src/util/string.c

   1 /* String handling functions */
   2
   3 #ifdef HAVE_CONFIG_H
   4 #include "config.h"
   5 #endif
   6
   7 #ifndef _GNU_SOURCE
   8 #define _GNU_SOURCE /* XXX: fseeko, ftello */
   9 #endif
  10
  11 #include <ctype.h>
  12 #include <stdarg.h>
  13 #include <stdio.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16
  17 #include "elinks.h"
  18
  19 #include "util/conv.h"
  20 #include "util/error.h"
  21 #include "util/memdebug.h"
  22 #include "util/memory.h"
  23 #include "util/string.h"
  24 #include "util/snprintf.h"
  25
  26
  27 /* This file looks to be slowly being overloaded by a lot of various stuff,
  28  * like memory managment, stubs, tools, granular and non-granular strings,
  29  * struct string object... Perhaps util/memory.* and util/stubs.* (stubs.h
  30  * probably included in elinks.h, it's important enough) would be nice to
  31  * have. --pasky */
  32
  33
  34 #define string_assert(f, l, x, o) \
  35         if ((assert_failed = !(x))) { \
  36                 errfile = f, errline = l, \
  37                 elinks_internal("[" o "] assertion " #x " failed!"); \
  38         }
  39
  40 #ifdef DEBUG_MEMLEAK
  41
  42 unsigned char *
  43 debug_memacpy(unsigned char *f, int l, unsigned char *src, int len)
  44 {
  45         unsigned char *m;
  46
  47         string_assert(f, l, len >= 0, "memacpy");
  48         if_assert_failed len = 0;
  49
  50         m = debug_mem_alloc(f, l, len + 1);
  51         if (!m) return NULL;
  52
  53         if (src && len) memcpy(m, src, len);
  54         m[len] = '\0';
  55
  56         return m;
  57 }
  58
  59 unsigned char *
  60 debug_stracpy(unsigned char *f, int l, unsigned char *src)
  61 {
  62         string_assert(f, l, src, "stracpy");
  63         if_assert_failed return NULL;
  64
  65         return debug_memacpy(f, l, src, strlen(src));
  66 }
  67
  68 #else /* DEBUG_MEMLEAK */
  69
  70 unsigned char *
  71 memacpy(unsigned char *src, int len)
  72 {
  73         unsigned char *m;
  74
  75         assertm(len >= 0, "[memacpy]");
  76         if_assert_failed { len = 0; }
  77
  78         m = mem_alloc(len + 1);
  79         if (!m) return NULL;
  80
  81         if (src && len) memcpy(m, src, len);
  82         m[len] = 0;
  83
  84         return m;
  85 }
  86
  87 unsigned char *
  88 stracpy(unsigned char *src)
  89 {
  90         assertm(src, "[stracpy]");
  91         if_assert_failed return NULL;
  92
  93         return memacpy(src, strlen(src));
  94 }
  95
  96 #endif /* DEBUG_MEMLEAK */
  97
  98
  99 void
 100 add_to_strn(unsigned char **dst, unsigned char *src)
 101 {
 102         unsigned char *newdst;
 103         int dstlen;
 104         int srclen;
 105
 106         assertm(*dst && src, "[add_to_strn]");
 107         if_assert_failed return;
 108
 109         dstlen = strlen(*dst);
 110         srclen = strlen(src) + 1; /* Include the NUL char! */
 111         newdst = mem_realloc(*dst, dstlen + srclen);
 112         if (!newdst) return;
 113
 114         memcpy(newdst + dstlen, src, srclen);
 115         *dst = newdst;
 116 }
 117
 118 unsigned char *
 119 insert_in_string(unsigned char **dst, int pos, unsigned char *seq, int seqlen)
 120 {
 121         int dstlen = strlen(*dst);
 122         unsigned char *string = mem_realloc(*dst, dstlen + seqlen + 1);
 123
 124         if (!string) return NULL;
 125
 126         memmove(string + pos + seqlen, string + pos, dstlen - pos + 1);
 127         memcpy(string + pos, seq, seqlen);
 128         *dst = string;
 129
 130         return string;
 131 }
 132
 133 unsigned char *
 134 straconcat(unsigned char *str, ...)
 135 {
 136         va_list ap;
 137         unsigned char *a;
 138         unsigned char *s;
 139         unsigned int len;
 140
 141         assertm(str, "[straconcat]");
 142         if_assert_failed { return NULL; }
 143
 144         len = strlen(str);
 145         s = mem_alloc(len + 1);
 146         if (!s) return NULL;
 147
 148         if (len) memcpy(s, str, len);
 149
 150         va_start(ap, str);
 151         while ((a = va_arg(ap, unsigned char *))) {
 152                 unsigned int l = strlen(a);
 153                 unsigned char *ns;
 154
 155                 if (!l) continue;
 156
 157                 ns = mem_realloc(s, len + 1 + l);
 158                 if (!ns) {
 159                         mem_free(s);
 160                         va_end(ap);
 161                         return NULL;
 162                 }
 163
 164                 s = ns;
 165                 memcpy(s + len, a, l);
 166                 len += l;
 167         }
 168         va_end(ap);
 169
 170         s[len] = '\0';
 171         return s;
 172 }
 173
 174 int
 175 xstrcmp(unsigned char *s1, unsigned char *s2)
 176 {
 177         if (!s1) return -!!s2;
 178         if (!s2) return 1;
 179         return strcmp(s1, s2);
 180 }
 181
 182 unsigned char *
 183 safe_strncpy(unsigned char *dst, const unsigned char *src, size_t dst_size)
 184 {
 185         assertm(dst && src && dst_size > 0, "[safe_strncpy]");
 186         if_assert_failed return NULL;
 187
 188         strncpy(dst, src, dst_size);
 189         dst[dst_size - 1] = '\0';
 190
 191         return dst;
 192 }
 193
 194 #define strlcmp_device(c,s1,n1,s2,n2,t1,t2) { \
 195         size_t p; \
 196         int d; \
 197  \
 198         /* XXX: The return value is inconsistent. Hrmpf. Making it consistent
 199          * would make the @n1 != @n2 case significantly more expensive, though.
 200          * So noone should better rely on the return value actually meaning
 201          * anything quantitatively. --pasky */ \
 202  \
 203         if (!s1 || !s2) \
 204                 return 1; \
 205  \
 206         /* n1,n2 is unsigned, so don't assume -1 < 0 ! >:) */ \
 207  \
 208         /* TODO: Don't precompute strlen()s but rather make the loop smarter.
 209          * --pasky */ \
 210         if (n1 == -1) n1 = strlen(s1); \
 211         if (n2 == -1) n2 = strlen(s2); \
 212  \
 213         string_assert(errfile, errline, n1 >= 0 && n2 >= 0, c); \
 214  \
 215         d = n1 - n2; \
 216         if (d) return d; \
 217  \
 218         for (p = 0; p < n1 && s1[p] && s2[p]; p++) { \
 219                 d = t1 - t2; \
 220                 if (d) return d; \
 221         } \
 222         return 0; \
 223 }
 224
 225 int
 226 elinks_strlcmp(const unsigned char *s1, size_t n1,
 227                const unsigned char *s2, size_t n2)
 228 {
 229         strlcmp_device("strlcmp", s1, n1, s2, n2, s1[p], s2[p]);
 230 }
 231
 232 int
 233 elinks_strlcasecmp(const unsigned char *s1, size_t n1,
 234                    const unsigned char *s2, size_t n2)
 235 {
 236         strlcmp_device("strlcasecmp", s1, n1, s2, n2, toupper(s1[p]), toupper(s2[p]));
 237 }
 238
 239
 240 /* The new string utilities: */
 241
 242 /* TODO Currently most of the functions use add_bytes_to_string() as a backend
 243  *      instead we should optimize each function. */
 244
 245 inline struct string *
 246 #ifdef DEBUG_MEMLEAK
 247 init_string__(unsigned char *file, int line, struct string *string)
 248 #else
 249 init_string(struct string *string)
 250 #endif
 251 {
 252         assertm(string, "[init_string]");
 253         if_assert_failed { return NULL; }
 254
 255         string->length = 0;
 256 #ifdef DEBUG_MEMLEAK
 257         string->source = debug_mem_alloc(file, line, STRING_GRANULARITY + 1);
 258 #else
 259         string->source = mem_alloc(STRING_GRANULARITY + 1);
 260 #endif
 261         if (!string->source) return NULL;
 262
 263         *string->source = '\0';
 264
 265         set_string_magic(string);
 266
 267         return string;
 268 }
 269
 270 inline void
 271 done_string(struct string *string)
 272 {
 273         assertm(string, "[done_string]");
 274         if_assert_failed { return; }
 275
 276         if (string->source) {
 277                 /* We only check the magic if we have to free anything so
 278                  * that done_string() can be called multiple times without
 279                  * blowing up something */
 280                 check_string_magic(string);
 281                 mem_free(string->source);
 282         }
 283
 284         /* Blast everything including the magic */
 285         memset(string, 0, sizeof(*string));
 286 }
 287
 288 inline struct string *
 289 add_to_string(struct string *string, const unsigned char *source)
 290 {
 291         assertm(string && source, "[add_to_string]");
 292         if_assert_failed { return NULL; }
 293
 294         check_string_magic(string);
 295
 296         if (!*source) return string;
 297
 298         return add_bytes_to_string(string, source, strlen(source));
 299 }
 300
 301 inline struct string *
 302 add_crlf_to_string(struct string *string)
 303 {
 304         assertm(string, "[add_crlf_to_string]");
 305         if_assert_failed { return NULL; }
 306
 307         check_string_magic(string);
 308
 309         if (!realloc_string(string, string->length + 2))
 310                 return NULL;
 311
 312         string->source[string->length++] = ASCII_CR;
 313         string->source[string->length++] = ASCII_LF;
 314         string->source[string->length]   = '\0';
 315
 316         return string;
 317 }
 318
 319 inline struct string *
 320 add_string_to_string(struct string *string, struct string *from)
 321 {
 322         assertm(string && from, "[add_string_to_string]");
 323         if_assert_failed { return NULL; }
 324
 325         check_string_magic(string);
 326         check_string_magic(from);
 327
 328         if (!*from->source) return NULL;
 329
 330         return add_bytes_to_string(string, from->source, from->length);
 331 }
 332
 333 struct string *
 334 add_file_to_string(struct string *string, unsigned char *filename)
 335 {
 336         FILE *file;
 337         off_t filelen;
 338         int newlength;
 339
 340         assertm(string && filename, "[add_file_to_string]");
 341         if_assert_failed { return NULL; }
 342
 343         check_string_magic(string);
 344
 345         file = fopen(filename, "rb");
 346         if (!file) return NULL;
 347
 348         if (fseeko(file, 0, SEEK_END)) goto err;
 349
 350         filelen = ftello(file);
 351         if (filelen == -1) goto err;
 352
 353         if (fseeko(file, 0, SEEK_SET)) goto err;
 354
 355         newlength = string->length + filelen;
 356         if (!realloc_string(string, newlength)) goto err;
 357
 358         string->length += fread(string->source + string->length, 1,
 359                                 (size_t) filelen, file);
 360         string->source[string->length] = 0;
 361         fclose(file);
 362
 363         if (string->length != newlength) goto err;
 364
 365         return string;
 366
 367 err:
 368         fclose(file);
 369
 370         return NULL;
 371 }
 372
 373 struct string *
 374 string_concat(struct string *string, ...)
 375 {
 376         va_list ap;
 377         unsigned char *source;
 378
 379         assertm(string, "[string_concat]");
 380         if_assert_failed { return NULL; }
 381
 382         check_string_magic(string);
 383
 384         va_start(ap, string);
 385         while ((source = va_arg(ap, unsigned char *)))
 386                 if (*source)
 387                         add_to_string(string, source);
 388
 389         va_end(ap);
 390
 391         return string;
 392 }
 393
 394 inline struct string *
 395 add_char_to_string(struct string *string, unsigned char character)
 396 {
 397         assertm(string && character, "[add_char_to_string]");
 398         if_assert_failed { return NULL; }
 399
 400         check_string_magic(string);
 401
 402         if (!realloc_string(string, string->length + 1))
 403                 return NULL;
 404
 405         string->source[string->length++] = character;
 406         string->source[string->length]   = '\0';
 407
 408         return string;
 409 }
 410
 411 inline struct string *
 412 add_xchar_to_string(struct string *string, unsigned char character, int times)
 413 {
 414         int newlength;
 415
 416         assertm(string && character && times >= 0, "[add_xchar_to_string]");
 417         if_assert_failed { return NULL; }
 418
 419         check_string_magic(string);
 420
 421         if (!times) return string;
 422
 423         newlength = string->length + times;
 424         if (!realloc_string(string, newlength))
 425                 return NULL;
 426
 427         memset(string->source + string->length, character, times);
 428         string->length = newlength;
 429         string->source[newlength] = '\0';
 430
 431         return string;
 432 }
 433
 434 /* Add printf-like format string to @string. */
 435 struct string *
 436 add_format_to_string(struct string *string, unsigned char *format, ...)
 437 {
 438         int newlength;
 439         int width;
 440         va_list ap;
 441         va_list ap2;
 442
 443         assertm(string && format, "[add_format_to_string]");
 444         if_assert_failed { return NULL; }
 445
 446         check_string_magic(string);
 447
 448         va_start(ap, format);
 449         VA_COPY(ap2, ap);
 450
 451         width = vsnprintf(NULL, 0, format, ap2);
 452         if (width <= 0) return NULL;
 453
 454         newlength = string->length + width;
 455         if (!realloc_string(string, newlength))
 456                 return NULL;
 457
 458         vsnprintf(&string->source[string->length], width + 1, format, ap);
 459
 460         va_end(ap);
 461
 462         string->length = newlength;
 463         string->source[newlength] = '\0';
 464
 465         return string;
 466 }
 467
 468 struct string *
 469 add_to_string_list(struct list_head *list, const unsigned char *source,
 470                    int length)
 471 {
 472         struct string_list_item *item;
 473         struct string *string;
 474
 475         assertm(list && source, "[add_to_string_list]");
 476         if_assert_failed return NULL;
 477
 478         item = mem_alloc(sizeof(*item));
 479         if (!item) return NULL;
 480
 481         string = &item->string;
 482         if (length < 0) length = strlen(source);
 483
 484         if (!init_string(string)
 485             || !add_bytes_to_string(string, source, length)) {
 486                 done_string(string);
 487                 mem_free(item);
 488                 return NULL;
 489         }
 490
 491         add_to_list_end(*list, item);
 492         return string;
 493 }
 494
 495 void
 496 free_string_list(struct list_head *list)
 497 {
 498         assertm(list, "[free_string_list]");
 499         if_assert_failed return;
 500
 501         while (!list_empty(*list)) {
 502                 struct string_list_item *item = list->next;
 503
 504                 del_from_list(item);
 505                 done_string(&item->string);
 506                 mem_free(item);
 507         }
 508 }