src/common/string-utils/string-utils.cpp

   1 /*
   2  * Copyright (C) 2017 Philippe Proulx <pproulx@efficios.com>
   3  *
   4  * SPDX-License-Identifier: LGPL-2.1-only
   5  *
   6  */
   7
   8 #define _LGPL_SOURCE
   9 #include <stdlib.h>
  10 #include <stdio.h>
  11 #include <string.h>
  12 #include <stdbool.h>
  13 #include <type_traits>
  14 #include <assert.h>
  15 #include <errno.h>
  16 #include <stdarg.h>
  17
  18 #include "string-utils.hpp"
  19 #include "../macros.hpp"
  20
  21 enum star_glob_pattern_type_flags {
  22         STAR_GLOB_PATTERN_TYPE_FLAG_NONE = 0,
  23         STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN = 1,
  24         STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY = 2,
  25 };
  26
  27 static
  28 star_glob_pattern_type_flags &operator|=(star_glob_pattern_type_flags &l,
  29                 star_glob_pattern_type_flags r)
  30 {
  31         using T = std::underlying_type<star_glob_pattern_type_flags>::type;
  32         l = static_cast<star_glob_pattern_type_flags> (
  33                 static_cast<T> (l) | static_cast<T> (r));
  34         return l;
  35 }
  36
  37 /*
  38  * Normalizes the star-only globbing pattern `pattern`, that is, crushes
  39  * consecutive `*` characters into a single `*`, avoiding `\*`.
  40  */
  41 void strutils_normalize_star_glob_pattern(char *pattern)
  42 {
  43         const char *p;
  44         char *np;
  45         bool got_star = false;
  46
  47         LTTNG_ASSERT(pattern);
  48
  49         for (p = pattern, np = pattern; *p != '\0'; p++) {
  50                 switch (*p) {
  51                 case '*':
  52                         if (got_star) {
  53                                 /* Avoid consecutive stars. */
  54                                 continue;
  55                         }
  56
  57                         got_star = true;
  58                         break;
  59                 case '\\':
  60                         /* Copy backslash character. */
  61                         *np = *p;
  62                         np++;
  63                         p++;
  64
  65                         if (*p == '\0') {
  66                                 goto end;
  67                         }
  68
  69                         /* fall through */
  70                 default:
  71                         got_star = false;
  72                         break;
  73                 }
  74
  75                 /* Copy single character. */
  76                 *np = *p;
  77                 np++;
  78         }
  79
  80 end:
  81         *np = '\0';
  82 }
  83
  84 static
  85 enum star_glob_pattern_type_flags strutils_test_glob_pattern(const char *pattern)
  86 {
  87         enum star_glob_pattern_type_flags ret =
  88                 STAR_GLOB_PATTERN_TYPE_FLAG_NONE;
  89         const char *p;
  90
  91         LTTNG_ASSERT(pattern);
  92
  93         for (p = pattern; *p != '\0'; p++) {
  94                 switch (*p) {
  95                 case '*':
  96                         ret = STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN;
  97
  98                         if (p[1] == '\0') {
  99                                 ret |= STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY;
 100                         }
 101
 102                         goto end;
 103                 case '\\':
 104                         p++;
 105
 106                         if (*p == '\0') {
 107                                 goto end;
 108                         }
 109                         break;
 110                 default:
 111                         break;
 112                 }
 113         }
 114
 115 end:
 116         return ret;
 117 }
 118
 119 /*
 120  * Returns true if `pattern` is a star-only globbing pattern, that is,
 121  * it contains at least one non-escaped `*`.
 122  */
 123 bool strutils_is_star_glob_pattern(const char *pattern)
 124 {
 125         return strutils_test_glob_pattern(pattern) &
 126                 STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN;
 127 }
 128
 129 /*
 130  * Returns true if `pattern` is a globbing pattern with a globbing,
 131  * non-escaped star only at its very end.
 132  */
 133 bool strutils_is_star_at_the_end_only_glob_pattern(const char *pattern)
 134 {
 135         return strutils_test_glob_pattern(pattern) &
 136                 STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY;
 137 }
 138
 139 /*
 140  * Unescapes the input string `input`, that is, in a `\x` sequence,
 141  * removes `\`. If `only_char` is not 0, only this character is
 142  * escaped.
 143  */
 144 char *strutils_unescape_string(const char *input, char only_char)
 145 {
 146         char *output;
 147         char *o;
 148         const char *i;
 149
 150         LTTNG_ASSERT(input);
 151         output = (char *) zmalloc(strlen(input) + 1);
 152         if (!output) {
 153                 goto end;
 154         }
 155
 156         for (i = input, o = output; *i != '\0'; i++) {
 157                 switch (*i) {
 158                 case '\\':
 159                         if (only_char && i[1] != only_char) {
 160                                 break;
 161                         }
 162
 163                         i++;
 164
 165                         if (*i == '\0') {
 166                                 /* Copy last `\`. */
 167                                 *o = '\\';
 168                                 o++;
 169                                 goto end;
 170                         }
 171                 default:
 172                         break;
 173                 }
 174
 175                 /* Copy single character. */
 176                 *o = *i;
 177                 o++;
 178         }
 179
 180 end:
 181         return output;
 182 }
 183
 184 /*
 185  * Frees a null-terminated array of strings, including each contained
 186  * string.
 187  */
 188 void strutils_free_null_terminated_array_of_strings(char **array)
 189 {
 190         char **item;
 191
 192         if (!array) {
 193                 return;
 194         }
 195
 196         for (item = array; *item; item++) {
 197                 free(*item);
 198         }
 199
 200         free(array);
 201 }
 202
 203 /*
 204  * Splits the input string `input` using the given delimiter `delim`.
 205  *
 206  * The return value is a dynamic pointer array that is assumed to be empty. The
 207  * array must be discarded by the caller by invoking
 208  * lttng_dynamic_pointer_array_reset().
 209  *
 210  * Empty substrings are part of the result. For example:
 211  *
 212  *     Input: ,hello,,there,
 213  *     Result:
 214  *       ``
 215  *       `hello`
 216  *       ``
 217  *       `there`
 218  *       ``
 219  *
 220  * If `escape_delim` is true, then `\,`, where `,` is the delimiter,
 221  * escapes the delimiter and is copied as `,` only in the resulting
 222  * substring. For example:
 223  *
 224  *     Input: hello\,world,zoom,\,hi
 225  *     Result:
 226  *       `hello,world`
 227  *       `zoom`
 228  *       `,hi`
 229  *
 230  * Other characters are not escaped (this is the caller's job if
 231  * needed). However they are considering during the parsing, that is,
 232  * `\x`, where `x` is any character, is copied as is to the resulting
 233  * substring, e.g.:
 234  *
 235  *     Input: hello\,wo\rld\\,zoom\,
 236  *     Result:
 237  *       `hello,wo\rld\\`
 238  *       `zoom,`
 239  *
 240  * If `escape_delim` is false, nothing at all is escaped, and `delim`,
 241  * when found in `input`, is always a delimiter, e.g.:
 242  *
 243  *     Input: hello\,world,zoom,\,hi
 244  *     Result:
 245  *       `hello\`
 246  *       `world`
 247  *       `zoom`
 248  *       `\`
 249  *       `hi`
 250  *
 251  * Returns -1 if there's an error.
 252  */
 253 int strutils_split(const char *input,
 254                 char delim,
 255                 bool escape_delim,
 256                 struct lttng_dynamic_pointer_array *out_strings)
 257 {
 258         int ret;
 259         size_t at;
 260         size_t number_of_substrings = 1;
 261         size_t longest_substring_len = 0;
 262         const char *s;
 263         const char *last;
 264
 265         LTTNG_ASSERT(input);
 266         LTTNG_ASSERT(!(escape_delim && delim == '\\'));
 267         LTTNG_ASSERT(delim != '\0');
 268         lttng_dynamic_pointer_array_init(out_strings, free);
 269
 270         /* First pass: count the number of substrings. */
 271         for (s = input, last = input - 1; *s != '\0'; s++) {
 272                 if (escape_delim && *s == '\\') {
 273                         /* Ignore following (escaped) character. */
 274                         s++;
 275
 276                         if (*s == '\0') {
 277                                 break;
 278                         }
 279
 280                         continue;
 281                 }
 282
 283                 if (*s == delim) {
 284                         size_t last_len = s - last - 1;
 285                         last = s;
 286                         number_of_substrings++;
 287
 288                         if (last_len > longest_substring_len) {
 289                                 longest_substring_len = last_len;
 290                         }
 291                 }
 292         }
 293
 294         if ((s - last - 1) > longest_substring_len) {
 295                 longest_substring_len = s - last - 1;
 296         }
 297
 298         /* Second pass: actually split and copy substrings. */
 299         for (at = 0, s = input; at < number_of_substrings; at++) {
 300                 const char *ss;
 301                 char *d;
 302                 char *substring = (char *) zmalloc(longest_substring_len + 1);
 303
 304                 if (!substring) {
 305                         goto error;
 306                 }
 307
 308                 ret = lttng_dynamic_pointer_array_add_pointer(
 309                                 out_strings, substring);
 310                 if (ret) {
 311                         free(substring);
 312                         goto error;
 313                 }
 314
 315                 /*
 316                  * Copy characters to substring until we find the next
 317                  * delimiter or the end of the input string.
 318                  */
 319                 for (ss = s, d = substring; *ss != '\0'; ss++) {
 320                         if (escape_delim && *ss == '\\') {
 321                                 if (ss[1] == delim) {
 322                                         /*
 323                                          * '\' followed by delimiter and
 324                                          * we need to escape this ('\'
 325                                          * won't be part of the
 326                                          * resulting substring).
 327                                          */
 328                                         ss++;
 329                                         *d = *ss;
 330                                         d++;
 331                                         continue;
 332                                 } else {
 333                                         /*
 334                                          * Copy '\' and the following
 335                                          * character.
 336                                          */
 337                                         *d = *ss;
 338                                         d++;
 339                                         ss++;
 340
 341                                         if (*ss == '\0') {
 342                                                 break;
 343                                         }
 344                                 }
 345                         } else if (*ss == delim) {
 346                                 /* We're done with this substring. */
 347                                 break;
 348                         }
 349
 350                         *d = *ss;
 351                         d++;
 352                 }
 353
 354                 /* Next substring starts after the last delimiter. */
 355                 s = ss + 1;
 356         }
 357
 358         ret = 0;
 359         goto end;
 360
 361 error:
 362         ret = -1;
 363 end:
 364         return ret;
 365 }
 366
 367 size_t strutils_array_of_strings_len(char * const *array)
 368 {
 369         char * const *item;
 370         size_t count = 0;
 371
 372         LTTNG_ASSERT(array);
 373
 374         for (item = array; *item; item++) {
 375                 count++;
 376         }
 377
 378         return count;
 379 }
 380
 381 int strutils_append_str(char **s, const char *append)
 382 {
 383         char *old = *s;
 384         char *new_str;
 385         size_t oldlen = (old == NULL) ? 0 : strlen(old);
 386         size_t appendlen = strlen(append);
 387
 388         new_str = (char *) zmalloc(oldlen + appendlen + 1);
 389         if (!new_str) {
 390                 return -ENOMEM;
 391         }
 392         if (oldlen) {
 393                 strcpy(new_str, old);
 394         }
 395         strcat(new_str, append);
 396         *s = new_str;
 397         free(old);
 398         return 0;
 399 }
 400
 401 int strutils_appendf(char **s, const char *fmt, ...)
 402 {
 403         char *new_str;
 404         size_t oldlen = (*s) ? strlen(*s) : 0;
 405         int ret;
 406         va_list args;
 407
 408         /* Compute length of formatted string we append. */
 409         va_start(args, fmt);
 410         ret = vsnprintf(NULL, 0, fmt, args);
 411         va_end(args);
 412
 413         if (ret == -1) {
 414                 goto end;
 415         }
 416
 417         /* Allocate space for old string + new string + \0. */
 418         new_str = (char *) zmalloc(oldlen + ret + 1);
 419         if (!new_str) {
 420                 ret = -ENOMEM;
 421                 goto end;
 422         }
 423
 424         /* Copy old string, if there was one. */
 425         if (oldlen) {
 426                 strcpy(new_str, *s);
 427         }
 428
 429         /* Format new string in-place. */
 430         va_start(args, fmt);
 431         ret = vsprintf(&new_str[oldlen], fmt, args);
 432         va_end(args);
 433
 434         if (ret == -1) {
 435                 ret = -1;
 436                 goto end;
 437         }
 438
 439         free(*s);
 440         *s = new_str;
 441         new_str = NULL;
 442
 443 end:
 444         return ret;
 445 }