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
  10 #include <assert.h>
  11 #include <errno.h>
  12 #include <string.h>
  13 #include <ctype.h>
  14
  15 //#define DEBUG_CHARS
  16 #define MAX_PUTBACK 3
  17
  18 static int         c;
  19 token_t            lexer_token;
  20 symbol_t          *symbol_L;
  21 static FILE       *input;
  22 static char        buf[1024 + MAX_PUTBACK];
  23 static const char *bufend;
  24 static const char *bufpos;
  25 static strset_t    stringset;
  26
  27 static void error_prefix_at(const char *input_name, unsigned linenr)
  28 {
  29         fprintf(stderr, "%s:%u: Error: ", input_name, linenr);
  30 }
  31
  32 static void error_prefix(void)
  33 {
  34         error_prefix_at(lexer_token.source_position.input_name,
  35                         lexer_token.source_position.linenr);
  36 }
  37
  38 static void parse_error(const char *msg)
  39 {
  40         error_prefix();
  41         fprintf(stderr, "%s\n", msg);
  42 }
  43
  44 static inline void next_real_char(void)
  45 {
  46         bufpos++;
  47         if(bufpos >= bufend) {
  48                 size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK,
  49                                  input);
  50                 if(s == 0) {
  51                         c = EOF;
  52                         return;
  53                 }
  54                 bufpos = buf + MAX_PUTBACK;
  55                 bufend = buf + MAX_PUTBACK + s;
  56         }
  57         c = *(bufpos);
  58 }
  59
  60 static inline void put_back(int pc)
  61 {
  62         assert(bufpos >= buf);
  63         assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc);
  64
  65         char *p = buf + (bufpos - buf);
  66         *p = pc;
  67
  68         /* going backwards in the buffer is legal as long as it's not more often
  69          * than MAX_PUTBACK */
  70         bufpos--;
  71
  72 #ifdef DEBUG_CHARS
  73         printf("putback '%c'\n", pc);
  74 #endif
  75 }
  76
  77 static inline void next_char(void);
  78
  79 #define MATCH_NEWLINE(code)                   \
  80         case '\r':                                \
  81                 next_char();                          \
  82                 if(c == '\n') {                       \
  83                         next_char();                      \
  84                 }                                     \
  85                 lexer_token.source_position.linenr++; \
  86                 code;                                 \
  87         case '\n':                                \
  88                 next_char();                          \
  89                 lexer_token.source_position.linenr++; \
  90                 code;
  91
  92 #define eat(c_type)  do { assert(c == c_type); next_char(); } while(0)
  93
  94 static void maybe_concat_lines(void)
  95 {
  96         eat('\\');
  97
  98         switch(c) {
  99         MATCH_NEWLINE(return;)
 100
 101         default:
 102                 break;
 103         }
 104
 105         put_back(c);
 106         c = '\\';
 107 }
 108
 109 static inline void next_char(void)
 110 {
 111         next_real_char();
 112
 113 #if 0
 114         /* filter trigraphs */
 115         if(UNLIKELY(c == '\\')) {
 116                 maybe_concat_lines();
 117                 goto end_of_next_char;
 118         }
 119
 120         if(LIKELY(c != '?'))
 121                 goto end_of_next_char;
 122
 123         next_real_char();
 124         if(LIKELY(c != '?')) {
 125                 put_back(c);
 126                 c = '?';
 127                 goto end_of_next_char;
 128         }
 129
 130         next_real_char();
 131         switch(c) {
 132         case '=': c = '#'; break;
 133         case '(': c = '['; break;
 134         case '/': c = '\\'; maybe_concat_lines(); break;
 135         case ')': c = ']'; break;
 136         case '\'': c = '^'; break;
 137         case '<': c = '{'; break;
 138         case '!': c = '|'; break;
 139         case '>': c = '}'; break;
 140         case '-': c = '~'; break;
 141         default:
 142                 put_back('?');
 143                 put_back(c);
 144                 c = '?';
 145                 break;
 146         }
 147
 148 end_of_next_char:
 149 #endif
 150         (void) maybe_concat_lines;
 151 #ifdef DEBUG_CHARS
 152         printf("nchar '%c'\n", c);
 153 #else
 154         ;
 155 #endif
 156 }
 157
 158 #define SYMBOL_CHARS  \
 159         case 'a':         \
 160         case 'b':         \
 161         case 'c':         \
 162         case 'd':         \
 163         case 'e':         \
 164         case 'f':         \
 165         case 'g':         \
 166         case 'h':         \
 167         case 'i':         \
 168         case 'j':         \
 169         case 'k':         \
 170         case 'l':         \
 171         case 'm':         \
 172         case 'n':         \
 173         case 'o':         \
 174         case 'p':         \
 175         case 'q':         \
 176         case 'r':         \
 177         case 's':         \
 178         case 't':         \
 179         case 'u':         \
 180         case 'v':         \
 181         case 'w':         \
 182         case 'x':         \
 183         case 'y':         \
 184         case 'z':         \
 185         case 'A':         \
 186         case 'B':         \
 187         case 'C':         \
 188         case 'D':         \
 189         case 'E':         \
 190         case 'F':         \
 191         case 'G':         \
 192         case 'H':         \
 193         case 'I':         \
 194         case 'J':         \
 195         case 'K':         \
 196         case 'L':         \
 197         case 'M':         \
 198         case 'N':         \
 199         case 'O':         \
 200         case 'P':         \
 201         case 'Q':         \
 202         case 'R':         \
 203         case 'S':         \
 204         case 'T':         \
 205         case 'U':         \
 206         case 'V':         \
 207         case 'W':         \
 208         case 'X':         \
 209         case 'Y':         \
 210         case 'Z':         \
 211         case '_':
 212
 213 #define DIGITS        \
 214         case '0':         \
 215         case '1':         \
 216         case '2':         \
 217         case '3':         \
 218         case '4':         \
 219         case '5':         \
 220         case '6':         \
 221         case '7':         \
 222         case '8':         \
 223         case '9':
 224
 225 static void parse_symbol(void)
 226 {
 227         symbol_t *symbol;
 228         char     *string;
 229
 230         obstack_1grow(&symbol_obstack, c);
 231         next_char();
 232
 233         while(1) {
 234                 switch(c) {
 235                 DIGITS
 236                 SYMBOL_CHARS
 237                         obstack_1grow(&symbol_obstack, c);
 238                         next_char();
 239                         break;
 240
 241                 default:
 242                         goto end_symbol;
 243                 }
 244         }
 245
 246 end_symbol:
 247         obstack_1grow(&symbol_obstack, '\0');
 248
 249         string = obstack_finish(&symbol_obstack);
 250         symbol = symbol_table_insert(string);
 251
 252         lexer_token.type     = symbol->ID;
 253         lexer_token.v.symbol = symbol;
 254
 255         if(symbol->string != string) {
 256                 obstack_free(&symbol_obstack, string);
 257         }
 258 }
 259
 260 static void parse_integer_suffix(void)
 261 {
 262         if(c == 'U' || c == 'U') {
 263                 /* TODO do something with the suffixes... */
 264                 next_char();
 265                 if(c == 'L' || c == 'l') {
 266                         next_char();
 267                         if(c == 'L' || c == 'l') {
 268                                 next_char();
 269                         }
 270                 }
 271         } else if(c == 'l' || c == 'L') {
 272                 next_char();
 273                 if(c == 'l' || c == 'L') {
 274                         next_char();
 275                         if(c == 'u' || c == 'U') {
 276                                 next_char();
 277                         }
 278                 } else if(c == 'u' || c == 'U') {
 279                         next_char();
 280                 }
 281         }
 282 }
 283
 284 static void parse_floating_suffix(void)
 285 {
 286         switch(c) {
 287         /* TODO: do something usefull with the suffixes... */
 288         case 'f':
 289         case 'F':
 290         case 'l':
 291         case 'L':
 292                 next_char();
 293                 break;
 294         default:
 295                 break;
 296         }
 297 }
 298
 299 static void parse_number_hex(void)
 300 {
 301         assert(c == 'x' || c == 'X');
 302         next_char();
 303
 304         if (!isdigit(c) &&
 305                 !('A' <= c && c <= 'F') &&
 306                 !('a' <= c && c <= 'f')) {
 307                 parse_error("premature end of hex number literal");
 308                 lexer_token.type = T_ERROR;
 309                 return;
 310         }
 311
 312         int value = 0;
 313         while(1) {
 314                 if (isdigit(c)) {
 315                         value = 16 * value + c - '0';
 316                 } else if ('A' <= c && c <= 'F') {
 317                         value = 16 * value + c - 'A' + 10;
 318                 } else if ('a' <= c && c <= 'f') {
 319                         value = 16 * value + c - 'a' + 10;
 320                 } else {
 321                         parse_integer_suffix();
 322
 323                         lexer_token.type       = T_INTEGER;
 324                         lexer_token.v.intvalue = value;
 325                         return;
 326                 }
 327                 next_char();
 328         }
 329
 330         if(c == '.' || c == 'p' || c == 'P') {
 331                 next_char();
 332                 panic("Hex floating point numbers not implemented yet");
 333         }
 334 }
 335
 336 static void parse_number_oct(void)
 337 {
 338         int value = 0;
 339         while(c >= '0' && c <= '7') {
 340                 value = 8 * value + c - '0';
 341                 next_char();
 342         }
 343         if (c == '8' || c == '9') {
 344                 parse_error("invalid octal number");
 345                 lexer_token.type = T_ERROR;
 346                 return;
 347         }
 348
 349         lexer_token.type       = T_INTEGER;
 350         lexer_token.v.intvalue = value;
 351
 352         parse_integer_suffix();
 353 }
 354
 355 static void parse_floatingpoint_exponent(long double value)
 356 {
 357         unsigned int expo = 0;
 358         long double  factor = 10.;
 359
 360         if(c == '-') {
 361                 next_char();
 362                 factor = 0.1;
 363         } else if(c == '+') {
 364                 next_char();
 365         }
 366
 367         while(c >= '0' && c <= '9') {
 368                 expo = 10 * expo + (c - '0');
 369                 next_char();
 370         }
 371
 372         while(1) {
 373                 if(expo & 1)
 374                         value *= factor;
 375                 expo >>= 1;
 376                 if(expo == 0)
 377                         break;
 378                 factor *= factor;
 379         }
 380
 381         lexer_token.type         = T_FLOATINGPOINT;
 382         lexer_token.v.floatvalue = value;
 383
 384         parse_floating_suffix();
 385 }
 386
 387 static void parse_floatingpoint_fract(int integer_part)
 388 {
 389         long double value  = integer_part;
 390         long double factor = 1.;
 391
 392         while(c >= '0' && c <= '9') {
 393                 factor *= 0.1;
 394                 value  += (c - '0') * factor;
 395                 next_char();
 396         }
 397
 398         if(c == 'e' || c == 'E') {
 399                 next_char();
 400                 parse_floatingpoint_exponent(value);
 401                 return;
 402         }
 403
 404         lexer_token.type         = T_FLOATINGPOINT;
 405         lexer_token.v.floatvalue = value;
 406
 407         parse_floating_suffix();
 408 }
 409
 410 static void parse_number_dec(void)
 411 {
 412         int value = 0;
 413
 414         while(isdigit(c)) {
 415                 value = 10 * value + c - '0';
 416                 next_char();
 417         }
 418
 419         if(c == '.') {
 420                 next_char();
 421                 parse_floatingpoint_fract(value);
 422                 return;
 423         }
 424         if(c == 'e' || c == 'E') {
 425                 next_char();
 426                 parse_floatingpoint_exponent(value);
 427                 return;
 428         }
 429         parse_integer_suffix();
 430
 431         lexer_token.type       = T_INTEGER;
 432         lexer_token.v.intvalue = value;
 433 }
 434
 435 static void parse_number(void)
 436 {
 437         if (c == '0') {
 438                 next_char();
 439                 switch (c) {
 440                         case 'X':
 441                         case 'x':
 442                                 parse_number_hex();
 443                                 break;
 444                         case '0':
 445                         case '1':
 446                         case '2':
 447                         case '3':
 448                         case '4':
 449                         case '5':
 450                         case '6':
 451                         case '7':
 452                                 parse_number_oct();
 453                                 break;
 454                         case '.':
 455                                 next_char();
 456                                 parse_floatingpoint_fract(0);
 457                                 break;
 458                         case 'e':
 459                         case 'E':
 460                                 parse_floatingpoint_exponent(0);
 461                                 break;
 462                         case '8':
 463                         case '9':
 464                                 next_char();
 465                                 parse_error("invalid octal number");
 466                                 lexer_token.type = T_ERROR;
 467                                 return;
 468                         default:
 469                                 put_back(c);
 470                                 c = '0';
 471                                 parse_number_dec();
 472                                 return;
 473                 }
 474         } else {
 475                 parse_number_dec();
 476         }
 477 }
 478
 479 static inline int is_octal_digit(int chr)
 480 {
 481         return '0' <= chr && chr <= '7';
 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
1049 void lexer_open_stream(FILE *stream, const char *input_name)
1050 {
1051         input                                  = stream;
1052         lexer_token.source_position.linenr     = 0;
1053         lexer_token.source_position.input_name = input_name;
1054
1055         symbol_L = symbol_table_insert("L");
1056
1057         /* place a virtual \n at the beginning so the lexer knows that we're
1058          * at the beginning of a line */
1059         c = '\n';
1060 }
1061
1062 void exit_lexer(void)
1063 {
1064         strset_destroy(&stringset);
1065 }
1066
1067 static __attribute__((unused))
1068 void dbg_pos(const source_position_t source_position)
1069 {
1070         fprintf(stdout, "%s:%d\n", source_position.input_name,
1071                 source_position.linenr);
1072         fflush(stdout);
1073 }