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

Annotation of src/usr.bin/less/charset.c, Revision 1.18

1.1       etheisen    1: /*
1.10      shadchin    2:  * Copyright (C) 1984-2012  Mark Nudelman
1.13      nicm        3:  * Modified for use with illumos by Garrett D'Amore.
                      4:  * Copyright 2014 Garrett D'Amore <garrett@damore.org>
1.1       etheisen    5:  *
1.5       millert     6:  * You may distribute under the terms of either the GNU General Public
                      7:  * License or the Less License, as specified in the README file.
1.1       etheisen    8:  *
1.10      shadchin    9:  * For more information, see the README file.
1.11      nicm       10:  */
1.1       etheisen   11:
                     12: /*
                     13:  * Functions to define the character set
                     14:  * and do things specific to the character set.
                     15:  */
                     16:
                     17: #include <ctype.h>
1.8       shadchin   18: #include <langinfo.h>
1.18    ! mmcc       19: #include <locale.h>
1.1       etheisen   20:
1.8       shadchin   21: #include "charset.h"
1.18    ! mmcc       22: #include "less.h"
1.8       shadchin   23:
1.11      nicm       24: int utf_mode = 0;
1.5       millert    25:
1.11      nicm       26: static const char *binfmt = NULL;
                     27: static const char *utfbinfmt = NULL;
                     28: int binattr = AT_STANDOUT;
1.1       etheisen   29:
1.11      nicm       30: static int
                     31: checkfmt(const char *s)
                     32: {
                     33:        char c;
                     34:        int seen = 0;
                     35:
                     36:        /* %[<flags>][<width>][.<prec>][<type>]<fmt> */
                     37:
                     38:        if (*s == '*') {                /* skip leading attribute if there */
                     39:                s++;
                     40:                if (strchr("dksu", *s) == NULL) {
                     41:                        return (-1);
                     42:                }
                     43:                s++;
                     44:        }
                     45:
                     46:        while ((c = *s++) != 0) {
                     47:                if (!isascii(c) || !isprint(c)) {
                     48:                        return (-1);
                     49:                }
                     50:                if (c != '%') {
                     51:                        continue;
                     52:                }
                     53:                if (*s == '%') {        /* % escaped with second % */
                     54:                        s++;
                     55:                        continue;
                     56:                }
                     57:                if (seen) {
                     58:                        return (-1);    /* 2nd % format item! */
                     59:                }
                     60:                while (strchr(" '+-0#", *s) != NULL) {  /* skip flags */
                     61:                        s++;
                     62:                }
                     63:                while (isdigit(*s)) {                   /* skip width */
                     64:                        s++;
                     65:                }
                     66:                if (*s == '.') {                /* skip precision */
                     67:                        s++;
                     68:                        while (isdigit(*s)) {
                     69:                                s++;
                     70:                        }
                     71:                }
                     72:                /* type width specifications, only "l", "h", and "hh" valid */
                     73:                if (*s == 'l') {
                     74:                        s++;
                     75:                } else if (*s == 'h') {
                     76:                        s++;
                     77:                        if (*s == 'h')
                     78:                                s++;
                     79:                }
                     80:
                     81:                if (strchr("cCdiouxX", *s) == NULL) {
                     82:                        /* bad or evil format character (%s, %n, etc.) */
                     83:                        return (-1);
                     84:                }
                     85:
                     86:                seen = 1;
                     87:        }
                     88:
                     89:        return (0);
                     90: }
1.1       etheisen   91:
                     92: /*
1.8       shadchin   93:  * Define the printing format for control (or binary utf) chars.
1.1       etheisen   94:  */
1.11      nicm       95: static void
                     96: setbinfmt(char *e, const char **fmtvarptr, const char *default_fmt)
1.1       etheisen   97: {
1.11      nicm       98:        const char *s;
                     99:
                    100:        if (((s = lgetenv(e)) == NULL) || (*s == 0)) {
                    101:                s = default_fmt;
                    102:                goto attr;
                    103:        }
                    104:
                    105:        if (s != NULL && *s != 0) {
                    106:                if (checkfmt(s) < 0) {
                    107:                        s = default_fmt;
                    108:                        goto attr;
1.8       shadchin  109:                }
                    110:        }
                    111:
1.1       etheisen  112:        /*
                    113:         * Select the attributes if it starts with "*".
                    114:         */
1.11      nicm      115: attr:
                    116:        if (*s == '*') {
                    117:                switch (s[1]) {
                    118:                case 'd':  binattr = AT_BOLD; break;
                    119:                case 'k':  binattr = AT_BLINK; break;
                    120:                case 's':  binattr = AT_STANDOUT; break;
1.1       etheisen  121:                case 'u':  binattr = AT_UNDERLINE; break;
1.11      nicm      122:                default:   binattr = AT_NORMAL; break;
1.1       etheisen  123:                }
                    124:                s += 2;
                    125:        }
1.8       shadchin  126:        *fmtvarptr = s;
1.1       etheisen  127: }
                    128:
                    129: /*
1.17      nicm      130:  * Initialize charset data structures.
1.1       etheisen  131:  */
1.17      nicm      132: void
                    133: init_charset(void)
1.1       etheisen  134: {
1.8       shadchin  135:        char *s;
                    136:
1.17      nicm      137:        setlocale(LC_ALL, "");
                    138:
1.8       shadchin  139:        s = nl_langinfo(CODESET);
1.16      tedu      140:        if (s && strcasecmp(s, "utf-8") == 0)
                    141:                utf_mode = 1;
1.8       shadchin  142:
1.11      nicm      143:        setbinfmt("LESSBINFMT", &binfmt, "*s<%02X>");
                    144:        setbinfmt("LESSUTFBINFMT", &utfbinfmt, "<U+%04lX>");
1.8       shadchin  145: }
                    146:
                    147: /*
1.1       etheisen  148:  * Is a given character a "binary" character?
                    149:  */
1.11      nicm      150: int
                    151: binary_char(LWCHAR c)
1.1       etheisen  152: {
1.8       shadchin  153:        if (utf_mode)
                    154:                return (is_ubin_char(c));
1.1       etheisen  155:        c &= 0377;
1.17      nicm      156:        return (!isprint((unsigned char)c) && !iscntrl((unsigned char)c));
1.1       etheisen  157: }
                    158:
                    159: /*
                    160:  * Is a given character a "control" character?
                    161:  */
1.11      nicm      162: int
                    163: control_char(LWCHAR c)
1.1       etheisen  164: {
                    165:        c &= 0377;
1.17      nicm      166:        return (iscntrl((unsigned char)c));
1.1       etheisen  167: }
                    168:
                    169: /*
                    170:  * Return the printable form of a character.
                    171:  * For example, in the "ascii" charset '\3' is printed as "^C".
                    172:  */
1.11      nicm      173: char *
                    174: prchar(LWCHAR c)
1.1       etheisen  175: {
1.8       shadchin  176:        /* {{ This buffer can be overrun if LESSBINFMT is a long string. }} */
                    177:        static char buf[32];
1.1       etheisen  178:
                    179:        c &= 0377;
1.8       shadchin  180:        if ((c < 128 || !utf_mode) && !control_char(c))
1.11      nicm      181:                (void) snprintf(buf, sizeof (buf), "%c", (int)c);
1.1       etheisen  182:        else if (c == ESC)
1.11      nicm      183:                (void) strlcpy(buf, "ESC", sizeof (buf));
                    184:        else if (c < 128 && !control_char(c ^ 0100))
                    185:                (void) snprintf(buf, sizeof (buf), "^%c", (int)(c ^ 0100));
1.1       etheisen  186:        else
1.11      nicm      187:                (void) snprintf(buf, sizeof (buf), binfmt, c);
1.8       shadchin  188:        return (buf);
                    189: }
                    190:
                    191: /*
                    192:  * Return the printable form of a UTF-8 character.
                    193:  */
1.11      nicm      194: char *
                    195: prutfchar(LWCHAR ch)
1.8       shadchin  196: {
                    197:        static char buf[32];
                    198:
1.11      nicm      199:        if (ch == ESC) {
                    200:                (void) strlcpy(buf, "ESC", sizeof (buf));
                    201:        } else if (ch < 128 && control_char(ch)) {
1.8       shadchin  202:                if (!control_char(ch ^ 0100))
1.11      nicm      203:                        (void) snprintf(buf, sizeof (buf), "^%c",
                    204:                            ((char)ch) ^ 0100);
1.8       shadchin  205:                else
1.11      nicm      206:                        (void) snprintf(buf, sizeof (buf), binfmt, (char)ch);
                    207:        } else if (is_ubin_char(ch)) {
                    208:                (void) snprintf(buf, sizeof (buf), utfbinfmt, ch);
                    209:        } else {
1.8       shadchin  210:                int len;
1.11      nicm      211:                if (ch >= 0x80000000) {
1.8       shadchin  212:                        len = 3;
                    213:                        ch = 0xFFFD;
1.11      nicm      214:                } else {
                    215:                        len =   (ch < 0x80) ? 1
                    216:                            : (ch < 0x800) ? 2
                    217:                            : (ch < 0x10000) ? 3
                    218:                            : (ch < 0x200000) ? 4
                    219:                            : (ch < 0x4000000) ? 5
                    220:                            : 6;
1.8       shadchin  221:                }
                    222:                buf[len] = '\0';
1.11      nicm      223:                if (len == 1) {
                    224:                        *buf = (char)ch;
                    225:                } else {
1.8       shadchin  226:                        *buf = ((1 << len) - 1) << (8 - len);
1.11      nicm      227:                        while (--len > 0) {
                    228:                                buf[len] = (char)(0x80 | (ch & 0x3F));
1.8       shadchin  229:                                ch >>= 6;
                    230:                        }
                    231:                        *buf |= ch;
                    232:                }
                    233:        }
1.1       etheisen  234:        return (buf);
                    235: }
1.6       millert   236:
1.8       shadchin  237: /*
                    238:  * Get the length of a UTF-8 character in bytes.
                    239:  */
1.11      nicm      240: int
                    241: utf_len(char ch)
1.8       shadchin  242: {
                    243:        if ((ch & 0x80) == 0)
1.11      nicm      244:                return (1);
1.8       shadchin  245:        if ((ch & 0xE0) == 0xC0)
1.11      nicm      246:                return (2);
1.8       shadchin  247:        if ((ch & 0xF0) == 0xE0)
1.11      nicm      248:                return (3);
1.8       shadchin  249:        if ((ch & 0xF8) == 0xF0)
1.11      nicm      250:                return (4);
1.8       shadchin  251:        if ((ch & 0xFC) == 0xF8)
1.11      nicm      252:                return (5);
1.8       shadchin  253:        if ((ch & 0xFE) == 0xFC)
1.11      nicm      254:                return (6);
1.8       shadchin  255:        /* Invalid UTF-8 encoding. */
1.11      nicm      256:        return (1);
1.8       shadchin  257: }
                    258:
                    259: /*
                    260:  * Is a UTF-8 character well-formed?
                    261:  */
1.11      nicm      262: int
                    263: is_utf8_well_formed(const char *s)
1.8       shadchin  264: {
                    265:        int i;
                    266:        int len;
                    267:
                    268:        if (IS_UTF8_INVALID(s[0]))
                    269:                return (0);
                    270:
1.11      nicm      271:        len = utf_len((char)s[0]);
1.8       shadchin  272:        if (len == 1)
                    273:                return (1);
1.11      nicm      274:        if (len == 2) {
                    275:                if ((unsigned char)(s[0]) < 0xC2)
                    276:                        return (0);
                    277:        } else {
1.8       shadchin  278:                unsigned char mask;
                    279:                mask = (~((1 << (8-len)) - 1)) & 0xFF;
                    280:                if (s[0] == mask && (s[1] & mask) == 0x80)
                    281:                        return (0);
                    282:        }
                    283:
                    284:        for (i = 1;  i < len;  i++)
                    285:                if (!IS_UTF8_TRAIL(s[i]))
                    286:                        return (0);
                    287:        return (1);
                    288: }
                    289:
                    290: /*
                    291:  * Get the value of a UTF-8 character.
                    292:  */
1.11      nicm      293: LWCHAR
                    294: get_wchar(const char *p)
1.8       shadchin  295: {
1.11      nicm      296:        switch (utf_len(p[0])) {
1.8       shadchin  297:        case 1:
                    298:        default:
                    299:                /* 0xxxxxxx */
                    300:                return (LWCHAR)
1.11      nicm      301:                    (p[0] & 0xFF);
1.8       shadchin  302:        case 2:
                    303:                /* 110xxxxx 10xxxxxx */
                    304:                return (LWCHAR) (
1.11      nicm      305:                    ((p[0] & 0x1F) << 6) |
                    306:                    (p[1] & 0x3F));
1.8       shadchin  307:        case 3:
                    308:                /* 1110xxxx 10xxxxxx 10xxxxxx */
                    309:                return (LWCHAR) (
1.11      nicm      310:                    ((p[0] & 0x0F) << 12) |
                    311:                    ((p[1] & 0x3F) << 6) |
                    312:                    (p[2] & 0x3F));
1.8       shadchin  313:        case 4:
                    314:                /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
                    315:                return (LWCHAR) (
1.11      nicm      316:                    ((p[0] & 0x07) << 18) |
                    317:                    ((p[1] & 0x3F) << 12) |
                    318:                    ((p[2] & 0x3F) << 6) |
                    319:                    (p[3] & 0x3F));
1.8       shadchin  320:        case 5:
                    321:                /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
                    322:                return (LWCHAR) (
1.11      nicm      323:                    ((p[0] & 0x03) << 24) |
                    324:                    ((p[1] & 0x3F) << 18) |
                    325:                    ((p[2] & 0x3F) << 12) |
                    326:                    ((p[3] & 0x3F) << 6) |
                    327:                    (p[4] & 0x3F));
1.8       shadchin  328:        case 6:
                    329:                /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
                    330:                return (LWCHAR) (
1.11      nicm      331:                    ((p[0] & 0x01) << 30) |
                    332:                    ((p[1] & 0x3F) << 24) |
                    333:                    ((p[2] & 0x3F) << 18) |
                    334:                    ((p[3] & 0x3F) << 12) |
                    335:                    ((p[4] & 0x3F) << 6) |
                    336:                    (p[5] & 0x3F));
1.8       shadchin  337:        }
                    338: }
                    339:
                    340: /*
                    341:  * Store a character into a UTF-8 string.
                    342:  */
1.11      nicm      343: void
                    344: put_wchar(char **pp, LWCHAR ch)
1.8       shadchin  345: {
1.11      nicm      346:        if (!utf_mode || ch < 0x80) {
1.8       shadchin  347:                /* 0xxxxxxx */
1.11      nicm      348:                *(*pp)++ = (char)ch;
                    349:        } else if (ch < 0x800) {
1.8       shadchin  350:                /* 110xxxxx 10xxxxxx */
1.11      nicm      351:                *(*pp)++ = (char)(0xC0 | ((ch >> 6) & 0x1F));
                    352:                *(*pp)++ = (char)(0x80 | (ch & 0x3F));
                    353:        } else if (ch < 0x10000) {
1.8       shadchin  354:                /* 1110xxxx 10xxxxxx 10xxxxxx */
1.11      nicm      355:                *(*pp)++ = (char)(0xE0 | ((ch >> 12) & 0x0F));
                    356:                *(*pp)++ = (char)(0x80 | ((ch >> 6) & 0x3F));
                    357:                *(*pp)++ = (char)(0x80 | (ch & 0x3F));
                    358:        } else if (ch < 0x200000) {
1.8       shadchin  359:                /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
1.11      nicm      360:                *(*pp)++ = (char)(0xF0 | ((ch >> 18) & 0x07));
                    361:                *(*pp)++ = (char)(0x80 | ((ch >> 12) & 0x3F));
                    362:                *(*pp)++ = (char)(0x80 | ((ch >> 6) & 0x3F));
                    363:                *(*pp)++ = (char)(0x80 | (ch & 0x3F));
                    364:        } else if (ch < 0x4000000) {
1.8       shadchin  365:                /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
1.11      nicm      366:                *(*pp)++ = (char)(0xF0 | ((ch >> 24) & 0x03));
                    367:                *(*pp)++ = (char)(0x80 | ((ch >> 18) & 0x3F));
                    368:                *(*pp)++ = (char)(0x80 | ((ch >> 12) & 0x3F));
                    369:                *(*pp)++ = (char)(0x80 | ((ch >> 6) & 0x3F));
                    370:                *(*pp)++ = (char)(0x80 | (ch & 0x3F));
                    371:        } else {
1.8       shadchin  372:                /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
1.11      nicm      373:                *(*pp)++ = (char)(0xF0 | ((ch >> 30) & 0x01));
                    374:                *(*pp)++ = (char)(0x80 | ((ch >> 24) & 0x3F));
                    375:                *(*pp)++ = (char)(0x80 | ((ch >> 18) & 0x3F));
                    376:                *(*pp)++ = (char)(0x80 | ((ch >> 12) & 0x3F));
                    377:                *(*pp)++ = (char)(0x80 | ((ch >> 6) & 0x3F));
                    378:                *(*pp)++ = (char)(0x80 | (ch & 0x3F));
1.8       shadchin  379:        }
                    380: }
                    381:
                    382: /*
                    383:  * Step forward or backward one character in a string.
                    384:  */
1.11      nicm      385: LWCHAR
                    386: step_char(char **pp, int dir, char *limit)
1.8       shadchin  387: {
                    388:        LWCHAR ch;
                    389:        int len;
                    390:        char *p = *pp;
                    391:
1.11      nicm      392:        if (!utf_mode) {
1.8       shadchin  393:                /* It's easy if chars are one byte. */
                    394:                if (dir > 0)
                    395:                        ch = (LWCHAR) ((p < limit) ? *p++ : 0);
                    396:                else
                    397:                        ch = (LWCHAR) ((p > limit) ? *--p : 0);
1.11      nicm      398:        } else if (dir > 0) {
1.8       shadchin  399:                len = utf_len(*p);
1.11      nicm      400:                if (p + len > limit) {
1.8       shadchin  401:                        ch = 0;
                    402:                        p = limit;
1.11      nicm      403:                } else {
1.8       shadchin  404:                        ch = get_wchar(p);
                    405:                        p += len;
                    406:                }
1.11      nicm      407:        } else {
1.8       shadchin  408:                while (p > limit && IS_UTF8_TRAIL(p[-1]))
                    409:                        p--;
                    410:                if (p > limit)
                    411:                        ch = get_wchar(--p);
                    412:                else
                    413:                        ch = 0;
                    414:        }
                    415:        *pp = p;
1.11      nicm      416:        return (ch);
1.8       shadchin  417: }
                    418:
                    419: /*
                    420:  * Unicode characters data
                    421:  */
                    422: struct wchar_range { LWCHAR first, last; };
                    423:
                    424: /*
                    425:  * Characters with general category values
                    426:  *     Mn: Mark, Nonspacing
                    427:  *     Me: Mark, Enclosing
                    428:  * Last synched with
                    429:  *     <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
                    430:  *     dated 2005-11-30T00:58:48Z
                    431:  */
                    432: static struct wchar_range comp_table[] = {
                    433:        {  0x0300,  0x036F} /* Mn */, {  0x0483,  0x0486} /* Mn */,
                    434:        {  0x0488,  0x0489} /* Me */,
                    435:        {  0x0591,  0x05BD} /* Mn */, {  0x05BF,  0x05BF} /* Mn */,
                    436:        {  0x05C1,  0x05C2} /* Mn */, {  0x05C4,  0x05C5} /* Mn */,
                    437:        {  0x05C7,  0x05C7} /* Mn */, {  0x0610,  0x0615} /* Mn */,
                    438:        {  0x064B,  0x065E} /* Mn */, {  0x0670,  0x0670} /* Mn */,
                    439:        {  0x06D6,  0x06DC} /* Mn */,
                    440:        {  0x06DE,  0x06DE} /* Me */,
                    441:        {  0x06DF,  0x06E4} /* Mn */, {  0x06E7,  0x06E8} /* Mn */,
                    442:        {  0x06EA,  0x06ED} /* Mn */, {  0x0711,  0x0711} /* Mn */,
                    443:        {  0x0730,  0x074A} /* Mn */, {  0x07A6,  0x07B0} /* Mn */,
                    444:        {  0x07EB,  0x07F3} /* Mn */, {  0x0901,  0x0902} /* Mn */,
                    445:        {  0x093C,  0x093C} /* Mn */, {  0x0941,  0x0948} /* Mn */,
                    446:        {  0x094D,  0x094D} /* Mn */, {  0x0951,  0x0954} /* Mn */,
                    447:        {  0x0962,  0x0963} /* Mn */, {  0x0981,  0x0981} /* Mn */,
                    448:        {  0x09BC,  0x09BC} /* Mn */, {  0x09C1,  0x09C4} /* Mn */,
                    449:        {  0x09CD,  0x09CD} /* Mn */, {  0x09E2,  0x09E3} /* Mn */,
                    450:        {  0x0A01,  0x0A02} /* Mn */, {  0x0A3C,  0x0A3C} /* Mn */,
                    451:        {  0x0A41,  0x0A42} /* Mn */, {  0x0A47,  0x0A48} /* Mn */,
                    452:        {  0x0A4B,  0x0A4D} /* Mn */, {  0x0A70,  0x0A71} /* Mn */,
                    453:        {  0x0A81,  0x0A82} /* Mn */, {  0x0ABC,  0x0ABC} /* Mn */,
                    454:        {  0x0AC1,  0x0AC5} /* Mn */, {  0x0AC7,  0x0AC8} /* Mn */,
                    455:        {  0x0ACD,  0x0ACD} /* Mn */, {  0x0AE2,  0x0AE3} /* Mn */,
                    456:        {  0x0B01,  0x0B01} /* Mn */, {  0x0B3C,  0x0B3C} /* Mn */,
                    457:        {  0x0B3F,  0x0B3F} /* Mn */, {  0x0B41,  0x0B43} /* Mn */,
                    458:        {  0x0B4D,  0x0B4D} /* Mn */, {  0x0B56,  0x0B56} /* Mn */,
                    459:        {  0x0B82,  0x0B82} /* Mn */, {  0x0BC0,  0x0BC0} /* Mn */,
                    460:        {  0x0BCD,  0x0BCD} /* Mn */, {  0x0C3E,  0x0C40} /* Mn */,
                    461:        {  0x0C46,  0x0C48} /* Mn */, {  0x0C4A,  0x0C4D} /* Mn */,
                    462:        {  0x0C55,  0x0C56} /* Mn */, {  0x0CBC,  0x0CBC} /* Mn */,
                    463:        {  0x0CBF,  0x0CBF} /* Mn */, {  0x0CC6,  0x0CC6} /* Mn */,
                    464:        {  0x0CCC,  0x0CCD} /* Mn */, {  0x0CE2,  0x0CE3} /* Mn */,
                    465:        {  0x0D41,  0x0D43} /* Mn */, {  0x0D4D,  0x0D4D} /* Mn */,
                    466:        {  0x0DCA,  0x0DCA} /* Mn */, {  0x0DD2,  0x0DD4} /* Mn */,
                    467:        {  0x0DD6,  0x0DD6} /* Mn */, {  0x0E31,  0x0E31} /* Mn */,
                    468:        {  0x0E34,  0x0E3A} /* Mn */, {  0x0E47,  0x0E4E} /* Mn */,
                    469:        {  0x0EB1,  0x0EB1} /* Mn */, {  0x0EB4,  0x0EB9} /* Mn */,
                    470:        {  0x0EBB,  0x0EBC} /* Mn */, {  0x0EC8,  0x0ECD} /* Mn */,
                    471:        {  0x0F18,  0x0F19} /* Mn */, {  0x0F35,  0x0F35} /* Mn */,
                    472:        {  0x0F37,  0x0F37} /* Mn */, {  0x0F39,  0x0F39} /* Mn */,
                    473:        {  0x0F71,  0x0F7E} /* Mn */, {  0x0F80,  0x0F84} /* Mn */,
                    474:        {  0x0F86,  0x0F87} /* Mn */, {  0x0F90,  0x0F97} /* Mn */,
                    475:        {  0x0F99,  0x0FBC} /* Mn */, {  0x0FC6,  0x0FC6} /* Mn */,
                    476:        {  0x102D,  0x1030} /* Mn */, {  0x1032,  0x1032} /* Mn */,
                    477:        {  0x1036,  0x1037} /* Mn */, {  0x1039,  0x1039} /* Mn */,
                    478:        {  0x1058,  0x1059} /* Mn */, {  0x135F,  0x135F} /* Mn */,
                    479:        {  0x1712,  0x1714} /* Mn */, {  0x1732,  0x1734} /* Mn */,
                    480:        {  0x1752,  0x1753} /* Mn */, {  0x1772,  0x1773} /* Mn */,
                    481:        {  0x17B7,  0x17BD} /* Mn */, {  0x17C6,  0x17C6} /* Mn */,
                    482:        {  0x17C9,  0x17D3} /* Mn */, {  0x17DD,  0x17DD} /* Mn */,
                    483:        {  0x180B,  0x180D} /* Mn */, {  0x18A9,  0x18A9} /* Mn */,
                    484:        {  0x1920,  0x1922} /* Mn */, {  0x1927,  0x1928} /* Mn */,
                    485:        {  0x1932,  0x1932} /* Mn */, {  0x1939,  0x193B} /* Mn */,
                    486:        {  0x1A17,  0x1A18} /* Mn */, {  0x1B00,  0x1B03} /* Mn */,
                    487:        {  0x1B34,  0x1B34} /* Mn */, {  0x1B36,  0x1B3A} /* Mn */,
                    488:        {  0x1B3C,  0x1B3C} /* Mn */, {  0x1B42,  0x1B42} /* Mn */,
                    489:        {  0x1B6B,  0x1B73} /* Mn */, {  0x1DC0,  0x1DCA} /* Mn */,
                    490:        {  0x1DFE,  0x1DFF} /* Mn */, {  0x20D0,  0x20DC} /* Mn */,
                    491:        {  0x20DD,  0x20E0} /* Me */,
                    492:        {  0x20E1,  0x20E1} /* Mn */,
                    493:        {  0x20E2,  0x20E4} /* Me */,
                    494:        {  0x20E5,  0x20EF} /* Mn */, {  0x302A,  0x302F} /* Mn */,
                    495:        {  0x3099,  0x309A} /* Mn */, {  0xA806,  0xA806} /* Mn */,
                    496:        {  0xA80B,  0xA80B} /* Mn */, {  0xA825,  0xA826} /* Mn */,
                    497:        {  0xFB1E,  0xFB1E} /* Mn */, {  0xFE00,  0xFE0F} /* Mn */,
                    498:        {  0xFE20,  0xFE23} /* Mn */, { 0x10A01, 0x10A03} /* Mn */,
                    499:        { 0x10A05, 0x10A06} /* Mn */, { 0x10A0C, 0x10A0F} /* Mn */,
                    500:        { 0x10A38, 0x10A3A} /* Mn */, { 0x10A3F, 0x10A3F} /* Mn */,
                    501:        { 0x1D167, 0x1D169} /* Mn */, { 0x1D17B, 0x1D182} /* Mn */,
                    502:        { 0x1D185, 0x1D18B} /* Mn */, { 0x1D1AA, 0x1D1AD} /* Mn */,
                    503:        { 0x1D242, 0x1D244} /* Mn */, { 0xE0100, 0xE01EF} /* Mn */,
                    504: };
                    505:
                    506: /*
                    507:  * Special pairs, not ranges.
                    508:  */
                    509: static struct wchar_range comb_table[] = {
1.11      nicm      510:        {0x0644, 0x0622}, {0x0644, 0x0623}, {0x0644, 0x0625}, {0x0644, 0x0627},
1.8       shadchin  511: };
                    512:
                    513: /*
                    514:  * Characters with general category values
                    515:  *     Cc: Other, Control
                    516:  *     Cf: Other, Format
                    517:  *     Cs: Other, Surrogate
                    518:  *     Co: Other, Private Use
                    519:  *     Cn: Other, Not Assigned
                    520:  *     Zl: Separator, Line
                    521:  *     Zp: Separator, Paragraph
                    522:  * Last synched with
                    523:  *     <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
                    524:  *     dated 2005-11-30T00:58:48Z
                    525:  */
                    526: static struct wchar_range ubin_table[] = {
1.11      nicm      527:        {  0x0000,  0x0007} /* Cc */,
                    528:        {  0x000B,  0x000C} /* Cc */,
                    529:        {  0x000E,  0x001A} /* Cc */,
                    530:        {  0x001C,  0x001F} /* Cc */,
1.8       shadchin  531:        {  0x007F,  0x009F} /* Cc */,
                    532: #if 0
                    533:        {  0x00AD,  0x00AD} /* Cf */,
                    534: #endif
                    535:        {  0x0370,  0x0373} /* Cn */, {  0x0376,  0x0379} /* Cn */,
                    536:        {  0x037F,  0x0383} /* Cn */, {  0x038B,  0x038B} /* Cn */,
                    537:        {  0x038D,  0x038D} /* Cn */, {  0x03A2,  0x03A2} /* Cn */,
                    538:        {  0x03CF,  0x03CF} /* Cn */, {  0x0487,  0x0487} /* Cn */,
                    539:        {  0x0514,  0x0530} /* Cn */, {  0x0557,  0x0558} /* Cn */,
                    540:        {  0x0560,  0x0560} /* Cn */, {  0x0588,  0x0588} /* Cn */,
                    541:        {  0x058B,  0x0590} /* Cn */, {  0x05C8,  0x05CF} /* Cn */,
                    542:        {  0x05EB,  0x05EF} /* Cn */, {  0x05F5,  0x05FF} /* Cn */,
                    543: #if 0
                    544:        {  0x0600,  0x0603} /* Cf */,
                    545: #endif
                    546:        {  0x0604,  0x060A} /* Cn */, {  0x0616,  0x061A} /* Cn */,
                    547:        {  0x061C,  0x061D} /* Cn */, {  0x0620,  0x0620} /* Cn */,
                    548:        {  0x063B,  0x063F} /* Cn */, {  0x065F,  0x065F} /* Cn */,
                    549: #if 0
                    550:        {  0x06DD,  0x06DD} /* Cf */,
                    551: #endif
                    552:        {  0x070E,  0x070E} /* Cn */,
                    553: #if 0
                    554:        {  0x070F,  0x070F} /* Cf */,
                    555: #endif
                    556:        {  0x074B,  0x074C} /* Cn */, {  0x076E,  0x077F} /* Cn */,
                    557:        {  0x07B2,  0x07BF} /* Cn */, {  0x07FB,  0x0900} /* Cn */,
                    558:        {  0x093A,  0x093B} /* Cn */, {  0x094E,  0x094F} /* Cn */,
                    559:        {  0x0955,  0x0957} /* Cn */, {  0x0971,  0x097A} /* Cn */,
                    560:        {  0x0980,  0x0980} /* Cn */, {  0x0984,  0x0984} /* Cn */,
                    561:        {  0x098D,  0x098E} /* Cn */, {  0x0991,  0x0992} /* Cn */,
                    562:        {  0x09A9,  0x09A9} /* Cn */, {  0x09B1,  0x09B1} /* Cn */,
                    563:        {  0x09B3,  0x09B5} /* Cn */, {  0x09BA,  0x09BB} /* Cn */,
                    564:        {  0x09C5,  0x09C6} /* Cn */, {  0x09C9,  0x09CA} /* Cn */,
                    565:        {  0x09CF,  0x09D6} /* Cn */, {  0x09D8,  0x09DB} /* Cn */,
                    566:        {  0x09DE,  0x09DE} /* Cn */, {  0x09E4,  0x09E5} /* Cn */,
                    567:        {  0x09FB,  0x0A00} /* Cn */, {  0x0A04,  0x0A04} /* Cn */,
                    568:        {  0x0A0B,  0x0A0E} /* Cn */, {  0x0A11,  0x0A12} /* Cn */,
                    569:        {  0x0A29,  0x0A29} /* Cn */, {  0x0A31,  0x0A31} /* Cn */,
                    570:        {  0x0A34,  0x0A34} /* Cn */, {  0x0A37,  0x0A37} /* Cn */,
                    571:        {  0x0A3A,  0x0A3B} /* Cn */, {  0x0A3D,  0x0A3D} /* Cn */,
                    572:        {  0x0A43,  0x0A46} /* Cn */, {  0x0A49,  0x0A4A} /* Cn */,
                    573:        {  0x0A4E,  0x0A58} /* Cn */, {  0x0A5D,  0x0A5D} /* Cn */,
                    574:        {  0x0A5F,  0x0A65} /* Cn */, {  0x0A75,  0x0A80} /* Cn */,
                    575:        {  0x0A84,  0x0A84} /* Cn */, {  0x0A8E,  0x0A8E} /* Cn */,
                    576:        {  0x0A92,  0x0A92} /* Cn */, {  0x0AA9,  0x0AA9} /* Cn */,
                    577:        {  0x0AB1,  0x0AB1} /* Cn */, {  0x0AB4,  0x0AB4} /* Cn */,
                    578:        {  0x0ABA,  0x0ABB} /* Cn */, {  0x0AC6,  0x0AC6} /* Cn */,
                    579:        {  0x0ACA,  0x0ACA} /* Cn */, {  0x0ACE,  0x0ACF} /* Cn */,
                    580:        {  0x0AD1,  0x0ADF} /* Cn */, {  0x0AE4,  0x0AE5} /* Cn */,
                    581:        {  0x0AF0,  0x0AF0} /* Cn */, {  0x0AF2,  0x0B00} /* Cn */,
                    582:        {  0x0B04,  0x0B04} /* Cn */, {  0x0B0D,  0x0B0E} /* Cn */,
                    583:        {  0x0B11,  0x0B12} /* Cn */, {  0x0B29,  0x0B29} /* Cn */,
                    584:        {  0x0B31,  0x0B31} /* Cn */, {  0x0B34,  0x0B34} /* Cn */,
                    585:        {  0x0B3A,  0x0B3B} /* Cn */, {  0x0B44,  0x0B46} /* Cn */,
                    586:        {  0x0B49,  0x0B4A} /* Cn */, {  0x0B4E,  0x0B55} /* Cn */,
                    587:        {  0x0B58,  0x0B5B} /* Cn */, {  0x0B5E,  0x0B5E} /* Cn */,
                    588:        {  0x0B62,  0x0B65} /* Cn */, {  0x0B72,  0x0B81} /* Cn */,
                    589:        {  0x0B84,  0x0B84} /* Cn */, {  0x0B8B,  0x0B8D} /* Cn */,
                    590:        {  0x0B91,  0x0B91} /* Cn */, {  0x0B96,  0x0B98} /* Cn */,
                    591:        {  0x0B9B,  0x0B9B} /* Cn */, {  0x0B9D,  0x0B9D} /* Cn */,
                    592:        {  0x0BA0,  0x0BA2} /* Cn */, {  0x0BA5,  0x0BA7} /* Cn */,
                    593:        {  0x0BAB,  0x0BAD} /* Cn */, {  0x0BBA,  0x0BBD} /* Cn */,
                    594:        {  0x0BC3,  0x0BC5} /* Cn */, {  0x0BC9,  0x0BC9} /* Cn */,
                    595:        {  0x0BCE,  0x0BD6} /* Cn */, {  0x0BD8,  0x0BE5} /* Cn */,
                    596:        {  0x0BFB,  0x0C00} /* Cn */, {  0x0C04,  0x0C04} /* Cn */,
                    597:        {  0x0C0D,  0x0C0D} /* Cn */, {  0x0C11,  0x0C11} /* Cn */,
                    598:        {  0x0C29,  0x0C29} /* Cn */, {  0x0C34,  0x0C34} /* Cn */,
                    599:        {  0x0C3A,  0x0C3D} /* Cn */, {  0x0C45,  0x0C45} /* Cn */,
                    600:        {  0x0C49,  0x0C49} /* Cn */, {  0x0C4E,  0x0C54} /* Cn */,
                    601:        {  0x0C57,  0x0C5F} /* Cn */, {  0x0C62,  0x0C65} /* Cn */,
                    602:        {  0x0C70,  0x0C81} /* Cn */, {  0x0C84,  0x0C84} /* Cn */,
                    603:        {  0x0C8D,  0x0C8D} /* Cn */, {  0x0C91,  0x0C91} /* Cn */,
                    604:        {  0x0CA9,  0x0CA9} /* Cn */, {  0x0CB4,  0x0CB4} /* Cn */,
                    605:        {  0x0CBA,  0x0CBB} /* Cn */, {  0x0CC5,  0x0CC5} /* Cn */,
                    606:        {  0x0CC9,  0x0CC9} /* Cn */, {  0x0CCE,  0x0CD4} /* Cn */,
                    607:        {  0x0CD7,  0x0CDD} /* Cn */, {  0x0CDF,  0x0CDF} /* Cn */,
                    608:        {  0x0CE4,  0x0CE5} /* Cn */, {  0x0CF0,  0x0CF0} /* Cn */,
                    609:        {  0x0CF3,  0x0D01} /* Cn */, {  0x0D04,  0x0D04} /* Cn */,
                    610:        {  0x0D0D,  0x0D0D} /* Cn */, {  0x0D11,  0x0D11} /* Cn */,
                    611:        {  0x0D29,  0x0D29} /* Cn */, {  0x0D3A,  0x0D3D} /* Cn */,
                    612:        {  0x0D44,  0x0D45} /* Cn */, {  0x0D49,  0x0D49} /* Cn */,
                    613:        {  0x0D4E,  0x0D56} /* Cn */, {  0x0D58,  0x0D5F} /* Cn */,
                    614:        {  0x0D62,  0x0D65} /* Cn */, {  0x0D70,  0x0D81} /* Cn */,
                    615:        {  0x0D84,  0x0D84} /* Cn */, {  0x0D97,  0x0D99} /* Cn */,
                    616:        {  0x0DB2,  0x0DB2} /* Cn */, {  0x0DBC,  0x0DBC} /* Cn */,
                    617:        {  0x0DBE,  0x0DBF} /* Cn */, {  0x0DC7,  0x0DC9} /* Cn */,
                    618:        {  0x0DCB,  0x0DCE} /* Cn */, {  0x0DD5,  0x0DD5} /* Cn */,
                    619:        {  0x0DD7,  0x0DD7} /* Cn */, {  0x0DE0,  0x0DF1} /* Cn */,
                    620:        {  0x0DF5,  0x0E00} /* Cn */, {  0x0E3B,  0x0E3E} /* Cn */,
                    621:        {  0x0E5C,  0x0E80} /* Cn */, {  0x0E83,  0x0E83} /* Cn */,
                    622:        {  0x0E85,  0x0E86} /* Cn */, {  0x0E89,  0x0E89} /* Cn */,
                    623:        {  0x0E8B,  0x0E8C} /* Cn */, {  0x0E8E,  0x0E93} /* Cn */,
                    624:        {  0x0E98,  0x0E98} /* Cn */, {  0x0EA0,  0x0EA0} /* Cn */,
                    625:        {  0x0EA4,  0x0EA4} /* Cn */, {  0x0EA6,  0x0EA6} /* Cn */,
                    626:        {  0x0EA8,  0x0EA9} /* Cn */, {  0x0EAC,  0x0EAC} /* Cn */,
                    627:        {  0x0EBA,  0x0EBA} /* Cn */, {  0x0EBE,  0x0EBF} /* Cn */,
                    628:        {  0x0EC5,  0x0EC5} /* Cn */, {  0x0EC7,  0x0EC7} /* Cn */,
                    629:        {  0x0ECE,  0x0ECF} /* Cn */, {  0x0EDA,  0x0EDB} /* Cn */,
                    630:        {  0x0EDE,  0x0EFF} /* Cn */, {  0x0F48,  0x0F48} /* Cn */,
                    631:        {  0x0F6B,  0x0F70} /* Cn */, {  0x0F8C,  0x0F8F} /* Cn */,
                    632:        {  0x0F98,  0x0F98} /* Cn */, {  0x0FBD,  0x0FBD} /* Cn */,
                    633:        {  0x0FCD,  0x0FCE} /* Cn */, {  0x0FD2,  0x0FFF} /* Cn */,
                    634:        {  0x1022,  0x1022} /* Cn */, {  0x1028,  0x1028} /* Cn */,
                    635:        {  0x102B,  0x102B} /* Cn */, {  0x1033,  0x1035} /* Cn */,
                    636:        {  0x103A,  0x103F} /* Cn */, {  0x105A,  0x109F} /* Cn */,
                    637:        {  0x10C6,  0x10CF} /* Cn */, {  0x10FD,  0x10FF} /* Cn */,
                    638:        {  0x115A,  0x115E} /* Cn */, {  0x11A3,  0x11A7} /* Cn */,
                    639:        {  0x11FA,  0x11FF} /* Cn */, {  0x1249,  0x1249} /* Cn */,
                    640:        {  0x124E,  0x124F} /* Cn */, {  0x1257,  0x1257} /* Cn */,
                    641:        {  0x1259,  0x1259} /* Cn */, {  0x125E,  0x125F} /* Cn */,
                    642:        {  0x1289,  0x1289} /* Cn */, {  0x128E,  0x128F} /* Cn */,
                    643:        {  0x12B1,  0x12B1} /* Cn */, {  0x12B6,  0x12B7} /* Cn */,
                    644:        {  0x12BF,  0x12BF} /* Cn */, {  0x12C1,  0x12C1} /* Cn */,
                    645:        {  0x12C6,  0x12C7} /* Cn */, {  0x12D7,  0x12D7} /* Cn */,
                    646:        {  0x1311,  0x1311} /* Cn */, {  0x1316,  0x1317} /* Cn */,
                    647:        {  0x135B,  0x135E} /* Cn */, {  0x137D,  0x137F} /* Cn */,
                    648:        {  0x139A,  0x139F} /* Cn */, {  0x13F5,  0x1400} /* Cn */,
                    649:        {  0x1677,  0x167F} /* Cn */, {  0x169D,  0x169F} /* Cn */,
                    650:        {  0x16F1,  0x16FF} /* Cn */, {  0x170D,  0x170D} /* Cn */,
                    651:        {  0x1715,  0x171F} /* Cn */, {  0x1737,  0x173F} /* Cn */,
                    652:        {  0x1754,  0x175F} /* Cn */, {  0x176D,  0x176D} /* Cn */,
                    653:        {  0x1771,  0x1771} /* Cn */, {  0x1774,  0x177F} /* Cn */,
                    654: #if 0
                    655:        {  0x17B4,  0x17B5} /* Cf */,
                    656: #endif
                    657:        {  0x17DE,  0x17DF} /* Cn */, {  0x17EA,  0x17EF} /* Cn */,
                    658:        {  0x17FA,  0x17FF} /* Cn */, {  0x180F,  0x180F} /* Cn */,
                    659:        {  0x181A,  0x181F} /* Cn */, {  0x1878,  0x187F} /* Cn */,
                    660:        {  0x18AA,  0x18FF} /* Cn */, {  0x191D,  0x191F} /* Cn */,
                    661:        {  0x192C,  0x192F} /* Cn */, {  0x193C,  0x193F} /* Cn */,
                    662:        {  0x1941,  0x1943} /* Cn */, {  0x196E,  0x196F} /* Cn */,
                    663:        {  0x1975,  0x197F} /* Cn */, {  0x19AA,  0x19AF} /* Cn */,
                    664:        {  0x19CA,  0x19CF} /* Cn */, {  0x19DA,  0x19DD} /* Cn */,
                    665:        {  0x1A1C,  0x1A1D} /* Cn */, {  0x1A20,  0x1AFF} /* Cn */,
                    666:        {  0x1B4C,  0x1B4F} /* Cn */, {  0x1B7D,  0x1CFF} /* Cn */,
                    667:        {  0x1DCB,  0x1DFD} /* Cn */, {  0x1E9C,  0x1E9F} /* Cn */,
                    668:        {  0x1EFA,  0x1EFF} /* Cn */, {  0x1F16,  0x1F17} /* Cn */,
                    669:        {  0x1F1E,  0x1F1F} /* Cn */, {  0x1F46,  0x1F47} /* Cn */,
                    670:        {  0x1F4E,  0x1F4F} /* Cn */, {  0x1F58,  0x1F58} /* Cn */,
                    671:        {  0x1F5A,  0x1F5A} /* Cn */, {  0x1F5C,  0x1F5C} /* Cn */,
                    672:        {  0x1F5E,  0x1F5E} /* Cn */, {  0x1F7E,  0x1F7F} /* Cn */,
                    673:        {  0x1FB5,  0x1FB5} /* Cn */, {  0x1FC5,  0x1FC5} /* Cn */,
                    674:        {  0x1FD4,  0x1FD5} /* Cn */, {  0x1FDC,  0x1FDC} /* Cn */,
                    675:        {  0x1FF0,  0x1FF1} /* Cn */, {  0x1FF5,  0x1FF5} /* Cn */,
                    676:        {  0x1FFF,  0x1FFF} /* Cn */,
                    677:        {  0x200B,  0x200F} /* Cf */,
                    678:        {  0x2028,  0x2028} /* Zl */,
                    679:        {  0x2029,  0x2029} /* Zp */,
                    680:        {  0x202A,  0x202E} /* Cf */,
                    681:        {  0x2060,  0x2063} /* Cf */,
                    682:        {  0x2064,  0x2069} /* Cn */,
                    683:        {  0x206A,  0x206F} /* Cf */,
                    684:        {  0x2072,  0x2073} /* Cn */, {  0x208F,  0x208F} /* Cn */,
                    685:        {  0x2095,  0x209F} /* Cn */, {  0x20B6,  0x20CF} /* Cn */,
                    686:        {  0x20F0,  0x20FF} /* Cn */, {  0x214F,  0x2152} /* Cn */,
                    687:        {  0x2185,  0x218F} /* Cn */, {  0x23E8,  0x23FF} /* Cn */,
                    688:        {  0x2427,  0x243F} /* Cn */, {  0x244B,  0x245F} /* Cn */,
                    689:        {  0x269D,  0x269F} /* Cn */, {  0x26B3,  0x2700} /* Cn */,
                    690:        {  0x2705,  0x2705} /* Cn */, {  0x270A,  0x270B} /* Cn */,
                    691:        {  0x2728,  0x2728} /* Cn */, {  0x274C,  0x274C} /* Cn */,
                    692:        {  0x274E,  0x274E} /* Cn */, {  0x2753,  0x2755} /* Cn */,
                    693:        {  0x2757,  0x2757} /* Cn */, {  0x275F,  0x2760} /* Cn */,
                    694:        {  0x2795,  0x2797} /* Cn */, {  0x27B0,  0x27B0} /* Cn */,
                    695:        {  0x27BF,  0x27BF} /* Cn */, {  0x27CB,  0x27CF} /* Cn */,
                    696:        {  0x27EC,  0x27EF} /* Cn */, {  0x2B1B,  0x2B1F} /* Cn */,
                    697:        {  0x2B24,  0x2BFF} /* Cn */, {  0x2C2F,  0x2C2F} /* Cn */,
                    698:        {  0x2C5F,  0x2C5F} /* Cn */, {  0x2C6D,  0x2C73} /* Cn */,
                    699:        {  0x2C78,  0x2C7F} /* Cn */, {  0x2CEB,  0x2CF8} /* Cn */,
                    700:        {  0x2D26,  0x2D2F} /* Cn */, {  0x2D66,  0x2D6E} /* Cn */,
                    701:        {  0x2D70,  0x2D7F} /* Cn */, {  0x2D97,  0x2D9F} /* Cn */,
                    702:        {  0x2DA7,  0x2DA7} /* Cn */, {  0x2DAF,  0x2DAF} /* Cn */,
                    703:        {  0x2DB7,  0x2DB7} /* Cn */, {  0x2DBF,  0x2DBF} /* Cn */,
                    704:        {  0x2DC7,  0x2DC7} /* Cn */, {  0x2DCF,  0x2DCF} /* Cn */,
                    705:        {  0x2DD7,  0x2DD7} /* Cn */, {  0x2DDF,  0x2DFF} /* Cn */,
                    706:        {  0x2E18,  0x2E1B} /* Cn */, {  0x2E1E,  0x2E7F} /* Cn */,
                    707:        {  0x2E9A,  0x2E9A} /* Cn */, {  0x2EF4,  0x2EFF} /* Cn */,
                    708:        {  0x2FD6,  0x2FEF} /* Cn */, {  0x2FFC,  0x2FFF} /* Cn */,
                    709:        {  0x3040,  0x3040} /* Cn */, {  0x3097,  0x3098} /* Cn */,
                    710:        {  0x3100,  0x3104} /* Cn */, {  0x312D,  0x3130} /* Cn */,
                    711:        {  0x318F,  0x318F} /* Cn */, {  0x31B8,  0x31BF} /* Cn */,
                    712:        {  0x31D0,  0x31EF} /* Cn */, {  0x321F,  0x321F} /* Cn */,
                    713:        {  0x3244,  0x324F} /* Cn */, {  0x32FF,  0x32FF} /* Cn */,
                    714:        {  0x4DB6,  0x4DBF} /* Cn */, {  0x9FBC,  0x9FFF} /* Cn */,
                    715:        {  0xA48D,  0xA48F} /* Cn */, {  0xA4C7,  0xA6FF} /* Cn */,
                    716:        {  0xA71B,  0xA71F} /* Cn */, {  0xA722,  0xA7FF} /* Cn */,
                    717:        {  0xA82C,  0xA83F} /* Cn */, {  0xA878,  0xABFF} /* Cn */,
                    718:        {  0xD7A4,  0xD7FF} /* Cn */,
                    719:        {  0xD800,  0xDFFF} /* Cs */,
                    720:        {  0xE000,  0xF8FF} /* Co */,
                    721:        {  0xFA2E,  0xFA2F} /* Cn */, {  0xFA6B,  0xFA6F} /* Cn */,
                    722:        {  0xFADA,  0xFAFF} /* Cn */, {  0xFB07,  0xFB12} /* Cn */,
                    723:        {  0xFB18,  0xFB1C} /* Cn */, {  0xFB37,  0xFB37} /* Cn */,
                    724:        {  0xFB3D,  0xFB3D} /* Cn */, {  0xFB3F,  0xFB3F} /* Cn */,
                    725:        {  0xFB42,  0xFB42} /* Cn */, {  0xFB45,  0xFB45} /* Cn */,
                    726:        {  0xFBB2,  0xFBD2} /* Cn */, {  0xFD40,  0xFD4F} /* Cn */,
                    727:        {  0xFD90,  0xFD91} /* Cn */, {  0xFDC8,  0xFDEF} /* Cn */,
                    728:        {  0xFDFE,  0xFDFF} /* Cn */, {  0xFE1A,  0xFE1F} /* Cn */,
                    729:        {  0xFE24,  0xFE2F} /* Cn */, {  0xFE53,  0xFE53} /* Cn */,
                    730:        {  0xFE67,  0xFE67} /* Cn */, {  0xFE6C,  0xFE6F} /* Cn */,
                    731:        {  0xFE75,  0xFE75} /* Cn */, {  0xFEFD,  0xFEFE} /* Cn */,
                    732:        {  0xFEFF,  0xFEFF} /* Cf */,
                    733:        {  0xFF00,  0xFF00} /* Cn */, {  0xFFBF,  0xFFC1} /* Cn */,
                    734:        {  0xFFC8,  0xFFC9} /* Cn */, {  0xFFD0,  0xFFD1} /* Cn */,
                    735:        {  0xFFD8,  0xFFD9} /* Cn */, {  0xFFDD,  0xFFDF} /* Cn */,
                    736:        {  0xFFE7,  0xFFE7} /* Cn */, {  0xFFEF,  0xFFF8} /* Cn */,
                    737:        {  0xFFF9,  0xFFFB} /* Cf */,
                    738:        {  0xFFFE,  0xFFFF} /* Cn */, { 0x1000C, 0x1000C} /* Cn */,
                    739:        { 0x10027, 0x10027} /* Cn */, { 0x1003B, 0x1003B} /* Cn */,
                    740:        { 0x1003E, 0x1003E} /* Cn */, { 0x1004E, 0x1004F} /* Cn */,
                    741:        { 0x1005E, 0x1007F} /* Cn */, { 0x100FB, 0x100FF} /* Cn */,
                    742:        { 0x10103, 0x10106} /* Cn */, { 0x10134, 0x10136} /* Cn */,
                    743:        { 0x1018B, 0x102FF} /* Cn */, { 0x1031F, 0x1031F} /* Cn */,
                    744:        { 0x10324, 0x1032F} /* Cn */, { 0x1034B, 0x1037F} /* Cn */,
                    745:        { 0x1039E, 0x1039E} /* Cn */, { 0x103C4, 0x103C7} /* Cn */,
                    746:        { 0x103D6, 0x103FF} /* Cn */,
                    747:        { 0x1049E, 0x1049F} /* Cn */, { 0x104AA, 0x107FF} /* Cn */,
                    748:        { 0x10806, 0x10807} /* Cn */, { 0x10809, 0x10809} /* Cn */,
                    749:        { 0x10836, 0x10836} /* Cn */, { 0x10839, 0x1083B} /* Cn */,
                    750:        { 0x1083D, 0x1083E} /* Cn */, { 0x10840, 0x108FF} /* Cn */,
                    751:        { 0x1091A, 0x1091E} /* Cn */, { 0x10920, 0x109FF} /* Cn */,
                    752:        { 0x10A04, 0x10A04} /* Cn */, { 0x10A07, 0x10A0B} /* Cn */,
                    753:        { 0x10A14, 0x10A14} /* Cn */, { 0x10A18, 0x10A18} /* Cn */,
                    754:        { 0x10A34, 0x10A37} /* Cn */, { 0x10A3B, 0x10A3E} /* Cn */,
                    755:        { 0x10A48, 0x10A4F} /* Cn */, { 0x10A59, 0x11FFF} /* Cn */,
                    756:        { 0x1236F, 0x123FF} /* Cn */, { 0x12463, 0x1246F} /* Cn */,
                    757:        { 0x12474, 0x1CFFF} /* Cn */, { 0x1D0F6, 0x1D0FF} /* Cn */,
                    758:        { 0x1D127, 0x1D129} /* Cn */,
                    759:        { 0x1D173, 0x1D17A} /* Cf */,
                    760:        { 0x1D1DE, 0x1D1FF} /* Cn */, { 0x1D246, 0x1D2FF} /* Cn */,
                    761:        { 0x1D357, 0x1D35F} /* Cn */, { 0x1D372, 0x1D3FF} /* Cn */,
                    762:        { 0x1D455, 0x1D455} /* Cn */, { 0x1D49D, 0x1D49D} /* Cn */,
                    763:        { 0x1D4A0, 0x1D4A1} /* Cn */, { 0x1D4A3, 0x1D4A4} /* Cn */,
                    764:        { 0x1D4A7, 0x1D4A8} /* Cn */, { 0x1D4AD, 0x1D4AD} /* Cn */,
                    765:        { 0x1D4BA, 0x1D4BA} /* Cn */, { 0x1D4BC, 0x1D4BC} /* Cn */,
                    766:        { 0x1D4C4, 0x1D4C4} /* Cn */, { 0x1D506, 0x1D506} /* Cn */,
                    767:        { 0x1D50B, 0x1D50C} /* Cn */, { 0x1D515, 0x1D515} /* Cn */,
                    768:        { 0x1D51D, 0x1D51D} /* Cn */, { 0x1D53A, 0x1D53A} /* Cn */,
                    769:        { 0x1D53F, 0x1D53F} /* Cn */, { 0x1D545, 0x1D545} /* Cn */,
                    770:        { 0x1D547, 0x1D549} /* Cn */, { 0x1D551, 0x1D551} /* Cn */,
                    771:        { 0x1D6A6, 0x1D6A7} /* Cn */, { 0x1D7CC, 0x1D7CD} /* Cn */,
                    772:        { 0x1D800, 0x1FFFF} /* Cn */, { 0x2A6D7, 0x2F7FF} /* Cn */,
                    773:        { 0x2FA1E, 0xE0000} /* Cn */,
                    774:        { 0xE0001, 0xE0001} /* Cf */,
                    775:        { 0xE0002, 0xE001F} /* Cn */,
                    776:        { 0xE0020, 0xE007F} /* Cf */,
                    777:        { 0xE0080, 0xE00FF} /* Cn */, { 0xE01F0, 0xEFFFF} /* Cn */,
                    778:        { 0xF0000, 0xFFFFD} /* Co */,
                    779:        { 0xFFFFE, 0xFFFFF} /* Cn */,
1.11      nicm      780:        { 0x100000, 0x10FFFD} /* Co */,
                    781:        { 0x10FFFE, 0x10FFFF} /* Cn */,
                    782:        { 0x110000, 0x7FFFFFFF} /* ISO 10646?? */
1.8       shadchin  783: };
                    784:
                    785: /*
                    786:  * Double width characters
                    787:  *     W: East Asian Wide
                    788:  *     F: East Asian Full-width
                    789:  * Unassigned code points may be included when they allow ranges to be merged.
                    790:  * Last synched with
                    791:  *     <http://www.unicode.org/Public/5.0.0/ucd/EastAsianWidth-5.0.0d2.txt>
                    792:  *     dated 2005-11-08T01:32:56Z
                    793:  */
                    794: static struct wchar_range wide_table[] = {
                    795:        {  0x1100,  0x115F} /* W */, {  0x2329,  0x232A} /* W */,
                    796:        {  0x2E80,  0x2FFB} /* W */,
                    797:        {  0x3000,  0x3000} /* F */,
                    798:        {  0x3001,  0x303E} /* W */, {  0x3041,  0x4DB5} /* W */,
                    799:        {  0x4E00,  0x9FBB} /* W */, {  0xA000,  0xA4C6} /* W */,
                    800:        {  0xAC00,  0xD7A3} /* W */, {  0xF900,  0xFAD9} /* W */,
                    801:        {  0xFE10,  0xFE19} /* W */, {  0xFE30,  0xFE6B} /* W */,
                    802:        {  0xFF01,  0xFF60} /* F */, {  0xFFE0,  0xFFE6} /* F */,
                    803:        { 0x20000, 0x2FFFD} /* W */, { 0x30000, 0x3FFFD} /* W */,
                    804: };
                    805:
1.11      nicm      806: static int
                    807: is_in_table(LWCHAR ch, struct wchar_range table[], int tsize)
1.8       shadchin  808: {
                    809:        int hi;
                    810:        int lo;
                    811:
                    812:        /* Binary search in the table. */
                    813:        if (ch < table[0].first)
1.11      nicm      814:                return (0);
1.8       shadchin  815:        lo = 0;
                    816:        hi = tsize - 1;
1.11      nicm      817:        while (lo <= hi) {
1.8       shadchin  818:                int mid = (lo + hi) / 2;
                    819:                if (ch > table[mid].last)
                    820:                        lo = mid + 1;
                    821:                else if (ch < table[mid].first)
                    822:                        hi = mid - 1;
                    823:                else
1.11      nicm      824:                        return (1);
1.8       shadchin  825:        }
1.11      nicm      826:        return (0);
1.8       shadchin  827: }
                    828:
                    829: /*
                    830:  * Is a character a UTF-8 composing character?
                    831:  * If a composing character follows any char, the two combine into one glyph.
                    832:  */
1.11      nicm      833: int
                    834: is_composing_char(LWCHAR ch)
1.8       shadchin  835: {
1.11      nicm      836:        return (is_in_table(ch, comp_table,
                    837:            (sizeof (comp_table) / sizeof (*comp_table))));
1.8       shadchin  838: }
                    839:
                    840: /*
                    841:  * Should this UTF-8 character be treated as binary?
                    842:  */
1.11      nicm      843: int
                    844: is_ubin_char(LWCHAR ch)
1.8       shadchin  845: {
1.11      nicm      846:        return (is_in_table(ch, ubin_table,
                    847:            (sizeof (ubin_table) / sizeof (*ubin_table))));
1.8       shadchin  848: }
                    849:
                    850: /*
                    851:  * Is this a double width UTF-8 character?
                    852:  */
1.11      nicm      853: int
                    854: is_wide_char(LWCHAR ch)
1.8       shadchin  855: {
1.11      nicm      856:        return (is_in_table(ch, wide_table,
                    857:            (sizeof (wide_table) / sizeof (*wide_table))));
1.8       shadchin  858: }
                    859:
                    860: /*
                    861:  * Is a character a UTF-8 combining character?
                    862:  * A combining char acts like an ordinary char, but if it follows
                    863:  * a specific char (not any char), the two combine into one glyph.
                    864:  */
1.11      nicm      865: int
                    866: is_combining_char(LWCHAR ch1, LWCHAR ch2)
1.8       shadchin  867: {
                    868:        /* The table is small; use linear search. */
                    869:        int i;
1.11      nicm      870:        for (i = 0;  i < sizeof (comb_table) / sizeof (*comb_table);  i++) {
1.8       shadchin  871:                if (ch1 == comb_table[i].first &&
                    872:                    ch2 == comb_table[i].last)
1.11      nicm      873:                        return (1);
1.8       shadchin  874:        }
1.11      nicm      875:        return (0);
1.8       shadchin  876: }