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