1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
56 next(STR *s)  in next()  argument
63 	switch (s->state) {  in next()
69 		switch (*s->str) {  in next()
71 			s->state = EOS;  in next()
74 			s->lastch = backslash(s, &is_octal);  in next()
77 			if (bracket(s))  in next()
78 				return (next(s));  in next()
81 			clen = mbrtowc(&wch, s->str, MB_LEN_MAX, NULL);  in next()
82 			if (clen == (size_t)-1 || clen == (size_t)-2 ||  in next()
86 			s->lastch = wch;  in next()
87 			s->str += clen;  in next()
92 		if (s->str[0] == '-' && genrange(s, is_octal))  in next()
93 			return (next(s));  in next()
96 		if (s->cnt-- == 0) {  in next()
97 			s->state = NORMAL;  in next()
98 			return (next(s));  in next()
100 		++s->lastch;  in next()
103 		if (s->cnt-- == 0) {  in next()
104 			s->state = NORMAL;  in next()
105 			return (next(s));  in next()
111 		s->cnt++;  in next()
112 		ch = nextwctype(s->lastch, s->cclass);  in next()
113 		if (ch == -1) {  in next()
114 			s->state = NORMAL;  in next()
115 			return (next(s));  in next()
117 		s->lastch = ch;  in next()
120 		if ((ch = s->set[s->cnt++]) == OOBCH) {  in next()
121 			s->state = NORMAL;  in next()
122 			return (next(s));  in next()
124 		s->lastch = ch;  in next()
133 bracket(STR *s)  in bracket()  argument
137 	switch (s->str[1]) {  in bracket()
139 		if ((p = strchr(s->str + 2, ']')) == NULL)  in bracket()
141 		if (*(p - 1) != ':' || p - s->str < 4)  in bracket()
143 		*(p - 1) = '\0';  in bracket()
144 		s->str += 2;  in bracket()
145 		genclass(s);  in bracket()
146 		s->str = p + 1;  in bracket()
149 		if (s->str[2] == '\0' || (p = strchr(s->str + 3, ']')) == NULL)  in bracket()
151 		if (*(p - 1) != '=' || p - s->str < 4)  in bracket()
153 		s->str += 2;  in bracket()
154 		genequiv(s);  in bracket()
158 		if ((p = strpbrk(s->str + 2, "*]")) == NULL)  in bracket()
162 		s->str += 1;  in bracket()
163 		genseq(s);  in bracket()
170 genclass(STR *s)  in genclass()  argument
173 	if ((s->cclass = wctype(s->str)) == 0)  in genclass()
174 		errx(1, "unknown class %s", s->str);  in genclass()
175 	s->cnt = 0;  in genclass()
176 	s->lastch = -1;		/* incremented before check in next() */  in genclass()
177 	if (strcmp(s->str, "upper") == 0)  in genclass()
178 		s->state = CCLASS_UPPER;  in genclass()
179 	else if (strcmp(s->str, "lower") == 0)  in genclass()
180 		s->state = CCLASS_LOWER;  in genclass()
182 		s->state = CCLASS;  in genclass()
186 genequiv(STR *s)  in genequiv()  argument
193 	if (*s->str == '\\') {  in genequiv()
194 		s->equiv[0] = backslash(s, NULL);  in genequiv()
195 		if (*s->str != '=')  in genequiv()
197 		s->str += 2;  in genequiv()
199 		clen = mbrtowc(&wc, s->str, MB_LEN_MAX, NULL);  in genequiv()
200 		if (clen == (size_t)-1 || clen == (size_t)-2 || clen == 0)  in genequiv()
202 		s->equiv[0] = wc;  in genequiv()
203 		if (s->str[clen] != '=')  in genequiv()
205 		s->str += clen + 2;  in genequiv()
213 	 * it fills the string with primary collation weight bytes. Only one-  in genequiv()
214 	 * to-one mappings are supported.  in genequiv()
217 	src[0] = (char)s->equiv[0];  in genequiv()
225 				s->equiv[p++] = i;  in genequiv()
227 		s->equiv[p] = OOBCH;  in genequiv()
230 	s->cnt = 0;  in genequiv()
231 	s->state = SET;  in genequiv()
232 	s->set = s->equiv;  in genequiv()
236 genrange(STR *s, int was_octal)  in genrange()  argument
245 	savestart = s->str;  in genrange()
246 	if (*++s->str == '\\')  in genrange()
247 		stopval = backslash(s, &octal);  in genrange()
249 		clen = mbrtowc(&wc, s->str, MB_LEN_MAX, NULL);  in genrange()
250 		if (clen == (size_t)-1 || clen == (size_t)-2)  in genrange()
253 		s->str += clen;  in genrange()
260 		if (stopval < s->lastch) {  in genrange()
261 			s->str = savestart;  in genrange()
264 		s->cnt = stopval - s->lastch + 1;  in genrange()
265 		s->state = RANGE;  in genrange()
266 		--s->lastch;  in genrange()
269 	if (charcoll((const void *)&stopval, (const void *)&(s->lastch)) < 0) {  in genrange()
270 		s->str = savestart;  in genrange()
273 	if ((s->set = p = malloc((NCHARS_SB + 1) * sizeof(int))) == NULL)  in genrange()
276 		if (charcoll((const void *)&cnt, (const void *)&(s->lastch)) >= 0 &&  in genrange()
280 	n = p - s->set;  in genrange()
282 	s->cnt = 0;  in genrange()
283 	s->state = SET;  in genrange()
285 		mergesort(s->set, n, sizeof(*(s->set)), charcoll);  in genrange()
290 genseq(STR *s)  in genseq()  argument
296 	if (s->which == STRING1)  in genseq()
299 	if (*s->str == '\\')  in genseq()
300 		s->lastch = backslash(s, NULL);  in genseq()
302 		clen = mbrtowc(&wc, s->str, MB_LEN_MAX, NULL);  in genseq()
303 		if (clen == (size_t)-1 || clen == (size_t)-2)  in genseq()
305 		s->lastch = wc;  in genseq()
306 		s->str += clen;  in genseq()
308 	if (*s->str != '*')  in genseq()
311 	switch (*++s->str) {  in genseq()
313 		s->cnt = backslash(s, NULL);  in genseq()
316 		s->cnt = 0;  in genseq()
317 		++s->str;  in genseq()
320 		if (isdigit((u_char)*s->str)) {  in genseq()
321 			s->cnt = strtol(s->str, &ep, 0);  in genseq()
323 				s->str = ep + 1;  in genseq()
331 	s->state = s->cnt ? SEQUENCE : INFINITE;  in genseq()
339 backslash(STR *s, int *is_octal)  in backslash()  argument
346 		ch = (u_char)*++s->str;  in backslash()
349 		val = val * 8 + ch - '0';  in backslash()
351 			++s->str;  in backslash()
361 		++s->str;  in backslash()
377 		case '\0':			/*  \" -> \ */  in backslash()
378 			s->state = EOS;  in backslash()
380 		default:			/* \x" -> x */  in backslash()