nsz Git - cparser/blob - lexer.c

   1 #include <config.h>
   2
   3 #include "lexer.h"
   4 #include "token_t.h"
   5 #include "symbol_table_t.h"
   6 #include "adt/error.h"
   7 #include "adt/strset.h"
   8 #include "adt/util.h"
   9 #include "type_t.h"
  10
  11 #include <assert.h>
  12 #include <errno.h>
  13 #include <string.h>
  14 #include <stdbool.h>
  15 #include <ctype.h>
  16
  17 //#define DEBUG_CHARS
  18 #define MAX_PUTBACK 3
  19
  20 static int         c;
  21 token_t            lexer_token;
  22 symbol_t          *symbol_L;
  23 static FILE       *input;
  24 static char        buf[1024 + MAX_PUTBACK];
  25 static const char *bufend;
  26 static const char *bufpos;
  27 static strset_t    stringset;
  28
  29 static type_t     *type_int        = NULL;
  30 static type_t     *type_uint       = NULL;
  31 static type_t     *type_long       = NULL;
  32 static type_t     *type_ulong      = NULL;
  33 static type_t     *type_longlong   = NULL;
  34 static type_t     *type_ulonglong  = NULL;
  35 static type_t     *type_float      = NULL;
  36 static type_t     *type_double     = NULL;
  37 static type_t     *type_longdouble = NULL;
  38
  39 static void error_prefix_at(const char *input_name, unsigned linenr)
  40 {
  41         fprintf(stderr, "%s:%u: Error: ", input_name, linenr);
  42 }
  43
  44 static void error_prefix(void)
  45 {
  46         error_prefix_at(lexer_token.source_position.input_name,
  47                         lexer_token.source_position.linenr);
  48 }
  49
  50 static void parse_error(const char *msg)
  51 {
  52         error_prefix();
  53         fprintf(stderr, "%s\n", msg);
  54 }
  55
  56 static inline void next_real_char(void)
  57 {
  58         bufpos++;
  59         if(bufpos >= bufend) {
  60                 size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK,
  61                                  input);
  62                 if(s == 0) {
  63                         c = EOF;
  64                         return;
  65                 }
  66                 bufpos = buf + MAX_PUTBACK;
  67                 bufend = buf + MAX_PUTBACK + s;
  68         }
  69         c = *(bufpos);
  70 }
  71
  72 static inline void put_back(int pc)
  73 {
  74         assert(bufpos >= buf);
  75         //assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc);
  76
  77         char *p = buf + (bufpos - buf);
  78         *p = pc;
  79
  80         /* going backwards in the buffer is legal as long as it's not more often
  81          * than MAX_PUTBACK */
  82         bufpos--;
  83
  84 #ifdef DEBUG_CHARS
  85         printf("putback '%c'\n", pc);
  86 #endif
  87 }
  88
  89 static inline void next_char(void);
  90
  91 #define MATCH_NEWLINE(code)                   \
  92         case '\r':                                \
  93                 next_char();                          \
  94                 if(c == '\n') {                       \
  95                         next_char();                      \
  96                 }                                     \
  97                 lexer_token.source_position.linenr++; \
  98                 code;                                 \
  99         case '\n':                                \
 100                 next_char();                          \
 101                 lexer_token.source_position.linenr++; \
 102                 code;
 103
 104 #define eat(c_type)  do { assert(c == c_type); next_char(); } while(0)
 105
 106 static void maybe_concat_lines(void)
 107 {
 108         eat('\\');
 109
 110         switch(c) {
 111         MATCH_NEWLINE(return;)
 112
 113         default:
 114                 break;
 115         }
 116
 117         put_back(c);
 118         c = '\\';
 119 }
 120
 121 static inline void next_char(void)
 122 {
 123         next_real_char();
 124
 125         /* filter trigraphs */
 126         if(UNLIKELY(c == '\\')) {
 127                 maybe_concat_lines();
 128                 goto end_of_next_char;
 129         }
 130
 131         if(LIKELY(c != '?'))
 132                 goto end_of_next_char;
 133
 134         next_real_char();
 135         if(LIKELY(c != '?')) {
 136                 put_back(c);
 137                 c = '?';
 138                 goto end_of_next_char;
 139         }
 140
 141         next_real_char();
 142         switch(c) {
 143         case '=': c = '#'; break;
 144         case '(': c = '['; break;
 145         case '/': c = '\\'; maybe_concat_lines(); break;
 146         case ')': c = ']'; break;
 147         case '\'': c = '^'; break;
 148         case '<': c = '{'; break;
 149         case '!': c = '|'; break;
 150         case '>': c = '}'; break;
 151         case '-': c = '~'; break;
 152         default:
 153                 put_back('?');
 154                 put_back(c);
 155                 c = '?';
 156                 break;
 157         }
 158
 159 end_of_next_char:;
 160 #ifdef DEBUG_CHARS
 161         printf("nchar '%c'\n", c);
 162 #endif
 163 }
 164
 165 #define SYMBOL_CHARS  \
 166         case 'a':         \
 167         case 'b':         \
 168         case 'c':         \
 169         case 'd':         \
 170         case 'e':         \
 171         case 'f':         \
 172         case 'g':         \
 173         case 'h':         \
 174         case 'i':         \
 175         case 'j':         \
 176         case 'k':         \
 177         case 'l':         \
 178         case 'm':         \
 179         case 'n':         \
 180         case 'o':         \
 181         case 'p':         \
 182         case 'q':         \
 183         case 'r':         \
 184         case 's':         \
 185         case 't':         \
 186         case 'u':         \
 187         case 'v':         \
 188         case 'w':         \
 189         case 'x':         \
 190         case 'y':         \
 191         case 'z':         \
 192         case 'A':         \
 193         case 'B':         \
 194         case 'C':         \
 195         case 'D':         \
 196         case 'E':         \
 197         case 'F':         \
 198         case 'G':         \
 199         case 'H':         \
 200         case 'I':         \
 201         case 'J':         \
 202         case 'K':         \
 203         case 'L':         \
 204         case 'M':         \
 205         case 'N':         \
 206         case 'O':         \
 207         case 'P':         \
 208         case 'Q':         \
 209         case 'R':         \
 210         case 'S':         \
 211         case 'T':         \
 212         case 'U':         \
 213         case 'V':         \
 214         case 'W':         \
 215         case 'X':         \
 216         case 'Y':         \
 217         case 'Z':         \
 218         case '_':
 219
 220 #define DIGITS        \
 221         case '0':         \
 222         case '1':         \
 223         case '2':         \
 224         case '3':         \
 225         case '4':         \
 226         case '5':         \
 227         case '6':         \
 228         case '7':         \
 229         case '8':         \
 230         case '9':
 231
 232 static void parse_symbol(void)
 233 {
 234         symbol_t *symbol;
 235         char     *string;
 236
 237         obstack_1grow(&symbol_obstack, c);
 238         next_char();
 239
 240         while(1) {
 241                 switch(c) {
 242                 DIGITS
 243                 SYMBOL_CHARS
 244                         obstack_1grow(&symbol_obstack, c);
 245                         next_char();
 246                         break;
 247
 248                 default:
 249                         goto end_symbol;
 250                 }
 251         }
 252
 253 end_symbol:
 254         obstack_1grow(&symbol_obstack, '\0');
 255
 256         string = obstack_finish(&symbol_obstack);
 257         symbol = symbol_table_insert(string);
 258
 259         lexer_token.type     = symbol->ID;
 260         lexer_token.v.symbol = symbol;
 261
 262         if(symbol->string != string) {
 263                 obstack_free(&symbol_obstack, string);
 264         }
 265 }
 266
 267 static void parse_integer_suffix(void)
 268 {
 269         if(c == 'U' || c == 'U') {
 270                 next_char();
 271                 if(c == 'L' || c == 'l') {
 272                         next_char();
 273                         if(c == 'L' || c == 'l') {
 274                                 next_char();
 275                                 lexer_token.datatype = type_ulonglong;
 276                         } else {
 277                                 lexer_token.datatype = type_ulong;
 278                         }
 279                 } else {
 280                         lexer_token.datatype = type_uint;
 281                 }
 282         } else if(c == 'l' || c == 'L') {
 283                 next_char();
 284                 if(c == 'l' || c == 'L') {
 285                         next_char();
 286                         if(c == 'u' || c == 'U') {
 287                                 next_char();
 288                                 lexer_token.datatype = type_ulonglong;
 289                         } else {
 290                                 lexer_token.datatype = type_longlong;
 291                         }
 292                 } else if(c == 'u' || c == 'U') {
 293                         next_char();
 294                         lexer_token.datatype = type_ulong;
 295                 } else {
 296                         lexer_token.datatype = type_int;
 297                 }
 298         } else {
 299                 lexer_token.datatype = type_int;
 300         }
 301 }
 302
 303 static void parse_floating_suffix(void)
 304 {
 305         switch(c) {
 306         /* TODO: do something usefull with the suffixes... */
 307         case 'f':
 308         case 'F':
 309                 next_char();
 310                 lexer_token.datatype = type_float;
 311                 break;
 312         case 'l':
 313         case 'L':
 314                 next_char();
 315                 lexer_token.datatype = type_longdouble;
 316                 break;
 317         default:
 318                 lexer_token.datatype = type_double;
 319                 break;
 320         }
 321 }
 322
 323 static inline bool is_hex_digit(int c)
 324 {
 325         return (c >= '0' && c <= '9')
 326                         || (c >= 'a' && c <= 'f')
 327                         || (c >= 'A' && c <= 'F');
 328 }
 329
 330 static void parse_number_hex(void)
 331 {
 332         assert(c == 'x' || c == 'X');
 333         next_char();
 334
 335         while(is_hex_digit(c)) {
 336                 obstack_1grow(&symbol_obstack, c);
 337                 next_char();
 338         }
 339         obstack_1grow(&symbol_obstack, '\0');
 340         char *string = obstack_finish(&symbol_obstack);
 341
 342         if(c == '.' || c == 'p' || c == 'P') {
 343                 next_char();
 344                 panic("Hex floating point numbers not implemented yet");
 345         }
 346         if(*string == '\0') {
 347                 parse_error("invalid hex number");
 348                 lexer_token.type = T_ERROR;
 349         }
 350
 351         char *endptr;
 352         lexer_token.type       = T_INTEGER;
 353         lexer_token.v.intvalue = strtoull(string, &endptr, 16);
 354         if(*endptr != '\0') {
 355                 parse_error("hex number literal too long");
 356         }
 357
 358         obstack_free(&symbol_obstack, string);
 359 }
 360
 361 static inline bool is_octal_digit(int chr)
 362 {
 363         return '0' <= chr && chr <= '7';
 364 }
 365
 366 static void parse_number_oct(void)
 367 {
 368         while(is_octal_digit(c)) {
 369                 obstack_1grow(&symbol_obstack, c);
 370                 next_char();
 371         }
 372         obstack_1grow(&symbol_obstack, '\0');
 373         char *string = obstack_finish(&symbol_obstack);
 374
 375         char *endptr;
 376         lexer_token.type       = T_INTEGER;
 377         lexer_token.v.intvalue = strtoull(string, &endptr, 8);
 378         if(*endptr != '\0') {
 379                 parse_error("octal number literal too long");
 380         }
 381
 382         obstack_free(&symbol_obstack, string);
 383         parse_integer_suffix();
 384 }
 385
 386 static void parse_number_dec(void)
 387 {
 388         bool is_float = false;
 389         while(isdigit(c)) {
 390                 obstack_1grow(&symbol_obstack, c);
 391                 next_char();
 392         }
 393
 394         if(c == '.') {
 395                 obstack_1grow(&symbol_obstack, '.');
 396                 next_char();
 397
 398                 while(isdigit(c)) {
 399                         obstack_1grow(&symbol_obstack, c);
 400                         next_char();
 401                 }
 402                 is_float = true;
 403         }
 404         if(c == 'e' || c == 'E') {
 405                 obstack_1grow(&symbol_obstack, 'e');
 406                 next_char();
 407
 408                 if(c == '-' || c == '+') {
 409                         obstack_1grow(&symbol_obstack, c);
 410                         next_char();
 411                 }
 412
 413                 while(isdigit(c)) {
 414                         obstack_1grow(&symbol_obstack, c);
 415                         next_char();
 416                 }
 417                 is_float = true;
 418         }
 419
 420         obstack_1grow(&symbol_obstack, '\0');
 421         char *string = obstack_finish(&symbol_obstack);
 422
 423         char *endptr;
 424         if(is_float) {
 425                 lexer_token.type         = T_FLOATINGPOINT;
 426                 lexer_token.v.floatvalue = strtold(string, &endptr);
 427
 428                 if(*endptr != '\0') {
 429                         parse_error("invalid number literal");
 430                 }
 431
 432                 parse_floating_suffix();
 433         } else {
 434                 lexer_token.type       = T_INTEGER;
 435                 lexer_token.v.intvalue = strtoull(string, &endptr, 10);
 436
 437                 if(*endptr != '\0') {
 438                         parse_error("invalid number literal");
 439                 }
 440
 441                 parse_integer_suffix();
 442         }
 443         obstack_free(&symbol_obstack, string);
 444 }
 445
 446 static void parse_number(void)
 447 {
 448         if (c == '0') {
 449                 next_char();
 450                 switch (c) {
 451                         case 'X':
 452                         case 'x':
 453                                 parse_number_hex();
 454                                 break;
 455                         case '0':
 456                         case '1':
 457                         case '2':
 458                         case '3':
 459                         case '4':
 460                         case '5':
 461                         case '6':
 462                         case '7':
 463                                 parse_number_oct();
 464                                 break;
 465                         case '8':
 466                         case '9':
 467                                 next_char();
 468                                 parse_error("invalid octal number");
 469                                 lexer_token.type = T_ERROR;
 470                                 return;
 471                         case '.':
 472                         case 'e':
 473                         case 'E':
 474                         default:
 475                                 obstack_1grow(&symbol_obstack, '0');
 476                                 parse_number_dec();
 477                                 return;
 478                 }
 479         } else {
 480                 parse_number_dec();
 481         }
 482 }
 483
 484 static int parse_octal_sequence(const int first_digit)
 485 {
 486         assert(is_octal_digit(first_digit));
 487         int value = first_digit - '0';
 488         if (!is_octal_digit(c)) return value;
 489         value = 8 * value + c - '0';
 490         next_char();
 491         if (!is_octal_digit(c)) return value;
 492         value = 8 * value + c - '0';
 493         next_char();
 494         return value;
 495 }
 496
 497 static int parse_hex_sequence(void)
 498 {
 499         int value = 0;
 500         while(1) {
 501                 if (c >= '0' && c <= '9') {
 502                         value = 16 * value + c - '0';
 503                 } else if ('A' <= c && c <= 'F') {
 504                         value = 16 * value + c - 'A' + 10;
 505                 } else if ('a' <= c && c <= 'f') {
 506                         value = 16 * value + c - 'a' + 10;
 507                 } else {
 508                         break;
 509                 }
 510                 next_char();
 511         }
 512
 513         return value;
 514 }
 515
 516 static int parse_escape_sequence(void)
 517 {
 518         eat('\\');
 519
 520         int ec = c;
 521         next_char();
 522
 523         switch(ec) {
 524         case '"':  return '"';
 525         case '\'': return '\'';
 526         case '\\': return '\\';
 527         case '?': return '\?';
 528         case 'a': return '\a';
 529         case 'b': return '\b';
 530         case 'f': return '\f';
 531         case 'n': return '\n';
 532         case 'r': return '\r';
 533         case 't': return '\t';
 534         case 'v': return '\v';
 535         case 'x':
 536                 return parse_hex_sequence();
 537         case '0':
 538         case '1':
 539         case '2':
 540         case '3':
 541         case '4':
 542         case '5':
 543         case '6':
 544         case '7':
 545                 return parse_octal_sequence(ec);
 546         case EOF:
 547                 parse_error("reached end of file while parsing escape sequence");
 548                 return EOF;
 549         default:
 550                 parse_error("unknown escape sequence");
 551                 return EOF;
 552         }
 553 }
 554
 555 const char *concat_strings(const char *s1, const char *s2)
 556 {
 557         size_t  len1   = strlen(s1);
 558         size_t  len2   = strlen(s2);
 559
 560         char   *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 561         memcpy(concat, s1, len1);
 562         memcpy(concat + len1, s2, len2 + 1);
 563
 564         const char *result = strset_insert(&stringset, concat);
 565         if(result != concat) {
 566                 obstack_free(&symbol_obstack, concat);
 567         }
 568
 569         return result;
 570 }
 571
 572 static void parse_string_literal(void)
 573 {
 574         unsigned    start_linenr = lexer_token.source_position.linenr;
 575         char       *string;
 576         const char *result;
 577
 578         assert(c == '"');
 579         next_char();
 580
 581         int tc;
 582         while(1) {
 583                 switch(c) {
 584                 case '\\':
 585                         tc = parse_escape_sequence();
 586                         obstack_1grow(&symbol_obstack, tc);
 587                         break;
 588
 589                 case EOF:
 590                         error_prefix_at(lexer_token.source_position.input_name,
 591                                         start_linenr);
 592                         fprintf(stderr, "string has no end\n");
 593                         lexer_token.type = T_ERROR;
 594                         return;
 595
 596                 case '"':
 597                         next_char();
 598                         goto end_of_string;
 599
 600                 default:
 601                         obstack_1grow(&symbol_obstack, c);
 602                         next_char();
 603                         break;
 604                 }
 605         }
 606
 607 end_of_string:
 608
 609         /* TODO: concatenate multiple strings separated by whitespace... */
 610
 611         /* add finishing 0 to the string */
 612         obstack_1grow(&symbol_obstack, '\0');
 613         string = obstack_finish(&symbol_obstack);
 614
 615         /* check if there is already a copy of the string */
 616         result = strset_insert(&stringset, string);
 617         if(result != string) {
 618                 obstack_free(&symbol_obstack, string);
 619         }
 620
 621         lexer_token.type     = T_STRING_LITERAL;
 622         lexer_token.v.string = result;
 623 }
 624
 625 static void parse_character_constant(void)
 626 {
 627         eat('\'');
 628
 629         int found_char = 0;
 630         while(1) {
 631                 switch(c) {
 632                 case '\\':
 633                         found_char = parse_escape_sequence();
 634                         break;
 635
 636                 MATCH_NEWLINE(
 637                         parse_error("newline while parsing character constant");
 638                         break;
 639                 )
 640
 641                 case '\'':
 642                         next_char();
 643                         goto end_of_char_constant;
 644
 645                 case EOF:
 646                         parse_error("EOF while parsing character constant");
 647                         lexer_token.type = T_ERROR;
 648                         return;
 649
 650                 default:
 651                         if(found_char != 0) {
 652                                 parse_error("more than 1 characters in character "
 653                                             "constant");
 654                                 goto end_of_char_constant;
 655                         } else {
 656                                 found_char = c;
 657                                 next_char();
 658                         }
 659                         break;
 660                 }
 661         }
 662
 663 end_of_char_constant:
 664         lexer_token.type       = T_INTEGER;
 665         lexer_token.v.intvalue = found_char;
 666 }
 667
 668 static void skip_multiline_comment(void)
 669 {
 670         unsigned start_linenr = lexer_token.source_position.linenr;
 671
 672         while(1) {
 673                 switch(c) {
 674                 case '*':
 675                         next_char();
 676                         if(c == '/') {
 677                                 next_char();
 678                                 return;
 679                         }
 680                         break;
 681
 682                 MATCH_NEWLINE(break;)
 683
 684                 case EOF:
 685                         error_prefix_at(lexer_token.source_position.input_name,
 686                                         start_linenr);
 687                         fprintf(stderr, "at end of file while looking for comment end\n");
 688                         return;
 689
 690                 default:
 691                         next_char();
 692                         break;
 693                 }
 694         }
 695 }
 696
 697 static void skip_line_comment(void)
 698 {
 699         while(1) {
 700                 switch(c) {
 701                 case EOF:
 702                         return;
 703
 704                 case '\n':
 705                 case '\r':
 706                         return;
 707
 708                 default:
 709                         next_char();
 710                         break;
 711                 }
 712         }
 713 }
 714
 715 static token_t pp_token;
 716
 717 static inline void next_pp_token(void)
 718 {
 719         lexer_next_preprocessing_token();
 720         pp_token = lexer_token;
 721 }
 722
 723 static void eat_until_newline(void)
 724 {
 725         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
 726                 next_pp_token();
 727         }
 728 }
 729
 730 static void error_directive(void)
 731 {
 732         error_prefix();
 733         fprintf(stderr, "#error directive: \n");
 734
 735         /* parse pp-tokens until new-line */
 736 }
 737
 738 static void define_directive(void)
 739 {
 740         lexer_next_preprocessing_token();
 741         if(lexer_token.type != T_IDENTIFIER) {
 742                 parse_error("expected identifier after #define\n");
 743                 eat_until_newline();
 744         }
 745 }
 746
 747 static void ifdef_directive(int is_ifndef)
 748 {
 749         (void) is_ifndef;
 750         lexer_next_preprocessing_token();
 751         //expect_identifier();
 752         //extect_newline();
 753 }
 754
 755 static void endif_directive(void)
 756 {
 757         //expect_newline();
 758 }
 759
 760 static void parse_line_directive(void)
 761 {
 762         if(pp_token.type != T_INTEGER) {
 763                 parse_error("expected integer");
 764         } else {
 765                 lexer_token.source_position.linenr = pp_token.v.intvalue - 1;
 766                 next_pp_token();
 767         }
 768         if(pp_token.type == T_STRING_LITERAL) {
 769                 lexer_token.source_position.input_name = pp_token.v.string;
 770                 next_pp_token();
 771         }
 772
 773         eat_until_newline();
 774 }
 775
 776 static void parse_preprocessor_identifier(void)
 777 {
 778         assert(pp_token.type == T_IDENTIFIER);
 779         symbol_t *symbol = pp_token.v.symbol;
 780
 781         switch(symbol->pp_ID) {
 782         case TP_include:
 783                 printf("include - enable header name parsing!\n");
 784                 break;
 785         case TP_define:
 786                 define_directive();
 787                 break;
 788         case TP_ifdef:
 789                 ifdef_directive(0);
 790                 break;
 791         case TP_ifndef:
 792                 ifdef_directive(1);
 793                 break;
 794         case TP_endif:
 795                 endif_directive();
 796                 break;
 797         case TP_line:
 798                 next_pp_token();
 799                 parse_line_directive();
 800                 break;
 801         case TP_if:
 802         case TP_else:
 803         case TP_elif:
 804         case TP_undef:
 805         case TP_error:
 806                 error_directive();
 807                 break;
 808         case TP_pragma:
 809                 break;
 810         }
 811 }
 812
 813 static void parse_preprocessor_directive(void)
 814 {
 815         next_pp_token();
 816
 817         switch(pp_token.type) {
 818         case T_IDENTIFIER:
 819                 parse_preprocessor_identifier();
 820                 break;
 821         case T_INTEGER:
 822                 parse_line_directive();
 823                 break;
 824         default:
 825                 parse_error("invalid preprocessor directive");
 826                 eat_until_newline();
 827                 break;
 828         }
 829 }
 830
 831 #define MAYBE_PROLOG                                       \
 832                         next_char();                                   \
 833                         while(1) {                                     \
 834                                 switch(c) {
 835
 836 #define MAYBE(ch, set_type)                                \
 837                                 case ch:                                   \
 838                                         next_char();                           \
 839                                         lexer_token.type = set_type;           \
 840                                         return;
 841
 842 #define ELSE_CODE(code)                                    \
 843                                 default:                                   \
 844                                         code;                                  \
 845                                 }                                          \
 846                         } /* end of while(1) */                        \
 847                         break;
 848
 849 #define ELSE(set_type)                                     \
 850                 ELSE_CODE(                                         \
 851                         lexer_token.type = set_type;                   \
 852                         return;                                        \
 853                 )
 854
 855 void lexer_next_preprocessing_token(void)
 856 {
 857         while(1) {
 858                 switch(c) {
 859                 case ' ':
 860                 case '\t':
 861                         next_char();
 862                         break;
 863
 864                 MATCH_NEWLINE(
 865                         lexer_token.type = '\n';
 866                         return;
 867                 )
 868
 869                 SYMBOL_CHARS
 870                         parse_symbol();
 871                         /* might be a wide string ( L"string" ) */
 872                         if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 873                            lexer_token.v.symbol == symbol_L)) {
 874                                 parse_string_literal();
 875                                 return;
 876                         }
 877                         return;
 878
 879                 DIGITS
 880                         parse_number();
 881                         return;
 882
 883                 case '"':
 884                         parse_string_literal();
 885                         return;
 886
 887                 case '\'':
 888                         parse_character_constant();
 889                         return;
 890
 891                 case '.':
 892                         MAYBE_PROLOG
 893                                 case '.':
 894                                         MAYBE_PROLOG
 895                                         MAYBE('.', T_DOTDOTDOT)
 896                                         ELSE_CODE(
 897                                                 put_back(c);
 898                                                 c = '.';
 899                                                 lexer_token.type = '.';
 900                                                 return;
 901                                         )
 902                         ELSE('.')
 903                 case '&':
 904                         MAYBE_PROLOG
 905                         MAYBE('&', T_ANDAND)
 906                         MAYBE('=', T_ANDEQUAL)
 907                         ELSE('&')
 908                 case '*':
 909                         MAYBE_PROLOG
 910                         MAYBE('=', T_ASTERISKEQUAL)
 911                         ELSE('*')
 912                 case '+':
 913                         MAYBE_PROLOG
 914                         MAYBE('+', T_PLUSPLUS)
 915                         MAYBE('=', T_PLUSEQUAL)
 916                         ELSE('+')
 917                 case '-':
 918                         MAYBE_PROLOG
 919                         MAYBE('>', T_MINUSGREATER)
 920                         MAYBE('-', T_MINUSMINUS)
 921                         MAYBE('=', T_MINUSEQUAL)
 922                         ELSE('-')
 923                 case '!':
 924                         MAYBE_PROLOG
 925                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
 926                         ELSE('!')
 927                 case '/':
 928                         MAYBE_PROLOG
 929                         MAYBE('=', T_SLASHEQUAL)
 930                                 case '*':
 931                                         next_char();
 932                                         skip_multiline_comment();
 933                                         lexer_next_preprocessing_token();
 934                                         return;
 935                                 case '/':
 936                                         next_char();
 937                                         skip_line_comment();
 938                                         lexer_next_preprocessing_token();
 939                                         return;
 940                         ELSE('/')
 941                 case '%':
 942                         MAYBE_PROLOG
 943                         MAYBE('>', T_PERCENTGREATER)
 944                         MAYBE('=', T_PERCENTEQUAL)
 945                                 case ':':
 946                                         MAYBE_PROLOG
 947                                                 case '%':
 948                                                         MAYBE_PROLOG
 949                                                         MAYBE(':', T_PERCENTCOLONPERCENTCOLON)
 950                                                         ELSE_CODE(
 951                                                                 put_back(c);
 952                                                                 c = '%';
 953                                                                 lexer_token.type = T_PERCENTCOLON;
 954                                                                 return;
 955                                                         )
 956                                         ELSE(T_PERCENTCOLON)
 957                         ELSE('%')
 958                 case '<':
 959                         MAYBE_PROLOG
 960                         MAYBE(':', T_LESSCOLON)
 961                         MAYBE('%', T_LESSPERCENT)
 962                         MAYBE('=', T_LESSEQUAL)
 963                                 case '<':
 964                                         MAYBE_PROLOG
 965                                         MAYBE('=', T_LESSLESSEQUAL)
 966                                         ELSE(T_LESSLESS)
 967                         ELSE('<')
 968                 case '>':
 969                         MAYBE_PROLOG
 970                         MAYBE('=', T_GREATEREQUAL)
 971                                 case '>':
 972                                         MAYBE_PROLOG
 973                                         MAYBE('=', T_GREATERGREATEREQUAL)
 974                                         ELSE(T_GREATERGREATER)
 975                         ELSE('>')
 976                 case '^':
 977                         MAYBE_PROLOG
 978                         MAYBE('=', T_CARETEQUAL)
 979                         ELSE('^')
 980                 case '|':
 981                         MAYBE_PROLOG
 982                         MAYBE('=', T_PIPEEQUAL)
 983                         MAYBE('|', T_PIPEPIPE)
 984                         ELSE('|')
 985                 case ':':
 986                         MAYBE_PROLOG
 987                         MAYBE('>', T_COLONGREATER)
 988                         ELSE(':')
 989                 case '=':
 990                         MAYBE_PROLOG
 991                         MAYBE('=', T_EQUALEQUAL)
 992                         ELSE('=')
 993                 case '#':
 994                         MAYBE_PROLOG
 995                         MAYBE('#', T_HASHHASH)
 996                         ELSE('#')
 997
 998                 case '?':
 999                 case '[':
1000                 case ']':
1001                 case '(':
1002                 case ')':
1003                 case '{':
1004                 case '}':
1005                 case '~':
1006                 case ';':
1007                 case ',':
1008                 case '\\':
1009                         lexer_token.type = c;
1010                         next_char();
1011                         return;
1012
1013                 case EOF:
1014                         lexer_token.type = T_EOF;
1015                         return;
1016
1017                 default:
1018                         next_char();
1019                         error_prefix();
1020                         fprintf(stderr, "unknown character '%c' found\n", c);
1021                         lexer_token.type = T_ERROR;
1022                         return;
1023                 }
1024         }
1025 }
1026
1027 void lexer_next_token(void)
1028 {
1029         lexer_next_preprocessing_token();
1030         if(lexer_token.type != '\n')
1031                 return;
1032
1033 newline_found:
1034         do {
1035                 lexer_next_preprocessing_token();
1036         } while(lexer_token.type == '\n');
1037
1038         if(lexer_token.type == '#') {
1039                 parse_preprocessor_directive();
1040                 goto newline_found;
1041         }
1042 }
1043
1044 void init_lexer(void)
1045 {
1046         strset_init(&stringset);
1047
1048         type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
1049         type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
1050         type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
1051         type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
1052         type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
1053                                           TYPE_QUALIFIER_CONST);
1054         type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
1055                                           TYPE_QUALIFIER_CONST);
1056
1057         type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
1058         type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
1059                                            TYPE_QUALIFIER_CONST);
1060         type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE,
1061                                            TYPE_QUALIFIER_CONST);
1062 }
1063
1064 void lexer_open_stream(FILE *stream, const char *input_name)
1065 {
1066         input                                  = stream;
1067         lexer_token.source_position.linenr     = 0;
1068         lexer_token.source_position.input_name = input_name;
1069
1070         symbol_L = symbol_table_insert("L");
1071
1072         /* place a virtual \n at the beginning so the lexer knows that we're
1073          * at the beginning of a line */
1074         c = '\n';
1075 }
1076
1077 void exit_lexer(void)
1078 {
1079         strset_destroy(&stringset);
1080 }
1081
1082 static __attribute__((unused))
1083 void dbg_pos(const source_position_t source_position)
1084 {
1085         fprintf(stdout, "%s:%d\n", source_position.input_name,
1086                 source_position.linenr);
1087         fflush(stdout);
1088 }