(UBLOCK_TAMIL_SUPPLEMENT, "Tamil_Sup Tamil_Supplement"),
|
|
Value(UBLOCK_WANCHO, "Wancho Wancho"),
|
|
Value(UBLOCK_CHORASMIAN, "Chorasmian Chorasmian"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_G, "CJK_Ext_G CJK_Unified_Ideographs_Extension_G"),
|
|
Value(UBLOCK_DIVES_AKURU, "Dives_Akuru Dives_Akuru"),
|
|
Value(UBLOCK_KHITAN_SMALL_SCRIPT, "Khitan_Small_Script Khitan_Small_Script"),
|
|
Value(UBLOCK_LISU_SUPPLEMENT, "Lisu_Sup Lisu_Supplement"),
|
|
Value(UBLOCK_SYMBOLS_FOR_LEGACY_COMPUTING, "Symbols_For_Legacy_Computing Symbols_For_Legacy_Computing"),
|
|
Value(UBLOCK_TANGUT_SUPPLEMENT, "Tangut_Sup Tangut_Supplement"),
|
|
Value(UBLOCK_YEZIDI, "Yezidi Yezidi"),
|
|
Value(UBLOCK_ARABIC_EXTENDED_B, "Arabic_Ext_B Arabic_Extended_B"),
|
|
Value(UBLOCK_CYPRO_MINOAN, "Cypro_Minoan Cypro_Minoan"),
|
|
Value(UBLOCK_ETHIOPIC_EXTENDED_B, "Ethiopic_Ext_B Ethiopic_Extended_B"),
|
|
Value(UBLOCK_KANA_EXTENDED_B, "Kana_Ext_B Kana_Extended_B"),
|
|
Value(UBLOCK_LATIN_EXTENDED_F, "Latin_Ext_F Latin_Extended_F"),
|
|
Value(UBLOCK_LATIN_EXTENDED_G, "Latin_Ext_G Latin_Extended_G"),
|
|
Value(UBLOCK_OLD_UYGHUR, "Old_Uyghur Old_Uyghur"),
|
|
Value(UBLOCK_TANGSA, "Tangsa Tangsa"),
|
|
Value(UBLOCK_TOTO, "Toto Toto"),
|
|
Value(UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED_A, "UCAS_Ext_A Unified_Canadian_Aboriginal_Syllabics_Extended_A"),
|
|
Value(UBLOCK_VITHKUQI, "Vithkuqi Vithkuqi"),
|
|
Value(UBLOCK_ZNAMENNY_MUSICAL_NOTATION, "Znamenny_Music Znamenny_Musical_Notation"),
|
|
Value(UBLOCK_ARABIC_EXTENDED_C, "Arabic_Ext_C Arabic_Extended_C"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_H, "CJK_Ext_H CJK_Unified_Ideographs_Extension_H"),
|
|
Value(UBLOCK_CYRILLIC_EXTENDED_D, "Cyrillic_Ext_D Cyrillic_Extended_D"),
|
|
Value(UBLOCK_DEVANAGARI_EXTENDED_A, "Devanagari_Ext_A Devanagari_Extended_A"),
|
|
Value(UBLOCK_KAKTOVIK_NUMERALS, "Kaktovik_Numerals Kaktovik_Numerals"),
|
|
Value(UBLOCK_KAWI, "Kawi Kawi"),
|
|
Value(UBLOCK_NAG_MUNDARI, "Nag_Mundari Nag_Mundari"),
|
|
Value(UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_I, "CJK_Ext_I CJK_Unified_Ideographs_Extension_I"),
|
|
Value(UBLOCK_EGYPTIAN_HIEROGLYPHS_EXTENDED_A, "Egyptian_Hieroglyphs_Ext_A Egyptian_Hieroglyphs_Extended_A"),
|
|
Value(UBLOCK_GARAY, "Garay Garay"),
|
|
Value(UBLOCK_GURUNG_KHEMA, "Gurung_Khema Gurung_Khema"),
|
|
Value(UBLOCK_KIRAT_RAI, "Kirat_Rai Kirat_Rai"),
|
|
Value(UBLOCK_MYANMAR_EXTENDED_C, "Myanmar_Ext_C Myanmar_Extended_C"),
|
|
Value(UBLOCK_OL_ONAL, "Ol_Onal Ol_Onal"),
|
|
Value(UBLOCK_SUNUWAR, "Sunuwar Sunuwar"),
|
|
Value(UBLOCK_SYMBOLS_FOR_LEGACY_COMPUTING_SUPPLEMENT, "Symbols_For_Legacy_Computing_Sup Symbols_For_Legacy_Computing_Supplement"),
|
|
Value(UBLOCK_TODHRI, "Todhri Todhri"),
|
|
Value(UBLOCK_TULU_TIGALARI, "Tulu_Tigalari Tulu_Tigalari"),
|
|
};
|
|
|
|
static const Value VALUES_ccc[58] = {
|
|
Value(0, "NR Not_Reordered"),
|
|
Value(1, "OV Overlay"),
|
|
Value(6, "HANR Han_Reading"),
|
|
Value(7, "NK Nukta"),
|
|
Value(8, "KV Kana_Voicing"),
|
|
Value(9, "VR Virama"),
|
|
Value(10, "CCC10 CCC10"),
|
|
Value(11, "CCC11 CCC11"),
|
|
Value(12, "CCC12 CCC12"),
|
|
Value(13, "CCC13 CCC13"),
|
|
Value(14, "CCC14 CCC14"),
|
|
Value(15, "CCC15 CCC15"),
|
|
Value(16, "CCC16 CCC16"),
|
|
Value(17, "CCC17 CCC17"),
|
|
Value(18, "CCC18 CCC18"),
|
|
Value(19, "CCC19 CCC19"),
|
|
Value(20, "CCC20 CCC20"),
|
|
Value(21, "CCC21 CCC21"),
|
|
Value(22, "CCC22 CCC22"),
|
|
Value(23, "CCC23 CCC23"),
|
|
Value(24, "CCC24 CCC24"),
|
|
Value(25, "CCC25 CCC25"),
|
|
Value(26, "CCC26 CCC26"),
|
|
Value(27, "CCC27 CCC27"),
|
|
Value(28, "CCC28 CCC28"),
|
|
Value(29, "CCC29 CCC29"),
|
|
Value(30, "CCC30 CCC30"),
|
|
Value(31, "CCC31 CCC31"),
|
|
Value(32, "CCC32 CCC32"),
|
|
Value(33, "CCC33 CCC33"),
|
|
Value(34, "CCC34 CCC34"),
|
|
Value(35, "CCC35 CCC35"),
|
|
Value(36, "CCC36 CCC36"),
|
|
Value(84, "CCC84 CCC84"),
|
|
Value(91, "CCC91 CCC91"),
|
|
Value(103, "CCC103 CCC103"),
|
|
Value(107, "CCC107 CCC107"),
|
|
Value(118, "CCC118 CCC118"),
|
|
Value(122, "CCC122 CCC122"),
|
|
Value(129, "CCC129 CCC129"),
|
|
Value(130, "CCC130 CCC130"),
|
|
Value(132, "CCC132 CCC132"),
|
|
Value(133, "CCC133 CCC133"),
|
|
Value(200, "ATBL Attached_Below_Left"),
|
|
Value(202, "ATB Attached_Below"),
|
|
Value(214, "ATA Attached_Above"),
|
|
Value(216, "ATAR Attached_Above_Right"),
|
|
Value(218, "BL Below_Left"),
|
|
Value(220, "B Below"),
|
|
Value(222, "BR Below_Right"),
|
|
Value(224, "L Left"),
|
|
Value(226, "R Right"),
|
|
Value(228, "AL Above_Left"),
|
|
Value(230, "A Above"),
|
|
Value(232, "AR Above_Right"),
|
|
Value(233, "DB Double_Below"),
|
|
Value(234, "DA Double_Above"),
|
|
Value(240, "IS Iota_Subscript"),
|
|
};
|
|
|
|
static const Value VALUES_dt[18] = {
|
|
Value(U_DT_NONE, "None None none"),
|
|
Value(U_DT_CANONICAL, "Can Canonical can"),
|
|
Value(U_DT_COMPAT, "Com Compat com"),
|
|
Value(U_DT_CIRCLE, "Enc Circle enc"),
|
|
Value(U_DT_FINAL, "Fin Final fin"),
|
|
Value(U_DT_FONT, "Font Font font"),
|
|
Value(U_DT_FRACTION, "Fra Fraction fra"),
|
|
Value(U_DT_INITIAL, "Init Initial init"),
|
|
Value(U_DT_ISOLATED, "Iso Isolated iso"),
|
|
Value(U_DT_MEDIAL, "Med Medial med"),
|
|
Value(U_DT_NARROW, "Nar Narrow nar"),
|
|
Value(U_DT_NOBREAK, "Nb Nobreak nb"),
|
|
Value(U_DT_SMALL, "Sml Small sml"),
|
|
Value(U_DT_SQUARE, "Sqr Square sqr"),
|
|
Value(U_DT_SUB, "Sub Sub sub"),
|
|
Value(U_DT_SUPER, "Sup Super sup"),
|
|
Value(U_DT_VERTICAL, "Vert Vertical vert"),
|
|
Value(U_DT_WIDE, "Wide Wide wide"),
|
|
};
|
|
|
|
static const Value VALUES_ea[6] = {
|
|
Value(U_EA_NEUTRAL, "N Neutral"),
|
|
Value(U_EA_AMBIGUOUS, "A Ambiguous"),
|
|
Value(U_EA_HALFWIDTH, "H Halfwidth"),
|
|
Value(U_EA_FULLWIDTH, "F Fullwidth"),
|
|
Value(U_EA_NARROW, "Na Narrow"),
|
|
Value(U_EA_WIDE, "W Wide"),
|
|
};
|
|
|
|
static const Value VALUES_gc[30] = {
|
|
Value(U_GENERAL_OTHER_TYPES, "Cn Unassigned"),
|
|
Value(U_UPPERCASE_LETTER, "Lu Uppercase_Letter"),
|
|
Value(U_LOWERCASE_LETTER, "Ll Lowercase_Letter"),
|
|
Value(U_TITLECASE_LETTER, "Lt Titlecase_Letter"),
|
|
Value(U_MODIFIER_LETTER, "Lm Modifier_Letter"),
|
|
Value(U_OTHER_LETTER, "Lo Other_Letter"),
|
|
Value(U_NON_SPACING_MARK, "Mn Nonspacing_Mark"),
|
|
Value(U_ENCLOSING_MARK, "Me Enclosing_Mark"),
|
|
Value(U_COMBINING_SPACING_MARK, "Mc Spacing_Mark"),
|
|
Value(U_DECIMAL_DIGIT_NUMBER, "Nd Decimal_Number digit"),
|
|
Value(U_LETTER_NUMBER, "Nl Letter_Number"),
|
|
Value(U_OTHER_NUMBER, "No Other_Number"),
|
|
Value(U_SPACE_SEPARATOR, "Zs Space_Separator"),
|
|
Value(U_LINE_SEPARATOR, "Zl Line_Separator"),
|
|
Value(U_PARAGRAPH_SEPARATOR, "Zp Paragraph_Separator"),
|
|
Value(U_CONTROL_CHAR, "Cc Control cntrl"),
|
|
Value(U_FORMAT_CHAR, "Cf Format"),
|
|
Value(U_PRIVATE_USE_CHAR, "Co Private_Use"),
|
|
Value(U_SURROGATE, "Cs Surrogate"),
|
|
Value(U_DASH_PUNCTUATION, "Pd Dash_Punctuation"),
|
|
Value(U_START_PUNCTUATION, "Ps Open_Punctuation"),
|
|
Value(U_END_PUNCTUATION, "Pe Close_Punctuation"),
|
|
Value(U_CONNECTOR_PUNCTUATION, "Pc Connector_Punctuation"),
|
|
Value(U_OTHER_PUNCTUATION, "Po Other_Punctuation"),
|
|
Value(U_MATH_SYMBOL, "Sm Math_Symbol"),
|
|
Value(U_CURRENCY_SYMBOL, "Sc Currency_Symbol"),
|
|
Value(U_MODIFIER_SYMBOL, "Sk Modifier_Symbol"),
|
|
Value(U_OTHER_SYMBOL, "So Other_Symbol"),
|
|
Value(U_INITIAL_PUNCTUATION, "Pi Initial_Punctuation"),
|
|
Value(U_FINAL_PUNCTUATION, "Pf Final_Punctuation"),
|
|
};
|
|
|
|
static const Value VALUES_jg[104] = {
|
|
Value(U_JG_NO_JOINING_GROUP, "No_Joining_Group No_Joining_Group"),
|
|
Value(U_JG_AIN, "Ain Ain"),
|
|
Value(U_JG_ALAPH, "Alaph Alaph"),
|
|
Value(U_JG_ALEF, "Alef Alef"),
|
|
Value(U_JG_BEH, "Beh Beh"),
|
|
Value(U_JG_BETH, "Beth Beth"),
|
|
Value(U_JG_DAL, "Dal Dal"),
|
|
Value(U_JG_DALATH_RISH, "Dalath_Rish Dalath_Rish"),
|
|
Value(U_JG_E, "E E"),
|
|
Value(U_JG_FEH, "Feh Feh"),
|
|
Value(U_JG_FINAL_SEMKATH, "Final_Semkath Final_Semkath"),
|
|
Value(U_JG_GAF, "Gaf Gaf"),
|
|
Value(U_JG_GAMAL, "Gamal Gamal"),
|
|
Value(U_JG_HAH, "Hah Hah"),
|
|
Value(U_JG_TEH_MARBUTA_GOAL, "Teh_Marbuta_Goal Teh_Marbuta_Goal Hamza_On_Heh_Goal"),
|
|
Value(U_JG_HE, "He He"),
|
|
Value(U_JG_HEH, "Heh Heh"),
|
|
Value(U_JG_HEH_GOAL, "Heh_Goal Heh_Goal"),
|
|
Value(U_JG_HETH, "Heth Heth"),
|
|
Value(U_JG_KAF, "Kaf Kaf"),
|
|
Value(U_JG_KAPH, "Kaph Kaph"),
|
|
Value(U_JG_KNOTTED_HEH, "Knotted_Heh Knotted_Heh"),
|
|
Value(U_JG_LAM, "Lam Lam"),
|
|
Value(U_JG_LAMADH, "Lamadh Lamadh"),
|
|
Value(U_JG_MEEM, "Meem Meem"),
|
|
Value(U_JG_MIM, "Mim Mim"),
|
|
Value(U_JG_NOON, "Noon Noon"),
|
|
Value(U_JG_NUN, "Nun Nun"),
|
|
Value(U_JG_PE, "Pe Pe"),
|
|
Value(U_JG_QAF, "Qaf Qaf"),
|
|
Value(U_JG_QAPH, "Qaph Qaph"),
|
|
Value(U_JG_REH, "Reh Reh"),
|
|
Value(U_JG_REVERSED_PE, "Reversed_Pe Reversed_Pe"),
|
|
Value(U_JG_SAD, "Sad Sad"),
|
|
Value(U_JG_SADHE, "Sadhe Sadhe"),
|
|
Value(U_JG_SEEN, "Seen Seen"),
|
|
Value(U_JG_SEMKATH, "Semkath Semkath"),
|
|
Value(U_JG_SHIN, "Shin Shin"),
|
|
Value(U_JG_SWASH_KAF, "Swash_Kaf Swash_Kaf"),
|
|
Value(U_JG_SYRIAC_WAW, "Syriac_Waw Syriac_Waw"),
|
|
Value(U_JG_TAH, "Tah Tah"),
|
|
Value(U_JG_TAW, "Taw Taw"),
|
|
Value(U_JG_TEH_MARBUTA, "Teh_Marbuta Teh_Marbuta"),
|
|
Value(U_JG_TETH, "Teth Teth"),
|
|
Value(U_JG_WAW, "Waw Waw"),
|
|
Value(U_JG_YEH, "Yeh Yeh"),
|
|
Value(U_JG_YEH_BARREE, "Yeh_Barree Yeh_Barree"),
|
|
Value(U_JG_YEH_WITH_TAIL, "Yeh_With_Tail Yeh_With_Tail"),
|
|
Value(U_JG_YUDH, "Yudh Yudh"),
|
|
Value(U_JG_YUDH_HE, "Yudh_He Yudh_He"),
|
|
Value(U_JG_ZAIN, "Zain Zain"),
|
|
Value(U_JG_FE, "Fe Fe"),
|
|
Value(U_JG_KHAPH, "Khaph Khaph"),
|
|
Value(U_JG_ZHAIN, "Zhain Zhain"),
|
|
Value(U_JG_BURUSHASKI_YEH_BARREE, "Burushaski_Yeh_Barree Burushaski_Yeh_Barree"),
|
|
Value(U_JG_FARSI_YEH, "Farsi_Yeh Farsi_Yeh"),
|
|
Value(U_JG_NYA, "Nya Nya"),
|
|
Value(U_JG_ROHINGYA_YEH, "Rohingya_Yeh Rohingya_Yeh"),
|
|
Value(U_JG_MANICHAEAN_ALEPH, "Manichaean_Aleph Manichaean_Aleph"),
|
|
Value(U_JG_MANICHAEAN_AYIN, "Manichaean_Ayin Manichaean_Ayin"),
|
|
Value(U_JG_MANICHAEAN_BETH, "Manichaean_Beth Manichaean_Beth"),
|
|
Value(U_JG_MANICHAEAN_DALETH, "Manichaean_Daleth Manichaean_Daleth"),
|
|
Value(U_JG_MANICHAEAN_DHAMEDH, "Manichaean_Dhamedh Manichaean_Dhamedh"),
|
|
Value(U_JG_MANICHAEAN_FIVE, "Manichaean_Five Manichaean_Five"),
|
|
Value(U_JG_MANICHAEAN_GIMEL, "Manichaean_Gimel Manichaean_Gimel"),
|
|
Value(U_JG_MANICHAEAN_HETH, "Manichaean_Heth Manichaean_Heth"),
|
|
Value(U_JG_MANICHAEAN_HUNDRED, "Manichaean_Hundred Manichaean_Hundred"),
|
|
Value(U_JG_MANICHAEAN_KAPH, "Manichaean_Kaph Manichaean_Kaph"),
|
|
Value(U_JG_MANICHAEAN_LAMEDH, "Manichaean_Lamedh Manichaean_Lamedh"),
|
|
Value(U_JG_MANICHAEAN_MEM, "Manichaean_Mem Manichaean_Mem"),
|
|
Value(U_JG_MANICHAEAN_NUN, "Manichaean_Nun Manichaean_Nun"),
|
|
Value(U_JG_MANICHAEAN_ONE, "Manichaean_One Manichaean_One"),
|
|
Value(U_JG_MANICHAEAN_PE, "Manichaean_Pe Manichaean_Pe"),
|
|
Value(U_JG_MANICHAEAN_QOPH, "Manichaean_Qoph Manichaean_Qoph"),
|
|
Value(U_JG_MANICHAEAN_RESH, "Manichaean_Resh Manichaean_Resh"),
|
|
Value(U_JG_MANICHAEAN_SADHE, "Manichaean_Sadhe Manichaean_Sadhe"),
|
|
Value(U_JG_MANICHAEAN_SAMEKH, "Manichaean_Samekh Manichaean_Samekh"),
|
|
Value(U_JG_MANICHAEAN_TAW, "Manichaean_Taw Manichaean_Taw"),
|
|
Value(U_JG_MANICHAEAN_TEN, "Manichaean_Ten Manichaean_Ten"),
|
|
Value(U_JG_MANICHAEAN_TETH, "Manichaean_Teth Manichaean_Teth"),
|
|
Value(U_JG_MANICHAEAN_THAMEDH, "Manichaean_Thamedh Manichaean_Thamedh"),
|
|
Value(U_JG_MANICHAEAN_TWENTY, "Manichaean_Twenty Manichaean_Twenty"),
|
|
Value(U_JG_MANICHAEAN_WAW, "Manichaean_Waw Manichaean_Waw"),
|
|
Value(U_JG_MANICHAEAN_YODH, "Manichaean_Yodh Manichaean_Yodh"),
|
|
Value(U_JG_MANICHAEAN_ZAYIN, "Manichaean_Zayin Manichaean_Zayin"),
|
|
Value(U_JG_STRAIGHT_WAW, "Straight_Waw Straight_Waw"),
|
|
Value(U_JG_AFRICAN_FEH, "African_Feh African_Feh"),
|
|
Value(U_JG_AFRICAN_NOON, "African_Noon African_Noon"),
|
|
Value(U_JG_AFRICAN_QAF, "African_Qaf African_Qaf"),
|
|
Value(U_JG_MALAYALAM_BHA, "Malayalam_Bha Malayalam_Bha"),
|
|
Value(U_JG_MALAYALAM_JA, "Malayalam_Ja Malayalam_Ja"),
|
|
Value(U_JG_MALAYALAM_LLA, "Malayalam_Lla Malayalam_Lla"),
|
|
Value(U_JG_MALAYALAM_LLLA, "Malayalam_Llla Malayalam_Llla"),
|
|
Value(U_JG_MALAYALAM_NGA, "Malayalam_Nga Malayalam_Nga"),
|
|
Value(U_JG_MALAYALAM_NNA, "Malayalam_Nna Malayalam_Nna"),
|
|
Value(U_JG_MALAYALAM_NNNA, "Malayalam_Nnna Malayalam_Nnna"),
|
|
Value(U_JG_MALAYALAM_NYA, "Malayalam_Nya Malayalam_Nya"),
|
|
Value(U_JG_MALAYALAM_RA, "Malayalam_Ra Malayalam_Ra"),
|
|
Value(U_JG_MALAYALAM_SSA, "Malayalam_Ssa Malayalam_Ssa"),
|
|
Value(U_JG_MALAYALAM_TTA, "Malayalam_Tta Malayalam_Tta"),
|
|
Value(U_JG_HANIFI_ROHINGYA_KINNA_YA, "Hanifi_Rohingya_Kinna_Ya Hanifi_Rohingya_Kinna_Ya"),
|
|
Value(U_JG_HANIFI_ROHINGYA_PA, "Hanifi_Rohingya_Pa Hanifi_Rohingya_Pa"),
|
|
Value(U_JG_THIN_YEH, "Thin_Yeh Thin_Yeh"),
|
|
Value(U_JG_VERTICAL_TAIL, "Vertical_Tail Vertical_Tail"),
|
|
};
|
|
|
|
static const Value VALUES_jt[6] = {
|
|
Value(U_JT_NON_JOINING, "U Non_Joining"),
|
|
Value(U_JT_JOIN_CAUSING, "C Join_Causing"),
|
|
Value(U_JT_DUAL_JOINING, "D Dual_Joining"),
|
|
Value(U_JT_LEFT_JOINING, "L Left_Joining"),
|
|
Value(U_JT_RIGHT_JOINING, "R Right_Joining"),
|
|
Value(U_JT_TRANSPARENT, "T Transparent"),
|
|
};
|
|
|
|
static const Value VALUES_lb[48] = {
|
|
Value(U_LB_UNKNOWN, "XX Unknown"),
|
|
Value(U_LB_AMBIGUOUS, "AI Ambiguous"),
|
|
Value(U_LB_ALPHABETIC, "AL Alphabetic"),
|
|
Value(U_LB_BREAK_BOTH, "B2 Break_Both"),
|
|
Value(U_LB_BREAK_AFTER, "BA Break_After"),
|
|
Value(U_LB_BREAK_BEFORE, "BB Break_Before"),
|
|
Value(U_LB_MANDATORY_BREAK, "BK Mandatory_Break"),
|
|
Value(U_LB_CONTINGENT_BREAK, "CB Contingent_Break"),
|
|
Value(U_LB_CLOSE_PUNCTUATION, "CL Close_Punctuation"),
|
|
Value(U_LB_COMBINING_MARK, "CM Combining_Mark"),
|
|
Value(U_LB_CARRIAGE_RETURN, "CR Carriage_Return"),
|
|
Value(U_LB_EXCLAMATION, "EX Exclamation"),
|
|
Value(U_LB_GLUE, "GL Glue"),
|
|
Value(U_LB_HYPHEN, "HY Hyphen"),
|
|
Value(U_LB_IDEOGRAPHIC, "ID Ideographic"),
|
|
Value(U_LB_INSEPARABLE, "IN Inseparable Inseperable"),
|
|
Value(U_LB_INFIX_NUMERIC, "IS Infix_Numeric"),
|
|
Value(U_LB_LINE_FEED, "LF Line_Feed"),
|
|
Value(U_LB_NONSTARTER, "NS Nonstarter"),
|
|
Value(U_LB_NUMERIC, "NU Numeric"),
|
|
Value(U_LB_OPEN_PUNCTUATION, "OP Open_Punctuation"),
|
|
Value(U_LB_POSTFIX_NUMERIC, "PO Postfix_Numeric"),
|
|
Value(U_LB_PREFIX_NUMERIC, "PR Prefix_Numeric"),
|
|
Value(U_LB_QUOTATION, "QU Quotation"),
|
|
Value(U_LB_COMPLEX_CONTEXT, "SA Complex_Context"),
|
|
Value(U_LB_SURROGATE, "SG Surrogate"),
|
|
Value(U_LB_SPACE, "SP Space"),
|
|
Value(U_LB_BREAK_SYMBOLS, "SY Break_Symbols"),
|
|
Value(U_LB_ZWSPACE, "ZW ZWSpace"),
|
|
Value(U_LB_NEXT_LINE, "NL Next_Line"),
|
|
Value(U_LB_WORD_JOINER, "WJ Word_Joiner"),
|
|
Value(U_LB_H2, "H2 H2"),
|
|
Value(U_LB_H3, "H3 H3"),
|
|
Value(U_LB_JL, "JL JL"),
|
|
Value(U_LB_JT, "JT JT"),
|
|
Value(U_LB_JV, "JV JV"),
|
|
Value(U_LB_CLOSE_PARENTHESIS, "CP Close_Parenthesis"),
|
|
Value(U_LB_CONDITIONAL_JAPANESE_STARTER, "CJ Conditional_Japanese_Starter"),
|
|
Value(U_LB_HEBREW_LETTER, "HL Hebrew_Letter"),
|
|
Value(U_LB_REGIONAL_INDICATOR, "RI Regional_Indicator"),
|
|
Value(U_LB_E_BASE, "EB E_Base"),
|
|
Value(U_LB_E_MODIFIER, "EM E_Modifier"),
|
|
Value(U_LB_ZWJ, "ZWJ ZWJ"),
|
|
Value(U_LB_AKSARA, "AK Aksara"),
|
|
Value(U_LB_AKSARA_PREBASE, "AP Aksara_Prebase"),
|
|
Value(U_LB_AKSARA_START, "AS Aksara_Start"),
|
|
Value(U_LB_VIRAMA_FINAL, "VF Virama_Final"),
|
|
Value(U_LB_VIRAMA, "VI Virama"),
|
|
};
|
|
|
|
static const Value VALUES_nt[4] = {
|
|
Value(U_NT_NONE, "None None"),
|
|
Value(U_NT_DECIMAL, "De Decimal"),
|
|
Value(U_NT_DIGIT, "Di Digit"),
|
|
Value(U_NT_NUMERIC, "Nu Numeric"),
|
|
};
|
|
|
|
static const Value VALUES_sc[208] = {
|
|
Value(USCRIPT_COMMON, "Zyyy Common"),
|
|
Value(USCRIPT_INHERITED, "Zinh Inherited Qaai"),
|
|
Value(USCRIPT_ARABIC, "Arab Arabic"),
|
|
Value(USCRIPT_ARMENIAN, "Armn Armenian"),
|
|
Value(USCRIPT_BENGALI, "Beng Bengali"),
|
|
Value(USCRIPT_BOPOMOFO, "Bopo Bopomofo"),
|
|
Value(USCRIPT_CHEROKEE, "Cher Cherokee"),
|
|
Value(USCRIPT_COPTIC, "Copt Coptic Qaac"),
|
|
Value(USCRIPT_CYRILLIC, "Cyrl Cyrillic"),
|
|
Value(USCRIPT_DESERET, "Dsrt Deseret"),
|
|
Value(USCRIPT_DEVANAGARI, "Deva Devanagari"),
|
|
Value(USCRIPT_ETHIOPIC, "Ethi Ethiopic"),
|
|
Value(USCRIPT_GEORGIAN, "Geor Georgian"),
|
|
Value(USCRIPT_GOTHIC, "Goth Gothic"),
|
|
Value(USCRIPT_GREEK, "Grek Greek"),
|
|
Value(USCRIPT_GUJARATI, "Gujr Gujarati"),
|
|
Value(USCRIPT_GURMUKHI, "Guru Gurmukhi"),
|
|
Value(USCRIPT_HAN, "Hani Han"),
|
|
Value(USCRIPT_HANGUL, "Hang Hangul"),
|
|
Value(USCRIPT_HEBREW, "Hebr Hebrew"),
|
|
Value(USCRIPT_HIRAGANA, "Hira Hiragana"),
|
|
Value(USCRIPT_KANNADA, "Knda Kannada"),
|
|
Value(USCRIPT_KATAKANA, "Kana Katakana"),
|
|
Value(USCRIPT_KHMER, "Khmr Khmer"),
|
|
Value(USCRIPT_LAO, "Laoo Lao"),
|
|
Value(USCRIPT_LATIN, "Latn Latin"),
|
|
Value(USCRIPT_MALAYALAM, "Mlym Malayalam"),
|
|
Value(USCRIPT_MONGOLIAN, "Mong Mongolian"),
|
|
Value(USCRIPT_MYANMAR, "Mymr Myanmar"),
|
|
Value(USCRIPT_OGHAM, "Ogam Ogham"),
|
|
Value(USCRIPT_OLD_ITALIC, "Ital Old_Italic"),
|
|
Value(USCRIPT_ORIYA, "Orya Oriya"),
|
|
Value(USCRIPT_RUNIC, "Runr Runic"),
|
|
Value(USCRIPT_SINHALA, "Sinh Sinhala"),
|
|
Value(USCRIPT_SYRIAC, "Syrc Syriac"),
|
|
Value(USCRIPT_TAMIL, "Taml Tamil"),
|
|
Value(USCRIPT_TELUGU, "Telu Telugu"),
|
|
Value(USCRIPT_THAANA, "Thaa Thaana"),
|
|
Value(USCRIPT_THAI, "Thai Thai"),
|
|
Value(USCRIPT_TIBETAN, "Tibt Tibetan"),
|
|
Value(USCRIPT_CANADIAN_ABORIGINAL, "Cans Canadian_Aboriginal"),
|
|
Value(USCRIPT_YI, "Yiii Yi"),
|
|
Value(USCRIPT_TAGALOG, "Tglg Tagalog"),
|
|
Value(USCRIPT_HANUNOO, "Hano Hanunoo"),
|
|
Value(USCRIPT_BUHID, "Buhd Buhid"),
|
|
Value(USCRIPT_TAGBANWA, "Tagb Tagbanwa"),
|
|
Value(USCRIPT_BRAILLE, "Brai Braille"),
|
|
Value(USCRIPT_CYPRIOT, "Cprt Cypriot"),
|
|
Value(USCRIPT_LIMBU, "Limb Limbu"),
|
|
Value(USCRIPT_LINEAR_B, "Linb Linear_B"),
|
|
Value(USCRIPT_OSMANYA, "Osma Osmanya"),
|
|
Value(USCRIPT_SHAVIAN, "Shaw Shavian"),
|
|
Value(USCRIPT_TAI_LE, "Tale Tai_Le"),
|
|
Value(USCRIPT_UGARITIC, "Ugar Ugaritic"),
|
|
Value(USCRIPT_KATAKANA_OR_HIRAGANA, "Hrkt Katakana_Or_Hiragana"),
|
|
Value(USCRIPT_BUGINESE, "Bugi Buginese"),
|
|
Value(USCRIPT_GLAGOLITIC, "Glag Glagolitic"),
|
|
Value(USCRIPT_KHAROSHTHI, "Khar Kharoshthi"),
|
|
Value(USCRIPT_SYLOTI_NAGRI, "Sylo Syloti_Nagri"),
|
|
Value(USCRIPT_NEW_TAI_LUE, "Talu New_Tai_Lue"),
|
|
Value(USCRIPT_TIFINAGH, "Tfng Tifinagh"),
|
|
Value(USCRIPT_OLD_PERSIAN, "Xpeo Old_Persian"),
|
|
Value(USCRIPT_BALINESE, "Bali Balinese"),
|
|
Value(USCRIPT_BATAK, "Batk Batak"),
|
|
Value(USCRIPT_BLISSYMBOLS, "Blis Blis"),
|
|
Value(USCRIPT_BRAHMI, "Brah Brahmi"),
|
|
Value(USCRIPT_CHAM, "Cham Cham"),
|
|
Value(USCRIPT_CIRTH, "Cirt Cirt"),
|
|
Value(USCRIPT_OLD_CHURCH_SLAVONIC_CYRILLIC, "Cyrs Cyrs"),
|
|
Value(USCRIPT_DEMOTIC_EGYPTIAN, "Egyd Egyd"),
|
|
Value(USCRIPT_HIERATIC_EGYPTIAN, "Egyh Egyh"),
|
|
Value(USCRIPT_EGYPTIAN_HIEROGLYPHS, "Egyp Egyptian_Hieroglyphs"),
|
|
Value(USCRIPT_KHUTSURI, "Geok Geok"),
|
|
Value(USCRIPT_SIMPLIFIED_HAN, "Hans Hans"),
|
|
Value(USCRIPT_TRADITIONAL_HAN, "Hant Hant"),
|
|
Value(USCRIPT_PAHAWH_HMONG, "Hmng Pahawh_Hmong"),
|
|
Value(USCRIPT_OLD_HUNGARIAN, "Hung Old_Hungarian"),
|
|
Value(USCRIPT_HARAPPAN_INDUS, "Inds Inds"),
|
|
Value(USCRIPT_JAVANESE, "Java Javanese"),
|
|
Value(USCRIPT_KAYAH_LI, "Kali Kayah_Li"),
|
|
Value(USCRIPT_LATIN_FRAKTUR, "Latf Latf"),
|
|
Value(USCRIPT_LATIN_GAELIC, "Latg Latg"),
|
|
Value(USCRIPT_LEPCHA, "Lepc Lepcha"),
|
|
Value(USCRIPT_LINEAR_A, "Lina Linear_A"),
|
|
Value(USCRIPT_MANDAIC, "Mand Mandaic"),
|
|
Value(USCRIPT_MAYAN_HIEROGLYPHS, "Maya Maya"),
|
|
Value(USCRIPT_MEROITIC_HIEROGLYPHS, "Mero Meroitic_Hieroglyphs"),
|
|
Value(USCRIPT_NKO, "Nkoo Nko"),
|
|
Value(USCRIPT_ORKHON, "Orkh Old_Turkic"),
|
|
Value(USCRIPT_OLD_PERMIC, "Perm Old_Permic"),
|
|
Value(USCRIPT_PHAGS_PA, "Phag Phags_Pa"),
|
|
Value(USCRIPT_PHOENICIAN, "Phnx Phoenician"),
|
|
Value(USCRIPT_MIAO, "Plrd Miao"),
|
|
Value(USCRIPT_RONGORONGO, "Roro Roro"),
|
|
Value(USCRIPT_SARATI, "Sara Sara"),
|
|
Value(USCRIPT_ESTRANGELO_SYRIAC, "Syre Syre"),
|
|
Value(USCRIPT_WESTERN_SYRIAC, "Syrj Syrj"),
|
|
Value(USCRIPT_EASTERN_SYRIAC, "Syrn Syrn"),
|
|
Value(USCRIPT_TENGWAR, "Teng Teng"),
|
|
Value(USCRIPT_VAI, "Vaii Vai"),
|
|
Value(USCRIPT_VISIBLE_SPEECH, "Visp Visp"),
|
|
Value(USCRIPT_CUNEIFORM, "Xsux Cuneiform"),
|
|
Value(USCRIPT_UNWRITTEN_LANGUAGES, "Zxxx Zxxx"),
|
|
Value(USCRIPT_UNKNOWN, "Zzzz Unknown"),
|
|
Value(USCRIPT_CARIAN, "Cari Carian"),
|
|
Value(USCRIPT_JAPANESE, "Jpan Jpan"),
|
|
Value(USCRIPT_LANNA, "Lana Tai_Tham"),
|
|
Value(USCRIPT_LYCIAN, "Lyci Lycian"),
|
|
Value(USCRIPT_LYDIAN, "Lydi Lydian"),
|
|
Value(USCRIPT_OL_CHIKI, "Olck Ol_Chiki"),
|
|
Value(USCRIPT_REJANG, "Rjng Rejang"),
|
|
Value(USCRIPT_SAURASHTRA, "Saur Saurashtra"),
|
|
Value(USCRIPT_SIGN_WRITING, "Sgnw SignWriting"),
|
|
Value(USCRIPT_SUNDANESE, "Sund Sundanese"),
|
|
Value(USCRIPT_MOON, "Moon Moon"),
|
|
Value(USCRIPT_MEITEI_MAYEK, "Mtei Meetei_Mayek"),
|
|
Value(USCRIPT_IMPERIAL_ARAMAIC, "Armi Imperial_Aramaic"),
|
|
Value(USCRIPT_AVESTAN, "Avst Avestan"),
|
|
Value(USCRIPT_CHAKMA, "Cakm Chakma"),
|
|
Value(USCRIPT_KOREAN, "Kore Kore"),
|
|
Value(USCRIPT_KAITHI, "Kthi Kaithi"),
|
|
Value(USCRIPT_MANICHAEAN, "Mani Manichaean"),
|
|
Value(USCRIPT_INSCRIPTIONAL_PAHLAVI, "Phli Inscriptional_Pahlavi"),
|
|
Value(USCRIPT_PSALTER_PAHLAVI, "Phlp Psalter_Pahlavi"),
|
|
Value(USCRIPT_BOOK_PAHLAVI, "Phlv Phlv"),
|
|
Value(USCRIPT_INSCRIPTIONAL_PARTHIAN, "Prti Inscriptional_Parthian"),
|
|
Value(USCRIPT_SAMARITAN, "Samr Samaritan"),
|
|
Value(USCRIPT_TAI_VIET, "Tavt Tai_Viet"),
|
|
Value(USCRIPT_MATHEMATICAL_NOTATION, "Zmth Zmth"),
|
|
Value(USCRIPT_SYMBOLS, "Zsym Zsym"),
|
|
Value(USCRIPT_BAMUM, "Bamu Bamum"),
|
|
Value(USCRIPT_LISU, "Lisu Lisu"),
|
|
Value(USCRIPT_NAKHI_GEBA, "Nkgb Nkgb"),
|
|
Value(USCRIPT_OLD_SOUTH_ARABIAN, "Sarb Old_South_Arabian"),
|
|
Value(USCRIPT_BASSA_VAH, "Bass Bassa_Vah"),
|
|
Value(USCRIPT_DUPLOYAN, "Dupl Duployan"),
|
|
Value(USCRIPT_ELBASAN, "Elba Elbasan"),
|
|
Value(USCRIPT_GRANTHA, "Gran Grantha"),
|
|
Value(USCRIPT_KPELLE, "Kpel Kpel"),
|
|
Value(USCRIPT_LOMA, "Loma Loma"),
|
|
Value(USCRIPT_MENDE, "Mend Mende_Kikakui"),
|
|
Value(USCRIPT_MEROITIC_CURSIVE, "Merc Meroitic_Cursive"),
|
|
Value(USCRIPT_OLD_NORTH_ARABIAN, "Narb Old_North_Arabian"),
|
|
Value(USCRIPT_NABATAEAN, "Nbat Nabataean"),
|
|
Value(USCRIPT_PALMYRENE, "Palm Palmyrene"),
|
|
Value(USCRIPT_KHUDAWADI, "Sind Khudawadi"),
|
|
Value(USCRIPT_WARANG_CITI, "Wara Warang_Citi"),
|
|
Value(USCRIPT_AFAKA, "Afak Afak"),
|
|
Value(USCRIPT_JURCHEN, "Jurc Jurc"),
|
|
Value(USCRIPT_MRO, "Mroo Mro"),
|
|
Value(USCRIPT_NUSHU, "Nshu Nushu"),
|
|
Value(USCRIPT_SHARADA, "Shrd Sharada"),
|
|
Value(USCRIPT_SORA_SOMPENG, "Sora Sora_Sompeng"),
|
|
Value(USCRIPT_TAKRI, "Takr Takri"),
|
|
Value(USCRIPT_TANGUT, "Tang Tangut"),
|
|
Value(USCRIPT_WOLEAI, "Wole Wole"),
|
|
Value(USCRIPT_ANATOLIAN_HIEROGLYPHS, "Hluw Anatolian_Hieroglyphs"),
|
|
Value(USCRIPT_KHOJKI, "Khoj Khojki"),
|
|
Value(USCRIPT_TIRHUTA, "Tirh Tirhuta"),
|
|
Value(USCRIPT_CAUCASIAN_ALBANIAN, "Aghb Caucasian_Albanian"),
|
|
Value(USCRIPT_MAHAJANI, "Mahj Mahajani"),
|
|
Value(USCRIPT_AHOM, "Ahom Ahom"),
|
|
Value(USCRIPT_HATRAN, "Hatr Hatran"),
|
|
Value(USCRIPT_MODI, "Modi Modi"),
|
|
Value(USCRIPT_MULTANI, "Mult Multani"),
|
|
Value(USCRIPT_PAU_CIN_HAU, "Pauc Pau_Cin_Hau"),
|
|
Value(USCRIPT_SIDDHAM, "Sidd Siddham"),
|
|
Value(USCRIPT_ADLAM, "Adlm Adlam"),
|
|
Value(USCRIPT_BHAIKSUKI, "Bhks Bhaiksuki"),
|
|
Value(USCRIPT_MARCHEN, "Marc Marchen"),
|
|
Value(USCRIPT_NEWA, "Newa Newa"),
|
|
Value(USCRIPT_OSAGE, "Osge Osage"),
|
|
Value(USCRIPT_HAN_WITH_BOPOMOFO, "Hanb Hanb"),
|
|
Value(USCRIPT_JAMO, "Jamo Jamo"),
|
|
Value(USCRIPT_SYMBOLS_EMOJI, "Zsye Zsye"),
|
|
Value(USCRIPT_MASARAM_GONDI, "Gonm Masaram_Gondi"),
|
|
Value(USCRIPT_SOYOMBO, "Soyo Soyombo"),
|
|
Value(USCRIPT_ZANABAZAR_SQUARE, "Zanb Zanabazar_Square"),
|
|
Value(USCRIPT_DOGRA, "Dogr Dogra"),
|
|
Value(USCRIPT_GUNJALA_GONDI, "Gong Gunjala_Gondi"),
|
|
Value(USCRIPT_MAKASAR, "Maka Makasar"),
|
|
Value(USCRIPT_MEDEFAIDRIN, "Medf Medefaidrin"),
|
|
Value(USCRIPT_HANIFI_ROHINGYA, "Rohg Hanifi_Rohingya"),
|
|
Value(USCRIPT_SOGDIAN, "Sogd Sogdian"),
|
|
Value(USCRIPT_OLD_SOGDIAN, "Sogo Old_Sogdian"),
|
|
Value(USCRIPT_ELYMAIC, "Elym Elymaic"),
|
|
Value(USCRIPT_NYIAKENG_PUACHUE_HMONG, "Hmnp Nyiakeng_Puachue_Hmong"),
|
|
Value(USCRIPT_NANDINAGARI, "Nand Nandinagari"),
|
|
Value(USCRIPT_WANCHO, "Wcho Wancho"),
|
|
Value(USCRIPT_CHORASMIAN, "Chrs Chorasmian"),
|
|
Value(USCRIPT_DIVES_AKURU, "Diak Dives_Akuru"),
|
|
Value(USCRIPT_KHITAN_SMALL_SCRIPT, "Kits Khitan_Small_Script"),
|
|
Value(USCRIPT_YEZIDI, "Yezi Yezidi"),
|
|
Value(USCRIPT_CYPRO_MINOAN, "Cpmn Cypro_Minoan"),
|
|
Value(USCRIPT_OLD_UYGHUR, "Ougr Old_Uyghur"),
|
|
Value(USCRIPT_TANGSA, "Tnsa Tangsa"),
|
|
Value(USCRIPT_TOTO, "Toto Toto"),
|
|
Value(USCRIPT_VITHKUQI, "Vith Vithkuqi"),
|
|
Value(USCRIPT_KAWI, "Kawi Kawi"),
|
|
Value(USCRIPT_NAG_MUNDARI, "Nagm Nag_Mundari"),
|
|
Value(USCRIPT_ARABIC_NASTALIQ, "Aran Aran"),
|
|
Value(USCRIPT_GARAY, "Gara Garay"),
|
|
Value(USCRIPT_GURUNG_KHEMA, "Gukh Gurung_Khema"),
|
|
Value(USCRIPT_KIRAT_RAI, "Krai Kirat_Rai"),
|
|
Value(USCRIPT_OL_ONAL, "Onao Ol_Onal"),
|
|
Value(USCRIPT_SUNUWAR, "Sunu Sunuwar"),
|
|
Value(USCRIPT_TODHRI, "Todr Todhri"),
|
|
Value(USCRIPT_TULU_TIGALARI, "Tutg Tulu_Tigalari"),
|
|
};
|
|
|
|
static const Value VALUES_hst[6] = {
|
|
Value(U_HST_NOT_APPLICABLE, "NA Not_Applicable"),
|
|
Value(U_HST_LEADING_JAMO, "L Leading_Jamo"),
|
|
Value(U_HST_VOWEL_JAMO, "V Vowel_Jamo"),
|
|
Value(U_HST_TRAILING_JAMO, "T Trailing_Jamo"),
|
|
Value(U_HST_LV_SYLLABLE, "LV LV_Syllable"),
|
|
Value(U_HST_LVT_SYLLABLE, "LVT LVT_Syllable"),
|
|
};
|
|
|
|
static const Value VALUES_NFD_QC[2] = {
|
|
Value(UNORM_NO, "N No"),
|
|
Value(UNORM_YES, "Y Yes"),
|
|
};
|
|
|
|
static const Value VALUES_NFKD_QC[2] = {
|
|
Value(UNORM_NO, "N No"),
|
|
Value(UNORM_YES, "Y Yes"),
|
|
};
|
|
|
|
static const Value VALUES_NFC_QC[3] = {
|
|
Value(UNORM_NO, "N No"),
|
|
Value(UNORM_YES, "Y Yes"),
|
|
Value(UNORM_MAYBE, "M Maybe"),
|
|
};
|
|
|
|
static const Value VALUES_NFKC_QC[3] = {
|
|
Value(UNORM_NO, "N No"),
|
|
Value(UNORM_YES, "Y Yes"),
|
|
Value(UNORM_MAYBE, "M Maybe"),
|
|
};
|
|
|
|
static const Value VALUES_lccc[58] = {
|
|
Value(0, "NR Not_Reordered"),
|
|
Value(1, "OV Overlay"),
|
|
Value(6, "HANR Han_Reading"),
|
|
Value(7, "NK Nukta"),
|
|
Value(8, "KV Kana_Voicing"),
|
|
Value(9, "VR Virama"),
|
|
Value(10, "CCC10 CCC10"),
|
|
Value(11, "CCC11 CCC11"),
|
|
Value(12, "CCC12 CCC12"),
|
|
Value(13, "CCC13 CCC13"),
|
|
Value(14, "CCC14 CCC14"),
|
|
Value(15, "CCC15 CCC15"),
|
|
Value(16, "CCC16 CCC16"),
|
|
Value(17, "CCC17 CCC17"),
|
|
Value(18, "CCC18 CCC18"),
|
|
Value(19, "CCC19 CCC19"),
|
|
Value(20, "CCC20 CCC20"),
|
|
Value(21, "CCC21 CCC21"),
|
|
Value(22, "CCC22 CCC22"),
|
|
Value(23, "CCC23 CCC23"),
|
|
Value(24, "CCC24 CCC24"),
|
|
Value(25, "CCC25 CCC25"),
|
|
Value(26, "CCC26 CCC26"),
|
|
Value(27, "CCC27 CCC27"),
|
|
Value(28, "CCC28 CCC28"),
|
|
Value(29, "CCC29 CCC29"),
|
|
Value(30, "CCC30 CCC30"),
|
|
Value(31, "CCC31 CCC31"),
|
|
Value(32, "CCC32 CCC32"),
|
|
Value(33, "CCC33 CCC33"),
|
|
Value(34, "CCC34 CCC34"),
|
|
Value(35, "CCC35 CCC35"),
|
|
Value(36, "CCC36 CCC36"),
|
|
Value(84, "CCC84 CCC84"),
|
|
Value(91, "CCC91 CCC91"),
|
|
Value(103, "CCC103 CCC103"),
|
|
Value(107, "CCC107 CCC107"),
|
|
Value(118, "CCC118 CCC118"),
|
|
Value(122, "CCC122 CCC122"),
|
|
Value(129, "CCC129 CCC129"),
|
|
Value(130, "CCC130 CCC130"),
|
|
Value(132, "CCC132 CCC132"),
|
|
Value(133, "CCC133 CCC133"),
|
|
Value(200, "ATBL Attached_Below_Left"),
|
|
Value(202, "ATB Attached_Below"),
|
|
Value(214, "ATA Attached_Above"),
|
|
Value(216, "ATAR Attached_Above_Right"),
|
|
Value(218, "BL Below_Left"),
|
|
Value(220, "B Below"),
|
|
Value(222, "BR Below_Right"),
|
|
Value(224, "L Left"),
|
|
Value(226, "R Right"),
|
|
Value(228, "AL Above_Left"),
|
|
Value(230, "A Above"),
|
|
Value(232, "AR Above_Right"),
|
|
Value(233, "DB Double_Below"),
|
|
Value(234, "DA Double_Above"),
|
|
Value(240, "IS Iota_Subscript"),
|
|
};
|
|
|
|
static const Value VALUES_tccc[58] = {
|
|
Value(0, "NR Not_Reordered"),
|
|
Value(1, "OV Overlay"),
|
|
Value(6, "HANR Han_Reading"),
|
|
Value(7, "NK Nukta"),
|
|
Value(8, "KV Kana_Voicing"),
|
|
Value(9, "VR Virama"),
|
|
Value(10, "CCC10 CCC10"),
|
|
Value(11, "CCC11 CCC11"),
|
|
Value(12, "CCC12 CCC12"),
|
|
Value(13, "CCC13 CCC13"),
|
|
Value(14, "CCC14 CCC14"),
|
|
Value(15, "CCC15 CCC15"),
|
|
Value(16, "CCC16 CCC16"),
|
|
Value(17, "CCC17 CCC17"),
|
|
Value(18, "CCC18 CCC18"),
|
|
Value(19, "CCC19 CCC19"),
|
|
Value(20, "CCC20 CCC20"),
|
|
Value(21, "CCC21 CCC21"),
|
|
Value(22, "CCC22 CCC22"),
|
|
Value(23, "CCC23 CCC23"),
|
|
Value(24, "CCC24 CCC24"),
|
|
Value(25, "CCC25 CCC25"),
|
|
Value(26, "CCC26 CCC26"),
|
|
Value(27, "CCC27 CCC27"),
|
|
Value(28, "CCC28 CCC28"),
|
|
Value(29, "CCC29 CCC29"),
|
|
Value(30, "CCC30 CCC30"),
|
|
Value(31, "CCC31 CCC31"),
|
|
Value(32, "CCC32 CCC32"),
|
|
Value(33, "CCC33 CCC33"),
|
|
Value(34, "CCC34 CCC34"),
|
|
Value(35, "CCC35 CCC35"),
|
|
Value(36, "CCC36 CCC36"),
|
|
Value(84, "CCC84 CCC84"),
|
|
Value(91, "CCC91 CCC91"),
|
|
Value(103, "CCC103 CCC103"),
|
|
Value(107, "CCC107 CCC107"),
|
|
Value(118, "CCC118 CCC118"),
|
|
Value(122, "CCC122 CCC122"),
|
|
Value(129, "CCC129 CCC129"),
|
|
Value(130, "CCC130 CCC130"),
|
|
Value(132, "CCC132 CCC132"),
|
|
Value(133, "CCC133 CCC133"),
|
|
Value(200, "ATBL Attached_Below_Left"),
|
|
Value(202, "ATB Attached_Below"),
|
|
Value(214, "ATA Attached_Above"),
|
|
Value(216, "ATAR Attached_Above_Right"),
|
|
Value(218, "BL Below_Left"),
|
|
Value(220, "B Below"),
|
|
Value(222, "BR Below_Right"),
|
|
Value(224, "L Left"),
|
|
Value(226, "R Right"),
|
|
Value(228, "AL Above_Left"),
|
|
Value(230, "A Above"),
|
|
Value(232, "AR Above_Right"),
|
|
Value(233, "DB Double_Below"),
|
|
Value(234, "DA Double_Above"),
|
|
Value(240, "IS Iota_Subscript"),
|
|
};
|
|
|
|
static const Value VALUES_GCB[18] = {
|
|
Value(U_GCB_OTHER, "XX Other"),
|
|
Value(U_GCB_CONTROL, "CN Control"),
|
|
Value(U_GCB_CR, "CR CR"),
|
|
Value(U_GCB_EXTEND, "EX Extend"),
|
|
Value(U_GCB_L, "L L"),
|
|
Value(U_GCB_LF, "LF LF"),
|
|
Value(U_GCB_LV, "LV LV"),
|
|
Value(U_GCB_LVT, "LVT LVT"),
|
|
Value(U_GCB_T, "T T"),
|
|
Value(U_GCB_V, "V V"),
|
|
Value(U_GCB_SPACING_MARK, "SM SpacingMark"),
|
|
Value(U_GCB_PREPEND, "PP Prepend"),
|
|
Value(U_GCB_REGIONAL_INDICATOR, "RI Regional_Indicator"),
|
|
Value(U_GCB_E_BASE, "EB E_Base"),
|
|
Value(U_GCB_E_BASE_GAZ, "EBG E_Base_GAZ"),
|
|
Value(U_GCB_E_MODIFIER, "EM E_Modifier"),
|
|
Value(U_GCB_GLUE_AFTER_ZWJ, "GAZ Glue_After_Zwj"),
|
|
Value(U_GCB_ZWJ, "ZWJ ZWJ"),
|
|
};
|
|
|
|
static const Value VALUES_SB[15] = {
|
|
Value(U_SB_OTHER, "XX Other"),
|
|
Value(U_SB_ATERM, "AT ATerm"),
|
|
Value(U_SB_CLOSE, "CL Close"),
|
|
Value(U_SB_FORMAT, "FO Format"),
|
|
Value(U_SB_LOWER, "LO Lower"),
|
|
Value(U_SB_NUMERIC, "NU Numeric"),
|
|
Value(U_SB_OLETTER, "LE OLetter"),
|
|
Value(U_SB_SEP, "SE Sep"),
|
|
Value(U_SB_SP, "SP Sp"),
|
|
Value(U_SB_STERM, "ST STerm"),
|
|
Value(U_SB_UPPER, "UP Upper"),
|
|
Value(U_SB_CR, "CR CR"),
|
|
Value(U_SB_EXTEND, "EX Extend"),
|
|
Value(U_SB_LF, "LF LF"),
|
|
Value(U_SB_SCONTINUE, "SC SContinue"),
|
|
};
|
|
|
|
static const Value VALUES_WB[23] = {
|
|
Value(U_WB_OTHER, "XX Other"),
|
|
Value(U_WB_ALETTER, "LE ALetter"),
|
|
Value(U_WB_FORMAT, "FO Format"),
|
|
Value(U_WB_KATAKANA, "KA Katakana"),
|
|
Value(U_WB_MIDLETTER, "ML MidLetter"),
|
|
Value(U_WB_MIDNUM, "MN MidNum"),
|
|
Value(U_WB_NUMERIC, "NU Numeric"),
|
|
Value(U_WB_EXTENDNUMLET, "EX ExtendNumLet"),
|
|
Value(U_WB_CR, "CR CR"),
|
|
Value(U_WB_EXTEND, "Extend Extend"),
|
|
Value(U_WB_LF, "LF LF"),
|
|
Value(U_WB_MIDNUMLET, "MB MidNumLet"),
|
|
Value(U_WB_NEWLINE, "NL Newline"),
|
|
Value(U_WB_REGIONAL_INDICATOR, "RI Regional_Indicator"),
|
|
Value(U_WB_HEBREW_LETTER, "HL Hebrew_Letter"),
|
|
Value(U_WB_SINGLE_QUOTE, "SQ Single_Quote"),
|
|
Value(U_WB_DOUBLE_QUOTE, "DQ Double_Quote"),
|
|
Value(U_WB_E_BASE, "EB E_Base"),
|
|
Value(U_WB_E_BASE_GAZ, "EBG E_Base_GAZ"),
|
|
Value(U_WB_E_MODIFIER, "EM E_Modifier"),
|
|
Value(U_WB_GLUE_AFTER_ZWJ, "GAZ Glue_After_Zwj"),
|
|
Value(U_WB_ZWJ, "ZWJ ZWJ"),
|
|
Value(U_WB_WSEGSPACE, "WSegSpace WSegSpace"),
|
|
};
|
|
|
|
static const Value VALUES_bpt[3] = {
|
|
Value(U_BPT_NONE, "n None"),
|
|
Value(U_BPT_OPEN, "o Open"),
|
|
Value(U_BPT_CLOSE, "c Close"),
|
|
};
|
|
|
|
static const Value VALUES_InPC[16] = {
|
|
Value(U_INPC_NA, "NA NA"),
|
|
Value(U_INPC_BOTTOM, "Bottom Bottom"),
|
|
Value(U_INPC_BOTTOM_AND_LEFT, "Bottom_And_Left Bottom_And_Left"),
|
|
Value(U_INPC_BOTTOM_AND_RIGHT, "Bottom_And_Right Bottom_And_Right"),
|
|
Value(U_INPC_LEFT, "Left Left"),
|
|
Value(U_INPC_LEFT_AND_RIGHT, "Left_And_Right Left_And_Right"),
|
|
Value(U_INPC_OVERSTRUCK, "Overstruck Overstruck"),
|
|
Value(U_INPC_RIGHT, "Right Right"),
|
|
Value(U_INPC_TOP, "Top Top"),
|
|
Value(U_INPC_TOP_AND_BOTTOM, "Top_And_Bottom Top_And_Bottom"),
|
|
Value(U_INPC_TOP_AND_BOTTOM_AND_RIGHT, "Top_And_Bottom_And_Right Top_And_Bottom_And_Right"),
|
|
Value(U_INPC_TOP_AND_LEFT, "Top_And_Left Top_And_Left"),
|
|
Value(U_INPC_TOP_AND_LEFT_AND_RIGHT, "Top_And_Left_And_Right Top_And_Left_And_Right"),
|
|
Value(U_INPC_TOP_AND_RIGHT, "Top_And_Right Top_And_Right"),
|
|
Value(U_INPC_VISUAL_ORDER_LEFT, "Visual_Order_Left Visual_Order_Left"),
|
|
Value(U_INPC_TOP_AND_BOTTOM_AND_LEFT, "Top_And_Bottom_And_Left Top_And_Bottom_And_Left"),
|
|
};
|
|
|
|
static const Value VALUES_InSC[37] = {
|
|
Value(U_INSC_OTHER, "Other Other"),
|
|
Value(U_INSC_AVAGRAHA, "Avagraha Avagraha"),
|
|
Value(U_INSC_BINDU, "Bindu Bindu"),
|
|
Value(U_INSC_BRAHMI_JOINING_NUMBER, "Brahmi_Joining_Number Brahmi_Joining_Number"),
|
|
Value(U_INSC_CANTILLATION_MARK, "Cantillation_Mark Cantillation_Mark"),
|
|
Value(U_INSC_CONSONANT, "Consonant Consonant"),
|
|
Value(U_INSC_CONSONANT_DEAD, "Consonant_Dead Consonant_Dead"),
|
|
Value(U_INSC_CONSONANT_FINAL, "Consonant_Final Consonant_Final"),
|
|
Value(U_INSC_CONSONANT_HEAD_LETTER, "Consonant_Head_Letter Consonant_Head_Letter"),
|
|
Value(U_INSC_CONSONANT_INITIAL_POSTFIXED, "Consonant_Initial_Postfixed Consonant_Initial_Postfixed"),
|
|
Value(U_INSC_CONSONANT_KILLER, "Consonant_Killer Consonant_Killer"),
|
|
Value(U_INSC_CONSONANT_MEDIAL, "Consonant_Medial Consonant_Medial"),
|
|
Value(U_INSC_CONSONANT_PLACEHOLDER, "Consonant_Placeholder Consonant_Placeholder"),
|
|
Value(U_INSC_CONSONANT_PRECEDING_REPHA, "Consonant_Preceding_Repha Consonant_Preceding_Repha"),
|
|
Value(U_INSC_CONSONANT_PREFIXED, "Consonant_Prefixed Consonant_Prefixed"),
|
|
Value(U_INSC_CONSONANT_SUBJOINED, "Consonant_Subjoined Consonant_Subjoined"),
|
|
Value(U_INSC_CONSONANT_SUCCEEDING_REPHA, "Consonant_Succeeding_Repha Consonant_Succeeding_Repha"),
|
|
Value(U_INSC_CONSONANT_WITH_STACKER, "Consonant_With_Stacker Consonant_With_Stacker"),
|
|
Value(U_INSC_GEMINATION_MARK, "Gemination_Mark Gemination_Mark"),
|
|
Value(U_INSC_INVISIBLE_STACKER, "Invisible_Stacker Invisible_Stacker"),
|
|
Value(U_INSC_JOINER, "Joiner Joiner"),
|
|
Value(U_INSC_MODIFYING_LETTER, "Modifying_Letter Modifying_Letter"),
|
|
Value(U_INSC_NON_JOINER, "Non_Joiner Non_Joiner"),
|
|
Value(U_INSC_NUKTA, "Nukta Nukta"),
|
|
Value(U_INSC_NUMBER, "Number Number"),
|
|
Value(U_INSC_NUMBER_JOINER, "Number_Joiner Number_Joiner"),
|
|
Value(U_INSC_PURE_KILLER, "Pure_Killer Pure_Killer"),
|
|
Value(U_INSC_REGISTER_SHIFTER, "Register_Shifter Register_Shifter"),
|
|
Value(U_INSC_SYLLABLE_MODIFIER, "Syllable_Modifier Syllable_Modifier"),
|
|
Value(U_INSC_TONE_LETTER, "Tone_Letter Tone_Letter"),
|
|
Value(U_INSC_TONE_MARK, "Tone_Mark Tone_Mark"),
|
|
Value(U_INSC_VIRAMA, "Virama Virama"),
|
|
Value(U_INSC_VISARGA, "Visarga Visarga"),
|
|
Value(U_INSC_VOWEL, "Vowel Vowel"),
|
|
Value(U_INSC_VOWEL_DEPENDENT, "Vowel_Dependent Vowel_Dependent"),
|
|
Value(U_INSC_VOWEL_INDEPENDENT, "Vowel_Independent Vowel_Independent"),
|
|
Value(U_INSC_REORDERING_KILLER, "Reordering_Killer Reordering_Killer"),
|
|
};
|
|
|
|
static const Value VALUES_vo[4] = {
|
|
Value(U_VO_ROTATED, "R Rotated"),
|
|
Value(U_VO_TRANSFORMED_ROTATED, "Tr Transformed_Rotated"),
|
|
Value(U_VO_TRANSFORMED_UPRIGHT, "Tu Transformed_Upright"),
|
|
Value(U_VO_UPRIGHT, "U Upright"),
|
|
};
|
|
|
|
static const Value VALUES_ID_Status[2] = {
|
|
Value(U_ID_STATUS_RESTRICTED, "Restricted Restricted"),
|
|
Value(U_ID_STATUS_ALLOWED, "Allowed Allowed"),
|
|
};
|
|
|
|
static const Value VALUES_InCB[4] = {
|
|
Value(U_INCB_NONE, "None None"),
|
|
Value(U_INCB_CONSONANT, "Consonant Consonant"),
|
|
Value(U_INCB_EXTEND, "Extend Extend"),
|
|
Value(U_INCB_LINKER, "Linker Linker"),
|
|
};
|
|
|
|
static const Value VALUES_gcm[38] = {
|
|
Value((int32_t)U_GC_C_MASK, "C Other"),
|
|
Value((int32_t)U_GC_CC_MASK, "Cc Control cntrl"),
|
|
Value((int32_t)U_GC_CF_MASK, "Cf Format"),
|
|
Value((int32_t)U_GC_CN_MASK, "Cn Unassigned"),
|
|
Value((int32_t)U_GC_CO_MASK, "Co Private_Use"),
|
|
Value((int32_t)U_GC_CS_MASK, "Cs Surrogate"),
|
|
Value((int32_t)U_GC_L_MASK, "L Letter"),
|
|
Value((int32_t)U_GC_LC_MASK, "LC Cased_Letter"),
|
|
Value((int32_t)U_GC_LL_MASK, "Ll Lowercase_Letter"),
|
|
Value((int32_t)U_GC_LM_MASK, "Lm Modifier_Letter"),
|
|
Value((int32_t)U_GC_LO_MASK, "Lo Other_Letter"),
|
|
Value((int32_t)U_GC_LT_MASK, "Lt Titlecase_Letter"),
|
|
Value((int32_t)U_GC_LU_MASK, "Lu Uppercase_Letter"),
|
|
Value((int32_t)U_GC_M_MASK, "M Mark Combining_Mark"),
|
|
Value((int32_t)U_GC_MC_MASK, "Mc Spacing_Mark"),
|
|
Value((int32_t)U_GC_ME_MASK, "Me Enclosing_Mark"),
|
|
Value((int32_t)U_GC_MN_MASK, "Mn Nonspacing_Mark"),
|
|
Value((int32_t)U_GC_N_MASK, "N Number"),
|
|
Value((int32_t)U_GC_ND_MASK, "Nd Decimal_Number digit"),
|
|
Value((int32_t)U_GC_NL_MASK, "Nl Letter_Number"),
|
|
Value((int32_t)U_GC_NO_MASK, "No Other_Number"),
|
|
Value((int32_t)U_GC_P_MASK, "P Punctuation punct"),
|
|
Value((int32_t)U_GC_PC_MASK, "Pc Connector_Punctuation"),
|
|
Value((int32_t)U_GC_PD_MASK, "Pd Dash_Punctuation"),
|
|
Value((int32_t)U_GC_PE_MASK, "Pe Close_Punctuation"),
|
|
Value((int32_t)U_GC_PF_MASK, "Pf Final_Punctuation"),
|
|
Value((int32_t)U_GC_PI_MASK, "Pi Initial_Punctuation"),
|
|
Value((int32_t)U_GC_PO_MASK, "Po Other_Punctuation"),
|
|
Value((int32_t)U_GC_PS_MASK, "Ps Open_Punctuation"),
|
|
Value((int32_t)U_GC_S_MASK, "S Symbol"),
|
|
Value((int32_t)U_GC_SC_MASK, "Sc Currency_Symbol"),
|
|
Value((int32_t)U_GC_SK_MASK, "Sk Modifier_Symbol"),
|
|
Value((int32_t)U_GC_SM_MASK, "Sm Math_Symbol"),
|
|
Value((int32_t)U_GC_SO_MASK, "So Other_Symbol"),
|
|
Value((int32_t)U_GC_Z_MASK, "Z Separator"),
|
|
Value((int32_t)U_GC_ZL_MASK, "Zl Line_Separator"),
|
|
Value((int32_t)U_GC_ZP_MASK, "Zp Paragraph_Separator"),
|
|
Value((int32_t)U_GC_ZS_MASK, "Zs Space_Separator"),
|
|
};
|
|
|
|
static const Value VALUES_ID_Type[12] = {
|
|
Value(U_ID_TYPE_NOT_CHARACTER, "Not_Character Not_Character"),
|
|
Value(U_ID_TYPE_DEPRECATED, "Deprecated Deprecated"),
|
|
Value(U_ID_TYPE_DEFAULT_IGNORABLE, "Default_Ignorable Default_Ignorable"),
|
|
Value(U_ID_TYPE_NOT_NFKC, "Not_NFKC Not_NFKC"),
|
|
Value(U_ID_TYPE_NOT_XID, "Not_XID Not_XID"),
|
|
Value(U_ID_TYPE_EXCLUSION, "Exclusion Exclusion"),
|
|
Value(U_ID_TYPE_OBSOLETE, "Obsolete Obsolete"),
|
|
Value(U_ID_TYPE_TECHNICAL, "Technical Technical"),
|
|
Value(U_ID_TYPE_UNCOMMON_USE, "Uncommon_Use Uncommon_Use"),
|
|
Value(U_ID_TYPE_LIMITED_USE, "Limited_Use Limited_Use"),
|
|
Value(U_ID_TYPE_INCLUSION, "Inclusion Inclusion"),
|
|
Value(U_ID_TYPE_RECOMMENDED, "Recommended Recommended"),
|
|
};
|
|
|
|
static const Property PROPERTIES[120] = {
|
|
Property(UCHAR_ALPHABETIC, "Alpha Alphabetic"),
|
|
Property(UCHAR_ASCII_HEX_DIGIT, "AHex ASCII_Hex_Digit"),
|
|
Property(UCHAR_BIDI_CONTROL, "Bidi_C Bidi_Control"),
|
|
Property(UCHAR_BIDI_MIRRORED, "Bidi_M Bidi_Mirrored"),
|
|
Property(UCHAR_DASH, "Dash Dash"),
|
|
Property(UCHAR_DEFAULT_IGNORABLE_CODE_POINT, "DI Default_Ignorable_Code_Point"),
|
|
Property(UCHAR_DEPRECATED, "Dep Deprecated"),
|
|
Property(UCHAR_DIACRITIC, "Dia Diacritic"),
|
|
Property(UCHAR_EXTENDER, "Ext Extender"),
|
|
Property(UCHAR_FULL_COMPOSITION_EXCLUSION, "Comp_Ex Full_Composition_Exclusion"),
|
|
Property(UCHAR_GRAPHEME_BASE, "Gr_Base Grapheme_Base"),
|
|
Property(UCHAR_GRAPHEME_EXTEND, "Gr_Ext Grapheme_Extend"),
|
|
Property(UCHAR_GRAPHEME_LINK, "Gr_Link Grapheme_Link"),
|
|
Property(UCHAR_HEX_DIGIT, "Hex Hex_Digit"),
|
|
Property(UCHAR_HYPHEN, "Hyphen Hyphen"),
|
|
Property(UCHAR_ID_CONTINUE, "IDC ID_Continue"),
|
|
Property(UCHAR_ID_START, "IDS ID_Start"),
|
|
Property(UCHAR_IDEOGRAPHIC, "Ideo Ideographic"),
|
|
Property(UCHAR_IDS_BINARY_OPERATOR, "IDSB IDS_Binary_Operator"),
|
|
Property(UCHAR_IDS_TRINARY_OPERATOR, "IDST IDS_Trinary_Operator"),
|
|
Property(UCHAR_JOIN_CONTROL, "Join_C Join_Control"),
|
|
Property(UCHAR_LOGICAL_ORDER_EXCEPTION, "LOE Logical_Order_Exception"),
|
|
Property(UCHAR_LOWERCASE, "Lower Lowercase"),
|
|
Property(UCHAR_MATH, "Math Math"),
|
|
Property(UCHAR_NONCHARACTER_CODE_POINT, "NChar Noncharacter_Code_Point"),
|
|
Property(UCHAR_QUOTATION_MARK, "QMark Quotation_Mark"),
|
|
Property(UCHAR_RADICAL, "Radical Radical"),
|
|
Property(UCHAR_SOFT_DOTTED, "SD Soft_Dotted"),
|
|
Property(UCHAR_TERMINAL_PUNCTUATION, "Term Terminal_Punctuation"),
|
|
Property(UCHAR_UNIFIED_IDEOGRAPH, "UIdeo Unified_Ideograph"),
|
|
Property(UCHAR_UPPERCASE, "Upper Uppercase"),
|
|
Property(UCHAR_WHITE_SPACE, "WSpace White_Space space"),
|
|
Property(UCHAR_XID_CONTINUE, "XIDC XID_Continue"),
|
|
Property(UCHAR_XID_START, "XIDS XID_Start"),
|
|
Property(UCHAR_CASE_SENSITIVE, "Sensitive Case_Sensitive"),
|
|
Property(UCHAR_S_TERM, "STerm Sentence_Terminal"),
|
|
Property(UCHAR_VARIATION_SELECTOR, "VS Variation_Selector"),
|
|
Property(UCHAR_NFD_INERT, "nfdinert NFD_Inert"),
|
|
Property(UCHAR_NFKD_INERT, "nfkdinert NFKD_Inert"),
|
|
Property(UCHAR_NFC_INERT, "nfcinert NFC_Inert"),
|
|
Property(UCHAR_NFKC_INERT, "nfkcinert NFKC_Inert"),
|
|
Property(UCHAR_SEGMENT_STARTER, "segstart Segment_Starter"),
|
|
Property(UCHAR_PATTERN_SYNTAX, "Pat_Syn Pattern_Syntax"),
|
|
Property(UCHAR_PATTERN_WHITE_SPACE, "Pat_WS Pattern_White_Space"),
|
|
Property(UCHAR_POSIX_ALNUM, " alnum"),
|
|
Property(UCHAR_POSIX_BLANK, " blank"),
|
|
Property(UCHAR_POSIX_GRAPH, " graph"),
|
|
Property(UCHAR_POSIX_PRINT, " print"),
|
|
Property(UCHAR_POSIX_XDIGIT, " xdigit"),
|
|
Property(UCHAR_CASED, "Cased Cased"),
|
|
Property(UCHAR_CASE_IGNORABLE, "CI Case_Ignorable"),
|
|
Property(UCHAR_CHANGES_WHEN_LOWERCASED, "CWL Changes_When_Lowercased"),
|
|
Property(UCHAR_CHANGES_WHEN_UPPERCASED, "CWU Changes_When_Uppercased"),
|
|
Property(UCHAR_CHANGES_WHEN_TITLECASED, "CWT Changes_When_Titlecased"),
|
|
Property(UCHAR_CHANGES_WHEN_CASEFOLDED, "CWCF Changes_When_Casefolded"),
|
|
Property(UCHAR_CHANGES_WHEN_CASEMAPPED, "CWCM Changes_When_Casemapped"),
|
|
Property(UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED, "CWKCF Changes_When_NFKC_Casefolded"),
|
|
Property(UCHAR_EMOJI, "Emoji Emoji"),
|
|
Property(UCHAR_EMOJI_PRESENTATION, "EPres Emoji_Presentation"),
|
|
Property(UCHAR_EMOJI_MODIFIER, "EMod Emoji_Modifier"),
|
|
Property(UCHAR_EMOJI_MODIFIER_BASE, "EBase Emoji_Modifier_Base"),
|
|
Property(UCHAR_EMOJI_COMPONENT, "EComp Emoji_Component"),
|
|
Property(UCHAR_REGIONAL_INDICATOR, "RI Regional_Indicator"),
|
|
Property(UCHAR_PREPENDED_CONCATENATION_MARK, "PCM Prepended_Concatenation_Mark"),
|
|
Property(UCHAR_EXTENDED_PICTOGRAPHIC, "ExtPict Extended_Pictographic"),
|
|
Property(UCHAR_BASIC_EMOJI, "Basic_Emoji Basic_Emoji"),
|
|
Property(UCHAR_EMOJI_KEYCAP_SEQUENCE, "Emoji_Keycap_Sequence Emoji_Keycap_Sequence"),
|
|
Property(UCHAR_RGI_EMOJI_MODIFIER_SEQUENCE, "RGI_Emoji_Modifier_Sequence RGI_Emoji_Modifier_Sequence"),
|
|
Property(UCHAR_RGI_EMOJI_FLAG_SEQUENCE, "RGI_Emoji_Flag_Sequence RGI_Emoji_Flag_Sequence"),
|
|
Property(UCHAR_RGI_EMOJI_TAG_SEQUENCE, "RGI_Emoji_Tag_Sequence RGI_Emoji_Tag_Sequence"),
|
|
Property(UCHAR_RGI_EMOJI_ZWJ_SEQUENCE, "RGI_Emoji_ZWJ_Sequence RGI_Emoji_ZWJ_Sequence"),
|
|
Property(UCHAR_RGI_EMOJI, "RGI_Emoji RGI_Emoji"),
|
|
Property(UCHAR_IDS_UNARY_OPERATOR, "IDSU IDS_Unary_Operator"),
|
|
Property(UCHAR_ID_COMPAT_MATH_START, "ID_Compat_Math_Start ID_Compat_Math_Start"),
|
|
Property(UCHAR_ID_COMPAT_MATH_CONTINUE, "ID_Compat_Math_Continue ID_Compat_Math_Continue"),
|
|
Property(UCHAR_BIDI_CLASS, "bc Bidi_Class", VALUES_bc, 23),
|
|
Property(UCHAR_BLOCK, "blk Block", VALUES_blk, 339),
|
|
Property(UCHAR_CANONICAL_COMBINING_CLASS, "ccc Canonical_Combining_Class", VALUES_ccc, 58),
|
|
Property(UCHAR_DECOMPOSITION_TYPE, "dt Decomposition_Type", VALUES_dt, 18),
|
|
Property(UCHAR_EAST_ASIAN_WIDTH, "ea East_Asian_Width", VALUES_ea, 6),
|
|
Property(UCHAR_GENERAL_CATEGORY, "gc General_Category", VALUES_gc, 30),
|
|
Property(UCHAR_JOINING_GROUP, "jg Joining_Group", VALUES_jg, 104),
|
|
Property(UCHAR_JOINING_TYPE, "jt Joining_Type", VALUES_jt, 6),
|
|
Property(UCHAR_LINE_BREAK, "lb Line_Break", VALUES_lb, 48),
|
|
Property(UCHAR_NUMERIC_TYPE, "nt Numeric_Type", VALUES_nt, 4),
|
|
Property(UCHAR_SCRIPT, "sc Script", VALUES_sc, 208),
|
|
Property(UCHAR_HANGUL_SYLLABLE_TYPE, "hst Hangul_Syllable_Type", VALUES_hst, 6),
|
|
Property(UCHAR_NFD_QUICK_CHECK, "NFD_QC NFD_Quick_Check", VALUES_NFD_QC, 2),
|
|
Property(UCHAR_NFKD_QUICK_CHECK, "NFKD_QC NFKD_Quick_Check", VALUES_NFKD_QC, 2),
|
|
Property(UCHAR_NFC_QUICK_CHECK, "NFC_QC NFC_Quick_Check", VALUES_NFC_QC, 3),
|
|
Property(UCHAR_NFKC_QUICK_CHECK, "NFKC_QC NFKC_Quick_Check", VALUES_NFKC_QC, 3),
|
|
Property(UCHAR_LEAD_CANONICAL_COMBINING_CLASS, "lccc Lead_Canonical_Combining_Class", VALUES_lccc, 58),
|
|
Property(UCHAR_TRAIL_CANONICAL_COMBINING_CLASS, "tccc Trail_Canonical_Combining_Class", VALUES_tccc, 58),
|
|
Property(UCHAR_GRAPHEME_CLUSTER_BREAK, "GCB Grapheme_Cluster_Break", VALUES_GCB, 18),
|
|
Property(UCHAR_SENTENCE_BREAK, "SB Sentence_Break", VALUES_SB, 15),
|
|
Property(UCHAR_WORD_BREAK, "WB Word_Break", VALUES_WB, 23),
|
|
Property(UCHAR_BIDI_PAIRED_BRACKET_TYPE, "bpt Bidi_Paired_Bracket_Type", VALUES_bpt, 3),
|
|
Property(UCHAR_INDIC_POSITIONAL_CATEGORY, "InPC Indic_Positional_Category", VALUES_InPC, 16),
|
|
Property(UCHAR_INDIC_SYLLABIC_CATEGORY, "InSC Indic_Syllabic_Category", VALUES_InSC, 37),
|
|
Property(UCHAR_VERTICAL_ORIENTATION, "vo Vertical_Orientation", VALUES_vo, 4),
|
|
Property(UCHAR_IDENTIFIER_STATUS, "ID_Status Identifier_Status", VALUES_ID_Status, 2),
|
|
Property(UCHAR_INDIC_CONJUNCT_BREAK, "InCB Indic_Conjunct_Break", VALUES_InCB, 4),
|
|
Property(UCHAR_GENERAL_CATEGORY_MASK, "gcm General_Category_Mask", VALUES_gcm, 38),
|
|
Property(UCHAR_NUMERIC_VALUE, "nv Numeric_Value"),
|
|
Property(UCHAR_AGE, "age Age"),
|
|
Property(UCHAR_BIDI_MIRRORING_GLYPH, "bmg Bidi_Mirroring_Glyph"),
|
|
Property(UCHAR_CASE_FOLDING, "cf Case_Folding"),
|
|
Property(UCHAR_ISO_COMMENT, "isc ISO_Comment"),
|
|
Property(UCHAR_LOWERCASE_MAPPING, "lc Lowercase_Mapping"),
|
|
Property(UCHAR_NAME, "na Name"),
|
|
Property(UCHAR_SIMPLE_CASE_FOLDING, "scf Simple_Case_Folding sfc"),
|
|
Property(UCHAR_SIMPLE_LOWERCASE_MAPPING, "slc Simple_Lowercase_Mapping"),
|
|
Property(UCHAR_SIMPLE_TITLECASE_MAPPING, "stc Simple_Titlecase_Mapping"),
|
|
Property(UCHAR_SIMPLE_UPPERCASE_MAPPING, "suc Simple_Uppercase_Mapping"),
|
|
Property(UCHAR_TITLECASE_MAPPING, "tc Titlecase_Mapping"),
|
|
Property(UCHAR_UNICODE_1_NAME, "na1 Unicode_1_Name"),
|
|
Property(UCHAR_UPPERCASE_MAPPING, "uc Uppercase_Mapping"),
|
|
Property(UCHAR_BIDI_PAIRED_BRACKET, "bpb Bidi_Paired_Bracket"),
|
|
Property(UCHAR_SCRIPT_EXTENSIONS, "scx Script_Extensions"),
|
|
Property(UCHAR_IDENTIFIER_TYPE, "ID_Type Identifier_Type", VALUES_ID_Type, 12),
|
|
};
|
|
|
|
const int32_t MAX_ALIASES = 4;
|