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 1.3 2003/10/04 20:36:47 hmp 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(int argc, char **argv)
 132         {
 133         int i;
 134
 135 #ifdef THINK_C
 136         argc = ccommand( &argv );
 137 #endif
 138 #ifdef NEED_ARGV_FIXUP
 139         argv_fixup( &argc, &argv );
 140 #endif
 141
 142         flexinit( argc, argv );
 143
 144         readin();
 145
 146         ntod();
 147
 148         for ( i = 1; i <= num_rules; ++i )
 149                 if ( ! rule_useful[i] && i != default_rule )
 150                         line_warning( _( "rule cannot be matched" ),
 151                                         rule_linenum[i] );
 152
 153         if ( spprdflt && ! reject && rule_useful[default_rule] )
 154                 line_warning(
 155                         _( "-s option given but default rule can be matched" ),
 156                         rule_linenum[default_rule] );
 157
 158         /* Generate the C state transition tables from the DFA. */
 159         make_tables();
 160
 161         /* Note, flexend does not return.  It exits with its argument
 162          * as status.
 163          */
 164         flexend( 0 );
 165
 166         return 0;       /* keep compilers/lint happy */
 167         }
 168
 169
 170 /* check_options - check user-specified options */
 171
 172 void check_options(void)
 173         {
 174         int i;
 175
 176         if ( lex_compat )
 177                 {
 178                 if ( C_plus_plus )
 179                         flexerror( _( "Can't use -+ with -l option" ) );
 180
 181                 if ( fulltbl || fullspd )
 182                         flexerror( _( "Can't use -f or -F with -l option" ) );
 183
 184                 /* Don't rely on detecting use of yymore() and REJECT,
 185                  * just assume they'll be used.
 186                  */
 187                 yymore_really_used = reject_really_used = true;
 188
 189                 yytext_is_array = true;
 190                 do_yylineno = true;
 191                 use_read = false;
 192                 }
 193
 194         if ( do_yylineno )
 195                 /* This should really be "maintain_backup_tables = true" */
 196                 reject_really_used = true;
 197
 198         if ( csize == unspecified )
 199                 {
 200                 if ( (fulltbl || fullspd) && ! useecs )
 201                         csize = DEFAULT_CSIZE;
 202                 else
 203                         csize = CSIZE;
 204                 }
 205
 206         if ( interactive == unspecified )
 207                 {
 208                 if ( fulltbl || fullspd )
 209                         interactive = false;
 210                 else
 211                         interactive = true;
 212                 }
 213
 214         if ( fulltbl || fullspd )
 215                 {
 216                 if ( usemecs )
 217                         flexerror(
 218                         _( "-Cf/-CF and -Cm don't make sense together" ) );
 219
 220                 if ( interactive )
 221                         flexerror( _( "-Cf/-CF and -I are incompatible" ) );
 222
 223                 if ( lex_compat )
 224                         flexerror(
 225                 _( "-Cf/-CF are incompatible with lex-compatibility mode" ) );
 226
 227                 if ( do_yylineno )
 228                         flexerror(
 229                         _( "-Cf/-CF and %option yylineno are incompatible" ) );
 230
 231                 if ( fulltbl && fullspd )
 232                         flexerror( _( "-Cf and -CF are mutually exclusive" ) );
 233                 }
 234
 235         if ( C_plus_plus && fullspd )
 236                 flexerror( _( "Can't use -+ with -CF option" ) );
 237
 238         if ( C_plus_plus && yytext_is_array )
 239                 {
 240                 warn( _( "%array incompatible with -+ option" ) );
 241                 yytext_is_array = false;
 242                 }
 243
 244         if ( useecs )
 245                 { /* Set up doubly-linked equivalence classes. */
 246
 247                 /* We loop all the way up to csize, since ecgroup[csize] is
 248                  * the position used for NUL characters.
 249                  */
 250                 ecgroup[1] = NIL;
 251
 252                 for ( i = 2; i <= csize; ++i )
 253                         {
 254                         ecgroup[i] = i - 1;
 255                         nextecm[i - 1] = i;
 256                         }
 257
 258                 nextecm[csize] = NIL;
 259                 }
 260
 261         else
 262                 {
 263                 /* Put everything in its own equivalence class. */
 264                 for ( i = 1; i <= csize; ++i )
 265                         {
 266                         ecgroup[i] = i;
 267                         nextecm[i] = BAD_SUBSCRIPT;     /* to catch errors */
 268                         }
 269                 }
 270
 271         if ( ! use_stdout )
 272                 {
 273                 FILE *prev_stdout;
 274
 275                 if ( ! did_outfilename )
 276                         {
 277                         char *suffix;
 278
 279                         if ( C_plus_plus )
 280                                 suffix = "cc";
 281                         else
 282                                 suffix = "c";
 283
 284                         sprintf( outfile_path, outfile_template,
 285                                 prefix, suffix );
 286
 287                         outfilename = outfile_path;
 288                         }
 289
 290                 prev_stdout = freopen( outfilename, "w", stdout );
 291
 292                 if ( prev_stdout == NULL )
 293                         lerrsf( _( "could not create %s" ), outfilename );
 294
 295                 outfile_created = 1;
 296                 }
 297
 298         if ( skelname && (skelfile = fopen( skelname, "r" )) == NULL )
 299                 lerrsf( _( "can't open skeleton file %s" ), skelname );
 300
 301         if ( strcmp( prefix, "yy" ) )
 302                 {
 303 #define GEN_PREFIX(name) out_str3( "#define yy%s %s%s\n", name, prefix, name )
 304                 if ( C_plus_plus )
 305                         GEN_PREFIX( "FlexLexer" );
 306                 else
 307                         {
 308                         GEN_PREFIX( "_create_buffer" );
 309                         GEN_PREFIX( "_delete_buffer" );
 310                         GEN_PREFIX( "_scan_buffer" );
 311                         GEN_PREFIX( "_scan_string" );
 312                         GEN_PREFIX( "_scan_bytes" );
 313                         GEN_PREFIX( "_flex_debug" );
 314                         GEN_PREFIX( "_init_buffer" );
 315                         GEN_PREFIX( "_flush_buffer" );
 316                         GEN_PREFIX( "_load_buffer_state" );
 317                         GEN_PREFIX( "_switch_to_buffer" );
 318                         GEN_PREFIX( "in" );
 319                         GEN_PREFIX( "leng" );
 320                         GEN_PREFIX( "lex" );
 321                         GEN_PREFIX( "out" );
 322                         GEN_PREFIX( "restart" );
 323                         GEN_PREFIX( "text" );
 324
 325                         if ( do_yylineno )
 326                                 GEN_PREFIX( "lineno" );
 327                         }
 328
 329                 if ( do_yywrap )
 330                         GEN_PREFIX( "wrap" );
 331
 332                 outn( "" );
 333                 }
 334
 335         if ( did_outfilename )
 336                 line_directive_out( stdout, 0 );
 337
 338         skelout();
 339         }
 340
 341
 342 /* flexend - terminate flex
 343  *
 344  * note
 345  *    This routine does not return.
 346  */
 347
 348 void flexend(int exit_status)
 349         {
 350         int tblsiz;
 351         int unlink();
 352
 353         if ( skelfile != NULL )
 354                 {
 355                 if ( ferror( skelfile ) )
 356                         lerrsf( _( "input error reading skeleton file %s" ),
 357                                 skelname );
 358
 359                 else if ( fclose( skelfile ) )
 360                         lerrsf( _( "error closing skeleton file %s" ),
 361                                 skelname );
 362                 }
 363
 364         if ( exit_status != 0 && outfile_created )
 365                 {
 366                 if ( ferror( stdout ) )
 367                         lerrsf( _( "error writing output file %s" ),
 368                                 outfilename );
 369
 370                 else if ( fclose( stdout ) )
 371                         lerrsf( _( "error closing output file %s" ),
 372                                 outfilename );
 373
 374                 else if ( unlink( outfilename ) )
 375                         lerrsf( _( "error deleting output file %s" ),
 376                                 outfilename );
 377                 }
 378
 379         if ( backing_up_report && backing_up_file )
 380                 {
 381                 if ( num_backing_up == 0 )
 382                         fprintf( backing_up_file, _( "No backing up.\n" ) );
 383                 else if ( fullspd || fulltbl )
 384                         fprintf( backing_up_file,
 385                                 _( "%d backing up (non-accepting) states.\n" ),
 386                                 num_backing_up );
 387                 else
 388                         fprintf( backing_up_file,
 389                                 _( "Compressed tables always back up.\n" ) );
 390
 391                 if ( ferror( backing_up_file ) )
 392                         lerrsf( _( "error writing backup file %s" ),
 393                                 backing_name );
 394
 395                 else if ( fclose( backing_up_file ) )
 396                         lerrsf( _( "error closing backup file %s" ),
 397                                 backing_name );
 398                 }
 399
 400         if ( printstats )
 401                 {
 402                 fprintf( stderr, _( "%s version %s usage statistics:\n" ),
 403                         program_name, flex_version );
 404
 405                 fprintf( stderr, _( "  scanner options: -" ) );
 406
 407                 if ( C_plus_plus )
 408                         putc( '+', stderr );
 409                 if ( backing_up_report )
 410                         putc( 'b', stderr );
 411                 if ( ddebug )
 412                         putc( 'd', stderr );
 413                 if ( caseins )
 414                         putc( 'i', stderr );
 415                 if ( lex_compat )
 416                         putc( 'l', stderr );
 417                 if ( performance_report > 0 )
 418                         putc( 'p', stderr );
 419                 if ( performance_report > 1 )
 420                         putc( 'p', stderr );
 421                 if ( spprdflt )
 422                         putc( 's', stderr );
 423                 if ( use_stdout )
 424                         putc( 't', stderr );
 425                 if ( printstats )
 426                         putc( 'v', stderr );    /* always true! */
 427                 if ( nowarn )
 428                         putc( 'w', stderr );
 429                 if ( interactive == false )
 430                         putc( 'B', stderr );
 431                 if ( interactive == true )
 432                         putc( 'I', stderr );
 433                 if ( ! gen_line_dirs )
 434                         putc( 'L', stderr );
 435                 if ( trace )
 436                         putc( 'T', stderr );
 437
 438                 if ( csize == unspecified )
 439                         /* We encountered an error fairly early on, so csize
 440                          * never got specified.  Define it now, to prevent
 441                          * bogus table sizes being written out below.
 442                          */
 443                         csize = 256;
 444
 445                 if ( csize == 128 )
 446                         putc( '7', stderr );
 447                 else
 448                         putc( '8', stderr );
 449
 450                 fprintf( stderr, " -C" );
 451
 452                 if ( long_align )
 453                         putc( 'a', stderr );
 454                 if ( fulltbl )
 455                         putc( 'f', stderr );
 456                 if ( fullspd )
 457                         putc( 'F', stderr );
 458                 if ( useecs )
 459                         putc( 'e', stderr );
 460                 if ( usemecs )
 461                         putc( 'm', stderr );
 462                 if ( use_read )
 463                         putc( 'r', stderr );
 464
 465                 if ( did_outfilename )
 466                         fprintf( stderr, " -o%s", outfilename );
 467
 468                 if ( skelname )
 469                         fprintf( stderr, " -S%s", skelname );
 470
 471                 if ( strcmp( prefix, "yy" ) )
 472                         fprintf( stderr, " -P%s", prefix );
 473
 474                 putc( '\n', stderr );
 475
 476                 fprintf( stderr, _( "  %d/%d NFA states\n" ),
 477                         lastnfa, current_mns );
 478                 fprintf( stderr, _( "  %d/%d DFA states (%d words)\n" ),
 479                         lastdfa, current_max_dfas, totnst );
 480                 fprintf( stderr, _( "  %d rules\n" ),
 481                 num_rules + num_eof_rules - 1 /* - 1 for def. rule */ );
 482
 483                 if ( num_backing_up == 0 )
 484                         fprintf( stderr, _( "  No backing up\n" ) );
 485                 else if ( fullspd || fulltbl )
 486                         fprintf( stderr,
 487                         _( "  %d backing-up (non-accepting) states\n" ),
 488                                 num_backing_up );
 489                 else
 490                         fprintf( stderr,
 491                                 _( "  Compressed tables always back-up\n" ) );
 492
 493                 if ( bol_needed )
 494                         fprintf( stderr,
 495                                 _( "  Beginning-of-line patterns used\n" ) );
 496
 497                 fprintf( stderr, _( "  %d/%d start conditions\n" ), lastsc,
 498                         current_max_scs );
 499                 fprintf( stderr,
 500                         _( "  %d epsilon states, %d double epsilon states\n" ),
 501                         numeps, eps2 );
 502
 503                 if ( lastccl == 0 )
 504                         fprintf( stderr, _( "  no character classes\n" ) );
 505                 else
 506                         fprintf( stderr,
 507 _( "  %d/%d character classes needed %d/%d words of storage, %d reused\n" ),
 508                                 lastccl, current_maxccls,
 509                                 cclmap[lastccl] + ccllen[lastccl],
 510                                 current_max_ccl_tbl_size, cclreuse );
 511
 512                 fprintf( stderr, _( "  %d state/nextstate pairs created\n" ),
 513                         numsnpairs );
 514                 fprintf( stderr, _( "  %d/%d unique/duplicate transitions\n" ),
 515                         numuniq, numdup );
 516
 517                 if ( fulltbl )
 518                         {
 519                         tblsiz = lastdfa * numecs;
 520                         fprintf( stderr, _( "  %d table entries\n" ), tblsiz );
 521                         }
 522
 523                 else
 524                         {
 525                         tblsiz = 2 * (lastdfa + numtemps) + 2 * tblend;
 526
 527                         fprintf( stderr,
 528                                 _( "  %d/%d base-def entries created\n" ),
 529                                 lastdfa + numtemps, current_max_dfas );
 530                         fprintf( stderr,
 531                         _( "  %d/%d (peak %d) nxt-chk entries created\n" ),
 532                                 tblend, current_max_xpairs, peakpairs );
 533                         fprintf( stderr,
 534                 _( "  %d/%d (peak %d) template nxt-chk entries created\n" ),
 535                                 numtemps * nummecs,
 536                                 current_max_template_xpairs,
 537                                 numtemps * numecs );
 538                         fprintf( stderr, _( "  %d empty table entries\n" ),
 539                                 nummt );
 540                         fprintf( stderr, _( "  %d protos created\n" ),
 541                                 numprots );
 542                         fprintf( stderr,
 543                                 _( "  %d templates created, %d uses\n" ),
 544                                 numtemps, tmpuses );
 545                         }
 546
 547                 if ( useecs )
 548                         {
 549                         tblsiz = tblsiz + csize;
 550                         fprintf( stderr,
 551                                 _( "  %d/%d equivalence classes created\n" ),
 552                                 numecs, csize );
 553                         }
 554
 555                 if ( usemecs )
 556                         {
 557                         tblsiz = tblsiz + numecs;
 558                         fprintf( stderr,
 559                         _( "  %d/%d meta-equivalence classes created\n" ),
 560                                 nummecs, csize );
 561                         }
 562
 563                 fprintf( stderr,
 564                 _( "  %d (%d saved) hash collisions, %d DFAs equal\n" ),
 565                         hshcol, hshsave, dfaeql );
 566                 fprintf( stderr, _( "  %d sets of reallocations needed\n" ),
 567                         num_reallocs );
 568                 fprintf( stderr, _( "  %d total table entries needed\n" ),
 569                         tblsiz );
 570                 }
 571
 572         exit( exit_status );
 573         }
 574
 575
 576 /* flexinit - initialize flex */
 577
 578 void flexinit(int argc, char **argv)
 579         {
 580         int i, sawcmpflag;
 581         char *arg;
 582
 583         printstats = syntaxerror = trace = spprdflt = caseins = false;
 584         lex_compat = C_plus_plus = backing_up_report = ddebug = fulltbl = false;
 585         fullspd = long_align = nowarn = yymore_used = continued_action = false;
 586         do_yylineno = yytext_is_array = in_rule = reject = do_stdinit = false;
 587         yymore_really_used = reject_really_used = unspecified;
 588         interactive = csize = unspecified;
 589         do_yywrap = gen_line_dirs = usemecs = useecs = true;
 590         performance_report = 0;
 591         did_outfilename = 0;
 592         prefix = "yy";
 593         yyclass = 0;
 594         use_read = use_stdout = false;
 595
 596         sawcmpflag = false;
 597
 598         /* Initialize dynamic array for holding the rule actions. */
 599         action_size = 2048;     /* default size of action array in bytes */
 600         action_array = allocate_character_array( action_size );
 601         defs1_offset = prolog_offset = action_offset = action_index = 0;
 602         action_array[0] = '\0';
 603
 604         program_name = argv[0];
 605
 606         if ( program_name[0] != '\0' &&
 607              program_name[strlen( program_name ) - 1] == '+' )
 608                 C_plus_plus = true;
 609
 610         /* read flags */
 611         for ( --argc, ++argv; argc ; --argc, ++argv )
 612                 {
 613                 arg = argv[0];
 614
 615                 if ( arg[0] != '-' || arg[1] == '\0' )
 616                         break;
 617
 618                 if ( arg[1] == '-' )
 619                         { /* --option */
 620                         if ( ! strcmp( arg, "--help" ) )
 621                                 arg = "-h";
 622
 623                         else if ( ! strcmp( arg, "--version" ) )
 624                                 arg = "-V";
 625
 626                         else if ( ! strcmp( arg, "--" ) )
 627                                 { /* end of options */
 628                                 --argc;
 629                                 ++argv;
 630                                 break;
 631                                 }
 632                         }
 633
 634                 for ( i = 1; arg[i] != '\0'; ++i )
 635                         switch ( arg[i] )
 636                                 {
 637                                 case '+':
 638                                         C_plus_plus = true;
 639                                         break;
 640
 641                                 case 'B':
 642                                         interactive = false;
 643                                         break;
 644
 645                                 case 'b':
 646                                         backing_up_report = true;
 647                                         break;
 648
 649                                 case 'c':
 650                                         break;
 651
 652                                 case 'C':
 653                                         if ( i != 1 )
 654                                                 flexerror(
 655                                 _( "-C flag must be given separately" ) );
 656
 657                                         if ( ! sawcmpflag )
 658                                                 {
 659                                                 useecs = false;
 660                                                 usemecs = false;
 661                                                 fulltbl = false;
 662                                                 sawcmpflag = true;
 663                                                 }
 664
 665                                         for ( ++i; arg[i] != '\0'; ++i )
 666                                                 switch ( arg[i] )
 667                                                         {
 668                                                         case 'a':
 669                                                                 long_align =
 670                                                                         true;
 671                                                                 break;
 672
 673                                                         case 'e':
 674                                                                 useecs = true;
 675                                                                 break;
 676
 677                                                         case 'F':
 678                                                                 fullspd = true;
 679                                                                 break;
 680
 681                                                         case 'f':
 682                                                                 fulltbl = true;
 683                                                                 break;
 684
 685                                                         case 'm':
 686                                                                 usemecs = true;
 687                                                                 break;
 688
 689                                                         case 'r':
 690                                                                 use_read = true;
 691                                                                 break;
 692
 693                                                         default:
 694                                                                 lerrif(
 695                                                 _( "unknown -C option '%c'" ),
 696                                                                 (int) arg[i] );
 697                                                                 break;
 698                                                         }
 699
 700                                         goto get_next_arg;
 701
 702                                 case 'd':
 703                                         ddebug = true;
 704                                         break;
 705
 706                                 case 'f':
 707                                         useecs = usemecs = false;
 708                                         use_read = fulltbl = true;
 709                                         break;
 710
 711                                 case 'F':
 712                                         useecs = usemecs = false;
 713                                         use_read = fullspd = true;
 714                                         break;
 715
 716                                 case '?':
 717                                 case 'h':
 718                                         usage();
 719                                         exit( 0 );
 720
 721                                 case 'I':
 722                                         interactive = true;
 723                                         break;
 724
 725                                 case 'i':
 726                                         caseins = true;
 727                                         break;
 728
 729                                 case 'l':
 730                                         lex_compat = true;
 731                                         break;
 732
 733                                 case 'L':
 734                                         gen_line_dirs = false;
 735                                         break;
 736
 737                                 case 'n':
 738                                         /* Stupid do-nothing deprecated
 739                                          * option.
 740                                          */
 741                                         break;
 742
 743                                 case 'o':
 744                                         if ( i != 1 )
 745                                                 flexerror(
 746                                 _( "-o flag must be given separately" ) );
 747
 748                                         outfilename = arg + i + 1;
 749                                         did_outfilename = 1;
 750                                         goto get_next_arg;
 751
 752                                 case 'P':
 753                                         if ( i != 1 )
 754                                                 flexerror(
 755                                 _( "-P flag must be given separately" ) );
 756
 757                                         prefix = arg + i + 1;
 758                                         goto get_next_arg;
 759
 760                                 case 'p':
 761                                         ++performance_report;
 762                                         break;
 763
 764                                 case 'S':
 765                                         if ( i != 1 )
 766                                                 flexerror(
 767                                 _( "-S flag must be given separately" ) );
 768
 769                                         skelname = arg + i + 1;
 770                                         goto get_next_arg;
 771
 772                                 case 's':
 773                                         spprdflt = true;
 774                                         break;
 775
 776                                 case 't':
 777                                         use_stdout = true;
 778                                         break;
 779
 780                                 case 'T':
 781                                         trace = true;
 782                                         break;
 783
 784                                 case 'v':
 785                                         printstats = true;
 786                                         break;
 787
 788                                 case 'V':
 789                                         printf( _( "%s version %s\n" ),
 790                                                 program_name, flex_version );
 791                                         exit( 0 );
 792
 793                                 case 'w':
 794                                         nowarn = true;
 795                                         break;
 796
 797                                 case '7':
 798                                         csize = 128;
 799                                         break;
 800
 801                                 case '8':
 802                                         csize = CSIZE;
 803                                         break;
 804
 805                                 default:
 806                                         fprintf( stderr,
 807                 _( "%s: unknown flag '%c'.  For usage, try\n\t%s --help\n" ),
 808                                                 program_name, (int) arg[i],
 809                                                 program_name );
 810                                         exit( 1 );
 811                                 }
 812
 813                 /* Used by -C, -S, -o, and -P flags in lieu of a "continue 2"
 814                  * control.
 815                  */
 816                 get_next_arg: ;
 817                 }
 818
 819         num_input_files = argc;
 820         input_files = argv;
 821         set_input_file( num_input_files > 0 ? input_files[0] : NULL );
 822
 823         lastccl = lastsc = lastdfa = lastnfa = 0;
 824         num_rules = num_eof_rules = default_rule = 0;
 825         numas = numsnpairs = tmpuses = 0;
 826         numecs = numeps = eps2 = num_reallocs = hshcol = dfaeql = totnst = 0;
 827         numuniq = numdup = hshsave = eofseen = datapos = dataline = 0;
 828         num_backing_up = onesp = numprots = 0;
 829         variable_trailing_context_rules = bol_needed = false;
 830
 831         out_linenum = linenum = sectnum = 1;
 832         firstprot = NIL;
 833
 834         /* Used in mkprot() so that the first proto goes in slot 1
 835          * of the proto queue.
 836          */
 837         lastprot = 1;
 838
 839         set_up_initial_allocations();
 840         }
 841
 842
 843 /* readin - read in the rules section of the input file(s) */
 844
 845 void readin(void)
 846         {
 847         static char yy_stdinit[] = "FILE *yyin = stdin, *yyout = stdout;";
 848         static char yy_nostdinit[] =
 849                 "FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;";
 850
 851         line_directive_out( (FILE *) 0, 1 );
 852
 853         if ( yyparse() )
 854                 {
 855                 pinpoint_message( _( "fatal parse error" ) );
 856                 flexend( 1 );
 857                 }
 858
 859         if ( syntaxerror )
 860                 flexend( 1 );
 861
 862         if ( backing_up_report )
 863                 {
 864                 backing_up_file = fopen( backing_name, "w" );
 865                 if ( backing_up_file == NULL )
 866                         lerrsf(
 867                         _( "could not create backing-up info file %s" ),
 868                                 backing_name );
 869                 }
 870
 871         else
 872                 backing_up_file = NULL;
 873
 874         if ( yymore_really_used == true )
 875                 yymore_used = true;
 876         else if ( yymore_really_used == false )
 877                 yymore_used = false;
 878
 879         if ( reject_really_used == true )
 880                 reject = true;
 881         else if ( reject_really_used == false )
 882                 reject = false;
 883
 884         if ( performance_report > 0 )
 885                 {
 886                 if ( lex_compat )
 887                         {
 888                         fprintf( stderr,
 889 _( "-l AT&T lex compatibility option entails a large performance penalty\n" ) );
 890                         fprintf( stderr,
 891 _( " and may be the actual source of other reported performance penalties\n" ) );
 892                         }
 893
 894                 else if ( do_yylineno )
 895                         {
 896                         fprintf( stderr,
 897         _( "%%option yylineno entails a large performance penalty\n" ) );
 898                         }
 899
 900                 if ( performance_report > 1 )
 901                         {
 902                         if ( interactive )
 903                                 fprintf( stderr,
 904         _( "-I (interactive) entails a minor performance penalty\n" ) );
 905
 906                         if ( yymore_used )
 907                                 fprintf( stderr,
 908                 _( "yymore() entails a minor performance penalty\n" ) );
 909                         }
 910
 911                 if ( reject )
 912                         fprintf( stderr,
 913                         _( "REJECT entails a large performance penalty\n" ) );
 914
 915                 if ( variable_trailing_context_rules )
 916                         fprintf( stderr,
 917 _( "Variable trailing context rules entail a large performance penalty\n" ) );
 918                 }
 919
 920         if ( reject )
 921                 real_reject = true;
 922
 923         if ( variable_trailing_context_rules )
 924                 reject = true;
 925
 926         if ( (fulltbl || fullspd) && reject )
 927                 {
 928                 if ( real_reject )
 929                         flexerror(
 930                                 _( "REJECT cannot be used with -f or -F" ) );
 931                 else if ( do_yylineno )
 932                         flexerror(
 933                         _( "%option yylineno cannot be used with -f or -F" ) );
 934                 else
 935                         flexerror(
 936         _( "variable trailing context rules cannot be used with -f or -F" ) );
 937                 }
 938
 939         if ( reject )
 940                 outn( "\n#define YY_USES_REJECT" );
 941
 942         if ( ! do_yywrap )
 943                 {
 944                 outn( "\n#define yywrap() 1" );
 945                 outn( "#define YY_SKIP_YYWRAP" );
 946                 }
 947
 948         if ( ddebug )
 949                 outn( "\n#define FLEX_DEBUG" );
 950
 951         if ( csize == 256 )
 952                 outn( "typedef unsigned char YY_CHAR;" );
 953         else
 954                 outn( "typedef char YY_CHAR;" );
 955
 956         if ( C_plus_plus )
 957                 {
 958                 outn( "#define yytext_ptr yytext" );
 959
 960                 if ( interactive )
 961                         outn( "#define YY_INTERACTIVE" );
 962                 }
 963
 964         else
 965                 {
 966                 if ( do_stdinit )
 967                         {
 968                         outn( "#ifdef VMS" );
 969                         outn( "#ifndef __VMS_POSIX" );
 970                         outn( yy_nostdinit );
 971                         outn( "#else" );
 972                         outn( yy_stdinit );
 973                         outn( "#endif" );
 974                         outn( "#else" );
 975                         outn( yy_stdinit );
 976                         outn( "#endif" );
 977                         }
 978
 979                 else
 980                         outn( yy_nostdinit );
 981                 }
 982
 983         if ( fullspd )
 984                 outn( "typedef yyconst struct yy_trans_info *yy_state_type;" );
 985         else if ( ! C_plus_plus )
 986                 outn( "typedef int yy_state_type;" );
 987
 988         if ( ddebug )
 989                 outn( "\n#define FLEX_DEBUG" );
 990
 991         if ( lex_compat )
 992                 outn( "#define YY_FLEX_LEX_COMPAT" );
 993
 994         if ( do_yylineno && ! C_plus_plus )
 995                 {
 996                 outn( "extern int yylineno;" );
 997                 outn( "int yylineno = 1;" );
 998                 }
 999
1000         if ( C_plus_plus )
1001                 {
1002                 outn( "\n#include <FlexLexer.h>" );
1003
1004                 if ( yyclass )
1005                         {
1006                         outn( "int yyFlexLexer::yylex()" );
1007                         outn( "\t{" );
1008                         outn(
1009 "\tLexerError( \"yyFlexLexer::yylex invoked but %option yyclass used\" );" );
1010                         outn( "\treturn 0;" );
1011                         outn( "\t}" );
1012
1013                         out_str( "\n#define YY_DECL int %s::yylex()\n",
1014                                 yyclass );
1015                         }
1016                 }
1017
1018         else
1019                 {
1020                 if ( yytext_is_array )
1021                         outn( "extern char yytext[];\n" );
1022
1023                 else
1024                         {
1025                         outn( "extern char *yytext;" );
1026                         outn( "#define yytext_ptr yytext" );
1027                         }
1028
1029                 if ( yyclass )
1030                         flexerror(
1031                 _( "%option yyclass only meaningful for C++ scanners" ) );
1032                 }
1033
1034         if ( useecs )
1035                 numecs = cre8ecs( nextecm, ecgroup, csize );
1036         else
1037                 numecs = csize;
1038
1039         /* Now map the equivalence class for NUL to its expected place. */
1040         ecgroup[0] = ecgroup[csize];
1041         NUL_ec = ABS( ecgroup[0] );
1042
1043         if ( useecs )
1044                 ccl2ecl();
1045         }
1046
1047
1048 /* set_up_initial_allocations - allocate memory for internal tables */
1049
1050 void set_up_initial_allocations(void)
1051         {
1052         current_mns = INITIAL_MNS;
1053         firstst = allocate_integer_array( current_mns );
1054         lastst = allocate_integer_array( current_mns );
1055         finalst = allocate_integer_array( current_mns );
1056         transchar = allocate_integer_array( current_mns );
1057         trans1 = allocate_integer_array( current_mns );
1058         trans2 = allocate_integer_array( current_mns );
1059         accptnum = allocate_integer_array( current_mns );
1060         assoc_rule = allocate_integer_array( current_mns );
1061         state_type = allocate_integer_array( current_mns );
1062
1063         current_max_rules = INITIAL_MAX_RULES;
1064         rule_type = allocate_integer_array( current_max_rules );
1065         rule_linenum = allocate_integer_array( current_max_rules );
1066         rule_useful = allocate_integer_array( current_max_rules );
1067
1068         current_max_scs = INITIAL_MAX_SCS;
1069         scset = allocate_integer_array( current_max_scs );
1070         scbol = allocate_integer_array( current_max_scs );
1071         scxclu = allocate_integer_array( current_max_scs );
1072         sceof = allocate_integer_array( current_max_scs );
1073         scname = allocate_char_ptr_array( current_max_scs );
1074
1075         current_maxccls = INITIAL_MAX_CCLS;
1076         cclmap = allocate_integer_array( current_maxccls );
1077         ccllen = allocate_integer_array( current_maxccls );
1078         cclng = allocate_integer_array( current_maxccls );
1079
1080         current_max_ccl_tbl_size = INITIAL_MAX_CCL_TBL_SIZE;
1081         ccltbl = allocate_Character_array( current_max_ccl_tbl_size );
1082
1083         current_max_dfa_size = INITIAL_MAX_DFA_SIZE;
1084
1085         current_max_xpairs = INITIAL_MAX_XPAIRS;
1086         nxt = allocate_integer_array( current_max_xpairs );
1087         chk = allocate_integer_array( current_max_xpairs );
1088
1089         current_max_template_xpairs = INITIAL_MAX_TEMPLATE_XPAIRS;
1090         tnxt = allocate_integer_array( current_max_template_xpairs );
1091
1092         current_max_dfas = INITIAL_MAX_DFAS;
1093         base = allocate_integer_array( current_max_dfas );
1094         def = allocate_integer_array( current_max_dfas );
1095         dfasiz = allocate_integer_array( current_max_dfas );
1096         accsiz = allocate_integer_array( current_max_dfas );
1097         dhash = allocate_integer_array( current_max_dfas );
1098         dss = allocate_int_ptr_array( current_max_dfas );
1099         dfaacc = allocate_dfaacc_union( current_max_dfas );
1100
1101         nultrans = (int *) 0;
1102         }
1103
1104
1105 void usage(void)
1106         {
1107         FILE *f = stdout;
1108
1109         fprintf( f,
1110 _( "%s [-bcdfhilnpstvwBFILTV78+? -C[aefFmr] -ooutput -Pprefix -Sskeleton]\n" ),
1111                 program_name );
1112         fprintf( f, _( "\t[--help --version] [file ...]\n" ) );
1113
1114         fprintf( f, _( "\t-b  generate backing-up information to %s\n" ),
1115                 backing_name );
1116         fprintf( f, _( "\t-c  do-nothing POSIX option\n" ) );
1117         fprintf( f, _( "\t-d  turn on debug mode in generated scanner\n" ) );
1118         fprintf( f, _( "\t-f  generate fast, large scanner\n" ) );
1119         fprintf( f, _( "\t-h  produce this help message\n" ) );
1120         fprintf( f, _( "\t-i  generate case-insensitive scanner\n" ) );
1121         fprintf( f, _( "\t-l  maximal compatibility with original lex\n" ) );
1122         fprintf( f, _( "\t-n  do-nothing POSIX option\n" ) );
1123         fprintf( f, _( "\t-p  generate performance report to stderr\n" ) );
1124         fprintf( f,
1125                 _( "\t-s  suppress default rule to ECHO unmatched text\n" ) );
1126
1127         if ( ! did_outfilename )
1128                 {
1129                 sprintf( outfile_path, outfile_template,
1130                         prefix, C_plus_plus ? "cc" : "c" );
1131                 outfilename = outfile_path;
1132                 }
1133
1134         fprintf( f,
1135                 _( "\t-t  write generated scanner on stdout instead of %s\n" ),
1136                 outfilename );
1137
1138         fprintf( f,
1139                 _( "\t-v  write summary of scanner statistics to f\n" ) );
1140         fprintf( f, _( "\t-w  do not generate warnings\n" ) );
1141         fprintf( f, _( "\t-B  generate batch scanner (opposite of -I)\n" ) );
1142         fprintf( f,
1143                 _( "\t-F  use alternative fast scanner representation\n" ) );
1144         fprintf( f,
1145                 _( "\t-I  generate interactive scanner (opposite of -B)\n" ) );
1146         fprintf( f, _( "\t-L  suppress #line directives in scanner\n" ) );
1147         fprintf( f, _( "\t-T  %s should run in trace mode\n" ), program_name );
1148         fprintf( f, _( "\t-V  report %s version\n" ), program_name );
1149         fprintf( f, _( "\t-7  generate 7-bit scanner\n" ) );
1150         fprintf( f, _( "\t-8  generate 8-bit scanner\n" ) );
1151         fprintf( f, _( "\t-+  generate C++ scanner class\n" ) );
1152         fprintf( f, _( "\t-?  produce this help message\n" ) );
1153         fprintf( f,
1154 _( "\t-C  specify degree of table compression (default is -Cem):\n" ) );
1155         fprintf( f,
1156 _( "\t\t-Ca  trade off larger tables for better memory alignment\n" ) );
1157         fprintf( f, _( "\t\t-Ce  construct equivalence classes\n" ) );
1158         fprintf( f,
1159 _( "\t\t-Cf  do not compress scanner tables; use -f representation\n" ) );
1160         fprintf( f,
1161 _( "\t\t-CF  do not compress scanner tables; use -F representation\n" ) );
1162         fprintf( f, _( "\t\t-Cm  construct meta-equivalence classes\n" ) );
1163         fprintf( f,
1164         _( "\t\t-Cr  use read() instead of stdio for scanner input\n" ) );
1165         fprintf( f, _( "\t-o  specify output filename\n" ) );
1166         fprintf( f, _( "\t-P  specify scanner prefix other than \"yy\"\n" ) );
1167         fprintf( f, _( "\t-S  specify skeleton file\n" ) );
1168         fprintf( f, _( "\t--help     produce this help message\n" ) );
1169         fprintf( f, _( "\t--version  report %s version\n" ), program_name );
1170         }