xref: /freebsd/contrib/less/charset.c (revision 0fca6ea1d4eea4c934cfff25ac9ee8ad6fe95583)
1 /*
2  * Copyright (C) 1984-2023  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, see the README file.
8  */
9 
10 
11 /*
12  * Functions to define the character set
13  * and do things specific to the character set.
14  */
15 
16 #include "less.h"
17 #if HAVE_LOCALE
18 #include <locale.h>
19 #include <ctype.h>
20 #include <langinfo.h>
21 #endif
22 
23 #include "charset.h"
24 #include "xbuf.h"
25 
26 #if MSDOS_COMPILER==WIN32C
27 #define WIN32_LEAN_AND_MEAN
28 #include <windows.h>
29 #endif
30 
31 extern int bs_mode;
32 
33 public int utf_mode = 0;
34 
35 /*
36  * Predefined character sets,
37  * selected by the LESSCHARSET environment variable.
38  */
39 struct charset {
40 	char *name;
41 	int *p_flag;
42 	char *desc;
43 } charsets[] = {
44 		{ "ascii",              NULL,       "8bcccbcc18b95.b" },
45 		{ "utf-8",              &utf_mode,  "8bcccbcc18b95.b126.bb" },
46 		{ "iso8859",            NULL,       "8bcccbcc18b95.33b." },
47 		{ "latin3",             NULL,       "8bcccbcc18b95.33b5.b8.b15.b4.b12.b18.b12.b." },
48 		{ "arabic",             NULL,       "8bcccbcc18b95.33b.3b.7b2.13b.3b.b26.5b19.b" },
49 		{ "greek",              NULL,       "8bcccbcc18b95.33b4.2b4.b3.b35.b44.b" },
50 		{ "greek2005",          NULL,       "8bcccbcc18b95.33b14.b35.b44.b" },
51 		{ "hebrew",             NULL,       "8bcccbcc18b95.33b.b29.32b28.2b2.b" },
52 		{ "koi8-r",             NULL,       "8bcccbcc18b95.b." },
53 		{ "KOI8-T",             NULL,       "8bcccbcc18b95.b8.b6.b8.b.b.5b7.3b4.b4.b3.b.b.3b." },
54 		{ "georgianps",         NULL,       "8bcccbcc18b95.3b11.4b12.2b." },
55 		{ "tcvn",               NULL,       "b..b...bcccbccbbb7.8b95.b48.5b." },
56 		{ "TIS-620",            NULL,       "8bcccbcc18b95.b.4b.11b7.8b." },
57 		{ "next",               NULL,       "8bcccbcc18b95.bb125.bb" },
58 		{ "dos",                NULL,       "8bcccbcc12bc5b95.b." },
59 		{ "windows-1251",       NULL,       "8bcccbcc12bc5b95.b24.b." },
60 		{ "windows-1252",       NULL,       "8bcccbcc12bc5b95.b.b11.b.2b12.b." },
61 		{ "windows-1255",       NULL,       "8bcccbcc12bc5b95.b.b8.b.5b9.b.4b." },
62 		{ "ebcdic",             NULL,       "5bc6bcc7bcc41b.9b7.9b5.b..8b6.10b6.b9.7b9.8b8.17b3.3b9.7b9.8b8.6b10.b.b.b." },
63 		{ "IBM-1047",           NULL,       "4cbcbc3b9cbccbccbb4c6bcc5b3cbbc4bc4bccbc191.b" },
64 		{ NULL, NULL, NULL }
65 };
66 
67 /*
68  * Support "locale charmap"/nl_langinfo(CODESET) values, as well as others.
69  */
70 struct cs_alias {
71 	char *name;
72 	char *oname;
73 } cs_aliases[] = {
74 	{ "UTF-8",              "utf-8" },
75 	{ "utf8",               "utf-8" },
76 	{ "UTF8",               "utf-8" },
77 	{ "ANSI_X3.4-1968",     "ascii" },
78 	{ "US-ASCII",           "ascii" },
79 	{ "latin1",             "iso8859" },
80 	{ "ISO-8859-1",         "iso8859" },
81 	{ "latin9",             "iso8859" },
82 	{ "ISO-8859-15",        "iso8859" },
83 	{ "latin2",             "iso8859" },
84 	{ "ISO-8859-2",         "iso8859" },
85 	{ "ISO-8859-3",         "latin3" },
86 	{ "latin4",             "iso8859" },
87 	{ "ISO-8859-4",         "iso8859" },
88 	{ "cyrillic",           "iso8859" },
89 	{ "ISO-8859-5",         "iso8859" },
90 	{ "ISO-8859-6",         "arabic" },
91 	{ "ISO-8859-7",         "greek" },
92 	{ "IBM9005",            "greek2005" },
93 	{ "ISO-8859-8",         "hebrew" },
94 	{ "latin5",             "iso8859" },
95 	{ "ISO-8859-9",         "iso8859" },
96 	{ "latin6",             "iso8859" },
97 	{ "ISO-8859-10",        "iso8859" },
98 	{ "latin7",             "iso8859" },
99 	{ "ISO-8859-13",        "iso8859" },
100 	{ "latin8",             "iso8859" },
101 	{ "ISO-8859-14",        "iso8859" },
102 	{ "latin10",            "iso8859" },
103 	{ "ISO-8859-16",        "iso8859" },
104 	{ "IBM437",             "dos" },
105 	{ "EBCDIC-US",          "ebcdic" },
106 	{ "IBM1047",            "IBM-1047" },
107 	{ "KOI8-R",             "koi8-r" },
108 	{ "KOI8-U",             "koi8-r" },
109 	{ "GEORGIAN-PS",        "georgianps" },
110 	{ "TCVN5712-1",         "tcvn" },
111 	{ "NEXTSTEP",           "next" },
112 	{ "windows",            "windows-1252" }, /* backward compatibility */
113 	{ "CP1251",             "windows-1251" },
114 	{ "CP1252",             "windows-1252" },
115 	{ "CP1255",             "windows-1255" },
116 	{ NULL, NULL }
117 };
118 
119 #define IS_BINARY_CHAR  01
120 #define IS_CONTROL_CHAR 02
121 
122 static char chardef[256];
123 static char *binfmt = NULL;
124 static char *utfbinfmt = NULL;
125 public int binattr = AT_STANDOUT|AT_COLOR_BIN;
126 
127 static struct xbuffer user_wide_array;
128 static struct xbuffer user_ubin_array;
129 static struct xbuffer user_compose_array;
130 static struct xbuffer user_prt_array;
131 static struct wchar_range_table user_wide_table;
132 static struct wchar_range_table user_ubin_table;
133 static struct wchar_range_table user_compose_table;
134 static struct wchar_range_table user_prt_table;
135 
136 /*
137  * Set a wchar_range_table to the table in an xbuffer.
138  */
139 static void wchar_range_table_set(struct wchar_range_table *tbl, struct xbuffer *arr)
140 {
141 	tbl->table = (struct wchar_range *) arr->data;
142 	tbl->count = arr->end / sizeof(struct wchar_range);
143 }
144 
145 /*
146  * Skip over a "U" or "U+" prefix before a hex codepoint.
147  */
148 static char * skip_uprefix(char *s)
149 {
150 	if (*s == 'U' || *s == 'u')
151 		if (*++s == '+') ++s;
152 	return s;
153 }
154 
155 /*
156  * Parse a dash-separated range of hex values.
157  */
158 static void wchar_range_get(char **ss, struct wchar_range *range)
159 {
160 	char *s = skip_uprefix(*ss);
161 	range->first = lstrtoul(s, &s, 16);
162 	if (s[0] == '-')
163 	{
164 		s = skip_uprefix(&s[1]);
165 		range->last = lstrtoul(s, &s, 16);
166 	} else
167 	{
168 		range->last = range->first;
169 	}
170 	*ss = s;
171 }
172 
173 /*
174  * Parse the LESSUTFCHARDEF variable.
175  */
176 static void ichardef_utf(char *s)
177 {
178 	xbuf_init(&user_wide_array);
179 	xbuf_init(&user_ubin_array);
180 	xbuf_init(&user_compose_array);
181 	xbuf_init(&user_prt_array);
182 
183 	if (s != NULL)
184 	{
185 		while (s[0] != '\0')
186 		{
187 			struct wchar_range range;
188 			wchar_range_get(&s, &range);
189 			if (range.last == 0)
190 			{
191 				error("invalid hex number(s) in LESSUTFCHARDEF", NULL_PARG);
192 				quit(QUIT_ERROR);
193 			}
194 			if (*s++ != ':')
195 			{
196 				error("missing colon in LESSUTFCHARDEF", NULL_PARG);
197 				quit(QUIT_ERROR);
198 			}
199 			switch (*s++)
200 			{
201 			case 'b':
202 				xbuf_add_data(&user_ubin_array, (unsigned char *) &range, sizeof(range));
203 				break;
204 			case 'c':
205 				xbuf_add_data(&user_compose_array, (unsigned char *) &range, sizeof(range));
206 				break;
207 			case 'w':
208 				xbuf_add_data(&user_wide_array, (unsigned char *) &range, sizeof(range));
209 				xbuf_add_data(&user_prt_array, (unsigned char *) &range, sizeof(range));
210 				break;
211 			case 'p': case '.':
212 				xbuf_add_data(&user_prt_array, (unsigned char *) &range, sizeof(range));
213 				break;
214 			case '\0':
215 				s--;
216 				break;
217 			default:
218 				/* Ignore unknown character attribute. */
219 				break;
220 			}
221 			if (s[0] == ',') ++s;
222 		}
223 	}
224 	wchar_range_table_set(&user_wide_table, &user_wide_array);
225 	wchar_range_table_set(&user_ubin_table, &user_ubin_array);
226 	wchar_range_table_set(&user_compose_table, &user_compose_array);
227 	wchar_range_table_set(&user_prt_table, &user_prt_array);
228 }
229 
230 /*
231  * Define a charset, given a description string.
232  * The string consists of 256 letters,
233  * one for each character in the charset.
234  * If the string is shorter than 256 letters, missing letters
235  * are taken to be identical to the last one.
236  * A decimal number followed by a letter is taken to be a
237  * repetition of the letter.
238  *
239  * Each letter is one of:
240  *      . normal character
241  *      b binary character
242  *      c control character
243  */
244 static void ichardef(char *s)
245 {
246 	char *cp;
247 	int n;
248 	char v;
249 
250 	n = 0;
251 	v = 0;
252 	cp = chardef;
253 	while (*s != '\0')
254 	{
255 		switch (*s++)
256 		{
257 		case '.':
258 			v = 0;
259 			break;
260 		case 'c':
261 			v = IS_CONTROL_CHAR;
262 			break;
263 		case 'b':
264 			v = IS_BINARY_CHAR|IS_CONTROL_CHAR;
265 			break;
266 
267 		case '0': case '1': case '2': case '3': case '4':
268 		case '5': case '6': case '7': case '8': case '9':
269 			if (ckd_mul(&n, n, 10) || ckd_add(&n, n, s[-1] - '0'))
270 				goto invalid_chardef;
271 			continue;
272 
273 		default:
274 		invalid_chardef:
275 			error("invalid chardef", NULL_PARG);
276 			quit(QUIT_ERROR);
277 			/*NOTREACHED*/
278 		}
279 
280 		do
281 		{
282 			if (cp >= chardef + sizeof(chardef))
283 			{
284 				error("chardef longer than 256", NULL_PARG);
285 				quit(QUIT_ERROR);
286 				/*NOTREACHED*/
287 			}
288 			*cp++ = v;
289 		} while (--n > 0);
290 		n = 0;
291 	}
292 
293 	while (cp < chardef + sizeof(chardef))
294 		*cp++ = v;
295 }
296 
297 /*
298  * Define a charset, given a charset name.
299  * The valid charset names are listed in the "charsets" array.
300  */
301 static int icharset(char *name, int no_error)
302 {
303 	struct charset *p;
304 	struct cs_alias *a;
305 
306 	if (name == NULL || *name == '\0')
307 		return (0);
308 
309 	/* First see if the name is an alias. */
310 	for (a = cs_aliases;  a->name != NULL;  a++)
311 	{
312 		if (strcmp(name, a->name) == 0)
313 		{
314 			name = a->oname;
315 			break;
316 		}
317 	}
318 
319 	for (p = charsets;  p->name != NULL;  p++)
320 	{
321 		if (strcmp(name, p->name) == 0)
322 		{
323 			ichardef(p->desc);
324 			if (p->p_flag != NULL)
325 			{
326 #if MSDOS_COMPILER==WIN32C
327 				*(p->p_flag) = 1 + (GetConsoleOutputCP() != CP_UTF8);
328 #else
329 				*(p->p_flag) = 1;
330 #endif
331 			}
332 			return (1);
333 		}
334 	}
335 
336 	if (!no_error) {
337 		error("invalid charset name", NULL_PARG);
338 		quit(QUIT_ERROR);
339 	}
340 	return (0);
341 }
342 
343 #if HAVE_LOCALE
344 /*
345  * Define a charset, given a locale name.
346  */
347 static void ilocale(void)
348 {
349 	int c;
350 
351 	for (c = 0;  c < (int) sizeof(chardef);  c++)
352 	{
353 		if (isprint(c))
354 			chardef[c] = 0;
355 		else if (iscntrl(c))
356 			chardef[c] = IS_CONTROL_CHAR;
357 		else
358 			chardef[c] = IS_BINARY_CHAR|IS_CONTROL_CHAR;
359 	}
360 }
361 #endif
362 
363 /*
364  * Define the printing format for control (or binary utf) chars.
365  */
366 public void setfmt(char *s, char **fmtvarptr, int *attrptr, char *default_fmt, int for_printf)
367 {
368 	if (s && utf_mode)
369 	{
370 		/* It would be too hard to account for width otherwise.  */
371 		char constant *t = s;
372 		while (*t)
373 		{
374 			if (*t < ' ' || *t > '~')
375 			{
376 				s = default_fmt;
377 				goto attr;
378 			}
379 			t++;
380 		}
381 	}
382 
383 	if (s == NULL || *s == '\0')
384 		s = default_fmt;
385 	else if (for_printf &&
386 	    ((*s == '*' && (s[1] == '\0' || s[2] == '\0' || strchr(s + 2, 'n'))) ||
387 	     (*s != '*' && strchr(s, 'n'))))
388 		/* %n is evil */
389 		s = default_fmt;
390 
391 	/*
392 	 * Select the attributes if it starts with "*".
393 	 */
394  attr:
395 	if (*s == '*' && s[1] != '\0')
396 	{
397 		switch (s[1])
398 		{
399 		case 'd':  *attrptr = AT_BOLD;      break;
400 		case 'k':  *attrptr = AT_BLINK;     break;
401 		case 's':  *attrptr = AT_STANDOUT;  break;
402 		case 'u':  *attrptr = AT_UNDERLINE; break;
403 		default:   *attrptr = AT_NORMAL;    break;
404 		}
405 		s += 2;
406 	}
407 	*fmtvarptr = s;
408 }
409 
410 /*
411  *
412  */
413 static void set_charset(void)
414 {
415 	char *s;
416 
417 #if MSDOS_COMPILER==WIN32C
418 	/*
419 	 * If the Windows console is using UTF-8, we'll use it too.
420 	 */
421 	if (GetConsoleOutputCP() == CP_UTF8)
422 		if (icharset("utf-8", 1))
423 			return;
424 #endif
425 
426 	ichardef_utf(lgetenv("LESSUTFCHARDEF"));
427 
428 	/*
429 	 * See if environment variable LESSCHARSET is defined.
430 	 */
431 	s = lgetenv("LESSCHARSET");
432 	if (icharset(s, 0))
433 		return;
434 
435 	/*
436 	 * LESSCHARSET is not defined: try LESSCHARDEF.
437 	 */
438 	s = lgetenv("LESSCHARDEF");
439 	if (!isnullenv(s))
440 	{
441 		ichardef(s);
442 		return;
443 	}
444 
445 #if HAVE_LOCALE
446 #ifdef CODESET
447 	/*
448 	 * Try using the codeset name as the charset name.
449 	 */
450 	s = nl_langinfo(CODESET);
451 	if (icharset(s, 1))
452 		return;
453 #endif
454 #endif
455 
456 #if HAVE_STRSTR
457 	/*
458 	 * Check whether LC_ALL, LC_CTYPE or LANG look like UTF-8 is used.
459 	 */
460 	if ((s = lgetenv("LC_ALL")) != NULL ||
461 	    (s = lgetenv("LC_CTYPE")) != NULL ||
462 	    (s = lgetenv("LANG")) != NULL)
463 	{
464 		if (   strstr(s, "UTF-8") != NULL || strstr(s, "utf-8") != NULL
465 		    || strstr(s, "UTF8")  != NULL || strstr(s, "utf8")  != NULL)
466 			if (icharset("utf-8", 1))
467 				return;
468 	}
469 #endif
470 
471 #if HAVE_LOCALE
472 	/*
473 	 * Get character definitions from locale functions,
474 	 * rather than from predefined charset entry.
475 	 */
476 	ilocale();
477 #else
478 #if MSDOS_COMPILER
479 	/*
480 	 * Default to "dos".
481 	 */
482 	(void) icharset("dos", 1);
483 #else
484 	/*
485 	 * Default to "latin1".
486 	 */
487 	(void) icharset("latin1", 1);
488 #endif
489 #endif
490 }
491 
492 /*
493  * Initialize charset data structures.
494  */
495 public void init_charset(void)
496 {
497 	char *s;
498 
499 #if HAVE_LOCALE
500 	setlocale(LC_ALL, "");
501 #endif
502 
503 	set_charset();
504 
505 	s = lgetenv("LESSBINFMT");
506 	setfmt(s, &binfmt, &binattr, "*s<%02X>", TRUE);
507 
508 	s = lgetenv("LESSUTFBINFMT");
509 	setfmt(s, &utfbinfmt, &binattr, "<U+%04lX>", TRUE);
510 }
511 
512 /*
513  * Is a given character a "binary" character?
514  */
515 public int binary_char(LWCHAR c)
516 {
517 	if (utf_mode)
518 		return (is_ubin_char(c));
519 	c &= 0377;
520 	return (chardef[c] & IS_BINARY_CHAR);
521 }
522 
523 /*
524  * Is a given character a "control" character?
525  */
526 public int control_char(LWCHAR c)
527 {
528 	c &= 0377;
529 	return (chardef[c] & IS_CONTROL_CHAR);
530 }
531 
532 /*
533  * Return the printable form of a character.
534  * For example, in the "ascii" charset '\3' is printed as "^C".
535  */
536 public char * prchar(LWCHAR c)
537 {
538 	/* {{ This buffer can be overrun if LESSBINFMT is a long string. }} */
539 	static char buf[MAX_PRCHAR_LEN+1];
540 
541 	c &= 0377;
542 	if ((c < 128 || !utf_mode) && !control_char(c))
543 		SNPRINTF1(buf, sizeof(buf), "%c", (int) c);
544 	else if (c == ESC)
545 		strcpy(buf, "ESC");
546 #if IS_EBCDIC_HOST
547 	else if (!binary_char(c) && c < 64)
548 		SNPRINTF1(buf, sizeof(buf), "^%c",
549 		/*
550 		 * This array roughly inverts CONTROL() #defined in less.h,
551 		 * and should be kept in sync with CONTROL() and IBM-1047.
552 		 */
553 		"@ABC.I.?...KLMNO"
554 		"PQRS.JH.XY.."
555 		"\\]^_"
556 		"......W[.....EFG"
557 		"..V....D....TU.Z"[c]);
558 #else
559 	else if (c < 128 && !control_char(c ^ 0100))
560 		SNPRINTF1(buf, sizeof(buf), "^%c", (int) (c ^ 0100));
561 #endif
562 	else
563 		SNPRINTF1(buf, sizeof(buf), binfmt, c);
564 	return (buf);
565 }
566 
567 /*
568  * Return the printable form of a UTF-8 character.
569  */
570 public char * prutfchar(LWCHAR ch)
571 {
572 	static char buf[MAX_PRCHAR_LEN+1];
573 
574 	if (ch == ESC)
575 		strcpy(buf, "ESC");
576 	else if (ch < 128 && control_char(ch))
577 	{
578 		if (!control_char(ch ^ 0100))
579 			SNPRINTF1(buf, sizeof(buf), "^%c", ((char) ch) ^ 0100);
580 		else
581 			SNPRINTF1(buf, sizeof(buf), binfmt, (char) ch);
582 	} else if (is_ubin_char(ch))
583 	{
584 		SNPRINTF1(buf, sizeof(buf), utfbinfmt, ch);
585 	} else
586 	{
587 		char *p = buf;
588 		if (ch >= 0x80000000)
589 			ch = 0xFFFD; /* REPLACEMENT CHARACTER */
590 		put_wchar(&p, ch);
591 		*p = '\0';
592 	}
593 	return (buf);
594 }
595 
596 /*
597  * Get the length of a UTF-8 character in bytes.
598  */
599 public int utf_len(int ch)
600 {
601 	if ((ch & 0x80) == 0)
602 		return 1;
603 	if ((ch & 0xE0) == 0xC0)
604 		return 2;
605 	if ((ch & 0xF0) == 0xE0)
606 		return 3;
607 	if ((ch & 0xF8) == 0xF0)
608 		return 4;
609 	if ((ch & 0xFC) == 0xF8)
610 		return 5;
611 	if ((ch & 0xFE) == 0xFC)
612 		return 6;
613 	/* Invalid UTF-8 encoding. */
614 	return 1;
615 }
616 
617 /*
618  * Does the parameter point to the lead byte of a well-formed UTF-8 character?
619  */
620 public int is_utf8_well_formed(char *ss, int slen)
621 {
622 	int i;
623 	int len;
624 	unsigned char *s = (unsigned char *) ss;
625 
626 	if (IS_UTF8_INVALID(s[0]))
627 		return (0);
628 
629 	len = utf_len(s[0]);
630 	if (len > slen)
631 		return (0);
632 	if (len == 1)
633 		return (1);
634 	if (len == 2)
635 	{
636 		if (s[0] < 0xC2)
637 		    return (0);
638 	} else
639 	{
640 		unsigned char mask;
641 		mask = (~((1 << (8-len)) - 1)) & 0xFF;
642 		if (s[0] == mask && (s[1] & mask) == 0x80)
643 			return (0);
644 	}
645 
646 	for (i = 1;  i < len;  i++)
647 		if (!IS_UTF8_TRAIL(s[i]))
648 			return (0);
649 	return (1);
650 }
651 
652 /*
653  * Skip bytes until a UTF-8 lead byte (11xxxxxx) or ASCII byte (0xxxxxxx) is found.
654  */
655 public void utf_skip_to_lead(char **pp, char *limit)
656 {
657 	do {
658 		++(*pp);
659 	} while (*pp < limit && !IS_UTF8_LEAD((*pp)[0] & 0377) && !IS_ASCII_OCTET((*pp)[0]));
660 }
661 
662 
663 /*
664  * Get the value of a UTF-8 character.
665  */
666 public LWCHAR get_wchar(constant char *p)
667 {
668 	switch (utf_len(p[0]))
669 	{
670 	case 1:
671 	default:
672 		/* 0xxxxxxx */
673 		return (LWCHAR)
674 			(p[0] & 0xFF);
675 	case 2:
676 		/* 110xxxxx 10xxxxxx */
677 		return (LWCHAR) (
678 			((p[0] & 0x1F) << 6) |
679 			(p[1] & 0x3F));
680 	case 3:
681 		/* 1110xxxx 10xxxxxx 10xxxxxx */
682 		return (LWCHAR) (
683 			((p[0] & 0x0F) << 12) |
684 			((p[1] & 0x3F) << 6) |
685 			(p[2] & 0x3F));
686 	case 4:
687 		/* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
688 		return (LWCHAR) (
689 			((p[0] & 0x07) << 18) |
690 			((p[1] & 0x3F) << 12) |
691 			((p[2] & 0x3F) << 6) |
692 			(p[3] & 0x3F));
693 	case 5:
694 		/* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
695 		return (LWCHAR) (
696 			((p[0] & 0x03) << 24) |
697 			((p[1] & 0x3F) << 18) |
698 			((p[2] & 0x3F) << 12) |
699 			((p[3] & 0x3F) << 6) |
700 			(p[4] & 0x3F));
701 	case 6:
702 		/* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
703 		return (LWCHAR) (
704 			((p[0] & 0x01) << 30) |
705 			((p[1] & 0x3F) << 24) |
706 			((p[2] & 0x3F) << 18) |
707 			((p[3] & 0x3F) << 12) |
708 			((p[4] & 0x3F) << 6) |
709 			(p[5] & 0x3F));
710 	}
711 }
712 
713 /*
714  * Store a character into a UTF-8 string.
715  */
716 public void put_wchar(char **pp, LWCHAR ch)
717 {
718 	if (!utf_mode || ch < 0x80)
719 	{
720 		/* 0xxxxxxx */
721 		*(*pp)++ = (char) ch;
722 	} else if (ch < 0x800)
723 	{
724 		/* 110xxxxx 10xxxxxx */
725 		*(*pp)++ = (char) (0xC0 | ((ch >> 6) & 0x1F));
726 		*(*pp)++ = (char) (0x80 | (ch & 0x3F));
727 	} else if (ch < 0x10000)
728 	{
729 		/* 1110xxxx 10xxxxxx 10xxxxxx */
730 		*(*pp)++ = (char) (0xE0 | ((ch >> 12) & 0x0F));
731 		*(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
732 		*(*pp)++ = (char) (0x80 | (ch & 0x3F));
733 	} else if (ch < 0x200000)
734 	{
735 		/* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
736 		*(*pp)++ = (char) (0xF0 | ((ch >> 18) & 0x07));
737 		*(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
738 		*(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
739 		*(*pp)++ = (char) (0x80 | (ch & 0x3F));
740 	} else if (ch < 0x4000000)
741 	{
742 		/* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
743 		*(*pp)++ = (char) (0xF0 | ((ch >> 24) & 0x03));
744 		*(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
745 		*(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
746 		*(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
747 		*(*pp)++ = (char) (0x80 | (ch & 0x3F));
748 	} else
749 	{
750 		/* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
751 		*(*pp)++ = (char) (0xF0 | ((ch >> 30) & 0x01));
752 		*(*pp)++ = (char) (0x80 | ((ch >> 24) & 0x3F));
753 		*(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
754 		*(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
755 		*(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
756 		*(*pp)++ = (char) (0x80 | (ch & 0x3F));
757 	}
758 }
759 
760 /*
761  * Step forward or backward one character in a string.
762  */
763 public LWCHAR step_char(char **pp, signed int dir, constant char *limit)
764 {
765 	LWCHAR ch;
766 	int len;
767 	char *p = *pp;
768 
769 	if (!utf_mode)
770 	{
771 		/* It's easy if chars are one byte. */
772 		if (dir > 0)
773 			ch = (LWCHAR) (unsigned char) ((p < limit) ? *p++ : 0);
774 		else
775 			ch = (LWCHAR) (unsigned char) ((p > limit) ? *--p : 0);
776 	} else if (dir > 0)
777 	{
778 		len = utf_len(*p);
779 		if (p + len > limit)
780 		{
781 			ch = 0;
782 			p = (char *) limit;
783 		} else
784 		{
785 			ch = get_wchar(p);
786 			p += len;
787 		}
788 	} else
789 	{
790 		while (p > limit && IS_UTF8_TRAIL(p[-1]))
791 			p--;
792 		if (p > limit)
793 			ch = get_wchar(--p);
794 		else
795 			ch = 0;
796 	}
797 	*pp = p;
798 	return ch;
799 }
800 
801 /*
802  * Unicode characters data
803  * Actual data is in the generated *.uni files.
804  */
805 
806 #define DECLARE_RANGE_TABLE_START(name) \
807 	static struct wchar_range name##_array[] = {
808 #define DECLARE_RANGE_TABLE_END(name) \
809 	}; struct wchar_range_table name##_table = { name##_array, sizeof(name##_array)/sizeof(*name##_array) };
810 
811 DECLARE_RANGE_TABLE_START(compose)
812 #include "compose.uni"
813 DECLARE_RANGE_TABLE_END(compose)
814 
815 DECLARE_RANGE_TABLE_START(ubin)
816 #include "ubin.uni"
817 DECLARE_RANGE_TABLE_END(ubin)
818 
819 DECLARE_RANGE_TABLE_START(wide)
820 #include "wide.uni"
821 DECLARE_RANGE_TABLE_END(wide)
822 
823 DECLARE_RANGE_TABLE_START(fmt)
824 #include "fmt.uni"
825 DECLARE_RANGE_TABLE_END(fmt)
826 
827 /* comb_table is special pairs, not ranges. */
828 static struct wchar_range comb_table[] = {
829 	{0x0644,0x0622}, {0x0644,0x0623}, {0x0644,0x0625}, {0x0644,0x0627},
830 };
831 
832 
833 static int is_in_table(LWCHAR ch, struct wchar_range_table *table)
834 {
835 	int hi;
836 	int lo;
837 
838 	/* Binary search in the table. */
839 	if (table->table == NULL || table->count == 0 || ch < table->table[0].first)
840 		return 0;
841 	lo = 0;
842 	hi = table->count - 1;
843 	while (lo <= hi)
844 	{
845 		int mid = (lo + hi) / 2;
846 		if (ch > table->table[mid].last)
847 			lo = mid + 1;
848 		else if (ch < table->table[mid].first)
849 			hi = mid - 1;
850 		else
851 			return 1;
852 	}
853 	return 0;
854 }
855 
856 /*
857  * Is a character a UTF-8 composing character?
858  * If a composing character follows any char, the two combine into one glyph.
859  */
860 public int is_composing_char(LWCHAR ch)
861 {
862 	if (is_in_table(ch, &user_prt_table)) return 0;
863 	return is_in_table(ch, &user_compose_table) ||
864 	       is_in_table(ch, &compose_table) ||
865 	       (bs_mode != BS_CONTROL && is_in_table(ch, &fmt_table));
866 }
867 
868 /*
869  * Should this UTF-8 character be treated as binary?
870  */
871 public int is_ubin_char(LWCHAR ch)
872 {
873 	if (is_in_table(ch, &user_prt_table)) return 0;
874 	return is_in_table(ch, &user_ubin_table) ||
875 	       is_in_table(ch, &ubin_table) ||
876 	       (bs_mode == BS_CONTROL && is_in_table(ch, &fmt_table));
877 }
878 
879 /*
880  * Is this a double width UTF-8 character?
881  */
882 public int is_wide_char(LWCHAR ch)
883 {
884 	return is_in_table(ch, &user_wide_table) ||
885 	       is_in_table(ch, &wide_table);
886 }
887 
888 /*
889  * Is a character a UTF-8 combining character?
890  * A combining char acts like an ordinary char, but if it follows
891  * a specific char (not any char), the two combine into one glyph.
892  */
893 public int is_combining_char(LWCHAR ch1, LWCHAR ch2)
894 {
895 	/* The table is small; use linear search. */
896 	int i;
897 	for (i = 0;  i < sizeof(comb_table)/sizeof(*comb_table);  i++)
898 	{
899 		if (ch1 == comb_table[i].first &&
900 		    ch2 == comb_table[i].last)
901 			return 1;
902 	}
903 	return 0;
904 }
905 
906