usr.bin/lex/main.c

   1 /* flex - tool to generate fast lexical analyzers */
   2
   3 /*-
   4  * Copyright (c) 1990 The Regents of the University of California.
   5  * All rights reserved.
   6  *
   7  * This code is derived from software contributed to Berkeley by
   8  * Vern Paxson.
   9  *
  10  * The United States Government has rights in this work pursuant
  11  * to contract no. DE-AC03-76SF00098 between the United States
  12  * Department of Energy and the University of California.
  13  *
  14  * Redistribution and use in source and binary forms are permitted provided
  15  * that: (1) source distributions retain this entire copyright notice and
  16  * comment, and (2) distributions including binaries display the following
  17  * acknowledgement:  ``This product includes software developed by the
  18  * University of California, Berkeley and its contributors'' in the
  19  * documentation or other materials provided with the distribution and in
  20  * all advertising materials mentioning features or use of this software.
  21  * Neither the name of the University nor the names of its contributors may
  22  * be used to endorse or promote products derived from this software without
  23  * specific prior written permission.
  24  * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
  25  * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
  26  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
  27  *
  28  * @(#) Copyright (c) 1990 The Regents of the University of California. All rights reserved.
  29  * $Header: /home/daffy/u0/vern/flex/RCS/main.c,v 2.64 96/05/25 20:42:42 vern Exp $
  30  */
  31
  32 /* $Header: /home/daffy/u0/vern/flex/RCS/main.c,v 2.64 96/05/25 20:42:42 vern Exp $ */
  33 /* $FreeBSD: src/usr.bin/lex/main.c,v 1.7 1999/10/27 07:56:45 obrien Exp $ */
  34 /* $DragonFly: src/usr.bin/lex/main.c,v 1.2 2003/06/17 04:29:28 dillon Exp $ */
  35
  36
  37 #include "flexdef.h"
  38 #include "version.h"
  39
  40 static char flex_version[] = FLEX_VERSION;
  41
  42
  43 /* declare functions that have forward references */
  44
  45 void flexinit PROTO((int, char**));
  46 void readin PROTO((void));
  47 void set_up_initial_allocations PROTO((void));
  48
  49 #ifdef NEED_ARGV_FIXUP
  50 extern void argv_fixup PROTO((int *, char ***));
  51 #endif
  52
  53
  54 /* these globals are all defined and commented in flexdef.h */
  55 int printstats, syntaxerror, eofseen, ddebug, trace, nowarn, spprdflt;
  56 int interactive, caseins, lex_compat, do_yylineno, useecs, fulltbl, usemecs;
  57 int fullspd, gen_line_dirs, performance_report, backing_up_report;
  58 int C_plus_plus, long_align, use_read, yytext_is_array, do_yywrap, csize;
  59 int yymore_used, reject, real_reject, continued_action, in_rule;
  60 int yymore_really_used, reject_really_used;
  61 int datapos, dataline, linenum, out_linenum;
  62 FILE *skelfile = NULL;
  63 int skel_ind = 0;
  64 char *action_array;
  65 int action_size, defs1_offset, prolog_offset, action_offset, action_index;
  66 char *infilename = NULL, *outfilename = NULL;
  67 int did_outfilename;
  68 char *prefix, *yyclass;
  69 int do_stdinit, use_stdout;
  70 int onestate[ONE_STACK_SIZE], onesym[ONE_STACK_SIZE];
  71 int onenext[ONE_STACK_SIZE], onedef[ONE_STACK_SIZE], onesp;
  72 int current_mns, current_max_rules;
  73 int num_rules, num_eof_rules, default_rule, lastnfa;
  74 int *firstst, *lastst, *finalst, *transchar, *trans1, *trans2;
  75 int *accptnum, *assoc_rule, *state_type;
  76 int *rule_type, *rule_linenum, *rule_useful;
  77 int current_state_type;
  78 int variable_trailing_context_rules;
  79 int numtemps, numprots, protprev[MSP], protnext[MSP], prottbl[MSP];
  80 int protcomst[MSP], firstprot, lastprot, protsave[PROT_SAVE_SIZE];
  81 int numecs, nextecm[CSIZE + 1], ecgroup[CSIZE + 1], nummecs, tecfwd[CSIZE + 1];
  82 int tecbck[CSIZE + 1];
  83 int lastsc, *scset, *scbol, *scxclu, *sceof;
  84 int current_max_scs;
  85 char **scname;
  86 int current_max_dfa_size, current_max_xpairs;
  87 int current_max_template_xpairs, current_max_dfas;
  88 int lastdfa, *nxt, *chk, *tnxt;
  89 int *base, *def, *nultrans, NUL_ec, tblend, firstfree, **dss, *dfasiz;
  90 union dfaacc_union *dfaacc;
  91 int *accsiz, *dhash, numas;
  92 int numsnpairs, jambase, jamstate;
  93 int lastccl, *cclmap, *ccllen, *cclng, cclreuse;
  94 int current_maxccls, current_max_ccl_tbl_size;
  95 Char *ccltbl;
  96 char nmstr[MAXLINE];
  97 int sectnum, nummt, hshcol, dfaeql, numeps, eps2, num_reallocs;
  98 int tmpuses, totnst, peakpairs, numuniq, numdup, hshsave;
  99 int num_backing_up, bol_needed;
 100 FILE *backing_up_file;
 101 int end_of_buffer_state;
 102 char **input_files;
 103 int num_input_files;
 104
 105 /* Make sure program_name is initialized so we don't crash if writing
 106  * out an error message before getting the program name from argv[0].
 107  */
 108 char *program_name = "flex";
 109
 110 #ifndef SHORT_FILE_NAMES
 111 static char *outfile_template = "lex.%s.%s";
 112 static char *backing_name = "lex.backup";
 113 #else
 114 static char *outfile_template = "lex%s.%s";
 115 static char *backing_name = "lex.bck";
 116 #endif
 117
 118 #ifdef THINK_C
 119 #include <console.h>
 120 #endif
 121
 122 #ifdef MS_DOS
 123 extern unsigned _stklen = 16384;
 124 #endif
 125
 126 static char outfile_path[MAXLINE];
 127 static int outfile_created = 0;
 128 static char *skelname = NULL;
 129
 130
 131 int main( argc, argv )
 132 int argc;
 133 char **argv;
 134         {
 135         int i;
 136
 137 #ifdef THINK_C
 138         argc = ccommand( &argv );
 139 #endif
 140 #ifdef NEED_ARGV_FIXUP
 141         argv_fixup( &argc, &argv );
 142 #endif
 143
 144         flexinit( argc, argv );
 145
 146         readin();
 147
 148         ntod();
 149
 150         for ( i = 1; i <= num_rules; ++i )
 151                 if ( ! rule_useful[i] && i != default_rule )
 152                         line_warning( _( "rule cannot be matched" ),
 153                                         rule_linenum[i] );
 154
 155         if ( spprdflt && ! reject && rule_useful[default_rule] )
 156                 line_warning(
 157                         _( "-s option given but default rule can be matched" ),
 158                         rule_linenum[default_rule] );
 159
 160         /* Generate the C state transition tables from the DFA. */
 161         make_tables();
 162
 163         /* Note, flexend does not return.  It exits with its argument
 164          * as status.
 165          */
 166         flexend( 0 );
 167
 168         return 0;       /* keep compilers/lint happy */
 169         }
 170
 171
 172 /* check_options - check user-specified options */
 173
 174 void check_options()
 175         {
 176         int i;
 177
 178         if ( lex_compat )
 179                 {
 180                 if ( C_plus_plus )
 181                         flexerror( _( "Can't use -+ with -l option" ) );
 182
 183                 if ( fulltbl || fullspd )
 184                         flexerror( _( "Can't use -f or -F with -l option" ) );
 185
 186                 /* Don't rely on detecting use of yymore() and REJECT,
 187                  * just assume they'll be used.
 188                  */
 189                 yymore_really_used = reject_really_used = true;
 190
 191                 yytext_is_array = true;
 192                 do_yylineno = true;
 193                 use_read = false;
 194                 }
 195
 196         if ( do_yylineno )
 197                 /* This should really be "maintain_backup_tables = true" */
 198                 reject_really_used = true;
 199
 200         if ( csize == unspecified )
 201                 {
 202                 if ( (fulltbl || fullspd) && ! useecs )
 203                         csize = DEFAULT_CSIZE;
 204                 else
 205                         csize = CSIZE;
 206                 }
 207
 208         if ( interactive == unspecified )
 209                 {
 210                 if ( fulltbl || fullspd )
 211                         interactive = false;
 212                 else
 213                         interactive = true;
 214                 }
 215
 216         if ( fulltbl || fullspd )
 217                 {
 218                 if ( usemecs )
 219                         flexerror(
 220                         _( "-Cf/-CF and -Cm don't make sense together" ) );
 221
 222                 if ( interactive )
 223                         flexerror( _( "-Cf/-CF and -I are incompatible" ) );
 224
 225                 if ( lex_compat )
 226                         flexerror(
 227                 _( "-Cf/-CF are incompatible with lex-compatibility mode" ) );
 228
 229                 if ( do_yylineno )
 230                         flexerror(
 231                         _( "-Cf/-CF and %option yylineno are incompatible" ) );
 232
 233                 if ( fulltbl && fullspd )
 234                         flexerror( _( "-Cf and -CF are mutually exclusive" ) );
 235                 }
 236
 237         if ( C_plus_plus && fullspd )
 238                 flexerror( _( "Can't use -+ with -CF option" ) );
 239
 240         if ( C_plus_plus && yytext_is_array )
 241                 {
 242                 warn( _( "%array incompatible with -+ option" ) );
 243                 yytext_is_array = false;
 244                 }
 245
 246         if ( useecs )
 247                 { /* Set up doubly-linked equivalence classes. */
 248
 249                 /* We loop all the way up to csize, since ecgroup[csize] is
 250                  * the position used for NUL characters.
 251                  */
 252                 ecgroup[1] = NIL;
 253
 254                 for ( i = 2; i <= csize; ++i )
 255                         {
 256                         ecgroup[i] = i - 1;
 257                         nextecm[i - 1] = i;
 258                         }
 259
 260                 nextecm[csize] = NIL;
 261                 }
 262
 263         else
 264                 {
 265                 /* Put everything in its own equivalence class. */
 266                 for ( i = 1; i <= csize; ++i )
 267                         {
 268                         ecgroup[i] = i;
 269                         nextecm[i] = BAD_SUBSCRIPT;     /* to catch errors */
 270                         }
 271                 }
 272
 273         if ( ! use_stdout )
 274                 {
 275                 FILE *prev_stdout;
 276
 277                 if ( ! did_outfilename )
 278                         {
 279                         char *suffix;
 280
 281                         if ( C_plus_plus )
 282                                 suffix = "cc";
 283                         else
 284                                 suffix = "c";
 285
 286                         sprintf( outfile_path, outfile_template,
 287                                 prefix, suffix );
 288
 289                         outfilename = outfile_path;
 290                         }
 291
 292                 prev_stdout = freopen( outfilename, "w", stdout );
 293
 294                 if ( prev_stdout == NULL )
 295                         lerrsf( _( "could not create %s" ), outfilename );
 296
 297                 outfile_created = 1;
 298                 }
 299
 300         if ( skelname && (skelfile = fopen( skelname, "r" )) == NULL )
 301                 lerrsf( _( "can't open skeleton file %s" ), skelname );
 302
 303         if ( strcmp( prefix, "yy" ) )
 304                 {
 305 #define GEN_PREFIX(name) out_str3( "#define yy%s %s%s\n", name, prefix, name )
 306                 if ( C_plus_plus )
 307                         GEN_PREFIX( "FlexLexer" );
 308                 else
 309                         {
 310                         GEN_PREFIX( "_create_buffer" );
 311                         GEN_PREFIX( "_delete_buffer" );
 312                         GEN_PREFIX( "_scan_buffer" );
 313                         GEN_PREFIX( "_scan_string" );
 314                         GEN_PREFIX( "_scan_bytes" );
 315                         GEN_PREFIX( "_flex_debug" );
 316                         GEN_PREFIX( "_init_buffer" );
 317                         GEN_PREFIX( "_flush_buffer" );
 318                         GEN_PREFIX( "_load_buffer_state" );
 319                         GEN_PREFIX( "_switch_to_buffer" );
 320                         GEN_PREFIX( "in" );
 321                         GEN_PREFIX( "leng" );
 322                         GEN_PREFIX( "lex" );
 323                         GEN_PREFIX( "out" );
 324                         GEN_PREFIX( "restart" );
 325                         GEN_PREFIX( "text" );
 326
 327                         if ( do_yylineno )
 328                                 GEN_PREFIX( "lineno" );
 329                         }
 330
 331                 if ( do_yywrap )
 332                         GEN_PREFIX( "wrap" );
 333
 334                 outn( "" );
 335                 }
 336
 337         if ( did_outfilename )
 338                 line_directive_out( stdout, 0 );
 339
 340         skelout();
 341         }
 342
 343
 344 /* flexend - terminate flex
 345  *
 346  * note
 347  *    This routine does not return.
 348  */
 349
 350 void flexend( exit_status )
 351 int exit_status;
 352
 353         {
 354         int tblsiz;
 355         int unlink();
 356
 357         if ( skelfile != NULL )
 358                 {
 359                 if ( ferror( skelfile ) )
 360                         lerrsf( _( "input error reading skeleton file %s" ),
 361                                 skelname );
 362
 363                 else if ( fclose( skelfile ) )
 364                         lerrsf( _( "error closing skeleton file %s" ),
 365                                 skelname );
 366                 }
 367
 368         if ( exit_status != 0 && outfile_created )
 369                 {
 370                 if ( ferror( stdout ) )
 371                         lerrsf( _( "error writing output file %s" ),
 372                                 outfilename );
 373
 374                 else if ( fclose( stdout ) )
 375                         lerrsf( _( "error closing output file %s" ),
 376                                 outfilename );
 377
 378                 else if ( unlink( outfilename ) )
 379                         lerrsf( _( "error deleting output file %s" ),
 380                                 outfilename );
 381                 }
 382
 383         if ( backing_up_report && backing_up_file )
 384                 {
 385                 if ( num_backing_up == 0 )
 386                         fprintf( backing_up_file, _( "No backing up.\n" ) );
 387                 else if ( fullspd || fulltbl )
 388                         fprintf( backing_up_file,
 389                                 _( "%d backing up (non-accepting) states.\n" ),
 390                                 num_backing_up );
 391                 else
 392                         fprintf( backing_up_file,
 393                                 _( "Compressed tables always back up.\n" ) );
 394
 395                 if ( ferror( backing_up_file ) )
 396                         lerrsf( _( "error writing backup file %s" ),
 397                                 backing_name );
 398
 399                 else if ( fclose( backing_up_file ) )
 400                         lerrsf( _( "error closing backup file %s" ),
 401                                 backing_name );
 402                 }
 403
 404         if ( printstats )
 405                 {
 406                 fprintf( stderr, _( "%s version %s usage statistics:\n" ),
 407                         program_name, flex_version );
 408
 409                 fprintf( stderr, _( "  scanner options: -" ) );
 410
 411                 if ( C_plus_plus )
 412                         putc( '+', stderr );
 413                 if ( backing_up_report )
 414                         putc( 'b', stderr );
 415                 if ( ddebug )
 416                         putc( 'd', stderr );
 417                 if ( caseins )
 418                         putc( 'i', stderr );
 419                 if ( lex_compat )
 420                         putc( 'l', stderr );
 421                 if ( performance_report > 0 )
 422                         putc( 'p', stderr );
 423                 if ( performance_report > 1 )
 424                         putc( 'p', stderr );
 425                 if ( spprdflt )
 426                         putc( 's', stderr );
 427                 if ( use_stdout )
 428                         putc( 't', stderr );
 429                 if ( printstats )
 430                         putc( 'v', stderr );    /* always true! */
 431                 if ( nowarn )
 432                         putc( 'w', stderr );
 433                 if ( interactive == false )
 434                         putc( 'B', stderr );
 435                 if ( interactive == true )
 436                         putc( 'I', stderr );
 437                 if ( ! gen_line_dirs )
 438                         putc( 'L', stderr );
 439                 if ( trace )
 440                         putc( 'T', stderr );
 441
 442                 if ( csize == unspecified )
 443                         /* We encountered an error fairly early on, so csize
 444                          * never got specified.  Define it now, to prevent
 445                          * bogus table sizes being written out below.
 446                          */
 447                         csize = 256;
 448
 449                 if ( csize == 128 )
 450                         putc( '7', stderr );
 451                 else
 452                         putc( '8', stderr );
 453
 454                 fprintf( stderr, " -C" );
 455
 456                 if ( long_align )
 457                         putc( 'a', stderr );
 458                 if ( fulltbl )
 459                         putc( 'f', stderr );
 460                 if ( fullspd )
 461                         putc( 'F', stderr );
 462                 if ( useecs )
 463                         putc( 'e', stderr );
 464                 if ( usemecs )
 465                         putc( 'm', stderr );
 466                 if ( use_read )
 467                         putc( 'r', stderr );
 468
 469                 if ( did_outfilename )
 470                         fprintf( stderr, " -o%s", outfilename );
 471
 472                 if ( skelname )
 473                         fprintf( stderr, " -S%s", skelname );
 474
 475                 if ( strcmp( prefix, "yy" ) )
 476                         fprintf( stderr, " -P%s", prefix );
 477
 478                 putc( '\n', stderr );
 479
 480                 fprintf( stderr, _( "  %d/%d NFA states\n" ),
 481                         lastnfa, current_mns );
 482                 fprintf( stderr, _( "  %d/%d DFA states (%d words)\n" ),
 483                         lastdfa, current_max_dfas, totnst );
 484                 fprintf( stderr, _( "  %d rules\n" ),
 485                 num_rules + num_eof_rules - 1 /* - 1 for def. rule */ );
 486
 487                 if ( num_backing_up == 0 )
 488                         fprintf( stderr, _( "  No backing up\n" ) );
 489                 else if ( fullspd || fulltbl )
 490                         fprintf( stderr,
 491                         _( "  %d backing-up (non-accepting) states\n" ),
 492                                 num_backing_up );
 493                 else
 494                         fprintf( stderr,
 495                                 _( "  Compressed tables always back-up\n" ) );
 496
 497                 if ( bol_needed )
 498                         fprintf( stderr,
 499                                 _( "  Beginning-of-line patterns used\n" ) );
 500
 501                 fprintf( stderr, _( "  %d/%d start conditions\n" ), lastsc,
 502                         current_max_scs );
 503                 fprintf( stderr,
 504                         _( "  %d epsilon states, %d double epsilon states\n" ),
 505                         numeps, eps2 );
 506
 507                 if ( lastccl == 0 )
 508                         fprintf( stderr, _( "  no character classes\n" ) );
 509                 else
 510                         fprintf( stderr,
 511 _( "  %d/%d character classes needed %d/%d words of storage, %d reused\n" ),
 512                                 lastccl, current_maxccls,
 513                                 cclmap[lastccl] + ccllen[lastccl],
 514                                 current_max_ccl_tbl_size, cclreuse );
 515
 516                 fprintf( stderr, _( "  %d state/nextstate pairs created\n" ),
 517                         numsnpairs );
 518                 fprintf( stderr, _( "  %d/%d unique/duplicate transitions\n" ),
 519                         numuniq, numdup );
 520
 521                 if ( fulltbl )
 522                         {
 523                         tblsiz = lastdfa * numecs;
 524                         fprintf( stderr, _( "  %d table entries\n" ), tblsiz );
 525                         }
 526
 527                 else
 528                         {
 529                         tblsiz = 2 * (lastdfa + numtemps) + 2 * tblend;
 530
 531                         fprintf( stderr,
 532                                 _( "  %d/%d base-def entries created\n" ),
 533                                 lastdfa + numtemps, current_max_dfas );
 534                         fprintf( stderr,
 535                         _( "  %d/%d (peak %d) nxt-chk entries created\n" ),
 536                                 tblend, current_max_xpairs, peakpairs );
 537                         fprintf( stderr,
 538                 _( "  %d/%d (peak %d) template nxt-chk entries created\n" ),
 539                                 numtemps * nummecs,
 540                                 current_max_template_xpairs,
 541                                 numtemps * numecs );
 542                         fprintf( stderr, _( "  %d empty table entries\n" ),
 543                                 nummt );
 544                         fprintf( stderr, _( "  %d protos created\n" ),
 545                                 numprots );
 546                         fprintf( stderr,
 547                                 _( "  %d templates created, %d uses\n" ),
 548                                 numtemps, tmpuses );
 549                         }
 550
 551                 if ( useecs )
 552                         {
 553                         tblsiz = tblsiz + csize;
 554                         fprintf( stderr,
 555                                 _( "  %d/%d equivalence classes created\n" ),
 556                                 numecs, csize );
 557                         }
 558
 559                 if ( usemecs )
 560                         {
 561                         tblsiz = tblsiz + numecs;
 562                         fprintf( stderr,
 563                         _( "  %d/%d meta-equivalence classes created\n" ),
 564                                 nummecs, csize );
 565                         }
 566
 567                 fprintf( stderr,
 568                 _( "  %d (%d saved) hash collisions, %d DFAs equal\n" ),
 569                         hshcol, hshsave, dfaeql );
 570                 fprintf( stderr, _( "  %d sets of reallocations needed\n" ),
 571                         num_reallocs );
 572                 fprintf( stderr, _( "  %d total table entries needed\n" ),
 573                         tblsiz );
 574                 }
 575
 576         exit( exit_status );
 577         }
 578
 579
 580 /* flexinit - initialize flex */
 581
 582 void flexinit( argc, argv )
 583 int argc;
 584 char **argv;
 585         {
 586         int i, sawcmpflag;
 587         char *arg;
 588
 589         printstats = syntaxerror = trace = spprdflt = caseins = false;
 590         lex_compat = C_plus_plus = backing_up_report = ddebug = fulltbl = false;
 591         fullspd = long_align = nowarn = yymore_used = continued_action = false;
 592         do_yylineno = yytext_is_array = in_rule = reject = do_stdinit = false;
 593         yymore_really_used = reject_really_used = unspecified;
 594         interactive = csize = unspecified;
 595         do_yywrap = gen_line_dirs = usemecs = useecs = true;
 596         performance_report = 0;
 597         did_outfilename = 0;
 598         prefix = "yy";
 599         yyclass = 0;
 600         use_read = use_stdout = false;
 601
 602         sawcmpflag = false;
 603
 604         /* Initialize dynamic array for holding the rule actions. */
 605         action_size = 2048;     /* default size of action array in bytes */
 606         action_array = allocate_character_array( action_size );
 607         defs1_offset = prolog_offset = action_offset = action_index = 0;
 608         action_array[0] = '\0';
 609
 610         program_name = argv[0];
 611
 612         if ( program_name[0] != '\0' &&
 613              program_name[strlen( program_name ) - 1] == '+' )
 614                 C_plus_plus = true;
 615
 616         /* read flags */
 617         for ( --argc, ++argv; argc ; --argc, ++argv )
 618                 {
 619                 arg = argv[0];
 620
 621                 if ( arg[0] != '-' || arg[1] == '\0' )
 622                         break;
 623
 624                 if ( arg[1] == '-' )
 625                         { /* --option */
 626                         if ( ! strcmp( arg, "--help" ) )
 627                                 arg = "-h";
 628
 629                         else if ( ! strcmp( arg, "--version" ) )
 630                                 arg = "-V";
 631
 632                         else if ( ! strcmp( arg, "--" ) )
 633                                 { /* end of options */
 634                                 --argc;
 635                                 ++argv;
 636                                 break;
 637                                 }
 638                         }
 639
 640                 for ( i = 1; arg[i] != '\0'; ++i )
 641                         switch ( arg[i] )
 642                                 {
 643                                 case '+':
 644                                         C_plus_plus = true;
 645                                         break;
 646
 647                                 case 'B':
 648                                         interactive = false;
 649                                         break;
 650
 651                                 case 'b':
 652                                         backing_up_report = true;
 653                                         break;
 654
 655                                 case 'c':
 656                                         break;
 657
 658                                 case 'C':
 659                                         if ( i != 1 )
 660                                                 flexerror(
 661                                 _( "-C flag must be given separately" ) );
 662
 663                                         if ( ! sawcmpflag )
 664                                                 {
 665                                                 useecs = false;
 666                                                 usemecs = false;
 667                                                 fulltbl = false;
 668                                                 sawcmpflag = true;
 669                                                 }
 670
 671                                         for ( ++i; arg[i] != '\0'; ++i )
 672                                                 switch ( arg[i] )
 673                                                         {
 674                                                         case 'a':
 675                                                                 long_align =
 676                                                                         true;
 677                                                                 break;
 678
 679                                                         case 'e':
 680                                                                 useecs = true;
 681                                                                 break;
 682
 683                                                         case 'F':
 684                                                                 fullspd = true;
 685                                                                 break;
 686
 687                                                         case 'f':
 688                                                                 fulltbl = true;
 689                                                                 break;
 690
 691                                                         case 'm':
 692                                                                 usemecs = true;
 693                                                                 break;
 694
 695                                                         case 'r':
 696                                                                 use_read = true;
 697                                                                 break;
 698
 699                                                         default:
 700                                                                 lerrif(
 701                                                 _( "unknown -C option '%c'" ),
 702                                                                 (int) arg[i] );
 703                                                                 break;
 704                                                         }
 705
 706                                         goto get_next_arg;
 707
 708                                 case 'd':
 709                                         ddebug = true;
 710                                         break;
 711
 712                                 case 'f':
 713                                         useecs = usemecs = false;
 714                                         use_read = fulltbl = true;
 715                                         break;
 716
 717                                 case 'F':
 718                                         useecs = usemecs = false;
 719                                         use_read = fullspd = true;
 720                                         break;
 721
 722                                 case '?':
 723                                 case 'h':
 724                                         usage();
 725                                         exit( 0 );
 726
 727                                 case 'I':
 728                                         interactive = true;
 729                                         break;
 730
 731                                 case 'i':
 732                                         caseins = true;
 733                                         break;
 734
 735                                 case 'l':
 736                                         lex_compat = true;
 737                                         break;
 738
 739                                 case 'L':
 740                                         gen_line_dirs = false;
 741                                         break;
 742
 743                                 case 'n':
 744                                         /* Stupid do-nothing deprecated
 745                                          * option.
 746                                          */
 747                                         break;
 748
 749                                 case 'o':
 750                                         if ( i != 1 )
 751                                                 flexerror(
 752                                 _( "-o flag must be given separately" ) );
 753
 754                                         outfilename = arg + i + 1;
 755                                         did_outfilename = 1;
 756                                         goto get_next_arg;
 757
 758                                 case 'P':
 759                                         if ( i != 1 )
 760                                                 flexerror(
 761                                 _( "-P flag must be given separately" ) );
 762
 763                                         prefix = arg + i + 1;
 764                                         goto get_next_arg;
 765
 766                                 case 'p':
 767                                         ++performance_report;
 768                                         break;
 769
 770                                 case 'S':
 771                                         if ( i != 1 )
 772                                                 flexerror(
 773                                 _( "-S flag must be given separately" ) );
 774
 775                                         skelname = arg + i + 1;
 776                                         goto get_next_arg;
 777
 778                                 case 's':
 779                                         spprdflt = true;
 780                                         break;
 781
 782                                 case 't':
 783                                         use_stdout = true;
 784                                         break;
 785
 786                                 case 'T':
 787                                         trace = true;
 788                                         break;
 789
 790                                 case 'v':
 791                                         printstats = true;
 792                                         break;
 793
 794                                 case 'V':
 795                                         printf( _( "%s version %s\n" ),
 796                                                 program_name, flex_version );
 797                                         exit( 0 );
 798
 799                                 case 'w':
 800                                         nowarn = true;
 801                                         break;
 802
 803                                 case '7':
 804                                         csize = 128;
 805                                         break;
 806
 807                                 case '8':
 808                                         csize = CSIZE;
 809                                         break;
 810
 811                                 default:
 812                                         fprintf( stderr,
 813                 _( "%s: unknown flag '%c'.  For usage, try\n\t%s --help\n" ),
 814                                                 program_name, (int) arg[i],
 815                                                 program_name );
 816                                         exit( 1 );
 817                                 }
 818
 819                 /* Used by -C, -S, -o, and -P flags in lieu of a "continue 2"
 820                  * control.
 821                  */
 822                 get_next_arg: ;
 823                 }
 824
 825         num_input_files = argc;
 826         input_files = argv;
 827         set_input_file( num_input_files > 0 ? input_files[0] : NULL );
 828
 829         lastccl = lastsc = lastdfa = lastnfa = 0;
 830         num_rules = num_eof_rules = default_rule = 0;
 831         numas = numsnpairs = tmpuses = 0;
 832         numecs = numeps = eps2 = num_reallocs = hshcol = dfaeql = totnst = 0;
 833         numuniq = numdup = hshsave = eofseen = datapos = dataline = 0;
 834         num_backing_up = onesp = numprots = 0;
 835         variable_trailing_context_rules = bol_needed = false;
 836
 837         out_linenum = linenum = sectnum = 1;
 838         firstprot = NIL;
 839
 840         /* Used in mkprot() so that the first proto goes in slot 1
 841          * of the proto queue.
 842          */
 843         lastprot = 1;
 844
 845         set_up_initial_allocations();
 846         }
 847
 848
 849 /* readin - read in the rules section of the input file(s) */
 850
 851 void readin()
 852         {
 853         static char yy_stdinit[] = "FILE *yyin = stdin, *yyout = stdout;";
 854         static char yy_nostdinit[] =
 855                 "FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;";
 856
 857         line_directive_out( (FILE *) 0, 1 );
 858
 859         if ( yyparse() )
 860                 {
 861                 pinpoint_message( _( "fatal parse error" ) );
 862                 flexend( 1 );
 863                 }
 864
 865         if ( syntaxerror )
 866                 flexend( 1 );
 867
 868         if ( backing_up_report )
 869                 {
 870                 backing_up_file = fopen( backing_name, "w" );
 871                 if ( backing_up_file == NULL )
 872                         lerrsf(
 873                         _( "could not create backing-up info file %s" ),
 874                                 backing_name );
 875                 }
 876
 877         else
 878                 backing_up_file = NULL;
 879
 880         if ( yymore_really_used == true )
 881                 yymore_used = true;
 882         else if ( yymore_really_used == false )
 883                 yymore_used = false;
 884
 885         if ( reject_really_used == true )
 886                 reject = true;
 887         else if ( reject_really_used == false )
 888                 reject = false;
 889
 890         if ( performance_report > 0 )
 891                 {
 892                 if ( lex_compat )
 893                         {
 894                         fprintf( stderr,
 895 _( "-l AT&T lex compatibility option entails a large performance penalty\n" ) );
 896                         fprintf( stderr,
 897 _( " and may be the actual source of other reported performance penalties\n" ) );
 898                         }
 899
 900                 else if ( do_yylineno )
 901                         {
 902                         fprintf( stderr,
 903         _( "%%option yylineno entails a large performance penalty\n" ) );
 904                         }
 905
 906                 if ( performance_report > 1 )
 907                         {
 908                         if ( interactive )
 909                                 fprintf( stderr,
 910         _( "-I (interactive) entails a minor performance penalty\n" ) );
 911
 912                         if ( yymore_used )
 913                                 fprintf( stderr,
 914                 _( "yymore() entails a minor performance penalty\n" ) );
 915                         }
 916
 917                 if ( reject )
 918                         fprintf( stderr,
 919                         _( "REJECT entails a large performance penalty\n" ) );
 920
 921                 if ( variable_trailing_context_rules )
 922                         fprintf( stderr,
 923 _( "Variable trailing context rules entail a large performance penalty\n" ) );
 924                 }
 925
 926         if ( reject )
 927                 real_reject = true;
 928
 929         if ( variable_trailing_context_rules )
 930                 reject = true;
 931
 932         if ( (fulltbl || fullspd) && reject )
 933                 {
 934                 if ( real_reject )
 935                         flexerror(
 936                                 _( "REJECT cannot be used with -f or -F" ) );
 937                 else if ( do_yylineno )
 938                         flexerror(
 939                         _( "%option yylineno cannot be used with -f or -F" ) );
 940                 else
 941                         flexerror(
 942         _( "variable trailing context rules cannot be used with -f or -F" ) );
 943                 }
 944
 945         if ( reject )
 946                 outn( "\n#define YY_USES_REJECT" );
 947
 948         if ( ! do_yywrap )
 949                 {
 950                 outn( "\n#define yywrap() 1" );
 951                 outn( "#define YY_SKIP_YYWRAP" );
 952                 }
 953
 954         if ( ddebug )
 955                 outn( "\n#define FLEX_DEBUG" );
 956
 957         if ( csize == 256 )
 958                 outn( "typedef unsigned char YY_CHAR;" );
 959         else
 960                 outn( "typedef char YY_CHAR;" );
 961
 962         if ( C_plus_plus )
 963                 {
 964                 outn( "#define yytext_ptr yytext" );
 965
 966                 if ( interactive )
 967                         outn( "#define YY_INTERACTIVE" );
 968                 }
 969
 970         else
 971                 {
 972                 if ( do_stdinit )
 973                         {
 974                         outn( "#ifdef VMS" );
 975                         outn( "#ifndef __VMS_POSIX" );
 976                         outn( yy_nostdinit );
 977                         outn( "#else" );
 978                         outn( yy_stdinit );
 979                         outn( "#endif" );
 980                         outn( "#else" );
 981                         outn( yy_stdinit );
 982                         outn( "#endif" );
 983                         }
 984
 985                 else
 986                         outn( yy_nostdinit );
 987                 }
 988
 989         if ( fullspd )
 990                 outn( "typedef yyconst struct yy_trans_info *yy_state_type;" );
 991         else if ( ! C_plus_plus )
 992                 outn( "typedef int yy_state_type;" );
 993
 994         if ( ddebug )
 995                 outn( "\n#define FLEX_DEBUG" );
 996
 997         if ( lex_compat )
 998                 outn( "#define YY_FLEX_LEX_COMPAT" );
 999
1000         if ( do_yylineno && ! C_plus_plus )
1001                 {
1002                 outn( "extern int yylineno;" );
1003                 outn( "int yylineno = 1;" );
1004                 }
1005
1006         if ( C_plus_plus )
1007                 {
1008                 outn( "\n#include <FlexLexer.h>" );
1009
1010                 if ( yyclass )
1011                         {
1012                         outn( "int yyFlexLexer::yylex()" );
1013                         outn( "\t{" );
1014                         outn(
1015 "\tLexerError( \"yyFlexLexer::yylex invoked but %option yyclass used\" );" );
1016                         outn( "\treturn 0;" );
1017                         outn( "\t}" );
1018
1019                         out_str( "\n#define YY_DECL int %s::yylex()\n",
1020                                 yyclass );
1021                         }
1022                 }
1023
1024         else
1025                 {
1026                 if ( yytext_is_array )
1027                         outn( "extern char yytext[];\n" );
1028
1029                 else
1030                         {
1031                         outn( "extern char *yytext;" );
1032                         outn( "#define yytext_ptr yytext" );
1033                         }
1034
1035                 if ( yyclass )
1036                         flexerror(
1037                 _( "%option yyclass only meaningful for C++ scanners" ) );
1038                 }
1039
1040         if ( useecs )
1041                 numecs = cre8ecs( nextecm, ecgroup, csize );
1042         else
1043                 numecs = csize;
1044
1045         /* Now map the equivalence class for NUL to its expected place. */
1046         ecgroup[0] = ecgroup[csize];
1047         NUL_ec = ABS( ecgroup[0] );
1048
1049         if ( useecs )
1050                 ccl2ecl();
1051         }
1052
1053
1054 /* set_up_initial_allocations - allocate memory for internal tables */
1055
1056 void set_up_initial_allocations()
1057         {
1058         current_mns = INITIAL_MNS;
1059         firstst = allocate_integer_array( current_mns );
1060         lastst = allocate_integer_array( current_mns );
1061         finalst = allocate_integer_array( current_mns );
1062         transchar = allocate_integer_array( current_mns );
1063         trans1 = allocate_integer_array( current_mns );
1064         trans2 = allocate_integer_array( current_mns );
1065         accptnum = allocate_integer_array( current_mns );
1066         assoc_rule = allocate_integer_array( current_mns );
1067         state_type = allocate_integer_array( current_mns );
1068
1069         current_max_rules = INITIAL_MAX_RULES;
1070         rule_type = allocate_integer_array( current_max_rules );
1071         rule_linenum = allocate_integer_array( current_max_rules );
1072         rule_useful = allocate_integer_array( current_max_rules );
1073
1074         current_max_scs = INITIAL_MAX_SCS;
1075         scset = allocate_integer_array( current_max_scs );
1076         scbol = allocate_integer_array( current_max_scs );
1077         scxclu = allocate_integer_array( current_max_scs );
1078         sceof = allocate_integer_array( current_max_scs );
1079         scname = allocate_char_ptr_array( current_max_scs );
1080
1081         current_maxccls = INITIAL_MAX_CCLS;
1082         cclmap = allocate_integer_array( current_maxccls );
1083         ccllen = allocate_integer_array( current_maxccls );
1084         cclng = allocate_integer_array( current_maxccls );
1085
1086         current_max_ccl_tbl_size = INITIAL_MAX_CCL_TBL_SIZE;
1087         ccltbl = allocate_Character_array( current_max_ccl_tbl_size );
1088
1089         current_max_dfa_size = INITIAL_MAX_DFA_SIZE;
1090
1091         current_max_xpairs = INITIAL_MAX_XPAIRS;
1092         nxt = allocate_integer_array( current_max_xpairs );
1093         chk = allocate_integer_array( current_max_xpairs );
1094
1095         current_max_template_xpairs = INITIAL_MAX_TEMPLATE_XPAIRS;
1096         tnxt = allocate_integer_array( current_max_template_xpairs );
1097
1098         current_max_dfas = INITIAL_MAX_DFAS;
1099         base = allocate_integer_array( current_max_dfas );
1100         def = allocate_integer_array( current_max_dfas );
1101         dfasiz = allocate_integer_array( current_max_dfas );
1102         accsiz = allocate_integer_array( current_max_dfas );
1103         dhash = allocate_integer_array( current_max_dfas );
1104         dss = allocate_int_ptr_array( current_max_dfas );
1105         dfaacc = allocate_dfaacc_union( current_max_dfas );
1106
1107         nultrans = (int *) 0;
1108         }
1109
1110
1111 void usage()
1112         {
1113         FILE *f = stdout;
1114
1115         fprintf( f,
1116 _( "%s [-bcdfhilnpstvwBFILTV78+? -C[aefFmr] -ooutput -Pprefix -Sskeleton]\n" ),
1117                 program_name );
1118         fprintf( f, _( "\t[--help --version] [file ...]\n" ) );
1119
1120         fprintf( f, _( "\t-b  generate backing-up information to %s\n" ),
1121                 backing_name );
1122         fprintf( f, _( "\t-c  do-nothing POSIX option\n" ) );
1123         fprintf( f, _( "\t-d  turn on debug mode in generated scanner\n" ) );
1124         fprintf( f, _( "\t-f  generate fast, large scanner\n" ) );
1125         fprintf( f, _( "\t-h  produce this help message\n" ) );
1126         fprintf( f, _( "\t-i  generate case-insensitive scanner\n" ) );
1127         fprintf( f, _( "\t-l  maximal compatibility with original lex\n" ) );
1128         fprintf( f, _( "\t-n  do-nothing POSIX option\n" ) );
1129         fprintf( f, _( "\t-p  generate performance report to stderr\n" ) );
1130         fprintf( f,
1131                 _( "\t-s  suppress default rule to ECHO unmatched text\n" ) );
1132
1133         if ( ! did_outfilename )
1134                 {
1135                 sprintf( outfile_path, outfile_template,
1136                         prefix, C_plus_plus ? "cc" : "c" );
1137                 outfilename = outfile_path;
1138                 }
1139
1140         fprintf( f,
1141                 _( "\t-t  write generated scanner on stdout instead of %s\n" ),
1142                 outfilename );
1143
1144         fprintf( f,
1145                 _( "\t-v  write summary of scanner statistics to f\n" ) );
1146         fprintf( f, _( "\t-w  do not generate warnings\n" ) );
1147         fprintf( f, _( "\t-B  generate batch scanner (opposite of -I)\n" ) );
1148         fprintf( f,
1149                 _( "\t-F  use alternative fast scanner representation\n" ) );
1150         fprintf( f,
1151                 _( "\t-I  generate interactive scanner (opposite of -B)\n" ) );
1152         fprintf( f, _( "\t-L  suppress #line directives in scanner\n" ) );
1153         fprintf( f, _( "\t-T  %s should run in trace mode\n" ), program_name );
1154         fprintf( f, _( "\t-V  report %s version\n" ), program_name );
1155         fprintf( f, _( "\t-7  generate 7-bit scanner\n" ) );
1156         fprintf( f, _( "\t-8  generate 8-bit scanner\n" ) );
1157         fprintf( f, _( "\t-+  generate C++ scanner class\n" ) );
1158         fprintf( f, _( "\t-?  produce this help message\n" ) );
1159         fprintf( f,
1160 _( "\t-C  specify degree of table compression (default is -Cem):\n" ) );
1161         fprintf( f,
1162 _( "\t\t-Ca  trade off larger tables for better memory alignment\n" ) );
1163         fprintf( f, _( "\t\t-Ce  construct equivalence classes\n" ) );
1164         fprintf( f,
1165 _( "\t\t-Cf  do not compress scanner tables; use -f representation\n" ) );
1166         fprintf( f,
1167 _( "\t\t-CF  do not compress scanner tables; use -F representation\n" ) );
1168         fprintf( f, _( "\t\t-Cm  construct meta-equivalence classes\n" ) );
1169         fprintf( f,
1170         _( "\t\t-Cr  use read() instead of stdio for scanner input\n" ) );
1171         fprintf( f, _( "\t-o  specify output filename\n" ) );
1172         fprintf( f, _( "\t-P  specify scanner prefix other than \"yy\"\n" ) );
1173         fprintf( f, _( "\t-S  specify skeleton file\n" ) );
1174         fprintf( f, _( "\t--help     produce this help message\n" ) );
1175         fprintf( f, _( "\t--version  report %s version\n" ), program_name );
1176         }