nsz Git - cparser/blob - lexer.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2008 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <config.h>
  21
  22 #include "diagnostic.h"
  23 #include "lexer.h"
  24 #include "symbol_t.h"
  25 #include "token_t.h"
  26 #include "symbol_table_t.h"
  27 #include "adt/error.h"
  28 #include "adt/strset.h"
  29 #include "adt/util.h"
  30 #include "types.h"
  31 #include "type_t.h"
  32 #include "target_architecture.h"
  33 #include "parser.h"
  34 #include "warning.h"
  35 #include "lang_features.h"
  36
  37 #include <assert.h>
  38 #include <errno.h>
  39 #include <string.h>
  40 #include <stdbool.h>
  41 #include <ctype.h>
  42
  43 //#define DEBUG_CHARS
  44 #define MAX_PUTBACK 3
  45
  46 #ifdef _WIN32
  47 /* No strtold on windows and no replacement yet */
  48 #define strtold(s, e) strtod(s, e)
  49 #endif
  50
  51 static int         c;
  52 token_t            lexer_token;
  53 symbol_t          *symbol_L;
  54 static FILE       *input;
  55 static char        buf[1024 + MAX_PUTBACK];
  56 static const char *bufend;
  57 static const char *bufpos;
  58 static strset_t    stringset;
  59 bool               allow_dollar_in_symbol = true;
  60
  61 /**
  62  * Prints a parse error message at the current token.
  63  *
  64  * @param msg   the error message
  65  */
  66 static void parse_error(const char *msg)
  67 {
  68         errorf(&lexer_token.source_position,  "%s", msg);
  69 }
  70
  71 /**
  72  * Prints an internal error message at the current token.
  73  *
  74  * @param msg   the error message
  75  */
  76 static NORETURN internal_error(const char *msg)
  77 {
  78         internal_errorf(&lexer_token.source_position,  "%s", msg);
  79 }
  80
  81 static inline void next_real_char(void)
  82 {
  83         assert(bufpos <= bufend);
  84         if (bufpos >= bufend) {
  85                 if (input == NULL) {
  86                         c = EOF;
  87                         return;
  88                 }
  89
  90                 size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK,
  91                                  input);
  92                 if(s == 0) {
  93                         c = EOF;
  94                         return;
  95                 }
  96                 bufpos = buf + MAX_PUTBACK;
  97                 bufend = buf + MAX_PUTBACK + s;
  98         }
  99         c = *bufpos++;
 100 }
 101
 102 /**
 103  * Put a character back into the buffer.
 104  *
 105  * @param pc  the character to put back
 106  */
 107 static inline void put_back(int pc)
 108 {
 109         assert(bufpos > buf);
 110         *(--bufpos - buf + buf) = (char) pc;
 111
 112 #ifdef DEBUG_CHARS
 113         printf("putback '%c'\n", pc);
 114 #endif
 115 }
 116
 117 static inline void next_char(void);
 118
 119 #define MATCH_NEWLINE(code)                   \
 120         case '\r':                                \
 121                 next_char();                          \
 122                 if(c == '\n') {                       \
 123                         next_char();                      \
 124                 }                                     \
 125                 lexer_token.source_position.linenr++; \
 126                 code                                  \
 127         case '\n':                                \
 128                 next_char();                          \
 129                 lexer_token.source_position.linenr++; \
 130                 code
 131
 132 #define eat(c_type)  do { assert(c == c_type); next_char(); } while(0)
 133
 134 static void maybe_concat_lines(void)
 135 {
 136         eat('\\');
 137
 138         switch(c) {
 139         MATCH_NEWLINE(return;)
 140
 141         default:
 142                 break;
 143         }
 144
 145         put_back(c);
 146         c = '\\';
 147 }
 148
 149 /**
 150  * Set c to the next input character, ie.
 151  * after expanding trigraphs.
 152  */
 153 static inline void next_char(void)
 154 {
 155         next_real_char();
 156
 157         /* filter trigraphs */
 158         if(UNLIKELY(c == '\\')) {
 159                 maybe_concat_lines();
 160                 goto end_of_next_char;
 161         }
 162
 163         if(LIKELY(c != '?'))
 164                 goto end_of_next_char;
 165
 166         next_real_char();
 167         if(LIKELY(c != '?')) {
 168                 put_back(c);
 169                 c = '?';
 170                 goto end_of_next_char;
 171         }
 172
 173         next_real_char();
 174         switch(c) {
 175         case '=': c = '#'; break;
 176         case '(': c = '['; break;
 177         case '/': c = '\\'; maybe_concat_lines(); break;
 178         case ')': c = ']'; break;
 179         case '\'': c = '^'; break;
 180         case '<': c = '{'; break;
 181         case '!': c = '|'; break;
 182         case '>': c = '}'; break;
 183         case '-': c = '~'; break;
 184         default:
 185                 put_back(c);
 186                 put_back('?');
 187                 c = '?';
 188                 break;
 189         }
 190
 191 end_of_next_char:;
 192 #ifdef DEBUG_CHARS
 193         printf("nchar '%c'\n", c);
 194 #endif
 195 }
 196
 197 #define SYMBOL_CHARS  \
 198         case '$': if (!allow_dollar_in_symbol) goto dollar_sign; \
 199         case 'a':         \
 200         case 'b':         \
 201         case 'c':         \
 202         case 'd':         \
 203         case 'e':         \
 204         case 'f':         \
 205         case 'g':         \
 206         case 'h':         \
 207         case 'i':         \
 208         case 'j':         \
 209         case 'k':         \
 210         case 'l':         \
 211         case 'm':         \
 212         case 'n':         \
 213         case 'o':         \
 214         case 'p':         \
 215         case 'q':         \
 216         case 'r':         \
 217         case 's':         \
 218         case 't':         \
 219         case 'u':         \
 220         case 'v':         \
 221         case 'w':         \
 222         case 'x':         \
 223         case 'y':         \
 224         case 'z':         \
 225         case 'A':         \
 226         case 'B':         \
 227         case 'C':         \
 228         case 'D':         \
 229         case 'E':         \
 230         case 'F':         \
 231         case 'G':         \
 232         case 'H':         \
 233         case 'I':         \
 234         case 'J':         \
 235         case 'K':         \
 236         case 'L':         \
 237         case 'M':         \
 238         case 'N':         \
 239         case 'O':         \
 240         case 'P':         \
 241         case 'Q':         \
 242         case 'R':         \
 243         case 'S':         \
 244         case 'T':         \
 245         case 'U':         \
 246         case 'V':         \
 247         case 'W':         \
 248         case 'X':         \
 249         case 'Y':         \
 250         case 'Z':         \
 251         case '_':
 252
 253 #define DIGITS        \
 254         case '0':         \
 255         case '1':         \
 256         case '2':         \
 257         case '3':         \
 258         case '4':         \
 259         case '5':         \
 260         case '6':         \
 261         case '7':         \
 262         case '8':         \
 263         case '9':
 264
 265 /**
 266  * Read a symbol from the input and build
 267  * the lexer_token.
 268  */
 269 static void parse_symbol(void)
 270 {
 271         symbol_t *symbol;
 272         char     *string;
 273
 274         obstack_1grow(&symbol_obstack, (char) c);
 275         next_char();
 276
 277         while(1) {
 278                 switch(c) {
 279                 DIGITS
 280                 SYMBOL_CHARS
 281                         obstack_1grow(&symbol_obstack, (char) c);
 282                         next_char();
 283                         break;
 284
 285                 default:
 286 dollar_sign:
 287                         goto end_symbol;
 288                 }
 289         }
 290
 291 end_symbol:
 292         obstack_1grow(&symbol_obstack, '\0');
 293
 294         string = obstack_finish(&symbol_obstack);
 295         symbol = symbol_table_insert(string);
 296
 297         lexer_token.type     = symbol->ID;
 298         lexer_token.v.symbol = symbol;
 299
 300         if(symbol->string != string) {
 301                 obstack_free(&symbol_obstack, string);
 302         }
 303 }
 304
 305 static void parse_integer_suffix(bool is_oct_hex)
 306 {
 307         bool is_unsigned  = false;
 308         bool min_long     = false;
 309         bool min_longlong = false;
 310
 311         if(c == 'U' || c == 'u') {
 312                 is_unsigned = true;
 313                 next_char();
 314                 if(c == 'L' || c == 'l') {
 315                         min_long = true;
 316                         next_char();
 317                         if(c == 'L' || c == 'l') {
 318                                 min_longlong = true;
 319                                 next_char();
 320                         }
 321                 }
 322         } else if(c == 'l' || c == 'L') {
 323                 min_long = true;
 324                 next_char();
 325                 if(c == 'l' || c == 'L') {
 326                         min_longlong = true;
 327                         next_char();
 328                         if(c == 'u' || c == 'U') {
 329                                 is_unsigned = true;
 330                                 next_char();
 331                         }
 332                 } else if(c == 'u' || c == 'U') {
 333                         is_unsigned = true;
 334                         next_char();
 335                         lexer_token.datatype = type_unsigned_long;
 336                 }
 337         }
 338
 339         if(!is_unsigned) {
 340                 long long v = lexer_token.v.intvalue;
 341                 if(!min_long) {
 342                         if(v >= TARGET_INT_MIN && v <= TARGET_INT_MAX) {
 343                                 lexer_token.datatype = type_int;
 344                                 return;
 345                         } else if(is_oct_hex && v >= 0 && v <= TARGET_UINT_MAX) {
 346                                 lexer_token.datatype = type_unsigned_int;
 347                                 return;
 348                         }
 349                 }
 350                 if(!min_longlong) {
 351                         if(v >= TARGET_LONG_MIN && v <= TARGET_LONG_MAX) {
 352                                 lexer_token.datatype = type_long;
 353                                 return;
 354                         } else if(is_oct_hex && v >= 0 && (unsigned long long)v <= (unsigned long long)TARGET_ULONG_MAX) {
 355                                 lexer_token.datatype = type_unsigned_long;
 356                                 return;
 357                         }
 358                 }
 359                 unsigned long long uv = (unsigned long long) v;
 360                 if(is_oct_hex && uv > (unsigned long long) TARGET_LONGLONG_MAX) {
 361                         lexer_token.datatype = type_unsigned_long_long;
 362                         return;
 363                 }
 364
 365                 lexer_token.datatype = type_long_long;
 366         } else {
 367                 unsigned long long v = (unsigned long long) lexer_token.v.intvalue;
 368                 if(!min_long && v <= TARGET_UINT_MAX) {
 369                         lexer_token.datatype = type_unsigned_int;
 370                         return;
 371                 }
 372                 if(!min_longlong && v <= TARGET_ULONG_MAX) {
 373                         lexer_token.datatype = type_unsigned_long;
 374                         return;
 375                 }
 376                 lexer_token.datatype = type_unsigned_long_long;
 377         }
 378 }
 379
 380 static void parse_floating_suffix(void)
 381 {
 382         switch(c) {
 383         /* TODO: do something useful with the suffixes... */
 384         case 'f':
 385         case 'F':
 386                 next_char();
 387                 lexer_token.datatype = type_float;
 388                 break;
 389         case 'l':
 390         case 'L':
 391                 next_char();
 392                 lexer_token.datatype = type_long_double;
 393                 break;
 394         default:
 395                 lexer_token.datatype = type_double;
 396                 break;
 397         }
 398 }
 399
 400 /**
 401  * A replacement for strtoull. Only those parts needed for
 402  * our parser are implemented.
 403  */
 404 static unsigned long long parse_int_string(const char *s, const char **endptr, int base) {
 405         unsigned long long v = 0;
 406
 407         switch (base) {
 408         case 16:
 409                 for (;; ++s) {
 410                         /* check for overrun */
 411                         if (v >= 0x1000000000000000ULL)
 412                                 break;
 413                         switch (tolower(*s)) {
 414                         case '0': v <<= 4; break;
 415                         case '1': v <<= 4; v |= 0x1; break;
 416                         case '2': v <<= 4; v |= 0x2; break;
 417                         case '3': v <<= 4; v |= 0x3; break;
 418                         case '4': v <<= 4; v |= 0x4; break;
 419                         case '5': v <<= 4; v |= 0x5; break;
 420                         case '6': v <<= 4; v |= 0x6; break;
 421                         case '7': v <<= 4; v |= 0x7; break;
 422                         case '8': v <<= 4; v |= 0x8; break;
 423                         case '9': v <<= 4; v |= 0x9; break;
 424                         case 'a': v <<= 4; v |= 0xa; break;
 425                         case 'b': v <<= 4; v |= 0xb; break;
 426                         case 'c': v <<= 4; v |= 0xc; break;
 427                         case 'd': v <<= 4; v |= 0xd; break;
 428                         case 'e': v <<= 4; v |= 0xe; break;
 429                         case 'f': v <<= 4; v |= 0xf; break;
 430                         default:
 431                                 goto end;
 432                         }
 433                 }
 434                 break;
 435         case 8:
 436                 for (;; ++s) {
 437                         /* check for overrun */
 438                         if (v >= 0x2000000000000000ULL)
 439                                 break;
 440                         switch (tolower(*s)) {
 441                         case '0': v <<= 3; break;
 442                         case '1': v <<= 3; v |= 1; break;
 443                         case '2': v <<= 3; v |= 2; break;
 444                         case '3': v <<= 3; v |= 3; break;
 445                         case '4': v <<= 3; v |= 4; break;
 446                         case '5': v <<= 3; v |= 5; break;
 447                         case '6': v <<= 3; v |= 6; break;
 448                         case '7': v <<= 3; v |= 7; break;
 449                         default:
 450                                 goto end;
 451                         }
 452                 }
 453                 break;
 454         case 10:
 455                 for (;; ++s) {
 456                         /* check for overrun */
 457                         if (v > 0x1999999999999999ULL)
 458                                 break;
 459                         switch (tolower(*s)) {
 460                         case '0': v *= 10; break;
 461                         case '1': v *= 10; v += 1; break;
 462                         case '2': v *= 10; v += 2; break;
 463                         case '3': v *= 10; v += 3; break;
 464                         case '4': v *= 10; v += 4; break;
 465                         case '5': v *= 10; v += 5; break;
 466                         case '6': v *= 10; v += 6; break;
 467                         case '7': v *= 10; v += 7; break;
 468                         case '8': v *= 10; v += 8; break;
 469                         case '9': v *= 10; v += 9; break;
 470                         default:
 471                                 goto end;
 472                         }
 473                 }
 474                 break;
 475         default:
 476                 assert(0);
 477                 break;
 478         }
 479 end:
 480         *endptr = s;
 481         return v;
 482 }
 483
 484 /**
 485  * Parses a hex number including hex floats and set the
 486  * lexer_token.
 487  */
 488 static void parse_number_hex(void)
 489 {
 490         assert(c == 'x' || c == 'X');
 491         next_char();
 492
 493         while(isxdigit(c)) {
 494                 obstack_1grow(&symbol_obstack, (char) c);
 495                 next_char();
 496         }
 497         obstack_1grow(&symbol_obstack, '\0');
 498         char *string = obstack_finish(&symbol_obstack);
 499
 500         if(c == '.' || c == 'p' || c == 'P') {
 501                 next_char();
 502                 internal_error("Hex floating point numbers not implemented yet");
 503         }
 504         if(*string == '\0') {
 505                 parse_error("invalid hex number");
 506                 lexer_token.type = T_ERROR;
 507         }
 508
 509         const char *endptr;
 510         lexer_token.type       = T_INTEGER;
 511         lexer_token.v.intvalue = parse_int_string(string, &endptr, 16);
 512         if(*endptr != '\0') {
 513                 parse_error("hex number literal too long");
 514         }
 515
 516         obstack_free(&symbol_obstack, string);
 517         parse_integer_suffix(true);
 518 }
 519
 520 /**
 521  * Returns true if the given char is a octal digit.
 522  *
 523  * @param char  the character to check
 524  */
 525 static inline bool is_octal_digit(int chr)
 526 {
 527         switch(chr) {
 528         case '0':
 529         case '1':
 530         case '2':
 531         case '3':
 532         case '4':
 533         case '5':
 534         case '6':
 535         case '7':
 536                 return true;
 537         default:
 538                 return false;
 539         }
 540 }
 541
 542 /**
 543  * Parses a octal number and set the lexer_token.
 544  */
 545 static void parse_number_oct(void)
 546 {
 547         while(is_octal_digit(c)) {
 548                 obstack_1grow(&symbol_obstack, (char) c);
 549                 next_char();
 550         }
 551         obstack_1grow(&symbol_obstack, '\0');
 552         char *string = obstack_finish(&symbol_obstack);
 553
 554         const char *endptr;
 555         lexer_token.type       = T_INTEGER;
 556         lexer_token.v.intvalue = parse_int_string(string, &endptr, 8);
 557         if(*endptr != '\0') {
 558                 parse_error("octal number literal too long");
 559         }
 560
 561         obstack_free(&symbol_obstack, string);
 562         parse_integer_suffix(true);
 563 }
 564
 565 /**
 566  * Parses a decimal including float number and set the
 567  * lexer_token.
 568  */
 569 static void parse_number_dec(void)
 570 {
 571         bool is_float = false;
 572         while(isdigit(c)) {
 573                 obstack_1grow(&symbol_obstack, (char) c);
 574                 next_char();
 575         }
 576
 577         if(c == '.') {
 578                 obstack_1grow(&symbol_obstack, '.');
 579                 next_char();
 580
 581                 while(isdigit(c)) {
 582                         obstack_1grow(&symbol_obstack, (char) c);
 583                         next_char();
 584                 }
 585                 is_float = true;
 586         }
 587         if(c == 'e' || c == 'E') {
 588                 obstack_1grow(&symbol_obstack, 'e');
 589                 next_char();
 590
 591                 if(c == '-' || c == '+') {
 592                         obstack_1grow(&symbol_obstack, (char) c);
 593                         next_char();
 594                 }
 595
 596                 while(isdigit(c)) {
 597                         obstack_1grow(&symbol_obstack, (char) c);
 598                         next_char();
 599                 }
 600                 is_float = true;
 601         }
 602
 603         obstack_1grow(&symbol_obstack, '\0');
 604         char *string = obstack_finish(&symbol_obstack);
 605
 606         if(is_float) {
 607                 char *endptr;
 608                 lexer_token.type         = T_FLOATINGPOINT;
 609                 lexer_token.v.floatvalue = strtold(string, &endptr);
 610
 611                 if(*endptr != '\0') {
 612                         parse_error("invalid number literal");
 613                 }
 614
 615                 parse_floating_suffix();
 616         } else {
 617                 const char *endptr;
 618                 lexer_token.type       = T_INTEGER;
 619                 lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 620
 621                 if(*endptr != '\0') {
 622                         parse_error("invalid number literal");
 623                 }
 624
 625                 parse_integer_suffix(false);
 626         }
 627         obstack_free(&symbol_obstack, string);
 628 }
 629
 630 /**
 631  * Parses a number and sets the lexer_token.
 632  */
 633 static void parse_number(void)
 634 {
 635         if (c == '0') {
 636                 next_char();
 637                 switch (c) {
 638                         case 'X':
 639                         case 'x':
 640                                 parse_number_hex();
 641                                 break;
 642                         case '0':
 643                         case '1':
 644                         case '2':
 645                         case '3':
 646                         case '4':
 647                         case '5':
 648                         case '6':
 649                         case '7':
 650                                 parse_number_oct();
 651                                 break;
 652                         case '8':
 653                         case '9':
 654                                 next_char();
 655                                 parse_error("invalid octal number");
 656                                 lexer_token.type = T_ERROR;
 657                                 return;
 658                         case '.':
 659                         case 'e':
 660                         case 'E':
 661                         default:
 662                                 obstack_1grow(&symbol_obstack, '0');
 663                                 parse_number_dec();
 664                                 return;
 665                 }
 666         } else {
 667                 parse_number_dec();
 668         }
 669 }
 670
 671 /**
 672  * Returns the value of a digit.
 673  * The only portable way to do it ...
 674  */
 675 static int digit_value(int digit) {
 676         switch (digit) {
 677         case '0': return 0;
 678         case '1': return 1;
 679         case '2': return 2;
 680         case '3': return 3;
 681         case '4': return 4;
 682         case '5': return 5;
 683         case '6': return 6;
 684         case '7': return 7;
 685         case '8': return 8;
 686         case '9': return 9;
 687         case 'a':
 688         case 'A': return 10;
 689         case 'b':
 690         case 'B': return 11;
 691         case 'c':
 692         case 'C': return 12;
 693         case 'd':
 694         case 'D': return 13;
 695         case 'e':
 696         case 'E': return 14;
 697         case 'f':
 698         case 'F': return 15;
 699         default:
 700                 internal_error("wrong character given");
 701         }
 702 }
 703
 704 /**
 705  * Parses an octal character sequence.
 706  *
 707  * @param first_digit  the already read first digit
 708  */
 709 static int parse_octal_sequence(const int first_digit)
 710 {
 711         assert(is_octal_digit(first_digit));
 712         int value = digit_value(first_digit);
 713         if (!is_octal_digit(c)) return value;
 714         value = 8 * value + digit_value(c);
 715         next_char();
 716         if (!is_octal_digit(c)) return value;
 717         value = 8 * value + digit_value(c);
 718         next_char();
 719
 720         if(char_is_signed) {
 721                 return (signed char) value;
 722         } else {
 723                 return (unsigned char) value;
 724         }
 725 }
 726
 727 /**
 728  * Parses a hex character sequence.
 729  */
 730 static int parse_hex_sequence(void)
 731 {
 732         int value = 0;
 733         while(isxdigit(c)) {
 734                 value = 16 * value + digit_value(c);
 735                 next_char();
 736         }
 737
 738         if(char_is_signed) {
 739                 return (signed char) value;
 740         } else {
 741                 return (unsigned char) value;
 742         }
 743 }
 744
 745 /**
 746  * Parse an escape sequence.
 747  */
 748 static int parse_escape_sequence(void)
 749 {
 750         eat('\\');
 751
 752         int ec = c;
 753         next_char();
 754
 755         switch(ec) {
 756         case '"':  return '"';
 757         case '\'': return '\'';
 758         case '\\': return '\\';
 759         case '?': return '\?';
 760         case 'a': return '\a';
 761         case 'b': return '\b';
 762         case 'f': return '\f';
 763         case 'n': return '\n';
 764         case 'r': return '\r';
 765         case 't': return '\t';
 766         case 'v': return '\v';
 767         case 'x':
 768                 return parse_hex_sequence();
 769         case '0':
 770         case '1':
 771         case '2':
 772         case '3':
 773         case '4':
 774         case '5':
 775         case '6':
 776         case '7':
 777                 return parse_octal_sequence(ec);
 778         case EOF:
 779                 parse_error("reached end of file while parsing escape sequence");
 780                 return EOF;
 781         default:
 782                 parse_error("unknown escape sequence");
 783                 return EOF;
 784         }
 785 }
 786
 787 /**
 788  * Concatenate two strings.
 789  */
 790 string_t concat_strings(const string_t *const s1, const string_t *const s2)
 791 {
 792         const size_t len1 = s1->size - 1;
 793         const size_t len2 = s2->size - 1;
 794
 795         char *const concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
 796         memcpy(concat, s1->begin, len1);
 797         memcpy(concat + len1, s2->begin, len2 + 1);
 798
 799 #if 0 /* TODO hash */
 800         const char *result = strset_insert(&stringset, concat);
 801         if(result != concat) {
 802                 obstack_free(&symbol_obstack, concat);
 803         }
 804
 805         return result;
 806 #else
 807         return (string_t){ concat, len1 + len2 + 1 };
 808 #endif
 809 }
 810
 811 /**
 812  * Concatenate a string and a wide string.
 813  */
 814 wide_string_t concat_string_wide_string(const string_t *const s1, const wide_string_t *const s2)
 815 {
 816         const size_t len1 = s1->size - 1;
 817         const size_t len2 = s2->size - 1;
 818
 819         wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat));
 820         const char *const src = s1->begin;
 821         for (size_t i = 0; i != len1; ++i) {
 822                 concat[i] = src[i];
 823         }
 824         memcpy(concat + len1, s2->begin, (len2 + 1) * sizeof(*concat));
 825
 826         return (wide_string_t){ concat, len1 + len2 + 1 };
 827 }
 828
 829 /**
 830  * Concatenate two wide strings.
 831  */
 832 wide_string_t concat_wide_strings(const wide_string_t *const s1, const wide_string_t *const s2)
 833 {
 834         const size_t len1 = s1->size - 1;
 835         const size_t len2 = s2->size - 1;
 836
 837         wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat));
 838         memcpy(concat,        s1->begin, len1       * sizeof(*concat));
 839         memcpy(concat + len1, s2->begin, (len2 + 1) * sizeof(*concat));
 840
 841         return (wide_string_t){ concat, len1 + len2 + 1 };
 842 }
 843
 844 /**
 845  * Concatenate a wide string and a string.
 846  */
 847 wide_string_t concat_wide_string_string(const wide_string_t *const s1, const string_t *const s2)
 848 {
 849         const size_t len1 = s1->size - 1;
 850         const size_t len2 = s2->size - 1;
 851
 852         wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat));
 853         memcpy(concat, s1->begin, len1 * sizeof(*concat));
 854         const char *const src = s2->begin;
 855         for (size_t i = 0; i != len2 + 1; ++i) {
 856                 concat[i] = src[i];
 857         }
 858
 859         return (wide_string_t){ concat, len1 + len2 + 1 };
 860 }
 861
 862 /**
 863  * Parse a string literal and set lexer_token.
 864  */
 865 static void parse_string_literal(void)
 866 {
 867         const unsigned start_linenr = lexer_token.source_position.linenr;
 868
 869         eat('"');
 870
 871         int tc;
 872         while(1) {
 873                 switch(c) {
 874                 case '\\':
 875                         tc = parse_escape_sequence();
 876                         obstack_1grow(&symbol_obstack, (char) tc);
 877                         break;
 878
 879                 case EOF: {
 880                         source_position_t source_position;
 881                         source_position.input_name = lexer_token.source_position.input_name;
 882                         source_position.linenr     = start_linenr;
 883                         errorf(&source_position, "string has no end");
 884                         lexer_token.type = T_ERROR;
 885                         return;
 886                 }
 887
 888                 case '"':
 889                         next_char();
 890                         goto end_of_string;
 891
 892                 default:
 893                         obstack_1grow(&symbol_obstack, (char) c);
 894                         next_char();
 895                         break;
 896                 }
 897         }
 898
 899 end_of_string:
 900
 901         /* TODO: concatenate multiple strings separated by whitespace... */
 902
 903         /* add finishing 0 to the string */
 904         obstack_1grow(&symbol_obstack, '\0');
 905         const size_t      size   = (size_t)obstack_object_size(&symbol_obstack);
 906         const char *const string = obstack_finish(&symbol_obstack);
 907
 908 #if 0 /* TODO hash */
 909         /* check if there is already a copy of the string */
 910         result = strset_insert(&stringset, string);
 911         if(result != string) {
 912                 obstack_free(&symbol_obstack, string);
 913         }
 914 #else
 915         const char *const result = string;
 916 #endif
 917
 918         lexer_token.type           = T_STRING_LITERAL;
 919         lexer_token.v.string.begin = result;
 920         lexer_token.v.string.size  = size;
 921 }
 922
 923 /**
 924  * Parse a wide character constant and set lexer_token.
 925  */
 926 static void parse_wide_character_constant(void)
 927 {
 928         const unsigned start_linenr = lexer_token.source_position.linenr;
 929
 930         eat('\'');
 931
 932         while(1) {
 933                 switch(c) {
 934                 case '\\': {
 935                         wchar_rep_t tc = parse_escape_sequence();
 936                         obstack_grow(&symbol_obstack, &tc, sizeof(tc));
 937                         break;
 938                 }
 939
 940                 MATCH_NEWLINE(
 941                         parse_error("newline while parsing character constant");
 942                         break;
 943                 )
 944
 945                 case '\'':
 946                         next_char();
 947                         goto end_of_wide_char_constant;
 948
 949                 case EOF: {
 950                         source_position_t source_position = lexer_token.source_position;
 951                         source_position.linenr = start_linenr;
 952                         errorf(&source_position, "EOF while parsing character constant");
 953                         lexer_token.type = T_ERROR;
 954                         return;
 955                 }
 956
 957                 default: {
 958                         wchar_rep_t tc = (wchar_rep_t) c;
 959                         obstack_grow(&symbol_obstack, &tc, sizeof(tc));
 960                         next_char();
 961                         break;
 962                 }
 963                 }
 964         }
 965
 966 end_of_wide_char_constant:;
 967         size_t             size   = (size_t) obstack_object_size(&symbol_obstack);
 968         assert(size % sizeof(wchar_rep_t) == 0);
 969         size /= sizeof(wchar_rep_t);
 970
 971         const wchar_rep_t *string = obstack_finish(&symbol_obstack);
 972
 973         lexer_token.type                = T_WIDE_CHARACTER_CONSTANT;
 974         lexer_token.v.wide_string.begin = string;
 975         lexer_token.v.wide_string.size  = size;
 976         lexer_token.datatype            = type_wchar_t;
 977 }
 978
 979 /**
 980  * Parse a wide string literal and set lexer_token.
 981  */
 982 static void parse_wide_string_literal(void)
 983 {
 984         const unsigned start_linenr = lexer_token.source_position.linenr;
 985
 986         assert(c == '"');
 987         next_char();
 988
 989         while(1) {
 990                 switch(c) {
 991                 case '\\': {
 992                         wchar_rep_t tc = parse_escape_sequence();
 993                         obstack_grow(&symbol_obstack, &tc, sizeof(tc));
 994                         break;
 995                 }
 996
 997                 case EOF: {
 998                         source_position_t source_position;
 999                         source_position.input_name = lexer_token.source_position.input_name;
1000                         source_position.linenr     = start_linenr;
1001                         errorf(&source_position, "string has no end");
1002                         lexer_token.type = T_ERROR;
1003                         return;
1004                 }
1005
1006                 case '"':
1007                         next_char();
1008                         goto end_of_string;
1009
1010                 default: {
1011                         wchar_rep_t tc = c;
1012                         obstack_grow(&symbol_obstack, &tc, sizeof(tc));
1013                         next_char();
1014                         break;
1015                 }
1016                 }
1017         }
1018
1019 end_of_string:;
1020
1021         /* TODO: concatenate multiple strings separated by whitespace... */
1022
1023         /* add finishing 0 to the string */
1024         wchar_rep_t nul = L'\0';
1025         obstack_grow(&symbol_obstack, &nul, sizeof(nul));
1026         const size_t             size   = (size_t)obstack_object_size(&symbol_obstack) / sizeof(wchar_rep_t);
1027         const wchar_rep_t *const string = obstack_finish(&symbol_obstack);
1028
1029 #if 0 /* TODO hash */
1030         /* check if there is already a copy of the string */
1031         const wchar_rep_t *const result = strset_insert(&stringset, string);
1032         if(result != string) {
1033                 obstack_free(&symbol_obstack, string);
1034         }
1035 #else
1036         const wchar_rep_t *const result = string;
1037 #endif
1038
1039         lexer_token.type                = T_WIDE_STRING_LITERAL;
1040         lexer_token.v.wide_string.begin = result;
1041         lexer_token.v.wide_string.size  = size;
1042 }
1043
1044 /**
1045  * Parse a character constant and set lexer_token.
1046  */
1047 static void parse_character_constant(void)
1048 {
1049         const unsigned start_linenr = lexer_token.source_position.linenr;
1050
1051         eat('\'');
1052
1053         while(1) {
1054                 switch(c) {
1055                 case '\\': {
1056                         int tc = parse_escape_sequence();
1057                         obstack_1grow(&symbol_obstack, (char) tc);
1058                         break;
1059                 }
1060
1061                 MATCH_NEWLINE(
1062                         parse_error("newline while parsing character constant");
1063                         break;
1064                 )
1065
1066                 case '\'':
1067                         next_char();
1068                         goto end_of_char_constant;
1069
1070                 case EOF: {
1071                         source_position_t source_position;
1072                         source_position.input_name = lexer_token.source_position.input_name;
1073                         source_position.linenr     = start_linenr;
1074                         errorf(&source_position, "EOF while parsing character constant");
1075                         lexer_token.type = T_ERROR;
1076                         return;
1077                 }
1078
1079                 default:
1080                         obstack_1grow(&symbol_obstack, (char) c);
1081                         next_char();
1082                         break;
1083
1084                 }
1085         }
1086
1087 end_of_char_constant:;
1088         const size_t      size   = (size_t)obstack_object_size(&symbol_obstack);
1089         const char *const string = obstack_finish(&symbol_obstack);
1090
1091         lexer_token.type           = T_CHARACTER_CONSTANT;
1092         lexer_token.v.string.begin = string;
1093         lexer_token.v.string.size  = size;
1094         lexer_token.datatype       = type_int;
1095 }
1096
1097 /**
1098  * Skip a multiline comment.
1099  */
1100 static void skip_multiline_comment(void)
1101 {
1102         unsigned start_linenr = lexer_token.source_position.linenr;
1103
1104         while(1) {
1105                 switch(c) {
1106                 case '/':
1107                         next_char();
1108                         if (c == '*') {
1109                                 /* TODO: nested comment, warn here */
1110                         }
1111                         break;
1112                 case '*':
1113                         next_char();
1114                         if(c == '/') {
1115                                 next_char();
1116                                 return;
1117                         }
1118                         break;
1119
1120                 MATCH_NEWLINE(break;)
1121
1122                 case EOF: {
1123                         source_position_t source_position;
1124                         source_position.input_name = lexer_token.source_position.input_name;
1125                         source_position.linenr     = start_linenr;
1126                         errorf(&source_position, "at end of file while looking for comment end");
1127                         return;
1128                 }
1129
1130                 default:
1131                         next_char();
1132                         break;
1133                 }
1134         }
1135 }
1136
1137 /**
1138  * Skip a single line comment.
1139  */
1140 static void skip_line_comment(void)
1141 {
1142         while(1) {
1143                 switch(c) {
1144                 case EOF:
1145                         return;
1146
1147                 case '\n':
1148                 case '\r':
1149                         return;
1150
1151                 default:
1152                         next_char();
1153                         break;
1154                 }
1155         }
1156 }
1157
1158 /** The current preprocessor token. */
1159 static token_t pp_token;
1160
1161 /**
1162  * Read the next preprocessor token.
1163  */
1164 static inline void next_pp_token(void)
1165 {
1166         lexer_next_preprocessing_token();
1167         pp_token = lexer_token;
1168 }
1169
1170 /**
1171  * Eat all preprocessor tokens until newline.
1172  */
1173 static void eat_until_newline(void)
1174 {
1175         while(pp_token.type != '\n' && pp_token.type != T_EOF) {
1176                 next_pp_token();
1177         }
1178 }
1179
1180 /**
1181  * Handle the define directive.
1182  */
1183 static void define_directive(void)
1184 {
1185         lexer_next_preprocessing_token();
1186         if(lexer_token.type != T_IDENTIFIER) {
1187                 parse_error("expected identifier after #define\n");
1188                 eat_until_newline();
1189         }
1190 }
1191
1192 /**
1193  * Handle the ifdef directive.
1194  */
1195 static void ifdef_directive(int is_ifndef)
1196 {
1197         (void) is_ifndef;
1198         lexer_next_preprocessing_token();
1199         //expect_identifier();
1200         //extect_newline();
1201 }
1202
1203 /**
1204  * Handle the endif directive.
1205  */
1206 static void endif_directive(void)
1207 {
1208         //expect_newline();
1209 }
1210
1211 /**
1212  * Parse the line directive.
1213  */
1214 static void parse_line_directive(void)
1215 {
1216         if(pp_token.type != T_INTEGER) {
1217                 parse_error("expected integer");
1218         } else {
1219                 lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
1220                 next_pp_token();
1221         }
1222         if(pp_token.type == T_STRING_LITERAL) {
1223                 lexer_token.source_position.input_name = pp_token.v.string.begin;
1224                 next_pp_token();
1225         }
1226
1227         eat_until_newline();
1228 }
1229
1230 /**
1231  * STDC pragmas.
1232  */
1233 typedef enum stdc_pragma_kind_t {
1234         STDC_UNKNOWN,
1235         STDC_FP_CONTRACT,
1236         STDC_FENV_ACCESS,
1237         STDC_CX_LIMITED_RANGE
1238 } stdc_pragma_kind_t;
1239
1240 /**
1241  * STDC pragma values.
1242  */
1243 typedef enum stdc_pragma_value_kind_t {
1244         STDC_VALUE_UNKNOWN,
1245         STDC_VALUE_ON,
1246         STDC_VALUE_OFF,
1247         STDC_VALUE_DEFAULT
1248 } stdc_pragma_value_kind_t;
1249
1250 /**
1251  * Parse a pragma directive.
1252  */
1253 static void parse_pragma(void) {
1254         bool unknown_pragma = true;
1255
1256         next_pp_token();
1257         if (pp_token.v.symbol->pp_ID == TP_STDC) {
1258                 stdc_pragma_kind_t kind = STDC_UNKNOWN;
1259                 /* a STDC pragma */
1260                 if (c_mode & _C99) {
1261                         next_pp_token();
1262
1263                         switch (pp_token.v.symbol->pp_ID) {
1264                         case TP_FP_CONTRACT:
1265                                 kind = STDC_FP_CONTRACT;
1266                                 break;
1267                         case TP_FENV_ACCESS:
1268                                 kind = STDC_FENV_ACCESS;
1269                                 break;
1270                         case TP_CX_LIMITED_RANGE:
1271                                 kind = STDC_CX_LIMITED_RANGE;
1272                                 break;
1273                         default:
1274                                 break;
1275                         }
1276                         if (kind != STDC_UNKNOWN) {
1277                                 stdc_pragma_value_kind_t value = STDC_VALUE_UNKNOWN;
1278                                 next_pp_token();
1279                                 switch (pp_token.v.symbol->pp_ID) {
1280                                 case TP_ON:
1281                                         value = STDC_VALUE_ON;
1282                                         break;
1283                                 case TP_OFF:
1284                                         value = STDC_VALUE_OFF;
1285                                         break;
1286                                 case TP_DEFAULT:
1287                                         value = STDC_VALUE_DEFAULT;
1288                                         break;
1289                                 default:
1290                                         break;
1291                                 }
1292                                 if (value != STDC_VALUE_UNKNOWN) {
1293                                         unknown_pragma = false;
1294                                 } else {
1295                                         errorf(&pp_token.source_position, "bad STDC pragma argument");
1296                                 }
1297                         }
1298                 }
1299         } else {
1300                 unknown_pragma = true;
1301         }
1302         eat_until_newline();
1303         if (unknown_pragma && warning.unknown_pragmas) {
1304                 warningf(&pp_token.source_position, "encountered unknown #pragma");
1305         }
1306 }
1307
1308 /**
1309  * Parse a preprocessor non-null directive.
1310  */
1311 static void parse_preprocessor_identifier(void)
1312 {
1313         assert(pp_token.type == T_IDENTIFIER);
1314         symbol_t *symbol = pp_token.v.symbol;
1315
1316         switch(symbol->pp_ID) {
1317         case TP_include:
1318                 printf("include - enable header name parsing!\n");
1319                 break;
1320         case TP_define:
1321                 define_directive();
1322                 break;
1323         case TP_ifdef:
1324                 ifdef_directive(0);
1325                 break;
1326         case TP_ifndef:
1327                 ifdef_directive(1);
1328                 break;
1329         case TP_endif:
1330                 endif_directive();
1331                 break;
1332         case TP_line:
1333                 next_pp_token();
1334                 parse_line_directive();
1335                 break;
1336         case TP_if:
1337         case TP_else:
1338         case TP_elif:
1339         case TP_undef:
1340         case TP_error:
1341                 /* TODO; output the rest of the line */
1342                 parse_error("#error directive: ");
1343                 break;
1344         case TP_pragma:
1345                 parse_pragma();
1346                 break;
1347         }
1348 }
1349
1350 /**
1351  * Parse a preprocessor directive.
1352  */
1353 static void parse_preprocessor_directive(void)
1354 {
1355         next_pp_token();
1356
1357         switch(pp_token.type) {
1358         case T_IDENTIFIER:
1359                 parse_preprocessor_identifier();
1360                 break;
1361         case T_INTEGER:
1362                 parse_line_directive();
1363                 break;
1364         case '\n':
1365                 /* NULL directive, see § 6.10.7 */
1366                 break;
1367         default:
1368                 parse_error("invalid preprocessor directive");
1369                 eat_until_newline();
1370                 break;
1371         }
1372 }
1373
1374 #define MAYBE_PROLOG                                       \
1375                         next_char();                                   \
1376                         while(1) {                                     \
1377                                 switch(c) {
1378
1379 #define MAYBE(ch, set_type)                                \
1380                                 case ch:                                   \
1381                                         next_char();                           \
1382                                         lexer_token.type = set_type;           \
1383                                         return;
1384
1385 #define ELSE_CODE(code)                                    \
1386                                 default:                                   \
1387                                         code                                   \
1388                                 }                                          \
1389                         } /* end of while(1) */                        \
1390                         break;
1391
1392 #define ELSE(set_type)                                     \
1393                 ELSE_CODE(                                         \
1394                         lexer_token.type = set_type;                   \
1395                         return;                                        \
1396                 )
1397
1398 void lexer_next_preprocessing_token(void)
1399 {
1400         while(1) {
1401                 switch(c) {
1402                 case ' ':
1403                 case '\t':
1404                         next_char();
1405                         break;
1406
1407                 MATCH_NEWLINE(
1408                         lexer_token.type = '\n';
1409                         return;
1410                 )
1411
1412                 SYMBOL_CHARS
1413                         parse_symbol();
1414                         /* might be a wide string ( L"string" ) */
1415                         if(lexer_token.type == T_IDENTIFIER &&
1416                             lexer_token.v.symbol == symbol_L) {
1417                             if(c == '"') {
1418                                         parse_wide_string_literal();
1419                                 } else if(c == '\'') {
1420                                         parse_wide_character_constant();
1421                                 }
1422                         }
1423                         return;
1424
1425                 DIGITS
1426                         parse_number();
1427                         return;
1428
1429                 case '"':
1430                         parse_string_literal();
1431                         return;
1432
1433                 case '\'':
1434                         parse_character_constant();
1435                         return;
1436
1437                 case '.':
1438                         MAYBE_PROLOG
1439                                 DIGITS
1440                                         put_back(c);
1441                                         c = '.';
1442                                         parse_number_dec();
1443                                         return;
1444
1445                                 case '.':
1446                                         MAYBE_PROLOG
1447                                         MAYBE('.', T_DOTDOTDOT)
1448                                         ELSE_CODE(
1449                                                 put_back(c);
1450                                                 c = '.';
1451                                                 lexer_token.type = '.';
1452                                                 return;
1453                                         )
1454                         ELSE('.')
1455                 case '&':
1456                         MAYBE_PROLOG
1457                         MAYBE('&', T_ANDAND)
1458                         MAYBE('=', T_ANDEQUAL)
1459                         ELSE('&')
1460                 case '*':
1461                         MAYBE_PROLOG
1462                         MAYBE('=', T_ASTERISKEQUAL)
1463                         ELSE('*')
1464                 case '+':
1465                         MAYBE_PROLOG
1466                         MAYBE('+', T_PLUSPLUS)
1467                         MAYBE('=', T_PLUSEQUAL)
1468                         ELSE('+')
1469                 case '-':
1470                         MAYBE_PROLOG
1471                         MAYBE('>', T_MINUSGREATER)
1472                         MAYBE('-', T_MINUSMINUS)
1473                         MAYBE('=', T_MINUSEQUAL)
1474                         ELSE('-')
1475                 case '!':
1476                         MAYBE_PROLOG
1477                         MAYBE('=', T_EXCLAMATIONMARKEQUAL)
1478                         ELSE('!')
1479                 case '/':
1480                         MAYBE_PROLOG
1481                         MAYBE('=', T_SLASHEQUAL)
1482                                 case '*':
1483                                         next_char();
1484                                         skip_multiline_comment();
1485                                         lexer_next_preprocessing_token();
1486                                         return;
1487                                 case '/':
1488                                         next_char();
1489                                         skip_line_comment();
1490                                         lexer_next_preprocessing_token();
1491                                         return;
1492                         ELSE('/')
1493                 case '%':
1494                         MAYBE_PROLOG
1495                         MAYBE('>', '}')
1496                         MAYBE('=', T_PERCENTEQUAL)
1497                                 case ':':
1498                                         MAYBE_PROLOG
1499                                                 case '%':
1500                                                         MAYBE_PROLOG
1501                                                         MAYBE(':', T_HASHHASH)
1502                                                         ELSE_CODE(
1503                                                                 put_back(c);
1504                                                                 c = '%';
1505                                                                 lexer_token.type = '#';
1506                                                                 return;
1507                                                         )
1508                                         ELSE('#')
1509                         ELSE('%')
1510                 case '<':
1511                         MAYBE_PROLOG
1512                         MAYBE(':', '[')
1513                         MAYBE('%', '{')
1514                         MAYBE('=', T_LESSEQUAL)
1515                                 case '<':
1516                                         MAYBE_PROLOG
1517                                         MAYBE('=', T_LESSLESSEQUAL)
1518                                         ELSE(T_LESSLESS)
1519                         ELSE('<')
1520                 case '>':
1521                         MAYBE_PROLOG
1522                         MAYBE('=', T_GREATEREQUAL)
1523                                 case '>':
1524                                         MAYBE_PROLOG
1525                                         MAYBE('=', T_GREATERGREATEREQUAL)
1526                                         ELSE(T_GREATERGREATER)
1527                         ELSE('>')
1528                 case '^':
1529                         MAYBE_PROLOG
1530                         MAYBE('=', T_CARETEQUAL)
1531                         ELSE('^')
1532                 case '|':
1533                         MAYBE_PROLOG
1534                         MAYBE('=', T_PIPEEQUAL)
1535                         MAYBE('|', T_PIPEPIPE)
1536                         ELSE('|')
1537                 case ':':
1538                         MAYBE_PROLOG
1539                         MAYBE('>', ']')
1540                         ELSE(':')
1541                 case '=':
1542                         MAYBE_PROLOG
1543                         MAYBE('=', T_EQUALEQUAL)
1544                         ELSE('=')
1545                 case '#':
1546                         MAYBE_PROLOG
1547                         MAYBE('#', T_HASHHASH)
1548                         ELSE('#')
1549
1550                 case '?':
1551                 case '[':
1552                 case ']':
1553                 case '(':
1554                 case ')':
1555                 case '{':
1556                 case '}':
1557                 case '~':
1558                 case ';':
1559                 case ',':
1560                 case '\\':
1561                         lexer_token.type = c;
1562                         next_char();
1563                         return;
1564
1565                 case EOF:
1566                         lexer_token.type = T_EOF;
1567                         return;
1568
1569                 default:
1570 dollar_sign:
1571                         errorf(&lexer_token.source_position, "unknown character '%c' found", c);
1572                         next_char();
1573                         lexer_token.type = T_ERROR;
1574                         return;
1575                 }
1576         }
1577 }
1578
1579 void lexer_next_token(void)
1580 {
1581         lexer_next_preprocessing_token();
1582
1583         while (lexer_token.type == '\n') {
1584 newline_found:
1585                 lexer_next_preprocessing_token();
1586         }
1587
1588         if (lexer_token.type == '#') {
1589                 parse_preprocessor_directive();
1590                 goto newline_found;
1591         }
1592 }
1593
1594 void init_lexer(void)
1595 {
1596         strset_init(&stringset);
1597         symbol_L = symbol_table_insert("L");
1598 }
1599
1600 void lexer_open_stream(FILE *stream, const char *input_name)
1601 {
1602         input                                  = stream;
1603         lexer_token.source_position.linenr     = 0;
1604         lexer_token.source_position.input_name = input_name;
1605
1606         bufpos = NULL;
1607         bufend = NULL;
1608
1609         /* place a virtual \n at the beginning so the lexer knows that we're
1610          * at the beginning of a line */
1611         c = '\n';
1612 }
1613
1614 void lexer_open_buffer(const char *buffer, size_t len, const char *input_name)
1615 {
1616         input                                  = NULL;
1617         lexer_token.source_position.linenr     = 0;
1618         lexer_token.source_position.input_name = input_name;
1619
1620         bufpos = buffer;
1621         bufend = buffer + len;
1622
1623         /* place a virtual \n at the beginning so the lexer knows that we're
1624          * at the beginning of a line */
1625         c = '\n';
1626 }
1627
1628 void exit_lexer(void)
1629 {
1630         strset_destroy(&stringset);
1631 }
1632
1633 static __attribute__((unused))
1634 void dbg_pos(const source_position_t source_position)
1635 {
1636         fprintf(stdout, "%s:%u\n", source_position.input_name,
1637                 source_position.linenr);
1638         fflush(stdout);
1639 }