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                                 goto end;
 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                                 goto end;
 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                                 goto end;
 400                         }
 401                 }
 402                 break;
 403         default:
 404                 assert(0);
 405                 break;
 406         }
 407 end:
 408         *endptr = s;
 409         return v;
 410 }
 411
 412 static void parse_number_hex(void)
 413 {
 414         assert(c == 'x' || c == 'X');
 415         next_char();
 416
 417         while(isxdigit(c)) {
 418                 obstack_1grow(&symbol_obstack, c);
 419                 next_char();
 420         }
 421         obstack_1grow(&symbol_obstack, '\0');
 422         char *string = obstack_finish(&symbol_obstack);
 423
 424         if(c == '.' || c == 'p' || c == 'P') {
 425                 next_char();
 426                 panic("Hex floating point numbers not implemented yet");
 427         }
 428         if(*string == '\0') {
 429                 parse_error("invalid hex number");
 430                 lexer_token.type = T_ERROR;
 431         }
 432
 433         const char *endptr;
 434         lexer_token.type       = T_INTEGER;
 435         lexer_token.v.intvalue = parse_int_string(string, &endptr, 16);
 436         if(*endptr != '\0') {
 437                 parse_error("hex number literal too long");
 438         }
 439
 440         obstack_free(&symbol_obstack, string);
 441         parse_integer_suffix();
 442 }
 443
 444 static inline bool is_octal_digit(int chr)
 445 {
 446         return '0' <= chr && chr <= '7';
 447 }
 448
 449 static void parse_number_oct(void)
 450 {
 451         while(is_octal_digit(c)) {
 452                 obstack_1grow(&symbol_obstack, c);
 453                 next_char();
 454         }
 455         obstack_1grow(&symbol_obstack, '\0');
 456         char *string = obstack_finish(&symbol_obstack);
 457
 458         const char *endptr;
 459         lexer_token.type       = T_INTEGER;
 460         lexer_token.v.intvalue = parse_int_string(string, &endptr, 8);
 461         if(*endptr != '\0') {
 462                 parse_error("octal number literal too long");
 463         }
 464
 465         obstack_free(&symbol_obstack, string);
 466         parse_integer_suffix();
 467 }
 468
 469 static void parse_number_dec(void)
 470 {
 471         bool is_float = false;
 472         while(isdigit(c)) {
 473                 obstack_1grow(&symbol_obstack, c);
 474                 next_char();
 475         }
 476
 477         if(c == '.') {
 478                 obstack_1grow(&symbol_obstack, '.');
 479                 next_char();
 480
 481                 while(isdigit(c)) {
 482                         obstack_1grow(&symbol_obstack, c);
 483                         next_char();
 484                 }
 485                 is_float = true;
 486         }
 487         if(c == 'e' || c == 'E') {
 488                 obstack_1grow(&symbol_obstack, 'e');
 489                 next_char();
 490
 491                 if(c == '-' || c == '+') {
 492                         obstack_1grow(&symbol_obstack, c);
 493                         next_char();
 494                 }
 495
 496                 while(isdigit(c)) {
 497                         obstack_1grow(&symbol_obstack, c);
 498                         next_char();
 499                 }
 500                 is_float = true;
 501         }
 502
 503         obstack_1grow(&symbol_obstack, '\0');
 504         char *string = obstack_finish(&symbol_obstack);
 505
 506         const char *endptr;
 507         if(is_float) {
 508                 lexer_token.type         = T_FLOATINGPOINT;
 509                 lexer_token.v.floatvalue = strtold(string, &endptr);
 510
 511                 if(*endptr != '\0') {
 512                         parse_error("invalid number literal");
 513                 }
 514
 515                 parse_floating_suffix();
 516         } else {
 517                 lexer_token.type       = T_INTEGER;
 518                 lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 519
 520                 if(*endptr != '\0') {
 521                         parse_error("invalid number literal");
 522                 }
 523
 524                 parse_integer_suffix();
 525         }
 526         obstack_free(&symbol_obstack, string);
 527 }
 528
 529 static void parse_number(void)
 530 {
 531         if (c == '0') {
 532                 next_char();
 533                 switch (c) {
 534                         case 'X':
 535                         case 'x':
 536                                 parse_number_hex();
 537                                 break;
 538                         case '0':
 539                         case '1':
 540                         case '2':
 541                         case '3':
 542                         case '4':
 543                         case '5':
 544                         case '6':
 545                         case '7':
 546                                 parse_number_oct();
 547                                 break;
 548                         case '8':
 549                         case '9':
 550                                 next_char();
 551                                 parse_error("invalid octal number");
 552                                 lexer_token.type = T_ERROR;
 553                                 return;
 554                         case '.':
 555                         case 'e':
 556                         case 'E':
 557                         default:
 558                                 obstack_1grow(&symbol_obstack, '0');
 559                                 parse_number_dec();
 560                                 return;
 561                 }
 562         } else {
 563                 parse_number_dec();
 564         }
 565 }
 566
 567 static int parse_octal_sequence(const int first_digit)
 568 {
 569         assert(is_octal_digit(first_digit));
 570         int value = first_digit - '0';
 571         if (!is_octal_digit(c)) return value;
 572         value = 8 * value + c - '0';
 573         next_char();
 574         if (!is_octal_digit(c)) return value;
 575         value = 8 * value + c - '0';
 576         next_char();
 577         return value;
 578 }
 579
 580 static int parse_hex_sequence(void)
 581 {
 582         int value = 0;
 583         while(1) {
 584                 if (c >= '0' && c <= '9') {
 585                         value = 16 * value + c - '0';
 586                 } else if ('A' <= c && c <= 'F') {
 587                         value = 16 * value + c - 'A' + 10;
 588                 } else if ('a' <= c && c <= 'f') {
 589                         value = 16 * value + c - 'a' + 10;
 590                 } else {
 591                         break;
 592                 }
 593                 next_char();
 594         }
 595
 596         return value;
 597 }
 598
 599 static int parse_escape_sequence(void)
 600 {
 601         eat('\\');
 602
 603         int ec = c;
 604         next_char();
 605
 606         switch(ec) {
 607         case '"':  return '"';
 608         case '\'': return '\'';
 609         case '\\': return '\\';
 610         case '?': return '\?';
 611         case 'a': return '\a';
 612         case 'b': return '\b';
 613         case 'f': return '\f';
 614         case 'n': return '\n';
 615         case 'r': return '\r';
 616         case 't': return '\t';
 617         case 'v': return '\v';
 618         case 'x':
 619                 return parse_hex_sequence();
 620         case '0':
 621         case '1':
 622         case '2':
 623         case '3':
 624         case '4':
 625         case '5':
 626         case '6':
 627         case '7':
 628                 return parse_octal_sequence(ec);
 629         case EOF:
 630                 parse_error("reached end of file while parsing escape sequence");
 631                 return EOF;
 632         default:
 633                 parse_error("unknown escape sequence");
 634                 return EOF;
 635         }
 636 }
 637
 638 const char *concat_strings(const char *s1, const char *s2)
 639 {
 640         size_t  len1   = strlen(s1);
 641         size_t  len2   = strlen(s2);
 642
 643         char   *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 644         memcpy(concat, s1, len1);
 645         memcpy(concat + len1, s2, len2 + 1);
 646
 647         const char *result = strset_insert(&stringset, concat);
 648         if(result != concat) {
 649                 obstack_free(&symbol_obstack, concat);
 650         }
 651
 652         return result;
 653 }
 654
 655 static void parse_string_literal(void)
 656 {
 657         unsigned    start_linenr = lexer_token.source_position.linenr;
 658         char       *string;
 659         const char *result;
 660
 661         assert(c == '"');
 662         next_char();
 663
 664         int tc;
 665         while(1) {
 666                 switch(c) {
 667                 case '\\':
 668                         tc = parse_escape_sequence();
 669                         obstack_1grow(&symbol_obstack, tc);
 670                         break;
 671
 672                 case EOF:
 673                         error_prefix_at(lexer_token.source_position.input_name,
 674                                         start_linenr);
 675                         fprintf(stderr, "string has no end\n");
 676                         lexer_token.type = T_ERROR;
 677                         return;
 678
 679                 case '"':
 680                         next_char();
 681                         goto end_of_string;
 682
 683                 default:
 684                         obstack_1grow(&symbol_obstack, c);
 685                         next_char();
 686                         break;
 687                 }
 688         }
 689
 690 end_of_string:
 691
 692         /* TODO: concatenate multiple strings separated by whitespace... */
 693
 694         /* add finishing 0 to the string */
 695         obstack_1grow(&symbol_obstack, '\0');
 696         string = obstack_finish(&symbol_obstack);
 697
 698         /* check if there is already a copy of the string */
 699         result = strset_insert(&stringset, string);
 700         if(result != string) {
 701                 obstack_free(&symbol_obstack, string);
 702         }
 703
 704         lexer_token.type     = T_STRING_LITERAL;
 705         lexer_token.v.string = result;
 706 }
 707
 708 static void parse_character_constant(void)
 709 {
 710         eat('\'');
 711
 712         int found_char = 0;
 713         while(1) {
 714                 switch(c) {
 715                 case '\\':
 716                         found_char = parse_escape_sequence();
 717                         break;
 718
 719                 MATCH_NEWLINE(
 720                         parse_error("newline while parsing character constant");
 721                         break;
 722                 )
 723
 724                 case '\'':
 725                         next_char();
 726                         goto end_of_char_constant;
 727
 728                 case EOF:
 729                         parse_error("EOF while parsing character constant");
 730                         lexer_token.type = T_ERROR;
 731                         return;
 732
 733                 default:
 734                         if(found_char != 0) {
 735                                 parse_error("more than 1 characters in character "
 736                                             "constant");
 737                                 goto end_of_char_constant;
 738                         } else {
 739                                 found_char = c;
 740                                 next_char();
 741                         }
 742                         break;
 743                 }
 744         }
 745
 746 end_of_char_constant:
 747         lexer_token.type       = T_INTEGER;
 748         lexer_token.v.intvalue = found_char;
 749 }
 750
 751 static void skip_multiline_comment(void)
 752 {
 753         unsigned start_linenr = lexer_token.source_position.linenr;
 754
 755         while(1) {
 756                 switch(c) {
 757                 case '*':
 758                         next_char();
 759                         if(c == '/') {
 760                                 next_char();
 761                                 return;
 762                         }
 763                         break;
 764
 765                 MATCH_NEWLINE(break;)
 766
 767                 case EOF:
 768                         error_prefix_at(lexer_token.source_position.input_name,
 769                                         start_linenr);
 770                         fprintf(stderr, "at end of file while looking for comment end\n");
 771                         return;
 772
 773                 default:
 774                         next_char();
 775                         break;
 776                 }
 777         }
 778 }
 779
 780 static void skip_line_comment(void)
 781 {
 782         while(1) {
 783                 switch(c) {
 784                 case EOF:
 785                         return;
 786
 787                 case '\n':
 788                 case '\r':
 789                         return;
 790
 791                 default:
 792                         next_char();
 793                         break;
 794                 }
 795         }
 796 }
 797
 798 static token_t pp_token;
 799
 800 static inline void next_pp_token(void)
 801 {
 802         lexer_next_preprocessing_token();
 803         pp_token = lexer_token;
 804 }
 805
 806 static void eat_until_newline(void)
 807 {
 808         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
 809                 next_pp_token();
 810         }
 811 }
 812
 813 static void error_directive(void)
 814 {
 815         error_prefix();
 816         fprintf(stderr, "#error directive: \n");
 817
 818         /* parse pp-tokens until new-line */
 819 }
 820
 821 static void define_directive(void)
 822 {
 823         lexer_next_preprocessing_token();
 824         if(lexer_token.type != T_IDENTIFIER) {
 825                 parse_error("expected identifier after #define\n");
 826                 eat_until_newline();
 827         }
 828 }
 829
 830 static void ifdef_directive(int is_ifndef)
 831 {
 832         (void) is_ifndef;
 833         lexer_next_preprocessing_token();
 834         //expect_identifier();
 835         //extect_newline();
 836 }
 837
 838 static void endif_directive(void)
 839 {
 840         //expect_newline();
 841 }
 842
 843 static void parse_line_directive(void)
 844 {
 845         if(pp_token.type != T_INTEGER) {
 846                 parse_error("expected integer");
 847         } else {
 848                 lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
 849                 next_pp_token();
 850         }
 851         if(pp_token.type == T_STRING_LITERAL) {
 852                 lexer_token.source_position.input_name = pp_token.v.string;
 853                 next_pp_token();
 854         }
 855
 856         eat_until_newline();
 857 }
 858
 859 static void parse_preprocessor_identifier(void)
 860 {
 861         assert(pp_token.type == T_IDENTIFIER);
 862         symbol_t *symbol = pp_token.v.symbol;
 863
 864         switch(symbol->pp_ID) {
 865         case TP_include:
 866                 printf("include - enable header name parsing!\n");
 867                 break;
 868         case TP_define:
 869                 define_directive();
 870                 break;
 871         case TP_ifdef:
 872                 ifdef_directive(0);
 873                 break;
 874         case TP_ifndef:
 875                 ifdef_directive(1);
 876                 break;
 877         case TP_endif:
 878                 endif_directive();
 879                 break;
 880         case TP_line:
 881                 next_pp_token();
 882                 parse_line_directive();
 883                 break;
 884         case TP_if:
 885         case TP_else:
 886         case TP_elif:
 887         case TP_undef:
 888         case TP_error:
 889                 error_directive();
 890                 break;
 891         case TP_pragma:
 892                 break;
 893         }
 894 }
 895
 896 static void parse_preprocessor_directive(void)
 897 {
 898         next_pp_token();
 899
 900         switch(pp_token.type) {
 901         case T_IDENTIFIER:
 902                 parse_preprocessor_identifier();
 903                 break;
 904         case T_INTEGER:
 905                 parse_line_directive();
 906                 break;
 907         default:
 908                 parse_error("invalid preprocessor directive");
 909                 eat_until_newline();
 910                 break;
 911         }
 912 }
 913
 914 #define MAYBE_PROLOG                                       \
 915                         next_char();                                   \
 916                         while(1) {                                     \
 917                                 switch(c) {
 918
 919 #define MAYBE(ch, set_type)                                \
 920                                 case ch:                                   \
 921                                         next_char();                           \
 922                                         lexer_token.type = set_type;           \
 923                                         return;
 924
 925 #define ELSE_CODE(code)                                    \
 926                                 default:                                   \
 927                                         code;                                  \
 928                                 }                                          \
 929                         } /* end of while(1) */                        \
 930                         break;
 931
 932 #define ELSE(set_type)                                     \
 933                 ELSE_CODE(                                         \
 934                         lexer_token.type = set_type;                   \
 935                         return;                                        \
 936                 )
 937
 938 void lexer_next_preprocessing_token(void)
 939 {
 940         while(1) {
 941                 switch(c) {
 942                 case ' ':
 943                 case '\t':
 944                         next_char();
 945                         break;
 946
 947                 MATCH_NEWLINE(
 948                         lexer_token.type = '\n';
 949                         return;
 950                 )
 951
 952                 SYMBOL_CHARS
 953                         parse_symbol();
 954                         /* might be a wide string ( L"string" ) */
 955                         if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 956                            lexer_token.v.symbol == symbol_L)) {
 957                                 parse_string_literal();
 958                                 return;
 959                         }
 960                         return;
 961
 962                 DIGITS
 963                         parse_number();
 964                         return;
 965
 966                 case '"':
 967                         parse_string_literal();
 968                         return;
 969
 970                 case '\'':
 971                         parse_character_constant();
 972                         return;
 973
 974                 case '.':
 975                         MAYBE_PROLOG
 976                                 case '.':
 977                                         MAYBE_PROLOG
 978                                         MAYBE('.', T_DOTDOTDOT)
 979                                         ELSE_CODE(
 980                                                 put_back(c);
 981                                                 c = '.';
 982                                                 lexer_token.type = '.';
 983                                                 return;
 984                                         )
 985                         ELSE('.')
 986                 case '&':
 987                         MAYBE_PROLOG
 988                         MAYBE('&', T_ANDAND)
 989                         MAYBE('=', T_ANDEQUAL)
 990                         ELSE('&')
 991                 case '*':
 992                         MAYBE_PROLOG
 993                         MAYBE('=', T_ASTERISKEQUAL)
 994                         ELSE('*')
 995                 case '+':
 996                         MAYBE_PROLOG
 997                         MAYBE('+', T_PLUSPLUS)
 998                         MAYBE('=', T_PLUSEQUAL)
 999                         ELSE('+')
1000                 case '-':
1001                         MAYBE_PROLOG
1002                         MAYBE('>', T_MINUSGREATER)
1003                         MAYBE('-', T_MINUSMINUS)
1004                         MAYBE('=', T_MINUSEQUAL)
1005                         ELSE('-')
1006                 case '!':
1007                         MAYBE_PROLOG
1008                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
1009                         ELSE('!')
1010                 case '/':
1011                         MAYBE_PROLOG
1012                         MAYBE('=', T_SLASHEQUAL)
1013                                 case '*':
1014                                         next_char();
1015                                         skip_multiline_comment();
1016                                         lexer_next_preprocessing_token();
1017                                         return;
1018                                 case '/':
1019                                         next_char();
1020                                         skip_line_comment();
1021                                         lexer_next_preprocessing_token();
1022                                         return;
1023                         ELSE('/')
1024                 case '%':
1025                         MAYBE_PROLOG
1026                         MAYBE('>', T_PERCENTGREATER)
1027                         MAYBE('=', T_PERCENTEQUAL)
1028                                 case ':':
1029                                         MAYBE_PROLOG
1030                                                 case '%':
1031                                                         MAYBE_PROLOG
1032                                                         MAYBE(':', T_PERCENTCOLONPERCENTCOLON)
1033                                                         ELSE_CODE(
1034                                                                 put_back(c);
1035                                                                 c = '%';
1036                                                                 lexer_token.type = T_PERCENTCOLON;
1037                                                                 return;
1038                                                         )
1039                                         ELSE(T_PERCENTCOLON)
1040                         ELSE('%')
1041                 case '<':
1042                         MAYBE_PROLOG
1043                         MAYBE(':', T_LESSCOLON)
1044                         MAYBE('%', T_LESSPERCENT)
1045                         MAYBE('=', T_LESSEQUAL)
1046                                 case '<':
1047                                         MAYBE_PROLOG
1048                                         MAYBE('=', T_LESSLESSEQUAL)
1049                                         ELSE(T_LESSLESS)
1050                         ELSE('<')
1051                 case '>':
1052                         MAYBE_PROLOG
1053                         MAYBE('=', T_GREATEREQUAL)
1054                                 case '>':
1055                                         MAYBE_PROLOG
1056                                         MAYBE('=', T_GREATERGREATEREQUAL)
1057                                         ELSE(T_GREATERGREATER)
1058                         ELSE('>')
1059                 case '^':
1060                         MAYBE_PROLOG
1061                         MAYBE('=', T_CARETEQUAL)
1062                         ELSE('^')
1063                 case '|':
1064                         MAYBE_PROLOG
1065                         MAYBE('=', T_PIPEEQUAL)
1066                         MAYBE('|', T_PIPEPIPE)
1067                         ELSE('|')
1068                 case ':':
1069                         MAYBE_PROLOG
1070                         MAYBE('>', T_COLONGREATER)
1071                         ELSE(':')
1072                 case '=':
1073                         MAYBE_PROLOG
1074                         MAYBE('=', T_EQUALEQUAL)
1075                         ELSE('=')
1076                 case '#':
1077                         MAYBE_PROLOG
1078                         MAYBE('#', T_HASHHASH)
1079                         ELSE('#')
1080
1081                 case '?':
1082                 case '[':
1083                 case ']':
1084                 case '(':
1085                 case ')':
1086                 case '{':
1087                 case '}':
1088                 case '~':
1089                 case ';':
1090                 case ',':
1091                 case '\\':
1092                         lexer_token.type = c;
1093                         next_char();
1094                         return;
1095
1096                 case EOF:
1097                         lexer_token.type = T_EOF;
1098                         return;
1099
1100                 default:
1101                         next_char();
1102                         error_prefix();
1103                         fprintf(stderr, "unknown character '%c' found\n", c);
1104                         lexer_token.type = T_ERROR;
1105                         return;
1106                 }
1107         }
1108 }
1109
1110 void lexer_next_token(void)
1111 {
1112         lexer_next_preprocessing_token();
1113         if(lexer_token.type != '\n')
1114                 return;
1115
1116 newline_found:
1117         do {
1118                 lexer_next_preprocessing_token();
1119         } while(lexer_token.type == '\n');
1120
1121         if(lexer_token.type == '#') {
1122                 parse_preprocessor_directive();
1123                 goto newline_found;
1124         }
1125 }
1126
1127 void init_lexer(void)
1128 {
1129         strset_init(&stringset);
1130
1131         type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
1132         type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
1133         type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
1134         type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
1135         type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
1136                                           TYPE_QUALIFIER_CONST);
1137         type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
1138                                           TYPE_QUALIFIER_CONST);
1139
1140         type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
1141         type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
1142                                            TYPE_QUALIFIER_CONST);
1143         type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE,
1144                                            TYPE_QUALIFIER_CONST);
1145 }
1146
1147 void lexer_open_stream(FILE *stream, const char *input_name)
1148 {
1149         input                                  = stream;
1150         lexer_token.source_position.linenr     = 0;
1151         lexer_token.source_position.input_name = input_name;
1152
1153         symbol_L = symbol_table_insert("L");
1154
1155         /* place a virtual \n at the beginning so the lexer knows that we're
1156          * at the beginning of a line */
1157         c = '\n';
1158 }
1159
1160 void exit_lexer(void)
1161 {
1162         strset_destroy(&stringset);
1163 }
1164
1165 static __attribute__((unused))
1166 void dbg_pos(const source_position_t source_position)
1167 {
1168         fprintf(stdout, "%s:%d\n", source_position.input_name,
1169                 source_position.linenr);
1170         fflush(stdout);
1171 }