src/common/string-utils/string-utils.c

   1 /*
   2  * Copyright (C) 2017 Philippe Proulx <pproulx@efficios.com>
   3  *
   4  * SPDX-License-Identifier: GPL-2.0-only
   5  *
   6  */
   7
   8 #define _LGPL_SOURCE
   9 #include <stdlib.h>
  10 #include <string.h>
  11 #include <stdbool.h>
  12 #include <assert.h>
  13
  14 #include "string-utils.h"
  15 #include "../macros.h"
  16
  17 enum star_glob_pattern_type_flags {
  18         STAR_GLOB_PATTERN_TYPE_FLAG_NONE = 0,
  19         STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN = 1,
  20         STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY = 2,
  21 };
  22
  23 /*
  24  * Normalizes the star-only globbing pattern `pattern`, that is, crushes
  25  * consecutive `*` characters into a single `*`, avoiding `\*`.
  26  */
  27 LTTNG_HIDDEN
  28 void strutils_normalize_star_glob_pattern(char *pattern)
  29 {
  30         const char *p;
  31         char *np;
  32         bool got_star = false;
  33
  34         assert(pattern);
  35
  36         for (p = pattern, np = pattern; *p != '\0'; p++) {
  37                 switch (*p) {
  38                 case '*':
  39                         if (got_star) {
  40                                 /* Avoid consecutive stars. */
  41                                 continue;
  42                         }
  43
  44                         got_star = true;
  45                         break;
  46                 case '\\':
  47                         /* Copy backslash character. */
  48                         *np = *p;
  49                         np++;
  50                         p++;
  51
  52                         if (*p == '\0') {
  53                                 goto end;
  54                         }
  55
  56                         /* Fall through default case. */
  57                 default:
  58                         got_star = false;
  59                         break;
  60                 }
  61
  62                 /* Copy single character. */
  63                 *np = *p;
  64                 np++;
  65         }
  66
  67 end:
  68         *np = '\0';
  69 }
  70
  71 static
  72 enum star_glob_pattern_type_flags strutils_test_glob_pattern(const char *pattern)
  73 {
  74         enum star_glob_pattern_type_flags ret =
  75                 STAR_GLOB_PATTERN_TYPE_FLAG_NONE;
  76         const char *p;
  77
  78         assert(pattern);
  79
  80         for (p = pattern; *p != '\0'; p++) {
  81                 switch (*p) {
  82                 case '*':
  83                         ret = STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN;
  84
  85                         if (p[1] == '\0') {
  86                                 ret |= STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY;
  87                         }
  88
  89                         goto end;
  90                 case '\\':
  91                         p++;
  92
  93                         if (*p == '\0') {
  94                                 goto end;
  95                         }
  96                         break;
  97                 default:
  98                         break;
  99                 }
 100         }
 101
 102 end:
 103         return ret;
 104 }
 105
 106 /*
 107  * Returns true if `pattern` is a star-only globbing pattern, that is,
 108  * it contains at least one non-escaped `*`.
 109  */
 110 LTTNG_HIDDEN
 111 bool strutils_is_star_glob_pattern(const char *pattern)
 112 {
 113         return strutils_test_glob_pattern(pattern) &
 114                 STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN;
 115 }
 116
 117 /*
 118  * Returns true if `pattern` is a globbing pattern with a globbing,
 119  * non-escaped star only at its very end.
 120  */
 121 LTTNG_HIDDEN
 122 bool strutils_is_star_at_the_end_only_glob_pattern(const char *pattern)
 123 {
 124         return strutils_test_glob_pattern(pattern) &
 125                 STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY;
 126 }
 127
 128 /*
 129  * Unescapes the input string `input`, that is, in a `\x` sequence,
 130  * removes `\`. If `only_char` is not 0, only this character is
 131  * escaped.
 132  */
 133 LTTNG_HIDDEN
 134 char *strutils_unescape_string(const char *input, char only_char)
 135 {
 136         char *output;
 137         char *o;
 138         const char *i;
 139
 140         assert(input);
 141         output = zmalloc(strlen(input) + 1);
 142         if (!output) {
 143                 goto end;
 144         }
 145
 146         for (i = input, o = output; *i != '\0'; i++) {
 147                 switch (*i) {
 148                 case '\\':
 149                         if (only_char && i[1] != only_char) {
 150                                 break;
 151                         }
 152
 153                         i++;
 154
 155                         if (*i == '\0') {
 156                                 /* Copy last `\`. */
 157                                 *o = '\\';
 158                                 o++;
 159                                 goto end;
 160                         }
 161                 default:
 162                         break;
 163                 }
 164
 165                 /* Copy single character. */
 166                 *o = *i;
 167                 o++;
 168         }
 169
 170 end:
 171         return output;
 172 }
 173
 174 /*
 175  * Frees a null-terminated array of strings, including each contained
 176  * string.
 177  */
 178 LTTNG_HIDDEN
 179 void strutils_free_null_terminated_array_of_strings(char **array)
 180 {
 181         char **item;
 182
 183         if (!array) {
 184                 return;
 185         }
 186
 187         for (item = array; *item; item++) {
 188                 free(*item);
 189         }
 190
 191         free(array);
 192 }
 193
 194 /*
 195  * Splits the input string `input` using the given delimiter `delim`.
 196  *
 197  * The return value is an allocated null-terminated array of the
 198  * resulting substrings (also allocated). You can free this array and
 199  * its content with strutils_free_null_terminated_array_of_strings(). You
 200  * can get the number of substrings in it with
 201  * strutils_array_of_strings_len().
 202  *
 203  * Empty substrings are part of the result. For example:
 204  *
 205  *     Input: ,hello,,there,
 206  *     Result:
 207  *       ``
 208  *       `hello`
 209  *       ``
 210  *       `there`
 211  *       ``
 212  *
 213  * If `escape_delim` is true, then `\,`, where `,` is the delimiter,
 214  * escapes the delimiter and is copied as `,` only in the resulting
 215  * substring. For example:
 216  *
 217  *     Input: hello\,world,zoom,\,hi
 218  *     Result:
 219  *       `hello,world`
 220  *       `zoom`
 221  *       `,hi`
 222  *
 223  * Other characters are not escaped (this is the caller's job if
 224  * needed). However they are considering during the parsing, that is,
 225  * `\x`, where `x` is any character, is copied as is to the resulting
 226  * substring, e.g.:
 227  *
 228  *     Input: hello\,wo\rld\\,zoom\,
 229  *     Result:
 230  *       `hello,wo\rld\\`
 231  *       `zoom,`
 232  *
 233  * If `escape_delim` is false, nothing at all is escaped, and `delim`,
 234  * when found in `input`, is always a delimiter, e.g.:
 235  *
 236  *     Input: hello\,world,zoom,\,hi
 237  *     Result:
 238  *       `hello\`
 239  *       `world`
 240  *       `zoom`
 241  *       `\`
 242  *       `hi`
 243  *
 244  * Returns NULL if there's an error.
 245  */
 246 LTTNG_HIDDEN
 247 char **strutils_split(const char *input, char delim, bool escape_delim)
 248 {
 249         size_t at;
 250         size_t number_of_substrings = 1;
 251         size_t longest_substring_len = 0;
 252         const char *s;
 253         const char *last;
 254         char **substrings = NULL;
 255
 256         assert(input);
 257         assert(!(escape_delim && delim == '\\'));
 258         assert(delim != '\0');
 259
 260         /* First pass: count the number of substrings. */
 261         for (s = input, last = input - 1; *s != '\0'; s++) {
 262                 if (escape_delim && *s == '\\') {
 263                         /* Ignore following (escaped) character. */
 264                         s++;
 265
 266                         if (*s == '\0') {
 267                                 break;
 268                         }
 269
 270                         continue;
 271                 }
 272
 273                 if (*s == delim) {
 274                         size_t last_len = s - last - 1;
 275                         last = s;
 276                         number_of_substrings++;
 277
 278                         if (last_len > longest_substring_len) {
 279                                 longest_substring_len = last_len;
 280                         }
 281                 }
 282         }
 283
 284         if ((s - last - 1) > longest_substring_len) {
 285                 longest_substring_len = s - last - 1;
 286         }
 287
 288         substrings = calloc(number_of_substrings + 1, sizeof(*substrings));
 289         if (!substrings) {
 290                 goto error;
 291         }
 292
 293         /* Second pass: actually split and copy substrings. */
 294         for (at = 0, s = input; at < number_of_substrings; at++) {
 295                 const char *ss;
 296                 char *d;
 297
 298                 substrings[at] = zmalloc(longest_substring_len + 1);
 299                 if (!substrings[at]) {
 300                         goto error;
 301                 }
 302
 303                 /*
 304                  * Copy characters to substring until we find the next
 305                  * delimiter or the end of the input string.
 306                  */
 307                 for (ss = s, d = substrings[at]; *ss != '\0'; ss++) {
 308                         if (escape_delim && *ss == '\\') {
 309                                 if (ss[1] == delim) {
 310                                         /*
 311                                          * '\' followed by delimiter and
 312                                          * we need to escape this ('\'
 313                                          * won't be part of the
 314                                          * resulting substring).
 315                                          */
 316                                         ss++;
 317                                         *d = *ss;
 318                                         d++;
 319                                         continue;
 320                                 } else {
 321                                         /*
 322                                          * Copy '\' and the following
 323                                          * character.
 324                                          */
 325                                         *d = *ss;
 326                                         d++;
 327                                         ss++;
 328
 329                                         if (*ss == '\0') {
 330                                                 break;
 331                                         }
 332                                 }
 333                         } else if (*ss == delim) {
 334                                 /* We're done with this substring. */
 335                                 break;
 336                         }
 337
 338                         *d = *ss;
 339                         d++;
 340                 }
 341
 342                 /* Next substring starts after the last delimiter. */
 343                 s = ss + 1;
 344         }
 345
 346         goto end;
 347
 348 error:
 349         strutils_free_null_terminated_array_of_strings(substrings);
 350         substrings = NULL;
 351 end:
 352         return substrings;
 353 }
 354
 355 LTTNG_HIDDEN
 356 size_t strutils_array_of_strings_len(char * const *array)
 357 {
 358         char * const *item;
 359         size_t count = 0;
 360
 361         assert(array);
 362
 363         for (item = array; *item; item++) {
 364                 count++;
 365         }
 366
 367         return count;
 368 }