1 b2cfc4e2 2003-09-30 devnull /*
  2 b2cfc4e2 2003-09-30 devnull  * The authors of this software are Rob Pike and Ken Thompson.
  3 b2cfc4e2 2003-09-30 devnull  *              Copyright (c) 2002 by Lucent Technologies.
  4 b2cfc4e2 2003-09-30 devnull  * Permission to use, copy, modify, and distribute this software for any
  5 b2cfc4e2 2003-09-30 devnull  * purpose without fee is hereby granted, provided that this entire notice
  6 b2cfc4e2 2003-09-30 devnull  * is included in all copies of any software which is or includes a copy
  7 b2cfc4e2 2003-09-30 devnull  * or modification of this software and in all copies of the supporting
  8 b2cfc4e2 2003-09-30 devnull  * documentation for such software.
  9 b2cfc4e2 2003-09-30 devnull  * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
 10 b2cfc4e2 2003-09-30 devnull  * WARRANTY.  IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY
 11 b2cfc4e2 2003-09-30 devnull  * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
 12 b2cfc4e2 2003-09-30 devnull  * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
 13 b2cfc4e2 2003-09-30 devnull  */
 14 b2cfc4e2 2003-09-30 devnull #include <stdarg.h>
 15 b2cfc4e2 2003-09-30 devnull #include <string.h>
 16 b2cfc4e2 2003-09-30 devnull #include "utf.h"
 17 b2cfc4e2 2003-09-30 devnull #include "fmt.h"
 18 b2cfc4e2 2003-09-30 devnull #include "fmtdef.h"
 19 b2cfc4e2 2003-09-30 devnull 
 20 b2cfc4e2 2003-09-30 devnull /*
 21 b2cfc4e2 2003-09-30 devnull  * How many bytes of output UTF will be produced by quoting (if necessary) this string?
 22 b2cfc4e2 2003-09-30 devnull  * How many runes? How much of the input will be consumed?
 23 b2cfc4e2 2003-09-30 devnull  * The parameter q is filled in by __quotesetup.
 24 b2cfc4e2 2003-09-30 devnull  * The string may be UTF or Runes (s or r).
 25 b2cfc4e2 2003-09-30 devnull  * Return count does not include NUL.
 26 b2cfc4e2 2003-09-30 devnull  * Terminate the scan at the first of:
 27 b2cfc4e2 2003-09-30 devnull  *	NUL in input
 28 b2cfc4e2 2003-09-30 devnull  *	count exceeded in input
 29 b2cfc4e2 2003-09-30 devnull  *	count exceeded on output
 30 b2cfc4e2 2003-09-30 devnull  * *ninp is set to number of input bytes accepted.
 31 b2cfc4e2 2003-09-30 devnull  * nin may be <0 initially, to avoid checking input by count.
 32 b2cfc4e2 2003-09-30 devnull  */
 33 b2cfc4e2 2003-09-30 devnull void
 34 b2cfc4e2 2003-09-30 devnull __quotesetup(char *s, Rune *r, int nin, int nout, Quoteinfo *q, int sharp, int runesout)
 35 b2cfc4e2 2003-09-30 devnull {
 36 b2cfc4e2 2003-09-30 devnull 	int w;
 37 b2cfc4e2 2003-09-30 devnull 	Rune c;
 38 b2cfc4e2 2003-09-30 devnull 
 39 b2cfc4e2 2003-09-30 devnull 	q->quoted = 0;
 40 b2cfc4e2 2003-09-30 devnull 	q->nbytesout = 0;
 41 b2cfc4e2 2003-09-30 devnull 	q->nrunesout = 0;
 42 b2cfc4e2 2003-09-30 devnull 	q->nbytesin = 0;
 43 b2cfc4e2 2003-09-30 devnull 	q->nrunesin = 0;
 44 b2cfc4e2 2003-09-30 devnull 	if(sharp || nin==0 || (s && *s=='\0') || (r && *r=='\0')){
 45 b2cfc4e2 2003-09-30 devnull 		if(nout < 2)
 46 b2cfc4e2 2003-09-30 devnull 			return;
 47 b2cfc4e2 2003-09-30 devnull 		q->quoted = 1;
 48 b2cfc4e2 2003-09-30 devnull 		q->nbytesout = 2;
 49 b2cfc4e2 2003-09-30 devnull 		q->nrunesout = 2;
 50 b2cfc4e2 2003-09-30 devnull 	}
 51 b2cfc4e2 2003-09-30 devnull 	for(; nin!=0; nin-=w){
 52 b2cfc4e2 2003-09-30 devnull 		if(s)
 53 b2cfc4e2 2003-09-30 devnull 			w = chartorune(&c, s);
 54 b2cfc4e2 2003-09-30 devnull 		else{
 55 b2cfc4e2 2003-09-30 devnull 			c = *r;
 56 b2cfc4e2 2003-09-30 devnull 			w = runelen(c);
 57 b2cfc4e2 2003-09-30 devnull 		}
 58 b2cfc4e2 2003-09-30 devnull 
 59 b2cfc4e2 2003-09-30 devnull 		if(c == '\0')
 60 b2cfc4e2 2003-09-30 devnull 			break;
 61 b2cfc4e2 2003-09-30 devnull 		if(runesout){
 62 b2cfc4e2 2003-09-30 devnull 			if(q->nrunesout+1 > nout)
 63 b2cfc4e2 2003-09-30 devnull 				break;
 64 b2cfc4e2 2003-09-30 devnull 		}else{
 65 b2cfc4e2 2003-09-30 devnull 			if(q->nbytesout+w > nout)
 66 b2cfc4e2 2003-09-30 devnull 				break;
 67 b2cfc4e2 2003-09-30 devnull 		}
 68 b2cfc4e2 2003-09-30 devnull 
 69 b2cfc4e2 2003-09-30 devnull 		if((c <= L' ') || (c == L'\'') || (fmtdoquote!=nil && fmtdoquote(c))){
 70 b2cfc4e2 2003-09-30 devnull 			if(!q->quoted){
 71 b2cfc4e2 2003-09-30 devnull 				if(runesout){
 72 b2cfc4e2 2003-09-30 devnull 					if(1+q->nrunesout+1+1 > nout)	/* no room for quotes */
 73 b2cfc4e2 2003-09-30 devnull 						break;
 74 b2cfc4e2 2003-09-30 devnull 				}else{
 75 b2cfc4e2 2003-09-30 devnull 					if(1+q->nbytesout+w+1 > nout)	/* no room for quotes */
 76 b2cfc4e2 2003-09-30 devnull 						break;
 77 b2cfc4e2 2003-09-30 devnull 				}
 78 b2cfc4e2 2003-09-30 devnull 				q->nrunesout += 2;	/* include quotes */
 79 b2cfc4e2 2003-09-30 devnull 				q->nbytesout += 2;	/* include quotes */
 80 b2cfc4e2 2003-09-30 devnull 				q->quoted = 1;
 81 b2cfc4e2 2003-09-30 devnull 			}
 82 b2cfc4e2 2003-09-30 devnull 			if(c == '\'')	{
 83 b2cfc4e2 2003-09-30 devnull 				if(runesout){
 84 b2cfc4e2 2003-09-30 devnull 					if(1+q->nrunesout+1 > nout)	/* no room for quotes */
 85 b2cfc4e2 2003-09-30 devnull 						break;
 86 b2cfc4e2 2003-09-30 devnull 				}else{
 87 b2cfc4e2 2003-09-30 devnull 					if(1+q->nbytesout+w > nout)	/* no room for quotes */
 88 b2cfc4e2 2003-09-30 devnull 						break;
 89 b2cfc4e2 2003-09-30 devnull 				}
 90 b2cfc4e2 2003-09-30 devnull 				q->nbytesout++;
 91 b2cfc4e2 2003-09-30 devnull 				q->nrunesout++;	/* quotes reproduce as two characters */
 92 b2cfc4e2 2003-09-30 devnull 			}
 93 b2cfc4e2 2003-09-30 devnull 		}
 94 b2cfc4e2 2003-09-30 devnull 
 95 b2cfc4e2 2003-09-30 devnull 		/* advance input */
 96 b2cfc4e2 2003-09-30 devnull 		if(s)
 97 b2cfc4e2 2003-09-30 devnull 			s += w;
 98 b2cfc4e2 2003-09-30 devnull 		else
 99 b2cfc4e2 2003-09-30 devnull 			r++;
100 b2cfc4e2 2003-09-30 devnull 		q->nbytesin += w;
101 b2cfc4e2 2003-09-30 devnull 		q->nrunesin++;
102 b2cfc4e2 2003-09-30 devnull 
103 b2cfc4e2 2003-09-30 devnull 		/* advance output */
104 b2cfc4e2 2003-09-30 devnull 		q->nbytesout += w;
105 b2cfc4e2 2003-09-30 devnull 		q->nrunesout++;
106 b2cfc4e2 2003-09-30 devnull 	}
107 b2cfc4e2 2003-09-30 devnull }
108 b2cfc4e2 2003-09-30 devnull 
109 b2cfc4e2 2003-09-30 devnull static int
110 b2cfc4e2 2003-09-30 devnull qstrfmt(char *sin, Rune *rin, Quoteinfo *q, Fmt *f)
111 b2cfc4e2 2003-09-30 devnull {
112 b2cfc4e2 2003-09-30 devnull 	Rune r, *rm, *rme;
113 b2cfc4e2 2003-09-30 devnull 	char *t, *s, *m, *me;
114 b2cfc4e2 2003-09-30 devnull 	Rune *rt, *rs;
115 b2cfc4e2 2003-09-30 devnull 	ulong fl;
116 b2cfc4e2 2003-09-30 devnull 	int nc, w;
117 b2cfc4e2 2003-09-30 devnull 
118 b2cfc4e2 2003-09-30 devnull 	m = sin;
119 b2cfc4e2 2003-09-30 devnull 	me = m + q->nbytesin;
120 b2cfc4e2 2003-09-30 devnull 	rm = rin;
121 b2cfc4e2 2003-09-30 devnull 	rme = rm + q->nrunesin;
122 b2cfc4e2 2003-09-30 devnull 
123 b2cfc4e2 2003-09-30 devnull 	w = f->width;
124 b2cfc4e2 2003-09-30 devnull 	fl = f->flags;
125 b2cfc4e2 2003-09-30 devnull 	if(f->runes){
126 b2cfc4e2 2003-09-30 devnull 		if(!(fl & FmtLeft) && __rfmtpad(f, w - q->nrunesout) < 0)
127 b2cfc4e2 2003-09-30 devnull 			return -1;
128 b2cfc4e2 2003-09-30 devnull 	}else{
129 b2cfc4e2 2003-09-30 devnull 		if(!(fl & FmtLeft) && __fmtpad(f, w - q->nbytesout) < 0)
130 b2cfc4e2 2003-09-30 devnull 			return -1;
131 b2cfc4e2 2003-09-30 devnull 	}
132 b2cfc4e2 2003-09-30 devnull 	t = (char*)f->to;
133 b2cfc4e2 2003-09-30 devnull 	s = (char*)f->stop;
134 b2cfc4e2 2003-09-30 devnull 	rt = (Rune*)f->to;
135 b2cfc4e2 2003-09-30 devnull 	rs = (Rune*)f->stop;
136 b2cfc4e2 2003-09-30 devnull 	if(f->runes)
137 b2cfc4e2 2003-09-30 devnull 		FMTRCHAR(f, rt, rs, '\'');
138 b2cfc4e2 2003-09-30 devnull 	else
139 b2cfc4e2 2003-09-30 devnull 		FMTRUNE(f, t, s, '\'');
140 b2cfc4e2 2003-09-30 devnull 	for(nc = q->nrunesin; nc > 0; nc--){
141 b2cfc4e2 2003-09-30 devnull 		if(sin){
142 b2cfc4e2 2003-09-30 devnull 			r = *(uchar*)m;
143 b2cfc4e2 2003-09-30 devnull 			if(r < Runeself)
144 b2cfc4e2 2003-09-30 devnull 				m++;
145 b2cfc4e2 2003-09-30 devnull 			else if((me - m) >= UTFmax || fullrune(m, me-m))
146 b2cfc4e2 2003-09-30 devnull 				m += chartorune(&r, m);
147 b2cfc4e2 2003-09-30 devnull 			else
148 b2cfc4e2 2003-09-30 devnull 				break;
149 b2cfc4e2 2003-09-30 devnull 		}else{
150 b2cfc4e2 2003-09-30 devnull 			if(rm >= rme)
151 b2cfc4e2 2003-09-30 devnull 				break;
152 b2cfc4e2 2003-09-30 devnull 			r = *(uchar*)rm++;
153 b2cfc4e2 2003-09-30 devnull 		}
154 b2cfc4e2 2003-09-30 devnull 		if(f->runes){
155 b2cfc4e2 2003-09-30 devnull 			FMTRCHAR(f, rt, rs, r);
156 b2cfc4e2 2003-09-30 devnull 			if(r == '\'')
157 b2cfc4e2 2003-09-30 devnull 				FMTRCHAR(f, rt, rs, r);
158 b2cfc4e2 2003-09-30 devnull 		}else{
159 b2cfc4e2 2003-09-30 devnull 			FMTRUNE(f, t, s, r);
160 b2cfc4e2 2003-09-30 devnull 			if(r == '\'')
161 b2cfc4e2 2003-09-30 devnull 				FMTRUNE(f, t, s, r);
162 b2cfc4e2 2003-09-30 devnull 		}
163 b2cfc4e2 2003-09-30 devnull 	}
164 b2cfc4e2 2003-09-30 devnull 
165 b2cfc4e2 2003-09-30 devnull 	if(f->runes){
166 b2cfc4e2 2003-09-30 devnull 		FMTRCHAR(f, rt, rs, '\'');
167 b2cfc4e2 2003-09-30 devnull 		USED(rs);
168 b2cfc4e2 2003-09-30 devnull 		f->nfmt += rt - (Rune *)f->to;
169 b2cfc4e2 2003-09-30 devnull 		f->to = rt;
170 b2cfc4e2 2003-09-30 devnull 		if(fl & FmtLeft && __rfmtpad(f, w - q->nrunesout) < 0)
171 b2cfc4e2 2003-09-30 devnull 			return -1;
172 b2cfc4e2 2003-09-30 devnull 	}else{
173 b2cfc4e2 2003-09-30 devnull 		FMTRUNE(f, t, s, '\'');
174 b2cfc4e2 2003-09-30 devnull 		USED(s);
175 b2cfc4e2 2003-09-30 devnull 		f->nfmt += t - (char *)f->to;
176 b2cfc4e2 2003-09-30 devnull 		f->to = t;
177 b2cfc4e2 2003-09-30 devnull 		if(fl & FmtLeft && __fmtpad(f, w - q->nbytesout) < 0)
178 b2cfc4e2 2003-09-30 devnull 			return -1;
179 b2cfc4e2 2003-09-30 devnull 	}
180 b2cfc4e2 2003-09-30 devnull 	return 0;
181 b2cfc4e2 2003-09-30 devnull }
182 b2cfc4e2 2003-09-30 devnull 
183 b2cfc4e2 2003-09-30 devnull int
184 b2cfc4e2 2003-09-30 devnull __quotestrfmt(int runesin, Fmt *f)
185 b2cfc4e2 2003-09-30 devnull {
186 b2cfc4e2 2003-09-30 devnull 	int outlen;
187 b2cfc4e2 2003-09-30 devnull 	Rune *r;
188 b2cfc4e2 2003-09-30 devnull 	char *s;
189 b2cfc4e2 2003-09-30 devnull 	Quoteinfo q;
190 b2cfc4e2 2003-09-30 devnull 
191 b2cfc4e2 2003-09-30 devnull 	f->flags &= ~FmtPrec;	/* ignored for %q %Q, so disable for %s %S in easy case */
192 b2cfc4e2 2003-09-30 devnull 	if(runesin){
193 b2cfc4e2 2003-09-30 devnull 		r = va_arg(f->args, Rune *);
194 b2cfc4e2 2003-09-30 devnull 		s = nil;
195 b2cfc4e2 2003-09-30 devnull 	}else{
196 b2cfc4e2 2003-09-30 devnull 		s = va_arg(f->args, char *);
197 b2cfc4e2 2003-09-30 devnull 		r = nil;
198 b2cfc4e2 2003-09-30 devnull 	}
199 b2cfc4e2 2003-09-30 devnull 	if(!s && !r)
200 b2cfc4e2 2003-09-30 devnull 		return __fmtcpy(f, (void*)"<nil>", 5, 5);
201 b2cfc4e2 2003-09-30 devnull 
202 b2cfc4e2 2003-09-30 devnull 	if(f->flush)
203 b2cfc4e2 2003-09-30 devnull 		outlen = 0x7FFFFFFF;	/* if we can flush, no output limit */
204 b2cfc4e2 2003-09-30 devnull 	else if(f->runes)
205 b2cfc4e2 2003-09-30 devnull 		outlen = (Rune*)f->stop - (Rune*)f->to;
206 b2cfc4e2 2003-09-30 devnull 	else
207 b2cfc4e2 2003-09-30 devnull 		outlen = (char*)f->stop - (char*)f->to;
208 b2cfc4e2 2003-09-30 devnull 
209 b2cfc4e2 2003-09-30 devnull 	__quotesetup(s, r, -1, outlen, &q, f->flags&FmtSharp, f->runes);
210 b2cfc4e2 2003-09-30 devnull //print("bytes in %d bytes out %d runes in %d runesout %d\n", q.nbytesin, q.nbytesout, q.nrunesin, q.nrunesout);
211 b2cfc4e2 2003-09-30 devnull 
212 b2cfc4e2 2003-09-30 devnull 	if(runesin){
213 b2cfc4e2 2003-09-30 devnull 		if(!q.quoted)
214 b2cfc4e2 2003-09-30 devnull 			return __fmtrcpy(f, r, q.nrunesin);
215 b2cfc4e2 2003-09-30 devnull 		return qstrfmt(nil, r, &q, f);
216 b2cfc4e2 2003-09-30 devnull 	}
217 b2cfc4e2 2003-09-30 devnull 
218 b2cfc4e2 2003-09-30 devnull 	if(!q.quoted)
219 b2cfc4e2 2003-09-30 devnull 		return __fmtcpy(f, s, q.nrunesin, q.nbytesin);
220 b2cfc4e2 2003-09-30 devnull 	return qstrfmt(s, nil, &q, f);
221 b2cfc4e2 2003-09-30 devnull }
222 b2cfc4e2 2003-09-30 devnull 
223 b2cfc4e2 2003-09-30 devnull int
224 b2cfc4e2 2003-09-30 devnull quotestrfmt(Fmt *f)
225 b2cfc4e2 2003-09-30 devnull {
226 b2cfc4e2 2003-09-30 devnull 	return __quotestrfmt(0, f);
227 b2cfc4e2 2003-09-30 devnull }
228 b2cfc4e2 2003-09-30 devnull 
229 b2cfc4e2 2003-09-30 devnull int
230 b2cfc4e2 2003-09-30 devnull quoterunestrfmt(Fmt *f)
231 b2cfc4e2 2003-09-30 devnull {
232 b2cfc4e2 2003-09-30 devnull 	return __quotestrfmt(1, f);
233 b2cfc4e2 2003-09-30 devnull }
234 b2cfc4e2 2003-09-30 devnull 
235 b2cfc4e2 2003-09-30 devnull void
236 b2cfc4e2 2003-09-30 devnull quotefmtinstall(void)
237 b2cfc4e2 2003-09-30 devnull {
238 b2cfc4e2 2003-09-30 devnull 	fmtinstall('q', quotestrfmt);
239 b2cfc4e2 2003-09-30 devnull 	fmtinstall('Q', quoterunestrfmt);
240 b2cfc4e2 2003-09-30 devnull }
241 b2cfc4e2 2003-09-30 devnull 
242 b2cfc4e2 2003-09-30 devnull int
243 b2cfc4e2 2003-09-30 devnull __needsquotes(char *s, int *quotelenp)
244 b2cfc4e2 2003-09-30 devnull {
245 b2cfc4e2 2003-09-30 devnull 	Quoteinfo q;
246 b2cfc4e2 2003-09-30 devnull 
247 b2cfc4e2 2003-09-30 devnull 	__quotesetup(s, nil, -1, 0x7FFFFFFF, &q, 0, 0);
248 b2cfc4e2 2003-09-30 devnull 	*quotelenp = q.nbytesout;
249 b2cfc4e2 2003-09-30 devnull 
250 b2cfc4e2 2003-09-30 devnull 	return q.quoted;
251 b2cfc4e2 2003-09-30 devnull }
252 b2cfc4e2 2003-09-30 devnull 
253 b2cfc4e2 2003-09-30 devnull int
254 b2cfc4e2 2003-09-30 devnull __runeneedsquotes(Rune *r, int *quotelenp)
255 b2cfc4e2 2003-09-30 devnull {
256 b2cfc4e2 2003-09-30 devnull 	Quoteinfo q;
257 b2cfc4e2 2003-09-30 devnull 
258 b2cfc4e2 2003-09-30 devnull 	__quotesetup(nil, r, -1, 0x7FFFFFFF, &q, 0, 0);
259 b2cfc4e2 2003-09-30 devnull 	*quotelenp = q.nrunesout;
260 b2cfc4e2 2003-09-30 devnull 
261 b2cfc4e2 2003-09-30 devnull 	return q.quoted;
262 b2cfc4e2 2003-09-30 devnull }