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         if(is_float) {
 507                 char *endptr;
 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                 const char *endptr;
 518                 lexer_token.type       = T_INTEGER;
 519                 lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 520
 521                 if(*endptr != '\0') {
 522                         parse_error("invalid number literal");
 523                 }
 524
 525                 parse_integer_suffix();
 526         }
 527         obstack_free(&symbol_obstack, string);
 528 }
 529
 530 static void parse_number(void)
 531 {
 532         if (c == '0') {
 533                 next_char();
 534                 switch (c) {
 535                         case 'X':
 536                         case 'x':
 537                                 parse_number_hex();
 538                                 break;
 539                         case '0':
 540                         case '1':
 541                         case '2':
 542                         case '3':
 543                         case '4':
 544                         case '5':
 545                         case '6':
 546                         case '7':
 547                                 parse_number_oct();
 548                                 break;
 549                         case '8':
 550                         case '9':
 551                                 next_char();
 552                                 parse_error("invalid octal number");
 553                                 lexer_token.type = T_ERROR;
 554                                 return;
 555                         case '.':
 556                         case 'e':
 557                         case 'E':
 558                         default:
 559                                 obstack_1grow(&symbol_obstack, '0');
 560                                 parse_number_dec();
 561                                 return;
 562                 }
 563         } else {
 564                 parse_number_dec();
 565         }
 566 }
 567
 568 static int parse_octal_sequence(const int first_digit)
 569 {
 570         assert(is_octal_digit(first_digit));
 571         int value = first_digit - '0';
 572         if (!is_octal_digit(c)) return value;
 573         value = 8 * value + c - '0';
 574         next_char();
 575         if (!is_octal_digit(c)) return value;
 576         value = 8 * value + c - '0';
 577         next_char();
 578         return value;
 579 }
 580
 581 static int parse_hex_sequence(void)
 582 {
 583         int value = 0;
 584         while(1) {
 585                 if (c >= '0' && c <= '9') {
 586                         value = 16 * value + c - '0';
 587                 } else if ('A' <= c && c <= 'F') {
 588                         value = 16 * value + c - 'A' + 10;
 589                 } else if ('a' <= c && c <= 'f') {
 590                         value = 16 * value + c - 'a' + 10;
 591                 } else {
 592                         break;
 593                 }
 594                 next_char();
 595         }
 596
 597         return value;
 598 }
 599
 600 static int parse_escape_sequence(void)
 601 {
 602         eat('\\');
 603
 604         int ec = c;
 605         next_char();
 606
 607         switch(ec) {
 608         case '"':  return '"';
 609         case '\'': return '\'';
 610         case '\\': return '\\';
 611         case '?': return '\?';
 612         case 'a': return '\a';
 613         case 'b': return '\b';
 614         case 'f': return '\f';
 615         case 'n': return '\n';
 616         case 'r': return '\r';
 617         case 't': return '\t';
 618         case 'v': return '\v';
 619         case 'x':
 620                 return parse_hex_sequence();
 621         case '0':
 622         case '1':
 623         case '2':
 624         case '3':
 625         case '4':
 626         case '5':
 627         case '6':
 628         case '7':
 629                 return parse_octal_sequence(ec);
 630         case EOF:
 631                 parse_error("reached end of file while parsing escape sequence");
 632                 return EOF;
 633         default:
 634                 parse_error("unknown escape sequence");
 635                 return EOF;
 636         }
 637 }
 638
 639 const char *concat_strings(const char *s1, const char *s2)
 640 {
 641         size_t  len1   = strlen(s1);
 642         size_t  len2   = strlen(s2);
 643
 644         char   *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 645         memcpy(concat, s1, len1);
 646         memcpy(concat + len1, s2, len2 + 1);
 647
 648         const char *result = strset_insert(&stringset, concat);
 649         if(result != concat) {
 650                 obstack_free(&symbol_obstack, concat);
 651         }
 652
 653         return result;
 654 }
 655
 656 static void parse_string_literal(void)
 657 {
 658         unsigned    start_linenr = lexer_token.source_position.linenr;
 659         char       *string;
 660         const char *result;
 661
 662         assert(c == '"');
 663         next_char();
 664
 665         int tc;
 666         while(1) {
 667                 switch(c) {
 668                 case '\\':
 669                         tc = parse_escape_sequence();
 670                         obstack_1grow(&symbol_obstack, tc);
 671                         break;
 672
 673                 case EOF:
 674                         error_prefix_at(lexer_token.source_position.input_name,
 675                                         start_linenr);
 676                         fprintf(stderr, "string has no end\n");
 677                         lexer_token.type = T_ERROR;
 678                         return;
 679
 680                 case '"':
 681                         next_char();
 682                         goto end_of_string;
 683
 684                 default:
 685                         obstack_1grow(&symbol_obstack, c);
 686                         next_char();
 687                         break;
 688                 }
 689         }
 690
 691 end_of_string:
 692
 693         /* TODO: concatenate multiple strings separated by whitespace... */
 694
 695         /* add finishing 0 to the string */
 696         obstack_1grow(&symbol_obstack, '\0');
 697         string = obstack_finish(&symbol_obstack);
 698
 699         /* check if there is already a copy of the string */
 700         result = strset_insert(&stringset, string);
 701         if(result != string) {
 702                 obstack_free(&symbol_obstack, string);
 703         }
 704
 705         lexer_token.type     = T_STRING_LITERAL;
 706         lexer_token.v.string = result;
 707 }
 708
 709 static void parse_character_constant(void)
 710 {
 711         eat('\'');
 712
 713         int found_char = 0;
 714         while(1) {
 715                 switch(c) {
 716                 case '\\':
 717                         found_char = parse_escape_sequence();
 718                         break;
 719
 720                 MATCH_NEWLINE(
 721                         parse_error("newline while parsing character constant");
 722                         break;
 723                 )
 724
 725                 case '\'':
 726                         next_char();
 727                         goto end_of_char_constant;
 728
 729                 case EOF:
 730                         parse_error("EOF while parsing character constant");
 731                         lexer_token.type = T_ERROR;
 732                         return;
 733
 734                 default:
 735                         if(found_char != 0) {
 736                                 parse_error("more than 1 characters in character "
 737                                             "constant");
 738                                 goto end_of_char_constant;
 739                         } else {
 740                                 found_char = c;
 741                                 next_char();
 742                         }
 743                         break;
 744                 }
 745         }
 746
 747 end_of_char_constant:
 748         lexer_token.type       = T_INTEGER;
 749         lexer_token.v.intvalue = found_char;
 750 }
 751
 752 static void skip_multiline_comment(void)
 753 {
 754         unsigned start_linenr = lexer_token.source_position.linenr;
 755
 756         while(1) {
 757                 switch(c) {
 758                 case '*':
 759                         next_char();
 760                         if(c == '/') {
 761                                 next_char();
 762                                 return;
 763                         }
 764                         break;
 765
 766                 MATCH_NEWLINE(break;)
 767
 768                 case EOF:
 769                         error_prefix_at(lexer_token.source_position.input_name,
 770                                         start_linenr);
 771                         fprintf(stderr, "at end of file while looking for comment end\n");
 772                         return;
 773
 774                 default:
 775                         next_char();
 776                         break;
 777                 }
 778         }
 779 }
 780
 781 static void skip_line_comment(void)
 782 {
 783         while(1) {
 784                 switch(c) {
 785                 case EOF:
 786                         return;
 787
 788                 case '\n':
 789                 case '\r':
 790                         return;
 791
 792                 default:
 793                         next_char();
 794                         break;
 795                 }
 796         }
 797 }
 798
 799 static token_t pp_token;
 800
 801 static inline void next_pp_token(void)
 802 {
 803         lexer_next_preprocessing_token();
 804         pp_token = lexer_token;
 805 }
 806
 807 static void eat_until_newline(void)
 808 {
 809         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
 810                 next_pp_token();
 811         }
 812 }
 813
 814 static void error_directive(void)
 815 {
 816         error_prefix();
 817         fprintf(stderr, "#error directive: \n");
 818
 819         /* parse pp-tokens until new-line */
 820 }
 821
 822 static void define_directive(void)
 823 {
 824         lexer_next_preprocessing_token();
 825         if(lexer_token.type != T_IDENTIFIER) {
 826                 parse_error("expected identifier after #define\n");
 827                 eat_until_newline();
 828         }
 829 }
 830
 831 static void ifdef_directive(int is_ifndef)
 832 {
 833         (void) is_ifndef;
 834         lexer_next_preprocessing_token();
 835         //expect_identifier();
 836         //extect_newline();
 837 }
 838
 839 static void endif_directive(void)
 840 {
 841         //expect_newline();
 842 }
 843
 844 static void parse_line_directive(void)
 845 {
 846         if(pp_token.type != T_INTEGER) {
 847                 parse_error("expected integer");
 848         } else {
 849                 lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
 850                 next_pp_token();
 851         }
 852         if(pp_token.type == T_STRING_LITERAL) {
 853                 lexer_token.source_position.input_name = pp_token.v.string;
 854                 next_pp_token();
 855         }
 856
 857         eat_until_newline();
 858 }
 859
 860 static void parse_preprocessor_identifier(void)
 861 {
 862         assert(pp_token.type == T_IDENTIFIER);
 863         symbol_t *symbol = pp_token.v.symbol;
 864
 865         switch(symbol->pp_ID) {
 866         case TP_include:
 867                 printf("include - enable header name parsing!\n");
 868                 break;
 869         case TP_define:
 870                 define_directive();
 871                 break;
 872         case TP_ifdef:
 873                 ifdef_directive(0);
 874                 break;
 875         case TP_ifndef:
 876                 ifdef_directive(1);
 877                 break;
 878         case TP_endif:
 879                 endif_directive();
 880                 break;
 881         case TP_line:
 882                 next_pp_token();
 883                 parse_line_directive();
 884                 break;
 885         case TP_if:
 886         case TP_else:
 887         case TP_elif:
 888         case TP_undef:
 889         case TP_error:
 890                 error_directive();
 891                 break;
 892         case TP_pragma:
 893                 break;
 894         }
 895 }
 896
 897 static void parse_preprocessor_directive(void)
 898 {
 899         next_pp_token();
 900
 901         switch(pp_token.type) {
 902         case T_IDENTIFIER:
 903                 parse_preprocessor_identifier();
 904                 break;
 905         case T_INTEGER:
 906                 parse_line_directive();
 907                 break;
 908         default:
 909                 parse_error("invalid preprocessor directive");
 910                 eat_until_newline();
 911                 break;
 912         }
 913 }
 914
 915 #define MAYBE_PROLOG                                       \
 916                         next_char();                                   \
 917                         while(1) {                                     \
 918                                 switch(c) {
 919
 920 #define MAYBE(ch, set_type)                                \
 921                                 case ch:                                   \
 922                                         next_char();                           \
 923                                         lexer_token.type = set_type;           \
 924                                         return;
 925
 926 #define ELSE_CODE(code)                                    \
 927                                 default:                                   \
 928                                         code;                                  \
 929                                 }                                          \
 930                         } /* end of while(1) */                        \
 931                         break;
 932
 933 #define ELSE(set_type)                                     \
 934                 ELSE_CODE(                                         \
 935                         lexer_token.type = set_type;                   \
 936                         return;                                        \
 937                 )
 938
 939 void lexer_next_preprocessing_token(void)
 940 {
 941         while(1) {
 942                 switch(c) {
 943                 case ' ':
 944                 case '\t':
 945                         next_char();
 946                         break;
 947
 948                 MATCH_NEWLINE(
 949                         lexer_token.type = '\n';
 950                         return;
 951                 )
 952
 953                 SYMBOL_CHARS
 954                         parse_symbol();
 955                         /* might be a wide string ( L"string" ) */
 956                         if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 957                            lexer_token.v.symbol == symbol_L)) {
 958                                 parse_string_literal();
 959                                 return;
 960                         }
 961                         return;
 962
 963                 DIGITS
 964                         parse_number();
 965                         return;
 966
 967                 case '"':
 968                         parse_string_literal();
 969                         return;
 970
 971                 case '\'':
 972                         parse_character_constant();
 973                         return;
 974
 975                 case '.':
 976                         MAYBE_PROLOG
 977                                 case '.':
 978                                         MAYBE_PROLOG
 979                                         MAYBE('.', T_DOTDOTDOT)
 980                                         ELSE_CODE(
 981                                                 put_back(c);
 982                                                 c = '.';
 983                                                 lexer_token.type = '.';
 984                                                 return;
 985                                         )
 986                         ELSE('.')
 987                 case '&':
 988                         MAYBE_PROLOG
 989                         MAYBE('&', T_ANDAND)
 990                         MAYBE('=', T_ANDEQUAL)
 991                         ELSE('&')
 992                 case '*':
 993                         MAYBE_PROLOG
 994                         MAYBE('=', T_ASTERISKEQUAL)
 995                         ELSE('*')
 996                 case '+':
 997                         MAYBE_PROLOG
 998                         MAYBE('+', T_PLUSPLUS)
 999                         MAYBE('=', T_PLUSEQUAL)
1000                         ELSE('+')
1001                 case '-':
1002                         MAYBE_PROLOG
1003                         MAYBE('>', T_MINUSGREATER)
1004                         MAYBE('-', T_MINUSMINUS)
1005                         MAYBE('=', T_MINUSEQUAL)
1006                         ELSE('-')
1007                 case '!':
1008                         MAYBE_PROLOG
1009                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
1010                         ELSE('!')
1011                 case '/':
1012                         MAYBE_PROLOG
1013                         MAYBE('=', T_SLASHEQUAL)
1014                                 case '*':
1015                                         next_char();
1016                                         skip_multiline_comment();
1017                                         lexer_next_preprocessing_token();
1018                                         return;
1019                                 case '/':
1020                                         next_char();
1021                                         skip_line_comment();
1022                                         lexer_next_preprocessing_token();
1023                                         return;
1024                         ELSE('/')
1025                 case '%':
1026                         MAYBE_PROLOG
1027                         MAYBE('>', T_PERCENTGREATER)
1028                         MAYBE('=', T_PERCENTEQUAL)
1029                                 case ':':
1030                                         MAYBE_PROLOG
1031                                                 case '%':
1032                                                         MAYBE_PROLOG
1033                                                         MAYBE(':', T_PERCENTCOLONPERCENTCOLON)
1034                                                         ELSE_CODE(
1035                                                                 put_back(c);
1036                                                                 c = '%';
1037                                                                 lexer_token.type = T_PERCENTCOLON;
1038                                                                 return;
1039                                                         )
1040                                         ELSE(T_PERCENTCOLON)
1041                         ELSE('%')
1042                 case '<':
1043                         MAYBE_PROLOG
1044                         MAYBE(':', T_LESSCOLON)
1045                         MAYBE('%', T_LESSPERCENT)
1046                         MAYBE('=', T_LESSEQUAL)
1047                                 case '<':
1048                                         MAYBE_PROLOG
1049                                         MAYBE('=', T_LESSLESSEQUAL)
1050                                         ELSE(T_LESSLESS)
1051                         ELSE('<')
1052                 case '>':
1053                         MAYBE_PROLOG
1054                         MAYBE('=', T_GREATEREQUAL)
1055                                 case '>':
1056                                         MAYBE_PROLOG
1057                                         MAYBE('=', T_GREATERGREATEREQUAL)
1058                                         ELSE(T_GREATERGREATER)
1059                         ELSE('>')
1060                 case '^':
1061                         MAYBE_PROLOG
1062                         MAYBE('=', T_CARETEQUAL)
1063                         ELSE('^')
1064                 case '|':
1065                         MAYBE_PROLOG
1066                         MAYBE('=', T_PIPEEQUAL)
1067                         MAYBE('|', T_PIPEPIPE)
1068                         ELSE('|')
1069                 case ':':
1070                         MAYBE_PROLOG
1071                         MAYBE('>', T_COLONGREATER)
1072                         ELSE(':')
1073                 case '=':
1074                         MAYBE_PROLOG
1075                         MAYBE('=', T_EQUALEQUAL)
1076                         ELSE('=')
1077                 case '#':
1078                         MAYBE_PROLOG
1079                         MAYBE('#', T_HASHHASH)
1080                         ELSE('#')
1081
1082                 case '?':
1083                 case '[':
1084                 case ']':
1085                 case '(':
1086                 case ')':
1087                 case '{':
1088                 case '}':
1089                 case '~':
1090                 case ';':
1091                 case ',':
1092                 case '\\':
1093                         lexer_token.type = c;
1094                         next_char();
1095                         return;
1096
1097                 case EOF:
1098                         lexer_token.type = T_EOF;
1099                         return;
1100
1101                 default:
1102                         next_char();
1103                         error_prefix();
1104                         fprintf(stderr, "unknown character '%c' found\n", c);
1105                         lexer_token.type = T_ERROR;
1106                         return;
1107                 }
1108         }
1109 }
1110
1111 void lexer_next_token(void)
1112 {
1113         lexer_next_preprocessing_token();
1114         if(lexer_token.type != '\n')
1115                 return;
1116
1117 newline_found:
1118         do {
1119                 lexer_next_preprocessing_token();
1120         } while(lexer_token.type == '\n');
1121
1122         if(lexer_token.type == '#') {
1123                 parse_preprocessor_directive();
1124                 goto newline_found;
1125         }
1126 }
1127
1128 void init_lexer(void)
1129 {
1130         strset_init(&stringset);
1131
1132         type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
1133         type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
1134         type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
1135         type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
1136         type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
1137                                           TYPE_QUALIFIER_CONST);
1138         type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
1139                                           TYPE_QUALIFIER_CONST);
1140
1141         type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
1142         type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
1143                                            TYPE_QUALIFIER_CONST);
1144         type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE,
1145                                            TYPE_QUALIFIER_CONST);
1146 }
1147
1148 void lexer_open_stream(FILE *stream, const char *input_name)
1149 {
1150         input                                  = stream;
1151         lexer_token.source_position.linenr     = 0;
1152         lexer_token.source_position.input_name = input_name;
1153
1154         symbol_L = symbol_table_insert("L");
1155
1156         /* place a virtual \n at the beginning so the lexer knows that we're
1157          * at the beginning of a line */
1158         c = '\n';
1159 }
1160
1161 void exit_lexer(void)
1162 {
1163         strset_destroy(&stringset);
1164 }
1165
1166 static __attribute__((unused))
1167 void dbg_pos(const source_position_t source_position)
1168 {
1169         fprintf(stdout, "%s:%d\n", source_position.input_name,
1170                 source_position.linenr);
1171         fflush(stdout);
1172 }