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 #define HAS_SIGNED_CHAR
  26 //#define HAS_UNSIGNED_CHAR
  27
  28 #if defined HAS_SIGNED_CHAR
  29 typedef signed char char_type;
  30 #elif defined HAS_UNSIGNED_CHAR
  31 typedef unsigned char char_type;
  32 #else
  33 #       error signedness of char not determined
  34 #endif
  35
  36 static int         c;
  37 token_t            lexer_token;
  38 symbol_t          *symbol_L;
  39 static FILE       *input;
  40 static char        buf[1024 + MAX_PUTBACK];
  41 static const char *bufend;
  42 static const char *bufpos;
  43 static strset_t    stringset;
  44
  45 static type_t     *type_int        = NULL;
  46 static type_t     *type_uint       = NULL;
  47 static type_t     *type_long       = NULL;
  48 static type_t     *type_ulong      = NULL;
  49 static type_t     *type_longlong   = NULL;
  50 static type_t     *type_ulonglong  = NULL;
  51 static type_t     *type_float      = NULL;
  52 static type_t     *type_double     = NULL;
  53 static type_t     *type_longdouble = NULL;
  54
  55 static void error_prefix_at(const char *input_name, unsigned linenr)
  56 {
  57         fprintf(stderr, "%s:%u: Error: ", input_name, linenr);
  58 }
  59
  60 static void error_prefix(void)
  61 {
  62         error_prefix_at(lexer_token.source_position.input_name,
  63                         lexer_token.source_position.linenr);
  64 }
  65
  66 static void parse_error(const char *msg)
  67 {
  68         error_prefix();
  69         fprintf(stderr, "%s\n", msg);
  70 }
  71
  72 static inline void next_real_char(void)
  73 {
  74         bufpos++;
  75         if(bufpos >= bufend) {
  76                 size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK,
  77                                  input);
  78                 if(s == 0) {
  79                         c = EOF;
  80                         return;
  81                 }
  82                 bufpos = buf + MAX_PUTBACK;
  83                 bufend = buf + MAX_PUTBACK + s;
  84         }
  85         c = *(bufpos);
  86 }
  87
  88 static inline void put_back(int pc)
  89 {
  90         assert(bufpos >= buf);
  91         //assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc);
  92
  93         char *p = buf + (bufpos - buf);
  94         *p = pc;
  95
  96         /* going backwards in the buffer is legal as long as it's not more often
  97          * than MAX_PUTBACK */
  98         bufpos--;
  99
 100 #ifdef DEBUG_CHARS
 101         printf("putback '%c'\n", pc);
 102 #endif
 103 }
 104
 105 static inline void next_char(void);
 106
 107 #define MATCH_NEWLINE(code)                   \
 108         case '\r':                                \
 109                 next_char();                          \
 110                 if(c == '\n') {                       \
 111                         next_char();                      \
 112                 }                                     \
 113                 lexer_token.source_position.linenr++; \
 114                 code;                                 \
 115         case '\n':                                \
 116                 next_char();                          \
 117                 lexer_token.source_position.linenr++; \
 118                 code;
 119
 120 #define eat(c_type)  do { assert(c == c_type); next_char(); } while(0)
 121
 122 static void maybe_concat_lines(void)
 123 {
 124         eat('\\');
 125
 126         switch(c) {
 127         MATCH_NEWLINE(return;)
 128
 129         default:
 130                 break;
 131         }
 132
 133         put_back(c);
 134         c = '\\';
 135 }
 136
 137 static inline void next_char(void)
 138 {
 139         next_real_char();
 140
 141         /* filter trigraphs */
 142         if(UNLIKELY(c == '\\')) {
 143                 maybe_concat_lines();
 144                 goto end_of_next_char;
 145         }
 146
 147         if(LIKELY(c != '?'))
 148                 goto end_of_next_char;
 149
 150         next_real_char();
 151         if(LIKELY(c != '?')) {
 152                 put_back(c);
 153                 c = '?';
 154                 goto end_of_next_char;
 155         }
 156
 157         next_real_char();
 158         switch(c) {
 159         case '=': c = '#'; break;
 160         case '(': c = '['; break;
 161         case '/': c = '\\'; maybe_concat_lines(); break;
 162         case ')': c = ']'; break;
 163         case '\'': c = '^'; break;
 164         case '<': c = '{'; break;
 165         case '!': c = '|'; break;
 166         case '>': c = '}'; break;
 167         case '-': c = '~'; break;
 168         default:
 169                 put_back('?');
 170                 put_back(c);
 171                 c = '?';
 172                 break;
 173         }
 174
 175 end_of_next_char:;
 176 #ifdef DEBUG_CHARS
 177         printf("nchar '%c'\n", c);
 178 #endif
 179 }
 180
 181 #define SYMBOL_CHARS  \
 182         case 'a':         \
 183         case 'b':         \
 184         case 'c':         \
 185         case 'd':         \
 186         case 'e':         \
 187         case 'f':         \
 188         case 'g':         \
 189         case 'h':         \
 190         case 'i':         \
 191         case 'j':         \
 192         case 'k':         \
 193         case 'l':         \
 194         case 'm':         \
 195         case 'n':         \
 196         case 'o':         \
 197         case 'p':         \
 198         case 'q':         \
 199         case 'r':         \
 200         case 's':         \
 201         case 't':         \
 202         case 'u':         \
 203         case 'v':         \
 204         case 'w':         \
 205         case 'x':         \
 206         case 'y':         \
 207         case 'z':         \
 208         case 'A':         \
 209         case 'B':         \
 210         case 'C':         \
 211         case 'D':         \
 212         case 'E':         \
 213         case 'F':         \
 214         case 'G':         \
 215         case 'H':         \
 216         case 'I':         \
 217         case 'J':         \
 218         case 'K':         \
 219         case 'L':         \
 220         case 'M':         \
 221         case 'N':         \
 222         case 'O':         \
 223         case 'P':         \
 224         case 'Q':         \
 225         case 'R':         \
 226         case 'S':         \
 227         case 'T':         \
 228         case 'U':         \
 229         case 'V':         \
 230         case 'W':         \
 231         case 'X':         \
 232         case 'Y':         \
 233         case 'Z':         \
 234         case '_':
 235
 236 #define DIGITS        \
 237         case '0':         \
 238         case '1':         \
 239         case '2':         \
 240         case '3':         \
 241         case '4':         \
 242         case '5':         \
 243         case '6':         \
 244         case '7':         \
 245         case '8':         \
 246         case '9':
 247
 248 static void parse_symbol(void)
 249 {
 250         symbol_t *symbol;
 251         char     *string;
 252
 253         obstack_1grow(&symbol_obstack, c);
 254         next_char();
 255
 256         while(1) {
 257                 switch(c) {
 258                 DIGITS
 259                 SYMBOL_CHARS
 260                         obstack_1grow(&symbol_obstack, c);
 261                         next_char();
 262                         break;
 263
 264                 default:
 265                         goto end_symbol;
 266                 }
 267         }
 268
 269 end_symbol:
 270         obstack_1grow(&symbol_obstack, '\0');
 271
 272         string = obstack_finish(&symbol_obstack);
 273         symbol = symbol_table_insert(string);
 274
 275         lexer_token.type     = symbol->ID;
 276         lexer_token.v.symbol = symbol;
 277
 278         if(symbol->string != string) {
 279                 obstack_free(&symbol_obstack, string);
 280         }
 281 }
 282
 283 static void parse_integer_suffix(void)
 284 {
 285         if(c == 'U' || c == 'u') {
 286                 next_char();
 287                 if(c == 'L' || c == 'l') {
 288                         next_char();
 289                         if(c == 'L' || c == 'l') {
 290                                 next_char();
 291                                 lexer_token.datatype = type_ulonglong;
 292                         } else {
 293                                 lexer_token.datatype = type_ulong;
 294                         }
 295                 } else {
 296                         lexer_token.datatype = type_uint;
 297                 }
 298         } else if(c == 'l' || c == 'L') {
 299                 next_char();
 300                 if(c == 'l' || c == 'L') {
 301                         next_char();
 302                         if(c == 'u' || c == 'U') {
 303                                 next_char();
 304                                 lexer_token.datatype = type_ulonglong;
 305                         } else {
 306                                 lexer_token.datatype = type_longlong;
 307                         }
 308                 } else if(c == 'u' || c == 'U') {
 309                         next_char();
 310                         lexer_token.datatype = type_ulong;
 311                 } else {
 312                         lexer_token.datatype = type_int;
 313                 }
 314         } else {
 315                 lexer_token.datatype = type_int;
 316         }
 317 }
 318
 319 static void parse_floating_suffix(void)
 320 {
 321         switch(c) {
 322         /* TODO: do something usefull with the suffixes... */
 323         case 'f':
 324         case 'F':
 325                 next_char();
 326                 lexer_token.datatype = type_float;
 327                 break;
 328         case 'l':
 329         case 'L':
 330                 next_char();
 331                 lexer_token.datatype = type_longdouble;
 332                 break;
 333         default:
 334                 lexer_token.datatype = type_double;
 335                 break;
 336         }
 337 }
 338
 339 /**
 340  * A replacement for strtoull. Only those parts needed for
 341  * our parser are implemented.
 342  */
 343 static unsigned long long parse_int_string(const char *s, const char **endptr, int base) {
 344         unsigned long long v = 0;
 345
 346         switch (base) {
 347         case 16:
 348                 for (;; ++s) {
 349                         /* check for overrun */
 350                         if (v >= 0x1000000000000000ULL)
 351                                 break;
 352                         switch (tolower(*s)) {
 353                         case '0': v <<= 4; break;
 354                         case '1': v <<= 4; v |= 0x1; break;
 355                         case '2': v <<= 4; v |= 0x2; break;
 356                         case '3': v <<= 4; v |= 0x3; break;
 357                         case '4': v <<= 4; v |= 0x4; break;
 358                         case '5': v <<= 4; v |= 0x5; break;
 359                         case '6': v <<= 4; v |= 0x6; break;
 360                         case '7': v <<= 4; v |= 0x7; break;
 361                         case '8': v <<= 4; v |= 0x8; break;
 362                         case '9': v <<= 4; v |= 0x9; break;
 363                         case 'a': v <<= 4; v |= 0xa; break;
 364                         case 'b': v <<= 4; v |= 0xb; break;
 365                         case 'c': v <<= 4; v |= 0xc; break;
 366                         case 'd': v <<= 4; v |= 0xd; break;
 367                         case 'e': v <<= 4; v |= 0xe; break;
 368                         case 'f': v <<= 4; v |= 0xf; break;
 369                         default:
 370                                 goto end;
 371                         }
 372                 }
 373                 break;
 374         case 8:
 375                 for (;; ++s) {
 376                         /* check for overrun */
 377                         if (v >= 0x2000000000000000ULL)
 378                                 break;
 379                         switch (tolower(*s)) {
 380                         case '0': v <<= 3; break;
 381                         case '1': v <<= 3; v |= 1; break;
 382                         case '2': v <<= 3; v |= 2; break;
 383                         case '3': v <<= 3; v |= 3; break;
 384                         case '4': v <<= 3; v |= 4; break;
 385                         case '5': v <<= 3; v |= 5; break;
 386                         case '6': v <<= 3; v |= 6; break;
 387                         case '7': v <<= 3; v |= 7; break;
 388                         default:
 389                                 goto end;
 390                         }
 391                 }
 392                 break;
 393         case 10:
 394                 for (;; ++s) {
 395                         /* check for overrun */
 396                         if (v > 0x1999999999999999ULL)
 397                                 break;
 398                         switch (tolower(*s)) {
 399                         case '0': v *= 10; break;
 400                         case '1': v *= 10; v += 1; break;
 401                         case '2': v *= 10; v += 2; break;
 402                         case '3': v *= 10; v += 3; break;
 403                         case '4': v *= 10; v += 4; break;
 404                         case '5': v *= 10; v += 5; break;
 405                         case '6': v *= 10; v += 6; break;
 406                         case '7': v *= 10; v += 7; break;
 407                         case '8': v *= 10; v += 8; break;
 408                         case '9': v *= 10; v += 9; break;
 409                         default:
 410                                 goto end;
 411                         }
 412                 }
 413                 break;
 414         default:
 415                 assert(0);
 416                 break;
 417         }
 418 end:
 419         *endptr = s;
 420         return v;
 421 }
 422
 423 static void parse_number_hex(void)
 424 {
 425         assert(c == 'x' || c == 'X');
 426         next_char();
 427
 428         while(isxdigit(c)) {
 429                 obstack_1grow(&symbol_obstack, c);
 430                 next_char();
 431         }
 432         obstack_1grow(&symbol_obstack, '\0');
 433         char *string = obstack_finish(&symbol_obstack);
 434
 435         if(c == '.' || c == 'p' || c == 'P') {
 436                 next_char();
 437                 panic("Hex floating point numbers not implemented yet");
 438         }
 439         if(*string == '\0') {
 440                 parse_error("invalid hex number");
 441                 lexer_token.type = T_ERROR;
 442         }
 443
 444         const char *endptr;
 445         lexer_token.type       = T_INTEGER;
 446         lexer_token.v.intvalue = parse_int_string(string, &endptr, 16);
 447         if(*endptr != '\0') {
 448                 parse_error("hex number literal too long");
 449         }
 450
 451         obstack_free(&symbol_obstack, string);
 452         parse_integer_suffix();
 453 }
 454
 455 static inline bool is_octal_digit(int chr)
 456 {
 457         return '0' <= chr && chr <= '7';
 458 }
 459
 460 static void parse_number_oct(void)
 461 {
 462         while(is_octal_digit(c)) {
 463                 obstack_1grow(&symbol_obstack, c);
 464                 next_char();
 465         }
 466         obstack_1grow(&symbol_obstack, '\0');
 467         char *string = obstack_finish(&symbol_obstack);
 468
 469         const char *endptr;
 470         lexer_token.type       = T_INTEGER;
 471         lexer_token.v.intvalue = parse_int_string(string, &endptr, 8);
 472         if(*endptr != '\0') {
 473                 parse_error("octal number literal too long");
 474         }
 475
 476         obstack_free(&symbol_obstack, string);
 477         parse_integer_suffix();
 478 }
 479
 480 static void parse_number_dec(void)
 481 {
 482         bool is_float = false;
 483         while(isdigit(c)) {
 484                 obstack_1grow(&symbol_obstack, c);
 485                 next_char();
 486         }
 487
 488         if(c == '.') {
 489                 obstack_1grow(&symbol_obstack, '.');
 490                 next_char();
 491
 492                 while(isdigit(c)) {
 493                         obstack_1grow(&symbol_obstack, c);
 494                         next_char();
 495                 }
 496                 is_float = true;
 497         }
 498         if(c == 'e' || c == 'E') {
 499                 obstack_1grow(&symbol_obstack, 'e');
 500                 next_char();
 501
 502                 if(c == '-' || c == '+') {
 503                         obstack_1grow(&symbol_obstack, c);
 504                         next_char();
 505                 }
 506
 507                 while(isdigit(c)) {
 508                         obstack_1grow(&symbol_obstack, c);
 509                         next_char();
 510                 }
 511                 is_float = true;
 512         }
 513
 514         obstack_1grow(&symbol_obstack, '\0');
 515         char *string = obstack_finish(&symbol_obstack);
 516
 517         if(is_float) {
 518                 char *endptr;
 519                 lexer_token.type         = T_FLOATINGPOINT;
 520                 lexer_token.v.floatvalue = strtold(string, &endptr);
 521
 522                 if(*endptr != '\0') {
 523                         parse_error("invalid number literal");
 524                 }
 525
 526                 parse_floating_suffix();
 527         } else {
 528                 const char *endptr;
 529                 lexer_token.type       = T_INTEGER;
 530                 lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 531
 532                 if(*endptr != '\0') {
 533                         parse_error("invalid number literal");
 534                 }
 535
 536                 parse_integer_suffix();
 537         }
 538         obstack_free(&symbol_obstack, string);
 539 }
 540
 541 static void parse_number(void)
 542 {
 543         if (c == '0') {
 544                 next_char();
 545                 switch (c) {
 546                         case 'X':
 547                         case 'x':
 548                                 parse_number_hex();
 549                                 break;
 550                         case '0':
 551                         case '1':
 552                         case '2':
 553                         case '3':
 554                         case '4':
 555                         case '5':
 556                         case '6':
 557                         case '7':
 558                                 parse_number_oct();
 559                                 break;
 560                         case '8':
 561                         case '9':
 562                                 next_char();
 563                                 parse_error("invalid octal number");
 564                                 lexer_token.type = T_ERROR;
 565                                 return;
 566                         case '.':
 567                         case 'e':
 568                         case 'E':
 569                         default:
 570                                 obstack_1grow(&symbol_obstack, '0');
 571                                 parse_number_dec();
 572                                 return;
 573                 }
 574         } else {
 575                 parse_number_dec();
 576         }
 577 }
 578
 579 static int parse_octal_sequence(const int first_digit)
 580 {
 581         assert(is_octal_digit(first_digit));
 582         int value = first_digit - '0';
 583         if (!is_octal_digit(c)) return value;
 584         value = 8 * value + c - '0';
 585         next_char();
 586         if (!is_octal_digit(c)) return value;
 587         value = 8 * value + c - '0';
 588         next_char();
 589         return (char_type)value;
 590 }
 591
 592 static int parse_hex_sequence(void)
 593 {
 594         int value = 0;
 595         while(1) {
 596                 if (c >= '0' && c <= '9') {
 597                         value = 16 * value + c - '0';
 598                 } else if ('A' <= c && c <= 'F') {
 599                         value = 16 * value + c - 'A' + 10;
 600                 } else if ('a' <= c && c <= 'f') {
 601                         value = 16 * value + c - 'a' + 10;
 602                 } else {
 603                         break;
 604                 }
 605                 next_char();
 606         }
 607
 608         return (char_type)value;
 609 }
 610
 611 static int parse_escape_sequence(void)
 612 {
 613         eat('\\');
 614
 615         int ec = c;
 616         next_char();
 617
 618         switch(ec) {
 619         case '"':  return '"';
 620         case '\'': return '\'';
 621         case '\\': return '\\';
 622         case '?': return '\?';
 623         case 'a': return '\a';
 624         case 'b': return '\b';
 625         case 'f': return '\f';
 626         case 'n': return '\n';
 627         case 'r': return '\r';
 628         case 't': return '\t';
 629         case 'v': return '\v';
 630         case 'x':
 631                 return parse_hex_sequence();
 632         case '0':
 633         case '1':
 634         case '2':
 635         case '3':
 636         case '4':
 637         case '5':
 638         case '6':
 639         case '7':
 640                 return parse_octal_sequence(ec);
 641         case EOF:
 642                 parse_error("reached end of file while parsing escape sequence");
 643                 return EOF;
 644         default:
 645                 parse_error("unknown escape sequence");
 646                 return EOF;
 647         }
 648 }
 649
 650 const char *concat_strings(const char *s1, const char *s2)
 651 {
 652         size_t  len1   = strlen(s1);
 653         size_t  len2   = strlen(s2);
 654
 655         char   *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 656         memcpy(concat, s1, len1);
 657         memcpy(concat + len1, s2, len2 + 1);
 658
 659         const char *result = strset_insert(&stringset, concat);
 660         if(result != concat) {
 661                 obstack_free(&symbol_obstack, concat);
 662         }
 663
 664         return result;
 665 }
 666
 667 static void parse_string_literal(void)
 668 {
 669         unsigned    start_linenr = lexer_token.source_position.linenr;
 670         char       *string;
 671         const char *result;
 672
 673         assert(c == '"');
 674         next_char();
 675
 676         int tc;
 677         while(1) {
 678                 switch(c) {
 679                 case '\\':
 680                         tc = parse_escape_sequence();
 681                         obstack_1grow(&symbol_obstack, tc);
 682                         break;
 683
 684                 case EOF:
 685                         error_prefix_at(lexer_token.source_position.input_name,
 686                                         start_linenr);
 687                         fprintf(stderr, "string has no end\n");
 688                         lexer_token.type = T_ERROR;
 689                         return;
 690
 691                 case '"':
 692                         next_char();
 693                         goto end_of_string;
 694
 695                 default:
 696                         obstack_1grow(&symbol_obstack, c);
 697                         next_char();
 698                         break;
 699                 }
 700         }
 701
 702 end_of_string:
 703
 704         /* TODO: concatenate multiple strings separated by whitespace... */
 705
 706         /* add finishing 0 to the string */
 707         obstack_1grow(&symbol_obstack, '\0');
 708         string = obstack_finish(&symbol_obstack);
 709
 710         /* check if there is already a copy of the string */
 711         result = strset_insert(&stringset, string);
 712         if(result != string) {
 713                 obstack_free(&symbol_obstack, string);
 714         }
 715
 716         lexer_token.type     = T_STRING_LITERAL;
 717         lexer_token.v.string = result;
 718 }
 719
 720 static void parse_character_constant(void)
 721 {
 722         eat('\'');
 723
 724         int found_char = 0;
 725         while(1) {
 726                 switch(c) {
 727                 case '\\':
 728                         found_char = parse_escape_sequence();
 729                         break;
 730
 731                 MATCH_NEWLINE(
 732                         parse_error("newline while parsing character constant");
 733                         break;
 734                 )
 735
 736                 case '\'':
 737                         next_char();
 738                         goto end_of_char_constant;
 739
 740                 case EOF:
 741                         parse_error("EOF while parsing character constant");
 742                         lexer_token.type = T_ERROR;
 743                         return;
 744
 745                 default:
 746                         if(found_char != 0) {
 747                                 parse_error("more than 1 characters in character "
 748                                             "constant");
 749                                 goto end_of_char_constant;
 750                         } else {
 751                                 found_char = c;
 752                                 next_char();
 753                         }
 754                         break;
 755                 }
 756         }
 757
 758 end_of_char_constant:
 759         lexer_token.type       = T_INTEGER;
 760         lexer_token.v.intvalue = found_char;
 761 }
 762
 763 static void skip_multiline_comment(void)
 764 {
 765         unsigned start_linenr = lexer_token.source_position.linenr;
 766
 767         while(1) {
 768                 switch(c) {
 769                 case '*':
 770                         next_char();
 771                         if(c == '/') {
 772                                 next_char();
 773                                 return;
 774                         }
 775                         break;
 776
 777                 MATCH_NEWLINE(break;)
 778
 779                 case EOF:
 780                         error_prefix_at(lexer_token.source_position.input_name,
 781                                         start_linenr);
 782                         fprintf(stderr, "at end of file while looking for comment end\n");
 783                         return;
 784
 785                 default:
 786                         next_char();
 787                         break;
 788                 }
 789         }
 790 }
 791
 792 static void skip_line_comment(void)
 793 {
 794         while(1) {
 795                 switch(c) {
 796                 case EOF:
 797                         return;
 798
 799                 case '\n':
 800                 case '\r':
 801                         return;
 802
 803                 default:
 804                         next_char();
 805                         break;
 806                 }
 807         }
 808 }
 809
 810 static token_t pp_token;
 811
 812 static inline void next_pp_token(void)
 813 {
 814         lexer_next_preprocessing_token();
 815         pp_token = lexer_token;
 816 }
 817
 818 static void eat_until_newline(void)
 819 {
 820         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
 821                 next_pp_token();
 822         }
 823 }
 824
 825 static void error_directive(void)
 826 {
 827         error_prefix();
 828         fprintf(stderr, "#error directive: \n");
 829
 830         /* parse pp-tokens until new-line */
 831 }
 832
 833 static void define_directive(void)
 834 {
 835         lexer_next_preprocessing_token();
 836         if(lexer_token.type != T_IDENTIFIER) {
 837                 parse_error("expected identifier after #define\n");
 838                 eat_until_newline();
 839         }
 840 }
 841
 842 static void ifdef_directive(int is_ifndef)
 843 {
 844         (void) is_ifndef;
 845         lexer_next_preprocessing_token();
 846         //expect_identifier();
 847         //extect_newline();
 848 }
 849
 850 static void endif_directive(void)
 851 {
 852         //expect_newline();
 853 }
 854
 855 static void parse_line_directive(void)
 856 {
 857         if(pp_token.type != T_INTEGER) {
 858                 parse_error("expected integer");
 859         } else {
 860                 lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
 861                 next_pp_token();
 862         }
 863         if(pp_token.type == T_STRING_LITERAL) {
 864                 lexer_token.source_position.input_name = pp_token.v.string;
 865                 next_pp_token();
 866         }
 867
 868         eat_until_newline();
 869 }
 870
 871 static void parse_preprocessor_identifier(void)
 872 {
 873         assert(pp_token.type == T_IDENTIFIER);
 874         symbol_t *symbol = pp_token.v.symbol;
 875
 876         switch(symbol->pp_ID) {
 877         case TP_include:
 878                 printf("include - enable header name parsing!\n");
 879                 break;
 880         case TP_define:
 881                 define_directive();
 882                 break;
 883         case TP_ifdef:
 884                 ifdef_directive(0);
 885                 break;
 886         case TP_ifndef:
 887                 ifdef_directive(1);
 888                 break;
 889         case TP_endif:
 890                 endif_directive();
 891                 break;
 892         case TP_line:
 893                 next_pp_token();
 894                 parse_line_directive();
 895                 break;
 896         case TP_if:
 897         case TP_else:
 898         case TP_elif:
 899         case TP_undef:
 900         case TP_error:
 901                 error_directive();
 902                 break;
 903         case TP_pragma:
 904                 break;
 905         }
 906 }
 907
 908 static void parse_preprocessor_directive(void)
 909 {
 910         next_pp_token();
 911
 912         switch(pp_token.type) {
 913         case T_IDENTIFIER:
 914                 parse_preprocessor_identifier();
 915                 break;
 916         case T_INTEGER:
 917                 parse_line_directive();
 918                 break;
 919         default:
 920                 parse_error("invalid preprocessor directive");
 921                 eat_until_newline();
 922                 break;
 923         }
 924 }
 925
 926 #define MAYBE_PROLOG                                       \
 927                         next_char();                                   \
 928                         while(1) {                                     \
 929                                 switch(c) {
 930
 931 #define MAYBE(ch, set_type)                                \
 932                                 case ch:                                   \
 933                                         next_char();                           \
 934                                         lexer_token.type = set_type;           \
 935                                         return;
 936
 937 #define ELSE_CODE(code)                                    \
 938                                 default:                                   \
 939                                         code;                                  \
 940                                 }                                          \
 941                         } /* end of while(1) */                        \
 942                         break;
 943
 944 #define ELSE(set_type)                                     \
 945                 ELSE_CODE(                                         \
 946                         lexer_token.type = set_type;                   \
 947                         return;                                        \
 948                 )
 949
 950 void lexer_next_preprocessing_token(void)
 951 {
 952         while(1) {
 953                 switch(c) {
 954                 case ' ':
 955                 case '\t':
 956                         next_char();
 957                         break;
 958
 959                 MATCH_NEWLINE(
 960                         lexer_token.type = '\n';
 961                         return;
 962                 )
 963
 964                 SYMBOL_CHARS
 965                         parse_symbol();
 966                         /* might be a wide string ( L"string" ) */
 967                         if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 968                            lexer_token.v.symbol == symbol_L)) {
 969                                 parse_string_literal();
 970                                 return;
 971                         }
 972                         return;
 973
 974                 DIGITS
 975                         parse_number();
 976                         return;
 977
 978                 case '"':
 979                         parse_string_literal();
 980                         return;
 981
 982                 case '\'':
 983                         parse_character_constant();
 984                         return;
 985
 986                 case '.':
 987                         MAYBE_PROLOG
 988                                 case '.':
 989                                         MAYBE_PROLOG
 990                                         MAYBE('.', T_DOTDOTDOT)
 991                                         ELSE_CODE(
 992                                                 put_back(c);
 993                                                 c = '.';
 994                                                 lexer_token.type = '.';
 995                                                 return;
 996                                         )
 997                         ELSE('.')
 998                 case '&':
 999                         MAYBE_PROLOG
1000                         MAYBE('&', T_ANDAND)
1001                         MAYBE('=', T_ANDEQUAL)
1002                         ELSE('&')
1003                 case '*':
1004                         MAYBE_PROLOG
1005                         MAYBE('=', T_ASTERISKEQUAL)
1006                         ELSE('*')
1007                 case '+':
1008                         MAYBE_PROLOG
1009                         MAYBE('+', T_PLUSPLUS)
1010                         MAYBE('=', T_PLUSEQUAL)
1011                         ELSE('+')
1012                 case '-':
1013                         MAYBE_PROLOG
1014                         MAYBE('>', T_MINUSGREATER)
1015                         MAYBE('-', T_MINUSMINUS)
1016                         MAYBE('=', T_MINUSEQUAL)
1017                         ELSE('-')
1018                 case '!':
1019                         MAYBE_PROLOG
1020                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
1021                         ELSE('!')
1022                 case '/':
1023                         MAYBE_PROLOG
1024                         MAYBE('=', T_SLASHEQUAL)
1025                                 case '*':
1026                                         next_char();
1027                                         skip_multiline_comment();
1028                                         lexer_next_preprocessing_token();
1029                                         return;
1030                                 case '/':
1031                                         next_char();
1032                                         skip_line_comment();
1033                                         lexer_next_preprocessing_token();
1034                                         return;
1035                         ELSE('/')
1036                 case '%':
1037                         MAYBE_PROLOG
1038                         MAYBE('>', T_PERCENTGREATER)
1039                         MAYBE('=', T_PERCENTEQUAL)
1040                                 case ':':
1041                                         MAYBE_PROLOG
1042                                                 case '%':
1043                                                         MAYBE_PROLOG
1044                                                         MAYBE(':', T_PERCENTCOLONPERCENTCOLON)
1045                                                         ELSE_CODE(
1046                                                                 put_back(c);
1047                                                                 c = '%';
1048                                                                 lexer_token.type = T_PERCENTCOLON;
1049                                                                 return;
1050                                                         )
1051                                         ELSE(T_PERCENTCOLON)
1052                         ELSE('%')
1053                 case '<':
1054                         MAYBE_PROLOG
1055                         MAYBE(':', T_LESSCOLON)
1056                         MAYBE('%', T_LESSPERCENT)
1057                         MAYBE('=', T_LESSEQUAL)
1058                                 case '<':
1059                                         MAYBE_PROLOG
1060                                         MAYBE('=', T_LESSLESSEQUAL)
1061                                         ELSE(T_LESSLESS)
1062                         ELSE('<')
1063                 case '>':
1064                         MAYBE_PROLOG
1065                         MAYBE('=', T_GREATEREQUAL)
1066                                 case '>':
1067                                         MAYBE_PROLOG
1068                                         MAYBE('=', T_GREATERGREATEREQUAL)
1069                                         ELSE(T_GREATERGREATER)
1070                         ELSE('>')
1071                 case '^':
1072                         MAYBE_PROLOG
1073                         MAYBE('=', T_CARETEQUAL)
1074                         ELSE('^')
1075                 case '|':
1076                         MAYBE_PROLOG
1077                         MAYBE('=', T_PIPEEQUAL)
1078                         MAYBE('|', T_PIPEPIPE)
1079                         ELSE('|')
1080                 case ':':
1081                         MAYBE_PROLOG
1082                         MAYBE('>', T_COLONGREATER)
1083                         ELSE(':')
1084                 case '=':
1085                         MAYBE_PROLOG
1086                         MAYBE('=', T_EQUALEQUAL)
1087                         ELSE('=')
1088                 case '#':
1089                         MAYBE_PROLOG
1090                         MAYBE('#', T_HASHHASH)
1091                         ELSE('#')
1092
1093                 case '?':
1094                 case '[':
1095                 case ']':
1096                 case '(':
1097                 case ')':
1098                 case '{':
1099                 case '}':
1100                 case '~':
1101                 case ';':
1102                 case ',':
1103                 case '\\':
1104                         lexer_token.type = c;
1105                         next_char();
1106                         return;
1107
1108                 case EOF:
1109                         lexer_token.type = T_EOF;
1110                         return;
1111
1112                 default:
1113                         next_char();
1114                         error_prefix();
1115                         fprintf(stderr, "unknown character '%c' found\n", c);
1116                         lexer_token.type = T_ERROR;
1117                         return;
1118                 }
1119         }
1120 }
1121
1122 void lexer_next_token(void)
1123 {
1124         lexer_next_preprocessing_token();
1125         if(lexer_token.type != '\n')
1126                 return;
1127
1128 newline_found:
1129         do {
1130                 lexer_next_preprocessing_token();
1131         } while(lexer_token.type == '\n');
1132
1133         if(lexer_token.type == '#') {
1134                 parse_preprocessor_directive();
1135                 goto newline_found;
1136         }
1137 }
1138
1139 void init_lexer(void)
1140 {
1141         strset_init(&stringset);
1142
1143         type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
1144         type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
1145         type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
1146         type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
1147         type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
1148                                           TYPE_QUALIFIER_CONST);
1149         type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
1150                                           TYPE_QUALIFIER_CONST);
1151
1152         type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
1153         type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
1154                                            TYPE_QUALIFIER_CONST);
1155         type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE,
1156                                            TYPE_QUALIFIER_CONST);
1157 }
1158
1159 void lexer_open_stream(FILE *stream, const char *input_name)
1160 {
1161         input                                  = stream;
1162         lexer_token.source_position.linenr     = 0;
1163         lexer_token.source_position.input_name = input_name;
1164
1165         symbol_L = symbol_table_insert("L");
1166
1167         /* place a virtual \n at the beginning so the lexer knows that we're
1168          * at the beginning of a line */
1169         c = '\n';
1170 }
1171
1172 void exit_lexer(void)
1173 {
1174         strset_destroy(&stringset);
1175 }
1176
1177 static __attribute__((unused))
1178 void dbg_pos(const source_position_t source_position)
1179 {
1180         fprintf(stdout, "%s:%d\n", source_position.input_name,
1181                 source_position.linenr);
1182         fflush(stdout);
1183 }