usr.bin/unifdef/unifdef.c

   1 /*
   2  * Copyright (c) 2002 - 2005 Tony Finch <dot@dotat.at>.  All rights reserved.
   3  *
   4  * This code is derived from software contributed to Berkeley by Dave Yost.
   5  * It was rewritten to support ANSI C by Tony Finch. The original version of
   6  * unifdef carried the following copyright notice. None of its code remains
   7  * in this version (though some of the names remain).
   8  *
   9  * Copyright (c) 1985, 1993
  10  *      The Regents of the University of California.  All rights reserved.
  11  *
  12  * Redistribution and use in source and binary forms, with or without
  13  * modification, are permitted provided that the following conditions
  14  * are met:
  15  * 1. Redistributions of source code must retain the above copyright
  16  *    notice, this list of conditions and the following disclaimer.
  17  * 2. Redistributions in binary form must reproduce the above copyright
  18  *    notice, this list of conditions and the following disclaimer in the
  19  *    documentation and/or other materials provided with the distribution.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31  * SUCH DAMAGE.
  32  *
  33  * @(#) Copyright (c) 1985, 1993\n\
  34  *      The Regents of the University of California.  All rights reserved.
  35  * @(#)unifdef.c        8.1 (Berkeley) 6/6/93
  36  *
  37  * $DragonFly: src/usr.bin/unifdef/unifdef.c,v 1.4 2007/06/26 07:47:28 hasso Exp $
  38  * $FreeBSD: src/usr.bin/unifdef/unifdef.c,v 1.20 2005/05/21 09:55:09 ru Exp $
  39  * $NetBSD: unifdef.c,v 1.8 2000/07/03 02:51:36 matt Exp $
  40  * $dotat: things/unifdef.c,v 1.171 2005/03/08 12:38:48 fanf2 Exp $"
  41  */
  42
  43 /*
  44  * unifdef - remove ifdef'ed lines
  45  *
  46  *  Wishlist:
  47  *      provide an option which will append the name of the
  48  *        appropriate symbol after #else's and #endif's
  49  *      provide an option which will check symbols after
  50  *        #else's and #endif's to see that they match their
  51  *        corresponding #ifdef or #ifndef
  52  *
  53  *   The first two items above require better buffer handling, which would
  54  *     also make it possible to handle all "dodgy" directives correctly.
  55  */
  56
  57 #include <ctype.h>
  58 #include <err.h>
  59 #include <stdarg.h>
  60 #include <stdbool.h>
  61 #include <stdio.h>
  62 #include <stdlib.h>
  63 #include <string.h>
  64 #include <unistd.h>
  65
  66 /* types of input lines: */
  67 typedef enum {
  68         LT_TRUEI,               /* a true #if with ignore flag */
  69         LT_FALSEI,              /* a false #if with ignore flag */
  70         LT_IF,                  /* an unknown #if */
  71         LT_TRUE,                /* a true #if */
  72         LT_FALSE,               /* a false #if */
  73         LT_ELIF,                /* an unknown #elif */
  74         LT_ELTRUE,              /* a true #elif */
  75         LT_ELFALSE,             /* a false #elif */
  76         LT_ELSE,                /* #else */
  77         LT_ENDIF,               /* #endif */
  78         LT_DODGY,               /* flag: directive is not on one line */
  79         LT_DODGY_LAST = LT_DODGY + LT_ENDIF,
  80         LT_PLAIN,               /* ordinary line */
  81         LT_EOF,                 /* end of file */
  82         LT_COUNT
  83 } Linetype;
  84
  85 static char const * const linetype_name[] = {
  86         "TRUEI", "FALSEI", "IF", "TRUE", "FALSE",
  87         "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF",
  88         "DODGY TRUEI", "DODGY FALSEI",
  89         "DODGY IF", "DODGY TRUE", "DODGY FALSE",
  90         "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE",
  91         "DODGY ELSE", "DODGY ENDIF",
  92         "PLAIN", "EOF"
  93 };
  94
  95 /* state of #if processing */
  96 typedef enum {
  97         IS_OUTSIDE,
  98         IS_FALSE_PREFIX,        /* false #if followed by false #elifs */
  99         IS_TRUE_PREFIX,         /* first non-false #(el)if is true */
 100         IS_PASS_MIDDLE,         /* first non-false #(el)if is unknown */
 101         IS_FALSE_MIDDLE,        /* a false #elif after a pass state */
 102         IS_TRUE_MIDDLE,         /* a true #elif after a pass state */
 103         IS_PASS_ELSE,           /* an else after a pass state */
 104         IS_FALSE_ELSE,          /* an else after a true state */
 105         IS_TRUE_ELSE,           /* an else after only false states */
 106         IS_FALSE_TRAILER,       /* #elifs after a true are false */
 107         IS_COUNT
 108 } Ifstate;
 109
 110 static char const * const ifstate_name[] = {
 111         "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX",
 112         "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE",
 113         "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE",
 114         "FALSE_TRAILER"
 115 };
 116
 117 /* state of comment parser */
 118 typedef enum {
 119         NO_COMMENT = false,     /* outside a comment */
 120         C_COMMENT,              /* in a comment like this one */
 121         CXX_COMMENT,            /* between // and end of line */
 122         STARTING_COMMENT,       /* just after slash-backslash-newline */
 123         FINISHING_COMMENT,      /* star-backslash-newline in a C comment */
 124         CHAR_LITERAL,           /* inside '' */
 125         STRING_LITERAL          /* inside "" */
 126 } Comment_state;
 127
 128 static char const * const comment_name[] = {
 129         "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING"
 130 };
 131
 132 /* state of preprocessor line parser */
 133 typedef enum {
 134         LS_START,               /* only space and comments on this line */
 135         LS_HASH,                /* only space, comments, and a hash */
 136         LS_DIRTY                /* this line can't be a preprocessor line */
 137 } Line_state;
 138
 139 static char const * const linestate_name[] = {
 140         "START", "HASH", "DIRTY"
 141 };
 142
 143 /*
 144  * Minimum translation limits from ISO/IEC 9899:1999 5.2.4.1
 145  */
 146 #define MAXDEPTH        64                      /* maximum #if nesting */
 147 #define MAXLINE         4096                    /* maximum length of line */
 148 #define MAXSYMS         4096                    /* maximum number of symbols */
 149
 150 /*
 151  * Sometimes when editing a keyword the replacement text is longer, so
 152  * we leave some space at the end of the tline buffer to accommodate this.
 153  */
 154 #define EDITSLOP        10
 155
 156 /*
 157  * Globals.
 158  */
 159
 160 static bool             complement;             /* -c: do the complement */
 161 static bool             debugging;              /* -d: debugging reports */
 162 static bool             iocccok;                /* -e: fewer IOCCC errors */
 163 static bool             killconsts;             /* -k: eval constant #ifs */
 164 static bool             lnblank;                /* -l: blank deleted lines */
 165 static bool             lnnum;                  /* -n: add #line directives */
 166 static bool             symlist;                /* -s: output symbol list */
 167 static bool             text;                   /* -t: this is a text file */
 168
 169 static const char      *symname[MAXSYMS];       /* symbol name */
 170 static const char      *value[MAXSYMS];         /* -Dsym=value */
 171 static bool             ignore[MAXSYMS];        /* -iDsym or -iUsym */
 172 static int              nsyms;                  /* number of symbols */
 173
 174 static FILE            *input;                  /* input file pointer */
 175 static const char      *filename;               /* input file name */
 176 static int              linenum;                /* current line number */
 177
 178 static char             tline[MAXLINE+EDITSLOP];/* input buffer plus space */
 179 static char            *keyword;                /* used for editing #elif's */
 180
 181 static Comment_state    incomment;              /* comment parser state */
 182 static Line_state       linestate;              /* #if line parser state */
 183 static Ifstate          ifstate[MAXDEPTH];      /* #if processor state */
 184 static bool             ignoring[MAXDEPTH];     /* ignore comments state */
 185 static int              stifline[MAXDEPTH];     /* start of current #if */
 186 static int              depth;                  /* current #if nesting */
 187 static int              delcount;               /* count of deleted lines */
 188 static bool             keepthis;               /* don't delete constant #if */
 189
 190 static int              exitstat;               /* program exit status */
 191
 192 static void             addsym(bool, bool, char *);
 193 static void             debug(const char *, ...);
 194 static void             done(void);
 195 static void             error(const char *);
 196 static int              findsym(const char *);
 197 static void             flushline(bool);
 198 static Linetype         getline(void);
 199 static Linetype         ifeval(const char **);
 200 static void             ignoreoff(void);
 201 static void             ignoreon(void);
 202 static void             keywordedit(const char *);
 203 static void             nest(void);
 204 static void             process(void);
 205 static const char      *skipcomment(const char *);
 206 static const char      *skipsym(const char *);
 207 static void             state(Ifstate);
 208 static int              strlcmp(const char *, const char *, size_t);
 209 static void             unnest(void);
 210 static void             usage(void);
 211
 212 #define endsym(c) (!isalpha((unsigned char)c) && !isdigit((unsigned char)c) && c != '_')
 213
 214 /*
 215  * The main program.
 216  */
 217 int
 218 main(int argc, char *argv[])
 219 {
 220         int opt;
 221
 222         while ((opt = getopt(argc, argv, "i:D:U:I:cdeklnst")) != -1)
 223                 switch (opt) {
 224                 case 'i': /* treat stuff controlled by these symbols as text */
 225                         /*
 226                          * For strict backwards-compatibility the U or D
 227                          * should be immediately after the -i but it doesn't
 228                          * matter much if we relax that requirement.
 229                          */
 230                         opt = *optarg++;
 231                         if (opt == 'D')
 232                                 addsym(true, true, optarg);
 233                         else if (opt == 'U')
 234                                 addsym(true, false, optarg);
 235                         else
 236                                 usage();
 237                         break;
 238                 case 'D': /* define a symbol */
 239                         addsym(false, true, optarg);
 240                         break;
 241                 case 'U': /* undef a symbol */
 242                         addsym(false, false, optarg);
 243                         break;
 244                 case 'I':
 245                         /* no-op for compatibility with cpp */
 246                         break;
 247                 case 'c': /* treat -D as -U and vice versa */
 248                         complement = true;
 249                         break;
 250                 case 'd':
 251                         debugging = true;
 252                         break;
 253                 case 'e': /* fewer errors from dodgy lines */
 254                         iocccok = true;
 255                         break;
 256                 case 'k': /* process constant #ifs */
 257                         killconsts = true;
 258                         break;
 259                 case 'l': /* blank deleted lines instead of omitting them */
 260                         lnblank = true;
 261                         break;
 262                 case 'n': /* add #line directive after deleted lines */
 263                         lnnum = true;
 264                         break;
 265                 case 's': /* only output list of symbols that control #ifs */
 266                         symlist = true;
 267                         break;
 268                 case 't': /* don't parse C comments */
 269                         text = true;
 270                         break;
 271                 default:
 272                         usage();
 273                 }
 274         argc -= optind;
 275         argv += optind;
 276         if (argc > 1) {
 277                 errx(2, "can only do one file");
 278         } else if (argc == 1 && strcmp(*argv, "-") != 0) {
 279                 filename = *argv;
 280                 input = fopen(filename, "r");
 281                 if (input == NULL)
 282                         err(2, "can't open %s", filename);
 283         } else {
 284                 filename = "[stdin]";
 285                 input = stdin;
 286         }
 287         process();
 288         abort(); /* bug */
 289 }
 290
 291 static void
 292 usage(void)
 293 {
 294         fprintf(stderr, "usage: unifdef [-cdeklnst] [-Ipath]"
 295             " [-Dsym[=val]] [-Usym] [-iDsym[=val]] [-iUsym] ... [file]\n");
 296         exit(2);
 297 }
 298
 299 /*
 300  * A state transition function alters the global #if processing state
 301  * in a particular way. The table below is indexed by the current
 302  * processing state and the type of the current line.
 303  *
 304  * Nesting is handled by keeping a stack of states; some transition
 305  * functions increase or decrease the depth. They also maintain the
 306  * ignore state on a stack. In some complicated cases they have to
 307  * alter the preprocessor directive, as follows.
 308  *
 309  * When we have processed a group that starts off with a known-false
 310  * #if/#elif sequence (which has therefore been deleted) followed by a
 311  * #elif that we don't understand and therefore must keep, we edit the
 312  * latter into a #if to keep the nesting correct.
 313  *
 314  * When we find a true #elif in a group, the following block will
 315  * always be kept and the rest of the sequence after the next #elif or
 316  * #else will be discarded. We edit the #elif into a #else and the
 317  * following directive to #endif since this has the desired behaviour.
 318  *
 319  * "Dodgy" directives are split across multiple lines, the most common
 320  * example being a multi-line comment hanging off the right of the
 321  * directive. We can handle them correctly only if there is no change
 322  * from printing to dropping (or vice versa) caused by that directive.
 323  * If the directive is the first of a group we have a choice between
 324  * failing with an error, or passing it through unchanged instead of
 325  * evaluating it. The latter is not the default to avoid questions from
 326  * users about unifdef unexpectedly leaving behind preprocessor directives.
 327  */
 328 typedef void state_fn(void);
 329
 330 /* report an error */
 331 static void Eelif (void) { error("Inappropriate #elif"); }
 332 static void Eelse (void) { error("Inappropriate #else"); }
 333 static void Eendif(void) { error("Inappropriate #endif"); }
 334 static void Eeof  (void) { error("Premature EOF"); }
 335 static void Eioccc(void) { error("Obfuscated preprocessor control line"); }
 336 /* plain line handling */
 337 static void print (void) { flushline(true); }
 338 static void drop  (void) { flushline(false); }
 339 /* output lacks group's start line */
 340 static void Strue (void) { drop();  ignoreoff(); state(IS_TRUE_PREFIX); }
 341 static void Sfalse(void) { drop();  ignoreoff(); state(IS_FALSE_PREFIX); }
 342 static void Selse (void) { drop();               state(IS_TRUE_ELSE); }
 343 /* print/pass this block */
 344 static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); }
 345 static void Pelse (void) { print();              state(IS_PASS_ELSE); }
 346 static void Pendif(void) { print(); unnest(); }
 347 /* discard this block */
 348 static void Dfalse(void) { drop();  ignoreoff(); state(IS_FALSE_TRAILER); }
 349 static void Delif (void) { drop();  ignoreoff(); state(IS_FALSE_MIDDLE); }
 350 static void Delse (void) { drop();               state(IS_FALSE_ELSE); }
 351 static void Dendif(void) { drop();  unnest(); }
 352 /* first line of group */
 353 static void Fdrop (void) { nest();  Dfalse(); }
 354 static void Fpass (void) { nest();  Pelif(); }
 355 static void Ftrue (void) { nest();  Strue(); }
 356 static void Ffalse(void) { nest();  Sfalse(); }
 357 /* variable pedantry for obfuscated lines */
 358 static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); }
 359 static void Oif   (void) { if (!iocccok) Eioccc(); Fpass(); }
 360 static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); }
 361 /* ignore comments in this block */
 362 static void Idrop (void) { Fdrop();  ignoreon(); }
 363 static void Itrue (void) { Ftrue();  ignoreon(); }
 364 static void Ifalse(void) { Ffalse(); ignoreon(); }
 365 /* edit this line */
 366 static void Mpass (void) { strncpy(keyword, "if  ", 4); Pelif(); }
 367 static void Mtrue (void) { keywordedit("else\n");  state(IS_TRUE_MIDDLE); }
 368 static void Melif (void) { keywordedit("endif\n"); state(IS_FALSE_TRAILER); }
 369 static void Melse (void) { keywordedit("endif\n"); state(IS_FALSE_ELSE); }
 370
 371 static state_fn * const trans_table[IS_COUNT][LT_COUNT] = {
 372 /* IS_OUTSIDE */
 373 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif,
 374   Oiffy, Oiffy, Fpass, Oif,   Oif,   Eelif, Eelif, Eelif, Eelse, Eendif,
 375   print, done },
 376 /* IS_FALSE_PREFIX */
 377 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif,
 378   Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc,
 379   drop,  Eeof },
 380 /* IS_TRUE_PREFIX */
 381 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif,
 382   Oiffy, Oiffy, Fpass, Oif,   Oif,   Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
 383   print, Eeof },
 384 /* IS_PASS_MIDDLE */
 385 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif,
 386   Oiffy, Oiffy, Fpass, Oif,   Oif,   Pelif, Oelif, Oelif, Pelse, Pendif,
 387   print, Eeof },
 388 /* IS_FALSE_MIDDLE */
 389 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif,
 390   Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
 391   drop,  Eeof },
 392 /* IS_TRUE_MIDDLE */
 393 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif,
 394   Oiffy, Oiffy, Fpass, Oif,   Oif,   Eioccc,Eioccc,Eioccc,Eioccc,Pendif,
 395   print, Eeof },
 396 /* IS_PASS_ELSE */
 397 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif,
 398   Oiffy, Oiffy, Fpass, Oif,   Oif,   Eelif, Eelif, Eelif, Eelse, Pendif,
 399   print, Eeof },
 400 /* IS_FALSE_ELSE */
 401 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif,
 402   Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc,
 403   drop,  Eeof },
 404 /* IS_TRUE_ELSE */
 405 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif,
 406   Oiffy, Oiffy, Fpass, Oif,   Oif,   Eelif, Eelif, Eelif, Eelse, Eioccc,
 407   print, Eeof },
 408 /* IS_FALSE_TRAILER */
 409 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif,
 410   Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc,
 411   drop,  Eeof }
 412 /*TRUEI  FALSEI IF     TRUE   FALSE  ELIF   ELTRUE ELFALSE ELSE  ENDIF
 413   TRUEI  FALSEI IF     TRUE   FALSE  ELIF   ELTRUE ELFALSE ELSE  ENDIF (DODGY)
 414   PLAIN  EOF */
 415 };
 416
 417 /*
 418  * State machine utility functions
 419  */
 420 static void
 421 done(void)
 422 {
 423         if (incomment)
 424                 error("EOF in comment");
 425         exit(exitstat);
 426 }
 427 static void
 428 ignoreoff(void)
 429 {
 430         if (depth == 0)
 431                 abort(); /* bug */
 432         ignoring[depth] = ignoring[depth-1];
 433 }
 434 static void
 435 ignoreon(void)
 436 {
 437         ignoring[depth] = true;
 438 }
 439 static void
 440 keywordedit(const char *replacement)
 441 {
 442         strlcpy(keyword, replacement, tline + sizeof(tline) - keyword);
 443         print();
 444 }
 445 static void
 446 nest(void)
 447 {
 448         depth += 1;
 449         if (depth >= MAXDEPTH)
 450                 error("Too many levels of nesting");
 451         stifline[depth] = linenum;
 452 }
 453 static void
 454 unnest(void)
 455 {
 456         if (depth == 0)
 457                 abort(); /* bug */
 458         depth -= 1;
 459 }
 460 static void
 461 state(Ifstate is)
 462 {
 463         ifstate[depth] = is;
 464 }
 465
 466 /*
 467  * Write a line to the output or not, according to command line options.
 468  */
 469 static void
 470 flushline(bool keep)
 471 {
 472         if (symlist)
 473                 return;
 474         if (keep ^ complement) {
 475                 if (lnnum && delcount > 0)
 476                         printf("#line %d\n", linenum);
 477                 fputs(tline, stdout);
 478                 delcount = 0;
 479         } else {
 480                 if (lnblank)
 481                         putc('\n', stdout);
 482                 exitstat = 1;
 483                 delcount += 1;
 484         }
 485 }
 486
 487 /*
 488  * The driver for the state machine.
 489  */
 490 static void
 491 process(void)
 492 {
 493         Linetype lineval;
 494
 495         for (;;) {
 496                 linenum++;
 497                 lineval = getline();
 498                 trans_table[ifstate[depth]][lineval]();
 499                 debug("process %s -> %s depth %d",
 500                     linetype_name[lineval],
 501                     ifstate_name[ifstate[depth]], depth);
 502         }
 503 }
 504
 505 /*
 506  * Parse a line and determine its type. We keep the preprocessor line
 507  * parser state between calls in the global variable linestate, with
 508  * help from skipcomment().
 509  */
 510 static Linetype
 511 getline(void)
 512 {
 513         const char *cp;
 514         int cursym;
 515         int kwlen;
 516         Linetype retval;
 517         Comment_state wascomment;
 518
 519         if (fgets(tline, MAXLINE, input) == NULL)
 520                 return (LT_EOF);
 521         retval = LT_PLAIN;
 522         wascomment = incomment;
 523         cp = skipcomment(tline);
 524         if (linestate == LS_START) {
 525                 if (*cp == '#') {
 526                         linestate = LS_HASH;
 527                         cp = skipcomment(cp + 1);
 528                 } else if (*cp != '\0')
 529                         linestate = LS_DIRTY;
 530         }
 531         if (!incomment && linestate == LS_HASH) {
 532                 keyword = tline + (cp - tline);
 533                 cp = skipsym(cp);
 534                 kwlen = cp - keyword;
 535                 /* no way can we deal with a continuation inside a keyword */
 536                 if (strncmp(cp, "\\\n", 2) == 0)
 537                         Eioccc();
 538                 if (strlcmp("ifdef", keyword, kwlen) == 0 ||
 539                     strlcmp("ifndef", keyword, kwlen) == 0) {
 540                         cp = skipcomment(cp);
 541                         if ((cursym = findsym(cp)) < 0)
 542                                 retval = LT_IF;
 543                         else {
 544                                 retval = (keyword[2] == 'n')
 545                                     ? LT_FALSE : LT_TRUE;
 546                                 if (value[cursym] == NULL)
 547                                         retval = (retval == LT_TRUE)
 548                                             ? LT_FALSE : LT_TRUE;
 549                                 if (ignore[cursym])
 550                                         retval = (retval == LT_TRUE)
 551                                             ? LT_TRUEI : LT_FALSEI;
 552                         }
 553                         cp = skipsym(cp);
 554                 } else if (strlcmp("if", keyword, kwlen) == 0)
 555                         retval = ifeval(&cp);
 556                 else if (strlcmp("elif", keyword, kwlen) == 0)
 557                         retval = ifeval(&cp) - LT_IF + LT_ELIF;
 558                 else if (strlcmp("else", keyword, kwlen) == 0)
 559                         retval = LT_ELSE;
 560                 else if (strlcmp("endif", keyword, kwlen) == 0)
 561                         retval = LT_ENDIF;
 562                 else {
 563                         linestate = LS_DIRTY;
 564                         retval = LT_PLAIN;
 565                 }
 566                 cp = skipcomment(cp);
 567                 if (*cp != '\0') {
 568                         linestate = LS_DIRTY;
 569                         if (retval == LT_TRUE || retval == LT_FALSE ||
 570                             retval == LT_TRUEI || retval == LT_FALSEI)
 571                                 retval = LT_IF;
 572                         if (retval == LT_ELTRUE || retval == LT_ELFALSE)
 573                                 retval = LT_ELIF;
 574                 }
 575                 if (retval != LT_PLAIN && (wascomment || incomment)) {
 576                         retval += LT_DODGY;
 577                         if (incomment)
 578                                 linestate = LS_DIRTY;
 579                 }
 580                 /* skipcomment should have changed the state */
 581                 if (linestate == LS_HASH)
 582                         abort(); /* bug */
 583         }
 584         if (linestate == LS_DIRTY) {
 585                 while (*cp != '\0')
 586                         cp = skipcomment(cp + 1);
 587         }
 588         debug("parser %s comment %s line",
 589             comment_name[incomment], linestate_name[linestate]);
 590         return (retval);
 591 }
 592
 593 /*
 594  * These are the binary operators that are supported by the expression
 595  * evaluator. Note that if support for division is added then we also
 596  * need short-circuiting booleans because of divide-by-zero.
 597  */
 598 static int op_lt(int a, int b) { return (a < b); }
 599 static int op_gt(int a, int b) { return (a > b); }
 600 static int op_le(int a, int b) { return (a <= b); }
 601 static int op_ge(int a, int b) { return (a >= b); }
 602 static int op_eq(int a, int b) { return (a == b); }
 603 static int op_ne(int a, int b) { return (a != b); }
 604 static int op_or(int a, int b) { return (a || b); }
 605 static int op_and(int a, int b) { return (a && b); }
 606
 607 /*
 608  * An evaluation function takes three arguments, as follows: (1) a pointer to
 609  * an element of the precedence table which lists the operators at the current
 610  * level of precedence; (2) a pointer to an integer which will receive the
 611  * value of the expression; and (3) a pointer to a char* that points to the
 612  * expression to be evaluated and that is updated to the end of the expression
 613  * when evaluation is complete. The function returns LT_FALSE if the value of
 614  * the expression is zero, LT_TRUE if it is non-zero, or LT_IF if the
 615  * expression could not be evaluated.
 616  */
 617 struct ops;
 618
 619 typedef Linetype eval_fn(const struct ops *, int *, const char **);
 620
 621 static eval_fn eval_table, eval_unary;
 622
 623 /*
 624  * The precedence table. Expressions involving binary operators are evaluated
 625  * in a table-driven way by eval_table. When it evaluates a subexpression it
 626  * calls the inner function with its first argument pointing to the next
 627  * element of the table. Innermost expressions have special non-table-driven
 628  * handling.
 629  */
 630 static const struct ops {
 631         eval_fn *inner;
 632         struct op {
 633                 const char *str;
 634                 int (*fn)(int, int);
 635         } op[5];
 636 } eval_ops[] = {
 637         { eval_table, { { "||", op_or } } },
 638         { eval_table, { { "&&", op_and } } },
 639         { eval_table, { { "==", op_eq },
 640                         { "!=", op_ne } } },
 641         { eval_unary, { { "<=", op_le },
 642                         { ">=", op_ge },
 643                         { "<", op_lt },
 644                         { ">", op_gt } } }
 645 };
 646
 647 /*
 648  * Function for evaluating the innermost parts of expressions,
 649  * viz. !expr (expr) defined(symbol) symbol number
 650  * We reset the keepthis flag when we find a non-constant subexpression.
 651  */
 652 static Linetype
 653 eval_unary(const struct ops *ops, int *valp, const char **cpp)
 654 {
 655         const char *cp;
 656         char *ep;
 657         int sym;
 658
 659         cp = skipcomment(*cpp);
 660         if (*cp == '!') {
 661                 debug("eval%d !", ops - eval_ops);
 662                 cp++;
 663                 if (eval_unary(ops, valp, &cp) == LT_IF)
 664                         return (LT_IF);
 665                 *valp = !*valp;
 666         } else if (*cp == '(') {
 667                 cp++;
 668                 debug("eval%d (", ops - eval_ops);
 669                 if (eval_table(eval_ops, valp, &cp) == LT_IF)
 670                         return (LT_IF);
 671                 cp = skipcomment(cp);
 672                 if (*cp++ != ')')
 673                         return (LT_IF);
 674         } else if (isdigit((unsigned char)*cp)) {
 675                 debug("eval%d number", ops - eval_ops);
 676                 *valp = strtol(cp, &ep, 0);
 677                 cp = skipsym(cp);
 678         } else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) {
 679                 cp = skipcomment(cp+7);
 680                 debug("eval%d defined", ops - eval_ops);
 681                 if (*cp++ != '(')
 682                         return (LT_IF);
 683                 cp = skipcomment(cp);
 684                 sym = findsym(cp);
 685                 if (sym < 0)
 686                         return (LT_IF);
 687                 *valp = (value[sym] != NULL);
 688                 cp = skipsym(cp);
 689                 cp = skipcomment(cp);
 690                 if (*cp++ != ')')
 691                         return (LT_IF);
 692                 keepthis = false;
 693         } else if (!endsym(*cp)) {
 694                 debug("eval%d symbol", ops - eval_ops);
 695                 sym = findsym(cp);
 696                 if (sym < 0)
 697                         return (LT_IF);
 698                 if (value[sym] == NULL)
 699                         *valp = 0;
 700                 else {
 701                         *valp = strtol(value[sym], &ep, 0);
 702                         if (*ep != '\0' || ep == value[sym])
 703                                 return (LT_IF);
 704                 }
 705                 cp = skipsym(cp);
 706                 keepthis = false;
 707         } else {
 708                 debug("eval%d bad expr", ops - eval_ops);
 709                 return (LT_IF);
 710         }
 711
 712         *cpp = cp;
 713         debug("eval%d = %d", ops - eval_ops, *valp);
 714         return (*valp ? LT_TRUE : LT_FALSE);
 715 }
 716
 717 /*
 718  * Table-driven evaluation of binary operators.
 719  */
 720 static Linetype
 721 eval_table(const struct ops *ops, int *valp, const char **cpp)
 722 {
 723         const struct op *op;
 724         const char *cp;
 725         int val;
 726
 727         debug("eval%d", ops - eval_ops);
 728         cp = *cpp;
 729         if (ops->inner(ops+1, valp, &cp) == LT_IF)
 730                 return (LT_IF);
 731         for (;;) {
 732                 cp = skipcomment(cp);
 733                 for (op = ops->op; op->str != NULL; op++)
 734                         if (strncmp(cp, op->str, strlen(op->str)) == 0)
 735                                 break;
 736                 if (op->str == NULL)
 737                         break;
 738                 cp += strlen(op->str);
 739                 debug("eval%d %s", ops - eval_ops, op->str);
 740                 if (ops->inner(ops+1, &val, &cp) == LT_IF)
 741                         return (LT_IF);
 742                 *valp = op->fn(*valp, val);
 743         }
 744
 745         *cpp = cp;
 746         debug("eval%d = %d", ops - eval_ops, *valp);
 747         return (*valp ? LT_TRUE : LT_FALSE);
 748 }
 749
 750 /*
 751  * Evaluate the expression on a #if or #elif line. If we can work out
 752  * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we
 753  * return just a generic LT_IF.
 754  */
 755 static Linetype
 756 ifeval(const char **cpp)
 757 {
 758         int ret;
 759         int val;
 760
 761         debug("eval %s", *cpp);
 762         keepthis = killconsts ? false : true;
 763         ret = eval_table(eval_ops, &val, cpp);
 764         debug("eval = %d", val);
 765         return (keepthis ? LT_IF : ret);
 766 }
 767
 768 /*
 769  * Skip over comments, strings, and character literals and stop at the
 770  * next character position that is not whitespace. Between calls we keep
 771  * the comment state in the global variable incomment, and we also adjust
 772  * the global variable linestate when we see a newline.
 773  * XXX: doesn't cope with the buffer splitting inside a state transition.
 774  */
 775 static const char *
 776 skipcomment(const char *cp)
 777 {
 778         if (text || ignoring[depth]) {
 779                 for (; isspace((unsigned char)*cp); cp++)
 780                         if (*cp == '\n')
 781                                 linestate = LS_START;
 782                 return (cp);
 783         }
 784         while (*cp != '\0')
 785                 /* don't reset to LS_START after a line continuation */
 786                 if (strncmp(cp, "\\\n", 2) == 0)
 787                         cp += 2;
 788                 else switch (incomment) {
 789                 case NO_COMMENT:
 790                         if (strncmp(cp, "/\\\n", 3) == 0) {
 791                                 incomment = STARTING_COMMENT;
 792                                 cp += 3;
 793                         } else if (strncmp(cp, "/*", 2) == 0) {
 794                                 incomment = C_COMMENT;
 795                                 cp += 2;
 796                         } else if (strncmp(cp, "//", 2) == 0) {
 797                                 incomment = CXX_COMMENT;
 798                                 cp += 2;
 799                         } else if (strncmp(cp, "\'", 1) == 0) {
 800                                 incomment = CHAR_LITERAL;
 801                                 linestate = LS_DIRTY;
 802                                 cp += 1;
 803                         } else if (strncmp(cp, "\"", 1) == 0) {
 804                                 incomment = STRING_LITERAL;
 805                                 linestate = LS_DIRTY;
 806                                 cp += 1;
 807                         } else if (strncmp(cp, "\n", 1) == 0) {
 808                                 linestate = LS_START;
 809                                 cp += 1;
 810                         } else if (strchr(" \t", *cp) != NULL) {
 811                                 cp += 1;
 812                         } else
 813                                 return (cp);
 814                         continue;
 815                 case CXX_COMMENT:
 816                         if (strncmp(cp, "\n", 1) == 0) {
 817                                 incomment = NO_COMMENT;
 818                                 linestate = LS_START;
 819                         }
 820                         cp += 1;
 821                         continue;
 822                 case CHAR_LITERAL:
 823                 case STRING_LITERAL:
 824                         if ((incomment == CHAR_LITERAL && cp[0] == '\'') ||
 825                             (incomment == STRING_LITERAL && cp[0] == '\"')) {
 826                                 incomment = NO_COMMENT;
 827                                 cp += 1;
 828                         } else if (cp[0] == '\\') {
 829                                 if (cp[1] == '\0')
 830                                         cp += 1;
 831                                 else
 832                                         cp += 2;
 833                         } else if (strncmp(cp, "\n", 1) == 0) {
 834                                 if (incomment == CHAR_LITERAL)
 835                                         error("unterminated char literal");
 836                                 else
 837                                         error("unterminated string literal");
 838                         } else
 839                                 cp += 1;
 840                         continue;
 841                 case C_COMMENT:
 842                         if (strncmp(cp, "*\\\n", 3) == 0) {
 843                                 incomment = FINISHING_COMMENT;
 844                                 cp += 3;
 845                         } else if (strncmp(cp, "*/", 2) == 0) {
 846                                 incomment = NO_COMMENT;
 847                                 cp += 2;
 848                         } else
 849                                 cp += 1;
 850                         continue;
 851                 case STARTING_COMMENT:
 852                         if (*cp == '*') {
 853                                 incomment = C_COMMENT;
 854                                 cp += 1;
 855                         } else if (*cp == '/') {
 856                                 incomment = CXX_COMMENT;
 857                                 cp += 1;
 858                         } else {
 859                                 incomment = NO_COMMENT;
 860                                 linestate = LS_DIRTY;
 861                         }
 862                         continue;
 863                 case FINISHING_COMMENT:
 864                         if (*cp == '/') {
 865                                 incomment = NO_COMMENT;
 866                                 cp += 1;
 867                         } else
 868                                 incomment = C_COMMENT;
 869                         continue;
 870                 default:
 871                         abort(); /* bug */
 872                 }
 873         return (cp);
 874 }
 875
 876 /*
 877  * Skip over an identifier.
 878  */
 879 static const char *
 880 skipsym(const char *cp)
 881 {
 882         while (!endsym(*cp))
 883                 ++cp;
 884         return (cp);
 885 }
 886
 887 /*
 888  * Look for the symbol in the symbol table. If it is found, we return
 889  * the symbol table index, else we return -1.
 890  */
 891 static int
 892 findsym(const char *str)
 893 {
 894         const char *cp;
 895         int symind;
 896
 897         cp = skipsym(str);
 898         if (cp == str)
 899                 return (-1);
 900         if (symlist) {
 901                 printf("%.*s\n", (int)(cp-str), str);
 902                 /* we don't care about the value of the symbol */
 903                 return (0);
 904         }
 905         for (symind = 0; symind < nsyms; ++symind) {
 906                 if (strlcmp(symname[symind], str, cp-str) == 0) {
 907                         debug("findsym %s %s", symname[symind],
 908                             value[symind] ? value[symind] : "");
 909                         return (symind);
 910                 }
 911         }
 912         return (-1);
 913 }
 914
 915 /*
 916  * Add a symbol to the symbol table.
 917  */
 918 static void
 919 addsym(bool ignorethis, bool definethis, char *sym)
 920 {
 921         int symind;
 922         char *val;
 923
 924         symind = findsym(sym);
 925         if (symind < 0) {
 926                 if (nsyms >= MAXSYMS)
 927                         errx(2, "too many symbols");
 928                 symind = nsyms++;
 929         }
 930         symname[symind] = sym;
 931         ignore[symind] = ignorethis;
 932         val = sym + (skipsym(sym) - sym);
 933         if (definethis) {
 934                 if (*val == '=') {
 935                         value[symind] = val+1;
 936                         *val = '\0';
 937                 } else if (*val == '\0')
 938                         value[symind] = "";
 939                 else
 940                         usage();
 941         } else {
 942                 if (*val != '\0')
 943                         usage();
 944                 value[symind] = NULL;
 945         }
 946 }
 947
 948 /*
 949  * Compare s with n characters of t.
 950  * The same as strncmp() except that it checks that s[n] == '\0'.
 951  */
 952 static int
 953 strlcmp(const char *s, const char *t, size_t n)
 954 {
 955         while (n-- && *t != '\0')
 956                 if (*s != *t)
 957                         return ((unsigned char)*s - (unsigned char)*t);
 958                 else
 959                         ++s, ++t;
 960         return ((unsigned char)*s);
 961 }
 962
 963 /*
 964  * Diagnostics.
 965  */
 966 static void
 967 debug(const char *msg, ...)
 968 {
 969         va_list ap;
 970
 971         if (debugging) {
 972                 va_start(ap, msg);
 973                 vwarnx(msg, ap);
 974                 va_end(ap);
 975         }
 976 }
 977
 978 static void
 979 error(const char *msg)
 980 {
 981         if (depth == 0)
 982                 warnx("%s: %d: %s", filename, linenum, msg);
 983         else
 984                 warnx("%s: %d: %s (#if line %d depth %d)",
 985                     filename, linenum, msg, stifline[depth], depth);
 986         errx(2, "output may be truncated");
 987 }