src/basic/escape.c

   1 /***
   2   This file is part of systemd.
   3
   4   Copyright 2010 Lennart Poettering
   5
   6   systemd is free software; you can redistribute it and/or modify it
   7   under the terms of the GNU Lesser General Public License as published by
   8   the Free Software Foundation; either version 2.1 of the License, or
   9   (at your option) any later version.
  10
  11   systemd is distributed in the hope that it will be useful, but
  12   WITHOUT ANY WARRANTY; without even the implied warranty of
  13   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14   Lesser General Public License for more details.
  15
  16   You should have received a copy of the GNU Lesser General Public License
  17   along with systemd; If not, see <http://www.gnu.org/licenses/>.
  18 ***/
  19
  20 #include <errno.h>
  21 #include <stdlib.h>
  22 #include <string.h>
  23
  24 #include "alloc-util.h"
  25 #include "escape.h"
  26 #include "hexdecoct.h"
  27 #include "macro.h"
  28 #include "utf8.h"
  29
  30 size_t cescape_char(char c, char *buf) {
  31         char * buf_old = buf;
  32
  33         switch (c) {
  34
  35                 case '\a':
  36                         *(buf++) = '\\';
  37                         *(buf++) = 'a';
  38                         break;
  39                 case '\b':
  40                         *(buf++) = '\\';
  41                         *(buf++) = 'b';
  42                         break;
  43                 case '\f':
  44                         *(buf++) = '\\';
  45                         *(buf++) = 'f';
  46                         break;
  47                 case '\n':
  48                         *(buf++) = '\\';
  49                         *(buf++) = 'n';
  50                         break;
  51                 case '\r':
  52                         *(buf++) = '\\';
  53                         *(buf++) = 'r';
  54                         break;
  55                 case '\t':
  56                         *(buf++) = '\\';
  57                         *(buf++) = 't';
  58                         break;
  59                 case '\v':
  60                         *(buf++) = '\\';
  61                         *(buf++) = 'v';
  62                         break;
  63                 case '\\':
  64                         *(buf++) = '\\';
  65                         *(buf++) = '\\';
  66                         break;
  67                 case '"':
  68                         *(buf++) = '\\';
  69                         *(buf++) = '"';
  70                         break;
  71                 case '\'':
  72                         *(buf++) = '\\';
  73                         *(buf++) = '\'';
  74                         break;
  75
  76                 default:
  77                         /* For special chars we prefer octal over
  78                          * hexadecimal encoding, simply because glib's
  79                          * g_strescape() does the same */
  80                         if ((c < ' ') || (c >= 127)) {
  81                                 *(buf++) = '\\';
  82                                 *(buf++) = octchar((unsigned char) c >> 6);
  83                                 *(buf++) = octchar((unsigned char) c >> 3);
  84                                 *(buf++) = octchar((unsigned char) c);
  85                         } else
  86                                 *(buf++) = c;
  87                         break;
  88         }
  89
  90         return buf - buf_old;
  91 }
  92
  93 char *cescape_length(const char *s, size_t n) {
  94         const char *f;
  95         char *r, *t;
  96
  97         assert(s || n == 0);
  98
  99         /* Does C style string escaping. May be reversed with
 100          * cunescape(). */
 101
 102         r = new(char, n*4 + 1);
 103         if (!r)
 104                 return NULL;
 105
 106         for (f = s, t = r; f < s + n; f++)
 107                 t += cescape_char(*f, t);
 108
 109         *t = 0;
 110
 111         return r;
 112 }
 113
 114 char *cescape(const char *s) {
 115         assert(s);
 116
 117         return cescape_length(s, strlen(s));
 118 }
 119
 120 int cunescape_one(const char *p, size_t length, char32_t *ret, bool *eight_bit) {
 121         int r = 1;
 122
 123         assert(p);
 124         assert(*p);
 125         assert(ret);
 126
 127         /* Unescapes C style. Returns the unescaped character in ret.
 128          * Sets *eight_bit to true if the escaped sequence either fits in
 129          * one byte in UTF-8 or is a non-unicode literal byte and should
 130          * instead be copied directly.
 131          */
 132
 133         if (length != (size_t) -1 && length < 1)
 134                 return -EINVAL;
 135
 136         switch (p[0]) {
 137
 138         case 'a':
 139                 *ret = '\a';
 140                 break;
 141         case 'b':
 142                 *ret = '\b';
 143                 break;
 144         case 'f':
 145                 *ret = '\f';
 146                 break;
 147         case 'n':
 148                 *ret = '\n';
 149                 break;
 150         case 'r':
 151                 *ret = '\r';
 152                 break;
 153         case 't':
 154                 *ret = '\t';
 155                 break;
 156         case 'v':
 157                 *ret = '\v';
 158                 break;
 159         case '\\':
 160                 *ret = '\\';
 161                 break;
 162         case '"':
 163                 *ret = '"';
 164                 break;
 165         case '\'':
 166                 *ret = '\'';
 167                 break;
 168
 169         case 's':
 170                 /* This is an extension of the XDG syntax files */
 171                 *ret = ' ';
 172                 break;
 173
 174         case 'x': {
 175                 /* hexadecimal encoding */
 176                 int a, b;
 177
 178                 if (length != (size_t) -1 && length < 3)
 179                         return -EINVAL;
 180
 181                 a = unhexchar(p[1]);
 182                 if (a < 0)
 183                         return -EINVAL;
 184
 185                 b = unhexchar(p[2]);
 186                 if (b < 0)
 187                         return -EINVAL;
 188
 189                 /* Don't allow NUL bytes */
 190                 if (a == 0 && b == 0)
 191                         return -EINVAL;
 192
 193                 *ret = (a << 4U) | b;
 194                 *eight_bit = true;
 195                 r = 3;
 196                 break;
 197         }
 198
 199         case 'u': {
 200                 /* C++11 style 16bit unicode */
 201
 202                 int a[4];
 203                 unsigned i;
 204                 uint32_t c;
 205
 206                 if (length != (size_t) -1 && length < 5)
 207                         return -EINVAL;
 208
 209                 for (i = 0; i < 4; i++) {
 210                         a[i] = unhexchar(p[1 + i]);
 211                         if (a[i] < 0)
 212                                 return a[i];
 213                 }
 214
 215                 c = ((uint32_t) a[0] << 12U) | ((uint32_t) a[1] << 8U) | ((uint32_t) a[2] << 4U) | (uint32_t) a[3];
 216
 217                 /* Don't allow 0 chars */
 218                 if (c == 0)
 219                         return -EINVAL;
 220
 221                 *ret = c;
 222                 r = 5;
 223                 break;
 224         }
 225
 226         case 'U': {
 227                 /* C++11 style 32bit unicode */
 228
 229                 int a[8];
 230                 unsigned i;
 231                 char32_t c;
 232
 233                 if (length != (size_t) -1 && length < 9)
 234                         return -EINVAL;
 235
 236                 for (i = 0; i < 8; i++) {
 237                         a[i] = unhexchar(p[1 + i]);
 238                         if (a[i] < 0)
 239                                 return a[i];
 240                 }
 241
 242                 c = ((uint32_t) a[0] << 28U) | ((uint32_t) a[1] << 24U) | ((uint32_t) a[2] << 20U) | ((uint32_t) a[3] << 16U) |
 243                     ((uint32_t) a[4] << 12U) | ((uint32_t) a[5] <<  8U) | ((uint32_t) a[6] <<  4U) |  (uint32_t) a[7];
 244
 245                 /* Don't allow 0 chars */
 246                 if (c == 0)
 247                         return -EINVAL;
 248
 249                 /* Don't allow invalid code points */
 250                 if (!unichar_is_valid(c))
 251                         return -EINVAL;
 252
 253                 *ret = c;
 254                 r = 9;
 255                 break;
 256         }
 257
 258         case '0':
 259         case '1':
 260         case '2':
 261         case '3':
 262         case '4':
 263         case '5':
 264         case '6':
 265         case '7': {
 266                 /* octal encoding */
 267                 int a, b, c;
 268                 char32_t m;
 269
 270                 if (length != (size_t) -1 && length < 3)
 271                         return -EINVAL;
 272
 273                 a = unoctchar(p[0]);
 274                 if (a < 0)
 275                         return -EINVAL;
 276
 277                 b = unoctchar(p[1]);
 278                 if (b < 0)
 279                         return -EINVAL;
 280
 281                 c = unoctchar(p[2]);
 282                 if (c < 0)
 283                         return -EINVAL;
 284
 285                 /* don't allow NUL bytes */
 286                 if (a == 0 && b == 0 && c == 0)
 287                         return -EINVAL;
 288
 289                 /* Don't allow bytes above 255 */
 290                 m = ((uint32_t) a << 6U) | ((uint32_t) b << 3U) | (uint32_t) c;
 291                 if (m > 255)
 292                         return -EINVAL;
 293
 294                 *ret = m;
 295                 *eight_bit = true;
 296                 r = 3;
 297                 break;
 298         }
 299
 300         default:
 301                 return -EINVAL;
 302         }
 303
 304         return r;
 305 }
 306
 307 int cunescape_length_with_prefix(const char *s, size_t length, const char *prefix, UnescapeFlags flags, char **ret) {
 308         char *r, *t;
 309         const char *f;
 310         size_t pl;
 311
 312         assert(s);
 313         assert(ret);
 314
 315         /* Undoes C style string escaping, and optionally prefixes it. */
 316
 317         pl = strlen_ptr(prefix);
 318
 319         r = new(char, pl+length+1);
 320         if (!r)
 321                 return -ENOMEM;
 322
 323         if (prefix)
 324                 memcpy(r, prefix, pl);
 325
 326         for (f = s, t = r + pl; f < s + length; f++) {
 327                 size_t remaining;
 328                 bool eight_bit = false;
 329                 char32_t u;
 330                 int k;
 331
 332                 remaining = s + length - f;
 333                 assert(remaining > 0);
 334
 335                 if (*f != '\\') {
 336                         /* A literal, copy verbatim */
 337                         *(t++) = *f;
 338                         continue;
 339                 }
 340
 341                 if (remaining == 1) {
 342                         if (flags & UNESCAPE_RELAX) {
 343                                 /* A trailing backslash, copy verbatim */
 344                                 *(t++) = *f;
 345                                 continue;
 346                         }
 347
 348                         free(r);
 349                         return -EINVAL;
 350                 }
 351
 352                 k = cunescape_one(f + 1, remaining - 1, &u, &eight_bit);
 353                 if (k < 0) {
 354                         if (flags & UNESCAPE_RELAX) {
 355                                 /* Invalid escape code, let's take it literal then */
 356                                 *(t++) = '\\';
 357                                 continue;
 358                         }
 359
 360                         free(r);
 361                         return k;
 362                 }
 363
 364                 f += k;
 365                 if (eight_bit)
 366                         /* One byte? Set directly as specified */
 367                         *(t++) = u;
 368                 else
 369                         /* Otherwise encode as multi-byte UTF-8 */
 370                         t += utf8_encode_unichar(t, u);
 371         }
 372
 373         *t = 0;
 374
 375         *ret = r;
 376         return t - r;
 377 }
 378
 379 int cunescape_length(const char *s, size_t length, UnescapeFlags flags, char **ret) {
 380         return cunescape_length_with_prefix(s, length, NULL, flags, ret);
 381 }
 382
 383 int cunescape(const char *s, UnescapeFlags flags, char **ret) {
 384         return cunescape_length(s, strlen(s), flags, ret);
 385 }
 386
 387 char *xescape(const char *s, const char *bad) {
 388         char *r, *t;
 389         const char *f;
 390
 391         /* Escapes all chars in bad, in addition to \ and all special
 392          * chars, in \xFF style escaping. May be reversed with
 393          * cunescape(). */
 394
 395         r = new(char, strlen(s) * 4 + 1);
 396         if (!r)
 397                 return NULL;
 398
 399         for (f = s, t = r; *f; f++) {
 400
 401                 if ((*f < ' ') || (*f >= 127) ||
 402                     (*f == '\\') || strchr(bad, *f)) {
 403                         *(t++) = '\\';
 404                         *(t++) = 'x';
 405                         *(t++) = hexchar(*f >> 4);
 406                         *(t++) = hexchar(*f);
 407                 } else
 408                         *(t++) = *f;
 409         }
 410
 411         *t = 0;
 412
 413         return r;
 414 }
 415
 416 char *octescape(const char *s, size_t len) {
 417         char *r, *t;
 418         const char *f;
 419
 420         /* Escapes all chars in bad, in addition to \ and " chars,
 421          * in \nnn style escaping. */
 422
 423         r = new(char, len * 4 + 1);
 424         if (!r)
 425                 return NULL;
 426
 427         for (f = s, t = r; f < s + len; f++) {
 428
 429                 if (*f < ' ' || *f >= 127 || IN_SET(*f, '\\', '"')) {
 430                         *(t++) = '\\';
 431                         *(t++) = '0' + (*f >> 6);
 432                         *(t++) = '0' + ((*f >> 3) & 8);
 433                         *(t++) = '0' + (*f & 8);
 434                 } else
 435                         *(t++) = *f;
 436         }
 437
 438         *t = 0;
 439
 440         return r;
 441
 442 }
 443
 444 static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad, bool escape_tab_nl) {
 445         assert(bad);
 446
 447         for (; *s; s++) {
 448                 if (escape_tab_nl && IN_SET(*s, '\n', '\t')) {
 449                         *(t++) = '\\';
 450                         *(t++) = *s == '\n' ? 'n' : 't';
 451                         continue;
 452                 }
 453
 454                 if (*s == '\\' || strchr(bad, *s))
 455                         *(t++) = '\\';
 456
 457                 *(t++) = *s;
 458         }
 459
 460         return t;
 461 }
 462
 463 char *shell_escape(const char *s, const char *bad) {
 464         char *r, *t;
 465
 466         r = new(char, strlen(s)*2+1);
 467         if (!r)
 468                 return NULL;
 469
 470         t = strcpy_backslash_escaped(r, s, bad, false);
 471         *t = 0;
 472
 473         return r;
 474 }
 475
 476 char* shell_maybe_quote(const char *s, EscapeStyle style) {
 477         const char *p;
 478         char *r, *t;
 479
 480         assert(s);
 481
 482         /* Encloses a string in quotes if necessary to make it OK as a shell
 483          * string. Note that we treat benign UTF-8 characters as needing
 484          * escaping too, but that should be OK. */
 485
 486         for (p = s; *p; p++)
 487                 if (*p <= ' ' ||
 488                     *p >= 127 ||
 489                     strchr(SHELL_NEED_QUOTES, *p))
 490                         break;
 491
 492         if (!*p)
 493                 return strdup(s);
 494
 495         r = new(char, (style == ESCAPE_POSIX) + 1 + strlen(s)*2 + 1 + 1);
 496         if (!r)
 497                 return NULL;
 498
 499         t = r;
 500         if (style == ESCAPE_BACKSLASH)
 501                 *(t++) = '"';
 502         else if (style == ESCAPE_POSIX) {
 503                 *(t++) = '$';
 504                 *(t++) = '\'';
 505         } else
 506                 assert_not_reached("Bad EscapeStyle");
 507
 508         t = mempcpy(t, s, p - s);
 509
 510         if (style == ESCAPE_BACKSLASH)
 511                 t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE, false);
 512         else
 513                 t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE_POSIX, true);
 514
 515         if (style == ESCAPE_BACKSLASH)
 516                 *(t++) = '"';
 517         else
 518                 *(t++) = '\'';
 519         *t = 0;
 520
 521         return r;
 522 }