1 /* $NetBSD: utf8.c,v 1.3 2016/08/03 15:24:28 jakllsch Exp $ */ 2 /* $OpenBSD: utf8.c,v 1.3 2016/05/30 12:57:21 schwarze Exp $ */ 3 /* 4 * Copyright (c) 2016 Ingo Schwarze <schwarze@openbsd.org> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 */ 18 19 #include "includes.h" 20 __RCSID("$NetBSD: utf8.c,v 1.3 2016/08/03 15:24:28 jakllsch Exp $"); 21 /* 22 * Utility functions for multibyte-character handling, 23 * in particular to sanitize untrusted strings for terminal output. 24 */ 25 26 #include <sys/types.h> 27 #include <langinfo.h> 28 #include <limits.h> 29 #include <stdarg.h> 30 #include <stdio.h> 31 #include <stdlib.h> 32 #include <string.h> 33 #include <vis.h> 34 #include <wchar.h> 35 36 #include "utf8.h" 37 38 static int dangerous_locale(void); 39 static int grow_dst(char **, size_t *, size_t, char **, size_t); 40 static int vasnmprintf(char **, size_t, int *, const char *, va_list) 41 __attribute__((format(printf, 4, 0))); 42 43 44 /* 45 * For US-ASCII and UTF-8 encodings, we can safely recover from 46 * encoding errors and from non-printable characters. For any 47 * other encodings, err to the side of caution and abort parsing: 48 * For state-dependent encodings, recovery is impossible. 49 * For arbitrary encodings, replacement of non-printable 50 * characters would be non-trivial and too fragile. 51 */ 52 53 static int 54 dangerous_locale(void) { 55 char *loc; 56 57 loc = nl_langinfo(CODESET); 58 return strcmp(loc, "US-ASCII") && strcmp(loc, "UTF-8"); 59 } 60 61 static int 62 grow_dst(char **dst, size_t *sz, size_t maxsz, char **dp, size_t need) 63 { 64 char *tp; 65 size_t tsz; 66 67 if (*dp + need < *dst + *sz) 68 return 0; 69 tsz = *sz + 128; 70 if (tsz > maxsz) 71 tsz = maxsz; 72 if ((tp = realloc(*dst, tsz)) == NULL) 73 return -1; 74 *dp = tp + (*dp - *dst); 75 *dst = tp; 76 *sz = tsz; 77 return 0; 78 } 79 80 /* 81 * The following two functions limit the number of bytes written, 82 * including the terminating '\0', to sz. Unless wp is NULL, 83 * they limit the number of display columns occupied to *wp. 84 * Whichever is reached first terminates the output string. 85 * To stay close to the standard interfaces, they return the number of 86 * non-NUL bytes that would have been written if both were unlimited. 87 * If wp is NULL, newline, carriage return, and tab are allowed; 88 * otherwise, the actual number of columns occupied by what was 89 * written is returned in *wp. 90 */ 91 92 static int 93 vasnmprintf(char **str, size_t maxsz, int *wp, const char *fmt, va_list ap) 94 { 95 char *src; /* Source string returned from vasprintf. */ 96 char *sp; /* Pointer into src. */ 97 char *dst; /* Destination string to be returned. */ 98 char *dp; /* Pointer into dst. */ 99 char *tp; /* Temporary pointer for dst. */ 100 size_t sz; /* Number of bytes allocated for dst. */ 101 wchar_t wc; /* Wide character at sp. */ 102 int len; /* Number of bytes in the character at sp. */ 103 int ret; /* Number of bytes needed to format src. */ 104 int width; /* Display width of the character wc. */ 105 int total_width, max_width, print; 106 107 src = NULL; 108 if ((ret = vasprintf(&src, fmt, ap)) <= 0) 109 goto fail; 110 111 sz = strlen(src) + 1; 112 if ((dst = malloc(sz)) == NULL) { 113 free(src); 114 goto fail; 115 } 116 117 if (maxsz > INT_MAX) 118 maxsz = INT_MAX; 119 120 sp = src; 121 dp = dst; 122 ret = 0; 123 print = 1; 124 total_width = 0; 125 max_width = wp == NULL ? INT_MAX : *wp; 126 while (*sp != '\0') { 127 if ((len = mbtowc(&wc, sp, MB_CUR_MAX)) == -1) { 128 (void)mbtowc(NULL, NULL, MB_CUR_MAX); 129 if (dangerous_locale()) { 130 ret = -1; 131 break; 132 } 133 len = 1; 134 width = -1; 135 } else if (wp == NULL && 136 (wc == L'\n' || wc == L'\r' || wc == L'\t')) { 137 /* 138 * Don't use width uninitialized; the actual 139 * value doesn't matter because total_width 140 * is only returned for wp != NULL. 141 */ 142 width = 0; 143 } else if ((width = wcwidth(wc)) == -1 && 144 dangerous_locale()) { 145 ret = -1; 146 break; 147 } 148 149 /* Valid, printable character. */ 150 151 if (width >= 0) { 152 if (print && (dp - dst >= (int)maxsz - len || 153 total_width > max_width - width)) 154 print = 0; 155 if (print) { 156 if (grow_dst(&dst, &sz, maxsz, 157 &dp, len) == -1) { 158 ret = -1; 159 break; 160 } 161 total_width += width; 162 memcpy(dp, sp, len); 163 dp += len; 164 } 165 sp += len; 166 if (ret >= 0) 167 ret += len; 168 continue; 169 } 170 171 /* Escaping required. */ 172 173 while (len > 0) { 174 if (print && (dp - dst >= (int)maxsz - 4 || 175 total_width > max_width - 4)) 176 print = 0; 177 if (print) { 178 if (grow_dst(&dst, &sz, maxsz, 179 &dp, 4) == -1) { 180 ret = -1; 181 break; 182 } 183 #ifndef VIS_ALL 184 #define VIS_ALL VIS_WHITE /* XXX */ 185 #endif 186 tp = vis(dp, *sp, VIS_OCTAL | VIS_ALL, 0); 187 width = tp - dp; 188 total_width += width; 189 dp = tp; 190 } else 191 width = 4; 192 len--; 193 sp++; 194 if (ret >= 0) 195 ret += width; 196 } 197 if (len > 0) 198 break; 199 } 200 free(src); 201 *dp = '\0'; 202 *str = dst; 203 if (wp != NULL) 204 *wp = total_width; 205 206 /* 207 * If the string was truncated by the width limit but 208 * would have fit into the size limit, the only sane way 209 * to report the problem is using the return value, such 210 * that the usual idiom "if (ret < 0 || ret >= sz) error" 211 * works as expected. 212 */ 213 214 if (ret < (int)maxsz && !print) 215 ret = -1; 216 return ret; 217 218 fail: 219 if (wp != NULL) 220 *wp = 0; 221 if (ret == 0) { 222 *str = src; 223 return 0; 224 } else { 225 *str = NULL; 226 return -1; 227 } 228 } 229 230 int 231 snmprintf(char *str, size_t sz, int *wp, const char *fmt, ...) 232 { 233 va_list ap; 234 char *cp; 235 int ret; 236 237 va_start(ap, fmt); 238 ret = vasnmprintf(&cp, sz, wp, fmt, ap); 239 va_end(ap); 240 if (cp != NULL) { 241 (void)strlcpy(str, cp, sz); 242 free(cp); 243 } else 244 *str = '\0'; 245 return ret; 246 } 247 248 /* 249 * To stay close to the standard interfaces, the following functions 250 * return the number of non-NUL bytes written. 251 */ 252 253 int 254 vfmprintf(FILE *stream, const char *fmt, va_list ap) 255 { 256 char *str; 257 int ret; 258 259 if ((ret = vasnmprintf(&str, INT_MAX, NULL, fmt, ap)) < 0) 260 return -1; 261 if (fputs(str, stream) == EOF) 262 ret = -1; 263 free(str); 264 return ret; 265 } 266 267 int 268 fmprintf(FILE *stream, const char *fmt, ...) 269 { 270 va_list ap; 271 int ret; 272 273 va_start(ap, fmt); 274 ret = vfmprintf(stream, fmt, ap); 275 va_end(ap); 276 return ret; 277 } 278 279 int 280 mprintf(const char *fmt, ...) 281 { 282 va_list ap; 283 int ret; 284 285 va_start(ap, fmt); 286 ret = vfmprintf(stdout, fmt, ap); 287 va_end(ap); 288 return ret; 289 } 290