contrib/ldns/parse.c

   1 /*
   2  * a generic (simple) parser. Use to parse rr's, private key
   3  * information and /etc/resolv.conf files
   4  *
   5  * a Net::DNS like library for C
   6  * LibDNS Team @ NLnet Labs
   7  * (c) NLnet Labs, 2005-2006
   8  * See the file LICENSE for the license
   9  */
  10 #include <ldns/config.h>
  11 #include <ldns/ldns.h>
  12
  13 #include <limits.h>
  14 #include <strings.h>
  15
  16 ldns_lookup_table ldns_directive_types[] = {
  17         { LDNS_DIR_TTL, "$TTL" },
  18         { LDNS_DIR_ORIGIN, "$ORIGIN" },
  19         { LDNS_DIR_INCLUDE, "$INCLUDE" },
  20         { 0, NULL }
  21 };
  22
  23 /* add max_limit here? */
  24 ssize_t
  25 ldns_fget_token(FILE *f, char *token, const char *delim, size_t limit)
  26 {
  27         return ldns_fget_token_l(f, token, delim, limit, NULL);
  28 }
  29
  30 ldns_status
  31 ldns_fget_token_l_st(FILE *f, char **token, size_t *limit, bool fixed
  32                     , const char *delim, int *line_nr)
  33 {
  34         int c, prev_c;
  35         int p; /* 0 -> no parentheses seen, >0 nr of ( seen */
  36         int com, quoted;
  37         char *t, *old_token;
  38         size_t i;
  39         const char *d;
  40         const char *del;
  41
  42         /* standard delimiters */
  43         if (!delim) {
  44                 /* from isspace(3) */
  45                 del = LDNS_PARSE_NORMAL;
  46         } else {
  47                 del = delim;
  48         }
  49         if (!token || !limit)
  50                 return LDNS_STATUS_NULL;
  51
  52         if (fixed) {
  53                 if (*token == NULL || *limit == 0)
  54                         return LDNS_STATUS_NULL;
  55
  56         } else if (*token == NULL) {
  57                 *limit = LDNS_MAX_LINELEN;
  58                 if (!(*token = LDNS_XMALLOC(char, *limit + 1)))
  59                         return LDNS_STATUS_MEM_ERR;
  60
  61         } else if (*limit == 0)
  62                 return LDNS_STATUS_ERR;
  63         p = 0;
  64         i = 0;
  65         com = 0;
  66         quoted = 0;
  67         prev_c = 0;
  68         t = *token;
  69         if (del[0] == '"') {
  70                 quoted = 1;
  71         }
  72         while ((c = getc(f)) != EOF) {
  73                 if (c == '\r') /* carriage return */
  74                         c = ' ';
  75                 if (c == '(' && prev_c != '\\' && !quoted) {
  76                         /* this only counts for non-comments */
  77                         if (com == 0) {
  78                                 p++;
  79                         }
  80                         prev_c = c;
  81                         continue;
  82                 }
  83
  84                 if (c == ')' && prev_c != '\\' && !quoted) {
  85                         /* this only counts for non-comments */
  86                         if (com == 0) {
  87                                 p--;
  88                         }
  89                         prev_c = c;
  90                         continue;
  91                 }
  92
  93                 if (p < 0) {
  94                         /* more ) then ( - close off the string */
  95                         *t = '\0';
  96                         return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY
  97                                       : LDNS_STATUS_OK;
  98                 }
  99
 100                 /* do something with comments ; */
 101                 if (c == ';' && quoted == 0) {
 102                         if (prev_c != '\\') {
 103                                 com = 1;
 104                         }
 105                 }
 106                 if (c == '\"' && com == 0 && prev_c != '\\') {
 107                         quoted = 1 - quoted;
 108                 }
 109
 110                 if (c == '\n' && com != 0) {
 111                         /* comments */
 112                         com = 0;
 113                         *t = ' ';
 114                         if (line_nr) {
 115                                 *line_nr = *line_nr + 1;
 116                         }
 117                         if (p == 0 && i > 0) {
 118                                 goto tokenread;
 119                         } else {
 120                                 prev_c = c;
 121                                 continue;
 122                         }
 123                 }
 124
 125                 if (com == 1) {
 126                         *t = ' ';
 127                         prev_c = c;
 128                         continue;
 129                 }
 130
 131                 if (c == '\n' && p != 0 && t > *token) {
 132                         /* in parentheses */
 133                         if (line_nr) {
 134                                 *line_nr = *line_nr + 1;
 135                         }
 136                         if (*limit > 0
 137                         &&  (i >= *limit || (size_t)(t - *token) >= *limit)) {
 138                                 if (fixed) {
 139                                         *t = '\0';
 140                                         return LDNS_STATUS_SYNTAX_ERR;
 141                                 }
 142                                 old_token = *token;
 143                                 *limit *= 2;
 144                                 *token = LDNS_XREALLOC(*token, char, *limit + 1);
 145                                 if (*token == NULL) {
 146                                         *token = old_token;
 147                                         *t = '\0';
 148                                         return LDNS_STATUS_MEM_ERR;
 149                                 }
 150                                 if (*token != old_token)
 151                                         t = *token + (t - old_token);
 152                         }
 153                         *t++ = ' ';
 154                         prev_c = c;
 155                         continue;
 156                 }
 157
 158                 /* check if we hit the delim */
 159                 for (d = del; *d; d++) {
 160                         if (c == *d && i > 0 && prev_c != '\\' && p == 0) {
 161                                 if (c == '\n' && line_nr) {
 162                                         *line_nr = *line_nr + 1;
 163                                 }
 164                                 goto tokenread;
 165                         }
 166                 }
 167                 if (c != '\0' && c != '\n') {
 168                         i++;
 169                 }
 170                 if (*limit > 0
 171                 &&  (i >= *limit || (size_t)(t - *token) >= *limit)) {
 172                         if (fixed) {
 173                                 *t = '\0';
 174                                 return LDNS_STATUS_SYNTAX_ERR;
 175                         }
 176                         old_token = *token;
 177                         *limit *= 2;
 178                         *token = LDNS_XREALLOC(*token, char, *limit + 1);
 179                         if (*token == NULL) {
 180                                 *token = old_token;
 181                                 *t = '\0';
 182                                 return LDNS_STATUS_MEM_ERR;
 183                         }
 184                         if (*token != old_token)
 185                                 t = *token + (t - old_token);
 186                 }
 187                 if (c != '\0' && c != '\n') {
 188                         *t++ = c;
 189                 }
 190                 if (c == '\n' && line_nr) {
 191                         *line_nr = *line_nr + 1;
 192                 }
 193                 if (c == '\\' && prev_c == '\\')
 194                         prev_c = 0;
 195                 else    prev_c = c;
 196         }
 197         *t = '\0';
 198         if (c == EOF) {
 199                 return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
 200         }
 201
 202         if (p != 0) {
 203                 return LDNS_STATUS_SYNTAX_ERR;
 204         }
 205         return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
 206
 207 tokenread:
 208         if(*del == '"') /* do not skip over quotes, they are significant */
 209                 ldns_fskipcs_l(f, del+1, line_nr);
 210         else    ldns_fskipcs_l(f, del, line_nr);
 211         *t = '\0';
 212         if (p != 0) {
 213                 return LDNS_STATUS_SYNTAX_ERR;
 214         }
 215         return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
 216 }
 217
 218
 219 ssize_t
 220 ldns_fget_token_l(FILE *f, char *token, const char *delim, size_t limit, int *line_nr)
 221 {
 222         if (limit == 0)
 223                 limit = LDNS_MAX_LINELEN;
 224         if (ldns_fget_token_l_st(f, &token, &limit, true, delim, line_nr))
 225                 return -1;
 226         else
 227                 return (ssize_t)strlen(token);
 228 }
 229
 230 ssize_t
 231 ldns_fget_keyword_data(FILE *f, const char *keyword, const char *k_del, char *data,
 232                const char *d_del, size_t data_limit)
 233 {
 234        return ldns_fget_keyword_data_l(f, keyword, k_del, data, d_del,
 235                        data_limit, NULL);
 236 }
 237
 238 ssize_t
 239 ldns_fget_keyword_data_l(FILE *f, const char *keyword, const char *k_del, char *data,
 240                const char *d_del, size_t data_limit, int *line_nr)
 241 {
 242        /* we assume: keyword|sep|data */
 243        char *fkeyword;
 244        ssize_t i;
 245
 246        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
 247                return -1;
 248        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
 249        if(!fkeyword)
 250                return -1;
 251
 252        i = ldns_fget_token(f, fkeyword, k_del, LDNS_MAX_KEYWORDLEN);
 253        if(i==0 || i==-1) {
 254                LDNS_FREE(fkeyword);
 255                return -1;
 256        }
 257
 258        /* case??? i instead of strlen? */
 259        if (strncmp(fkeyword, keyword, LDNS_MAX_KEYWORDLEN - 1) == 0) {
 260                /* whee! */
 261                /* printf("%s\n%s\n", "Matching keyword", fkeyword); */
 262                i = ldns_fget_token_l(f, data, d_del, data_limit, line_nr);
 263                LDNS_FREE(fkeyword);
 264                return i;
 265        } else {
 266                /*printf("no match for %s (read: %s)\n", keyword, fkeyword);*/
 267                LDNS_FREE(fkeyword);
 268                return -1;
 269        }
 270 }
 271
 272
 273 ssize_t
 274 ldns_bget_token(ldns_buffer *b, char *token, const char *delim, size_t limit)
 275 {
 276         int c, lc;
 277         int p; /* 0 -> no parentheses seen, >0 nr of ( seen */
 278         int com, quoted;
 279         char *t;
 280         size_t i;
 281         const char *d;
 282         const char *del;
 283
 284         /* standard delimiters */
 285         if (!delim) {
 286                 /* from isspace(3) */
 287                 del = LDNS_PARSE_NORMAL;
 288         } else {
 289                 del = delim;
 290         }
 291
 292         p = 0;
 293         i = 0;
 294         com = 0;
 295         quoted = 0;
 296         t = token;
 297         lc = 0;
 298         if (del[0] == '"') {
 299                 quoted = 1;
 300         }
 301
 302         while ((c = ldns_bgetc(b)) != EOF) {
 303                 if (c == '\r') /* carriage return */
 304                         c = ' ';
 305                 if (c == '(' && lc != '\\' && !quoted) {
 306                         /* this only counts for non-comments */
 307                         if (com == 0) {
 308                                 p++;
 309                         }
 310                         lc = c;
 311                         continue;
 312                 }
 313
 314                 if (c == ')' && lc != '\\' && !quoted) {
 315                         /* this only counts for non-comments */
 316                         if (com == 0) {
 317                                 p--;
 318                         }
 319                         lc = c;
 320                         continue;
 321                 }
 322
 323                 if (p < 0) {
 324                         /* more ) then ( */
 325                         *t = '\0';
 326                         return 0;
 327                 }
 328
 329                 /* do something with comments ; */
 330                 if (c == ';' && quoted == 0) {
 331                         if (lc != '\\') {
 332                                 com = 1;
 333                         }
 334                 }
 335                 if (c == '"' && com == 0 && lc != '\\') {
 336                         quoted = 1 - quoted;
 337                 }
 338
 339                 if (c == '\n' && com != 0) {
 340                         /* comments */
 341                         com = 0;
 342                         *t = ' ';
 343                         lc = c;
 344                         continue;
 345                 }
 346
 347                 if (com == 1) {
 348                         *t = ' ';
 349                         lc = c;
 350                         continue;
 351                 }
 352
 353                 if (c == '\n' && p != 0) {
 354                         /* in parentheses */
 355                         *t++ = ' ';
 356                         lc = c;
 357                         continue;
 358                 }
 359
 360                 /* check if we hit the delim */
 361                 for (d = del; *d; d++) {
 362                         if (c == *d && lc != '\\' && p == 0) {
 363                                 goto tokenread;
 364                         }
 365                 }
 366
 367                 i++;
 368                 if (limit > 0 && (i >= limit || (size_t)(t-token) >= limit)) {
 369                         *t = '\0';
 370                         return -1;
 371                 }
 372                 *t++ = c;
 373
 374                 if (c == '\\' && lc == '\\') {
 375                         lc = 0;
 376                 } else {
 377                         lc = c;
 378                 }
 379         }
 380         *t = '\0';
 381         if (i == 0) {
 382                 /* nothing read */
 383                 return -1;
 384         }
 385         if (p != 0) {
 386                 return -1;
 387         }
 388         return (ssize_t)i;
 389
 390 tokenread:
 391         if(*del == '"') /* do not skip over quotes, they are significant */
 392                 ldns_bskipcs(b, del+1);
 393         else    ldns_bskipcs(b, del);
 394         *t = '\0';
 395
 396         if (p != 0) {
 397                 return -1;
 398         }
 399         return (ssize_t)i;
 400 }
 401
 402
 403 void
 404 ldns_bskipcs(ldns_buffer *buffer, const char *s)
 405 {
 406         bool found;
 407         char c;
 408         const char *d;
 409
 410         while(ldns_buffer_available_at(buffer, buffer->_position, sizeof(char))) {
 411                 c = (char) ldns_buffer_read_u8_at(buffer, buffer->_position);
 412                 found = false;
 413                 for (d = s; *d; d++) {
 414                         if (*d == c) {
 415                                 found = true;
 416                         }
 417                 }
 418                 if (found && buffer->_limit > buffer->_position) {
 419                         buffer->_position += sizeof(char);
 420                 } else {
 421                         return;
 422                 }
 423         }
 424 }
 425
 426 void
 427 ldns_fskipcs(FILE *fp, const char *s)
 428 {
 429         ldns_fskipcs_l(fp, s, NULL);
 430 }
 431
 432 void
 433 ldns_fskipcs_l(FILE *fp, const char *s, int *line_nr)
 434 {
 435         bool found;
 436         int c;
 437         const char *d;
 438
 439         while ((c = fgetc(fp)) != EOF) {
 440                 if (line_nr && c == '\n') {
 441                         *line_nr = *line_nr + 1;
 442                 }
 443                 found = false;
 444                 for (d = s; *d; d++) {
 445                         if (*d == c) {
 446                                 found = true;
 447                         }
 448                 }
 449                 if (!found) {
 450                         /* with getc, we've read too far */
 451                         ungetc(c, fp);
 452                         return;
 453                 }
 454         }
 455 }
 456
 457 ssize_t
 458 ldns_bget_keyword_data(ldns_buffer *b, const char *keyword, const char *k_del, char
 459 *data, const char *d_del, size_t data_limit)
 460 {
 461        /* we assume: keyword|sep|data */
 462        char *fkeyword;
 463        ssize_t i;
 464
 465        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
 466                return -1;
 467        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
 468        if(!fkeyword)
 469                return -1; /* out of memory */
 470
 471        i = ldns_bget_token(b, fkeyword, k_del, data_limit);
 472        if(i==0 || i==-1) {
 473                LDNS_FREE(fkeyword);
 474                return -1; /* nothing read */
 475        }
 476
 477        /* case??? */
 478        if (strncmp(fkeyword, keyword, strlen(keyword)) == 0) {
 479                LDNS_FREE(fkeyword);
 480                /* whee, the match! */
 481                /* retrieve it's data */
 482                i = ldns_bget_token(b, data, d_del, 0);
 483                return i;
 484        } else {
 485                LDNS_FREE(fkeyword);
 486                return -1;
 487        }
 488 }
 489