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 #ifdef _WIN32
  21 /* No strtold on windows and no replacement yet */
  22 #define strtold(s, e) strtod(s, e)
  23 #endif
  24
  25 static int         c;
  26 token_t            lexer_token;
  27 symbol_t          *symbol_L;
  28 static FILE       *input;
  29 static char        buf[1024 + MAX_PUTBACK];
  30 static const char *bufend;
  31 static const char *bufpos;
  32 static strset_t    stringset;
  33
  34 static type_t     *type_int        = NULL;
  35 static type_t     *type_uint       = NULL;
  36 static type_t     *type_long       = NULL;
  37 static type_t     *type_ulong      = NULL;
  38 static type_t     *type_longlong   = NULL;
  39 static type_t     *type_ulonglong  = NULL;
  40 static type_t     *type_float      = NULL;
  41 static type_t     *type_double     = NULL;
  42 static type_t     *type_longdouble = NULL;
  43
  44 static void error_prefix_at(const char *input_name, unsigned linenr)
  45 {
  46         fprintf(stderr, "%s:%u: Error: ", input_name, linenr);
  47 }
  48
  49 static void error_prefix(void)
  50 {
  51         error_prefix_at(lexer_token.source_position.input_name,
  52                         lexer_token.source_position.linenr);
  53 }
  54
  55 static void parse_error(const char *msg)
  56 {
  57         error_prefix();
  58         fprintf(stderr, "%s\n", msg);
  59 }
  60
  61 static inline void next_real_char(void)
  62 {
  63         bufpos++;
  64         if(bufpos >= bufend) {
  65                 size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK,
  66                                  input);
  67                 if(s == 0) {
  68                         c = EOF;
  69                         return;
  70                 }
  71                 bufpos = buf + MAX_PUTBACK;
  72                 bufend = buf + MAX_PUTBACK + s;
  73         }
  74         c = *(bufpos);
  75 }
  76
  77 static inline void put_back(int pc)
  78 {
  79         assert(bufpos >= buf);
  80         //assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc);
  81
  82         char *p = buf + (bufpos - buf);
  83         *p = pc;
  84
  85         /* going backwards in the buffer is legal as long as it's not more often
  86          * than MAX_PUTBACK */
  87         bufpos--;
  88
  89 #ifdef DEBUG_CHARS
  90         printf("putback '%c'\n", pc);
  91 #endif
  92 }
  93
  94 static inline void next_char(void);
  95
  96 #define MATCH_NEWLINE(code)                   \
  97         case '\r':                                \
  98                 next_char();                          \
  99                 if(c == '\n') {                       \
 100                         next_char();                      \
 101                 }                                     \
 102                 lexer_token.source_position.linenr++; \
 103                 code;                                 \
 104         case '\n':                                \
 105                 next_char();                          \
 106                 lexer_token.source_position.linenr++; \
 107                 code;
 108
 109 #define eat(c_type)  do { assert(c == c_type); next_char(); } while(0)
 110
 111 static void maybe_concat_lines(void)
 112 {
 113         eat('\\');
 114
 115         switch(c) {
 116         MATCH_NEWLINE(return;)
 117
 118         default:
 119                 break;
 120         }
 121
 122         put_back(c);
 123         c = '\\';
 124 }
 125
 126 static inline void next_char(void)
 127 {
 128         next_real_char();
 129
 130         /* filter trigraphs */
 131         if(UNLIKELY(c == '\\')) {
 132                 maybe_concat_lines();
 133                 goto end_of_next_char;
 134         }
 135
 136         if(LIKELY(c != '?'))
 137                 goto end_of_next_char;
 138
 139         next_real_char();
 140         if(LIKELY(c != '?')) {
 141                 put_back(c);
 142                 c = '?';
 143                 goto end_of_next_char;
 144         }
 145
 146         next_real_char();
 147         switch(c) {
 148         case '=': c = '#'; break;
 149         case '(': c = '['; break;
 150         case '/': c = '\\'; maybe_concat_lines(); break;
 151         case ')': c = ']'; break;
 152         case '\'': c = '^'; break;
 153         case '<': c = '{'; break;
 154         case '!': c = '|'; break;
 155         case '>': c = '}'; break;
 156         case '-': c = '~'; break;
 157         default:
 158                 put_back('?');
 159                 put_back(c);
 160                 c = '?';
 161                 break;
 162         }
 163
 164 end_of_next_char:;
 165 #ifdef DEBUG_CHARS
 166         printf("nchar '%c'\n", c);
 167 #endif
 168 }
 169
 170 #define SYMBOL_CHARS  \
 171         case 'a':         \
 172         case 'b':         \
 173         case 'c':         \
 174         case 'd':         \
 175         case 'e':         \
 176         case 'f':         \
 177         case 'g':         \
 178         case 'h':         \
 179         case 'i':         \
 180         case 'j':         \
 181         case 'k':         \
 182         case 'l':         \
 183         case 'm':         \
 184         case 'n':         \
 185         case 'o':         \
 186         case 'p':         \
 187         case 'q':         \
 188         case 'r':         \
 189         case 's':         \
 190         case 't':         \
 191         case 'u':         \
 192         case 'v':         \
 193         case 'w':         \
 194         case 'x':         \
 195         case 'y':         \
 196         case 'z':         \
 197         case 'A':         \
 198         case 'B':         \
 199         case 'C':         \
 200         case 'D':         \
 201         case 'E':         \
 202         case 'F':         \
 203         case 'G':         \
 204         case 'H':         \
 205         case 'I':         \
 206         case 'J':         \
 207         case 'K':         \
 208         case 'L':         \
 209         case 'M':         \
 210         case 'N':         \
 211         case 'O':         \
 212         case 'P':         \
 213         case 'Q':         \
 214         case 'R':         \
 215         case 'S':         \
 216         case 'T':         \
 217         case 'U':         \
 218         case 'V':         \
 219         case 'W':         \
 220         case 'X':         \
 221         case 'Y':         \
 222         case 'Z':         \
 223         case '_':
 224
 225 #define DIGITS        \
 226         case '0':         \
 227         case '1':         \
 228         case '2':         \
 229         case '3':         \
 230         case '4':         \
 231         case '5':         \
 232         case '6':         \
 233         case '7':         \
 234         case '8':         \
 235         case '9':
 236
 237 static void parse_symbol(void)
 238 {
 239         symbol_t *symbol;
 240         char     *string;
 241
 242         obstack_1grow(&symbol_obstack, c);
 243         next_char();
 244
 245         while(1) {
 246                 switch(c) {
 247                 DIGITS
 248                 SYMBOL_CHARS
 249                         obstack_1grow(&symbol_obstack, c);
 250                         next_char();
 251                         break;
 252
 253                 default:
 254                         goto end_symbol;
 255                 }
 256         }
 257
 258 end_symbol:
 259         obstack_1grow(&symbol_obstack, '\0');
 260
 261         string = obstack_finish(&symbol_obstack);
 262         symbol = symbol_table_insert(string);
 263
 264         lexer_token.type     = symbol->ID;
 265         lexer_token.v.symbol = symbol;
 266
 267         if(symbol->string != string) {
 268                 obstack_free(&symbol_obstack, string);
 269         }
 270 }
 271
 272 static void parse_integer_suffix(void)
 273 {
 274         if(c == 'U' || c == 'u') {
 275                 next_char();
 276                 if(c == 'L' || c == 'l') {
 277                         next_char();
 278                         if(c == 'L' || c == 'l') {
 279                                 next_char();
 280                                 lexer_token.datatype = type_ulonglong;
 281                         } else {
 282                                 lexer_token.datatype = type_ulong;
 283                         }
 284                 } else {
 285                         lexer_token.datatype = type_uint;
 286                 }
 287         } else if(c == 'l' || c == 'L') {
 288                 next_char();
 289                 if(c == 'l' || c == 'L') {
 290                         next_char();
 291                         if(c == 'u' || c == 'U') {
 292                                 next_char();
 293                                 lexer_token.datatype = type_ulonglong;
 294                         } else {
 295                                 lexer_token.datatype = type_longlong;
 296                         }
 297                 } else if(c == 'u' || c == 'U') {
 298                         next_char();
 299                         lexer_token.datatype = type_ulong;
 300                 } else {
 301                         lexer_token.datatype = type_int;
 302                 }
 303         } else {
 304                 lexer_token.datatype = type_int;
 305         }
 306 }
 307
 308 static void parse_floating_suffix(void)
 309 {
 310         switch(c) {
 311         /* TODO: do something usefull with the suffixes... */
 312         case 'f':
 313         case 'F':
 314                 next_char();
 315                 lexer_token.datatype = type_float;
 316                 break;
 317         case 'l':
 318         case 'L':
 319                 next_char();
 320                 lexer_token.datatype = type_longdouble;
 321                 break;
 322         default:
 323                 lexer_token.datatype = type_double;
 324                 break;
 325         }
 326 }
 327
 328 /**
 329  * A replacement for strtoull. Only those parts needed for
 330  * our parser are implemented.
 331  */
 332 static unsigned long long parse_int_string(const char *s, const char **endptr, int base) {
 333         unsigned long long v = 0;
 334
 335         switch (base) {
 336         case 16:
 337                 for (;; ++s) {
 338                         /* check for overrun */
 339                         if (v <= 0x1000000000000000ULL)
 340                                 break;
 341                         switch (tolower(*s)) {
 342                         case '0': v <<= 4; break;
 343                         case '1': v <<= 4; v |= 0x1; break;
 344                         case '2': v <<= 4; v |= 0x2; break;
 345                         case '3': v <<= 4; v |= 0x3; break;
 346                         case '4': v <<= 4; v |= 0x4; break;
 347                         case '5': v <<= 4; v |= 0x5; break;
 348                         case '6': v <<= 4; v |= 0x6; break;
 349                         case '7': v <<= 4; v |= 0x7; break;
 350                         case '8': v <<= 4; v |= 0x8; break;
 351                         case '9': v <<= 4; v |= 0x9; break;
 352                         case 'a': v <<= 4; v |= 0xa; break;
 353                         case 'b': v <<= 4; v |= 0xb; break;
 354                         case 'c': v <<= 4; v |= 0xc; break;
 355                         case 'd': v <<= 4; v |= 0xd; break;
 356                         case 'e': v <<= 4; v |= 0xe; break;
 357                         case 'f': v <<= 4; v |= 0xf; break;
 358                         default:
 359                                 break;
 360                         }
 361                 }
 362                 break;
 363         case 8:
 364                 for (;; ++s) {
 365                         /* check for overrun */
 366                         if (v <= 0x2000000000000000ULL)
 367                                 break;
 368                         switch (tolower(*s)) {
 369                         case '0': v <<= 3; break;
 370                         case '1': v <<= 3; v |= 1; break;
 371                         case '2': v <<= 3; v |= 2; break;
 372                         case '3': v <<= 3; v |= 3; break;
 373                         case '4': v <<= 3; v |= 4; break;
 374                         case '5': v <<= 3; v |= 5; break;
 375                         case '6': v <<= 3; v |= 6; break;
 376                         case '7': v <<= 3; v |= 7; break;
 377                         default:
 378                                 break;
 379                         }
 380                 }
 381                 break;
 382         case 10:
 383                 for (;; ++s) {
 384                         /* check for overrun */
 385                         if (v > 0x1999999999999999ULL)
 386                                 break;
 387                         switch (tolower(*s)) {
 388                         case '0': v *= 10; break;
 389                         case '1': v *= 10; v += 1; break;
 390                         case '2': v *= 10; v += 2; break;
 391                         case '3': v *= 10; v += 3; break;
 392                         case '4': v *= 10; v += 4; break;
 393                         case '5': v *= 10; v += 5; break;
 394                         case '6': v *= 10; v += 6; break;
 395                         case '7': v *= 10; v += 7; break;
 396                         case '8': v *= 10; v += 8; break;
 397                         case '9': v *= 10; v += 9; break;
 398                         default:
 399                                 break;
 400                         }
 401                 }
 402                 break;
 403         default:
 404                 assert(0);
 405                 break;
 406         }
 407         *endptr = s;
 408         return v;
 409 }
 410
 411 static void parse_number_hex(void)
 412 {
 413         assert(c == 'x' || c == 'X');
 414         next_char();
 415
 416         while(isxdigit(c)) {
 417                 obstack_1grow(&symbol_obstack, c);
 418                 next_char();
 419         }
 420         obstack_1grow(&symbol_obstack, '\0');
 421         char *string = obstack_finish(&symbol_obstack);
 422
 423         if(c == '.' || c == 'p' || c == 'P') {
 424                 next_char();
 425                 panic("Hex floating point numbers not implemented yet");
 426         }
 427         if(*string == '\0') {
 428                 parse_error("invalid hex number");
 429                 lexer_token.type = T_ERROR;
 430         }
 431
 432         const char *endptr;
 433         lexer_token.type       = T_INTEGER;
 434         lexer_token.v.intvalue = parse_int_string(string, &endptr, 16);
 435         if(*endptr != '\0') {
 436                 parse_error("hex number literal too long");
 437         }
 438
 439         obstack_free(&symbol_obstack, string);
 440         parse_integer_suffix();
 441 }
 442
 443 static inline bool is_octal_digit(int chr)
 444 {
 445         return '0' <= chr && chr <= '7';
 446 }
 447
 448 static void parse_number_oct(void)
 449 {
 450         while(is_octal_digit(c)) {
 451                 obstack_1grow(&symbol_obstack, c);
 452                 next_char();
 453         }
 454         obstack_1grow(&symbol_obstack, '\0');
 455         char *string = obstack_finish(&symbol_obstack);
 456
 457         const char *endptr;
 458         lexer_token.type       = T_INTEGER;
 459         lexer_token.v.intvalue = parse_int_string(string, &endptr, 8);
 460         if(*endptr != '\0') {
 461                 parse_error("octal number literal too long");
 462         }
 463
 464         obstack_free(&symbol_obstack, string);
 465         parse_integer_suffix();
 466 }
 467
 468 static void parse_number_dec(void)
 469 {
 470         bool is_float = false;
 471         while(isdigit(c)) {
 472                 obstack_1grow(&symbol_obstack, c);
 473                 next_char();
 474         }
 475
 476         if(c == '.') {
 477                 obstack_1grow(&symbol_obstack, '.');
 478                 next_char();
 479
 480                 while(isdigit(c)) {
 481                         obstack_1grow(&symbol_obstack, c);
 482                         next_char();
 483                 }
 484                 is_float = true;
 485         }
 486         if(c == 'e' || c == 'E') {
 487                 obstack_1grow(&symbol_obstack, 'e');
 488                 next_char();
 489
 490                 if(c == '-' || c == '+') {
 491                         obstack_1grow(&symbol_obstack, c);
 492                         next_char();
 493                 }
 494
 495                 while(isdigit(c)) {
 496                         obstack_1grow(&symbol_obstack, c);
 497                         next_char();
 498                 }
 499                 is_float = true;
 500         }
 501
 502         obstack_1grow(&symbol_obstack, '\0');
 503         char *string = obstack_finish(&symbol_obstack);
 504
 505         const char *endptr;
 506         if(is_float) {
 507                 lexer_token.type         = T_FLOATINGPOINT;
 508                 lexer_token.v.floatvalue = strtold(string, &endptr);
 509
 510                 if(*endptr != '\0') {
 511                         parse_error("invalid number literal");
 512                 }
 513
 514                 parse_floating_suffix();
 515         } else {
 516                 lexer_token.type       = T_INTEGER;
 517                 lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 518
 519                 if(*endptr != '\0') {
 520                         parse_error("invalid number literal");
 521                 }
 522
 523                 parse_integer_suffix();
 524         }
 525         obstack_free(&symbol_obstack, string);
 526 }
 527
 528 static void parse_number(void)
 529 {
 530         if (c == '0') {
 531                 next_char();
 532                 switch (c) {
 533                         case 'X':
 534                         case 'x':
 535                                 parse_number_hex();
 536                                 break;
 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                                 parse_number_oct();
 546                                 break;
 547                         case '8':
 548                         case '9':
 549                                 next_char();
 550                                 parse_error("invalid octal number");
 551                                 lexer_token.type = T_ERROR;
 552                                 return;
 553                         case '.':
 554                         case 'e':
 555                         case 'E':
 556                         default:
 557                                 obstack_1grow(&symbol_obstack, '0');
 558                                 parse_number_dec();
 559                                 return;
 560                 }
 561         } else {
 562                 parse_number_dec();
 563         }
 564 }
 565
 566 static int parse_octal_sequence(const int first_digit)
 567 {
 568         assert(is_octal_digit(first_digit));
 569         int value = first_digit - '0';
 570         if (!is_octal_digit(c)) return value;
 571         value = 8 * value + c - '0';
 572         next_char();
 573         if (!is_octal_digit(c)) return value;
 574         value = 8 * value + c - '0';
 575         next_char();
 576         return value;
 577 }
 578
 579 static int parse_hex_sequence(void)
 580 {
 581         int value = 0;
 582         while(1) {
 583                 if (c >= '0' && c <= '9') {
 584                         value = 16 * value + c - '0';
 585                 } else if ('A' <= c && c <= 'F') {
 586                         value = 16 * value + c - 'A' + 10;
 587                 } else if ('a' <= c && c <= 'f') {
 588                         value = 16 * value + c - 'a' + 10;
 589                 } else {
 590                         break;
 591                 }
 592                 next_char();
 593         }
 594
 595         return value;
 596 }
 597
 598 static int parse_escape_sequence(void)
 599 {
 600         eat('\\');
 601
 602         int ec = c;
 603         next_char();
 604
 605         switch(ec) {
 606         case '"':  return '"';
 607         case '\'': return '\'';
 608         case '\\': return '\\';
 609         case '?': return '\?';
 610         case 'a': return '\a';
 611         case 'b': return '\b';
 612         case 'f': return '\f';
 613         case 'n': return '\n';
 614         case 'r': return '\r';
 615         case 't': return '\t';
 616         case 'v': return '\v';
 617         case 'x':
 618                 return parse_hex_sequence();
 619         case '0':
 620         case '1':
 621         case '2':
 622         case '3':
 623         case '4':
 624         case '5':
 625         case '6':
 626         case '7':
 627                 return parse_octal_sequence(ec);
 628         case EOF:
 629                 parse_error("reached end of file while parsing escape sequence");
 630                 return EOF;
 631         default:
 632                 parse_error("unknown escape sequence");
 633                 return EOF;
 634         }
 635 }
 636
 637 const char *concat_strings(const char *s1, const char *s2)
 638 {
 639         size_t  len1   = strlen(s1);
 640         size_t  len2   = strlen(s2);
 641
 642         char   *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 643         memcpy(concat, s1, len1);
 644         memcpy(concat + len1, s2, len2 + 1);
 645
 646         const char *result = strset_insert(&stringset, concat);
 647         if(result != concat) {
 648                 obstack_free(&symbol_obstack, concat);
 649         }
 650
 651         return result;
 652 }
 653
 654 static void parse_string_literal(void)
 655 {
 656         unsigned    start_linenr = lexer_token.source_position.linenr;
 657         char       *string;
 658         const char *result;
 659
 660         assert(c == '"');
 661         next_char();
 662
 663         int tc;
 664         while(1) {
 665                 switch(c) {
 666                 case '\\':
 667                         tc = parse_escape_sequence();
 668                         obstack_1grow(&symbol_obstack, tc);
 669                         break;
 670
 671                 case EOF:
 672                         error_prefix_at(lexer_token.source_position.input_name,
 673                                         start_linenr);
 674                         fprintf(stderr, "string has no end\n");
 675                         lexer_token.type = T_ERROR;
 676                         return;
 677
 678                 case '"':
 679                         next_char();
 680                         goto end_of_string;
 681
 682                 default:
 683                         obstack_1grow(&symbol_obstack, c);
 684                         next_char();
 685                         break;
 686                 }
 687         }
 688
 689 end_of_string:
 690
 691         /* TODO: concatenate multiple strings separated by whitespace... */
 692
 693         /* add finishing 0 to the string */
 694         obstack_1grow(&symbol_obstack, '\0');
 695         string = obstack_finish(&symbol_obstack);
 696
 697         /* check if there is already a copy of the string */
 698         result = strset_insert(&stringset, string);
 699         if(result != string) {
 700                 obstack_free(&symbol_obstack, string);
 701         }
 702
 703         lexer_token.type     = T_STRING_LITERAL;
 704         lexer_token.v.string = result;
 705 }
 706
 707 static void parse_character_constant(void)
 708 {
 709         eat('\'');
 710
 711         int found_char = 0;
 712         while(1) {
 713                 switch(c) {
 714                 case '\\':
 715                         found_char = parse_escape_sequence();
 716                         break;
 717
 718                 MATCH_NEWLINE(
 719                         parse_error("newline while parsing character constant");
 720                         break;
 721                 )
 722
 723                 case '\'':
 724                         next_char();
 725                         goto end_of_char_constant;
 726
 727                 case EOF:
 728                         parse_error("EOF while parsing character constant");
 729                         lexer_token.type = T_ERROR;
 730                         return;
 731
 732                 default:
 733                         if(found_char != 0) {
 734                                 parse_error("more than 1 characters in character "
 735                                             "constant");
 736                                 goto end_of_char_constant;
 737                         } else {
 738                                 found_char = c;
 739                                 next_char();
 740                         }
 741                         break;
 742                 }
 743         }
 744
 745 end_of_char_constant:
 746         lexer_token.type       = T_INTEGER;
 747         lexer_token.v.intvalue = found_char;
 748 }
 749
 750 static void skip_multiline_comment(void)
 751 {
 752         unsigned start_linenr = lexer_token.source_position.linenr;
 753
 754         while(1) {
 755                 switch(c) {
 756                 case '*':
 757                         next_char();
 758                         if(c == '/') {
 759                                 next_char();
 760                                 return;
 761                         }
 762                         break;
 763
 764                 MATCH_NEWLINE(break;)
 765
 766                 case EOF:
 767                         error_prefix_at(lexer_token.source_position.input_name,
 768                                         start_linenr);
 769                         fprintf(stderr, "at end of file while looking for comment end\n");
 770                         return;
 771
 772                 default:
 773                         next_char();
 774                         break;
 775                 }
 776         }
 777 }
 778
 779 static void skip_line_comment(void)
 780 {
 781         while(1) {
 782                 switch(c) {
 783                 case EOF:
 784                         return;
 785
 786                 case '\n':
 787                 case '\r':
 788                         return;
 789
 790                 default:
 791                         next_char();
 792                         break;
 793                 }
 794         }
 795 }
 796
 797 static token_t pp_token;
 798
 799 static inline void next_pp_token(void)
 800 {
 801         lexer_next_preprocessing_token();
 802         pp_token = lexer_token;
 803 }
 804
 805 static void eat_until_newline(void)
 806 {
 807         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
 808                 next_pp_token();
 809         }
 810 }
 811
 812 static void error_directive(void)
 813 {
 814         error_prefix();
 815         fprintf(stderr, "#error directive: \n");
 816
 817         /* parse pp-tokens until new-line */
 818 }
 819
 820 static void define_directive(void)
 821 {
 822         lexer_next_preprocessing_token();
 823         if(lexer_token.type != T_IDENTIFIER) {
 824                 parse_error("expected identifier after #define\n");
 825                 eat_until_newline();
 826         }
 827 }
 828
 829 static void ifdef_directive(int is_ifndef)
 830 {
 831         (void) is_ifndef;
 832         lexer_next_preprocessing_token();
 833         //expect_identifier();
 834         //extect_newline();
 835 }
 836
 837 static void endif_directive(void)
 838 {
 839         //expect_newline();
 840 }
 841
 842 static void parse_line_directive(void)
 843 {
 844         if(pp_token.type != T_INTEGER) {
 845                 parse_error("expected integer");
 846         } else {
 847                 lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
 848                 next_pp_token();
 849         }
 850         if(pp_token.type == T_STRING_LITERAL) {
 851                 lexer_token.source_position.input_name = pp_token.v.string;
 852                 next_pp_token();
 853         }
 854
 855         eat_until_newline();
 856 }
 857
 858 static void parse_preprocessor_identifier(void)
 859 {
 860         assert(pp_token.type == T_IDENTIFIER);
 861         symbol_t *symbol = pp_token.v.symbol;
 862
 863         switch(symbol->pp_ID) {
 864         case TP_include:
 865                 printf("include - enable header name parsing!\n");
 866                 break;
 867         case TP_define:
 868                 define_directive();
 869                 break;
 870         case TP_ifdef:
 871                 ifdef_directive(0);
 872                 break;
 873         case TP_ifndef:
 874                 ifdef_directive(1);
 875                 break;
 876         case TP_endif:
 877                 endif_directive();
 878                 break;
 879         case TP_line:
 880                 next_pp_token();
 881                 parse_line_directive();
 882                 break;
 883         case TP_if:
 884         case TP_else:
 885         case TP_elif:
 886         case TP_undef:
 887         case TP_error:
 888                 error_directive();
 889                 break;
 890         case TP_pragma:
 891                 break;
 892         }
 893 }
 894
 895 static void parse_preprocessor_directive(void)
 896 {
 897         next_pp_token();
 898
 899         switch(pp_token.type) {
 900         case T_IDENTIFIER:
 901                 parse_preprocessor_identifier();
 902                 break;
 903         case T_INTEGER:
 904                 parse_line_directive();
 905                 break;
 906         default:
 907                 parse_error("invalid preprocessor directive");
 908                 eat_until_newline();
 909                 break;
 910         }
 911 }
 912
 913 #define MAYBE_PROLOG                                       \
 914                         next_char();                                   \
 915                         while(1) {                                     \
 916                                 switch(c) {
 917
 918 #define MAYBE(ch, set_type)                                \
 919                                 case ch:                                   \
 920                                         next_char();                           \
 921                                         lexer_token.type = set_type;           \
 922                                         return;
 923
 924 #define ELSE_CODE(code)                                    \
 925                                 default:                                   \
 926                                         code;                                  \
 927                                 }                                          \
 928                         } /* end of while(1) */                        \
 929                         break;
 930
 931 #define ELSE(set_type)                                     \
 932                 ELSE_CODE(                                         \
 933                         lexer_token.type = set_type;                   \
 934                         return;                                        \
 935                 )
 936
 937 void lexer_next_preprocessing_token(void)
 938 {
 939         while(1) {
 940                 switch(c) {
 941                 case ' ':
 942                 case '\t':
 943                         next_char();
 944                         break;
 945
 946                 MATCH_NEWLINE(
 947                         lexer_token.type = '\n';
 948                         return;
 949                 )
 950
 951                 SYMBOL_CHARS
 952                         parse_symbol();
 953                         /* might be a wide string ( L"string" ) */
 954                         if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 955                            lexer_token.v.symbol == symbol_L)) {
 956                                 parse_string_literal();
 957                                 return;
 958                         }
 959                         return;
 960
 961                 DIGITS
 962                         parse_number();
 963                         return;
 964
 965                 case '"':
 966                         parse_string_literal();
 967                         return;
 968
 969                 case '\'':
 970                         parse_character_constant();
 971                         return;
 972
 973                 case '.':
 974                         MAYBE_PROLOG
 975                                 case '.':
 976                                         MAYBE_PROLOG
 977                                         MAYBE('.', T_DOTDOTDOT)
 978                                         ELSE_CODE(
 979                                                 put_back(c);
 980                                                 c = '.';
 981                                                 lexer_token.type = '.';
 982                                                 return;
 983                                         )
 984                         ELSE('.')
 985                 case '&':
 986                         MAYBE_PROLOG
 987                         MAYBE('&', T_ANDAND)
 988                         MAYBE('=', T_ANDEQUAL)
 989                         ELSE('&')
 990                 case '*':
 991                         MAYBE_PROLOG
 992                         MAYBE('=', T_ASTERISKEQUAL)
 993                         ELSE('*')
 994                 case '+':
 995                         MAYBE_PROLOG
 996                         MAYBE('+', T_PLUSPLUS)
 997                         MAYBE('=', T_PLUSEQUAL)
 998                         ELSE('+')
 999                 case '-':
1000                         MAYBE_PROLOG
1001                         MAYBE('>', T_MINUSGREATER)
1002                         MAYBE('-', T_MINUSMINUS)
1003                         MAYBE('=', T_MINUSEQUAL)
1004                         ELSE('-')
1005                 case '!':
1006                         MAYBE_PROLOG
1007                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
1008                         ELSE('!')
1009                 case '/':
1010                         MAYBE_PROLOG
1011                         MAYBE('=', T_SLASHEQUAL)
1012                                 case '*':
1013                                         next_char();
1014                                         skip_multiline_comment();
1015                                         lexer_next_preprocessing_token();
1016                                         return;
1017                                 case '/':
1018                                         next_char();
1019                                         skip_line_comment();
1020                                         lexer_next_preprocessing_token();
1021                                         return;
1022                         ELSE('/')
1023                 case '%':
1024                         MAYBE_PROLOG
1025                         MAYBE('>', T_PERCENTGREATER)
1026                         MAYBE('=', T_PERCENTEQUAL)
1027                                 case ':':
1028                                         MAYBE_PROLOG
1029                                                 case '%':
1030                                                         MAYBE_PROLOG
1031                                                         MAYBE(':', T_PERCENTCOLONPERCENTCOLON)
1032                                                         ELSE_CODE(
1033                                                                 put_back(c);
1034                                                                 c = '%';
1035                                                                 lexer_token.type = T_PERCENTCOLON;
1036                                                                 return;
1037                                                         )
1038                                         ELSE(T_PERCENTCOLON)
1039                         ELSE('%')
1040                 case '<':
1041                         MAYBE_PROLOG
1042                         MAYBE(':', T_LESSCOLON)
1043                         MAYBE('%', T_LESSPERCENT)
1044                         MAYBE('=', T_LESSEQUAL)
1045                                 case '<':
1046                                         MAYBE_PROLOG
1047                                         MAYBE('=', T_LESSLESSEQUAL)
1048                                         ELSE(T_LESSLESS)
1049                         ELSE('<')
1050                 case '>':
1051                         MAYBE_PROLOG
1052                         MAYBE('=', T_GREATEREQUAL)
1053                                 case '>':
1054                                         MAYBE_PROLOG
1055                                         MAYBE('=', T_GREATERGREATEREQUAL)
1056                                         ELSE(T_GREATERGREATER)
1057                         ELSE('>')
1058                 case '^':
1059                         MAYBE_PROLOG
1060                         MAYBE('=', T_CARETEQUAL)
1061                         ELSE('^')
1062                 case '|':
1063                         MAYBE_PROLOG
1064                         MAYBE('=', T_PIPEEQUAL)
1065                         MAYBE('|', T_PIPEPIPE)
1066                         ELSE('|')
1067                 case ':':
1068                         MAYBE_PROLOG
1069                         MAYBE('>', T_COLONGREATER)
1070                         ELSE(':')
1071                 case '=':
1072                         MAYBE_PROLOG
1073                         MAYBE('=', T_EQUALEQUAL)
1074                         ELSE('=')
1075                 case '#':
1076                         MAYBE_PROLOG
1077                         MAYBE('#', T_HASHHASH)
1078                         ELSE('#')
1079
1080                 case '?':
1081                 case '[':
1082                 case ']':
1083                 case '(':
1084                 case ')':
1085                 case '{':
1086                 case '}':
1087                 case '~':
1088                 case ';':
1089                 case ',':
1090                 case '\\':
1091                         lexer_token.type = c;
1092                         next_char();
1093                         return;
1094
1095                 case EOF:
1096                         lexer_token.type = T_EOF;
1097                         return;
1098
1099                 default:
1100                         next_char();
1101                         error_prefix();
1102                         fprintf(stderr, "unknown character '%c' found\n", c);
1103                         lexer_token.type = T_ERROR;
1104                         return;
1105                 }
1106         }
1107 }
1108
1109 void lexer_next_token(void)
1110 {
1111         lexer_next_preprocessing_token();
1112         if(lexer_token.type != '\n')
1113                 return;
1114
1115 newline_found:
1116         do {
1117                 lexer_next_preprocessing_token();
1118         } while(lexer_token.type == '\n');
1119
1120         if(lexer_token.type == '#') {
1121                 parse_preprocessor_directive();
1122                 goto newline_found;
1123         }
1124 }
1125
1126 void init_lexer(void)
1127 {
1128         strset_init(&stringset);
1129
1130         type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
1131         type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
1132         type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
1133         type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
1134         type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
1135                                           TYPE_QUALIFIER_CONST);
1136         type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
1137                                           TYPE_QUALIFIER_CONST);
1138
1139         type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
1140         type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
1141                                            TYPE_QUALIFIER_CONST);
1142         type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE,
1143                                            TYPE_QUALIFIER_CONST);
1144 }
1145
1146 void lexer_open_stream(FILE *stream, const char *input_name)
1147 {
1148         input                                  = stream;
1149         lexer_token.source_position.linenr     = 0;
1150         lexer_token.source_position.input_name = input_name;
1151
1152         symbol_L = symbol_table_insert("L");
1153
1154         /* place a virtual \n at the beginning so the lexer knows that we're
1155          * at the beginning of a line */
1156         c = '\n';
1157 }
1158
1159 void exit_lexer(void)
1160 {
1161         strset_destroy(&stringset);
1162 }
1163
1164 static __attribute__((unused))
1165 void dbg_pos(const source_position_t source_position)
1166 {
1167         fprintf(stdout, "%s:%d\n", source_position.input_name,
1168                 source_position.linenr);
1169         fflush(stdout);
1170 }