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 ssize_t
  31 ldns_fget_token_l(FILE *f, char *token, const char *delim, size_t limit, int *line_nr)
  32 {
  33         int c, prev_c;
  34         int p; /* 0 -> no parenthese seen, >0 nr of ( seen */
  35         int com, quoted;
  36         char *t;
  37         size_t i;
  38         const char *d;
  39         const char *del;
  40
  41         /* standard delimeters */
  42         if (!delim) {
  43                 /* from isspace(3) */
  44                 del = LDNS_PARSE_NORMAL;
  45         } else {
  46                 del = delim;
  47         }
  48
  49         p = 0;
  50         i = 0;
  51         com = 0;
  52         quoted = 0;
  53         prev_c = 0;
  54         t = token;
  55         if (del[0] == '"') {
  56                 quoted = 1;
  57         }
  58         while ((c = getc(f)) != EOF) {
  59                 if (c == '\r') /* carriage return */
  60                         c = ' ';
  61                 if (c == '(' && prev_c != '\\' && !quoted) {
  62                         /* this only counts for non-comments */
  63                         if (com == 0) {
  64                                 p++;
  65                         }
  66                         prev_c = c;
  67                         continue;
  68                 }
  69
  70                 if (c == ')' && prev_c != '\\' && !quoted) {
  71                         /* this only counts for non-comments */
  72                         if (com == 0) {
  73                                 p--;
  74                         }
  75                         prev_c = c;
  76                         continue;
  77                 }
  78
  79                 if (p < 0) {
  80                         /* more ) then ( - close off the string */
  81                         *t = '\0';
  82                         return 0;
  83                 }
  84
  85                 /* do something with comments ; */
  86                 if (c == ';' && quoted == 0) {
  87                         if (prev_c != '\\') {
  88                                 com = 1;
  89                         }
  90                 }
  91                 if (c == '\"' && com == 0 && prev_c != '\\') {
  92                         quoted = 1 - quoted;
  93                 }
  94
  95                 if (c == '\n' && com != 0) {
  96                         /* comments */
  97                         com = 0;
  98                         *t = ' ';
  99                         if (line_nr) {
 100                                 *line_nr = *line_nr + 1;
 101                         }
 102                         if (p == 0 && i > 0) {
 103                                 goto tokenread;
 104                         } else {
 105                                 prev_c = c;
 106                                 continue;
 107                         }
 108                 }
 109
 110                 if (com == 1) {
 111                         *t = ' ';
 112                         prev_c = c;
 113                         continue;
 114                 }
 115
 116                 if (c == '\n' && p != 0 && t > token) {
 117                         /* in parentheses */
 118                         if (line_nr) {
 119                                 *line_nr = *line_nr + 1;
 120                         }
 121                         *t++ = ' ';
 122                         prev_c = c;
 123                         continue;
 124                 }
 125
 126                 /* check if we hit the delim */
 127                 for (d = del; *d; d++) {
 128                         if (c == *d && i > 0 && prev_c != '\\' && p == 0) {
 129                                 if (c == '\n' && line_nr) {
 130                                         *line_nr = *line_nr + 1;
 131                                 }
 132                                 goto tokenread;
 133                         }
 134                 }
 135                 if (c != '\0' && c != '\n') {
 136                         i++;
 137                 }
 138                 if (limit > 0 && i >= limit) {
 139                         *t = '\0';
 140                         return -1;
 141                 }
 142                 if (c != '\0' && c != '\n') {
 143                         *t++ = c;
 144                 }
 145                 if (c == '\\' && prev_c == '\\')
 146                         prev_c = 0;
 147                 else    prev_c = c;
 148         }
 149         *t = '\0';
 150         if (c == EOF) {
 151                 return (ssize_t)i;
 152         }
 153
 154         if (i == 0) {
 155                 /* nothing read */
 156                 return -1;
 157         }
 158         if (p != 0) {
 159                 return -1;
 160         }
 161         return (ssize_t)i;
 162
 163 tokenread:
 164         ldns_fskipcs_l(f, del, line_nr);
 165         *t = '\0';
 166         if (p != 0) {
 167                 return -1;
 168         }
 169
 170         return (ssize_t)i;
 171 }
 172
 173 ssize_t
 174 ldns_fget_keyword_data(FILE *f, const char *keyword, const char *k_del, char *data,
 175                const char *d_del, size_t data_limit)
 176 {
 177        return ldns_fget_keyword_data_l(f, keyword, k_del, data, d_del,
 178                        data_limit, NULL);
 179 }
 180
 181 ssize_t
 182 ldns_fget_keyword_data_l(FILE *f, const char *keyword, const char *k_del, char *data,
 183                const char *d_del, size_t data_limit, int *line_nr)
 184 {
 185        /* we assume: keyword|sep|data */
 186        char *fkeyword;
 187        ssize_t i;
 188
 189        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
 190                return -1;
 191        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
 192        if(!fkeyword)
 193                return -1;
 194
 195        i = ldns_fget_token(f, fkeyword, k_del, LDNS_MAX_KEYWORDLEN);
 196        if(i==0 || i==-1) {
 197                LDNS_FREE(fkeyword);
 198                return -1;
 199        }
 200
 201        /* case??? i instead of strlen? */
 202        if (strncmp(fkeyword, keyword, LDNS_MAX_KEYWORDLEN - 1) == 0) {
 203                /* whee! */
 204                /* printf("%s\n%s\n", "Matching keyword", fkeyword); */
 205                i = ldns_fget_token_l(f, data, d_del, data_limit, line_nr);
 206                LDNS_FREE(fkeyword);
 207                return i;
 208        } else {
 209                /*printf("no match for %s (read: %s)\n", keyword, fkeyword);*/
 210                LDNS_FREE(fkeyword);
 211                return -1;
 212        }
 213 }
 214
 215
 216 ssize_t
 217 ldns_bget_token(ldns_buffer *b, char *token, const char *delim, size_t limit)
 218 {
 219         int c, lc;
 220         int p; /* 0 -> no parenthese seen, >0 nr of ( seen */
 221         int com, quoted;
 222         char *t;
 223         size_t i;
 224         const char *d;
 225         const char *del;
 226
 227         /* standard delimiters */
 228         if (!delim) {
 229                 /* from isspace(3) */
 230                 del = LDNS_PARSE_NORMAL;
 231         } else {
 232                 del = delim;
 233         }
 234
 235         p = 0;
 236         i = 0;
 237         com = 0;
 238         quoted = 0;
 239         t = token;
 240         lc = 0;
 241         if (del[0] == '"') {
 242                 quoted = 1;
 243         }
 244
 245         while ((c = ldns_bgetc(b)) != EOF) {
 246                 if (c == '\r') /* carriage return */
 247                         c = ' ';
 248                 if (c == '(' && lc != '\\' && !quoted) {
 249                         /* this only counts for non-comments */
 250                         if (com == 0) {
 251                                 p++;
 252                         }
 253                         lc = c;
 254                         continue;
 255                 }
 256
 257                 if (c == ')' && lc != '\\' && !quoted) {
 258                         /* this only counts for non-comments */
 259                         if (com == 0) {
 260                                 p--;
 261                         }
 262                         lc = c;
 263                         continue;
 264                 }
 265
 266                 if (p < 0) {
 267                         /* more ) then ( */
 268                         *t = '\0';
 269                         return 0;
 270                 }
 271
 272                 /* do something with comments ; */
 273                 if (c == ';' && quoted == 0) {
 274                         if (lc != '\\') {
 275                                 com = 1;
 276                         }
 277                 }
 278                 if (c == '"' && com == 0 && lc != '\\') {
 279                         quoted = 1 - quoted;
 280                 }
 281
 282                 if (c == '\n' && com != 0) {
 283                         /* comments */
 284                         com = 0;
 285                         *t = ' ';
 286                         lc = c;
 287                         continue;
 288                 }
 289
 290                 if (com == 1) {
 291                         *t = ' ';
 292                         lc = c;
 293                         continue;
 294                 }
 295
 296                 if (c == '\n' && p != 0) {
 297                         /* in parentheses */
 298                         *t++ = ' ';
 299                         lc = c;
 300                         continue;
 301                 }
 302
 303                 /* check if we hit the delim */
 304                 for (d = del; *d; d++) {
 305                         if (c == *d && lc != '\\' && p == 0) {
 306                                 goto tokenread;
 307                         }
 308                 }
 309
 310                 i++;
 311                 if (limit > 0 && i >= limit) {
 312                         *t = '\0';
 313                         return -1;
 314                 }
 315                 *t++ = c;
 316
 317                 if (c == '\\' && lc == '\\') {
 318                         lc = 0;
 319                 } else {
 320                         lc = c;
 321                 }
 322         }
 323         *t = '\0';
 324         if (i == 0) {
 325                 /* nothing read */
 326                 return -1;
 327         }
 328         if (p != 0) {
 329                 return -1;
 330         }
 331         return (ssize_t)i;
 332
 333 tokenread:
 334         ldns_bskipcs(b, del);
 335         *t = '\0';
 336
 337         if (p != 0) {
 338                 return -1;
 339         }
 340         return (ssize_t)i;
 341 }
 342
 343
 344 void
 345 ldns_bskipcs(ldns_buffer *buffer, const char *s)
 346 {
 347         bool found;
 348         char c;
 349         const char *d;
 350
 351         while(ldns_buffer_available_at(buffer, buffer->_position, sizeof(char))) {
 352                 c = (char) ldns_buffer_read_u8_at(buffer, buffer->_position);
 353                 found = false;
 354                 for (d = s; *d; d++) {
 355                         if (*d == c) {
 356                                 found = true;
 357                         }
 358                 }
 359                 if (found && buffer->_limit > buffer->_position) {
 360                         buffer->_position += sizeof(char);
 361                 } else {
 362                         return;
 363                 }
 364         }
 365 }
 366
 367 void
 368 ldns_fskipcs(FILE *fp, const char *s)
 369 {
 370         ldns_fskipcs_l(fp, s, NULL);
 371 }
 372
 373 void
 374 ldns_fskipcs_l(FILE *fp, const char *s, int *line_nr)
 375 {
 376         bool found;
 377         int c;
 378         const char *d;
 379
 380         while ((c = fgetc(fp)) != EOF) {
 381                 if (line_nr && c == '\n') {
 382                         *line_nr = *line_nr + 1;
 383                 }
 384                 found = false;
 385                 for (d = s; *d; d++) {
 386                         if (*d == c) {
 387                                 found = true;
 388                         }
 389                 }
 390                 if (!found) {
 391                         /* with getc, we've read too far */
 392                         ungetc(c, fp);
 393                         return;
 394                 }
 395         }
 396 }
 397
 398 ssize_t
 399 ldns_bget_keyword_data(ldns_buffer *b, const char *keyword, const char *k_del, char
 400 *data, const char *d_del, size_t data_limit)
 401 {
 402        /* we assume: keyword|sep|data */
 403        char *fkeyword;
 404        ssize_t i;
 405
 406        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
 407                return -1;
 408        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
 409        if(!fkeyword)
 410                return -1; /* out of memory */
 411
 412        i = ldns_bget_token(b, fkeyword, k_del, data_limit);
 413        if(i==0 || i==-1) {
 414                LDNS_FREE(fkeyword);
 415                return -1; /* nothing read */
 416        }
 417
 418        /* case??? */
 419        if (strncmp(fkeyword, keyword, strlen(keyword)) == 0) {
 420                LDNS_FREE(fkeyword);
 421                /* whee, the match! */
 422                /* retrieve it's data */
 423                i = ldns_bget_token(b, data, d_del, 0);
 424                return i;
 425        } else {
 426                LDNS_FREE(fkeyword);
 427                return -1;
 428        }
 429 }
 430