src/basic/extract-word.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   2 /***
   3   Copyright 2010 Lennart Poettering
   4 ***/
   5
   6 #include <errno.h>
   7 #include <stdarg.h>
   8 #include <stdbool.h>
   9 #include <stddef.h>
  10 #include <stdint.h>
  11 #include <stdlib.h>
  12 #include <string.h>
  13 #include <syslog.h>
  14
  15 #include "alloc-util.h"
  16 #include "escape.h"
  17 #include "extract-word.h"
  18 #include "log.h"
  19 #include "macro.h"
  20 #include "string-util.h"
  21 #include "utf8.h"
  22
  23 int extract_first_word(const char **p, char **ret, const char *separators, ExtractFlags flags) {
  24         _cleanup_free_ char *s = NULL;
  25         size_t allocated = 0, sz = 0;
  26         char c;
  27         int r;
  28
  29         char quote = 0;                 /* 0 or ' or " */
  30         bool backslash = false;         /* whether we've just seen a backslash */
  31
  32         assert(p);
  33         assert(ret);
  34
  35         /* Bail early if called after last value or with no input */
  36         if (!*p)
  37                 goto finish;
  38         c = **p;
  39
  40         if (!separators)
  41                 separators = WHITESPACE;
  42
  43         /* Parses the first word of a string, and returns it in
  44          * *ret. Removes all quotes in the process. When parsing fails
  45          * (because of an uneven number of quotes or similar), leaves
  46          * the pointer *p at the first invalid character. */
  47
  48         if (flags & EXTRACT_DONT_COALESCE_SEPARATORS)
  49                 if (!GREEDY_REALLOC(s, allocated, sz+1))
  50                         return -ENOMEM;
  51
  52         for (;; (*p)++, c = **p) {
  53                 if (c == 0)
  54                         goto finish_force_terminate;
  55                 else if (strchr(separators, c)) {
  56                         if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
  57                                 (*p)++;
  58                                 goto finish_force_next;
  59                         }
  60                 } else {
  61                         /* We found a non-blank character, so we will always
  62                          * want to return a string (even if it is empty),
  63                          * allocate it here. */
  64                         if (!GREEDY_REALLOC(s, allocated, sz+1))
  65                                 return -ENOMEM;
  66                         break;
  67                 }
  68         }
  69
  70         for (;; (*p)++, c = **p) {
  71                 if (backslash) {
  72                         if (!GREEDY_REALLOC(s, allocated, sz+7))
  73                                 return -ENOMEM;
  74
  75                         if (c == 0) {
  76                                 if ((flags & EXTRACT_CUNESCAPE_RELAX) &&
  77                                     (!quote || flags & EXTRACT_RELAX)) {
  78                                         /* If we find an unquoted trailing backslash and we're in
  79                                          * EXTRACT_CUNESCAPE_RELAX mode, keep it verbatim in the
  80                                          * output.
  81                                          *
  82                                          * Unbalanced quotes will only be allowed in EXTRACT_RELAX
  83                                          * mode, EXTRACT_CUNESCAPE_RELAX mode does not allow them.
  84                                          */
  85                                         s[sz++] = '\\';
  86                                         goto finish_force_terminate;
  87                                 }
  88                                 if (flags & EXTRACT_RELAX)
  89                                         goto finish_force_terminate;
  90                                 return -EINVAL;
  91                         }
  92
  93                         if (flags & EXTRACT_CUNESCAPE) {
  94                                 bool eight_bit = false;
  95                                 char32_t u;
  96
  97                                 r = cunescape_one(*p, (size_t) -1, &u, &eight_bit);
  98                                 if (r < 0) {
  99                                         if (flags & EXTRACT_CUNESCAPE_RELAX) {
 100                                                 s[sz++] = '\\';
 101                                                 s[sz++] = c;
 102                                         } else
 103                                                 return -EINVAL;
 104                                 } else {
 105                                         (*p) += r - 1;
 106
 107                                         if (eight_bit)
 108                                                 s[sz++] = u;
 109                                         else
 110                                                 sz += utf8_encode_unichar(s + sz, u);
 111                                 }
 112                         } else
 113                                 s[sz++] = c;
 114
 115                         backslash = false;
 116
 117                 } else if (quote) {     /* inside either single or double quotes */
 118                         for (;; (*p)++, c = **p) {
 119                                 if (c == 0) {
 120                                         if (flags & EXTRACT_RELAX)
 121                                                 goto finish_force_terminate;
 122                                         return -EINVAL;
 123                                 } else if (c == quote) {        /* found the end quote */
 124                                         quote = 0;
 125                                         break;
 126                                 } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
 127                                         backslash = true;
 128                                         break;
 129                                 } else {
 130                                         if (!GREEDY_REALLOC(s, allocated, sz+2))
 131                                                 return -ENOMEM;
 132
 133                                         s[sz++] = c;
 134                                 }
 135                         }
 136
 137                 } else {
 138                         for (;; (*p)++, c = **p) {
 139                                 if (c == 0)
 140                                         goto finish_force_terminate;
 141                                 else if (IN_SET(c, '\'', '"') && (flags & EXTRACT_QUOTES)) {
 142                                         quote = c;
 143                                         break;
 144                                 } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
 145                                         backslash = true;
 146                                         break;
 147                                 } else if (strchr(separators, c)) {
 148                                         if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
 149                                                 (*p)++;
 150                                                 goto finish_force_next;
 151                                         }
 152                                         /* Skip additional coalesced separators. */
 153                                         for (;; (*p)++, c = **p) {
 154                                                 if (c == 0)
 155                                                         goto finish_force_terminate;
 156                                                 if (!strchr(separators, c))
 157                                                         break;
 158                                         }
 159                                         goto finish;
 160
 161                                 } else {
 162                                         if (!GREEDY_REALLOC(s, allocated, sz+2))
 163                                                 return -ENOMEM;
 164
 165                                         s[sz++] = c;
 166                                 }
 167                         }
 168                 }
 169         }
 170
 171 finish_force_terminate:
 172         *p = NULL;
 173 finish:
 174         if (!s) {
 175                 *p = NULL;
 176                 *ret = NULL;
 177                 return 0;
 178         }
 179
 180 finish_force_next:
 181         s[sz] = 0;
 182         *ret = TAKE_PTR(s);
 183
 184         return 1;
 185 }
 186
 187 #if 0 /// UNNEEDED by elogind
 188 int extract_first_word_and_warn(
 189                 const char **p,
 190                 char **ret,
 191                 const char *separators,
 192                 ExtractFlags flags,
 193                 const char *unit,
 194                 const char *filename,
 195                 unsigned line,
 196                 const char *rvalue) {
 197
 198         /* Try to unquote it, if it fails, warn about it and try again
 199          * but this time using EXTRACT_CUNESCAPE_RELAX to keep the
 200          * backslashes verbatim in invalid escape sequences. */
 201
 202         const char *save;
 203         int r;
 204
 205         save = *p;
 206         r = extract_first_word(p, ret, separators, flags);
 207         if (r >= 0)
 208                 return r;
 209
 210         if (r == -EINVAL && !(flags & EXTRACT_CUNESCAPE_RELAX)) {
 211
 212                 /* Retry it with EXTRACT_CUNESCAPE_RELAX. */
 213                 *p = save;
 214                 r = extract_first_word(p, ret, separators, flags|EXTRACT_CUNESCAPE_RELAX);
 215                 if (r >= 0) {
 216                         /* It worked this time, hence it must have been an invalid escape sequence. */
 217                         log_syntax(unit, LOG_WARNING, filename, line, EINVAL, "Ignoring unknown escape sequences: \"%s\"", *ret);
 218                         return r;
 219                 }
 220
 221                 /* If it's still EINVAL; then it must be unbalanced quoting, report this. */
 222                 if (r == -EINVAL)
 223                         return log_syntax(unit, LOG_ERR, filename, line, r, "Unbalanced quoting, ignoring: \"%s\"", rvalue);
 224         }
 225
 226         /* Can be any error, report it */
 227         return log_syntax(unit, LOG_ERR, filename, line, r, "Unable to decode word \"%s\", ignoring: %m", rvalue);
 228 }
 229
 230 /* We pass ExtractFlags as unsigned int (to avoid undefined behaviour when passing
 231  * an object that undergoes default argument promotion as an argument to va_start).
 232  * Let's make sure that ExtractFlags fits into an unsigned int. */
 233 assert_cc(sizeof(enum ExtractFlags) <= sizeof(unsigned));
 234
 235 int extract_many_words(const char **p, const char *separators, unsigned flags, ...) {
 236         va_list ap;
 237         char **l;
 238         int n = 0, i, c, r;
 239
 240         /* Parses a number of words from a string, stripping any
 241          * quotes if necessary. */
 242
 243         assert(p);
 244
 245         /* Count how many words are expected */
 246         va_start(ap, flags);
 247         for (;;) {
 248                 if (!va_arg(ap, char **))
 249                         break;
 250                 n++;
 251         }
 252         va_end(ap);
 253
 254         if (n <= 0)
 255                 return 0;
 256
 257         /* Read all words into a temporary array */
 258         l = newa0(char*, n);
 259         for (c = 0; c < n; c++) {
 260
 261                 r = extract_first_word(p, &l[c], separators, flags);
 262                 if (r < 0) {
 263                         int j;
 264
 265                         for (j = 0; j < c; j++)
 266                                 free(l[j]);
 267
 268                         return r;
 269                 }
 270
 271                 if (r == 0)
 272                         break;
 273         }
 274
 275         /* If we managed to parse all words, return them in the passed
 276          * in parameters */
 277         va_start(ap, flags);
 278         for (i = 0; i < n; i++) {
 279                 char **v;
 280
 281                 v = va_arg(ap, char **);
 282                 assert(v);
 283
 284                 *v = l[i];
 285         }
 286         va_end(ap);
 287
 288         return c;
 289 }
 290 #endif // 0