src/basic/extract-word.c

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