]>
Commit | Line | Data |
---|---|---|
53e1b683 | 1 | /* SPDX-License-Identifier: LGPL-2.1+ */ |
4f5dd394 | 2 | |
11c3a366 TA |
3 | #include <errno.h> |
4 | #include <stdlib.h> | |
5 | #include <string.h> | |
6 | ||
b5efdb8a | 7 | #include "alloc-util.h" |
e4e73a63 LP |
8 | #include "escape.h" |
9 | #include "hexdecoct.h" | |
11c3a366 | 10 | #include "macro.h" |
4f5dd394 | 11 | #include "utf8.h" |
4f5dd394 | 12 | |
b778252b ZJS |
13 | int cescape_char(char c, char *buf) { |
14 | char *buf_old = buf; | |
4f5dd394 | 15 | |
76a35973 LP |
16 | /* Needs space for 4 characters in the buffer */ |
17 | ||
4f5dd394 LP |
18 | switch (c) { |
19 | ||
20 | case '\a': | |
21 | *(buf++) = '\\'; | |
22 | *(buf++) = 'a'; | |
23 | break; | |
24 | case '\b': | |
25 | *(buf++) = '\\'; | |
26 | *(buf++) = 'b'; | |
27 | break; | |
28 | case '\f': | |
29 | *(buf++) = '\\'; | |
30 | *(buf++) = 'f'; | |
31 | break; | |
32 | case '\n': | |
33 | *(buf++) = '\\'; | |
34 | *(buf++) = 'n'; | |
35 | break; | |
36 | case '\r': | |
37 | *(buf++) = '\\'; | |
38 | *(buf++) = 'r'; | |
39 | break; | |
40 | case '\t': | |
41 | *(buf++) = '\\'; | |
42 | *(buf++) = 't'; | |
43 | break; | |
44 | case '\v': | |
45 | *(buf++) = '\\'; | |
46 | *(buf++) = 'v'; | |
47 | break; | |
48 | case '\\': | |
49 | *(buf++) = '\\'; | |
50 | *(buf++) = '\\'; | |
51 | break; | |
52 | case '"': | |
53 | *(buf++) = '\\'; | |
54 | *(buf++) = '"'; | |
55 | break; | |
56 | case '\'': | |
57 | *(buf++) = '\\'; | |
58 | *(buf++) = '\''; | |
59 | break; | |
60 | ||
61 | default: | |
62 | /* For special chars we prefer octal over | |
63 | * hexadecimal encoding, simply because glib's | |
64 | * g_strescape() does the same */ | |
65 | if ((c < ' ') || (c >= 127)) { | |
66 | *(buf++) = '\\'; | |
67 | *(buf++) = octchar((unsigned char) c >> 6); | |
68 | *(buf++) = octchar((unsigned char) c >> 3); | |
69 | *(buf++) = octchar((unsigned char) c); | |
70 | } else | |
71 | *(buf++) = c; | |
72 | break; | |
73 | } | |
74 | ||
75 | return buf - buf_old; | |
76 | } | |
77 | ||
a5ef3638 | 78 | char *cescape_length(const char *s, size_t n) { |
4f5dd394 | 79 | const char *f; |
a5ef3638 | 80 | char *r, *t; |
4f5dd394 | 81 | |
a5ef3638 | 82 | assert(s || n == 0); |
4f5dd394 LP |
83 | |
84 | /* Does C style string escaping. May be reversed with | |
85 | * cunescape(). */ | |
86 | ||
a5ef3638 | 87 | r = new(char, n*4 + 1); |
4f5dd394 LP |
88 | if (!r) |
89 | return NULL; | |
90 | ||
a5ef3638 | 91 | for (f = s, t = r; f < s + n; f++) |
4f5dd394 LP |
92 | t += cescape_char(*f, t); |
93 | ||
94 | *t = 0; | |
95 | ||
96 | return r; | |
97 | } | |
98 | ||
a5ef3638 LP |
99 | char *cescape(const char *s) { |
100 | assert(s); | |
101 | ||
102 | return cescape_length(s, strlen(s)); | |
103 | } | |
104 | ||
c932fb71 | 105 | int cunescape_one(const char *p, size_t length, char32_t *ret, bool *eight_bit) { |
4f5dd394 LP |
106 | int r = 1; |
107 | ||
108 | assert(p); | |
109 | assert(*p); | |
110 | assert(ret); | |
111 | ||
3565e095 ZJS |
112 | /* Unescapes C style. Returns the unescaped character in ret. |
113 | * Sets *eight_bit to true if the escaped sequence either fits in | |
114 | * one byte in UTF-8 or is a non-unicode literal byte and should | |
115 | * instead be copied directly. | |
116 | */ | |
4f5dd394 LP |
117 | |
118 | if (length != (size_t) -1 && length < 1) | |
119 | return -EINVAL; | |
120 | ||
121 | switch (p[0]) { | |
122 | ||
123 | case 'a': | |
124 | *ret = '\a'; | |
125 | break; | |
126 | case 'b': | |
127 | *ret = '\b'; | |
128 | break; | |
129 | case 'f': | |
130 | *ret = '\f'; | |
131 | break; | |
132 | case 'n': | |
133 | *ret = '\n'; | |
134 | break; | |
135 | case 'r': | |
136 | *ret = '\r'; | |
137 | break; | |
138 | case 't': | |
139 | *ret = '\t'; | |
140 | break; | |
141 | case 'v': | |
142 | *ret = '\v'; | |
143 | break; | |
144 | case '\\': | |
145 | *ret = '\\'; | |
146 | break; | |
147 | case '"': | |
148 | *ret = '"'; | |
149 | break; | |
150 | case '\'': | |
151 | *ret = '\''; | |
152 | break; | |
153 | ||
154 | case 's': | |
155 | /* This is an extension of the XDG syntax files */ | |
156 | *ret = ' '; | |
157 | break; | |
158 | ||
159 | case 'x': { | |
160 | /* hexadecimal encoding */ | |
161 | int a, b; | |
162 | ||
163 | if (length != (size_t) -1 && length < 3) | |
164 | return -EINVAL; | |
165 | ||
166 | a = unhexchar(p[1]); | |
167 | if (a < 0) | |
168 | return -EINVAL; | |
169 | ||
170 | b = unhexchar(p[2]); | |
171 | if (b < 0) | |
172 | return -EINVAL; | |
173 | ||
174 | /* Don't allow NUL bytes */ | |
175 | if (a == 0 && b == 0) | |
176 | return -EINVAL; | |
177 | ||
3565e095 ZJS |
178 | *ret = (a << 4U) | b; |
179 | *eight_bit = true; | |
4f5dd394 LP |
180 | r = 3; |
181 | break; | |
182 | } | |
183 | ||
184 | case 'u': { | |
185 | /* C++11 style 16bit unicode */ | |
186 | ||
187 | int a[4]; | |
da6053d0 | 188 | size_t i; |
4f5dd394 LP |
189 | uint32_t c; |
190 | ||
191 | if (length != (size_t) -1 && length < 5) | |
192 | return -EINVAL; | |
193 | ||
194 | for (i = 0; i < 4; i++) { | |
195 | a[i] = unhexchar(p[1 + i]); | |
196 | if (a[i] < 0) | |
197 | return a[i]; | |
198 | } | |
199 | ||
200 | c = ((uint32_t) a[0] << 12U) | ((uint32_t) a[1] << 8U) | ((uint32_t) a[2] << 4U) | (uint32_t) a[3]; | |
201 | ||
202 | /* Don't allow 0 chars */ | |
203 | if (c == 0) | |
204 | return -EINVAL; | |
205 | ||
3565e095 | 206 | *ret = c; |
4f5dd394 LP |
207 | r = 5; |
208 | break; | |
209 | } | |
210 | ||
211 | case 'U': { | |
212 | /* C++11 style 32bit unicode */ | |
213 | ||
214 | int a[8]; | |
da6053d0 | 215 | size_t i; |
c932fb71 | 216 | char32_t c; |
4f5dd394 LP |
217 | |
218 | if (length != (size_t) -1 && length < 9) | |
219 | return -EINVAL; | |
220 | ||
221 | for (i = 0; i < 8; i++) { | |
222 | a[i] = unhexchar(p[1 + i]); | |
223 | if (a[i] < 0) | |
224 | return a[i]; | |
225 | } | |
226 | ||
dcd12626 LP |
227 | c = ((uint32_t) a[0] << 28U) | ((uint32_t) a[1] << 24U) | ((uint32_t) a[2] << 20U) | ((uint32_t) a[3] << 16U) | |
228 | ((uint32_t) a[4] << 12U) | ((uint32_t) a[5] << 8U) | ((uint32_t) a[6] << 4U) | (uint32_t) a[7]; | |
4f5dd394 LP |
229 | |
230 | /* Don't allow 0 chars */ | |
231 | if (c == 0) | |
232 | return -EINVAL; | |
233 | ||
234 | /* Don't allow invalid code points */ | |
235 | if (!unichar_is_valid(c)) | |
236 | return -EINVAL; | |
237 | ||
3565e095 | 238 | *ret = c; |
4f5dd394 LP |
239 | r = 9; |
240 | break; | |
241 | } | |
242 | ||
243 | case '0': | |
244 | case '1': | |
245 | case '2': | |
246 | case '3': | |
247 | case '4': | |
248 | case '5': | |
249 | case '6': | |
250 | case '7': { | |
251 | /* octal encoding */ | |
252 | int a, b, c; | |
c932fb71 | 253 | char32_t m; |
4f5dd394 LP |
254 | |
255 | if (length != (size_t) -1 && length < 3) | |
256 | return -EINVAL; | |
257 | ||
258 | a = unoctchar(p[0]); | |
259 | if (a < 0) | |
260 | return -EINVAL; | |
261 | ||
262 | b = unoctchar(p[1]); | |
263 | if (b < 0) | |
264 | return -EINVAL; | |
265 | ||
266 | c = unoctchar(p[2]); | |
267 | if (c < 0) | |
268 | return -EINVAL; | |
269 | ||
270 | /* don't allow NUL bytes */ | |
271 | if (a == 0 && b == 0 && c == 0) | |
272 | return -EINVAL; | |
273 | ||
274 | /* Don't allow bytes above 255 */ | |
dcd12626 | 275 | m = ((uint32_t) a << 6U) | ((uint32_t) b << 3U) | (uint32_t) c; |
4f5dd394 LP |
276 | if (m > 255) |
277 | return -EINVAL; | |
278 | ||
279 | *ret = m; | |
3565e095 | 280 | *eight_bit = true; |
4f5dd394 LP |
281 | r = 3; |
282 | break; | |
283 | } | |
284 | ||
285 | default: | |
286 | return -EINVAL; | |
287 | } | |
288 | ||
289 | return r; | |
290 | } | |
291 | ||
292 | int cunescape_length_with_prefix(const char *s, size_t length, const char *prefix, UnescapeFlags flags, char **ret) { | |
293 | char *r, *t; | |
294 | const char *f; | |
295 | size_t pl; | |
296 | ||
297 | assert(s); | |
298 | assert(ret); | |
299 | ||
300 | /* Undoes C style string escaping, and optionally prefixes it. */ | |
301 | ||
7bf7ce28 | 302 | pl = strlen_ptr(prefix); |
4f5dd394 LP |
303 | |
304 | r = new(char, pl+length+1); | |
305 | if (!r) | |
306 | return -ENOMEM; | |
307 | ||
308 | if (prefix) | |
309 | memcpy(r, prefix, pl); | |
310 | ||
311 | for (f = s, t = r + pl; f < s + length; f++) { | |
312 | size_t remaining; | |
3565e095 | 313 | bool eight_bit = false; |
c932fb71 | 314 | char32_t u; |
4f5dd394 LP |
315 | int k; |
316 | ||
317 | remaining = s + length - f; | |
318 | assert(remaining > 0); | |
319 | ||
320 | if (*f != '\\') { | |
629ff674 | 321 | /* A literal, copy verbatim */ |
4f5dd394 LP |
322 | *(t++) = *f; |
323 | continue; | |
324 | } | |
325 | ||
326 | if (remaining == 1) { | |
327 | if (flags & UNESCAPE_RELAX) { | |
328 | /* A trailing backslash, copy verbatim */ | |
329 | *(t++) = *f; | |
330 | continue; | |
331 | } | |
332 | ||
333 | free(r); | |
334 | return -EINVAL; | |
335 | } | |
336 | ||
3565e095 | 337 | k = cunescape_one(f + 1, remaining - 1, &u, &eight_bit); |
4f5dd394 LP |
338 | if (k < 0) { |
339 | if (flags & UNESCAPE_RELAX) { | |
340 | /* Invalid escape code, let's take it literal then */ | |
341 | *(t++) = '\\'; | |
342 | continue; | |
343 | } | |
344 | ||
345 | free(r); | |
346 | return k; | |
347 | } | |
348 | ||
3565e095 ZJS |
349 | f += k; |
350 | if (eight_bit) | |
351 | /* One byte? Set directly as specified */ | |
352 | *(t++) = u; | |
4f5dd394 | 353 | else |
3565e095 | 354 | /* Otherwise encode as multi-byte UTF-8 */ |
4f5dd394 | 355 | t += utf8_encode_unichar(t, u); |
4f5dd394 LP |
356 | } |
357 | ||
358 | *t = 0; | |
359 | ||
360 | *ret = r; | |
361 | return t - r; | |
362 | } | |
363 | ||
364 | int cunescape_length(const char *s, size_t length, UnescapeFlags flags, char **ret) { | |
365 | return cunescape_length_with_prefix(s, length, NULL, flags, ret); | |
366 | } | |
367 | ||
368 | int cunescape(const char *s, UnescapeFlags flags, char **ret) { | |
369 | return cunescape_length(s, strlen(s), flags, ret); | |
370 | } | |
371 | ||
372 | char *xescape(const char *s, const char *bad) { | |
373 | char *r, *t; | |
374 | const char *f; | |
375 | ||
376 | /* Escapes all chars in bad, in addition to \ and all special | |
377 | * chars, in \xFF style escaping. May be reversed with | |
378 | * cunescape(). */ | |
379 | ||
380 | r = new(char, strlen(s) * 4 + 1); | |
381 | if (!r) | |
382 | return NULL; | |
383 | ||
384 | for (f = s, t = r; *f; f++) { | |
385 | ||
386 | if ((*f < ' ') || (*f >= 127) || | |
387 | (*f == '\\') || strchr(bad, *f)) { | |
388 | *(t++) = '\\'; | |
389 | *(t++) = 'x'; | |
390 | *(t++) = hexchar(*f >> 4); | |
391 | *(t++) = hexchar(*f); | |
392 | } else | |
393 | *(t++) = *f; | |
394 | } | |
395 | ||
396 | *t = 0; | |
397 | ||
398 | return r; | |
399 | } | |
400 | ||
95052df3 ZJS |
401 | char *octescape(const char *s, size_t len) { |
402 | char *r, *t; | |
403 | const char *f; | |
404 | ||
405 | /* Escapes all chars in bad, in addition to \ and " chars, | |
406 | * in \nnn style escaping. */ | |
407 | ||
408 | r = new(char, len * 4 + 1); | |
409 | if (!r) | |
410 | return NULL; | |
411 | ||
412 | for (f = s, t = r; f < s + len; f++) { | |
413 | ||
4c701096 | 414 | if (*f < ' ' || *f >= 127 || IN_SET(*f, '\\', '"')) { |
95052df3 ZJS |
415 | *(t++) = '\\'; |
416 | *(t++) = '0' + (*f >> 6); | |
417 | *(t++) = '0' + ((*f >> 3) & 8); | |
418 | *(t++) = '0' + (*f & 8); | |
419 | } else | |
420 | *(t++) = *f; | |
421 | } | |
422 | ||
423 | *t = 0; | |
424 | ||
425 | return r; | |
426 | ||
427 | } | |
428 | ||
804ee07c | 429 | static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad, bool escape_tab_nl) { |
4f5dd394 LP |
430 | assert(bad); |
431 | ||
432 | for (; *s; s++) { | |
804ee07c ZJS |
433 | if (escape_tab_nl && IN_SET(*s, '\n', '\t')) { |
434 | *(t++) = '\\'; | |
435 | *(t++) = *s == '\n' ? 'n' : 't'; | |
436 | continue; | |
437 | } | |
438 | ||
4f5dd394 LP |
439 | if (*s == '\\' || strchr(bad, *s)) |
440 | *(t++) = '\\'; | |
441 | ||
442 | *(t++) = *s; | |
443 | } | |
444 | ||
445 | return t; | |
446 | } | |
447 | ||
448 | char *shell_escape(const char *s, const char *bad) { | |
449 | char *r, *t; | |
450 | ||
451 | r = new(char, strlen(s)*2+1); | |
452 | if (!r) | |
453 | return NULL; | |
454 | ||
804ee07c | 455 | t = strcpy_backslash_escaped(r, s, bad, false); |
4f5dd394 LP |
456 | *t = 0; |
457 | ||
458 | return r; | |
459 | } | |
460 | ||
804ee07c | 461 | char* shell_maybe_quote(const char *s, EscapeStyle style) { |
4f5dd394 LP |
462 | const char *p; |
463 | char *r, *t; | |
464 | ||
465 | assert(s); | |
466 | ||
804ee07c ZJS |
467 | /* Encloses a string in quotes if necessary to make it OK as a shell |
468 | * string. Note that we treat benign UTF-8 characters as needing | |
469 | * escaping too, but that should be OK. */ | |
4f5dd394 LP |
470 | |
471 | for (p = s; *p; p++) | |
472 | if (*p <= ' ' || | |
473 | *p >= 127 || | |
474 | strchr(SHELL_NEED_QUOTES, *p)) | |
475 | break; | |
476 | ||
477 | if (!*p) | |
478 | return strdup(s); | |
479 | ||
804ee07c | 480 | r = new(char, (style == ESCAPE_POSIX) + 1 + strlen(s)*2 + 1 + 1); |
4f5dd394 LP |
481 | if (!r) |
482 | return NULL; | |
483 | ||
484 | t = r; | |
804ee07c ZJS |
485 | if (style == ESCAPE_BACKSLASH) |
486 | *(t++) = '"'; | |
487 | else if (style == ESCAPE_POSIX) { | |
488 | *(t++) = '$'; | |
489 | *(t++) = '\''; | |
490 | } else | |
491 | assert_not_reached("Bad EscapeStyle"); | |
492 | ||
4f5dd394 LP |
493 | t = mempcpy(t, s, p - s); |
494 | ||
804ee07c ZJS |
495 | if (style == ESCAPE_BACKSLASH) |
496 | t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE, false); | |
497 | else | |
498 | t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE_POSIX, true); | |
4f5dd394 | 499 | |
804ee07c ZJS |
500 | if (style == ESCAPE_BACKSLASH) |
501 | *(t++) = '"'; | |
502 | else | |
503 | *(t++) = '\''; | |
4f5dd394 LP |
504 | *t = 0; |
505 | ||
506 | return r; | |
507 | } |