Line data Source code
1 : /* SPDX-License-Identifier: LGPL-2.1+ */
2 :
3 : #include <errno.h>
4 : #include <stdarg.h>
5 : #include <stdbool.h>
6 : #include <stddef.h>
7 : #include <stdint.h>
8 : #include <stdlib.h>
9 : #include <string.h>
10 : #include <syslog.h>
11 :
12 : #include "alloc-util.h"
13 : #include "escape.h"
14 : #include "extract-word.h"
15 : #include "log.h"
16 : #include "macro.h"
17 : #include "string-util.h"
18 : #include "utf8.h"
19 :
20 14697 : int extract_first_word(const char **p, char **ret, const char *separators, ExtractFlags flags) {
21 14697 : _cleanup_free_ char *s = NULL;
22 14697 : size_t allocated = 0, sz = 0;
23 : char c;
24 : int r;
25 :
26 14697 : char quote = 0; /* 0 or ' or " */
27 14697 : bool backslash = false; /* whether we've just seen a backslash */
28 :
29 14697 : assert(p);
30 14697 : assert(ret);
31 :
32 : /* Bail early if called after last value or with no input */
33 14697 : if (!*p)
34 2573 : goto finish;
35 12124 : c = **p;
36 :
37 12124 : if (!separators)
38 5887 : separators = WHITESPACE;
39 :
40 : /* Parses the first word of a string, and returns it in
41 : * *ret. Removes all quotes in the process. When parsing fails
42 : * (because of an uneven number of quotes or similar), leaves
43 : * the pointer *p at the first invalid character. */
44 :
45 12124 : if (flags & EXTRACT_DONT_COALESCE_SEPARATORS)
46 2196 : if (!GREEDY_REALLOC(s, allocated, sz+1))
47 0 : return -ENOMEM;
48 :
49 99 : for (;; (*p)++, c = **p) {
50 12223 : if (c == 0)
51 24 : goto finish_force_terminate;
52 12199 : else if (strchr(separators, c)) {
53 105 : if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
54 6 : (*p)++;
55 6 : goto finish_force_next;
56 : }
57 : } else {
58 : /* We found a non-blank character, so we will always
59 : * want to return a string (even if it is empty),
60 : * allocate it here. */
61 12094 : if (!GREEDY_REALLOC(s, allocated, sz+1))
62 0 : return -ENOMEM;
63 12094 : break;
64 : }
65 : }
66 :
67 430 : for (;; (*p)++, c = **p) {
68 12524 : if (backslash) {
69 91 : if (!GREEDY_REALLOC(s, allocated, sz+7))
70 0 : return -ENOMEM;
71 :
72 91 : if (c == 0) {
73 27 : if ((flags & EXTRACT_CUNESCAPE_RELAX) &&
74 6 : (!quote || flags & EXTRACT_RELAX)) {
75 : /* If we find an unquoted trailing backslash and we're in
76 : * EXTRACT_CUNESCAPE_RELAX mode, keep it verbatim in the
77 : * output.
78 : *
79 : * Unbalanced quotes will only be allowed in EXTRACT_RELAX
80 : * mode, EXTRACT_CUNESCAPE_RELAX mode does not allow them.
81 : */
82 8 : s[sz++] = '\\';
83 8 : goto finish_force_terminate;
84 : }
85 19 : if (flags & EXTRACT_RELAX)
86 5 : goto finish_force_terminate;
87 14 : return -EINVAL;
88 : }
89 :
90 64 : if (flags & EXTRACT_CUNESCAPE) {
91 48 : bool eight_bit = false;
92 : char32_t u;
93 :
94 48 : r = cunescape_one(*p, (size_t) -1, &u, &eight_bit);
95 48 : if (r < 0) {
96 22 : if (flags & EXTRACT_CUNESCAPE_RELAX) {
97 14 : s[sz++] = '\\';
98 14 : s[sz++] = c;
99 : } else
100 8 : return -EINVAL;
101 : } else {
102 26 : (*p) += r - 1;
103 :
104 26 : if (eight_bit)
105 5 : s[sz++] = u;
106 : else
107 21 : sz += utf8_encode_unichar(s + sz, u);
108 : }
109 : } else
110 16 : s[sz++] = c;
111 :
112 56 : backslash = false;
113 :
114 12433 : } else if (quote) { /* inside either single or double quotes */
115 1037 : for (;; (*p)++, c = **p) {
116 1208 : if (c == 0) {
117 26 : if (flags & EXTRACT_RELAX)
118 3 : goto finish_force_terminate;
119 23 : return -EINVAL;
120 1182 : } else if (c == quote) { /* found the end quote */
121 122 : quote = 0;
122 122 : break;
123 1060 : } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
124 23 : backslash = true;
125 23 : break;
126 : } else {
127 1037 : if (!GREEDY_REALLOC(s, allocated, sz+2))
128 0 : return -ENOMEM;
129 :
130 1037 : s[sz++] = c;
131 : }
132 : }
133 :
134 : } else {
135 192211 : for (;; (*p)++, c = **p) {
136 204473 : if (c == 0)
137 2726 : goto finish_force_terminate;
138 201747 : else if (IN_SET(c, '\'', '"') && (flags & EXTRACT_UNQUOTE)) {
139 161 : quote = c;
140 161 : break;
141 201586 : } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
142 68 : backslash = true;
143 68 : break;
144 201518 : } else if (strchr(separators, c)) {
145 9307 : if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
146 1750 : (*p)++;
147 1750 : goto finish_force_next;
148 : }
149 : /* Skip additional coalesced separators. */
150 10276 : for (;; (*p)++, c = **p) {
151 17833 : if (c == 0)
152 44 : goto finish_force_terminate;
153 17789 : if (!strchr(separators, c))
154 7513 : break;
155 : }
156 7513 : goto finish;
157 :
158 : } else {
159 192211 : if (!GREEDY_REALLOC(s, allocated, sz+2))
160 0 : return -ENOMEM;
161 :
162 192211 : s[sz++] = c;
163 : }
164 : }
165 : }
166 : }
167 :
168 2810 : finish_force_terminate:
169 2810 : *p = NULL;
170 12896 : finish:
171 12896 : if (!s) {
172 2593 : *p = NULL;
173 2593 : *ret = NULL;
174 2593 : return 0;
175 : }
176 :
177 10303 : finish_force_next:
178 12059 : s[sz] = 0;
179 12059 : *ret = TAKE_PTR(s);
180 :
181 12059 : return 1;
182 : }
183 :
184 141 : int extract_first_word_and_warn(
185 : const char **p,
186 : char **ret,
187 : const char *separators,
188 : ExtractFlags flags,
189 : const char *unit,
190 : const char *filename,
191 : unsigned line,
192 : const char *rvalue) {
193 :
194 : /* Try to unquote it, if it fails, warn about it and try again
195 : * but this time using EXTRACT_CUNESCAPE_RELAX to keep the
196 : * backslashes verbatim in invalid escape sequences. */
197 :
198 : const char *save;
199 : int r;
200 :
201 141 : save = *p;
202 141 : r = extract_first_word(p, ret, separators, flags);
203 141 : if (r >= 0)
204 125 : return r;
205 :
206 16 : if (r == -EINVAL && !(flags & EXTRACT_CUNESCAPE_RELAX)) {
207 :
208 : /* Retry it with EXTRACT_CUNESCAPE_RELAX. */
209 16 : *p = save;
210 16 : r = extract_first_word(p, ret, separators, flags|EXTRACT_CUNESCAPE_RELAX);
211 16 : if (r >= 0) {
212 : /* It worked this time, hence it must have been an invalid escape sequence. */
213 9 : log_syntax(unit, LOG_WARNING, filename, line, EINVAL, "Ignoring unknown escape sequences: \"%s\"", *ret);
214 9 : return r;
215 : }
216 :
217 : /* If it's still EINVAL; then it must be unbalanced quoting, report this. */
218 7 : if (r == -EINVAL)
219 7 : return log_syntax(unit, LOG_ERR, filename, line, r, "Unbalanced quoting, ignoring: \"%s\"", rvalue);
220 : }
221 :
222 : /* Can be any error, report it */
223 0 : return log_syntax(unit, LOG_ERR, filename, line, r, "Unable to decode word \"%s\", ignoring: %m", rvalue);
224 : }
225 :
226 : /* We pass ExtractFlags as unsigned int (to avoid undefined behaviour when passing
227 : * an object that undergoes default argument promotion as an argument to va_start).
228 : * Let's make sure that ExtractFlags fits into an unsigned int. */
229 : assert_cc(sizeof(enum ExtractFlags) <= sizeof(unsigned));
230 :
231 90 : int extract_many_words(const char **p, const char *separators, unsigned flags, ...) {
232 : va_list ap;
233 : char **l;
234 90 : int n = 0, i, c, r;
235 :
236 : /* Parses a number of words from a string, stripping any
237 : * quotes if necessary. */
238 :
239 90 : assert(p);
240 :
241 : /* Count how many words are expected */
242 90 : va_start(ap, flags);
243 : for (;;) {
244 599 : if (!va_arg(ap, char **))
245 90 : break;
246 509 : n++;
247 : }
248 90 : va_end(ap);
249 :
250 90 : if (n <= 0)
251 1 : return 0;
252 :
253 : /* Read all words into a temporary array */
254 89 : l = newa0(char*, n);
255 482 : for (c = 0; c < n; c++) {
256 :
257 429 : r = extract_first_word(p, &l[c], separators, flags);
258 429 : if (r < 0) {
259 : int j;
260 :
261 4 : for (j = 0; j < c; j++)
262 2 : free(l[j]);
263 :
264 2 : return r;
265 : }
266 :
267 427 : if (r == 0)
268 34 : break;
269 : }
270 :
271 : /* If we managed to parse all words, return them in the passed
272 : * in parameters */
273 87 : va_start(ap, flags);
274 584 : for (i = 0; i < n; i++) {
275 : char **v;
276 :
277 497 : v = va_arg(ap, char **);
278 497 : assert(v);
279 :
280 497 : *v = l[i];
281 : }
282 87 : va_end(ap);
283 :
284 87 : return c;
285 : }
|
