blob: 43326c12d18d1b86c7fc8b01471964751fa7aee5 [file] [log] [blame]
# Override values For Indic_Syllabic_Category
# Not derivable
# Initial version based on Unicode 7.0 by Andrew Glass 2014-03-17
# Updated for Unicode 10.0 by Andrew Glass 2017-07-25
# Updated for Unicode 12.1 by Andrew Glass 2019-05-24
# Updated for Unicode 13.0 by Andrew Glass 2020-07-28
# Updated for Unicode 14.0 by Andrew Glass 2021-09-25
# Updated for Unicode 15.0 by Andrew Glass 2022-09-16
# Updated for Unicode 15.1 by Andrew Glass 2023-09-14
# ================================================
# OVERRIDES TO ASSIGNED VALUES
# ================================================
# Indic_Syllabic_Category=Bindu
193A ; Bindu # Mn LIMBU SIGN KEMPHRENG
AA29 ; Bindu # Mn CHAM VOWEL SIGN AA
10A0D ; Bindu # Mn KHAROSHTHI SIGN DOUBLE RING BELOW
# ================================================
# Indic_Syllabic_Category=Consonant
19C1..19C7 ; Consonant # Lo [7] NEW TAI LUE LETTER FINAL V..NEW TAI LUE LETTER FINAL B # Reassigned to avoid clustering with a base consonant
25CC ; Consonant # So DOTTED CIRCLE #Reassigned to allow it to cluster as a generic base
# ================================================
# Indic_Syllabic_Category=Consonant_Dead
0F7F ; Consonant_Dead # Mc TIBETAN SIGN RNAM BCAD # reassigned so that visarga can form an independent cluster, but see #19
# ================================================
# Indic_Syllabic_Category=Consonant_Final_Modifier
1C36 ; Consonant_Final_Modifier # Mn LEPCHA SIGN RAN
# ================================================
# Indic_Syllabic_Category=Gemination_Mark
11134 ; Gemination_Mark # Mc CHAKMA MAAYYAA
# ================================================
# Indic_Syllabic_Category=Nukta
0F71 ; Nukta # Mn TIBETAN VOWEL SIGN AA # Reassigned to get this before an above vowel, but see #22
1BF2..1BF3 ; Nukta # Mc [2] BATAK PANGOLAT..BATAK PANONGONAN # see USE issue #20
# ================================================
# Indic_Syllabic_Category=Tone_Mark
1A7B..1A7C ; Tone_Mark # Mn [2] TAI THAM SIGN MAI SAM..TAI THAM SIGN KHUEN-LUE KARAN
1A7F ; Tone_Mark # Mn TAI THAM COMBINING CRYPTOGRAMMIC DOT
# ================================================
# Indic_Syllabic_Category=Vowel_Independent
AAB1 ; Vowel_Independent # Lo TAI VIET VOWEL AA
AABA ; Vowel_Independent # Lo TAI VIET VOWEL UA
AABD ; Vowel_Independent # Lo TAI VIET VOWEL AN
# ================================================
# ================================================
# VALUES NOT ASSIGNED IN Indic_Syllabic_Category
# ================================================
# ================================================
# Indic_Syllabic_Category=Consonant
0800..0815 ; Consonant # Lo [22] SAMARITAN LETTER ALAF..SAMARITAN LETTER TAAF
0840..0858 ; Consonant # Lo [25] MANDAIC LETTER HALQA..MANDAIC LETTER AIN
0F00..0F01 ; Consonant # Lo [2] TIBETAN SYLLABLE OM..TIBETAN MARK GTER YIG MGO TRUNCATED
0F04..0F06 ; Consonant # Po TIBETAN MARK INITIAL YIG MGO MDUN MA..TIBETAN MARK CARET YIG MGO PHUR SHAD MA
1800 ; Consonant # Po MONGOLIAN BIRGA # Reassigned so that legacy Birga + MFVS sequences still work
1807 ; Consonant # Po MONGOLIAN SIBE SYLLABLE BOUNDARY MARKER
180A ; Consonant # Po MONGOLIAN NIRUGU
1820..1878 ; Consonant # Lo [88] MONGOLIAN LETTER A..MONGOLIAN LETTER CHA WITH TWO DOTS
1843 ; Consonant # Lm MONGOLIAN LETTER TODO LONG VOWEL SIGN
2D30..2D67 ; Consonant # Lo [56] TIFINAGH LETTER YA..TIFINAGH LETTER YO
2D6F ; Consonant # Lm TIFINAGH MODIFIER LETTER LABIALIZATION MARK
10570..1057A ; Consonant # Lo [11] VITHKUQI CAPITAL LETTER A..VITHKUQI CAPITAL LETTER GA
1057C..1058A ; Consonant # Lo [15] VITHKUQI CAPITAL LETTER HA..VITHKUQI CAPITAL LETTER RE
1058C..10592 ; Consonant # Lo [7] VITHKUQI CAPITAL LETTER SE..VITHKUQI CAPITAL LETTER XE
10594..10595 ; Consonant # Lo [2] VITHKUQI CAPITAL LETTER Y..VITHKUQI CAPITAL LETTER ZE
10597..105A1 ; Consonant # Lo [11] VITHKUQI SMALL LETTER A..VITHKUQI SMALL LETTER GA
105A3..105B1 ; Consonant # Lo [15] VITHKUQI SMALL LETTER HA..VITHKUQI SMALL LETTER RE
105B3..105B9 ; Consonant # Lo [7] VITHKUQI SMALL LETTER SE..VITHKUQI SMALL LETTER XE
105BB..105BC ; Consonant # Lo [2] VITHKUQI SMALL LETTER Y..VITHKUQI SMALL LETTER ZE
10AC0..10AC7 ; Consonant # Lo [8] MANICHAEAN LETTER ALEPH..MANICHAEAN LETTER WAW
10AC9..10AE4 ; Consonant # Lo [28] MANICHAEAN LETTER ZAYIN..MANICHAEAN LETTER TAW
10D00..10D23 ; Consonant # Lo [36] HANIFI ROHINGYA LETTER A..HANIFI ROHINGYA MARK NA KHONNA
10E80..10EA9 ; Consonant # Lo [42] YEZIDI LETTER ELIF..YEZIDI LETTER ET
10EB0..10EB1 ; Consonant # Lo [2] YEZIDI LETTER LAM WITH DOT ABOVE..YEZIDI LETTER YOT WITH CIRCUMFLEX ABOVE
10F30..10F45 ; Consonant # Lo [22] SOGDIAN LETTER ALEPH..SOGDIAN INDEPENDENT SHIN
10F70..10F81 ; Consonant # Lo [18] OLD UYGHUR LETTER ALEPH..OLD UYGHUR LETTER LESH
111DA ; Consonant # Lo SHARADA EKAM
16B00..16B2F ; Consonant # Lo [48] PAHAWH HMONG VOWEL KEEB..PAHAWH HMONG CONSONANT CAU
16F00..16F4A ; Consonant # Lo [75] MIAO LETTER PA..MIAO LETTER RTE
16FE4 ; Consonant # Mn KHITAN SMALL SCRIPT FILLER # Avoids Mn pushing this into VOWEL class
18B00..18CD5 ; Consonant # Lo [470] KHITAN SMALL SCRIPT CHARACTER-18B00..KHITAN SMALL SCRIPT CHARACTER-18CD5
1BC00..1BC6A ; Consonant # Lo [107] DUPLOYAN LETTER H..DUPLOYAN LETTER VOCALIC M
1BC70..1BC7C ; Consonant # Lo [13] DUPLOYAN AFFIX LEFT HORIZONTAL SECANT..DUPLOYAN AFFIX ATTACHED TANGENT HOOK
1BC80..1BC88 ; Consonant # Lo [9] DUPLOYAN AFFIX HIGH ACUTE..DUPLOYAN AFFIX HIGH VERTICAL
1BC90..1BC99 ; Consonant # Lo [10] DUPLOYAN AFFIX LOW ACUTE..DUPLOYAN AFFIX LOW ARROW
1E100..1E12C ; Consonant # Lo [45] NYIAKENG PUACHUE HMONG LETTER MA..NYIAKENG PUACHUE HMONG LETTER W
1E137..1E13D ; Consonant # Lm [7] NYIAKENG PUACHUE HMONG SIGN FOR PERSON..NYIAKENG PUACHUE HMONG SYLLABLE LENGTHENER
1E14E ; Consonant # Lo NYIAKENG PUACHUE HMONG LOGOGRAM NYAJ
1E14F ; Consonant # So NYIAKENG PUACHUE HMONG CIRCLED CA
1E290..1E2AD ; Consonant # Lo [30] TOTO LETTER PA..TOTO LETTER A
1E2C0..1E2EB ; Consonant # Lo [44] WANCHO LETTER AA..WANCHO LETTER YIH
1E4D0..1E4EA ; Consonant # Lo [27] NAG MUNDARI LETTER O..NAG MUNDARI LETTER ELL
1E4EB ; Consonant # Lm NAG MUNDARI SIGN OJOD
1E900..1E921 ; Consonant # Lu [34] ADLAM CAPITAL LETTER ALIF..ADLAM CAPITAL LETTER SHA
1E922..1E943 ; Consonant # Ll [34] ADLAM SMALL LETTER ALIF..ADLAM SMALL LETTER SHA
1E94B ; Consonant # Lm ADLAM NASALIZATION MARK
# ================================================
# Indic_Syllabic_Category=Consonant_Placeholder
1880..1884 ; Consonant_Placeholder # Lo [5] MONGOLIAN LETTER ALI GALI ANUSVARA ONE..MONGOLIAN LETTER ALI GALI INVERTED UBADAMA
# ================================================
# Indic_Syllabic_Category=Gemination_Mark
10D27 ; Gemination_Mark # Mn HANIFI ROHINGYA SIGN TASSI
# ================================================
# Indic_Syllabic_Category=Modifying_Letter
FE00..FE0F ; Modifying_Letter # Mn [16] VARIATION SELECTOR-1..VARIATION SELECTOR-16# Need to treat them as isolated bases so they don't merge with a cluster in invalid scenarios
16F50 ; Modifying_Letter # Lo MIAO LETTER NASALIZATION
# ================================================
# Indic_Syllabic_Category=Nukta
0859..085B ; Nukta # Mn [3] MANDAIC AFFRICATION MARK..MANDAIC GEMINATION MARK
0F39 ; Nukta # Mn TIBETAN MARK TSA -PHRU # NOW IN UNICODE 10.0
1885..1886 ; Nukta # Mn [2] MONGOLIAN LETTER ALI GALI BALUDA..MONGOLIAN LETTER ALI GALI THREE BALUDA
18A9 ; Nukta # Mn MONGOLIAN LETTER ALI GALI DAGALGA
10AE5..10AE6 ; Nukta # Mn [2] MANICHAEAN ABBREVIATION MARK ABOVE..MANICHAEAN ABBREVIATION MARK BELOW
16F4F ; Nukta # Mn MIAO SIGN CONSONANT MODIFIER BAR
1BC9D..1BC9E ; Nukta # Mn [2] DUPLOYAN THICK LETTER SELECTOR..DUPLOYAN DOUBLE MARK
1E944..1E94A ; Nukta # Mn [7] ADLAM ALIF LENGTHENER..ADLAM NUKTA
10F82..10F85 ; Nukta # Mn [4] OLD UYGHUR COMBINING DOT ABOVE..OLD UYGHUR COMBINING TWO DOTS BELOW
# ================================================
# Indic_Syllabic_Category=Number
10D30..10D39 ; Number # Nd [10] HANIFI ROHINGYA DIGIT ZERO..HANIFI ROHINGYA DIGIT NINE
10F51..10F54 ; Number # No [4] SOGDIAN NUMBER ONE..SOGDIAN NUMBER ONE HUNDRED
16AC0..16AC9 ; Number # Nd [10] TANGSA DIGIT ZERO..TANGSA DIGIT NINE
1E140..1E149 ; Number # Nd [10] NYIAKENG PUACHUE HMONG DIGIT ZERO..NYIAKENG PUACHUE HMONG DIGIT NINE
1E2F0..1E2F9 ; Number # Nd [10] WANCHO DIGIT ZERO..WANCHO DIGIT NINE
1E4F0..1E4F9 ; Number # Nd [10] NAG MUNDARI DIGIT ZERO..NAG MUNDARI DIGIT NINE
1E950..1E959 ; Number # Nd [10] ADLAM DIGIT ZERO..ADLAM DIGIT NINE
# ================================================
# Indic_Syllabic_Category=Tone_Mark
07EB..07F3 ; Tone_Mark # Mn [9] NKO COMBINING SHORT HIGH TONE..NKO COMBINING DOUBLE DOT ABOVE
07FD ; Tone_Mark # Mn NKO DANTAYALAN
0F86..0F87 ; Tone_Mark # Mn [2] TIBETAN SIGN LCI RTAGS..TIBETAN SIGN YANG RTAGS
17CF ; Tone_Mark # Mn KHMER SIGN AHSDA
10D24..10D26 ; Tone_Mark # Mn [3] HANIFI ROHINGYA SIGN HARBAHAY..HANIFI ROHINGYA SIGN TANA
10F46..10F50 ; Tone_Mark # Mn [11] SOGDIAN COMBINING DOT BELOW..SOGDIAN COMBINING STROKE BELOW
16B30..16B36 ; Tone_Mark # Mn [7] PAHAWH HMONG MARK CIM TUB..PAHAWH HMONG MARK CIM TAUM
16F8F..16F92 ; Tone_Mark # Mn [4] MIAO TONE RIGHT..MIAO TONE BELOW
1E130..1E136 ; Tone_Mark # Mn [7] NYIAKENG PUACHUE HMONG TONE-B..NYIAKENG PUACHUE HMONG TONE-D
1E2AE ; Tone_Mark # Mn TOTO SIGN RISING TONE
1E2EC..1E2EF ; Tone_Mark # Mn [4] WANCHO TONE TUP..WANCHO TONE KOINI
# ================================================
# Indic_Syllabic_Category=Virama
2D7F ; Virama # Mn TIFINAGH CONSONANT JOINER
# ================================================
# Indic_Syllabic_Category=Vowel_Independent
AAB1 ; Vowel_Independent # Lo TAI VIET VOWEL AA
AABA ; Vowel_Independent # Lo TAI VIET VOWEL UA
AABD ; Vowel_Independent # Lo TAI VIET VOWEL AN
# ================================================
# Indic_Syllabic_Category=Vowel_Dependent
0B55 ; Vowel_Dependent # Mn ORIYA SIGN OVERLINE
10EAB..10EAC ; Vowel_Dependent # Mn [2] YEZIDI COMBINING HAMZA MARK..YEZIDI COMBINING MADDA MARK
16F51..16F87 ; Vowel_Dependent # Mc [55] MIAO SIGN ASPIRATION..MIAO VOWEL SIGN UI
1E4EC..1E4EF ; Vowel_Dependent # Mn [4] NAG MUNDARI SIGN MUHOR..NAG MUNDARI SIGN SUTUH
# ================================================
# Indic_Syllabic_Category=Cantillation_Mark
1CF8..1CF9 ; Cantillation_Mark # Mn [2] VEDIC TONE RING ABOVE..VEDIC TONE DOUBLE RING ABOVE
# ================================================
# Indic_Syllabic_Category=Symbol_Modifier
1B6B..1B73 ; Symbol_Modifier # Mn [9] BALINESE MUSICAL SYMBOL COMBINING TEGEH..BALINESE MUSICAL SYMBOL COMBINING GONG
# ================================================
# ================================================
# PROPERTIES NOT ASSIGNED IN Indic_Syllabic_Category
# ================================================
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph
13000..1342F ; Hieroglyph # Lo [1072] EGYPTIAN HIEROGLYPH A001..EGYPTIAN HIEROGLYPH V011D
1343C..1343F ; Hieroglyph # Cf [4] EGYPTIAN HIEROGLYPH BEGIN ENCLOSURE..END WALLED ENCLOSURE
13441..13446 ; Hieroglyph # Lo [6] EGYPTIAN HIEROGLYPH FULL BLANK..HIEROGLYPH WIDE LOST SIGN
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Joiner
13430..13436 ; Hieroglyph_Joiner # Cf [7] EGYPTIAN HIEROGLYPH VERTICAL JOINER..EGYPTIAN HIEROGLYPH OVERLAY MIDDLE
13439..1343B ; Hieroglyph_Joiner # Cf [3] EGYPTIAN HIEROGLYPH INSERT AT MIDDLE..EGYPTIAN HIEROGLYPH INSERT AT BOTTOM
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Mark_Begin
005B ; Hieroglyph_Mark_Begin # Ps LEFT SQUARE BRACKET
007B ; Hieroglyph_Mark_Begin # Ps LEFT CURLY BRACKET
27E6 ; Hieroglyph_Mark_Begin # Ps MATHEMATICAL LEFT WHITE SQUARE BRACKET
27E8 ; Hieroglyph_Mark_Begin # Ps MATHEMATICAL LEFT ANGLE BRACKET
2E22 ; Hieroglyph_Mark_Begin # Ps TOP LEFT HALF BRACKET
2E24 ; Hieroglyph_Mark_Begin # Ps BOTTOM LEFT HALF BRACKET
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Mark_End
005D ; Hieroglyph_Mark_End # Pe RIGHT SQUARE BRACKET
007D ; Hieroglyph_Mark_End # Pe RIGHT CURLY BRACKET
27E7 ; Hieroglyph_Mark_End # Pe MATHEMATICAL RIGHT WHITE SQUARE BRACKET
27E9 ; Hieroglyph_Mark_End # Pe MATHEMATICAL RIGHT ANGLE BRACKET
2E23 ; Hieroglyph_Mark_End # Pe TOP RIGHT HALF BRACKET
2E25 ; Hieroglyph_Mark_End # Pe BOTTOM RIGHT HALF BRACKET
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Segment_Begin
13437 ; Hieroglyph_Segment_Begin # Cf EGYPTIAN HIEROGLYPH BEGIN SEGMENT
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Segment_End
13438 ; Hieroglyph_Segment_End # Cf EGYPTIAN HIEROGLYPH END SEGMENT
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Mirror
13440 ; Hieroglyph_Mirror # Mn EGYPTIAN HIEROGLYPH MIRROR HORIZONTALLY
# ================================================
# USE, Extended_Syllabic_Category=Hieroglyph_Modifier
13447..13455 ; Hieroglyph_Modifier # Mn [15] EGYPTIAN HIEROGLYPH MODIFIER DAMAGED AT TOP START..EGYPTIAN HIEROGLYPH MODIFIER DAMAGED
# ================================================
# eof