[BACK]Return to utf8.c CVS log [TXT][DIR] Up to [local] / src / usr.bin / tmux

Annotation of src/usr.bin/tmux/utf8.c, Revision 1.25

1.25    ! nicm        1: /* $OpenBSD: utf8.c,v 1.24 2015/11/14 12:03:23 nicm Exp $ */
1.1       nicm        2:
                      3: /*
                      4:  * Copyright (c) 2008 Nicholas Marriott <nicm@users.sourceforge.net>
                      5:  *
                      6:  * Permission to use, copy, modify, and distribute this software for any
                      7:  * purpose with or without fee is hereby granted, provided that the above
                      8:  * copyright notice and this permission notice appear in all copies.
                      9:  *
                     10:  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
                     11:  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
                     12:  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
                     13:  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
                     14:  * WHATSOEVER RESULTING FROM LOSS OF MIND, USE, DATA OR PROFITS, WHETHER
                     15:  * IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
                     16:  * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
                     17:  */
                     18:
                     19: #include <sys/types.h>
                     20:
1.11      nicm       21: #include <stdlib.h>
1.1       nicm       22: #include <string.h>
1.9       nicm       23: #include <vis.h>
1.1       nicm       24:
                     25: #include "tmux.h"
                     26:
                     27: struct utf8_width_entry {
                     28:        u_int   first;
                     29:        u_int   last;
                     30:
                     31:        int     width;
                     32:
                     33:        struct utf8_width_entry *left;
                     34:        struct utf8_width_entry *right;
                     35: };
                     36:
1.14      schwarze   37: /* Sorted, then repeatedly split in the middle to balance the tree. */
1.17      nicm       38: static struct utf8_width_entry utf8_width_table[] = {
1.14      schwarze   39:        { 0x00b41, 0x00b44, 0, NULL, NULL },
                     40:        { 0x008e4, 0x00902, 0, NULL, NULL },
                     41:        { 0x006d6, 0x006dd, 0, NULL, NULL },
                     42:        { 0x005c4, 0x005c5, 0, NULL, NULL },
1.1       nicm       43:        { 0x00591, 0x005bd, 0, NULL, NULL },
1.14      schwarze   44:        { 0x00300, 0x0036f, 0, NULL, NULL },
                     45:        { 0x00483, 0x00489, 0, NULL, NULL },
1.1       nicm       46:        { 0x005bf, 0x005bf, 0, NULL, NULL },
1.14      schwarze   47:        { 0x005c1, 0x005c2, 0, NULL, NULL },
                     48:        { 0x00610, 0x0061a, 0, NULL, NULL },
                     49:        { 0x00600, 0x00605, 0, NULL, NULL },
                     50:        { 0x005c7, 0x005c7, 0, NULL, NULL },
                     51:        { 0x0064b, 0x0065f, 0, NULL, NULL },
                     52:        { 0x0061c, 0x0061c, 0, NULL, NULL },
                     53:        { 0x00670, 0x00670, 0, NULL, NULL },
                     54:        { 0x007a6, 0x007b0, 0, NULL, NULL },
1.1       nicm       55:        { 0x006ea, 0x006ed, 0, NULL, NULL },
1.14      schwarze   56:        { 0x006df, 0x006e4, 0, NULL, NULL },
                     57:        { 0x006e7, 0x006e8, 0, NULL, NULL },
                     58:        { 0x00711, 0x00711, 0, NULL, NULL },
                     59:        { 0x0070f, 0x0070f, 0, NULL, NULL },
                     60:        { 0x00730, 0x0074a, 0, NULL, NULL },
                     61:        { 0x0081b, 0x00823, 0, NULL, NULL },
                     62:        { 0x007eb, 0x007f3, 0, NULL, NULL },
                     63:        { 0x00816, 0x00819, 0, NULL, NULL },
                     64:        { 0x00829, 0x0082d, 0, NULL, NULL },
                     65:        { 0x00825, 0x00827, 0, NULL, NULL },
                     66:        { 0x00859, 0x0085b, 0, NULL, NULL },
                     67:        { 0x00a41, 0x00a42, 0, NULL, NULL },
                     68:        { 0x00981, 0x00981, 0, NULL, NULL },
                     69:        { 0x00941, 0x00948, 0, NULL, NULL },
                     70:        { 0x0093a, 0x0093a, 0, NULL, NULL },
                     71:        { 0x0093c, 0x0093c, 0, NULL, NULL },
                     72:        { 0x00951, 0x00957, 0, NULL, NULL },
                     73:        { 0x0094d, 0x0094d, 0, NULL, NULL },
1.1       nicm       74:        { 0x00962, 0x00963, 0, NULL, NULL },
1.14      schwarze   75:        { 0x009e2, 0x009e3, 0, NULL, NULL },
                     76:        { 0x009c1, 0x009c4, 0, NULL, NULL },
                     77:        { 0x009bc, 0x009bc, 0, NULL, NULL },
                     78:        { 0x009cd, 0x009cd, 0, NULL, NULL },
                     79:        { 0x00a01, 0x00a02, 0, NULL, NULL },
                     80:        { 0x00a3c, 0x00a3c, 0, NULL, NULL },
                     81:        { 0x00ac1, 0x00ac5, 0, NULL, NULL },
                     82:        { 0x00a70, 0x00a71, 0, NULL, NULL },
                     83:        { 0x00a4b, 0x00a4d, 0, NULL, NULL },
                     84:        { 0x00a47, 0x00a48, 0, NULL, NULL },
                     85:        { 0x00a51, 0x00a51, 0, NULL, NULL },
                     86:        { 0x00a81, 0x00a82, 0, NULL, NULL },
                     87:        { 0x00a75, 0x00a75, 0, NULL, NULL },
                     88:        { 0x00abc, 0x00abc, 0, NULL, NULL },
                     89:        { 0x00ae2, 0x00ae3, 0, NULL, NULL },
                     90:        { 0x00ac7, 0x00ac8, 0, NULL, NULL },
                     91:        { 0x00acd, 0x00acd, 0, NULL, NULL },
                     92:        { 0x00b3c, 0x00b3c, 0, NULL, NULL },
1.1       nicm       93:        { 0x00b01, 0x00b01, 0, NULL, NULL },
1.14      schwarze   94:        { 0x00b3f, 0x00b3f, 0, NULL, NULL },
                     95:        { 0x03190, 0x031ba, 2, NULL, NULL },
                     96:        { 0x017c9, 0x017d3, 0, NULL, NULL },
                     97:        { 0x00ec8, 0x00ecd, 0, NULL, NULL },
                     98:        { 0x00cc6, 0x00cc6, 0, NULL, NULL },
                     99:        { 0x00c3e, 0x00c40, 0, NULL, NULL },
                    100:        { 0x00b82, 0x00b82, 0, NULL, NULL },
                    101:        { 0x00b56, 0x00b56, 0, NULL, NULL },
                    102:        { 0x00b4d, 0x00b4d, 0, NULL, NULL },
                    103:        { 0x00b62, 0x00b63, 0, NULL, NULL },
                    104:        { 0x00bcd, 0x00bcd, 0, NULL, NULL },
                    105:        { 0x00bc0, 0x00bc0, 0, NULL, NULL },
                    106:        { 0x00c00, 0x00c00, 0, NULL, NULL },
                    107:        { 0x00c62, 0x00c63, 0, NULL, NULL },
1.1       nicm      108:        { 0x00c4a, 0x00c4d, 0, NULL, NULL },
1.14      schwarze  109:        { 0x00c46, 0x00c48, 0, NULL, NULL },
                    110:        { 0x00c55, 0x00c56, 0, NULL, NULL },
                    111:        { 0x00cbc, 0x00cbc, 0, NULL, NULL },
                    112:        { 0x00c81, 0x00c81, 0, NULL, NULL },
1.1       nicm      113:        { 0x00cbf, 0x00cbf, 0, NULL, NULL },
                    114:        { 0x00dd2, 0x00dd4, 0, NULL, NULL },
1.14      schwarze  115:        { 0x00d41, 0x00d44, 0, NULL, NULL },
                    116:        { 0x00ce2, 0x00ce3, 0, NULL, NULL },
                    117:        { 0x00ccc, 0x00ccd, 0, NULL, NULL },
                    118:        { 0x00d01, 0x00d01, 0, NULL, NULL },
                    119:        { 0x00d62, 0x00d63, 0, NULL, NULL },
                    120:        { 0x00d4d, 0x00d4d, 0, NULL, NULL },
                    121:        { 0x00dca, 0x00dca, 0, NULL, NULL },
                    122:        { 0x00e47, 0x00e4e, 0, NULL, NULL },
                    123:        { 0x00e31, 0x00e31, 0, NULL, NULL },
                    124:        { 0x00dd6, 0x00dd6, 0, NULL, NULL },
                    125:        { 0x00e34, 0x00e3a, 0, NULL, NULL },
                    126:        { 0x00eb4, 0x00eb9, 0, NULL, NULL },
                    127:        { 0x00eb1, 0x00eb1, 0, NULL, NULL },
                    128:        { 0x00ebb, 0x00ebc, 0, NULL, NULL },
                    129:        { 0x0105e, 0x01060, 0, NULL, NULL },
                    130:        { 0x00f8d, 0x00f97, 0, NULL, NULL },
                    131:        { 0x00f39, 0x00f39, 0, NULL, NULL },
                    132:        { 0x00f35, 0x00f35, 0, NULL, NULL },
                    133:        { 0x00f18, 0x00f19, 0, NULL, NULL },
                    134:        { 0x00f37, 0x00f37, 0, NULL, NULL },
                    135:        { 0x00f80, 0x00f84, 0, NULL, NULL },
1.1       nicm      136:        { 0x00f71, 0x00f7e, 0, NULL, NULL },
1.14      schwarze  137:        { 0x00f86, 0x00f87, 0, NULL, NULL },
                    138:        { 0x01032, 0x01037, 0, NULL, NULL },
                    139:        { 0x00fc6, 0x00fc6, 0, NULL, NULL },
                    140:        { 0x00f99, 0x00fbc, 0, NULL, NULL },
                    141:        { 0x0102d, 0x01030, 0, NULL, NULL },
                    142:        { 0x0103d, 0x0103e, 0, NULL, NULL },
                    143:        { 0x01039, 0x0103a, 0, NULL, NULL },
                    144:        { 0x01058, 0x01059, 0, NULL, NULL },
                    145:        { 0x0135d, 0x0135f, 0, NULL, NULL },
                    146:        { 0x01085, 0x01086, 0, NULL, NULL },
                    147:        { 0x01071, 0x01074, 0, NULL, NULL },
                    148:        { 0x01082, 0x01082, 0, NULL, NULL },
                    149:        { 0x0109d, 0x0109d, 0, NULL, NULL },
                    150:        { 0x0108d, 0x0108d, 0, NULL, NULL },
                    151:        { 0x01100, 0x011ff, 2, NULL, NULL },
                    152:        { 0x01772, 0x01773, 0, NULL, NULL },
                    153:        { 0x01732, 0x01734, 0, NULL, NULL },
                    154:        { 0x01712, 0x01714, 0, NULL, NULL },
1.1       nicm      155:        { 0x01752, 0x01753, 0, NULL, NULL },
1.14      schwarze  156:        { 0x017b7, 0x017bd, 0, NULL, NULL },
                    157:        { 0x017b4, 0x017b5, 0, NULL, NULL },
                    158:        { 0x017c6, 0x017c6, 0, NULL, NULL },
                    159:        { 0x01c2c, 0x01c33, 0, NULL, NULL },
                    160:        { 0x01a7f, 0x01a7f, 0, NULL, NULL },
                    161:        { 0x01a17, 0x01a18, 0, NULL, NULL },
                    162:        { 0x01920, 0x01922, 0, NULL, NULL },
                    163:        { 0x0180b, 0x0180e, 0, NULL, NULL },
1.1       nicm      164:        { 0x017dd, 0x017dd, 0, NULL, NULL },
1.14      schwarze  165:        { 0x018a9, 0x018a9, 0, NULL, NULL },
                    166:        { 0x01932, 0x01932, 0, NULL, NULL },
1.1       nicm      167:        { 0x01927, 0x01928, 0, NULL, NULL },
                    168:        { 0x01939, 0x0193b, 0, NULL, NULL },
1.14      schwarze  169:        { 0x01a60, 0x01a60, 0, NULL, NULL },
                    170:        { 0x01a56, 0x01a56, 0, NULL, NULL },
                    171:        { 0x01a1b, 0x01a1b, 0, NULL, NULL },
                    172:        { 0x01a58, 0x01a5e, 0, NULL, NULL },
                    173:        { 0x01a65, 0x01a6c, 0, NULL, NULL },
                    174:        { 0x01a62, 0x01a62, 0, NULL, NULL },
                    175:        { 0x01a73, 0x01a7c, 0, NULL, NULL },
                    176:        { 0x01b80, 0x01b81, 0, NULL, NULL },
                    177:        { 0x01b36, 0x01b3a, 0, NULL, NULL },
                    178:        { 0x01b00, 0x01b03, 0, NULL, NULL },
                    179:        { 0x01ab0, 0x01abe, 0, NULL, NULL },
                    180:        { 0x01b34, 0x01b34, 0, NULL, NULL },
                    181:        { 0x01b42, 0x01b42, 0, NULL, NULL },
1.1       nicm      182:        { 0x01b3c, 0x01b3c, 0, NULL, NULL },
1.14      schwarze  183:        { 0x01b6b, 0x01b73, 0, NULL, NULL },
                    184:        { 0x01be6, 0x01be6, 0, NULL, NULL },
                    185:        { 0x01ba8, 0x01ba9, 0, NULL, NULL },
                    186:        { 0x01ba2, 0x01ba5, 0, NULL, NULL },
                    187:        { 0x01bab, 0x01bad, 0, NULL, NULL },
                    188:        { 0x01bed, 0x01bed, 0, NULL, NULL },
                    189:        { 0x01be8, 0x01be9, 0, NULL, NULL },
                    190:        { 0x01bef, 0x01bf1, 0, NULL, NULL },
                    191:        { 0x02329, 0x0232a, 2, NULL, NULL },
                    192:        { 0x01dc0, 0x01df5, 0, NULL, NULL },
                    193:        { 0x01ce2, 0x01ce8, 0, NULL, NULL },
                    194:        { 0x01cd0, 0x01cd2, 0, NULL, NULL },
                    195:        { 0x01c36, 0x01c37, 0, NULL, NULL },
                    196:        { 0x01cd4, 0x01ce0, 0, NULL, NULL },
                    197:        { 0x01cf4, 0x01cf4, 0, NULL, NULL },
                    198:        { 0x01ced, 0x01ced, 0, NULL, NULL },
                    199:        { 0x01cf8, 0x01cf9, 0, NULL, NULL },
                    200:        { 0x02060, 0x02064, 0, NULL, NULL },
                    201:        { 0x0200b, 0x0200f, 0, NULL, NULL },
                    202:        { 0x01dfc, 0x01dff, 0, NULL, NULL },
                    203:        { 0x0202a, 0x0202e, 0, NULL, NULL },
                    204:        { 0x02066, 0x0206f, 0, NULL, NULL },
                    205:        { 0x020d0, 0x020f0, 0, NULL, NULL },
                    206:        { 0x03001, 0x03029, 2, NULL, NULL },
                    207:        { 0x02e80, 0x02e99, 2, NULL, NULL },
                    208:        { 0x02d7f, 0x02d7f, 0, NULL, NULL },
                    209:        { 0x02cef, 0x02cf1, 0, NULL, NULL },
                    210:        { 0x02de0, 0x02dff, 0, NULL, NULL },
                    211:        { 0x02f00, 0x02fd5, 2, NULL, NULL },
                    212:        { 0x02e9b, 0x02ef3, 2, NULL, NULL },
                    213:        { 0x02ff0, 0x02ffb, 2, NULL, NULL },
1.1       nicm      214:        { 0x03099, 0x0309a, 0, NULL, NULL },
1.14      schwarze  215:        { 0x0302e, 0x0303e, 2, NULL, NULL },
                    216:        { 0x0302a, 0x0302d, 0, NULL, NULL },
                    217:        { 0x03041, 0x03096, 2, NULL, NULL },
                    218:        { 0x03105, 0x0312d, 2, NULL, NULL },
                    219:        { 0x0309b, 0x030ff, 2, NULL, NULL },
                    220:        { 0x03131, 0x0318e, 2, NULL, NULL },
1.1       nicm      221:        { 0x10a3f, 0x10a3f, 0, NULL, NULL },
1.14      schwarze  222:        { 0x0aa4c, 0x0aa4c, 0, NULL, NULL },
1.1       nicm      223:        { 0x0a825, 0x0a826, 0, NULL, NULL },
1.14      schwarze  224:        { 0x0a490, 0x0a4c6, 2, NULL, NULL },
                    225:        { 0x03250, 0x032fe, 2, NULL, NULL },
                    226:        { 0x031f0, 0x0321e, 2, NULL, NULL },
                    227:        { 0x031c0, 0x031e3, 2, NULL, NULL },
                    228:        { 0x03220, 0x03247, 2, NULL, NULL },
                    229:        { 0x04e00, 0x09fcc, 2, NULL, NULL },
                    230:        { 0x03300, 0x04db5, 2, NULL, NULL },
                    231:        { 0x0a000, 0x0a48c, 2, NULL, NULL },
                    232:        { 0x0a6f0, 0x0a6f1, 0, NULL, NULL },
                    233:        { 0x0a674, 0x0a67d, 0, NULL, NULL },
                    234:        { 0x0a66f, 0x0a672, 0, NULL, NULL },
                    235:        { 0x0a69f, 0x0a69f, 0, NULL, NULL },
                    236:        { 0x0a806, 0x0a806, 0, NULL, NULL },
                    237:        { 0x0a802, 0x0a802, 0, NULL, NULL },
                    238:        { 0x0a80b, 0x0a80b, 0, NULL, NULL },
                    239:        { 0x0a9b6, 0x0a9b9, 0, NULL, NULL },
                    240:        { 0x0a947, 0x0a951, 0, NULL, NULL },
                    241:        { 0x0a8e0, 0x0a8f1, 0, NULL, NULL },
                    242:        { 0x0a8c4, 0x0a8c4, 0, NULL, NULL },
                    243:        { 0x0a926, 0x0a92d, 0, NULL, NULL },
                    244:        { 0x0a980, 0x0a982, 0, NULL, NULL },
                    245:        { 0x0a960, 0x0a97c, 2, NULL, NULL },
                    246:        { 0x0a9b3, 0x0a9b3, 0, NULL, NULL },
                    247:        { 0x0aa29, 0x0aa2e, 0, NULL, NULL },
                    248:        { 0x0a9bc, 0x0a9bc, 0, NULL, NULL },
                    249:        { 0x0a9e5, 0x0a9e5, 0, NULL, NULL },
                    250:        { 0x0aa35, 0x0aa36, 0, NULL, NULL },
                    251:        { 0x0aa31, 0x0aa32, 0, NULL, NULL },
                    252:        { 0x0aa43, 0x0aa43, 0, NULL, NULL },
                    253:        { 0x0fb1e, 0x0fb1e, 0, NULL, NULL },
                    254:        { 0x0aaf6, 0x0aaf6, 0, NULL, NULL },
                    255:        { 0x0aab7, 0x0aab8, 0, NULL, NULL },
                    256:        { 0x0aab0, 0x0aab0, 0, NULL, NULL },
                    257:        { 0x0aa7c, 0x0aa7c, 0, NULL, NULL },
                    258:        { 0x0aab2, 0x0aab4, 0, NULL, NULL },
                    259:        { 0x0aac1, 0x0aac1, 0, NULL, NULL },
                    260:        { 0x0aabe, 0x0aabf, 0, NULL, NULL },
                    261:        { 0x0aaec, 0x0aaed, 0, NULL, NULL },
                    262:        { 0x0ac00, 0x0d7a3, 2, NULL, NULL },
                    263:        { 0x0abe8, 0x0abe8, 0, NULL, NULL },
                    264:        { 0x0abe5, 0x0abe5, 0, NULL, NULL },
                    265:        { 0x0abed, 0x0abed, 0, NULL, NULL },
                    266:        { 0x0f900, 0x0fa6d, 2, NULL, NULL },
1.24      nicm      267:        { 0x0d800, 0x0dfff, 0, NULL, NULL },
1.14      schwarze  268:        { 0x0fa70, 0x0fad9, 2, NULL, NULL },
                    269:        { 0x0fff9, 0x0fffb, 0, NULL, NULL },
                    270:        { 0x0fe30, 0x0fe52, 2, NULL, NULL },
1.1       nicm      271:        { 0x0fe10, 0x0fe19, 2, NULL, NULL },
1.14      schwarze  272:        { 0x0fe00, 0x0fe0f, 0, NULL, NULL },
                    273:        { 0x0fe20, 0x0fe2d, 0, NULL, NULL },
                    274:        { 0x0fe68, 0x0fe6b, 2, NULL, NULL },
                    275:        { 0x0fe54, 0x0fe66, 2, NULL, NULL },
                    276:        { 0x0feff, 0x0feff, 0, NULL, NULL },
                    277:        { 0x10a01, 0x10a03, 0, NULL, NULL },
                    278:        { 0x102e0, 0x102e0, 0, NULL, NULL },
                    279:        { 0x101fd, 0x101fd, 0, NULL, NULL },
                    280:        { 0x10376, 0x1037a, 0, NULL, NULL },
                    281:        { 0x10a0c, 0x10a0f, 0, NULL, NULL },
1.1       nicm      282:        { 0x10a05, 0x10a06, 0, NULL, NULL },
1.14      schwarze  283:        { 0x10a38, 0x10a3a, 0, NULL, NULL },
                    284:        { 0x11633, 0x1163a, 0, NULL, NULL },
                    285:        { 0x11236, 0x11237, 0, NULL, NULL },
                    286:        { 0x11100, 0x11102, 0, NULL, NULL },
                    287:        { 0x1107f, 0x11081, 0, NULL, NULL },
                    288:        { 0x11001, 0x11001, 0, NULL, NULL },
                    289:        { 0x10ae5, 0x10ae6, 0, NULL, NULL },
                    290:        { 0x11038, 0x11046, 0, NULL, NULL },
                    291:        { 0x110b9, 0x110ba, 0, NULL, NULL },
                    292:        { 0x110b3, 0x110b6, 0, NULL, NULL },
                    293:        { 0x110bd, 0x110bd, 0, NULL, NULL },
                    294:        { 0x11180, 0x11181, 0, NULL, NULL },
                    295:        { 0x1112d, 0x11134, 0, NULL, NULL },
                    296:        { 0x11127, 0x1112b, 0, NULL, NULL },
                    297:        { 0x11173, 0x11173, 0, NULL, NULL },
                    298:        { 0x1122f, 0x11231, 0, NULL, NULL },
                    299:        { 0x111b6, 0x111be, 0, NULL, NULL },
                    300:        { 0x11234, 0x11234, 0, NULL, NULL },
                    301:        { 0x11370, 0x11374, 0, NULL, NULL },
                    302:        { 0x11301, 0x11301, 0, NULL, NULL },
                    303:        { 0x112df, 0x112df, 0, NULL, NULL },
                    304:        { 0x112e3, 0x112ea, 0, NULL, NULL },
                    305:        { 0x11340, 0x11340, 0, NULL, NULL },
                    306:        { 0x1133c, 0x1133c, 0, NULL, NULL },
                    307:        { 0x11366, 0x1136c, 0, NULL, NULL },
                    308:        { 0x114c2, 0x114c3, 0, NULL, NULL },
                    309:        { 0x114ba, 0x114ba, 0, NULL, NULL },
                    310:        { 0x114b3, 0x114b8, 0, NULL, NULL },
                    311:        { 0x114bf, 0x114c0, 0, NULL, NULL },
                    312:        { 0x115bc, 0x115bd, 0, NULL, NULL },
                    313:        { 0x115b2, 0x115b5, 0, NULL, NULL },
                    314:        { 0x115bf, 0x115c0, 0, NULL, NULL },
                    315:        { 0x1d1aa, 0x1d1ad, 0, NULL, NULL },
                    316:        { 0x16b30, 0x16b36, 0, NULL, NULL },
                    317:        { 0x116ad, 0x116ad, 0, NULL, NULL },
                    318:        { 0x1163f, 0x11640, 0, NULL, NULL },
                    319:        { 0x1163d, 0x1163d, 0, NULL, NULL },
                    320:        { 0x116ab, 0x116ab, 0, NULL, NULL },
                    321:        { 0x116b7, 0x116b7, 0, NULL, NULL },
                    322:        { 0x116b0, 0x116b5, 0, NULL, NULL },
                    323:        { 0x16af0, 0x16af4, 0, NULL, NULL },
                    324:        { 0x1bca0, 0x1bca3, 0, NULL, NULL },
                    325:        { 0x1b000, 0x1b001, 2, NULL, NULL },
                    326:        { 0x16f8f, 0x16f92, 0, NULL, NULL },
                    327:        { 0x1bc9d, 0x1bc9e, 0, NULL, NULL },
                    328:        { 0x1d173, 0x1d182, 0, NULL, NULL },
                    329:        { 0x1d167, 0x1d169, 0, NULL, NULL },
                    330:        { 0x1d185, 0x1d18b, 0, NULL, NULL },
                    331:        { 0x2a700, 0x2b734, 2, NULL, NULL },
                    332:        { 0x1f210, 0x1f23a, 2, NULL, NULL },
                    333:        { 0x1e8d0, 0x1e8d6, 0, NULL, NULL },
                    334:        { 0x1d242, 0x1d244, 0, NULL, NULL },
                    335:        { 0x1f200, 0x1f202, 2, NULL, NULL },
                    336:        { 0x1f250, 0x1f251, 2, NULL, NULL },
                    337:        { 0x1f240, 0x1f248, 2, NULL, NULL },
                    338:        { 0x20000, 0x2a6d6, 2, NULL, NULL },
                    339:        { 0xe0020, 0xe007f, 0, NULL, NULL },
                    340:        { 0x2f800, 0x2fa1d, 2, NULL, NULL },
                    341:        { 0x2b740, 0x2b81d, 2, NULL, NULL },
                    342:        { 0xe0001, 0xe0001, 0, NULL, NULL },
                    343:        { 0xf0000, 0xffffd, 0, NULL, NULL },
                    344:        { 0xe0100, 0xe01ef, 0, NULL, NULL },
                    345:        { 0x100000, 0x10fffd, 0, NULL, NULL },
1.1       nicm      346: };
1.17      nicm      347: static struct utf8_width_entry *utf8_width_root = NULL;
1.1       nicm      348:
1.17      nicm      349: static void    utf8_build(void);
1.1       nicm      350:
1.11      nicm      351: /* Set a single character. */
                    352: void
1.19      nicm      353: utf8_set(struct utf8_data *ud, u_char ch)
1.11      nicm      354: {
1.20      nicm      355:        u_int   i;
                    356:
1.19      nicm      357:        *ud->data = ch;
1.25    ! nicm      358:        ud->have = 1;
1.19      nicm      359:        ud->size = 1;
1.11      nicm      360:
1.19      nicm      361:        ud->width = 1;
1.20      nicm      362:
                    363:        for (i = ud->size; i < sizeof ud->data; i++)
                    364:                ud->data[i] = '\0';
                    365: }
                    366:
                    367: /* Copy UTF-8 character. */
                    368: void
                    369: utf8_copy(struct utf8_data *to, const struct utf8_data *from)
                    370: {
                    371:        u_int   i;
                    372:
                    373:        memcpy(to, from, sizeof *to);
                    374:
                    375:        for (i = to->size; i < sizeof to->data; i++)
                    376:                to->data[i] = '\0';
1.11      nicm      377: }
                    378:
1.4       nicm      379: /*
                    380:  * Open UTF-8 sequence.
                    381:  *
                    382:  * 11000010-11011111 C2-DF start of 2-byte sequence
                    383:  * 11100000-11101111 E0-EF start of 3-byte sequence
                    384:  * 11110000-11110100 F0-F4 start of 4-byte sequence
                    385:  */
1.23      nicm      386: enum utf8_state
1.19      nicm      387: utf8_open(struct utf8_data *ud, u_char ch)
1.4       nicm      388: {
1.19      nicm      389:        memset(ud, 0, sizeof *ud);
1.4       nicm      390:        if (ch >= 0xc2 && ch <= 0xdf)
1.19      nicm      391:                ud->size = 2;
1.4       nicm      392:        else if (ch >= 0xe0 && ch <= 0xef)
1.19      nicm      393:                ud->size = 3;
1.4       nicm      394:        else if (ch >= 0xf0 && ch <= 0xf4)
1.19      nicm      395:                ud->size = 4;
1.4       nicm      396:        else
1.23      nicm      397:                return (UTF8_ERROR);
1.19      nicm      398:        utf8_append(ud, ch);
1.23      nicm      399:        return (UTF8_MORE);
1.4       nicm      400: }
                    401:
1.23      nicm      402: /* Append character to UTF-8, closing if finished. */
                    403: enum utf8_state
1.19      nicm      404: utf8_append(struct utf8_data *ud, u_char ch)
1.4       nicm      405: {
1.19      nicm      406:        if (ud->have >= ud->size)
1.4       nicm      407:                fatalx("UTF-8 character overflow");
1.19      nicm      408:        if (ud->size > sizeof ud->data)
1.4       nicm      409:                fatalx("UTF-8 character size too large");
                    410:
1.21      nicm      411:        if (ud->have != 0 && (ch & 0xc0) != 0x80)
                    412:                ud->width = 0xff;
                    413:
1.19      nicm      414:        ud->data[ud->have++] = ch;
                    415:        if (ud->have != ud->size)
1.23      nicm      416:                return (UTF8_MORE);
1.4       nicm      417:
1.21      nicm      418:        if (ud->width == 0xff)
1.23      nicm      419:                return (UTF8_ERROR);
1.19      nicm      420:        ud->width = utf8_width(utf8_combine(ud));
1.23      nicm      421:        return (UTF8_DONE);
1.1       nicm      422: }
                    423:
1.4       nicm      424: /* Build UTF-8 width tree. */
1.17      nicm      425: static void
1.1       nicm      426: utf8_build(void)
                    427: {
                    428:        struct utf8_width_entry **ptr, *item, *node;
1.17      nicm      429:        u_int                     i;
1.1       nicm      430:
                    431:        for (i = 0; i < nitems(utf8_width_table); i++) {
                    432:                item = &utf8_width_table[i];
                    433:
                    434:                ptr = &utf8_width_root;
                    435:                while (*ptr != NULL) {
                    436:                        node = *ptr;
                    437:                        if (item->last < node->first)
1.13      nicm      438:                                ptr = &node->left;
1.1       nicm      439:                        else if (item->first > node->last)
1.13      nicm      440:                                ptr = &node->right;
1.1       nicm      441:                }
                    442:                *ptr = item;
                    443:        }
                    444: }
                    445:
1.17      nicm      446: /* Lookup width of UTF-8 data in tree. */
                    447: u_int
                    448: utf8_width(u_int uc)
                    449: {
                    450:        struct utf8_width_entry *item;
                    451:
                    452:        if (utf8_width_root == NULL)
                    453:                utf8_build();
                    454:
                    455:        item = utf8_width_root;
                    456:        while (item != NULL) {
                    457:                if (uc < item->first)
                    458:                        item = item->left;
                    459:                else if (uc > item->last)
                    460:                        item = item->right;
                    461:                else
                    462:                        return (item->width);
                    463:        }
                    464:        return (1);
                    465: }
                    466:
1.4       nicm      467: /* Combine UTF-8 into 32-bit Unicode. */
1.1       nicm      468: u_int
1.19      nicm      469: utf8_combine(const struct utf8_data *ud)
1.1       nicm      470: {
1.22      nicm      471:        u_int   uc;
1.1       nicm      472:
1.22      nicm      473:        uc = 0xfffd;
1.19      nicm      474:        switch (ud->size) {
1.4       nicm      475:        case 1:
1.22      nicm      476:                uc = ud->data[0];
1.4       nicm      477:                break;
                    478:        case 2:
1.22      nicm      479:                uc = ud->data[1] & 0x3f;
                    480:                uc |= (ud->data[0] & 0x1f) << 6;
1.4       nicm      481:                break;
                    482:        case 3:
1.22      nicm      483:                uc = ud->data[2] & 0x3f;
                    484:                uc |= (ud->data[1] & 0x3f) << 6;
                    485:                uc |= (ud->data[0] & 0xf) << 12;
1.4       nicm      486:                break;
                    487:        case 4:
1.22      nicm      488:                uc = ud->data[3] & 0x3f;
                    489:                uc |= (ud->data[2] & 0x3f) << 6;
                    490:                uc |= (ud->data[1] & 0x3f) << 12;
                    491:                uc |= (ud->data[0] & 0x7) << 18;
1.4       nicm      492:                break;
1.1       nicm      493:        }
1.22      nicm      494:        return (uc);
1.15      nicm      495: }
                    496:
1.17      nicm      497: /* Split 32-bit Unicode into UTF-8. */
1.23      nicm      498: enum utf8_state
1.19      nicm      499: utf8_split(u_int uc, struct utf8_data *ud)
1.15      nicm      500: {
                    501:        if (uc < 0x7f) {
1.19      nicm      502:                ud->size = 1;
                    503:                ud->data[0] = uc;
1.15      nicm      504:        } else if (uc < 0x7ff) {
1.19      nicm      505:                ud->size = 2;
                    506:                ud->data[0] = 0xc0 | ((uc >> 6) & 0x1f);
                    507:                ud->data[1] = 0x80 | (uc & 0x3f);
1.15      nicm      508:        } else if (uc < 0xffff) {
1.19      nicm      509:                ud->size = 3;
                    510:                ud->data[0] = 0xe0 | ((uc >> 12) & 0xf);
                    511:                ud->data[1] = 0x80 | ((uc >> 6) & 0x3f);
                    512:                ud->data[2] = 0x80 | (uc & 0x3f);
1.15      nicm      513:        } else if (uc < 0x1fffff) {
1.19      nicm      514:                ud->size = 4;
                    515:                ud->data[0] = 0xf0 | ((uc >> 18) & 0x7);
                    516:                ud->data[1] = 0x80 | ((uc >> 12) & 0x3f);
                    517:                ud->data[2] = 0x80 | ((uc >> 6) & 0x3f);
                    518:                ud->data[3] = 0x80 | (uc & 0x3f);
1.15      nicm      519:        } else
1.23      nicm      520:                return (UTF8_ERROR);
1.19      nicm      521:        ud->width = utf8_width(uc);
1.23      nicm      522:        return (UTF8_DONE);
1.6       nicm      523: }
                    524:
                    525: /* Split a two-byte UTF-8 character. */
                    526: u_int
                    527: utf8_split2(u_int uc, u_char *ptr)
                    528: {
                    529:        if (uc > 0x7f) {
                    530:                ptr[0] = (uc >> 6) | 0xc0;
                    531:                ptr[1] = (uc & 0x3f) | 0x80;
                    532:                return (2);
                    533:        }
                    534:        ptr[0] = uc;
1.1       nicm      535:        return (1);
1.9       nicm      536: }
                    537:
                    538: /*
                    539:  * Encode len characters from src into dst, which is guaranteed to have four
                    540:  * bytes available for each character from src (for \abc or UTF-8) plus space
                    541:  * for \0.
                    542:  */
                    543: int
                    544: utf8_strvis(char *dst, const char *src, size_t len, int flag)
                    545: {
1.19      nicm      546:        struct utf8_data         ud;
1.9       nicm      547:        const char              *start, *end;
1.23      nicm      548:        enum utf8_state          more;
1.9       nicm      549:        size_t                   i;
                    550:
                    551:        start = dst;
                    552:        end = src + len;
                    553:
                    554:        while (src < end) {
1.23      nicm      555:                if ((more = utf8_open(&ud, *src)) == UTF8_MORE) {
                    556:                        while (++src < end && more == UTF8_MORE)
1.19      nicm      557:                                more = utf8_append(&ud, *src);
1.23      nicm      558:                        if (more == UTF8_DONE) {
1.9       nicm      559:                                /* UTF-8 character finished. */
1.19      nicm      560:                                for (i = 0; i < ud.size; i++)
                    561:                                        *dst++ = ud.data[i];
1.9       nicm      562:                                continue;
                    563:                        }
1.23      nicm      564:                        /* Not a complete, valid UTF-8 character. */
                    565:                        src -= ud.have;
1.9       nicm      566:                }
                    567:                if (src < end - 1)
                    568:                        dst = vis(dst, src[0], flag, src[1]);
                    569:                else if (src < end)
                    570:                        dst = vis(dst, src[0], flag, '\0');
                    571:                src++;
                    572:        }
                    573:
                    574:        *dst = '\0';
                    575:        return (dst - start);
1.16      nicm      576: }
                    577:
                    578: /*
                    579:  * Sanitize a string, changing any UTF-8 characters to '_'. Caller should free
                    580:  * the returned string. Anything not valid printable ASCII or UTF-8 is
                    581:  * stripped.
                    582:  */
                    583: char *
                    584: utf8_sanitize(const char *src)
                    585: {
                    586:        char                    *dst;
                    587:        size_t                   n;
1.23      nicm      588:        enum utf8_state          more;
1.19      nicm      589:        struct utf8_data         ud;
1.16      nicm      590:        u_int                    i;
                    591:
                    592:        dst = NULL;
                    593:
                    594:        n = 0;
                    595:        while (*src != '\0') {
                    596:                dst = xreallocarray(dst, n + 1, sizeof *dst);
1.23      nicm      597:                if ((more = utf8_open(&ud, *src)) == UTF8_MORE) {
                    598:                        while (*++src != '\0' && more == UTF8_MORE)
1.19      nicm      599:                                more = utf8_append(&ud, *src);
1.23      nicm      600:                        if (more == UTF8_DONE) {
1.19      nicm      601:                                dst = xreallocarray(dst, n + ud.width,
1.16      nicm      602:                                    sizeof *dst);
1.19      nicm      603:                                for (i = 0; i < ud.width; i++)
1.16      nicm      604:                                        dst[n++] = '_';
                    605:                                continue;
                    606:                        }
1.19      nicm      607:                        src -= ud.have;
1.16      nicm      608:                }
                    609:                if (*src > 0x1f && *src < 0x7f)
1.21      nicm      610:                        dst[n++] = *src;
1.23      nicm      611:                else
                    612:                        dst[n++] = '_';
1.16      nicm      613:                src++;
                    614:        }
                    615:
                    616:        dst = xreallocarray(dst, n + 1, sizeof *dst);
                    617:        dst[n] = '\0';
                    618:        return (dst);
1.11      nicm      619: }
                    620:
                    621: /*
                    622:  * Convert a string into a buffer of UTF-8 characters. Terminated by size == 0.
                    623:  * Caller frees.
                    624:  */
                    625: struct utf8_data *
                    626: utf8_fromcstr(const char *src)
                    627: {
                    628:        struct utf8_data        *dst;
                    629:        size_t                   n;
1.23      nicm      630:        enum utf8_state          more;
1.11      nicm      631:
                    632:        dst = NULL;
                    633:
                    634:        n = 0;
                    635:        while (*src != '\0') {
1.12      nicm      636:                dst = xreallocarray(dst, n + 1, sizeof *dst);
1.23      nicm      637:                if ((more = utf8_open(&dst[n], *src)) == UTF8_MORE) {
                    638:                        while (*++src != '\0' && more == UTF8_MORE)
1.11      nicm      639:                                more = utf8_append(&dst[n], *src);
1.23      nicm      640:                        if (more == UTF8_DONE) {
1.11      nicm      641:                                n++;
                    642:                                continue;
                    643:                        }
                    644:                        src -= dst[n].have;
                    645:                }
1.23      nicm      646:                utf8_set(&dst[n], *src);
                    647:                n++;
1.11      nicm      648:                src++;
                    649:        }
                    650:
1.12      nicm      651:        dst = xreallocarray(dst, n + 1, sizeof *dst);
1.11      nicm      652:        dst[n].size = 0;
                    653:        return (dst);
                    654: }
                    655:
                    656: /* Convert from a buffer of UTF-8 characters into a string. Caller frees. */
                    657: char *
                    658: utf8_tocstr(struct utf8_data *src)
                    659: {
                    660:        char    *dst;
                    661:        size_t   n;
                    662:
                    663:        dst = NULL;
                    664:
                    665:        n = 0;
                    666:        for(; src->size != 0; src++) {
1.12      nicm      667:                dst = xreallocarray(dst, n + src->size, 1);
1.11      nicm      668:                memcpy(dst + n, src->data, src->size);
                    669:                n += src->size;
                    670:        }
                    671:
1.12      nicm      672:        dst = xreallocarray(dst, n + 1, 1);
1.11      nicm      673:        dst[n] = '\0';
                    674:        return (dst);
                    675: }
                    676:
                    677: /* Get width of UTF-8 string. */
                    678: u_int
                    679: utf8_cstrwidth(const char *s)
                    680: {
                    681:        struct utf8_data        tmp;
                    682:        u_int                   width;
1.23      nicm      683:        enum utf8_state         more;
1.11      nicm      684:
                    685:        width = 0;
                    686:        while (*s != '\0') {
1.23      nicm      687:                if ((more = utf8_open(&tmp, *s)) == UTF8_MORE) {
                    688:                        while (*++s != '\0' && more == UTF8_MORE)
1.11      nicm      689:                                more = utf8_append(&tmp, *s);
1.23      nicm      690:                        if (more == UTF8_DONE) {
1.11      nicm      691:                                width += tmp.width;
                    692:                                continue;
                    693:                        }
                    694:                        s -= tmp.have;
                    695:                }
1.23      nicm      696:                if (*s > 0x1f && *s != 0x7f)
1.21      nicm      697:                        width++;
1.11      nicm      698:                s++;
                    699:        }
                    700:        return (width);
                    701: }
                    702:
                    703: /* Trim UTF-8 string to width. Caller frees. */
                    704: char *
                    705: utf8_trimcstr(const char *s, u_int width)
                    706: {
                    707:        struct utf8_data        *tmp, *next;
                    708:        char                    *out;
                    709:        u_int                    at;
                    710:
                    711:        tmp = utf8_fromcstr(s);
                    712:
                    713:        at = 0;
                    714:        for (next = tmp; next->size != 0; next++) {
                    715:                if (at + next->width > width) {
                    716:                        next->size = 0;
                    717:                        break;
                    718:                }
                    719:                at += next->width;
                    720:        }
                    721:
                    722:        out = utf8_tocstr(tmp);
                    723:        free(tmp);
1.18      nicm      724:        return (out);
                    725: }
                    726:
                    727: /* Pad UTF-8 string to width. Caller frees. */
                    728: char *
                    729: utf8_padcstr(const char *s, u_int width)
                    730: {
                    731:        size_t   slen;
                    732:        char    *out;
                    733:        u_int     n, i;
                    734:
                    735:        n = utf8_cstrwidth(s);
                    736:        if (n >= width)
                    737:                return (xstrdup(s));
                    738:
                    739:        slen = strlen(s);
                    740:        out = xmalloc(slen + 1 + (width - n));
                    741:        memcpy(out, s, slen);
                    742:        for (i = n; i < width; i++)
                    743:                out[slen++] = ' ';
                    744:        out[slen] = '\0';
1.11      nicm      745:        return (out);
1.1       nicm      746: }