Commit | Line | Data |
---|---|---|
65d7b23c FF |
1 | // Implementation of wcwidth(3) - file is developed at: |
2 | // https://github.com/termux/wcwidth/blob/master/wcwidth.c | |
3 | #include <stdbool.h> | |
4 | #include <stdlib.h> | |
5 | ||
6 | struct width_interval { | |
7 | int start; | |
8 | int end; | |
9 | }; | |
10 | ||
11 | // From https://github.com/jquast/wcwidth/blob/master/wcwidth/table_zero.py | |
12 | // at commit 0d7de112202cc8b2ebe9232ff4a5c954f19d561a (2016-07-02): | |
13 | static struct width_interval ZERO_WIDTH[] = { | |
14 | {0x0300, 0x036f}, // Combining Grave Accent ..Combining Latin Small Le | |
15 | {0x0483, 0x0489}, // Combining Cyrillic Titlo..Combining Cyrillic Milli | |
16 | {0x0591, 0x05bd}, // Hebrew Accent Etnahta ..Hebrew Point Meteg | |
17 | {0x05bf, 0x05bf}, // Hebrew Point Rafe ..Hebrew Point Rafe | |
18 | {0x05c1, 0x05c2}, // Hebrew Point Shin Dot ..Hebrew Point Sin Dot | |
19 | {0x05c4, 0x05c5}, // Hebrew Mark Upper Dot ..Hebrew Mark Lower Dot | |
20 | {0x05c7, 0x05c7}, // Hebrew Point Qamats Qata..Hebrew Point Qamats Qata | |
21 | {0x0610, 0x061a}, // Arabic Sign Sallallahou ..Arabic Small Kasra | |
22 | {0x064b, 0x065f}, // Arabic Fathatan ..Arabic Wavy Hamza Below | |
23 | {0x0670, 0x0670}, // Arabic Letter Superscrip..Arabic Letter Superscrip | |
24 | {0x06d6, 0x06dc}, // Arabic Small High Ligatu..Arabic Small High Seen | |
25 | {0x06df, 0x06e4}, // Arabic Small High Rounde..Arabic Small High Madda | |
26 | {0x06e7, 0x06e8}, // Arabic Small High Yeh ..Arabic Small High Noon | |
27 | {0x06ea, 0x06ed}, // Arabic Empty Centre Low ..Arabic Small Low Meem | |
28 | {0x0711, 0x0711}, // Syriac Letter Superscrip..Syriac Letter Superscrip | |
29 | {0x0730, 0x074a}, // Syriac Pthaha Above ..Syriac Barrekh | |
30 | {0x07a6, 0x07b0}, // Thaana Abafili ..Thaana Sukun | |
31 | {0x07eb, 0x07f3}, // Nko Combining Sh||t High..Nko Combining Double Dot | |
32 | {0x0816, 0x0819}, // Samaritan Mark In ..Samaritan Mark Dagesh | |
33 | {0x081b, 0x0823}, // Samaritan Mark Epentheti..Samaritan Vowel Sign A | |
34 | {0x0825, 0x0827}, // Samaritan Vowel Sign Sho..Samaritan Vowel Sign U | |
35 | {0x0829, 0x082d}, // Samaritan Vowel Sign Lon..Samaritan Mark Nequdaa | |
36 | {0x0859, 0x085b}, // Mandaic Affrication Mark..Mandaic Gemination Mark | |
37 | {0x08d4, 0x08e1}, // (nil) .. | |
38 | {0x08e3, 0x0902}, // Arabic Turned Damma Belo..Devanagari Sign Anusvara | |
39 | {0x093a, 0x093a}, // Devanagari Vowel Sign Oe..Devanagari Vowel Sign Oe | |
40 | {0x093c, 0x093c}, // Devanagari Sign Nukta ..Devanagari Sign Nukta | |
41 | {0x0941, 0x0948}, // Devanagari Vowel Sign U ..Devanagari Vowel Sign Ai | |
42 | {0x094d, 0x094d}, // Devanagari Sign Virama ..Devanagari Sign Virama | |
43 | {0x0951, 0x0957}, // Devanagari Stress Sign U..Devanagari Vowel Sign Uu | |
44 | {0x0962, 0x0963}, // Devanagari Vowel Sign Vo..Devanagari Vowel Sign Vo | |
45 | {0x0981, 0x0981}, // Bengali Sign Candrabindu..Bengali Sign Candrabindu | |
46 | {0x09bc, 0x09bc}, // Bengali Sign Nukta ..Bengali Sign Nukta | |
47 | {0x09c1, 0x09c4}, // Bengali Vowel Sign U ..Bengali Vowel Sign Vocal | |
48 | {0x09cd, 0x09cd}, // Bengali Sign Virama ..Bengali Sign Virama | |
49 | {0x09e2, 0x09e3}, // Bengali Vowel Sign Vocal..Bengali Vowel Sign Vocal | |
50 | {0x0a01, 0x0a02}, // Gurmukhi Sign Adak Bindi..Gurmukhi Sign Bindi | |
51 | {0x0a3c, 0x0a3c}, // Gurmukhi Sign Nukta ..Gurmukhi Sign Nukta | |
52 | {0x0a41, 0x0a42}, // Gurmukhi Vowel Sign U ..Gurmukhi Vowel Sign Uu | |
53 | {0x0a47, 0x0a48}, // Gurmukhi Vowel Sign Ee ..Gurmukhi Vowel Sign Ai | |
54 | {0x0a4b, 0x0a4d}, // Gurmukhi Vowel Sign Oo ..Gurmukhi Sign Virama | |
55 | {0x0a51, 0x0a51}, // Gurmukhi Sign Udaat ..Gurmukhi Sign Udaat | |
56 | {0x0a70, 0x0a71}, // Gurmukhi Tippi ..Gurmukhi Addak | |
57 | {0x0a75, 0x0a75}, // Gurmukhi Sign Yakash ..Gurmukhi Sign Yakash | |
58 | {0x0a81, 0x0a82}, // Gujarati Sign Candrabind..Gujarati Sign Anusvara | |
59 | {0x0abc, 0x0abc}, // Gujarati Sign Nukta ..Gujarati Sign Nukta | |
60 | {0x0ac1, 0x0ac5}, // Gujarati Vowel Sign U ..Gujarati Vowel Sign Cand | |
61 | {0x0ac7, 0x0ac8}, // Gujarati Vowel Sign E ..Gujarati Vowel Sign Ai | |
62 | {0x0acd, 0x0acd}, // Gujarati Sign Virama ..Gujarati Sign Virama | |
63 | {0x0ae2, 0x0ae3}, // Gujarati Vowel Sign Voca..Gujarati Vowel Sign Voca | |
64 | {0x0b01, 0x0b01}, // ||iya Sign Candrabindu ..||iya Sign Candrabindu | |
65 | {0x0b3c, 0x0b3c}, // ||iya Sign Nukta ..||iya Sign Nukta | |
66 | {0x0b3f, 0x0b3f}, // ||iya Vowel Sign I ..||iya Vowel Sign I | |
67 | {0x0b41, 0x0b44}, // ||iya Vowel Sign U ..||iya Vowel Sign Vocalic | |
68 | {0x0b4d, 0x0b4d}, // ||iya Sign Virama ..||iya Sign Virama | |
69 | {0x0b56, 0x0b56}, // ||iya Ai Length Mark ..||iya Ai Length Mark | |
70 | {0x0b62, 0x0b63}, // ||iya Vowel Sign Vocalic..||iya Vowel Sign Vocalic | |
71 | {0x0b82, 0x0b82}, // Tamil Sign Anusvara ..Tamil Sign Anusvara | |
72 | {0x0bc0, 0x0bc0}, // Tamil Vowel Sign Ii ..Tamil Vowel Sign Ii | |
73 | {0x0bcd, 0x0bcd}, // Tamil Sign Virama ..Tamil Sign Virama | |
74 | {0x0c00, 0x0c00}, // Telugu Sign Combining Ca..Telugu Sign Combining Ca | |
75 | {0x0c3e, 0x0c40}, // Telugu Vowel Sign Aa ..Telugu Vowel Sign Ii | |
76 | {0x0c46, 0x0c48}, // Telugu Vowel Sign E ..Telugu Vowel Sign Ai | |
77 | {0x0c4a, 0x0c4d}, // Telugu Vowel Sign O ..Telugu Sign Virama | |
78 | {0x0c55, 0x0c56}, // Telugu Length Mark ..Telugu Ai Length Mark | |
79 | {0x0c62, 0x0c63}, // Telugu Vowel Sign Vocali..Telugu Vowel Sign Vocali | |
80 | {0x0c81, 0x0c81}, // Kannada Sign Candrabindu..Kannada Sign Candrabindu | |
81 | {0x0cbc, 0x0cbc}, // Kannada Sign Nukta ..Kannada Sign Nukta | |
82 | {0x0cbf, 0x0cbf}, // Kannada Vowel Sign I ..Kannada Vowel Sign I | |
83 | {0x0cc6, 0x0cc6}, // Kannada Vowel Sign E ..Kannada Vowel Sign E | |
84 | {0x0ccc, 0x0ccd}, // Kannada Vowel Sign Au ..Kannada Sign Virama | |
85 | {0x0ce2, 0x0ce3}, // Kannada Vowel Sign Vocal..Kannada Vowel Sign Vocal | |
86 | {0x0d01, 0x0d01}, // Malayalam Sign Candrabin..Malayalam Sign Candrabin | |
87 | {0x0d41, 0x0d44}, // Malayalam Vowel Sign U ..Malayalam Vowel Sign Voc | |
88 | {0x0d4d, 0x0d4d}, // Malayalam Sign Virama ..Malayalam Sign Virama | |
89 | {0x0d62, 0x0d63}, // Malayalam Vowel Sign Voc..Malayalam Vowel Sign Voc | |
90 | {0x0dca, 0x0dca}, // Sinhala Sign Al-lakuna ..Sinhala Sign Al-lakuna | |
91 | {0x0dd2, 0x0dd4}, // Sinhala Vowel Sign Ketti..Sinhala Vowel Sign Ketti | |
92 | {0x0dd6, 0x0dd6}, // Sinhala Vowel Sign Diga ..Sinhala Vowel Sign Diga | |
93 | {0x0e31, 0x0e31}, // Thai Character Mai Han-a..Thai Character Mai Han-a | |
94 | {0x0e34, 0x0e3a}, // Thai Character Sara I ..Thai Character Phinthu | |
95 | {0x0e47, 0x0e4e}, // Thai Character Maitaikhu..Thai Character Yamakkan | |
96 | {0x0eb1, 0x0eb1}, // Lao Vowel Sign Mai Kan ..Lao Vowel Sign Mai Kan | |
97 | {0x0eb4, 0x0eb9}, // Lao Vowel Sign I ..Lao Vowel Sign Uu | |
98 | {0x0ebb, 0x0ebc}, // Lao Vowel Sign Mai Kon ..Lao Semivowel Sign Lo | |
99 | {0x0ec8, 0x0ecd}, // Lao Tone Mai Ek ..Lao Niggahita | |
100 | {0x0f18, 0x0f19}, // Tibetan Astrological Sig..Tibetan Astrological Sig | |
101 | {0x0f35, 0x0f35}, // Tibetan Mark Ngas Bzung ..Tibetan Mark Ngas Bzung | |
102 | {0x0f37, 0x0f37}, // Tibetan Mark Ngas Bzung ..Tibetan Mark Ngas Bzung | |
103 | {0x0f39, 0x0f39}, // Tibetan Mark Tsa -phru ..Tibetan Mark Tsa -phru | |
104 | {0x0f71, 0x0f7e}, // Tibetan Vowel Sign Aa ..Tibetan Sign Rjes Su Nga | |
105 | {0x0f80, 0x0f84}, // Tibetan Vowel Sign Rever..Tibetan Mark Halanta | |
106 | {0x0f86, 0x0f87}, // Tibetan Sign Lci Rtags ..Tibetan Sign Yang Rtags | |
107 | {0x0f8d, 0x0f97}, // Tibetan Subjoined Sign L..Tibetan Subjoined Letter | |
108 | {0x0f99, 0x0fbc}, // Tibetan Subjoined Letter..Tibetan Subjoined Letter | |
109 | {0x0fc6, 0x0fc6}, // Tibetan Symbol Padma Gda..Tibetan Symbol Padma Gda | |
110 | {0x102d, 0x1030}, // Myanmar Vowel Sign I ..Myanmar Vowel Sign Uu | |
111 | {0x1032, 0x1037}, // Myanmar Vowel Sign Ai ..Myanmar Sign Dot Below | |
112 | {0x1039, 0x103a}, // Myanmar Sign Virama ..Myanmar Sign Asat | |
113 | {0x103d, 0x103e}, // Myanmar Consonant Sign M..Myanmar Consonant Sign M | |
114 | {0x1058, 0x1059}, // Myanmar Vowel Sign Vocal..Myanmar Vowel Sign Vocal | |
115 | {0x105e, 0x1060}, // Myanmar Consonant Sign M..Myanmar Consonant Sign M | |
116 | {0x1071, 0x1074}, // Myanmar Vowel Sign Geba ..Myanmar Vowel Sign Kayah | |
117 | {0x1082, 0x1082}, // Myanmar Consonant Sign S..Myanmar Consonant Sign S | |
118 | {0x1085, 0x1086}, // Myanmar Vowel Sign Shan ..Myanmar Vowel Sign Shan | |
119 | {0x108d, 0x108d}, // Myanmar Sign Shan Counci..Myanmar Sign Shan Counci | |
120 | {0x109d, 0x109d}, // Myanmar Vowel Sign Aiton..Myanmar Vowel Sign Aiton | |
121 | {0x135d, 0x135f}, // Ethiopic Combining Gemin..Ethiopic Combining Gemin | |
122 | {0x1712, 0x1714}, // Tagalog Vowel Sign I ..Tagalog Sign Virama | |
123 | {0x1732, 0x1734}, // Hanunoo Vowel Sign I ..Hanunoo Sign Pamudpod | |
124 | {0x1752, 0x1753}, // Buhid Vowel Sign I ..Buhid Vowel Sign U | |
125 | {0x1772, 0x1773}, // Tagbanwa Vowel Sign I ..Tagbanwa Vowel Sign U | |
126 | {0x17b4, 0x17b5}, // Khmer Vowel Inherent Aq ..Khmer Vowel Inherent Aa | |
127 | {0x17b7, 0x17bd}, // Khmer Vowel Sign I ..Khmer Vowel Sign Ua | |
128 | {0x17c6, 0x17c6}, // Khmer Sign Nikahit ..Khmer Sign Nikahit | |
129 | {0x17c9, 0x17d3}, // Khmer Sign Muusikatoan ..Khmer Sign Bathamasat | |
130 | {0x17dd, 0x17dd}, // Khmer Sign Atthacan ..Khmer Sign Atthacan | |
131 | {0x180b, 0x180d}, // Mongolian Free Variation..Mongolian Free Variation | |
132 | {0x1885, 0x1886}, // Mongolian Letter Ali Gal..Mongolian Letter Ali Gal | |
133 | {0x18a9, 0x18a9}, // Mongolian Letter Ali Gal..Mongolian Letter Ali Gal | |
134 | {0x1920, 0x1922}, // Limbu Vowel Sign A ..Limbu Vowel Sign U | |
135 | {0x1927, 0x1928}, // Limbu Vowel Sign E ..Limbu Vowel Sign O | |
136 | {0x1932, 0x1932}, // Limbu Small Letter Anusv..Limbu Small Letter Anusv | |
137 | {0x1939, 0x193b}, // Limbu Sign Mukphreng ..Limbu Sign Sa-i | |
138 | {0x1a17, 0x1a18}, // Buginese Vowel Sign I ..Buginese Vowel Sign U | |
139 | {0x1a1b, 0x1a1b}, // Buginese Vowel Sign Ae ..Buginese Vowel Sign Ae | |
140 | {0x1a56, 0x1a56}, // Tai Tham Consonant Sign ..Tai Tham Consonant Sign | |
141 | {0x1a58, 0x1a5e}, // Tai Tham Sign Mai Kang L..Tai Tham Consonant Sign | |
142 | {0x1a60, 0x1a60}, // Tai Tham Sign Sakot ..Tai Tham Sign Sakot | |
143 | {0x1a62, 0x1a62}, // Tai Tham Vowel Sign Mai ..Tai Tham Vowel Sign Mai | |
144 | {0x1a65, 0x1a6c}, // Tai Tham Vowel Sign I ..Tai Tham Vowel Sign Oa B | |
145 | {0x1a73, 0x1a7c}, // Tai Tham Vowel Sign Oa A..Tai Tham Sign Khuen-lue | |
146 | {0x1a7f, 0x1a7f}, // Tai Tham Combining Crypt..Tai Tham Combining Crypt | |
147 | {0x1ab0, 0x1abe}, // Combining Doubled Circum..Combining Parentheses Ov | |
148 | {0x1b00, 0x1b03}, // Balinese Sign Ulu Ricem ..Balinese Sign Surang | |
149 | {0x1b34, 0x1b34}, // Balinese Sign Rerekan ..Balinese Sign Rerekan | |
150 | {0x1b36, 0x1b3a}, // Balinese Vowel Sign Ulu ..Balinese Vowel Sign Ra R | |
151 | {0x1b3c, 0x1b3c}, // Balinese Vowel Sign La L..Balinese Vowel Sign La L | |
152 | {0x1b42, 0x1b42}, // Balinese Vowel Sign Pepe..Balinese Vowel Sign Pepe | |
153 | {0x1b6b, 0x1b73}, // Balinese Musical Symbol ..Balinese Musical Symbol | |
154 | {0x1b80, 0x1b81}, // Sundanese Sign Panyecek ..Sundanese Sign Panglayar | |
155 | {0x1ba2, 0x1ba5}, // Sundanese Consonant Sign..Sundanese Vowel Sign Pan | |
156 | {0x1ba8, 0x1ba9}, // Sundanese Vowel Sign Pam..Sundanese Vowel Sign Pan | |
157 | {0x1bab, 0x1bad}, // Sundanese Sign Virama ..Sundanese Consonant Sign | |
158 | {0x1be6, 0x1be6}, // Batak Sign Tompi ..Batak Sign Tompi | |
159 | {0x1be8, 0x1be9}, // Batak Vowel Sign Pakpak ..Batak Vowel Sign Ee | |
160 | {0x1bed, 0x1bed}, // Batak Vowel Sign Karo O ..Batak Vowel Sign Karo O | |
161 | {0x1bef, 0x1bf1}, // Batak Vowel Sign U F|| S..Batak Consonant Sign H | |
162 | {0x1c2c, 0x1c33}, // Lepcha Vowel Sign E ..Lepcha Consonant Sign T | |
163 | {0x1c36, 0x1c37}, // Lepcha Sign Ran ..Lepcha Sign Nukta | |
164 | {0x1cd0, 0x1cd2}, // Vedic Tone Karshana ..Vedic Tone Prenkha | |
165 | {0x1cd4, 0x1ce0}, // Vedic Sign Yajurvedic Mi..Vedic Tone Rigvedic Kash | |
166 | {0x1ce2, 0x1ce8}, // Vedic Sign Visarga Svari..Vedic Sign Visarga Anuda | |
167 | {0x1ced, 0x1ced}, // Vedic Sign Tiryak ..Vedic Sign Tiryak | |
168 | {0x1cf4, 0x1cf4}, // Vedic Tone Candra Above ..Vedic Tone Candra Above | |
169 | {0x1cf8, 0x1cf9}, // Vedic Tone Ring Above ..Vedic Tone Double Ring A | |
170 | {0x1dc0, 0x1df5}, // Combining Dotted Grave A..Combining Up Tack Above | |
171 | {0x1dfb, 0x1dff}, // (nil) ..Combining Right Arrowhea | |
172 | {0x20d0, 0x20f0}, // Combining Left Harpoon A..Combining Asterisk Above | |
173 | {0x2cef, 0x2cf1}, // Coptic Combining Ni Abov..Coptic Combining Spiritu | |
174 | {0x2d7f, 0x2d7f}, // Tifinagh Consonant Joine..Tifinagh Consonant Joine | |
175 | {0x2de0, 0x2dff}, // Combining Cyrillic Lette..Combining Cyrillic Lette | |
176 | {0x302a, 0x302d}, // Ideographic Level Tone M..Ideographic Entering Ton | |
177 | {0x3099, 0x309a}, // Combining Katakana-hirag..Combining Katakana-hirag | |
178 | {0xa66f, 0xa672}, // Combining Cyrillic Vzmet..Combining Cyrillic Thous | |
179 | {0xa674, 0xa67d}, // Combining Cyrillic Lette..Combining Cyrillic Payer | |
180 | {0xa69e, 0xa69f}, // Combining Cyrillic Lette..Combining Cyrillic Lette | |
181 | {0xa6f0, 0xa6f1}, // Bamum Combining Mark Koq..Bamum Combining Mark Tuk | |
182 | {0xa802, 0xa802}, // Syloti Nagri Sign Dvisva..Syloti Nagri Sign Dvisva | |
183 | {0xa806, 0xa806}, // Syloti Nagri Sign Hasant..Syloti Nagri Sign Hasant | |
184 | {0xa80b, 0xa80b}, // Syloti Nagri Sign Anusva..Syloti Nagri Sign Anusva | |
185 | {0xa825, 0xa826}, // Syloti Nagri Vowel Sign ..Syloti Nagri Vowel Sign | |
186 | {0xa8c4, 0xa8c5}, // Saurashtra Sign Virama .. | |
187 | {0xa8e0, 0xa8f1}, // Combining Devanagari Dig..Combining Devanagari Sig | |
188 | {0xa926, 0xa92d}, // Kayah Li Vowel Ue ..Kayah Li Tone Calya Plop | |
189 | {0xa947, 0xa951}, // Rejang Vowel Sign I ..Rejang Consonant Sign R | |
190 | {0xa980, 0xa982}, // Javanese Sign Panyangga ..Javanese Sign Layar | |
191 | {0xa9b3, 0xa9b3}, // Javanese Sign Cecak Telu..Javanese Sign Cecak Telu | |
192 | {0xa9b6, 0xa9b9}, // Javanese Vowel Sign Wulu..Javanese Vowel Sign Suku | |
193 | {0xa9bc, 0xa9bc}, // Javanese Vowel Sign Pepe..Javanese Vowel Sign Pepe | |
194 | {0xa9e5, 0xa9e5}, // Myanmar Sign Shan Saw ..Myanmar Sign Shan Saw | |
195 | {0xaa29, 0xaa2e}, // Cham Vowel Sign Aa ..Cham Vowel Sign Oe | |
196 | {0xaa31, 0xaa32}, // Cham Vowel Sign Au ..Cham Vowel Sign Ue | |
197 | {0xaa35, 0xaa36}, // Cham Consonant Sign La ..Cham Consonant Sign Wa | |
198 | {0xaa43, 0xaa43}, // Cham Consonant Sign Fina..Cham Consonant Sign Fina | |
199 | {0xaa4c, 0xaa4c}, // Cham Consonant Sign Fina..Cham Consonant Sign Fina | |
200 | {0xaa7c, 0xaa7c}, // Myanmar Sign Tai Laing T..Myanmar Sign Tai Laing T | |
201 | {0xaab0, 0xaab0}, // Tai Viet Mai Kang ..Tai Viet Mai Kang | |
202 | {0xaab2, 0xaab4}, // Tai Viet Vowel I ..Tai Viet Vowel U | |
203 | {0xaab7, 0xaab8}, // Tai Viet Mai Khit ..Tai Viet Vowel Ia | |
204 | {0xaabe, 0xaabf}, // Tai Viet Vowel Am ..Tai Viet Tone Mai Ek | |
205 | {0xaac1, 0xaac1}, // Tai Viet Tone Mai Tho ..Tai Viet Tone Mai Tho | |
206 | {0xaaec, 0xaaed}, // Meetei Mayek Vowel Sign ..Meetei Mayek Vowel Sign | |
207 | {0xaaf6, 0xaaf6}, // Meetei Mayek Virama ..Meetei Mayek Virama | |
208 | {0xabe5, 0xabe5}, // Meetei Mayek Vowel Sign ..Meetei Mayek Vowel Sign | |
209 | {0xabe8, 0xabe8}, // Meetei Mayek Vowel Sign ..Meetei Mayek Vowel Sign | |
210 | {0xabed, 0xabed}, // Meetei Mayek Apun Iyek ..Meetei Mayek Apun Iyek | |
211 | {0xfb1e, 0xfb1e}, // Hebrew Point Judeo-spani..Hebrew Point Judeo-spani | |
212 | {0xfe00, 0xfe0f}, // Variation Select||-1 ..Variation Select||-16 | |
213 | {0xfe20, 0xfe2f}, // Combining Ligature Left ..Combining Cyrillic Titlo | |
214 | {0x101fd, 0x101fd}, // Phaistos Disc Sign Combi..Phaistos Disc Sign Combi | |
215 | {0x102e0, 0x102e0}, // Coptic Epact Thousands M..Coptic Epact Thousands M | |
216 | {0x10376, 0x1037a}, // Combining Old Permic Let..Combining Old Permic Let | |
217 | {0x10a01, 0x10a03}, // Kharoshthi Vowel Sign I ..Kharoshthi Vowel Sign Vo | |
218 | {0x10a05, 0x10a06}, // Kharoshthi Vowel Sign E ..Kharoshthi Vowel Sign O | |
219 | {0x10a0c, 0x10a0f}, // Kharoshthi Vowel Length ..Kharoshthi Sign Visarga | |
220 | {0x10a38, 0x10a3a}, // Kharoshthi Sign Bar Abov..Kharoshthi Sign Dot Belo | |
221 | {0x10a3f, 0x10a3f}, // Kharoshthi Virama ..Kharoshthi Virama | |
222 | {0x10ae5, 0x10ae6}, // Manichaean Abbreviation ..Manichaean Abbreviation | |
223 | {0x11001, 0x11001}, // Brahmi Sign Anusvara ..Brahmi Sign Anusvara | |
224 | {0x11038, 0x11046}, // Brahmi Vowel Sign Aa ..Brahmi Virama | |
225 | {0x1107f, 0x11081}, // Brahmi Number Joiner ..Kaithi Sign Anusvara | |
226 | {0x110b3, 0x110b6}, // Kaithi Vowel Sign U ..Kaithi Vowel Sign Ai | |
227 | {0x110b9, 0x110ba}, // Kaithi Sign Virama ..Kaithi Sign Nukta | |
228 | {0x11100, 0x11102}, // Chakma Sign Candrabindu ..Chakma Sign Visarga | |
229 | {0x11127, 0x1112b}, // Chakma Vowel Sign A ..Chakma Vowel Sign Uu | |
230 | {0x1112d, 0x11134}, // Chakma Vowel Sign Ai ..Chakma Maayyaa | |
231 | {0x11173, 0x11173}, // Mahajani Sign Nukta ..Mahajani Sign Nukta | |
232 | {0x11180, 0x11181}, // Sharada Sign Candrabindu..Sharada Sign Anusvara | |
233 | {0x111b6, 0x111be}, // Sharada Vowel Sign U ..Sharada Vowel Sign O | |
234 | {0x111ca, 0x111cc}, // Sharada Sign Nukta ..Sharada Extra Sh||t Vowe | |
235 | {0x1122f, 0x11231}, // Khojki Vowel Sign U ..Khojki Vowel Sign Ai | |
236 | {0x11234, 0x11234}, // Khojki Sign Anusvara ..Khojki Sign Anusvara | |
237 | {0x11236, 0x11237}, // Khojki Sign Nukta ..Khojki Sign Shadda | |
238 | {0x1123e, 0x1123e}, // (nil) .. | |
239 | {0x112df, 0x112df}, // Khudawadi Sign Anusvara ..Khudawadi Sign Anusvara | |
240 | {0x112e3, 0x112ea}, // Khudawadi Vowel Sign U ..Khudawadi Sign Virama | |
241 | {0x11300, 0x11301}, // Grantha Sign Combining A..Grantha Sign Candrabindu | |
242 | {0x1133c, 0x1133c}, // Grantha Sign Nukta ..Grantha Sign Nukta | |
243 | {0x11340, 0x11340}, // Grantha Vowel Sign Ii ..Grantha Vowel Sign Ii | |
244 | {0x11366, 0x1136c}, // Combining Grantha Digit ..Combining Grantha Digit | |
245 | {0x11370, 0x11374}, // Combining Grantha Letter..Combining Grantha Letter | |
246 | {0x11438, 0x1143f}, // (nil) .. | |
247 | {0x11442, 0x11444}, // (nil) .. | |
248 | {0x11446, 0x11446}, // (nil) .. | |
249 | {0x114b3, 0x114b8}, // Tirhuta Vowel Sign U ..Tirhuta Vowel Sign Vocal | |
250 | {0x114ba, 0x114ba}, // Tirhuta Vowel Sign Sh||t..Tirhuta Vowel Sign Sh||t | |
251 | {0x114bf, 0x114c0}, // Tirhuta Sign Candrabindu..Tirhuta Sign Anusvara | |
252 | {0x114c2, 0x114c3}, // Tirhuta Sign Virama ..Tirhuta Sign Nukta | |
253 | {0x115b2, 0x115b5}, // Siddham Vowel Sign U ..Siddham Vowel Sign Vocal | |
254 | {0x115bc, 0x115bd}, // Siddham Sign Candrabindu..Siddham Sign Anusvara | |
255 | {0x115bf, 0x115c0}, // Siddham Sign Virama ..Siddham Sign Nukta | |
256 | {0x115dc, 0x115dd}, // Siddham Vowel Sign Alter..Siddham Vowel Sign Alter | |
257 | {0x11633, 0x1163a}, // Modi Vowel Sign U ..Modi Vowel Sign Ai | |
258 | {0x1163d, 0x1163d}, // Modi Sign Anusvara ..Modi Sign Anusvara | |
259 | {0x1163f, 0x11640}, // Modi Sign Virama ..Modi Sign Ardhacandra | |
260 | {0x116ab, 0x116ab}, // Takri Sign Anusvara ..Takri Sign Anusvara | |
261 | {0x116ad, 0x116ad}, // Takri Vowel Sign Aa ..Takri Vowel Sign Aa | |
262 | {0x116b0, 0x116b5}, // Takri Vowel Sign U ..Takri Vowel Sign Au | |
263 | {0x116b7, 0x116b7}, // Takri Sign Nukta ..Takri Sign Nukta | |
264 | {0x1171d, 0x1171f}, // Ahom Consonant Sign Medi..Ahom Consonant Sign Medi | |
265 | {0x11722, 0x11725}, // Ahom Vowel Sign I ..Ahom Vowel Sign Uu | |
266 | {0x11727, 0x1172b}, // Ahom Vowel Sign Aw ..Ahom Sign Killer | |
267 | {0x11c30, 0x11c36}, // (nil) .. | |
268 | {0x11c38, 0x11c3d}, // (nil) .. | |
269 | {0x11c3f, 0x11c3f}, // (nil) .. | |
270 | {0x11c92, 0x11ca7}, // (nil) .. | |
271 | {0x11caa, 0x11cb0}, // (nil) .. | |
272 | {0x11cb2, 0x11cb3}, // (nil) .. | |
273 | {0x11cb5, 0x11cb6}, // (nil) .. | |
274 | {0x16af0, 0x16af4}, // Bassa Vah Combining High..Bassa Vah Combining High | |
275 | {0x16b30, 0x16b36}, // Pahawh Hmong Mark Cim Tu..Pahawh Hmong Mark Cim Ta | |
276 | {0x16f8f, 0x16f92}, // Miao Tone Right ..Miao Tone Below | |
277 | {0x1bc9d, 0x1bc9e}, // Duployan Thick Letter Se..Duployan Double Mark | |
278 | {0x1d167, 0x1d169}, // Musical Symbol Combining..Musical Symbol Combining | |
279 | {0x1d17b, 0x1d182}, // Musical Symbol Combining..Musical Symbol Combining | |
280 | {0x1d185, 0x1d18b}, // Musical Symbol Combining..Musical Symbol Combining | |
281 | {0x1d1aa, 0x1d1ad}, // Musical Symbol Combining..Musical Symbol Combining | |
282 | {0x1d242, 0x1d244}, // Combining Greek Musical ..Combining Greek Musical | |
283 | {0x1da00, 0x1da36}, // Signwriting Head Rim ..Signwriting Air Sucking | |
284 | {0x1da3b, 0x1da6c}, // Signwriting Mouth Closed..Signwriting Excitement | |
285 | {0x1da75, 0x1da75}, // Signwriting Upper Body T..Signwriting Upper Body T | |
286 | {0x1da84, 0x1da84}, // Signwriting Location Hea..Signwriting Location Hea | |
287 | {0x1da9b, 0x1da9f}, // Signwriting Fill Modifie..Signwriting Fill Modifie | |
288 | {0x1daa1, 0x1daaf}, // Signwriting Rotation Mod..Signwriting Rotation Mod | |
289 | {0x1e000, 0x1e006}, // (nil) .. | |
290 | {0x1e008, 0x1e018}, // (nil) .. | |
291 | {0x1e01b, 0x1e021}, // (nil) .. | |
292 | {0x1e023, 0x1e024}, // (nil) .. | |
293 | {0x1e026, 0x1e02a}, // (nil) .. | |
294 | {0x1e8d0, 0x1e8d6}, // Mende Kikakui Combining ..Mende Kikakui Combining | |
295 | {0x1e944, 0x1e94a}, // (nil) .. | |
296 | {0xe0100, 0xe01ef}, // Variation Select||-17 ..Variation Select||-256 | |
297 | }; | |
298 | ||
299 | // https://github.com/jquast/wcwidth/blob/master/wcwidth/table_wide.py | |
300 | // at commit 0d7de112202cc8b2ebe9232ff4a5c954f19d561a (2016-07-02): | |
301 | static struct width_interval WIDE_EASTASIAN[] = { | |
302 | {0x1100, 0x115f}, // Hangul Choseong Kiyeok ..Hangul Choseong Filler | |
303 | {0x231a, 0x231b}, // Watch ..Hourglass | |
304 | {0x2329, 0x232a}, // Left-pointing Angle Brac..Right-pointing Angle Bra | |
305 | {0x23e9, 0x23ec}, // Black Right-pointing Dou..Black Down-pointing Doub | |
306 | {0x23f0, 0x23f0}, // Alarm Clock ..Alarm Clock | |
307 | {0x23f3, 0x23f3}, // Hourglass With Flowing S..Hourglass With Flowing S | |
308 | {0x25fd, 0x25fe}, // White Medium Small Squar..Black Medium Small Squar | |
309 | {0x2614, 0x2615}, // Umbrella With Rain Drops..Hot Beverage | |
310 | {0x2648, 0x2653}, // Aries ..Pisces | |
311 | {0x267f, 0x267f}, // Wheelchair Symbol ..Wheelchair Symbol | |
312 | {0x2693, 0x2693}, // Anch|| ..Anch|| | |
313 | {0x26a1, 0x26a1}, // High Voltage Sign ..High Voltage Sign | |
314 | {0x26aa, 0x26ab}, // Medium White Circle ..Medium Black Circle | |
315 | {0x26bd, 0x26be}, // Soccer Ball ..Baseball | |
316 | {0x26c4, 0x26c5}, // Snowman Without Snow ..Sun Behind Cloud | |
317 | {0x26ce, 0x26ce}, // Ophiuchus ..Ophiuchus | |
318 | {0x26d4, 0x26d4}, // No Entry ..No Entry | |
319 | {0x26ea, 0x26ea}, // Church ..Church | |
320 | {0x26f2, 0x26f3}, // Fountain ..Flag In Hole | |
321 | {0x26f5, 0x26f5}, // Sailboat ..Sailboat | |
322 | {0x26fa, 0x26fa}, // Tent ..Tent | |
323 | {0x26fd, 0x26fd}, // Fuel Pump ..Fuel Pump | |
324 | {0x2705, 0x2705}, // White Heavy Check Mark ..White Heavy Check Mark | |
325 | {0x270a, 0x270b}, // Raised Fist ..Raised Hand | |
326 | {0x2728, 0x2728}, // Sparkles ..Sparkles | |
327 | {0x274c, 0x274c}, // Cross Mark ..Cross Mark | |
328 | {0x274e, 0x274e}, // Negative Squared Cross M..Negative Squared Cross M | |
329 | {0x2753, 0x2755}, // Black Question Mark ||na..White Exclamation Mark O | |
330 | {0x2757, 0x2757}, // Heavy Exclamation Mark S..Heavy Exclamation Mark S | |
331 | {0x2795, 0x2797}, // Heavy Plus Sign ..Heavy Division Sign | |
332 | {0x27b0, 0x27b0}, // Curly Loop ..Curly Loop | |
333 | {0x27bf, 0x27bf}, // Double Curly Loop ..Double Curly Loop | |
334 | {0x2b1b, 0x2b1c}, // Black Large Square ..White Large Square | |
335 | {0x2b50, 0x2b50}, // White Medium Star ..White Medium Star | |
336 | {0x2b55, 0x2b55}, // Heavy Large Circle ..Heavy Large Circle | |
337 | {0x2e80, 0x2e99}, // Cjk Radical Repeat ..Cjk Radical Rap | |
338 | {0x2e9b, 0x2ef3}, // Cjk Radical Choke ..Cjk Radical C-simplified | |
339 | {0x2f00, 0x2fd5}, // Kangxi Radical One ..Kangxi Radical Flute | |
340 | {0x2ff0, 0x2ffb}, // Ideographic Description ..Ideographic Description | |
341 | {0x3000, 0x303e}, // Ideographic Space ..Ideographic Variation In | |
342 | {0x3041, 0x3096}, // Hiragana Letter Small A ..Hiragana Letter Small Ke | |
343 | {0x3099, 0x30ff}, // Combining Katakana-hirag..Katakana Digraph Koto | |
344 | {0x3105, 0x312d}, // Bopomofo Letter B ..Bopomofo Letter Ih | |
345 | {0x3131, 0x318e}, // Hangul Letter Kiyeok ..Hangul Letter Araeae | |
346 | {0x3190, 0x31ba}, // Ideographic Annotation L..Bopomofo Letter Zy | |
347 | {0x31c0, 0x31e3}, // Cjk Stroke T ..Cjk Stroke Q | |
348 | {0x31f0, 0x321e}, // Katakana Letter Small Ku..Parenthesized K||ean Cha | |
349 | {0x3220, 0x3247}, // Parenthesized Ideograph ..Circled Ideograph Koto | |
350 | {0x3250, 0x32fe}, // Partnership Sign ..Circled Katakana Wo | |
351 | {0x3300, 0x4dbf}, // Square Apaato .. | |
352 | {0x4e00, 0xa48c}, // Cjk Unified Ideograph-4e..Yi Syllable Yyr | |
353 | {0xa490, 0xa4c6}, // Yi Radical Qot ..Yi Radical Ke | |
354 | {0xa960, 0xa97c}, // Hangul Choseong Tikeut-m..Hangul Choseong Ssangyeo | |
355 | {0xac00, 0xd7a3}, // Hangul Syllable Ga ..Hangul Syllable Hih | |
356 | {0xf900, 0xfaff}, // Cjk Compatibility Ideogr.. | |
357 | {0xfe10, 0xfe19}, // Presentation F||m F|| Ve..Presentation F||m F|| Ve | |
358 | {0xfe30, 0xfe52}, // Presentation F||m F|| Ve..Small Full Stop | |
359 | {0xfe54, 0xfe66}, // Small Semicolon ..Small Equals Sign | |
360 | {0xfe68, 0xfe6b}, // Small Reverse Solidus ..Small Commercial At | |
361 | {0xff01, 0xff60}, // Fullwidth Exclamation Ma..Fullwidth Right White Pa | |
362 | {0xffe0, 0xffe6}, // Fullwidth Cent Sign ..Fullwidth Won Sign | |
363 | {0x16fe0, 0x16fe0}, // (nil) .. | |
364 | {0x17000, 0x187ec}, // (nil) .. | |
365 | {0x18800, 0x18af2}, // (nil) .. | |
366 | {0x1b000, 0x1b001}, // Katakana Letter Archaic ..Hiragana Letter Archaic | |
367 | {0x1f004, 0x1f004}, // Mahjong Tile Red Dragon ..Mahjong Tile Red Dragon | |
368 | {0x1f0cf, 0x1f0cf}, // Playing Card Black Joker..Playing Card Black Joker | |
369 | {0x1f18e, 0x1f18e}, // Negative Squared Ab ..Negative Squared Ab | |
370 | {0x1f191, 0x1f19a}, // Squared Cl ..Squared Vs | |
371 | {0x1f200, 0x1f202}, // Square Hiragana Hoka ..Squared Katakana Sa | |
372 | {0x1f210, 0x1f23b}, // Squared Cjk Unified Ideo.. | |
373 | {0x1f240, 0x1f248}, // T||toise Shell Bracketed..T||toise Shell Bracketed | |
374 | {0x1f250, 0x1f251}, // Circled Ideograph Advant..Circled Ideograph Accept | |
375 | {0x1f300, 0x1f320}, // Cyclone ..Shooting Star | |
376 | {0x1f32d, 0x1f335}, // Hot Dog ..Cactus | |
377 | {0x1f337, 0x1f37c}, // Tulip ..Baby Bottle | |
378 | {0x1f37e, 0x1f393}, // Bottle With Popping C||k..Graduation Cap | |
379 | {0x1f3a0, 0x1f3ca}, // Carousel H||se ..Swimmer | |
380 | {0x1f3cf, 0x1f3d3}, // Cricket Bat And Ball ..Table Tennis Paddle And | |
381 | {0x1f3e0, 0x1f3f0}, // House Building ..European Castle | |
382 | {0x1f3f4, 0x1f3f4}, // Waving Black Flag ..Waving Black Flag | |
383 | {0x1f3f8, 0x1f43e}, // Badminton Racquet And Sh..Paw Prints | |
384 | {0x1f440, 0x1f440}, // Eyes ..Eyes | |
385 | {0x1f442, 0x1f4fc}, // Ear ..Videocassette | |
386 | {0x1f4ff, 0x1f53d}, // Prayer Beads ..Down-pointing Small Red | |
387 | {0x1f54b, 0x1f54e}, // Kaaba ..Men||ah With Nine Branch | |
388 | {0x1f550, 0x1f567}, // Clock Face One Oclock ..Clock Face Twelve-thirty | |
389 | {0x1f57a, 0x1f57a}, // (nil) .. | |
390 | {0x1f595, 0x1f596}, // Reversed Hand With Middl..Raised Hand With Part Be | |
391 | {0x1f5a4, 0x1f5a4}, // (nil) .. | |
392 | {0x1f5fb, 0x1f64f}, // Mount Fuji ..Person With Folded Hands | |
393 | {0x1f680, 0x1f6c5}, // Rocket ..Left Luggage | |
394 | {0x1f6cc, 0x1f6cc}, // Sleeping Accommodation ..Sleeping Accommodation | |
395 | {0x1f6d0, 0x1f6d2}, // Place Of W||ship .. | |
396 | {0x1f6eb, 0x1f6ec}, // Airplane Departure ..Airplane Arriving | |
397 | {0x1f6f4, 0x1f6f6}, // (nil) .. | |
398 | {0x1f910, 0x1f91e}, // Zipper-mouth Face .. | |
399 | {0x1f920, 0x1f927}, // (nil) .. | |
400 | {0x1f930, 0x1f930}, // (nil) .. | |
401 | {0x1f933, 0x1f93e}, // (nil) .. | |
402 | {0x1f940, 0x1f94b}, // (nil) .. | |
403 | {0x1f950, 0x1f95e}, // (nil) .. | |
404 | {0x1f980, 0x1f991}, // Crab .. | |
405 | {0x1f9c0, 0x1f9c0}, // Cheese Wedge ..Cheese Wedge | |
406 | {0x20000, 0x2fffd}, // Cjk Unified Ideograph-20.. | |
407 | {0x30000, 0x3fffd}, // (nil) .. | |
408 | }; | |
409 | ||
410 | static bool intable(struct width_interval* table, int table_length, int c) { | |
411 | // First quick check f|| Latin1 etc. characters. | |
412 | if (c < table[0].start) return false; | |
413 | ||
414 | // Binary search in table. | |
415 | int bot = 0; | |
416 | int top = table_length - 1; | |
417 | while (top >= bot) { | |
418 | int mid = (bot + top) / 2; | |
419 | if (table[mid].end < c) { | |
420 | bot = mid + 1; | |
421 | } else if (table[mid].start > c) { | |
422 | top = mid - 1; | |
423 | } else { | |
424 | return true; | |
425 | } | |
426 | } | |
427 | return false; | |
428 | } | |
429 | ||
430 | int wcwidth(wchar_t ucs) { | |
431 | if (ucs == 0 || | |
432 | ucs == 0x034F || | |
433 | (0x200B <= ucs && ucs <= 0x200F) || | |
434 | ucs == 0x2028 || | |
435 | ucs == 0x2029 || | |
436 | (0x202A <= ucs && ucs <= 0x202E) || | |
437 | (0x2060 <= ucs && ucs <= 0x2063)) { | |
438 | return 0; | |
439 | } | |
440 | ||
441 | // C0/C1 control characters | |
442 | if (ucs < 32 || (0x07F <= ucs && ucs < 0x0A0)) return -1; | |
443 | ||
444 | // combining characters with zero width | |
445 | if (intable(ZERO_WIDTH, sizeof(ZERO_WIDTH)/sizeof(struct width_interval), ucs)) return 0; | |
446 | ||
447 | return intable(WIDE_EASTASIAN, sizeof(WIDE_EASTASIAN)/sizeof(struct width_interval), ucs) ? 2 : 1; | |
448 | } |