utf8.c revision 323134 
1/* $OpenBSD: utf8.c,v 1.3 2016/05/30 12:57:21 schwarze Exp $ */
2/*
3 * Copyright (c) 2016 Ingo Schwarze <schwarze@openbsd.org>
4 *
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
8 *
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 */
17
18/*
19 * Utility functions for multibyte-character handling,
20 * in particular to sanitize untrusted strings for terminal output.
21 */
22
23#include "includes.h"
24
25#include <sys/types.h>
26#ifdef HAVE_LANGINFO_H
27# include <langinfo.h>
28#endif
29#include <limits.h>
30#include <locale.h>
31#include <stdarg.h>
32#include <stdio.h>
33#include <stdlib.h>
34#include <string.h>
35#if defined(HAVE_STRNVIS) && defined(HAVE_VIS_H) && !defined(BROKEN_STRNVIS)
36# include <vis.h>
37#endif
38#ifdef HAVE_WCHAR_H
39# include <wchar.h>
40#endif
41
42#include "utf8.h"
43
44static int	 dangerous_locale(void);
45static int	 grow_dst(char **, size_t *, size_t, char **, size_t);
46static int	 vasnmprintf(char **, size_t, int *, const char *, va_list);
47
48
49/*
50 * For US-ASCII and UTF-8 encodings, we can safely recover from
51 * encoding errors and from non-printable characters.  For any
52 * other encodings, err to the side of caution and abort parsing:
53 * For state-dependent encodings, recovery is impossible.
54 * For arbitrary encodings, replacement of non-printable
55 * characters would be non-trivial and too fragile.
56 */
57
58static int
59dangerous_locale(void) {
60	char	*loc;
61
62	loc = nl_langinfo(CODESET);
63	return strcmp(loc, "US-ASCII") && strcmp(loc, "UTF-8");
64}
65
66static int
67grow_dst(char **dst, size_t *sz, size_t maxsz, char **dp, size_t need)
68{
69	char	*tp;
70	size_t	 tsz;
71
72	if (*dp + need < *dst + *sz)
73		return 0;
74	tsz = *sz + 128;
75	if (tsz > maxsz)
76		tsz = maxsz;
77	if ((tp = realloc(*dst, tsz)) == NULL)
78		return -1;
79	*dp = tp + (*dp - *dst);
80	*dst = tp;
81	*sz = tsz;
82	return 0;
83}
84
85/*
86 * The following two functions limit the number of bytes written,
87 * including the terminating '\0', to sz.  Unless wp is NULL,
88 * they limit the number of display columns occupied to *wp.
89 * Whichever is reached first terminates the output string.
90 * To stay close to the standard interfaces, they return the number of
91 * non-NUL bytes that would have been written if both were unlimited.
92 * If wp is NULL, newline, carriage return, and tab are allowed;
93 * otherwise, the actual number of columns occupied by what was
94 * written is returned in *wp.
95 */
96
97static int
98vasnmprintf(char **str, size_t maxsz, int *wp, const char *fmt, va_list ap)
99{
100	char	*src;	/* Source string returned from vasprintf. */
101	char	*sp;	/* Pointer into src. */
102	char	*dst;	/* Destination string to be returned. */
103	char	*dp;	/* Pointer into dst. */
104	char	*tp;	/* Temporary pointer for dst. */
105	size_t	 sz;	/* Number of bytes allocated for dst. */
106	wchar_t	 wc;	/* Wide character at sp. */
107	int	 len;	/* Number of bytes in the character at sp. */
108	int	 ret;	/* Number of bytes needed to format src. */
109	int	 width;	/* Display width of the character wc. */
110	int	 total_width, max_width, print;
111
112	src = NULL;
113	if ((ret = vasprintf(&src, fmt, ap)) <= 0)
114		goto fail;
115
116	sz = strlen(src) + 1;
117	if ((dst = malloc(sz)) == NULL) {
118		free(src);
119		goto fail;
120	}
121
122	if (maxsz > INT_MAX)
123		maxsz = INT_MAX;
124
125	sp = src;
126	dp = dst;
127	ret = 0;
128	print = 1;
129	total_width = 0;
130	max_width = wp == NULL ? INT_MAX : *wp;
131	while (*sp != '\0') {
132		if ((len = mbtowc(&wc, sp, MB_CUR_MAX)) == -1) {
133			(void)mbtowc(NULL, NULL, MB_CUR_MAX);
134			if (dangerous_locale()) {
135				ret = -1;
136				break;
137			}
138			len = 1;
139			width = -1;
140		} else if (wp == NULL &&
141		    (wc == L'\n' || wc == L'\r' || wc == L'\t')) {
142			/*
143			 * Don't use width uninitialized; the actual
144			 * value doesn't matter because total_width
145			 * is only returned for wp != NULL.
146			 */
147			width = 0;
148		} else if ((width = wcwidth(wc)) == -1 &&
149		    dangerous_locale()) {
150			ret = -1;
151			break;
152		}
153
154		/* Valid, printable character. */
155
156		if (width >= 0) {
157			if (print && (dp - dst >= (int)maxsz - len ||
158			    total_width > max_width - width))
159				print = 0;
160			if (print) {
161				if (grow_dst(&dst, &sz, maxsz,
162				    &dp, len) == -1) {
163					ret = -1;
164					break;
165				}
166				total_width += width;
167				memcpy(dp, sp, len);
168				dp += len;
169			}
170			sp += len;
171			if (ret >= 0)
172				ret += len;
173			continue;
174		}
175
176		/* Escaping required. */
177
178		while (len > 0) {
179			if (print && (dp - dst >= (int)maxsz - 4 ||
180			    total_width > max_width - 4))
181				print = 0;
182			if (print) {
183				if (grow_dst(&dst, &sz, maxsz,
184				    &dp, 4) == -1) {
185					ret = -1;
186					break;
187				}
188				tp = vis(dp, *sp, VIS_OCTAL | VIS_ALL, 0);
189				width = tp - dp;
190				total_width += width;
191				dp = tp;
192			} else
193				width = 4;
194			len--;
195			sp++;
196			if (ret >= 0)
197				ret += width;
198		}
199		if (len > 0)
200			break;
201	}
202	free(src);
203	*dp = '\0';
204	*str = dst;
205	if (wp != NULL)
206		*wp = total_width;
207
208	/*
209	 * If the string was truncated by the width limit but
210	 * would have fit into the size limit, the only sane way
211	 * to report the problem is using the return value, such
212	 * that the usual idiom "if (ret < 0 || ret >= sz) error"
213	 * works as expected.
214	 */
215
216	if (ret < (int)maxsz && !print)
217		ret = -1;
218	return ret;
219
220fail:
221	if (wp != NULL)
222		*wp = 0;
223	if (ret == 0) {
224		*str = src;
225		return 0;
226	} else {
227		*str = NULL;
228		return -1;
229	}
230}
231
232int
233snmprintf(char *str, size_t sz, int *wp, const char *fmt, ...)
234{
235	va_list	 ap;
236	char	*cp;
237	int	 ret;
238
239	va_start(ap, fmt);
240	ret = vasnmprintf(&cp, sz, wp, fmt, ap);
241	va_end(ap);
242	if (cp != NULL) {
243		(void)strlcpy(str, cp, sz);
244		free(cp);
245	} else
246		*str = '\0';
247	return ret;
248}
249
250/*
251 * To stay close to the standard interfaces, the following functions
252 * return the number of non-NUL bytes written.
253 */
254
255int
256vfmprintf(FILE *stream, const char *fmt, va_list ap)
257{
258	char	*str;
259	int	 ret;
260
261	if ((ret = vasnmprintf(&str, INT_MAX, NULL, fmt, ap)) < 0)
262		return -1;
263	if (fputs(str, stream) == EOF)
264		ret = -1;
265	free(str);
266	return ret;
267}
268
269int
270fmprintf(FILE *stream, const char *fmt, ...)
271{
272	va_list	 ap;
273	int	 ret;
274
275	va_start(ap, fmt);
276	ret = vfmprintf(stream, fmt, ap);
277	va_end(ap);
278	return ret;
279}
280
281int
282mprintf(const char *fmt, ...)
283{
284	va_list	 ap;
285	int	 ret;
286
287	va_start(ap, fmt);
288	ret = vfmprintf(stdout, fmt, ap);
289	va_end(ap);
290	return ret;
291}
292
293/*
294 * Set up libc for multibyte output in the user's chosen locale.
295 *
296 * XXX: we are known to have problems with Turkish (i/I confusion) so we
297 *      deliberately fall back to the C locale for now. Longer term we should
298 *      always prefer to select C.[encoding] if possible, but there's no
299 *      standardisation in locales between systems, so we'll need to survey
300 *      what's out there first.
301 */
302void
303msetlocale(void)
304{
305	const char *vars[] = { "LC_ALL", "LC_CTYPE", "LANG", NULL };
306	char *cp;
307	int i;
308
309	/*
310	 * We can't yet cope with dotless/dotted I in Turkish locales,
311	 * so fall back to the C locale for these.
312	 */
313	for (i = 0; vars[i] != NULL; i++) {
314		if ((cp = getenv(vars[i])) == NULL)
315			continue;
316		if (strncasecmp(cp, "TR", 2) != 0)
317			break;
318		/*
319		 * If we're in a UTF-8 locale then prefer to use
320		 * the C.UTF-8 locale (or equivalent) if it exists.
321		 */
322		if ((strcasestr(cp, "UTF-8") != NULL ||
323		    strcasestr(cp, "UTF8") != NULL) &&
324		    (setlocale(LC_CTYPE, "C.UTF-8") != NULL ||
325		    setlocale(LC_CTYPE, "POSIX.UTF-8") != NULL))
326			return;
327		setlocale(LC_CTYPE, "C");
328		return;
329	}
330	/* We can handle this locale */
331	setlocale(LC_CTYPE, "");
332}
333