1 /* Copyright (c) 2013, Vsevolod Stakhov 2 * All rights reserved. 3 * 4 * Redistribution and use in source and binary forms, with or without 5 * modification, are permitted provided that the following conditions are met: 6 * * Redistributions of source code must retain the above copyright 7 * notice, this list of conditions and the following disclaimer. 8 * * Redistributions in binary form must reproduce the above copyright 9 * notice, this list of conditions and the following disclaimer in the 10 * documentation and/or other materials provided with the distribution. 11 * 12 * THIS SOFTWARE IS PROVIDED ''AS IS'' AND ANY 13 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 14 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 15 * DISCLAIMED. IN NO EVENT SHALL AUTHOR BE LIABLE FOR ANY 16 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 17 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 18 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 19 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 20 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 21 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 22 */ 23 24 #include "ucl.h" 25 #include "ucl_internal.h" 26 #include "ucl_chartable.h" 27 28 /** 29 * @file ucl_parser.c 30 * The implementation of ucl parser 31 */ 32 33 struct ucl_parser_saved_state { 34 unsigned int line; 35 unsigned int column; 36 size_t remain; 37 const unsigned char *pos; 38 }; 39 40 /** 41 * Move up to len characters 42 * @param parser 43 * @param begin 44 * @param len 45 * @return new position in chunk 46 */ 47 #define ucl_chunk_skipc(chunk, p) do{ \ 48 if (*(p) == '\n') { \ 49 (chunk)->line ++; \ 50 (chunk)->column = 0; \ 51 } \ 52 else (chunk)->column ++; \ 53 (p++); \ 54 (chunk)->pos ++; \ 55 (chunk)->remain --; \ 56 } while (0) 57 58 static inline void 59 ucl_set_err (struct ucl_parser *parser, int code, const char *str, UT_string **err) 60 { 61 const char *fmt_string, *filename; 62 struct ucl_chunk *chunk = parser->chunks; 63 64 if (parser->cur_file) { 65 filename = parser->cur_file; 66 } 67 else { 68 filename = "<unknown>"; 69 } 70 if (chunk->pos < chunk->end) { 71 if (isgraph (*chunk->pos)) { 72 fmt_string = "error while parsing %s: " 73 "line: %d, column: %d - '%s', character: '%c'"; 74 } 75 else { 76 fmt_string = "error while parsing %s: " 77 "line: %d, column: %d - '%s', character: '0x%02x'"; 78 } 79 ucl_create_err (err, fmt_string, 80 filename, chunk->line, chunk->column, 81 str, *chunk->pos); 82 } 83 else { 84 ucl_create_err (err, "error while parsing %s: at the end of chunk: %s", 85 filename, str); 86 } 87 } 88 89 /** 90 * Skip all comments from the current pos resolving nested and multiline comments 91 * @param parser 92 * @return 93 */ 94 static bool 95 ucl_skip_comments (struct ucl_parser *parser) 96 { 97 struct ucl_chunk *chunk = parser->chunks; 98 const unsigned char *p; 99 int comments_nested = 0; 100 bool quoted = false; 101 102 p = chunk->pos; 103 104 start: 105 if (chunk->remain > 0 && *p == '#') { 106 if (parser->state != UCL_STATE_SCOMMENT && 107 parser->state != UCL_STATE_MCOMMENT) { 108 while (p < chunk->end) { 109 if (*p == '\n') { 110 ucl_chunk_skipc (chunk, p); 111 goto start; 112 } 113 ucl_chunk_skipc (chunk, p); 114 } 115 } 116 } 117 else if (chunk->remain >= 2 && *p == '/') { 118 if (p[1] == '*') { 119 ucl_chunk_skipc (chunk, p); 120 comments_nested ++; 121 ucl_chunk_skipc (chunk, p); 122 123 while (p < chunk->end) { 124 if (*p == '"' && *(p - 1) != '\\') { 125 quoted = !quoted; 126 } 127 128 if (!quoted) { 129 if (*p == '*') { 130 ucl_chunk_skipc (chunk, p); 131 if (*p == '/') { 132 comments_nested --; 133 if (comments_nested == 0) { 134 ucl_chunk_skipc (chunk, p); 135 goto start; 136 } 137 } 138 ucl_chunk_skipc (chunk, p); 139 } 140 else if (p[0] == '/' && chunk->remain >= 2 && p[1] == '*') { 141 comments_nested ++; 142 ucl_chunk_skipc (chunk, p); 143 ucl_chunk_skipc (chunk, p); 144 continue; 145 } 146 } 147 ucl_chunk_skipc (chunk, p); 148 } 149 if (comments_nested != 0) { 150 ucl_set_err (parser, UCL_ENESTED, 151 "unfinished multiline comment", &parser->err); 152 return false; 153 } 154 } 155 } 156 157 return true; 158 } 159 160 /** 161 * Return multiplier for a character 162 * @param c multiplier character 163 * @param is_bytes if true use 1024 multiplier 164 * @return multiplier 165 */ 166 static inline unsigned long 167 ucl_lex_num_multiplier (const unsigned char c, bool is_bytes) { 168 const struct { 169 char c; 170 long mult_normal; 171 long mult_bytes; 172 } multipliers[] = { 173 {'m', 1000 * 1000, 1024 * 1024}, 174 {'k', 1000, 1024}, 175 {'g', 1000 * 1000 * 1000, 1024 * 1024 * 1024} 176 }; 177 int i; 178 179 for (i = 0; i < 3; i ++) { 180 if (tolower (c) == multipliers[i].c) { 181 if (is_bytes) { 182 return multipliers[i].mult_bytes; 183 } 184 return multipliers[i].mult_normal; 185 } 186 } 187 188 return 1; 189 } 190 191 192 /** 193 * Return multiplier for time scaling 194 * @param c 195 * @return 196 */ 197 static inline double 198 ucl_lex_time_multiplier (const unsigned char c) { 199 const struct { 200 char c; 201 double mult; 202 } multipliers[] = { 203 {'m', 60}, 204 {'h', 60 * 60}, 205 {'d', 60 * 60 * 24}, 206 {'w', 60 * 60 * 24 * 7}, 207 {'y', 60 * 60 * 24 * 7 * 365} 208 }; 209 int i; 210 211 for (i = 0; i < 5; i ++) { 212 if (tolower (c) == multipliers[i].c) { 213 return multipliers[i].mult; 214 } 215 } 216 217 return 1; 218 } 219 220 /** 221 * Return true if a character is a end of an atom 222 * @param c 223 * @return 224 */ 225 static inline bool 226 ucl_lex_is_atom_end (const unsigned char c) 227 { 228 return ucl_test_character (c, UCL_CHARACTER_VALUE_END); 229 } 230 231 static inline bool 232 ucl_lex_is_comment (const unsigned char c1, const unsigned char c2) 233 { 234 if (c1 == '/') { 235 if (c2 == '*') { 236 return true; 237 } 238 } 239 else if (c1 == '#') { 240 return true; 241 } 242 return false; 243 } 244 245 /** 246 * Check variable found 247 * @param parser 248 * @param ptr 249 * @param remain 250 * @param out_len 251 * @param strict 252 * @param found 253 * @return 254 */ 255 static inline const char * 256 ucl_check_variable_safe (struct ucl_parser *parser, const char *ptr, size_t remain, 257 size_t *out_len, bool strict, bool *found) 258 { 259 struct ucl_variable *var; 260 unsigned char *dst; 261 size_t dstlen; 262 bool need_free = false; 263 264 LL_FOREACH (parser->variables, var) { 265 if (strict) { 266 if (remain == var->var_len) { 267 if (memcmp (ptr, var->var, var->var_len) == 0) { 268 *out_len += var->value_len; 269 *found = true; 270 return (ptr + var->var_len); 271 } 272 } 273 } 274 else { 275 if (remain >= var->var_len) { 276 if (memcmp (ptr, var->var, var->var_len) == 0) { 277 *out_len += var->value_len; 278 *found = true; 279 return (ptr + var->var_len); 280 } 281 } 282 } 283 } 284 285 /* XXX: can only handle ${VAR} */ 286 if (!(*found) && parser->var_handler != NULL && strict) { 287 /* Call generic handler */ 288 if (parser->var_handler (ptr, remain, &dst, &dstlen, &need_free, 289 parser->var_data)) { 290 *found = true; 291 if (need_free) { 292 free (dst); 293 } 294 return (ptr + remain); 295 } 296 } 297 298 return ptr; 299 } 300 301 /** 302 * Check for a variable in a given string 303 * @param parser 304 * @param ptr 305 * @param remain 306 * @param out_len 307 * @param vars_found 308 * @return 309 */ 310 static const char * 311 ucl_check_variable (struct ucl_parser *parser, const char *ptr, 312 size_t remain, size_t *out_len, bool *vars_found) 313 { 314 const char *p, *end, *ret = ptr; 315 bool found = false; 316 317 if (*ptr == '{') { 318 /* We need to match the variable enclosed in braces */ 319 p = ptr + 1; 320 end = ptr + remain; 321 while (p < end) { 322 if (*p == '}') { 323 ret = ucl_check_variable_safe (parser, ptr + 1, p - ptr - 1, 324 out_len, true, &found); 325 if (found) { 326 /* {} must be excluded actually */ 327 ret ++; 328 if (!*vars_found) { 329 *vars_found = true; 330 } 331 } 332 else { 333 *out_len += 2; 334 } 335 break; 336 } 337 p ++; 338 } 339 } 340 else if (*ptr != '$') { 341 /* Not count escaped dollar sign */ 342 ret = ucl_check_variable_safe (parser, ptr, remain, out_len, false, &found); 343 if (found && !*vars_found) { 344 *vars_found = true; 345 } 346 if (!found) { 347 (*out_len) ++; 348 } 349 } 350 else { 351 ret ++; 352 (*out_len) ++; 353 } 354 355 return ret; 356 } 357 358 /** 359 * Expand a single variable 360 * @param parser 361 * @param ptr 362 * @param remain 363 * @param dest 364 * @return 365 */ 366 static const char * 367 ucl_expand_single_variable (struct ucl_parser *parser, const char *ptr, 368 size_t remain, unsigned char **dest) 369 { 370 unsigned char *d = *dest, *dst; 371 const char *p = ptr + 1, *ret; 372 struct ucl_variable *var; 373 size_t dstlen; 374 bool need_free = false; 375 bool found = false; 376 bool strict = false; 377 378 ret = ptr + 1; 379 remain --; 380 381 if (*p == '$') { 382 *d++ = *p++; 383 *dest = d; 384 return p; 385 } 386 else if (*p == '{') { 387 p ++; 388 strict = true; 389 ret += 2; 390 remain -= 2; 391 } 392 393 LL_FOREACH (parser->variables, var) { 394 if (remain >= var->var_len) { 395 if (memcmp (p, var->var, var->var_len) == 0) { 396 memcpy (d, var->value, var->value_len); 397 ret += var->var_len; 398 d += var->value_len; 399 found = true; 400 break; 401 } 402 } 403 } 404 if (!found) { 405 if (strict && parser->var_handler != NULL) { 406 if (parser->var_handler (ptr, remain, &dst, &dstlen, &need_free, 407 parser->var_data)) { 408 memcpy (d, dst, dstlen); 409 ret += dstlen; 410 d += remain; 411 found = true; 412 } 413 } 414 415 /* Leave variable as is */ 416 if (!found) { 417 if (strict) { 418 /* Copy '${' */ 419 memcpy (d, ptr, 2); 420 d += 2; 421 ret --; 422 } 423 else { 424 memcpy (d, ptr, 1); 425 d ++; 426 } 427 } 428 } 429 430 *dest = d; 431 return ret; 432 } 433 434 /** 435 * Expand variables in string 436 * @param parser 437 * @param dst 438 * @param src 439 * @param in_len 440 * @return 441 */ 442 static ssize_t 443 ucl_expand_variable (struct ucl_parser *parser, unsigned char **dst, 444 const char *src, size_t in_len) 445 { 446 const char *p, *end = src + in_len; 447 unsigned char *d; 448 size_t out_len = 0; 449 bool vars_found = false; 450 451 p = src; 452 while (p != end) { 453 if (*p == '$') { 454 p = ucl_check_variable (parser, p + 1, end - p - 1, &out_len, &vars_found); 455 } 456 else { 457 p ++; 458 out_len ++; 459 } 460 } 461 462 if (!vars_found) { 463 /* Trivial case */ 464 *dst = NULL; 465 return in_len; 466 } 467 468 *dst = UCL_ALLOC (out_len + 1); 469 if (*dst == NULL) { 470 return in_len; 471 } 472 473 d = *dst; 474 p = src; 475 while (p != end) { 476 if (*p == '$') { 477 p = ucl_expand_single_variable (parser, p, end - p, &d); 478 } 479 else { 480 *d++ = *p++; 481 } 482 } 483 484 *d = '\0'; 485 486 return out_len; 487 } 488 489 /** 490 * Store or copy pointer to the trash stack 491 * @param parser parser object 492 * @param src src string 493 * @param dst destination buffer (trash stack pointer) 494 * @param dst_const const destination pointer (e.g. value of object) 495 * @param in_len input length 496 * @param need_unescape need to unescape source (and copy it) 497 * @param need_lowercase need to lowercase value (and copy) 498 * @param need_expand need to expand variables (and copy as well) 499 * @return output length (excluding \0 symbol) 500 */ 501 static inline ssize_t 502 ucl_copy_or_store_ptr (struct ucl_parser *parser, 503 const unsigned char *src, unsigned char **dst, 504 const char **dst_const, size_t in_len, 505 bool need_unescape, bool need_lowercase, bool need_expand) 506 { 507 ssize_t ret = -1, tret; 508 unsigned char *tmp; 509 510 if (need_unescape || need_lowercase || 511 (need_expand && parser->variables != NULL) || 512 !(parser->flags & UCL_PARSER_ZEROCOPY)) { 513 /* Copy string */ 514 *dst = UCL_ALLOC (in_len + 1); 515 if (*dst == NULL) { 516 ucl_set_err (parser, 0, "cannot allocate memory for a string", 517 &parser->err); 518 return false; 519 } 520 if (need_lowercase) { 521 ret = ucl_strlcpy_tolower (*dst, src, in_len + 1); 522 } 523 else { 524 ret = ucl_strlcpy_unsafe (*dst, src, in_len + 1); 525 } 526 527 if (need_unescape) { 528 ret = ucl_unescape_json_string (*dst, ret); 529 } 530 if (need_expand) { 531 tmp = *dst; 532 tret = ret; 533 ret = ucl_expand_variable (parser, dst, tmp, ret); 534 if (*dst == NULL) { 535 /* Nothing to expand */ 536 *dst = tmp; 537 ret = tret; 538 } 539 else { 540 /* Free unexpanded value */ 541 UCL_FREE (in_len + 1, tmp); 542 } 543 } 544 *dst_const = *dst; 545 } 546 else { 547 *dst_const = src; 548 ret = in_len; 549 } 550 551 return ret; 552 } 553 554 /** 555 * Create and append an object at the specified level 556 * @param parser 557 * @param is_array 558 * @param level 559 * @return 560 */ 561 static inline ucl_object_t * 562 ucl_add_parser_stack (ucl_object_t *obj, struct ucl_parser *parser, bool is_array, int level) 563 { 564 struct ucl_stack *st; 565 566 if (!is_array) { 567 if (obj == NULL) { 568 obj = ucl_object_new_full (UCL_OBJECT, parser->chunks->priority); 569 } 570 else { 571 obj->type = UCL_OBJECT; 572 } 573 obj->value.ov = ucl_hash_create (); 574 parser->state = UCL_STATE_KEY; 575 } 576 else { 577 if (obj == NULL) { 578 obj = ucl_object_new_full (UCL_ARRAY, parser->chunks->priority); 579 } 580 else { 581 obj->type = UCL_ARRAY; 582 } 583 parser->state = UCL_STATE_VALUE; 584 } 585 586 st = UCL_ALLOC (sizeof (struct ucl_stack)); 587 if (st == NULL) { 588 ucl_set_err (parser, 0, "cannot allocate memory for an object", 589 &parser->err); 590 ucl_object_unref (obj); 591 return NULL; 592 } 593 st->obj = obj; 594 st->level = level; 595 LL_PREPEND (parser->stack, st); 596 parser->cur_obj = obj; 597 598 return obj; 599 } 600 601 int 602 ucl_maybe_parse_number (ucl_object_t *obj, 603 const char *start, const char *end, const char **pos, 604 bool allow_double, bool number_bytes, bool allow_time) 605 { 606 const char *p = start, *c = start; 607 char *endptr; 608 bool got_dot = false, got_exp = false, need_double = false, 609 is_time = false, valid_start = false, is_hex = false, 610 is_neg = false; 611 double dv = 0; 612 int64_t lv = 0; 613 614 if (*p == '-') { 615 is_neg = true; 616 c ++; 617 p ++; 618 } 619 while (p < end) { 620 if (is_hex && isxdigit (*p)) { 621 p ++; 622 } 623 else if (isdigit (*p)) { 624 valid_start = true; 625 p ++; 626 } 627 else if (!is_hex && (*p == 'x' || *p == 'X')) { 628 is_hex = true; 629 allow_double = false; 630 c = p + 1; 631 } 632 else if (allow_double) { 633 if (p == c) { 634 /* Empty digits sequence, not a number */ 635 *pos = start; 636 return EINVAL; 637 } 638 else if (*p == '.') { 639 if (got_dot) { 640 /* Double dots, not a number */ 641 *pos = start; 642 return EINVAL; 643 } 644 else { 645 got_dot = true; 646 need_double = true; 647 p ++; 648 } 649 } 650 else if (*p == 'e' || *p == 'E') { 651 if (got_exp) { 652 /* Double exp, not a number */ 653 *pos = start; 654 return EINVAL; 655 } 656 else { 657 got_exp = true; 658 need_double = true; 659 p ++; 660 if (p >= end) { 661 *pos = start; 662 return EINVAL; 663 } 664 if (!isdigit (*p) && *p != '+' && *p != '-') { 665 /* Wrong exponent sign */ 666 *pos = start; 667 return EINVAL; 668 } 669 else { 670 p ++; 671 } 672 } 673 } 674 else { 675 /* Got the end of the number, need to check */ 676 break; 677 } 678 } 679 else { 680 break; 681 } 682 } 683 684 if (!valid_start) { 685 *pos = start; 686 return EINVAL; 687 } 688 689 errno = 0; 690 if (need_double) { 691 dv = strtod (c, &endptr); 692 } 693 else { 694 if (is_hex) { 695 lv = strtoimax (c, &endptr, 16); 696 } 697 else { 698 lv = strtoimax (c, &endptr, 10); 699 } 700 } 701 if (errno == ERANGE) { 702 *pos = start; 703 return ERANGE; 704 } 705 706 /* Now check endptr */ 707 if (endptr == NULL || ucl_lex_is_atom_end (*endptr) || *endptr == '\0') { 708 p = endptr; 709 goto set_obj; 710 } 711 712 if (endptr < end && endptr != start) { 713 p = endptr; 714 switch (*p) { 715 case 'm': 716 case 'M': 717 case 'g': 718 case 'G': 719 case 'k': 720 case 'K': 721 if (end - p >= 2) { 722 if (p[1] == 's' || p[1] == 'S') { 723 /* Milliseconds */ 724 if (!need_double) { 725 need_double = true; 726 dv = lv; 727 } 728 is_time = true; 729 if (p[0] == 'm' || p[0] == 'M') { 730 dv /= 1000.; 731 } 732 else { 733 dv *= ucl_lex_num_multiplier (*p, false); 734 } 735 p += 2; 736 goto set_obj; 737 } 738 else if (number_bytes || (p[1] == 'b' || p[1] == 'B')) { 739 /* Bytes */ 740 if (need_double) { 741 need_double = false; 742 lv = dv; 743 } 744 lv *= ucl_lex_num_multiplier (*p, true); 745 p += 2; 746 goto set_obj; 747 } 748 else if (ucl_lex_is_atom_end (p[1])) { 749 if (need_double) { 750 dv *= ucl_lex_num_multiplier (*p, false); 751 } 752 else { 753 lv *= ucl_lex_num_multiplier (*p, number_bytes); 754 } 755 p ++; 756 goto set_obj; 757 } 758 else if (allow_time && end - p >= 3) { 759 if (tolower (p[0]) == 'm' && 760 tolower (p[1]) == 'i' && 761 tolower (p[2]) == 'n') { 762 /* Minutes */ 763 if (!need_double) { 764 need_double = true; 765 dv = lv; 766 } 767 is_time = true; 768 dv *= 60.; 769 p += 3; 770 goto set_obj; 771 } 772 } 773 } 774 else { 775 if (need_double) { 776 dv *= ucl_lex_num_multiplier (*p, false); 777 } 778 else { 779 lv *= ucl_lex_num_multiplier (*p, number_bytes); 780 } 781 p ++; 782 goto set_obj; 783 } 784 break; 785 case 'S': 786 case 's': 787 if (allow_time && 788 (p == end - 1 || ucl_lex_is_atom_end (p[1]))) { 789 if (!need_double) { 790 need_double = true; 791 dv = lv; 792 } 793 p ++; 794 is_time = true; 795 goto set_obj; 796 } 797 break; 798 case 'h': 799 case 'H': 800 case 'd': 801 case 'D': 802 case 'w': 803 case 'W': 804 case 'Y': 805 case 'y': 806 if (allow_time && 807 (p == end - 1 || ucl_lex_is_atom_end (p[1]))) { 808 if (!need_double) { 809 need_double = true; 810 dv = lv; 811 } 812 is_time = true; 813 dv *= ucl_lex_time_multiplier (*p); 814 p ++; 815 goto set_obj; 816 } 817 break; 818 case '\t': 819 case ' ': 820 while (p < end && ucl_test_character(*p, UCL_CHARACTER_WHITESPACE)) { 821 p++; 822 } 823 if (ucl_lex_is_atom_end(*p)) 824 goto set_obj; 825 break; 826 } 827 } 828 else if (endptr == end) { 829 /* Just a number at the end of chunk */ 830 p = endptr; 831 goto set_obj; 832 } 833 834 *pos = c; 835 return EINVAL; 836 837 set_obj: 838 if (allow_double && (need_double || is_time)) { 839 if (!is_time) { 840 obj->type = UCL_FLOAT; 841 } 842 else { 843 obj->type = UCL_TIME; 844 } 845 obj->value.dv = is_neg ? (-dv) : dv; 846 } 847 else { 848 obj->type = UCL_INT; 849 obj->value.iv = is_neg ? (-lv) : lv; 850 } 851 *pos = p; 852 return 0; 853 } 854 855 /** 856 * Parse possible number 857 * @param parser 858 * @param chunk 859 * @return true if a number has been parsed 860 */ 861 static bool 862 ucl_lex_number (struct ucl_parser *parser, 863 struct ucl_chunk *chunk, ucl_object_t *obj) 864 { 865 const unsigned char *pos; 866 int ret; 867 868 ret = ucl_maybe_parse_number (obj, chunk->pos, chunk->end, (const char **)&pos, 869 true, false, ((parser->flags & UCL_PARSER_NO_TIME) == 0)); 870 871 if (ret == 0) { 872 chunk->remain -= pos - chunk->pos; 873 chunk->column += pos - chunk->pos; 874 chunk->pos = pos; 875 return true; 876 } 877 else if (ret == ERANGE) { 878 ucl_set_err (parser, ERANGE, "numeric value out of range", &parser->err); 879 } 880 881 return false; 882 } 883 884 /** 885 * Parse quoted string with possible escapes 886 * @param parser 887 * @param chunk 888 * @return true if a string has been parsed 889 */ 890 static bool 891 ucl_lex_json_string (struct ucl_parser *parser, 892 struct ucl_chunk *chunk, bool *need_unescape, bool *ucl_escape, bool *var_expand) 893 { 894 const unsigned char *p = chunk->pos; 895 unsigned char c; 896 int i; 897 898 while (p < chunk->end) { 899 c = *p; 900 if (c < 0x1F) { 901 /* Unmasked control character */ 902 if (c == '\n') { 903 ucl_set_err (parser, UCL_ESYNTAX, "unexpected newline", 904 &parser->err); 905 } 906 else { 907 ucl_set_err (parser, UCL_ESYNTAX, "unexpected control character", 908 &parser->err); 909 } 910 return false; 911 } 912 else if (c == '\\') { 913 ucl_chunk_skipc (chunk, p); 914 c = *p; 915 if (p >= chunk->end) { 916 ucl_set_err (parser, UCL_ESYNTAX, "unfinished escape character", 917 &parser->err); 918 return false; 919 } 920 else if (ucl_test_character (c, UCL_CHARACTER_ESCAPE)) { 921 if (c == 'u') { 922 ucl_chunk_skipc (chunk, p); 923 for (i = 0; i < 4 && p < chunk->end; i ++) { 924 if (!isxdigit (*p)) { 925 ucl_set_err (parser, UCL_ESYNTAX, "invalid utf escape", 926 &parser->err); 927 return false; 928 } 929 ucl_chunk_skipc (chunk, p); 930 } 931 if (p >= chunk->end) { 932 ucl_set_err (parser, UCL_ESYNTAX, "unfinished escape character", 933 &parser->err); 934 return false; 935 } 936 } 937 else { 938 ucl_chunk_skipc (chunk, p); 939 } 940 } 941 *need_unescape = true; 942 *ucl_escape = true; 943 continue; 944 } 945 else if (c == '"') { 946 ucl_chunk_skipc (chunk, p); 947 return true; 948 } 949 else if (ucl_test_character (c, UCL_CHARACTER_UCL_UNSAFE)) { 950 *ucl_escape = true; 951 } 952 else if (c == '$') { 953 *var_expand = true; 954 } 955 ucl_chunk_skipc (chunk, p); 956 } 957 958 ucl_set_err (parser, UCL_ESYNTAX, "no quote at the end of json string", 959 &parser->err); 960 return false; 961 } 962 963 static void 964 ucl_parser_append_elt (struct ucl_parser *parser, ucl_hash_t *cont, 965 ucl_object_t *top, 966 ucl_object_t *elt) 967 { 968 ucl_object_t *nobj; 969 970 if ((parser->flags & UCL_PARSER_NO_IMPLICIT_ARRAYS) == 0) { 971 /* Implicit array */ 972 top->flags |= UCL_OBJECT_MULTIVALUE; 973 DL_APPEND (top, elt); 974 } 975 else { 976 if ((top->flags & UCL_OBJECT_MULTIVALUE) != 0) { 977 /* Just add to the explicit array */ 978 DL_APPEND (top->value.av, elt); 979 } 980 else { 981 /* Convert to an array */ 982 ucl_hash_delete (cont, top); 983 nobj = ucl_object_typed_new (UCL_ARRAY); 984 nobj->key = top->key; 985 nobj->keylen = top->keylen; 986 nobj->flags |= UCL_OBJECT_MULTIVALUE; 987 DL_APPEND (nobj->value.av, top); 988 DL_APPEND (nobj->value.av, elt); 989 ucl_hash_insert (cont, nobj, nobj->key, nobj->keylen); 990 } 991 } 992 } 993 994 /** 995 * Parse a key in an object 996 * @param parser 997 * @param chunk 998 * @return true if a key has been parsed 999 */ 1000 static bool 1001 ucl_parse_key (struct ucl_parser *parser, struct ucl_chunk *chunk, bool *next_key, bool *end_of_object) 1002 { 1003 const unsigned char *p, *c = NULL, *end, *t; 1004 const char *key = NULL; 1005 bool got_quote = false, got_eq = false, got_semicolon = false, 1006 need_unescape = false, ucl_escape = false, var_expand = false, 1007 got_content = false, got_sep = false; 1008 ucl_object_t *nobj, *tobj; 1009 ucl_hash_t *container; 1010 ssize_t keylen; 1011 1012 p = chunk->pos; 1013 1014 if (*p == '.') { 1015 /* It is macro actually */ 1016 ucl_chunk_skipc (chunk, p); 1017 parser->prev_state = parser->state; 1018 parser->state = UCL_STATE_MACRO_NAME; 1019 return true; 1020 } 1021 while (p < chunk->end) { 1022 /* 1023 * A key must start with alpha, number, '/' or '_' and end with space character 1024 */ 1025 if (c == NULL) { 1026 if (chunk->remain >= 2 && ucl_lex_is_comment (p[0], p[1])) { 1027 if (!ucl_skip_comments (parser)) { 1028 return false; 1029 } 1030 p = chunk->pos; 1031 } 1032 else if (ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE)) { 1033 ucl_chunk_skipc (chunk, p); 1034 } 1035 else if (ucl_test_character (*p, UCL_CHARACTER_KEY_START)) { 1036 /* The first symbol */ 1037 c = p; 1038 ucl_chunk_skipc (chunk, p); 1039 got_content = true; 1040 } 1041 else if (*p == '"') { 1042 /* JSON style key */ 1043 c = p + 1; 1044 got_quote = true; 1045 got_content = true; 1046 ucl_chunk_skipc (chunk, p); 1047 } 1048 else if (*p == '}') { 1049 /* We have actually end of an object */ 1050 *end_of_object = true; 1051 return true; 1052 } 1053 else if (*p == '.') { 1054 ucl_chunk_skipc (chunk, p); 1055 parser->prev_state = parser->state; 1056 parser->state = UCL_STATE_MACRO_NAME; 1057 return true; 1058 } 1059 else { 1060 /* Invalid identifier */ 1061 ucl_set_err (parser, UCL_ESYNTAX, "key must begin with a letter", 1062 &parser->err); 1063 return false; 1064 } 1065 } 1066 else { 1067 /* Parse the body of a key */ 1068 if (!got_quote) { 1069 if (ucl_test_character (*p, UCL_CHARACTER_KEY)) { 1070 got_content = true; 1071 ucl_chunk_skipc (chunk, p); 1072 } 1073 else if (ucl_test_character (*p, UCL_CHARACTER_KEY_SEP)) { 1074 end = p; 1075 break; 1076 } 1077 else { 1078 ucl_set_err (parser, UCL_ESYNTAX, "invalid character in a key", 1079 &parser->err); 1080 return false; 1081 } 1082 } 1083 else { 1084 /* We need to parse json like quoted string */ 1085 if (!ucl_lex_json_string (parser, chunk, &need_unescape, &ucl_escape, &var_expand)) { 1086 return false; 1087 } 1088 /* Always escape keys obtained via json */ 1089 end = chunk->pos - 1; 1090 p = chunk->pos; 1091 break; 1092 } 1093 } 1094 } 1095 1096 if (p >= chunk->end && got_content) { 1097 ucl_set_err (parser, UCL_ESYNTAX, "unfinished key", &parser->err); 1098 return false; 1099 } 1100 else if (!got_content) { 1101 return true; 1102 } 1103 *end_of_object = false; 1104 /* We are now at the end of the key, need to parse the rest */ 1105 while (p < chunk->end) { 1106 if (ucl_test_character (*p, UCL_CHARACTER_WHITESPACE)) { 1107 ucl_chunk_skipc (chunk, p); 1108 } 1109 else if (*p == '=') { 1110 if (!got_eq && !got_semicolon) { 1111 ucl_chunk_skipc (chunk, p); 1112 got_eq = true; 1113 } 1114 else { 1115 ucl_set_err (parser, UCL_ESYNTAX, "unexpected '=' character", 1116 &parser->err); 1117 return false; 1118 } 1119 } 1120 else if (*p == ':') { 1121 if (!got_eq && !got_semicolon) { 1122 ucl_chunk_skipc (chunk, p); 1123 got_semicolon = true; 1124 } 1125 else { 1126 ucl_set_err (parser, UCL_ESYNTAX, "unexpected ':' character", 1127 &parser->err); 1128 return false; 1129 } 1130 } 1131 else if (chunk->remain >= 2 && ucl_lex_is_comment (p[0], p[1])) { 1132 /* Check for comment */ 1133 if (!ucl_skip_comments (parser)) { 1134 return false; 1135 } 1136 p = chunk->pos; 1137 } 1138 else { 1139 /* Start value */ 1140 break; 1141 } 1142 } 1143 1144 if (p >= chunk->end && got_content) { 1145 ucl_set_err (parser, UCL_ESYNTAX, "unfinished key", &parser->err); 1146 return false; 1147 } 1148 1149 got_sep = got_semicolon || got_eq; 1150 1151 if (!got_sep) { 1152 /* 1153 * Maybe we have more keys nested, so search for termination character. 1154 * Possible choices: 1155 * 1) key1 key2 ... keyN [:=] value <- we treat that as error 1156 * 2) key1 ... keyN {} or [] <- we treat that as nested objects 1157 * 3) key1 value[;,\n] <- we treat that as linear object 1158 */ 1159 t = p; 1160 *next_key = false; 1161 while (ucl_test_character (*t, UCL_CHARACTER_WHITESPACE)) { 1162 t ++; 1163 } 1164 /* Check first non-space character after a key */ 1165 if (*t != '{' && *t != '[') { 1166 while (t < chunk->end) { 1167 if (*t == ',' || *t == ';' || *t == '\n' || *t == '\r') { 1168 break; 1169 } 1170 else if (*t == '{' || *t == '[') { 1171 *next_key = true; 1172 break; 1173 } 1174 t ++; 1175 } 1176 } 1177 } 1178 1179 /* Create a new object */ 1180 nobj = ucl_object_new_full (UCL_NULL, parser->chunks->priority); 1181 keylen = ucl_copy_or_store_ptr (parser, c, &nobj->trash_stack[UCL_TRASH_KEY], 1182 &key, end - c, need_unescape, parser->flags & UCL_PARSER_KEY_LOWERCASE, false); 1183 if (keylen == -1) { 1184 ucl_object_unref (nobj); 1185 return false; 1186 } 1187 else if (keylen == 0) { 1188 ucl_set_err (parser, UCL_ESYNTAX, "empty keys are not allowed", &parser->err); 1189 ucl_object_unref (nobj); 1190 return false; 1191 } 1192 1193 container = parser->stack->obj->value.ov; 1194 nobj->key = key; 1195 nobj->keylen = keylen; 1196 tobj = __DECONST (ucl_object_t *, ucl_hash_search_obj (container, nobj)); 1197 if (tobj == NULL) { 1198 container = ucl_hash_insert_object (container, nobj); 1199 nobj->prev = nobj; 1200 nobj->next = NULL; 1201 parser->stack->obj->len ++; 1202 } 1203 else { 1204 /* 1205 * The logic here is the following: 1206 * 1207 * - if we have two objects with the same priority, then we form an 1208 * implicit or explicit array 1209 * - if a new object has bigger priority, then we overwrite an old one 1210 * - if a new object has lower priority, then we ignore it 1211 */ 1212 unsigned priold = ucl_object_get_priority (tobj), 1213 prinew = ucl_object_get_priority (nobj); 1214 if (priold == prinew) { 1215 ucl_parser_append_elt (parser, container, tobj, nobj); 1216 } 1217 else if (priold > prinew) { 1218 ucl_object_unref (nobj); 1219 return true; 1220 } 1221 else { 1222 ucl_hash_replace (container, tobj, nobj); 1223 ucl_object_unref (tobj); 1224 } 1225 } 1226 1227 if (ucl_escape) { 1228 nobj->flags |= UCL_OBJECT_NEED_KEY_ESCAPE; 1229 } 1230 parser->stack->obj->value.ov = container; 1231 1232 parser->cur_obj = nobj; 1233 1234 return true; 1235 } 1236 1237 /** 1238 * Parse a cl string 1239 * @param parser 1240 * @param chunk 1241 * @return true if a key has been parsed 1242 */ 1243 static bool 1244 ucl_parse_string_value (struct ucl_parser *parser, 1245 struct ucl_chunk *chunk, bool *var_expand, bool *need_unescape) 1246 { 1247 const unsigned char *p; 1248 enum { 1249 UCL_BRACE_ROUND = 0, 1250 UCL_BRACE_SQUARE, 1251 UCL_BRACE_FIGURE 1252 }; 1253 int braces[3][2] = {{0, 0}, {0, 0}, {0, 0}}; 1254 1255 p = chunk->pos; 1256 1257 while (p < chunk->end) { 1258 1259 /* Skip pairs of figure braces */ 1260 if (*p == '{') { 1261 braces[UCL_BRACE_FIGURE][0] ++; 1262 } 1263 else if (*p == '}') { 1264 braces[UCL_BRACE_FIGURE][1] ++; 1265 if (braces[UCL_BRACE_FIGURE][1] <= braces[UCL_BRACE_FIGURE][0]) { 1266 /* This is not a termination symbol, continue */ 1267 ucl_chunk_skipc (chunk, p); 1268 continue; 1269 } 1270 } 1271 /* Skip pairs of square braces */ 1272 else if (*p == '[') { 1273 braces[UCL_BRACE_SQUARE][0] ++; 1274 } 1275 else if (*p == ']') { 1276 braces[UCL_BRACE_SQUARE][1] ++; 1277 if (braces[UCL_BRACE_SQUARE][1] <= braces[UCL_BRACE_SQUARE][0]) { 1278 /* This is not a termination symbol, continue */ 1279 ucl_chunk_skipc (chunk, p); 1280 continue; 1281 } 1282 } 1283 else if (*p == '$') { 1284 *var_expand = true; 1285 } 1286 else if (*p == '\\') { 1287 *need_unescape = true; 1288 ucl_chunk_skipc (chunk, p); 1289 if (p < chunk->end) { 1290 ucl_chunk_skipc (chunk, p); 1291 } 1292 continue; 1293 } 1294 1295 if (ucl_lex_is_atom_end (*p) || (chunk->remain >= 2 && ucl_lex_is_comment (p[0], p[1]))) { 1296 break; 1297 } 1298 ucl_chunk_skipc (chunk, p); 1299 } 1300 1301 return true; 1302 } 1303 1304 /** 1305 * Parse multiline string ending with \n{term}\n 1306 * @param parser 1307 * @param chunk 1308 * @param term 1309 * @param term_len 1310 * @return size of multiline string or 0 in case of error 1311 */ 1312 static int 1313 ucl_parse_multiline_string (struct ucl_parser *parser, 1314 struct ucl_chunk *chunk, const unsigned char *term, 1315 int term_len, unsigned char const **beg, 1316 bool *var_expand) 1317 { 1318 const unsigned char *p, *c, *tend; 1319 bool newline = false; 1320 int len = 0; 1321 1322 p = chunk->pos; 1323 1324 c = p; 1325 1326 while (p < chunk->end) { 1327 if (newline) { 1328 if (chunk->end - p < term_len) { 1329 return 0; 1330 } 1331 else if (memcmp (p, term, term_len) == 0) { 1332 tend = p + term_len; 1333 if (*tend != '\n' && *tend != ';' && *tend != ',') { 1334 /* Incomplete terminator */ 1335 ucl_chunk_skipc (chunk, p); 1336 continue; 1337 } 1338 len = p - c; 1339 chunk->remain -= term_len; 1340 chunk->pos = p + term_len; 1341 chunk->column = term_len; 1342 *beg = c; 1343 break; 1344 } 1345 } 1346 if (*p == '\n') { 1347 newline = true; 1348 } 1349 else { 1350 if (*p == '$') { 1351 *var_expand = true; 1352 } 1353 newline = false; 1354 } 1355 ucl_chunk_skipc (chunk, p); 1356 } 1357 1358 return len; 1359 } 1360 1361 static ucl_object_t* 1362 ucl_get_value_object (struct ucl_parser *parser) 1363 { 1364 ucl_object_t *t, *obj = NULL; 1365 1366 if (parser->stack->obj->type == UCL_ARRAY) { 1367 /* Object must be allocated */ 1368 obj = ucl_object_new_full (UCL_NULL, parser->chunks->priority); 1369 t = parser->stack->obj->value.av; 1370 DL_APPEND (t, obj); 1371 parser->cur_obj = obj; 1372 parser->stack->obj->value.av = t; 1373 parser->stack->obj->len ++; 1374 } 1375 else { 1376 /* Object has been already allocated */ 1377 obj = parser->cur_obj; 1378 } 1379 1380 return obj; 1381 } 1382 1383 /** 1384 * Handle value data 1385 * @param parser 1386 * @param chunk 1387 * @return 1388 */ 1389 static bool 1390 ucl_parse_value (struct ucl_parser *parser, struct ucl_chunk *chunk) 1391 { 1392 const unsigned char *p, *c; 1393 ucl_object_t *obj = NULL; 1394 unsigned int stripped_spaces; 1395 int str_len; 1396 bool need_unescape = false, ucl_escape = false, var_expand = false; 1397 1398 p = chunk->pos; 1399 1400 /* Skip any spaces and comments */ 1401 if (ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE) || 1402 (chunk->remain >= 2 && ucl_lex_is_comment (p[0], p[1]))) { 1403 while (p < chunk->end && ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE)) { 1404 ucl_chunk_skipc (chunk, p); 1405 } 1406 if (!ucl_skip_comments (parser)) { 1407 return false; 1408 } 1409 p = chunk->pos; 1410 } 1411 1412 while (p < chunk->end) { 1413 c = p; 1414 switch (*p) { 1415 case '"': 1416 obj = ucl_get_value_object (parser); 1417 ucl_chunk_skipc (chunk, p); 1418 if (!ucl_lex_json_string (parser, chunk, &need_unescape, &ucl_escape, &var_expand)) { 1419 return false; 1420 } 1421 str_len = chunk->pos - c - 2; 1422 obj->type = UCL_STRING; 1423 if ((str_len = ucl_copy_or_store_ptr (parser, c + 1, &obj->trash_stack[UCL_TRASH_VALUE], 1424 &obj->value.sv, str_len, need_unescape, false, var_expand)) == -1) { 1425 return false; 1426 } 1427 obj->len = str_len; 1428 parser->state = UCL_STATE_AFTER_VALUE; 1429 p = chunk->pos; 1430 return true; 1431 break; 1432 case '{': 1433 obj = ucl_get_value_object (parser); 1434 /* We have a new object */ 1435 obj = ucl_add_parser_stack (obj, parser, false, parser->stack->level); 1436 if (obj == NULL) { 1437 return false; 1438 } 1439 1440 ucl_chunk_skipc (chunk, p); 1441 return true; 1442 break; 1443 case '[': 1444 obj = ucl_get_value_object (parser); 1445 /* We have a new array */ 1446 obj = ucl_add_parser_stack (obj, parser, true, parser->stack->level); 1447 if (obj == NULL) { 1448 return false; 1449 } 1450 1451 ucl_chunk_skipc (chunk, p); 1452 return true; 1453 break; 1454 case ']': 1455 /* We have the array ending */ 1456 if (parser->stack && parser->stack->obj->type == UCL_ARRAY) { 1457 parser->state = UCL_STATE_AFTER_VALUE; 1458 return true; 1459 } 1460 else { 1461 goto parse_string; 1462 } 1463 break; 1464 case '<': 1465 obj = ucl_get_value_object (parser); 1466 /* We have something like multiline value, which must be <<[A-Z]+\n */ 1467 if (chunk->end - p > 3) { 1468 if (memcmp (p, "<<", 2) == 0) { 1469 p += 2; 1470 /* We allow only uppercase characters in multiline definitions */ 1471 while (p < chunk->end && *p >= 'A' && *p <= 'Z') { 1472 p ++; 1473 } 1474 if (*p =='\n') { 1475 /* Set chunk positions and start multiline parsing */ 1476 c += 2; 1477 chunk->remain -= p - c; 1478 chunk->pos = p + 1; 1479 chunk->column = 0; 1480 chunk->line ++; 1481 if ((str_len = ucl_parse_multiline_string (parser, chunk, c, 1482 p - c, &c, &var_expand)) == 0) { 1483 ucl_set_err (parser, UCL_ESYNTAX, 1484 "unterminated multiline value", &parser->err); 1485 return false; 1486 } 1487 obj->type = UCL_STRING; 1488 if ((str_len = ucl_copy_or_store_ptr (parser, c, &obj->trash_stack[UCL_TRASH_VALUE], 1489 &obj->value.sv, str_len - 1, false, false, var_expand)) == -1) { 1490 return false; 1491 } 1492 obj->len = str_len; 1493 parser->state = UCL_STATE_AFTER_VALUE; 1494 return true; 1495 } 1496 } 1497 } 1498 /* Fallback to ordinary strings */ 1499 default: 1500 parse_string: 1501 if (obj == NULL) { 1502 obj = ucl_get_value_object (parser); 1503 } 1504 /* Parse atom */ 1505 if (ucl_test_character (*p, UCL_CHARACTER_VALUE_DIGIT_START)) { 1506 if (!ucl_lex_number (parser, chunk, obj)) { 1507 if (parser->state == UCL_STATE_ERROR) { 1508 return false; 1509 } 1510 } 1511 else { 1512 parser->state = UCL_STATE_AFTER_VALUE; 1513 return true; 1514 } 1515 /* Fallback to normal string */ 1516 } 1517 1518 if (!ucl_parse_string_value (parser, chunk, &var_expand, &need_unescape)) { 1519 return false; 1520 } 1521 /* Cut trailing spaces */ 1522 stripped_spaces = 0; 1523 while (ucl_test_character (*(chunk->pos - 1 - stripped_spaces), 1524 UCL_CHARACTER_WHITESPACE)) { 1525 stripped_spaces ++; 1526 } 1527 str_len = chunk->pos - c - stripped_spaces; 1528 if (str_len <= 0) { 1529 ucl_set_err (parser, 0, "string value must not be empty", 1530 &parser->err); 1531 return false; 1532 } 1533 else if (str_len == 4 && memcmp (c, "null", 4) == 0) { 1534 obj->len = 0; 1535 obj->type = UCL_NULL; 1536 } 1537 else if (!ucl_maybe_parse_boolean (obj, c, str_len)) { 1538 obj->type = UCL_STRING; 1539 if ((str_len = ucl_copy_or_store_ptr (parser, c, &obj->trash_stack[UCL_TRASH_VALUE], 1540 &obj->value.sv, str_len, need_unescape, 1541 false, var_expand)) == -1) { 1542 return false; 1543 } 1544 obj->len = str_len; 1545 } 1546 parser->state = UCL_STATE_AFTER_VALUE; 1547 p = chunk->pos; 1548 1549 return true; 1550 break; 1551 } 1552 } 1553 1554 return true; 1555 } 1556 1557 /** 1558 * Handle after value data 1559 * @param parser 1560 * @param chunk 1561 * @return 1562 */ 1563 static bool 1564 ucl_parse_after_value (struct ucl_parser *parser, struct ucl_chunk *chunk) 1565 { 1566 const unsigned char *p; 1567 bool got_sep = false; 1568 struct ucl_stack *st; 1569 1570 p = chunk->pos; 1571 1572 while (p < chunk->end) { 1573 if (ucl_test_character (*p, UCL_CHARACTER_WHITESPACE)) { 1574 /* Skip whitespaces */ 1575 ucl_chunk_skipc (chunk, p); 1576 } 1577 else if (chunk->remain >= 2 && ucl_lex_is_comment (p[0], p[1])) { 1578 /* Skip comment */ 1579 if (!ucl_skip_comments (parser)) { 1580 return false; 1581 } 1582 /* Treat comment as a separator */ 1583 got_sep = true; 1584 p = chunk->pos; 1585 } 1586 else if (ucl_test_character (*p, UCL_CHARACTER_VALUE_END)) { 1587 if (*p == '}' || *p == ']') { 1588 if (parser->stack == NULL) { 1589 ucl_set_err (parser, UCL_ESYNTAX, 1590 "end of array or object detected without corresponding start", 1591 &parser->err); 1592 return false; 1593 } 1594 if ((*p == '}' && parser->stack->obj->type == UCL_OBJECT) || 1595 (*p == ']' && parser->stack->obj->type == UCL_ARRAY)) { 1596 1597 /* Pop all nested objects from a stack */ 1598 st = parser->stack; 1599 parser->stack = st->next; 1600 UCL_FREE (sizeof (struct ucl_stack), st); 1601 1602 while (parser->stack != NULL) { 1603 st = parser->stack; 1604 if (st->next == NULL || st->next->level == st->level) { 1605 break; 1606 } 1607 parser->stack = st->next; 1608 UCL_FREE (sizeof (struct ucl_stack), st); 1609 } 1610 } 1611 else { 1612 ucl_set_err (parser, UCL_ESYNTAX, 1613 "unexpected terminating symbol detected", 1614 &parser->err); 1615 return false; 1616 } 1617 1618 if (parser->stack == NULL) { 1619 /* Ignore everything after a top object */ 1620 return true; 1621 } 1622 else { 1623 ucl_chunk_skipc (chunk, p); 1624 } 1625 got_sep = true; 1626 } 1627 else { 1628 /* Got a separator */ 1629 got_sep = true; 1630 ucl_chunk_skipc (chunk, p); 1631 } 1632 } 1633 else { 1634 /* Anything else */ 1635 if (!got_sep) { 1636 ucl_set_err (parser, UCL_ESYNTAX, "delimiter is missing", 1637 &parser->err); 1638 return false; 1639 } 1640 return true; 1641 } 1642 } 1643 1644 return true; 1645 } 1646 1647 /** 1648 * Handle macro data 1649 * @param parser 1650 * @param chunk 1651 * @return 1652 */ 1653 static bool 1654 ucl_parse_macro_value (struct ucl_parser *parser, 1655 struct ucl_chunk *chunk, struct ucl_macro *macro, 1656 unsigned char const **macro_start, size_t *macro_len) 1657 { 1658 const unsigned char *p, *c; 1659 bool need_unescape = false, ucl_escape = false, var_expand = false; 1660 1661 p = chunk->pos; 1662 1663 switch (*p) { 1664 case '"': 1665 /* We have macro value encoded in quotes */ 1666 c = p; 1667 ucl_chunk_skipc (chunk, p); 1668 if (!ucl_lex_json_string (parser, chunk, &need_unescape, &ucl_escape, &var_expand)) { 1669 return false; 1670 } 1671 1672 *macro_start = c + 1; 1673 *macro_len = chunk->pos - c - 2; 1674 p = chunk->pos; 1675 break; 1676 case '{': 1677 /* We got a multiline macro body */ 1678 ucl_chunk_skipc (chunk, p); 1679 /* Skip spaces at the beginning */ 1680 while (p < chunk->end) { 1681 if (ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE)) { 1682 ucl_chunk_skipc (chunk, p); 1683 } 1684 else { 1685 break; 1686 } 1687 } 1688 c = p; 1689 while (p < chunk->end) { 1690 if (*p == '}') { 1691 break; 1692 } 1693 ucl_chunk_skipc (chunk, p); 1694 } 1695 *macro_start = c; 1696 *macro_len = p - c; 1697 ucl_chunk_skipc (chunk, p); 1698 break; 1699 default: 1700 /* Macro is not enclosed in quotes or braces */ 1701 c = p; 1702 while (p < chunk->end) { 1703 if (ucl_lex_is_atom_end (*p)) { 1704 break; 1705 } 1706 ucl_chunk_skipc (chunk, p); 1707 } 1708 *macro_start = c; 1709 *macro_len = p - c; 1710 break; 1711 } 1712 1713 /* We are at the end of a macro */ 1714 /* Skip ';' and space characters and return to previous state */ 1715 while (p < chunk->end) { 1716 if (!ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE) && *p != ';') { 1717 break; 1718 } 1719 ucl_chunk_skipc (chunk, p); 1720 } 1721 return true; 1722 } 1723 1724 /** 1725 * Parse macro arguments as UCL object 1726 * @param parser parser structure 1727 * @param chunk the current data chunk 1728 * @return 1729 */ 1730 static ucl_object_t * 1731 ucl_parse_macro_arguments (struct ucl_parser *parser, 1732 struct ucl_chunk *chunk) 1733 { 1734 ucl_object_t *res = NULL; 1735 struct ucl_parser *params_parser; 1736 int obraces = 1, ebraces = 0, state = 0; 1737 const unsigned char *p, *c; 1738 size_t args_len = 0; 1739 struct ucl_parser_saved_state saved; 1740 1741 saved.column = chunk->column; 1742 saved.line = chunk->line; 1743 saved.pos = chunk->pos; 1744 saved.remain = chunk->remain; 1745 p = chunk->pos; 1746 1747 if (*p != '(' || chunk->remain < 2) { 1748 return NULL; 1749 } 1750 1751 /* Set begin and start */ 1752 ucl_chunk_skipc (chunk, p); 1753 c = p; 1754 1755 while ((p) < (chunk)->end) { 1756 switch (state) { 1757 case 0: 1758 /* Parse symbols and check for '(', ')' and '"' */ 1759 if (*p == '(') { 1760 obraces ++; 1761 } 1762 else if (*p == ')') { 1763 ebraces ++; 1764 } 1765 else if (*p == '"') { 1766 state = 1; 1767 } 1768 /* Check pairing */ 1769 if (obraces == ebraces) { 1770 state = 99; 1771 } 1772 else { 1773 args_len ++; 1774 } 1775 /* Check overflow */ 1776 if (chunk->remain == 0) { 1777 goto restore_chunk; 1778 } 1779 ucl_chunk_skipc (chunk, p); 1780 break; 1781 case 1: 1782 /* We have quote character, so skip all but quotes */ 1783 if (*p == '"' && *(p - 1) != '\\') { 1784 state = 0; 1785 } 1786 if (chunk->remain == 0) { 1787 goto restore_chunk; 1788 } 1789 ucl_chunk_skipc (chunk, p); 1790 break; 1791 case 99: 1792 /* 1793 * We have read the full body of arguments, so we need to parse and set 1794 * object from that 1795 */ 1796 params_parser = ucl_parser_new (parser->flags); 1797 if (!ucl_parser_add_chunk (params_parser, c, args_len)) { 1798 ucl_set_err (parser, UCL_ESYNTAX, "macro arguments parsing error", 1799 &parser->err); 1800 } 1801 else { 1802 res = ucl_parser_get_object (params_parser); 1803 } 1804 ucl_parser_free (params_parser); 1805 1806 return res; 1807 1808 break; 1809 } 1810 } 1811 1812 return res; 1813 1814 restore_chunk: 1815 chunk->column = saved.column; 1816 chunk->line = saved.line; 1817 chunk->pos = saved.pos; 1818 chunk->remain = saved.remain; 1819 1820 return NULL; 1821 } 1822 1823 #define SKIP_SPACES_COMMENTS(parser, chunk, p) do { \ 1824 while ((p) < (chunk)->end) { \ 1825 if (!ucl_test_character (*(p), UCL_CHARACTER_WHITESPACE_UNSAFE)) { \ 1826 if ((chunk)->remain >= 2 && ucl_lex_is_comment ((p)[0], (p)[1])) { \ 1827 if (!ucl_skip_comments (parser)) { \ 1828 return false; \ 1829 } \ 1830 p = (chunk)->pos; \ 1831 } \ 1832 break; \ 1833 } \ 1834 ucl_chunk_skipc (chunk, p); \ 1835 } \ 1836 } while(0) 1837 1838 /** 1839 * Handle the main states of rcl parser 1840 * @param parser parser structure 1841 * @param data the pointer to the beginning of a chunk 1842 * @param len the length of a chunk 1843 * @return true if chunk has been parsed and false in case of error 1844 */ 1845 static bool 1846 ucl_state_machine (struct ucl_parser *parser) 1847 { 1848 ucl_object_t *obj, *macro_args; 1849 struct ucl_chunk *chunk = parser->chunks; 1850 const unsigned char *p, *c = NULL, *macro_start = NULL; 1851 unsigned char *macro_escaped; 1852 size_t macro_len = 0; 1853 struct ucl_macro *macro = NULL; 1854 bool next_key = false, end_of_object = false, ret; 1855 1856 if (parser->top_obj == NULL) { 1857 if (*chunk->pos == '[') { 1858 obj = ucl_add_parser_stack (NULL, parser, true, 0); 1859 } 1860 else { 1861 obj = ucl_add_parser_stack (NULL, parser, false, 0); 1862 } 1863 if (obj == NULL) { 1864 return false; 1865 } 1866 parser->top_obj = obj; 1867 parser->cur_obj = obj; 1868 parser->state = UCL_STATE_INIT; 1869 } 1870 1871 p = chunk->pos; 1872 while (chunk->pos < chunk->end) { 1873 switch (parser->state) { 1874 case UCL_STATE_INIT: 1875 /* 1876 * At the init state we can either go to the parse array or object 1877 * if we got [ or { correspondingly or can just treat new data as 1878 * a key of newly created object 1879 */ 1880 if (!ucl_skip_comments (parser)) { 1881 parser->prev_state = parser->state; 1882 parser->state = UCL_STATE_ERROR; 1883 return false; 1884 } 1885 else { 1886 /* Skip any spaces */ 1887 while (p < chunk->end && ucl_test_character (*p, 1888 UCL_CHARACTER_WHITESPACE_UNSAFE)) { 1889 ucl_chunk_skipc (chunk, p); 1890 } 1891 p = chunk->pos; 1892 if (*p == '[') { 1893 parser->state = UCL_STATE_VALUE; 1894 ucl_chunk_skipc (chunk, p); 1895 } 1896 else { 1897 parser->state = UCL_STATE_KEY; 1898 if (*p == '{') { 1899 ucl_chunk_skipc (chunk, p); 1900 } 1901 } 1902 } 1903 break; 1904 case UCL_STATE_KEY: 1905 /* Skip any spaces */ 1906 while (p < chunk->end && ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE)) { 1907 ucl_chunk_skipc (chunk, p); 1908 } 1909 if (*p == '}') { 1910 /* We have the end of an object */ 1911 parser->state = UCL_STATE_AFTER_VALUE; 1912 continue; 1913 } 1914 if (parser->stack == NULL) { 1915 /* No objects are on stack, but we want to parse a key */ 1916 ucl_set_err (parser, UCL_ESYNTAX, "top object is finished but the parser " 1917 "expects a key", &parser->err); 1918 parser->prev_state = parser->state; 1919 parser->state = UCL_STATE_ERROR; 1920 return false; 1921 } 1922 if (!ucl_parse_key (parser, chunk, &next_key, &end_of_object)) { 1923 parser->prev_state = parser->state; 1924 parser->state = UCL_STATE_ERROR; 1925 return false; 1926 } 1927 if (end_of_object) { 1928 p = chunk->pos; 1929 parser->state = UCL_STATE_AFTER_VALUE; 1930 continue; 1931 } 1932 else if (parser->state != UCL_STATE_MACRO_NAME) { 1933 if (next_key && parser->stack->obj->type == UCL_OBJECT) { 1934 /* Parse more keys and nest objects accordingly */ 1935 obj = ucl_add_parser_stack (parser->cur_obj, parser, false, 1936 parser->stack->level + 1); 1937 if (obj == NULL) { 1938 return false; 1939 } 1940 } 1941 else { 1942 parser->state = UCL_STATE_VALUE; 1943 } 1944 } 1945 else { 1946 c = chunk->pos; 1947 } 1948 p = chunk->pos; 1949 break; 1950 case UCL_STATE_VALUE: 1951 /* We need to check what we do have */ 1952 if (!ucl_parse_value (parser, chunk)) { 1953 parser->prev_state = parser->state; 1954 parser->state = UCL_STATE_ERROR; 1955 return false; 1956 } 1957 /* State is set in ucl_parse_value call */ 1958 p = chunk->pos; 1959 break; 1960 case UCL_STATE_AFTER_VALUE: 1961 if (!ucl_parse_after_value (parser, chunk)) { 1962 parser->prev_state = parser->state; 1963 parser->state = UCL_STATE_ERROR; 1964 return false; 1965 } 1966 if (parser->stack != NULL) { 1967 if (parser->stack->obj->type == UCL_OBJECT) { 1968 parser->state = UCL_STATE_KEY; 1969 } 1970 else { 1971 /* Array */ 1972 parser->state = UCL_STATE_VALUE; 1973 } 1974 } 1975 else { 1976 /* Skip everything at the end */ 1977 return true; 1978 } 1979 p = chunk->pos; 1980 break; 1981 case UCL_STATE_MACRO_NAME: 1982 if (!ucl_test_character (*p, UCL_CHARACTER_WHITESPACE_UNSAFE) && 1983 *p != '(') { 1984 ucl_chunk_skipc (chunk, p); 1985 } 1986 else if (p - c > 0) { 1987 /* We got macro name */ 1988 macro_len = (size_t)(p - c); 1989 HASH_FIND (hh, parser->macroes, c, macro_len, macro); 1990 if (macro == NULL) { 1991 ucl_create_err (&parser->err, "error on line %d at column %d: " 1992 "unknown macro: '%.*s', character: '%c'", 1993 chunk->line, chunk->column, (int)(p - c), c, *chunk->pos); 1994 parser->state = UCL_STATE_ERROR; 1995 return false; 1996 } 1997 /* Now we need to skip all spaces */ 1998 SKIP_SPACES_COMMENTS(parser, chunk, p); 1999 parser->state = UCL_STATE_MACRO; 2000 } 2001 break; 2002 case UCL_STATE_MACRO: 2003 if (*chunk->pos == '(') { 2004 macro_args = ucl_parse_macro_arguments (parser, chunk); 2005 p = chunk->pos; 2006 if (macro_args) { 2007 SKIP_SPACES_COMMENTS(parser, chunk, p); 2008 } 2009 } 2010 else { 2011 macro_args = NULL; 2012 } 2013 if (!ucl_parse_macro_value (parser, chunk, macro, 2014 ¯o_start, ¯o_len)) { 2015 parser->prev_state = parser->state; 2016 parser->state = UCL_STATE_ERROR; 2017 return false; 2018 } 2019 macro_len = ucl_expand_variable (parser, ¯o_escaped, 2020 macro_start, macro_len); 2021 parser->state = parser->prev_state; 2022 if (macro_escaped == NULL) { 2023 ret = macro->handler (macro_start, macro_len, macro_args, 2024 macro->ud); 2025 } 2026 else { 2027 ret = macro->handler (macro_escaped, macro_len, macro_args, 2028 macro->ud); 2029 UCL_FREE (macro_len + 1, macro_escaped); 2030 } 2031 p = chunk->pos; 2032 if (macro_args) { 2033 ucl_object_unref (macro_args); 2034 } 2035 if (!ret) { 2036 return false; 2037 } 2038 break; 2039 default: 2040 /* TODO: add all states */ 2041 ucl_set_err (parser, UCL_EINTERNAL, 2042 "internal error: parser is in an unknown state", &parser->err); 2043 parser->state = UCL_STATE_ERROR; 2044 return false; 2045 } 2046 } 2047 2048 return true; 2049 } 2050 2051 struct ucl_parser* 2052 ucl_parser_new (int flags) 2053 { 2054 struct ucl_parser *new; 2055 2056 new = UCL_ALLOC (sizeof (struct ucl_parser)); 2057 if (new == NULL) { 2058 return NULL; 2059 } 2060 memset (new, 0, sizeof (struct ucl_parser)); 2061 2062 ucl_parser_register_macro (new, "include", ucl_include_handler, new); 2063 ucl_parser_register_macro (new, "try_include", ucl_try_include_handler, new); 2064 ucl_parser_register_macro (new, "includes", ucl_includes_handler, new); 2065 2066 new->flags = flags; 2067 2068 /* Initial assumption about filevars */ 2069 ucl_parser_set_filevars (new, NULL, false); 2070 2071 return new; 2072 } 2073 2074 2075 void 2076 ucl_parser_register_macro (struct ucl_parser *parser, const char *macro, 2077 ucl_macro_handler handler, void* ud) 2078 { 2079 struct ucl_macro *new; 2080 2081 if (macro == NULL || handler == NULL) { 2082 return; 2083 } 2084 new = UCL_ALLOC (sizeof (struct ucl_macro)); 2085 if (new == NULL) { 2086 return; 2087 } 2088 memset (new, 0, sizeof (struct ucl_macro)); 2089 new->handler = handler; 2090 new->name = strdup (macro); 2091 new->ud = ud; 2092 HASH_ADD_KEYPTR (hh, parser->macroes, new->name, strlen (new->name), new); 2093 } 2094 2095 void 2096 ucl_parser_register_variable (struct ucl_parser *parser, const char *var, 2097 const char *value) 2098 { 2099 struct ucl_variable *new = NULL, *cur; 2100 2101 if (var == NULL) { 2102 return; 2103 } 2104 2105 /* Find whether a variable already exists */ 2106 LL_FOREACH (parser->variables, cur) { 2107 if (strcmp (cur->var, var) == 0) { 2108 new = cur; 2109 break; 2110 } 2111 } 2112 2113 if (value == NULL) { 2114 2115 if (new != NULL) { 2116 /* Remove variable */ 2117 DL_DELETE (parser->variables, new); 2118 free (new->var); 2119 free (new->value); 2120 UCL_FREE (sizeof (struct ucl_variable), new); 2121 } 2122 else { 2123 /* Do nothing */ 2124 return; 2125 } 2126 } 2127 else { 2128 if (new == NULL) { 2129 new = UCL_ALLOC (sizeof (struct ucl_variable)); 2130 if (new == NULL) { 2131 return; 2132 } 2133 memset (new, 0, sizeof (struct ucl_variable)); 2134 new->var = strdup (var); 2135 new->var_len = strlen (var); 2136 new->value = strdup (value); 2137 new->value_len = strlen (value); 2138 2139 DL_APPEND (parser->variables, new); 2140 } 2141 else { 2142 free (new->value); 2143 new->value = strdup (value); 2144 new->value_len = strlen (value); 2145 } 2146 } 2147 } 2148 2149 void 2150 ucl_parser_set_variables_handler (struct ucl_parser *parser, 2151 ucl_variable_handler handler, void *ud) 2152 { 2153 parser->var_handler = handler; 2154 parser->var_data = ud; 2155 } 2156 2157 bool 2158 ucl_parser_add_chunk_priority (struct ucl_parser *parser, const unsigned char *data, 2159 size_t len, unsigned priority) 2160 { 2161 struct ucl_chunk *chunk; 2162 2163 if (data == NULL) { 2164 ucl_create_err (&parser->err, "invalid chunk added"); 2165 return false; 2166 } 2167 if (len == 0) { 2168 parser->top_obj = ucl_object_new_full (UCL_OBJECT, priority); 2169 return true; 2170 } 2171 if (parser->state != UCL_STATE_ERROR) { 2172 chunk = UCL_ALLOC (sizeof (struct ucl_chunk)); 2173 if (chunk == NULL) { 2174 ucl_create_err (&parser->err, "cannot allocate chunk structure"); 2175 return false; 2176 } 2177 chunk->begin = data; 2178 chunk->remain = len; 2179 chunk->pos = chunk->begin; 2180 chunk->end = chunk->begin + len; 2181 chunk->line = 1; 2182 chunk->column = 0; 2183 chunk->priority = priority; 2184 LL_PREPEND (parser->chunks, chunk); 2185 parser->recursion ++; 2186 if (parser->recursion > UCL_MAX_RECURSION) { 2187 ucl_create_err (&parser->err, "maximum include nesting limit is reached: %d", 2188 parser->recursion); 2189 return false; 2190 } 2191 return ucl_state_machine (parser); 2192 } 2193 2194 ucl_create_err (&parser->err, "a parser is in an invalid state"); 2195 2196 return false; 2197 } 2198 2199 bool 2200 ucl_parser_add_chunk (struct ucl_parser *parser, const unsigned char *data, 2201 size_t len) 2202 { 2203 return ucl_parser_add_chunk_priority (parser, data, len, 0); 2204 } 2205 2206 bool 2207 ucl_parser_add_string (struct ucl_parser *parser, const char *data, 2208 size_t len) 2209 { 2210 if (data == NULL) { 2211 ucl_create_err (&parser->err, "invalid string added"); 2212 return false; 2213 } 2214 if (len == 0) { 2215 len = strlen (data); 2216 } 2217 2218 return ucl_parser_add_chunk (parser, (const unsigned char *)data, len); 2219 } 2220