?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Fix typo.
[dragonfly.git] / contrib / less-4 / charset.c
blob37d344de69f45436da8231cfdbcff020761b08be
1 /*
2 * Copyright (C) 1984-2007 Mark Nudelman
3 *
4 * You may distribute under the terms of either the GNU General Public
5 * License or the Less License, as specified in the README file.
6 *
7 * For more information about less, or for information on how to
8 * contact the author, see the README file.
9 */
10
11
12 /*
13 * Functions to define the character set
14 * and do things specific to the character set.
15 */
16
17 #include "less.h"
18 #if HAVE_LOCALE
19 #include <locale.h>
20 #include <ctype.h>
21 #include <langinfo.h>
22 #endif
23
24 #include "charset.h"
25
26 public int utf_mode = 0;
27
28 /*
29 * Predefined character sets,
30 * selected by the LESSCHARSET environment variable.
31 */
32 struct charset {
33 char *name;
34 int *p_flag;
35 char *desc;
36 } charsets[] = {
37 { "ascii", NULL, "8bcccbcc18b95.b" },
38 { "utf-8", &utf_mode, "8bcccbcc18b95.b126.bb" },
39 { "iso8859", NULL, "8bcccbcc18b95.33b." },
40 { "latin3", NULL, "8bcccbcc18b95.33b5.b8.b15.b4.b12.b18.b12.b." },
41 { "arabic", NULL, "8bcccbcc18b95.33b.3b.7b2.13b.3b.b26.5b19.b" },
42 { "greek", NULL, "8bcccbcc18b95.33b4.2b4.b3.b35.b44.b" },
43 { "greek2005", NULL, "8bcccbcc18b95.33b14.b35.b44.b" },
44 { "hebrew", NULL, "8bcccbcc18b95.33b.b29.32b28.2b2.b" },
45 { "koi8-r", NULL, "8bcccbcc18b95.b." },
46 { "KOI8-T", NULL, "8bcccbcc18b95.b8.b6.b8.b.b.5b7.3b4.b4.b3.b.b.3b." },
47 { "georgianps", NULL, "8bcccbcc18b95.3b11.4b12.2b." },
48 { "tcvn", NULL, "b..b...bcccbccbbb7.8b95.b48.5b." },
49 { "TIS-620", NULL, "8bcccbcc18b95.b.4b.11b7.8b." },
50 { "next", NULL, "8bcccbcc18b95.bb125.bb" },
51 { "dos", NULL, "8bcccbcc12bc5b95.b." },
52 { "windows-1251", NULL, "8bcccbcc12bc5b95.b24.b." },
53 { "windows-1252", NULL, "8bcccbcc12bc5b95.b.b11.b.2b12.b." },
54 { "windows-1255", NULL, "8bcccbcc12bc5b95.b.b8.b.5b9.b.4b." },
55 { "ebcdic", NULL, "5bc6bcc7bcc41b.9b7.9b5.b..8b6.10b6.b9.7b9.8b8.17b3.3b9.7b9.8b8.6b10.b.b.b." },
56 { "IBM-1047", NULL, "4cbcbc3b9cbccbccbb4c6bcc5b3cbbc4bc4bccbc191.b" },
57 { NULL, NULL, NULL }
58 };
59
60 /*
61 * Support "locale charmap"/nl_langinfo(CODESET) values, as well as others.
62 */
63 struct cs_alias {
64 char *name;
65 char *oname;
66 } cs_aliases[] = {
67 { "UTF-8", "utf-8" },
68 { "ANSI_X3.4-1968", "ascii" },
69 { "US-ASCII", "ascii" },
70 { "latin1", "iso8859" },
71 { "ISO-8859-1", "iso8859" },
72 { "latin9", "iso8859" },
73 { "ISO-8859-15", "iso8859" },
74 { "latin2", "iso8859" },
75 { "ISO-8859-2", "iso8859" },
76 { "ISO-8859-3", "latin3" },
77 { "latin4", "iso8859" },
78 { "ISO-8859-4", "iso8859" },
79 { "cyrillic", "iso8859" },
80 { "ISO-8859-5", "iso8859" },
81 { "ISO-8859-6", "arabic" },
82 { "ISO-8859-7", "greek" },
83 { "IBM9005", "greek2005" },
84 { "ISO-8859-8", "hebrew" },
85 { "latin5", "iso8859" },
86 { "ISO-8859-9", "iso8859" },
87 { "latin6", "iso8859" },
88 { "ISO-8859-10", "iso8859" },
89 { "latin7", "iso8859" },
90 { "ISO-8859-13", "iso8859" },
91 { "latin8", "iso8859" },
92 { "ISO-8859-14", "iso8859" },
93 { "latin10", "iso8859" },
94 { "ISO-8859-16", "iso8859" },
95 { "IBM437", "dos" },
96 { "EBCDIC-US", "ebcdic" },
97 { "IBM1047", "IBM-1047" },
98 { "KOI8-R", "koi8-r" },
99 { "KOI8-U", "koi8-r" },
100 { "GEORGIAN-PS", "georgianps" },
101 { "TCVN5712-1", "tcvn" },
102 { "NEXTSTEP", "next" },
103 { "windows", "windows-1252" }, /* backward compatibility */
104 { "CP1251", "windows-1251" },
105 { "CP1252", "windows-1252" },
106 { "CP1255", "windows-1255" },
107 { NULL, NULL }
108 };
109
110 #define IS_BINARY_CHAR 01
111 #define IS_CONTROL_CHAR 02
112
113 static char chardef[256];
114 static char *binfmt = NULL;
115 static char *utfbinfmt = NULL;
116 public int binattr = AT_STANDOUT;
117
118
119 /*
120 * Define a charset, given a description string.
121 * The string consists of 256 letters,
122 * one for each character in the charset.
123 * If the string is shorter than 256 letters, missing letters
124 * are taken to be identical to the last one.
125 * A decimal number followed by a letter is taken to be a
126 * repetition of the letter.
127 *
128 * Each letter is one of:
129 * . normal character
130 * b binary character
131 * c control character
132 */
133 static void
134 ichardef(s)
135 char *s;
136 {
137 register char *cp;
138 register int n;
139 register char v;
140
141 n = 0;
142 v = 0;
143 cp = chardef;
144 while (*s != '\0')
145 {
146 switch (*s++)
147 {
148 case '.':
149 v = 0;
150 break;
151 case 'c':
152 v = IS_CONTROL_CHAR;
153 break;
154 case 'b':
155 v = IS_BINARY_CHAR|IS_CONTROL_CHAR;
156 break;
157
158 case '0': case '1': case '2': case '3': case '4':
159 case '5': case '6': case '7': case '8': case '9':
160 n = (10 * n) + (s[-1] - '0');
161 continue;
162
163 default:
164 error("invalid chardef", NULL_PARG);
165 quit(QUIT_ERROR);
166 /*NOTREACHED*/
167 }
168
169 do
170 {
171 if (cp >= chardef + sizeof(chardef))
172 {
173 error("chardef longer than 256", NULL_PARG);
174 quit(QUIT_ERROR);
175 /*NOTREACHED*/
176 }
177 *cp++ = v;
178 } while (--n > 0);
179 n = 0;
180 }
181
182 while (cp < chardef + sizeof(chardef))
183 *cp++ = v;
184 }
185
186 /*
187 * Define a charset, given a charset name.
188 * The valid charset names are listed in the "charsets" array.
189 */
190 static int
191 icharset(name, no_error)
192 register char *name;
193 int no_error;
194 {
195 register struct charset *p;
196 register struct cs_alias *a;
197
198 if (name == NULL || *name == '\0')
199 return (0);
200
201 /* First see if the name is an alias. */
202 for (a = cs_aliases; a->name != NULL; a++)
203 {
204 if (strcmp(name, a->name) == 0)
205 {
206 name = a->oname;
207 break;
208 }
209 }
210
211 for (p = charsets; p->name != NULL; p++)
212 {
213 if (strcmp(name, p->name) == 0)
214 {
215 ichardef(p->desc);
216 if (p->p_flag != NULL)
217 *(p->p_flag) = 1;
218 return (1);
219 }
220 }
221
222 if (!no_error) {
223 error("invalid charset name", NULL_PARG);
224 quit(QUIT_ERROR);
225 }
226 return (0);
227 }
228
229 #if HAVE_LOCALE
230 /*
231 * Define a charset, given a locale name.
232 */
233 static void
234 ilocale()
235 {
236 register int c;
237
238 for (c = 0; c < (int) sizeof(chardef); c++)
239 {
240 if (isprint(c))
241 chardef[c] = 0;
242 else if (iscntrl(c))
243 chardef[c] = IS_CONTROL_CHAR;
244 else
245 chardef[c] = IS_BINARY_CHAR|IS_CONTROL_CHAR;
246 }
247 }
248 #endif
249
250 /*
251 * Define the printing format for control (or binary utf) chars.
252 */
253 static void
254 setbinfmt(s, fmtvarptr, default_fmt)
255 char *s;
256 char **fmtvarptr;
257 char *default_fmt;
258 {
259 if (s && utf_mode)
260 {
261 /* It would be too hard to account for width otherwise. */
262 char *t = s;
263 while (*t)
264 {
265 if (*t < ' ' || *t > '~')
266 {
267 s = default_fmt;
268 goto attr;
269 }
270 t++;
271 }
272 }
273
274 /* %n is evil */
275 if (s == NULL || *s == '\0' ||
276 (*s == '*' && (s[1] == '\0' || s[2] == '\0' || strchr(s + 2, 'n'))) ||
277 (*s != '*' && strchr(s, 'n')))
278 s = default_fmt;
279
280 /*
281 * Select the attributes if it starts with "*".
282 */
283 attr:
284 if (*s == '*')
285 {
286 switch (s[1])
287 {
288 case 'd': binattr = AT_BOLD; break;
289 case 'k': binattr = AT_BLINK; break;
290 case 's': binattr = AT_STANDOUT; break;
291 case 'u': binattr = AT_UNDERLINE; break;
292 default: binattr = AT_NORMAL; break;
293 }
294 s += 2;
295 }
296 *fmtvarptr = s;
297 }
298
299 /*
300 *
301 */
302 static void
303 set_charset()
304 {
305 char *s;
306
307 /*
308 * See if environment variable LESSCHARSET is defined.
309 */
310 s = lgetenv("LESSCHARSET");
311 if (icharset(s, 0))
312 return;
313
314 /*
315 * LESSCHARSET is not defined: try LESSCHARDEF.
316 */
317 s = lgetenv("LESSCHARDEF");
318 if (s != NULL && *s != '\0')
319 {
320 ichardef(s);
321 return;
322 }
323
324 #if HAVE_LOCALE
325 #ifdef CODESET
326 /*
327 * Try using the codeset name as the charset name.
328 */
329 s = nl_langinfo(CODESET);
330 if (icharset(s, 1))
331 return;
332 #endif
333 #endif
334
335 #if HAVE_STRSTR
336 /*
337 * Check whether LC_ALL, LC_CTYPE or LANG look like UTF-8 is used.
338 */
339 if ((s = lgetenv("LC_ALL")) != NULL ||
340 (s = lgetenv("LC_CTYPE")) != NULL ||
341 (s = lgetenv("LANG")) != NULL)
342 {
343 if ( strstr(s, "UTF-8") != NULL || strstr(s, "utf-8") != NULL
344 || strstr(s, "UTF8") != NULL || strstr(s, "utf8") != NULL)
345 if (icharset("utf-8", 1))
346 return;
347 }
348 #endif
349
350 #if HAVE_LOCALE
351 /*
352 * Get character definitions from locale functions,
353 * rather than from predefined charset entry.
354 */
355 ilocale();
356 #if MSDOS_COMPILER
357 /*
358 * Default to "dos".
359 */
360 (void) icharset("dos", 1);
361 #else
362 /*
363 * Default to "latin1".
364 */
365 (void) icharset("latin1", 1);
366 #endif
367 #endif
368 }
369
370 /*
371 * Initialize charset data structures.
372 */
373 public void
374 init_charset()
375 {
376 char *s;
377
378 #if HAVE_LOCALE
379 setlocale(LC_ALL, "");
380 #endif
381
382 set_charset();
383
384 s = lgetenv("LESSBINFMT");
385 setbinfmt(s, &binfmt, "*s<%02X>");
386
387 s = lgetenv("LESSUTFBINFMT");
388 setbinfmt(s, &utfbinfmt, "<U+%04lX>");
389 }
390
391 /*
392 * Is a given character a "binary" character?
393 */
394 public int
395 binary_char(c)
396 unsigned char c;
397 {
398 c &= 0377;
399 return (chardef[c] & IS_BINARY_CHAR);
400 }
401
402 /*
403 * Is a given character a "control" character?
404 */
405 public int
406 control_char(c)
407 int c;
408 {
409 c &= 0377;
410 return (chardef[c] & IS_CONTROL_CHAR);
411 }
412
413 /*
414 * Return the printable form of a character.
415 * For example, in the "ascii" charset '\3' is printed as "^C".
416 */
417 public char *
418 prchar(c)
419 int c;
420 {
421 /* {{ This buffer can be overrun if LESSBINFMT is a long string. }} */
422 static char buf[32];
423
424 c &= 0377;
425 if ((c < 128 || !utf_mode) && !control_char(c))
426 SNPRINTF1(buf, sizeof(buf), "%c", c);
427 else if (c == ESC)
428 strcpy(buf, "ESC");
429 #if IS_EBCDIC_HOST
430 else if (!binary_char(c) && c < 64)
431 SNPRINTF1(buf, sizeof(buf), "^%c",
432 /*
433 * This array roughly inverts CONTROL() #defined in less.h,
434 * and should be kept in sync with CONTROL() and IBM-1047.
435 */
436 "@ABC.I.?...KLMNO"
437 "PQRS.JH.XY.."
438 "\\]^_"
439 "......W[.....EFG"
440 "..V....D....TU.Z"[c]);
441 #else
442 else if (c < 128 && !control_char(c ^ 0100))
443 SNPRINTF1(buf, sizeof(buf), "^%c", c ^ 0100);
444 #endif
445 else
446 SNPRINTF1(buf, sizeof(buf), binfmt, c);
447 return (buf);
448 }
449
450 /*
451 * Return the printable form of a UTF-8 character.
452 */
453 public char *
454 prutfchar(ch)
455 LWCHAR ch;
456 {
457 static char buf[32];
458
459 if (ch == ESC)
460 strcpy(buf, "ESC");
461 else if (ch < 128 && control_char(ch))
462 {
463 if (!control_char(ch ^ 0100))
464 SNPRINTF1(buf, sizeof(buf), "^%c", ((char) ch) ^ 0100);
465 else
466 SNPRINTF1(buf, sizeof(buf), binfmt, (char) ch);
467 } else if (is_ubin_char(ch))
468 SNPRINTF1(buf, sizeof(buf), utfbinfmt, ch);
469 else
470 {
471 int len;
472 if (ch >= 0x80000000)
473 {
474 len = 3;
475 ch = 0xFFFD;
476 } else
477 {
478 len = (ch < 0x80) ? 1
479 : (ch < 0x800) ? 2
480 : (ch < 0x10000) ? 3
481 : (ch < 0x200000) ? 4
482 : (ch < 0x4000000) ? 5
483 : 6;
484 }
485 buf[len] = '\0';
486 if (len == 1)
487 *buf = (char) ch;
488 else
489 {
490 *buf = ((1 << len) - 1) << (8 - len);
491 while (--len > 0)
492 {
493 buf[len] = (char) (0x80 | (ch & 0x3F));
494 ch >>= 6;
495 }
496 *buf |= ch;
497 }
498 }
499 return (buf);
500 }
501
502 /*
503 * Get the length of a UTF-8 character in bytes.
504 */
505 public int
506 utf_len(ch)
507 char ch;
508 {
509 if ((ch & 0x80) == 0)
510 return 1;
511 if ((ch & 0xE0) == 0xC0)
512 return 2;
513 if ((ch & 0xF0) == 0xE0)
514 return 3;
515 if ((ch & 0xF8) == 0xF0)
516 return 4;
517 if ((ch & 0xFC) == 0xF8)
518 return 5;
519 if ((ch & 0xFE) == 0xFC)
520 return 6;
521 /* Invalid UTF-8 encoding. */
522 return 1;
523 }
524
525 /*
526 * Is a UTF-8 character well-formed?
527 */
528 public int
529 is_utf8_well_formed(s)
530 unsigned char *s;
531 {
532 int i;
533 int len;
534
535 if (IS_UTF8_INVALID(s[0]))
536 return (0);
537
538 len = utf_len((char) s[0]);
539 if (len == 1)
540 return (1);
541 if (len == 2)
542 {
543 if (s[0] < 0xC2)
544 return (0);
545 } else
546 {
547 unsigned char mask;
548 mask = (~((1 << (8-len)) - 1)) & 0xFF;
549 if (s[0] == mask && (s[1] & mask) == 0x80)
550 return (0);
551 }
552
553 for (i = 1; i < len; i++)
554 if (!IS_UTF8_TRAIL(s[i]))
555 return (0);
556 return (1);
557 }
558
559 /*
560 * Get the value of a UTF-8 character.
561 */
562 public LWCHAR
563 get_wchar(p)
564 char *p;
565 {
566 switch (utf_len(p[0]))
567 {
568 case 1:
569 default:
570 /* 0xxxxxxx */
571 return (LWCHAR)
572 (p[0] & 0xFF);
573 case 2:
574 /* 110xxxxx 10xxxxxx */
575 return (LWCHAR) (
576 ((p[0] & 0x1F) << 6) |
577 (p[1] & 0x3F));
578 case 3:
579 /* 1110xxxx 10xxxxxx 10xxxxxx */
580 return (LWCHAR) (
581 ((p[0] & 0x0F) << 12) |
582 ((p[1] & 0x3F) << 6) |
583 (p[2] & 0x3F));
584 case 4:
585 /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
586 return (LWCHAR) (
587 ((p[0] & 0x07) << 18) |
588 ((p[1] & 0x3F) << 12) |
589 ((p[2] & 0x3F) << 6) |
590 (p[3] & 0x3F));
591 case 5:
592 /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
593 return (LWCHAR) (
594 ((p[0] & 0x03) << 24) |
595 ((p[1] & 0x3F) << 18) |
596 ((p[2] & 0x3F) << 12) |
597 ((p[3] & 0x3F) << 6) |
598 (p[4] & 0x3F));
599 case 6:
600 /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
601 return (LWCHAR) (
602 ((p[0] & 0x01) << 30) |
603 ((p[1] & 0x3F) << 24) |
604 ((p[2] & 0x3F) << 18) |
605 ((p[3] & 0x3F) << 12) |
606 ((p[4] & 0x3F) << 6) |
607 (p[5] & 0x3F));
608 }
609 }
610
611 /*
612 * Store a character into a UTF-8 string.
613 */
614 public void
615 put_wchar(pp, ch)
616 char **pp;
617 LWCHAR ch;
618 {
619 if (!utf_mode || ch < 0x80)
620 {
621 /* 0xxxxxxx */
622 *(*pp)++ = (char) ch;
623 } else if (ch < 0x800)
624 {
625 /* 110xxxxx 10xxxxxx */
626 *(*pp)++ = (char) (0xC0 | ((ch >> 6) & 0x1F));
627 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
628 } else if (ch < 0x10000)
629 {
630 /* 1110xxxx 10xxxxxx 10xxxxxx */
631 *(*pp)++ = (char) (0xE0 | ((ch >> 12) & 0x0F));
632 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
633 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
634 } else if (ch < 0x200000)
635 {
636 /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
637 *(*pp)++ = (char) (0xF0 | ((ch >> 18) & 0x07));
638 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
639 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
640 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
641 } else if (ch < 0x4000000)
642 {
643 /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
644 *(*pp)++ = (char) (0xF0 | ((ch >> 24) & 0x03));
645 *(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
646 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
647 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
648 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
649 } else
650 {
651 /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
652 *(*pp)++ = (char) (0xF0 | ((ch >> 30) & 0x01));
653 *(*pp)++ = (char) (0x80 | ((ch >> 24) & 0x3F));
654 *(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
655 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
656 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
657 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
658 }
659 }
660
661 /*
662 * Step forward or backward one character in a string.
663 */
664 public LWCHAR
665 step_char(pp, dir, limit)
666 char **pp;
667 signed int dir;
668 char *limit;
669 {
670 LWCHAR ch;
671 int len;
672 char *p = *pp;
673
674 if (!utf_mode)
675 {
676 /* It's easy if chars are one byte. */
677 if (dir > 0)
678 ch = (LWCHAR) ((p < limit) ? *p++ : 0);
679 else
680 ch = (LWCHAR) ((p > limit) ? *--p : 0);
681 } else if (dir > 0)
682 {
683 len = utf_len(*p);
684 if (p + len > limit)
685 {
686 ch = 0;
687 p = limit;
688 } else
689 {
690 ch = get_wchar(p);
691 p += len;
692 }
693 } else
694 {
695 while (p > limit && IS_UTF8_TRAIL(p[-1]))
696 p--;
697 if (p > limit)
698 ch = get_wchar(--p);
699 else
700 ch = 0;
701 }
702 *pp = p;
703 return ch;
704 }
705
706 /*
707 * Unicode characters data
708 */
709 struct wchar_range { LWCHAR first, last; };
710
711 /*
712 * Characters with general category values
713 * Mn: Mark, Nonspacing
714 * Me: Mark, Enclosing
715 * Last synched with
716 * <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
717 * dated 2005-11-30T00:58:48Z
718 */
719 static struct wchar_range comp_table[] = {
720 { 0x0300, 0x036F} /* Mn */, { 0x0483, 0x0486} /* Mn */,
721 { 0x0488, 0x0489} /* Me */,
722 { 0x0591, 0x05BD} /* Mn */, { 0x05BF, 0x05BF} /* Mn */,
723 { 0x05C1, 0x05C2} /* Mn */, { 0x05C4, 0x05C5} /* Mn */,
724 { 0x05C7, 0x05C7} /* Mn */, { 0x0610, 0x0615} /* Mn */,
725 { 0x064B, 0x065E} /* Mn */, { 0x0670, 0x0670} /* Mn */,
726 { 0x06D6, 0x06DC} /* Mn */,
727 { 0x06DE, 0x06DE} /* Me */,
728 { 0x06DF, 0x06E4} /* Mn */, { 0x06E7, 0x06E8} /* Mn */,
729 { 0x06EA, 0x06ED} /* Mn */, { 0x0711, 0x0711} /* Mn */,
730 { 0x0730, 0x074A} /* Mn */, { 0x07A6, 0x07B0} /* Mn */,
731 { 0x07EB, 0x07F3} /* Mn */, { 0x0901, 0x0902} /* Mn */,
732 { 0x093C, 0x093C} /* Mn */, { 0x0941, 0x0948} /* Mn */,
733 { 0x094D, 0x094D} /* Mn */, { 0x0951, 0x0954} /* Mn */,
734 { 0x0962, 0x0963} /* Mn */, { 0x0981, 0x0981} /* Mn */,
735 { 0x09BC, 0x09BC} /* Mn */, { 0x09C1, 0x09C4} /* Mn */,
736 { 0x09CD, 0x09CD} /* Mn */, { 0x09E2, 0x09E3} /* Mn */,
737 { 0x0A01, 0x0A02} /* Mn */, { 0x0A3C, 0x0A3C} /* Mn */,
738 { 0x0A41, 0x0A42} /* Mn */, { 0x0A47, 0x0A48} /* Mn */,
739 { 0x0A4B, 0x0A4D} /* Mn */, { 0x0A70, 0x0A71} /* Mn */,
740 { 0x0A81, 0x0A82} /* Mn */, { 0x0ABC, 0x0ABC} /* Mn */,
741 { 0x0AC1, 0x0AC5} /* Mn */, { 0x0AC7, 0x0AC8} /* Mn */,
742 { 0x0ACD, 0x0ACD} /* Mn */, { 0x0AE2, 0x0AE3} /* Mn */,
743 { 0x0B01, 0x0B01} /* Mn */, { 0x0B3C, 0x0B3C} /* Mn */,
744 { 0x0B3F, 0x0B3F} /* Mn */, { 0x0B41, 0x0B43} /* Mn */,
745 { 0x0B4D, 0x0B4D} /* Mn */, { 0x0B56, 0x0B56} /* Mn */,
746 { 0x0B82, 0x0B82} /* Mn */, { 0x0BC0, 0x0BC0} /* Mn */,
747 { 0x0BCD, 0x0BCD} /* Mn */, { 0x0C3E, 0x0C40} /* Mn */,
748 { 0x0C46, 0x0C48} /* Mn */, { 0x0C4A, 0x0C4D} /* Mn */,
749 { 0x0C55, 0x0C56} /* Mn */, { 0x0CBC, 0x0CBC} /* Mn */,
750 { 0x0CBF, 0x0CBF} /* Mn */, { 0x0CC6, 0x0CC6} /* Mn */,
751 { 0x0CCC, 0x0CCD} /* Mn */, { 0x0CE2, 0x0CE3} /* Mn */,
752 { 0x0D41, 0x0D43} /* Mn */, { 0x0D4D, 0x0D4D} /* Mn */,
753 { 0x0DCA, 0x0DCA} /* Mn */, { 0x0DD2, 0x0DD4} /* Mn */,
754 { 0x0DD6, 0x0DD6} /* Mn */, { 0x0E31, 0x0E31} /* Mn */,
755 { 0x0E34, 0x0E3A} /* Mn */, { 0x0E47, 0x0E4E} /* Mn */,
756 { 0x0EB1, 0x0EB1} /* Mn */, { 0x0EB4, 0x0EB9} /* Mn */,
757 { 0x0EBB, 0x0EBC} /* Mn */, { 0x0EC8, 0x0ECD} /* Mn */,
758 { 0x0F18, 0x0F19} /* Mn */, { 0x0F35, 0x0F35} /* Mn */,
759 { 0x0F37, 0x0F37} /* Mn */, { 0x0F39, 0x0F39} /* Mn */,
760 { 0x0F71, 0x0F7E} /* Mn */, { 0x0F80, 0x0F84} /* Mn */,
761 { 0x0F86, 0x0F87} /* Mn */, { 0x0F90, 0x0F97} /* Mn */,
762 { 0x0F99, 0x0FBC} /* Mn */, { 0x0FC6, 0x0FC6} /* Mn */,
763 { 0x102D, 0x1030} /* Mn */, { 0x1032, 0x1032} /* Mn */,
764 { 0x1036, 0x1037} /* Mn */, { 0x1039, 0x1039} /* Mn */,
765 { 0x1058, 0x1059} /* Mn */, { 0x135F, 0x135F} /* Mn */,
766 { 0x1712, 0x1714} /* Mn */, { 0x1732, 0x1734} /* Mn */,
767 { 0x1752, 0x1753} /* Mn */, { 0x1772, 0x1773} /* Mn */,
768 { 0x17B7, 0x17BD} /* Mn */, { 0x17C6, 0x17C6} /* Mn */,
769 { 0x17C9, 0x17D3} /* Mn */, { 0x17DD, 0x17DD} /* Mn */,
770 { 0x180B, 0x180D} /* Mn */, { 0x18A9, 0x18A9} /* Mn */,
771 { 0x1920, 0x1922} /* Mn */, { 0x1927, 0x1928} /* Mn */,
772 { 0x1932, 0x1932} /* Mn */, { 0x1939, 0x193B} /* Mn */,
773 { 0x1A17, 0x1A18} /* Mn */, { 0x1B00, 0x1B03} /* Mn */,
774 { 0x1B34, 0x1B34} /* Mn */, { 0x1B36, 0x1B3A} /* Mn */,
775 { 0x1B3C, 0x1B3C} /* Mn */, { 0x1B42, 0x1B42} /* Mn */,
776 { 0x1B6B, 0x1B73} /* Mn */, { 0x1DC0, 0x1DCA} /* Mn */,
777 { 0x1DFE, 0x1DFF} /* Mn */, { 0x20D0, 0x20DC} /* Mn */,
778 { 0x20DD, 0x20E0} /* Me */,
779 { 0x20E1, 0x20E1} /* Mn */,
780 { 0x20E2, 0x20E4} /* Me */,
781 { 0x20E5, 0x20EF} /* Mn */, { 0x302A, 0x302F} /* Mn */,
782 { 0x3099, 0x309A} /* Mn */, { 0xA806, 0xA806} /* Mn */,
783 { 0xA80B, 0xA80B} /* Mn */, { 0xA825, 0xA826} /* Mn */,
784 { 0xFB1E, 0xFB1E} /* Mn */, { 0xFE00, 0xFE0F} /* Mn */,
785 { 0xFE20, 0xFE23} /* Mn */, { 0x10A01, 0x10A03} /* Mn */,
786 { 0x10A05, 0x10A06} /* Mn */, { 0x10A0C, 0x10A0F} /* Mn */,
787 { 0x10A38, 0x10A3A} /* Mn */, { 0x10A3F, 0x10A3F} /* Mn */,
788 { 0x1D167, 0x1D169} /* Mn */, { 0x1D17B, 0x1D182} /* Mn */,
789 { 0x1D185, 0x1D18B} /* Mn */, { 0x1D1AA, 0x1D1AD} /* Mn */,
790 { 0x1D242, 0x1D244} /* Mn */, { 0xE0100, 0xE01EF} /* Mn */,
791 };
792
793 /*
794 * Special pairs, not ranges.
795 */
796 static struct wchar_range comb_table[] = {
797 {0x0644,0x0622}, {0x0644,0x0623}, {0x0644,0x0625}, {0x0644,0x0627},
798 };
799
800 /*
801 * Characters with general category values
802 * Cc: Other, Control
803 * Cf: Other, Format
804 * Cs: Other, Surrogate
805 * Co: Other, Private Use
806 * Cn: Other, Not Assigned
807 * Zl: Separator, Line
808 * Zp: Separator, Paragraph
809 * Last synched with
810 * <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
811 * dated 2005-11-30T00:58:48Z
812 */
813 static struct wchar_range ubin_table[] = {
814 { 0x0000, 0x001F} /* Cc */, { 0x007F, 0x009F} /* Cc */,
815 #if 0
816 { 0x00AD, 0x00AD} /* Cf */,
817 #endif
818 { 0x0370, 0x0373} /* Cn */, { 0x0376, 0x0379} /* Cn */,
819 { 0x037F, 0x0383} /* Cn */, { 0x038B, 0x038B} /* Cn */,
820 { 0x038D, 0x038D} /* Cn */, { 0x03A2, 0x03A2} /* Cn */,
821 { 0x03CF, 0x03CF} /* Cn */, { 0x0487, 0x0487} /* Cn */,
822 { 0x0514, 0x0530} /* Cn */, { 0x0557, 0x0558} /* Cn */,
823 { 0x0560, 0x0560} /* Cn */, { 0x0588, 0x0588} /* Cn */,
824 { 0x058B, 0x0590} /* Cn */, { 0x05C8, 0x05CF} /* Cn */,
825 { 0x05EB, 0x05EF} /* Cn */, { 0x05F5, 0x05FF} /* Cn */,
826 #if 0
827 { 0x0600, 0x0603} /* Cf */,
828 #endif
829 { 0x0604, 0x060A} /* Cn */, { 0x0616, 0x061A} /* Cn */,
830 { 0x061C, 0x061D} /* Cn */, { 0x0620, 0x0620} /* Cn */,
831 { 0x063B, 0x063F} /* Cn */, { 0x065F, 0x065F} /* Cn */,
832 #if 0
833 { 0x06DD, 0x06DD} /* Cf */,
834 #endif
835 { 0x070E, 0x070E} /* Cn */,
836 #if 0
837 { 0x070F, 0x070F} /* Cf */,
838 #endif
839 { 0x074B, 0x074C} /* Cn */, { 0x076E, 0x077F} /* Cn */,
840 { 0x07B2, 0x07BF} /* Cn */, { 0x07FB, 0x0900} /* Cn */,
841 { 0x093A, 0x093B} /* Cn */, { 0x094E, 0x094F} /* Cn */,
842 { 0x0955, 0x0957} /* Cn */, { 0x0971, 0x097A} /* Cn */,
843 { 0x0980, 0x0980} /* Cn */, { 0x0984, 0x0984} /* Cn */,
844 { 0x098D, 0x098E} /* Cn */, { 0x0991, 0x0992} /* Cn */,
845 { 0x09A9, 0x09A9} /* Cn */, { 0x09B1, 0x09B1} /* Cn */,
846 { 0x09B3, 0x09B5} /* Cn */, { 0x09BA, 0x09BB} /* Cn */,
847 { 0x09C5, 0x09C6} /* Cn */, { 0x09C9, 0x09CA} /* Cn */,
848 { 0x09CF, 0x09D6} /* Cn */, { 0x09D8, 0x09DB} /* Cn */,
849 { 0x09DE, 0x09DE} /* Cn */, { 0x09E4, 0x09E5} /* Cn */,
850 { 0x09FB, 0x0A00} /* Cn */, { 0x0A04, 0x0A04} /* Cn */,
851 { 0x0A0B, 0x0A0E} /* Cn */, { 0x0A11, 0x0A12} /* Cn */,
852 { 0x0A29, 0x0A29} /* Cn */, { 0x0A31, 0x0A31} /* Cn */,
853 { 0x0A34, 0x0A34} /* Cn */, { 0x0A37, 0x0A37} /* Cn */,
854 { 0x0A3A, 0x0A3B} /* Cn */, { 0x0A3D, 0x0A3D} /* Cn */,
855 { 0x0A43, 0x0A46} /* Cn */, { 0x0A49, 0x0A4A} /* Cn */,
856 { 0x0A4E, 0x0A58} /* Cn */, { 0x0A5D, 0x0A5D} /* Cn */,
857 { 0x0A5F, 0x0A65} /* Cn */, { 0x0A75, 0x0A80} /* Cn */,
858 { 0x0A84, 0x0A84} /* Cn */, { 0x0A8E, 0x0A8E} /* Cn */,
859 { 0x0A92, 0x0A92} /* Cn */, { 0x0AA9, 0x0AA9} /* Cn */,
860 { 0x0AB1, 0x0AB1} /* Cn */, { 0x0AB4, 0x0AB4} /* Cn */,
861 { 0x0ABA, 0x0ABB} /* Cn */, { 0x0AC6, 0x0AC6} /* Cn */,
862 { 0x0ACA, 0x0ACA} /* Cn */, { 0x0ACE, 0x0ACF} /* Cn */,
863 { 0x0AD1, 0x0ADF} /* Cn */, { 0x0AE4, 0x0AE5} /* Cn */,
864 { 0x0AF0, 0x0AF0} /* Cn */, { 0x0AF2, 0x0B00} /* Cn */,
865 { 0x0B04, 0x0B04} /* Cn */, { 0x0B0D, 0x0B0E} /* Cn */,
866 { 0x0B11, 0x0B12} /* Cn */, { 0x0B29, 0x0B29} /* Cn */,
867 { 0x0B31, 0x0B31} /* Cn */, { 0x0B34, 0x0B34} /* Cn */,
868 { 0x0B3A, 0x0B3B} /* Cn */, { 0x0B44, 0x0B46} /* Cn */,
869 { 0x0B49, 0x0B4A} /* Cn */, { 0x0B4E, 0x0B55} /* Cn */,
870 { 0x0B58, 0x0B5B} /* Cn */, { 0x0B5E, 0x0B5E} /* Cn */,
871 { 0x0B62, 0x0B65} /* Cn */, { 0x0B72, 0x0B81} /* Cn */,
872 { 0x0B84, 0x0B84} /* Cn */, { 0x0B8B, 0x0B8D} /* Cn */,
873 { 0x0B91, 0x0B91} /* Cn */, { 0x0B96, 0x0B98} /* Cn */,
874 { 0x0B9B, 0x0B9B} /* Cn */, { 0x0B9D, 0x0B9D} /* Cn */,
875 { 0x0BA0, 0x0BA2} /* Cn */, { 0x0BA5, 0x0BA7} /* Cn */,
876 { 0x0BAB, 0x0BAD} /* Cn */, { 0x0BBA, 0x0BBD} /* Cn */,
877 { 0x0BC3, 0x0BC5} /* Cn */, { 0x0BC9, 0x0BC9} /* Cn */,
878 { 0x0BCE, 0x0BD6} /* Cn */, { 0x0BD8, 0x0BE5} /* Cn */,
879 { 0x0BFB, 0x0C00} /* Cn */, { 0x0C04, 0x0C04} /* Cn */,
880 { 0x0C0D, 0x0C0D} /* Cn */, { 0x0C11, 0x0C11} /* Cn */,
881 { 0x0C29, 0x0C29} /* Cn */, { 0x0C34, 0x0C34} /* Cn */,
882 { 0x0C3A, 0x0C3D} /* Cn */, { 0x0C45, 0x0C45} /* Cn */,
883 { 0x0C49, 0x0C49} /* Cn */, { 0x0C4E, 0x0C54} /* Cn */,
884 { 0x0C57, 0x0C5F} /* Cn */, { 0x0C62, 0x0C65} /* Cn */,
885 { 0x0C70, 0x0C81} /* Cn */, { 0x0C84, 0x0C84} /* Cn */,
886 { 0x0C8D, 0x0C8D} /* Cn */, { 0x0C91, 0x0C91} /* Cn */,
887 { 0x0CA9, 0x0CA9} /* Cn */, { 0x0CB4, 0x0CB4} /* Cn */,
888 { 0x0CBA, 0x0CBB} /* Cn */, { 0x0CC5, 0x0CC5} /* Cn */,
889 { 0x0CC9, 0x0CC9} /* Cn */, { 0x0CCE, 0x0CD4} /* Cn */,
890 { 0x0CD7, 0x0CDD} /* Cn */, { 0x0CDF, 0x0CDF} /* Cn */,
891 { 0x0CE4, 0x0CE5} /* Cn */, { 0x0CF0, 0x0CF0} /* Cn */,
892 { 0x0CF3, 0x0D01} /* Cn */, { 0x0D04, 0x0D04} /* Cn */,
893 { 0x0D0D, 0x0D0D} /* Cn */, { 0x0D11, 0x0D11} /* Cn */,
894 { 0x0D29, 0x0D29} /* Cn */, { 0x0D3A, 0x0D3D} /* Cn */,
895 { 0x0D44, 0x0D45} /* Cn */, { 0x0D49, 0x0D49} /* Cn */,
896 { 0x0D4E, 0x0D56} /* Cn */, { 0x0D58, 0x0D5F} /* Cn */,
897 { 0x0D62, 0x0D65} /* Cn */, { 0x0D70, 0x0D81} /* Cn */,
898 { 0x0D84, 0x0D84} /* Cn */, { 0x0D97, 0x0D99} /* Cn */,
899 { 0x0DB2, 0x0DB2} /* Cn */, { 0x0DBC, 0x0DBC} /* Cn */,
900 { 0x0DBE, 0x0DBF} /* Cn */, { 0x0DC7, 0x0DC9} /* Cn */,
901 { 0x0DCB, 0x0DCE} /* Cn */, { 0x0DD5, 0x0DD5} /* Cn */,
902 { 0x0DD7, 0x0DD7} /* Cn */, { 0x0DE0, 0x0DF1} /* Cn */,
903 { 0x0DF5, 0x0E00} /* Cn */, { 0x0E3B, 0x0E3E} /* Cn */,
904 { 0x0E5C, 0x0E80} /* Cn */, { 0x0E83, 0x0E83} /* Cn */,
905 { 0x0E85, 0x0E86} /* Cn */, { 0x0E89, 0x0E89} /* Cn */,
906 { 0x0E8B, 0x0E8C} /* Cn */, { 0x0E8E, 0x0E93} /* Cn */,
907 { 0x0E98, 0x0E98} /* Cn */, { 0x0EA0, 0x0EA0} /* Cn */,
908 { 0x0EA4, 0x0EA4} /* Cn */, { 0x0EA6, 0x0EA6} /* Cn */,
909 { 0x0EA8, 0x0EA9} /* Cn */, { 0x0EAC, 0x0EAC} /* Cn */,
910 { 0x0EBA, 0x0EBA} /* Cn */, { 0x0EBE, 0x0EBF} /* Cn */,
911 { 0x0EC5, 0x0EC5} /* Cn */, { 0x0EC7, 0x0EC7} /* Cn */,
912 { 0x0ECE, 0x0ECF} /* Cn */, { 0x0EDA, 0x0EDB} /* Cn */,
913 { 0x0EDE, 0x0EFF} /* Cn */, { 0x0F48, 0x0F48} /* Cn */,
914 { 0x0F6B, 0x0F70} /* Cn */, { 0x0F8C, 0x0F8F} /* Cn */,
915 { 0x0F98, 0x0F98} /* Cn */, { 0x0FBD, 0x0FBD} /* Cn */,
916 { 0x0FCD, 0x0FCE} /* Cn */, { 0x0FD2, 0x0FFF} /* Cn */,
917 { 0x1022, 0x1022} /* Cn */, { 0x1028, 0x1028} /* Cn */,
918 { 0x102B, 0x102B} /* Cn */, { 0x1033, 0x1035} /* Cn */,
919 { 0x103A, 0x103F} /* Cn */, { 0x105A, 0x109F} /* Cn */,
920 { 0x10C6, 0x10CF} /* Cn */, { 0x10FD, 0x10FF} /* Cn */,
921 { 0x115A, 0x115E} /* Cn */, { 0x11A3, 0x11A7} /* Cn */,
922 { 0x11FA, 0x11FF} /* Cn */, { 0x1249, 0x1249} /* Cn */,
923 { 0x124E, 0x124F} /* Cn */, { 0x1257, 0x1257} /* Cn */,
924 { 0x1259, 0x1259} /* Cn */, { 0x125E, 0x125F} /* Cn */,
925 { 0x1289, 0x1289} /* Cn */, { 0x128E, 0x128F} /* Cn */,
926 { 0x12B1, 0x12B1} /* Cn */, { 0x12B6, 0x12B7} /* Cn */,
927 { 0x12BF, 0x12BF} /* Cn */, { 0x12C1, 0x12C1} /* Cn */,
928 { 0x12C6, 0x12C7} /* Cn */, { 0x12D7, 0x12D7} /* Cn */,
929 { 0x1311, 0x1311} /* Cn */, { 0x1316, 0x1317} /* Cn */,
930 { 0x135B, 0x135E} /* Cn */, { 0x137D, 0x137F} /* Cn */,
931 { 0x139A, 0x139F} /* Cn */, { 0x13F5, 0x1400} /* Cn */,
932 { 0x1677, 0x167F} /* Cn */, { 0x169D, 0x169F} /* Cn */,
933 { 0x16F1, 0x16FF} /* Cn */, { 0x170D, 0x170D} /* Cn */,
934 { 0x1715, 0x171F} /* Cn */, { 0x1737, 0x173F} /* Cn */,
935 { 0x1754, 0x175F} /* Cn */, { 0x176D, 0x176D} /* Cn */,
936 { 0x1771, 0x1771} /* Cn */, { 0x1774, 0x177F} /* Cn */,
937 #if 0
938 { 0x17B4, 0x17B5} /* Cf */,
939 #endif
940 { 0x17DE, 0x17DF} /* Cn */, { 0x17EA, 0x17EF} /* Cn */,
941 { 0x17FA, 0x17FF} /* Cn */, { 0x180F, 0x180F} /* Cn */,
942 { 0x181A, 0x181F} /* Cn */, { 0x1878, 0x187F} /* Cn */,
943 { 0x18AA, 0x18FF} /* Cn */, { 0x191D, 0x191F} /* Cn */,
944 { 0x192C, 0x192F} /* Cn */, { 0x193C, 0x193F} /* Cn */,
945 { 0x1941, 0x1943} /* Cn */, { 0x196E, 0x196F} /* Cn */,
946 { 0x1975, 0x197F} /* Cn */, { 0x19AA, 0x19AF} /* Cn */,
947 { 0x19CA, 0x19CF} /* Cn */, { 0x19DA, 0x19DD} /* Cn */,
948 { 0x1A1C, 0x1A1D} /* Cn */, { 0x1A20, 0x1AFF} /* Cn */,
949 { 0x1B4C, 0x1B4F} /* Cn */, { 0x1B7D, 0x1CFF} /* Cn */,
950 { 0x1DCB, 0x1DFD} /* Cn */, { 0x1E9C, 0x1E9F} /* Cn */,
951 { 0x1EFA, 0x1EFF} /* Cn */, { 0x1F16, 0x1F17} /* Cn */,
952 { 0x1F1E, 0x1F1F} /* Cn */, { 0x1F46, 0x1F47} /* Cn */,
953 { 0x1F4E, 0x1F4F} /* Cn */, { 0x1F58, 0x1F58} /* Cn */,
954 { 0x1F5A, 0x1F5A} /* Cn */, { 0x1F5C, 0x1F5C} /* Cn */,
955 { 0x1F5E, 0x1F5E} /* Cn */, { 0x1F7E, 0x1F7F} /* Cn */,
956 { 0x1FB5, 0x1FB5} /* Cn */, { 0x1FC5, 0x1FC5} /* Cn */,
957 { 0x1FD4, 0x1FD5} /* Cn */, { 0x1FDC, 0x1FDC} /* Cn */,
958 { 0x1FF0, 0x1FF1} /* Cn */, { 0x1FF5, 0x1FF5} /* Cn */,
959 { 0x1FFF, 0x1FFF} /* Cn */,
960 { 0x200B, 0x200F} /* Cf */,
961 { 0x2028, 0x2028} /* Zl */,
962 { 0x2029, 0x2029} /* Zp */,
963 { 0x202A, 0x202E} /* Cf */,
964 { 0x2060, 0x2063} /* Cf */,
965 { 0x2064, 0x2069} /* Cn */,
966 { 0x206A, 0x206F} /* Cf */,
967 { 0x2072, 0x2073} /* Cn */, { 0x208F, 0x208F} /* Cn */,
968 { 0x2095, 0x209F} /* Cn */, { 0x20B6, 0x20CF} /* Cn */,
969 { 0x20F0, 0x20FF} /* Cn */, { 0x214F, 0x2152} /* Cn */,
970 { 0x2185, 0x218F} /* Cn */, { 0x23E8, 0x23FF} /* Cn */,
971 { 0x2427, 0x243F} /* Cn */, { 0x244B, 0x245F} /* Cn */,
972 { 0x269D, 0x269F} /* Cn */, { 0x26B3, 0x2700} /* Cn */,
973 { 0x2705, 0x2705} /* Cn */, { 0x270A, 0x270B} /* Cn */,
974 { 0x2728, 0x2728} /* Cn */, { 0x274C, 0x274C} /* Cn */,
975 { 0x274E, 0x274E} /* Cn */, { 0x2753, 0x2755} /* Cn */,
976 { 0x2757, 0x2757} /* Cn */, { 0x275F, 0x2760} /* Cn */,
977 { 0x2795, 0x2797} /* Cn */, { 0x27B0, 0x27B0} /* Cn */,
978 { 0x27BF, 0x27BF} /* Cn */, { 0x27CB, 0x27CF} /* Cn */,
979 { 0x27EC, 0x27EF} /* Cn */, { 0x2B1B, 0x2B1F} /* Cn */,
980 { 0x2B24, 0x2BFF} /* Cn */, { 0x2C2F, 0x2C2F} /* Cn */,
981 { 0x2C5F, 0x2C5F} /* Cn */, { 0x2C6D, 0x2C73} /* Cn */,
982 { 0x2C78, 0x2C7F} /* Cn */, { 0x2CEB, 0x2CF8} /* Cn */,
983 { 0x2D26, 0x2D2F} /* Cn */, { 0x2D66, 0x2D6E} /* Cn */,
984 { 0x2D70, 0x2D7F} /* Cn */, { 0x2D97, 0x2D9F} /* Cn */,
985 { 0x2DA7, 0x2DA7} /* Cn */, { 0x2DAF, 0x2DAF} /* Cn */,
986 { 0x2DB7, 0x2DB7} /* Cn */, { 0x2DBF, 0x2DBF} /* Cn */,
987 { 0x2DC7, 0x2DC7} /* Cn */, { 0x2DCF, 0x2DCF} /* Cn */,
988 { 0x2DD7, 0x2DD7} /* Cn */, { 0x2DDF, 0x2DFF} /* Cn */,
989 { 0x2E18, 0x2E1B} /* Cn */, { 0x2E1E, 0x2E7F} /* Cn */,
990 { 0x2E9A, 0x2E9A} /* Cn */, { 0x2EF4, 0x2EFF} /* Cn */,
991 { 0x2FD6, 0x2FEF} /* Cn */, { 0x2FFC, 0x2FFF} /* Cn */,
992 { 0x3040, 0x3040} /* Cn */, { 0x3097, 0x3098} /* Cn */,
993 { 0x3100, 0x3104} /* Cn */, { 0x312D, 0x3130} /* Cn */,
994 { 0x318F, 0x318F} /* Cn */, { 0x31B8, 0x31BF} /* Cn */,
995 { 0x31D0, 0x31EF} /* Cn */, { 0x321F, 0x321F} /* Cn */,
996 { 0x3244, 0x324F} /* Cn */, { 0x32FF, 0x32FF} /* Cn */,
997 { 0x4DB6, 0x4DBF} /* Cn */, { 0x9FBC, 0x9FFF} /* Cn */,
998 { 0xA48D, 0xA48F} /* Cn */, { 0xA4C7, 0xA6FF} /* Cn */,
999 { 0xA71B, 0xA71F} /* Cn */, { 0xA722, 0xA7FF} /* Cn */,
1000 { 0xA82C, 0xA83F} /* Cn */, { 0xA878, 0xABFF} /* Cn */,
1001 { 0xD7A4, 0xD7FF} /* Cn */,
1002 { 0xD800, 0xDFFF} /* Cs */,
1003 { 0xE000, 0xF8FF} /* Co */,
1004 { 0xFA2E, 0xFA2F} /* Cn */, { 0xFA6B, 0xFA6F} /* Cn */,
1005 { 0xFADA, 0xFAFF} /* Cn */, { 0xFB07, 0xFB12} /* Cn */,
1006 { 0xFB18, 0xFB1C} /* Cn */, { 0xFB37, 0xFB37} /* Cn */,
1007 { 0xFB3D, 0xFB3D} /* Cn */, { 0xFB3F, 0xFB3F} /* Cn */,
1008 { 0xFB42, 0xFB42} /* Cn */, { 0xFB45, 0xFB45} /* Cn */,
1009 { 0xFBB2, 0xFBD2} /* Cn */, { 0xFD40, 0xFD4F} /* Cn */,
1010 { 0xFD90, 0xFD91} /* Cn */, { 0xFDC8, 0xFDEF} /* Cn */,
1011 { 0xFDFE, 0xFDFF} /* Cn */, { 0xFE1A, 0xFE1F} /* Cn */,
1012 { 0xFE24, 0xFE2F} /* Cn */, { 0xFE53, 0xFE53} /* Cn */,
1013 { 0xFE67, 0xFE67} /* Cn */, { 0xFE6C, 0xFE6F} /* Cn */,
1014 { 0xFE75, 0xFE75} /* Cn */, { 0xFEFD, 0xFEFE} /* Cn */,
1015 { 0xFEFF, 0xFEFF} /* Cf */,
1016 { 0xFF00, 0xFF00} /* Cn */, { 0xFFBF, 0xFFC1} /* Cn */,
1017 { 0xFFC8, 0xFFC9} /* Cn */, { 0xFFD0, 0xFFD1} /* Cn */,
1018 { 0xFFD8, 0xFFD9} /* Cn */, { 0xFFDD, 0xFFDF} /* Cn */,
1019 { 0xFFE7, 0xFFE7} /* Cn */, { 0xFFEF, 0xFFF8} /* Cn */,
1020 { 0xFFF9, 0xFFFB} /* Cf */,
1021 { 0xFFFE, 0xFFFF} /* Cn */, { 0x1000C, 0x1000C} /* Cn */,
1022 { 0x10027, 0x10027} /* Cn */, { 0x1003B, 0x1003B} /* Cn */,
1023 { 0x1003E, 0x1003E} /* Cn */, { 0x1004E, 0x1004F} /* Cn */,
1024 { 0x1005E, 0x1007F} /* Cn */, { 0x100FB, 0x100FF} /* Cn */,
1025 { 0x10103, 0x10106} /* Cn */, { 0x10134, 0x10136} /* Cn */,
1026 { 0x1018B, 0x102FF} /* Cn */, { 0x1031F, 0x1031F} /* Cn */,
1027 { 0x10324, 0x1032F} /* Cn */, { 0x1034B, 0x1037F} /* Cn */,
1028 { 0x1039E, 0x1039E} /* Cn */, { 0x103C4, 0x103C7} /* Cn */,
1029 { 0x103D6, 0x103FF} /* Cn */,
1030 { 0x1049E, 0x1049F} /* Cn */, { 0x104AA, 0x107FF} /* Cn */,
1031 { 0x10806, 0x10807} /* Cn */, { 0x10809, 0x10809} /* Cn */,
1032 { 0x10836, 0x10836} /* Cn */, { 0x10839, 0x1083B} /* Cn */,
1033 { 0x1083D, 0x1083E} /* Cn */, { 0x10840, 0x108FF} /* Cn */,
1034 { 0x1091A, 0x1091E} /* Cn */, { 0x10920, 0x109FF} /* Cn */,
1035 { 0x10A04, 0x10A04} /* Cn */, { 0x10A07, 0x10A0B} /* Cn */,
1036 { 0x10A14, 0x10A14} /* Cn */, { 0x10A18, 0x10A18} /* Cn */,
1037 { 0x10A34, 0x10A37} /* Cn */, { 0x10A3B, 0x10A3E} /* Cn */,
1038 { 0x10A48, 0x10A4F} /* Cn */, { 0x10A59, 0x11FFF} /* Cn */,
1039 { 0x1236F, 0x123FF} /* Cn */, { 0x12463, 0x1246F} /* Cn */,
1040 { 0x12474, 0x1CFFF} /* Cn */, { 0x1D0F6, 0x1D0FF} /* Cn */,
1041 { 0x1D127, 0x1D129} /* Cn */,
1042 { 0x1D173, 0x1D17A} /* Cf */,
1043 { 0x1D1DE, 0x1D1FF} /* Cn */, { 0x1D246, 0x1D2FF} /* Cn */,
1044 { 0x1D357, 0x1D35F} /* Cn */, { 0x1D372, 0x1D3FF} /* Cn */,
1045 { 0x1D455, 0x1D455} /* Cn */, { 0x1D49D, 0x1D49D} /* Cn */,
1046 { 0x1D4A0, 0x1D4A1} /* Cn */, { 0x1D4A3, 0x1D4A4} /* Cn */,
1047 { 0x1D4A7, 0x1D4A8} /* Cn */, { 0x1D4AD, 0x1D4AD} /* Cn */,
1048 { 0x1D4BA, 0x1D4BA} /* Cn */, { 0x1D4BC, 0x1D4BC} /* Cn */,
1049 { 0x1D4C4, 0x1D4C4} /* Cn */, { 0x1D506, 0x1D506} /* Cn */,
1050 { 0x1D50B, 0x1D50C} /* Cn */, { 0x1D515, 0x1D515} /* Cn */,
1051 { 0x1D51D, 0x1D51D} /* Cn */, { 0x1D53A, 0x1D53A} /* Cn */,
1052 { 0x1D53F, 0x1D53F} /* Cn */, { 0x1D545, 0x1D545} /* Cn */,
1053 { 0x1D547, 0x1D549} /* Cn */, { 0x1D551, 0x1D551} /* Cn */,
1054 { 0x1D6A6, 0x1D6A7} /* Cn */, { 0x1D7CC, 0x1D7CD} /* Cn */,
1055 { 0x1D800, 0x1FFFF} /* Cn */, { 0x2A6D7, 0x2F7FF} /* Cn */,
1056 { 0x2FA1E, 0xE0000} /* Cn */,
1057 { 0xE0001, 0xE0001} /* Cf */,
1058 { 0xE0002, 0xE001F} /* Cn */,
1059 { 0xE0020, 0xE007F} /* Cf */,
1060 { 0xE0080, 0xE00FF} /* Cn */, { 0xE01F0, 0xEFFFF} /* Cn */,
1061 { 0xF0000, 0xFFFFD} /* Co */,
1062 { 0xFFFFE, 0xFFFFF} /* Cn */,
1063 {0x100000,0x10FFFD} /* Co */,
1064 {0x10FFFE,0x10FFFF} /* Cn */,
1065 {0x110000,0x7FFFFFFF} /* ISO 10646?? */
1066 };
1067
1068 /*
1069 * Double width characters
1070 * W: East Asian Wide
1071 * F: East Asian Full-width
1072 * Unassigned code points may be included when they allow ranges to be merged.
1073 * Last synched with
1074 * <http://www.unicode.org/Public/5.0.0/ucd/EastAsianWidth-5.0.0d2.txt>
1075 * dated 2005-11-08T01:32:56Z
1076 */
1077 static struct wchar_range wide_table[] = {
1078 { 0x1100, 0x115F} /* W */, { 0x2329, 0x232A} /* W */,
1079 { 0x2E80, 0x2FFB} /* W */,
1080 { 0x3000, 0x3000} /* F */,
1081 { 0x3001, 0x303E} /* W */, { 0x3041, 0x4DB5} /* W */,
1082 { 0x4E00, 0x9FBB} /* W */, { 0xA000, 0xA4C6} /* W */,
1083 { 0xAC00, 0xD7A3} /* W */, { 0xF900, 0xFAD9} /* W */,
1084 { 0xFE10, 0xFE19} /* W */, { 0xFE30, 0xFE6B} /* W */,
1085 { 0xFF01, 0xFF60} /* F */, { 0xFFE0, 0xFFE6} /* F */,
1086 { 0x20000, 0x2FFFD} /* W */, { 0x30000, 0x3FFFD} /* W */,
1087 };
1088
1089 static int
1090 is_in_table(ch, table, tsize)
1091 LWCHAR ch;
1092 struct wchar_range table[];
1093 int tsize;
1094 {
1095 int hi;
1096 int lo;
1097
1098 /* Binary search in the table. */
1099 if (ch < table[0].first)
1100 return 0;
1101 lo = 0;
1102 hi = tsize - 1;
1103 while (lo <= hi)
1104 {
1105 int mid = (lo + hi) / 2;
1106 if (ch > table[mid].last)
1107 lo = mid + 1;
1108 else if (ch < table[mid].first)
1109 hi = mid - 1;
1110 else
1111 return 1;
1112 }
1113 return 0;
1114 }
1115
1116 /*
1117 * Is a character a UTF-8 composing character?
1118 * If a composing character follows any char, the two combine into one glyph.
1119 */
1120 public int
1121 is_composing_char(ch)
1122 LWCHAR ch;
1123 {
1124 return is_in_table(ch, comp_table, (sizeof(comp_table) / sizeof(*comp_table)));
1125 }
1126
1127 /*
1128 * Should this UTF-8 character be treated as binary?
1129 */
1130 public int
1131 is_ubin_char(ch)
1132 LWCHAR ch;
1133 {
1134 return is_in_table(ch, ubin_table, (sizeof(ubin_table) / sizeof(*ubin_table)));
1135 }
1136
1137 /*
1138 * Is this a double width UTF-8 character?
1139 */
1140 public int
1141 is_wide_char(ch)
1142 LWCHAR ch;
1143 {
1144 return is_in_table(ch, wide_table, (sizeof(wide_table) / sizeof(*wide_table)));
1145 }
1146
1147 /*
1148 * Is a character a UTF-8 combining character?
1149 * A combining char acts like an ordinary char, but if it follows
1150 * a specific char (not any char), the two combine into one glyph.
1151 */
1152 public int
1153 is_combining_char(ch1, ch2)
1154 LWCHAR ch1;
1155 LWCHAR ch2;
1156 {
1157 /* The table is small; use linear search. */
1158 int i;
1159 for (i = 0; i < sizeof(comb_table)/sizeof(*comb_table); i++)
1160 {
1161 if (ch1 == comb_table[i].first &&
1162 ch2 == comb_table[i].last)
1163 return 1;
1164 }
1165 return 0;
1166 }
1167
DragonFly BSD