| 1 | |
| 2 | * Copyright (c) 2002 - 2005 Tony Finch <[email protected]>. 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 |
|
| 34 | #include <sys/cdefs.h>
|
| 35 |
|
| 36 | #ifndef lint
|
| 37 | #if 0 |
| 38 | static const char copyright[] =
|
| 39 | "@(#) Copyright (c) 1985, 1993\n\
|
| 40 | The Regents of the University of California. All rights reserved.\n";
|
| 41 | #endif
|
| 42 | #ifdef __IDSTRING
|
| 43 | __IDSTRING(Berkeley, "@(#)unifdef.c 8.1 (Berkeley) 6/6/93");
|
| 44 | __IDSTRING(NetBSD, "$NetBSD: unifdef.c,v 1.8 2000/07/03 02:51:36 matt Exp $");
|
| 45 | __IDSTRING(dotat, "$dotat: things/unifdef.c,v 1.171 2005/03/08 12:38:48 fanf2 Exp $");
|
| 46 | #endif
|
| 47 | #endif
|
| 48 | #ifdef __FBSDID
|
| 49 | __FBSDID("$FreeBSD: /repoman/r/ncvs/src/usr.bin/unifdef/unifdef.c,v 1.20 2005/05/21 09:55:09 ru Exp $");
|
| 50 | #endif
|
| 51 |
|
| 52 | |
| 53 | * unifdef - remove ifdef'ed lines
|
| 54 | *
|
| 55 | * Wishlist:
|
| 56 | * provide an option which will append the name of the
|
| 57 | * appropriate symbol after #else's and #endif's
|
| 58 | * provide an option which will check symbols after
|
| 59 | * #else's and #endif's to see that they match their
|
| 60 | * corresponding #ifdef or #ifndef
|
| 61 | *
|
| 62 | * The first two items above require better buffer handling, which would
|
| 63 | * also make it possible to handle all "dodgy" directives correctly.
|
| 64 |
|
| 65 |
|
| 66 | #include <ctype.h>
|
| 67 | #include <err.h>
|
| 68 | #include <stdarg.h>
|
| 69 | #include <stdbool.h>
|
| 70 | #include <stdio.h>
|
| 71 | #include <stdlib.h>
|
| 72 | #include <string.h>
|
| 73 | #include <unistd.h>
|
| 74 |
|
| 75 | size_t strlcpy(char *dst, const char *src, size_t siz);
|
| 76 |
|
| 77 |
|
| 78 | typedef enum {
|
| 79 | LT_TRUEI,
|
| 80 | LT_FALSEI,
|
| 81 | LT_IF,
|
| 82 | LT_TRUE,
|
| 83 | LT_FALSE,
|
| 84 | LT_ELIF,
|
| 85 | LT_ELTRUE,
|
| 86 | LT_ELFALSE,
|
| 87 | LT_ELSE,
|
| 88 | LT_ENDIF,
|
| 89 | LT_DODGY,
|
| 90 | LT_DODGY_LAST = LT_DODGY + LT_ENDIF,
|
| 91 | LT_PLAIN,
|
| 92 | LT_EOF,
|
| 93 | LT_COUNT
|
| 94 | } Linetype;
|
| 95 |
|
| 96 | static char const * const linetype_name[] = {
|
| 97 | "TRUEI", "FALSEI", "IF", "TRUE", "FALSE",
|
| 98 | "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF",
|
| 99 | "DODGY TRUEI", "DODGY FALSEI",
|
| 100 | "DODGY IF", "DODGY TRUE", "DODGY FALSE",
|
| 101 | "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE",
|
| 102 | "DODGY ELSE", "DODGY ENDIF",
|
| 103 | "PLAIN", "EOF"
|
| 104 | };
|
| 105 |
|
| 106 |
|
| 107 | typedef enum {
|
| 108 | IS_OUTSIDE,
|
| 109 | IS_FALSE_PREFIX,
|
| 110 | IS_TRUE_PREFIX,
|
| 111 | IS_PASS_MIDDLE,
|
| 112 | IS_FALSE_MIDDLE,
|
| 113 | IS_TRUE_MIDDLE,
|
| 114 | IS_PASS_ELSE,
|
| 115 | IS_FALSE_ELSE,
|
| 116 | IS_TRUE_ELSE,
|
| 117 | IS_FALSE_TRAILER,
|
| 118 | IS_COUNT
|
| 119 | } Ifstate;
|
| 120 |
|
| 121 | static char const * const ifstate_name[] = {
|
| 122 | "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX",
|
| 123 | "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE",
|
| 124 | "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE",
|
| 125 | "FALSE_TRAILER"
|
| 126 | };
|
| 127 |
|
| 128 |
|
| 129 | typedef enum {
|
| 130 | NO_COMMENT = false,
|
| 131 | C_COMMENT,
|
| 132 | CXX_COMMENT,
|
| 133 | STARTING_COMMENT,
|
| 134 | FINISHING_COMMENT,
|
| 135 | CHAR_LITERAL,
|
| 136 | STRING_LITERAL
|
| 137 | } Comment_state;
|
| 138 |
|
| 139 | static char const * const comment_name[] = {
|
| 140 | "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING"
|
| 141 | };
|
| 142 |
|
| 143 |
|
| 144 | typedef enum {
|
| 145 | LS_START,
|
| 146 | LS_HASH,
|
| 147 | LS_DIRTY
|
| 148 | } Line_state;
|
| 149 |
|
| 150 | static char const * const linestate_name[] = {
|
| 151 | "START", "HASH", "DIRTY"
|
| 152 | };
|
| 153 |
|
| 154 | |
| 155 | * Minimum translation limits from ISO/IEC 9899:1999 5.2.4.1
|
| 156 |
|
| 157 | #define MAXDEPTH 64
|
| 158 | #define MAXLINE 4096
|
| 159 | #define MAXSYMS 4096
|
| 160 |
|
| 161 | |
| 162 | * Sometimes when editing a keyword the replacement text is longer, so
|
| 163 | * we leave some space at the end of the tline buffer to accommodate this.
|
| 164 |
|
| 165 | #define EDITSLOP 10
|
| 166 |
|
| 167 | |
| 168 | * Globals.
|
| 169 |
|
| 170 |
|
| 171 | static bool complement;
|
| 172 | static bool debugging;
|
| 173 | static bool iocccok;
|
| 174 | static bool killconsts;
|
| 175 | static bool lnblank;
|
| 176 | static bool lnnum;
|
| 177 | static bool symlist;
|
| 178 | static bool text;
|
| 179 |
|
| 180 | static const char *symname[MAXSYMS];
|
| 181 | static const char *value[MAXSYMS];
|
| 182 | static bool ignore[MAXSYMS];
|
| 183 | static int nsyms;
|
| 184 |
|
| 185 | static FILE *input;
|
| 186 | static const char *filename;
|
| 187 | static int linenum;
|
| 188 |
|
| 189 | static char tline[MAXLINE+EDITSLOP];
|
| 190 | static char *keyword;
|
| 191 |
|
| 192 | static Comment_state incomment;
|
| 193 | static Line_state linestate;
|
| 194 | static Ifstate ifstate[MAXDEPTH];
|
| 195 | static bool ignoring[MAXDEPTH];
|
| 196 | static int stifline[MAXDEPTH];
|
| 197 | static int depth;
|
| 198 | static int delcount;
|
| 199 | static bool keepthis;
|
| 200 |
|
| 201 | static int exitstat;
|
| 202 |
|
| 203 | static void addsym(bool, bool, char *);
|
| 204 | static void debug(const char *, ...);
|
| 205 | static void done(void);
|
| 206 | static void error(const char *);
|
| 207 | static int findsym(const char *);
|
| 208 | static void flushline(bool);
|
| 209 | static Linetype getline(void);
|
| 210 | static Linetype ifeval(const char **);
|
| 211 | static void ignoreoff(void);
|
| 212 | static void ignoreon(void);
|
| 213 | static void keywordedit(const char *);
|
| 214 | static void nest(void);
|
| 215 | static void process(void);
|
| 216 | static const char *skipcomment(const char *);
|
| 217 | static const char *skipsym(const char *);
|
| 218 | static void state(Ifstate);
|
| 219 | static int strlcmp(const char *, const char *, size_t);
|
| 220 | static void unnest(void);
|
| 221 | static void usage(void);
|
| 222 |
|
| 223 | #define endsym(c) (!isalpha((unsigned char)c) && !isdigit((unsigned char)c) && c != '_')
|
| 224 |
|
| 225 | |
| 226 | * The main program.
|
| 227 |
|
| 228 | int
|
| 229 | main(int argc, char *argv[])
|
| 230 | {
|
| 231 | int opt;
|
| 232 |
|
| 233 | while ((opt = getopt(argc, argv, "i:D:U:I:cdeklnst")) != -1)
|
| 234 | switch (opt) {
|
| 235 | case 'i':
|
| 236 | |
| 237 | * For strict backwards-compatibility the U or D
|
| 238 | * should be immediately after the -i but it doesn't
|
| 239 | * matter much if we relax that requirement.
|
| 240 |
|
| 241 | opt = *optarg++;
|
| 242 | if (opt == 'D')
|
| 243 | addsym(true, true, optarg);
|
| 244 | else if (opt == 'U')
|
| 245 | addsym(true, false, optarg);
|
| 246 | else
|
| 247 | usage();
|
| 248 | break;
|
| 249 | case 'D':
|
| 250 | addsym(false, true, optarg);
|
| 251 | break;
|
| 252 | case 'U':
|
| 253 | addsym(false, false, optarg);
|
| 254 | break;
|
| 255 | case 'I':
|
| 256 |
|
| 257 | break;
|
| 258 | case 'c':
|
| 259 | complement = true;
|
| 260 | break;
|
| 261 | case 'd':
|
| 262 | debugging = true;
|
| 263 | break;
|
| 264 | case 'e':
|
| 265 | iocccok = true;
|
| 266 | break;
|
| 267 | case 'k':
|
| 268 | killconsts = true;
|
| 269 | break;
|
| 270 | case 'l':
|
| 271 | lnblank = true;
|
| 272 | break;
|
| 273 | case 'n':
|
| 274 | lnnum = true;
|
| 275 | break;
|
| 276 | case 's':
|
| 277 | symlist = true;
|
| 278 | break;
|
| 279 | case 't':
|
| 280 | text = true;
|
| 281 | break;
|
| 282 | default:
|
| 283 | usage();
|
| 284 | }
|
| 285 | argc -= optind;
|
| 286 | argv += optind;
|
| 287 | if (argc > 1) {
|
| 288 | errx(2, "can only do one file");
|
| 289 | } else if (argc == 1 && strcmp(*argv, "-") != 0) {
|
| 290 | filename = *argv;
|
| 291 | input = fopen(filename, "r");
|
| 292 | if (input == NULL)
|
| 293 | err(2, "can't open %s", filename);
|
| 294 | } else {
|
| 295 | filename = "[stdin]";
|
| 296 | input = stdin;
|
| 297 | }
|
| 298 | process();
|
| 299 | abort();
|
| 300 | }
|
| 301 |
|
| 302 | static void
|
| 303 | usage(void)
|
| 304 | {
|
| 305 | fprintf(stderr, "usage: unifdef [-cdeklnst] [-Ipath]"
|
| 306 | " [-Dsym[=val]] [-Usym] [-iDsym[=val]] [-iUsym] ... [file]\n");
|
| 307 | exit(2);
|
| 308 | }
|
| 309 |
|
| 310 | |
| 311 | * A state transition function alters the global #if processing state
|
| 312 | * in a particular way. The table below is indexed by the current
|
| 313 | * processing state and the type of the current line.
|
| 314 | *
|
| 315 | * Nesting is handled by keeping a stack of states; some transition
|
| 316 | * functions increase or decrease the depth. They also maintain the
|
| 317 | * ignore state on a stack. In some complicated cases they have to
|
| 318 | * alter the preprocessor directive, as follows.
|
| 319 | *
|
| 320 | * When we have processed a group that starts off with a known-false
|
| 321 | * #if/#elif sequence (which has therefore been deleted) followed by a
|
| 322 | * #elif that we don't understand and therefore must keep, we edit the
|
| 323 | * latter into a #if to keep the nesting correct.
|
| 324 | *
|
| 325 | * When we find a true #elif in a group, the following block will
|
| 326 | * always be kept and the rest of the sequence after the next #elif or
|
| 327 | * #else will be discarded. We edit the #elif into a #else and the
|
| 328 | * following directive to #endif since this has the desired behaviour.
|
| 329 | *
|
| 330 | * "Dodgy" directives are split across multiple lines, the most common
|
| 331 | * example being a multi-line comment hanging off the right of the
|
| 332 | * directive. We can handle them correctly only if there is no change
|
| 333 | * from printing to dropping (or vice versa) caused by that directive.
|
| 334 | * If the directive is the first of a group we have a choice between
|
| 335 | * failing with an error, or passing it through unchanged instead of
|
| 336 | * evaluating it. The latter is not the default to avoid questions from
|
| 337 | * users about unifdef unexpectedly leaving behind preprocessor directives.
|
| 338 |
|
| 339 | typedef void state_fn(void);
|
| 340 |
|
| 341 |
|
| 342 | static void Eelif (void) { error("Inappropriate #elif"); }
|
| 343 | static void Eelse (void) { error("Inappropriate #else"); }
|
| 344 | static void Eendif(void) { error("Inappropriate #endif"); }
|
| 345 | static void Eeof (void) { error("Premature EOF"); }
|
| 346 | static void Eioccc(void) { error("Obfuscated preprocessor control line"); }
|
| 347 |
|
| 348 | static void print (void) { flushline(true); }
|
| 349 | static void drop (void) { flushline(false); }
|
| 350 |
|
| 351 | static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); }
|
| 352 | static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); }
|
| 353 | static void Selse (void) { drop(); state(IS_TRUE_ELSE); }
|
| 354 |
|
| 355 | static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); }
|
| 356 | static void Pelse (void) { print(); state(IS_PASS_ELSE); }
|
| 357 | static void Pendif(void) { print(); unnest(); }
|
| 358 |
|
| 359 | static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); }
|
| 360 | static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); }
|
| 361 | static void Delse (void) { drop(); state(IS_FALSE_ELSE); }
|
| 362 | static void Dendif(void) { drop(); unnest(); }
|
| 363 |
|
| 364 | static void Fdrop (void) { nest(); Dfalse(); }
|
| 365 | static void Fpass (void) { nest(); Pelif(); }
|
| 366 | static void Ftrue (void) { nest(); Strue(); }
|
| 367 | static void Ffalse(void) { nest(); Sfalse(); }
|
| 368 |
|
| 369 | static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); }
|
| 370 | static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); }
|
| 371 | static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); }
|
| 372 |
|
| 373 | static void Idrop (void) { Fdrop(); ignoreon(); }
|
| 374 | static void Itrue (void) { Ftrue(); ignoreon(); }
|
| 375 | static void Ifalse(void) { Ffalse(); ignoreon(); }
|
| 376 |
|
| 377 | static void Mpass (void) { strncpy(keyword, "if ", 4); Pelif(); }
|
| 378 | static void Mtrue (void) { keywordedit("else\n"); state(IS_TRUE_MIDDLE); }
|
| 379 | static void Melif (void) { keywordedit("endif\n"); state(IS_FALSE_TRAILER); }
|
| 380 | static void Melse (void) { keywordedit("endif\n"); state(IS_FALSE_ELSE); }
|
| 381 |
|
| 382 | static state_fn * const trans_table[IS_COUNT][LT_COUNT] = {
|
| 383 |
|
| 384 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif,
|
| 385 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eendif,
|
| 386 | print, done },
|
| 387 |
|
| 388 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif,
|
| 389 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc,
|
| 390 | drop, Eeof },
|
| 391 |
|
| 392 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif,
|
| 393 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
|
| 394 | print, Eeof },
|
| 395 |
|
| 396 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif,
|
| 397 | Oiffy, Oiffy, Fpass, Oif, Oif, Pelif, Oelif, Oelif, Pelse, Pendif,
|
| 398 | print, Eeof },
|
| 399 |
|
| 400 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif,
|
| 401 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
|
| 402 | drop, Eeof },
|
| 403 |
|
| 404 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif,
|
| 405 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Pendif,
|
| 406 | print, Eeof },
|
| 407 |
|
| 408 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif,
|
| 409 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Pendif,
|
| 410 | print, Eeof },
|
| 411 |
|
| 412 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif,
|
| 413 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc,
|
| 414 | drop, Eeof },
|
| 415 |
|
| 416 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif,
|
| 417 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eioccc,
|
| 418 | print, Eeof },
|
| 419 |
|
| 420 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif,
|
| 421 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc,
|
| 422 | drop, Eeof }
|
| 423 | |
| 424 | TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF (DODGY)
|
| 425 | PLAIN EOF
|
| 426 | };
|
| 427 |
|
| 428 | |
| 429 | * State machine utility functions
|
| 430 |
|
| 431 | static void
|
| 432 | done(void)
|
| 433 | {
|
| 434 | if (incomment)
|
| 435 | error("EOF in comment");
|
| 436 | exit(exitstat);
|
| 437 | }
|
| 438 | static void
|
| 439 | ignoreoff(void)
|
| 440 | {
|
| 441 | if (depth == 0)
|
| 442 | abort();
|
| 443 | ignoring[depth] = ignoring[depth-1];
|
| 444 | }
|
| 445 | static void
|
| 446 | ignoreon(void)
|
| 447 | {
|
| 448 | ignoring[depth] = true;
|
| 449 | }
|
| 450 | static void
|
| 451 | keywordedit(const char *replacement)
|
| 452 | {
|
| 453 | size_t size = tline + sizeof(tline) - keyword;
|
| 454 | char *dst = keyword;
|
| 455 | const char *src = replacement;
|
| 456 | if (size != 0) {
|
| 457 | while ((--size != 0) && (*src != '\0'))
|
| 458 | *dst++ = *src++;
|
| 459 | *dst = '\0';
|
| 460 | }
|
| 461 | print();
|
| 462 | }
|
| 463 | static void
|
| 464 | nest(void)
|
| 465 | {
|
| 466 | depth += 1;
|
| 467 | if (depth >= MAXDEPTH)
|
| 468 | error("Too many levels of nesting");
|
| 469 | stifline[depth] = linenum;
|
| 470 | }
|
| 471 | static void
|
| 472 | unnest(void)
|
| 473 | {
|
| 474 | if (depth == 0)
|
| 475 | abort();
|
| 476 | depth -= 1;
|
| 477 | }
|
| 478 | static void
|
| 479 | state(Ifstate is)
|
| 480 | {
|
| 481 | ifstate[depth] = is;
|
| 482 | }
|
| 483 |
|
| 484 | |
| 485 | * Write a line to the output or not, according to command line options.
|
| 486 |
|
| 487 | static void
|
| 488 | flushline(bool keep)
|
| 489 | {
|
| 490 | if (symlist)
|
| 491 | return;
|
| 492 | if (keep ^ complement) {
|
| 493 | if (lnnum && delcount > 0)
|
| 494 | printf("#line %d\n", linenum);
|
| 495 | fputs(tline, stdout);
|
| 496 | delcount = 0;
|
| 497 | } else {
|
| 498 | if (lnblank)
|
| 499 | putc('\n', stdout);
|
| 500 | exitstat = 1;
|
| 501 | delcount += 1;
|
| 502 | }
|
| 503 | }
|
| 504 |
|
| 505 | |
| 506 | * The driver for the state machine.
|
| 507 |
|
| 508 | static void
|
| 509 | process(void)
|
| 510 | {
|
| 511 | Linetype lineval;
|
| 512 |
|
| 513 | for (;;) {
|
| 514 | linenum++;
|
| 515 | lineval = getline();
|
| 516 | trans_table[ifstate[depth]][lineval]();
|
| 517 | debug("process %s -> %s depth %d",
|
| 518 | linetype_name[lineval],
|
| 519 | ifstate_name[ifstate[depth]], depth);
|
| 520 | }
|
| 521 | }
|
| 522 |
|
| 523 | |
| 524 | * Parse a line and determine its type. We keep the preprocessor line
|
| 525 | * parser state between calls in the global variable linestate, with
|
| 526 | * help from skipcomment().
|
| 527 |
|
| 528 | static Linetype
|
| 529 | getline(void)
|
| 530 | {
|
| 531 | const char *cp;
|
| 532 | int cursym;
|
| 533 | int kwlen;
|
| 534 | Linetype retval;
|
| 535 | Comment_state wascomment;
|
| 536 |
|
| 537 | if (fgets(tline, MAXLINE, input) == NULL)
|
| 538 | return (LT_EOF);
|
| 539 | retval = LT_PLAIN;
|
| 540 | wascomment = incomment;
|
| 541 | cp = skipcomment(tline);
|
| 542 | if (linestate == LS_START) {
|
| 543 | if (*cp == '#') {
|
| 544 | linestate = LS_HASH;
|
| 545 | cp = skipcomment(cp + 1);
|
| 546 | } else if (*cp != '\0')
|
| 547 | linestate = LS_DIRTY;
|
| 548 | }
|
| 549 | if (!incomment && linestate == LS_HASH) {
|
| 550 | keyword = tline + (cp - tline);
|
| 551 | cp = skipsym(cp);
|
| 552 | kwlen = cp - keyword;
|
| 553 |
|
| 554 | if (strncmp(cp, "\\\n", 2) == 0)
|
| 555 | Eioccc();
|
| 556 | if (strlcmp("ifdef", keyword, kwlen) == 0 ||
|
| 557 | strlcmp("ifndef", keyword, kwlen) == 0) {
|
| 558 | cp = skipcomment(cp);
|
| 559 | if ((cursym = findsym(cp)) < 0)
|
| 560 | retval = LT_IF;
|
| 561 | else {
|
| 562 | retval = (keyword[2] == 'n')
|
| 563 | ? LT_FALSE : LT_TRUE;
|
| 564 | if (value[cursym] == NULL)
|
| 565 | retval = (retval == LT_TRUE)
|
| 566 | ? LT_FALSE : LT_TRUE;
|
| 567 | if (ignore[cursym])
|
| 568 | retval = (retval == LT_TRUE)
|
| 569 | ? LT_TRUEI : LT_FALSEI;
|
| 570 | }
|
| 571 | cp = skipsym(cp);
|
| 572 | } else if (strlcmp("if", keyword, kwlen) == 0)
|
| 573 | retval = ifeval(&cp);
|
| 574 | else if (strlcmp("elif", keyword, kwlen) == 0)
|
| 575 | retval = ifeval(&cp) - LT_IF + LT_ELIF;
|
| 576 | else if (strlcmp("else", keyword, kwlen) == 0)
|
| 577 | retval = LT_ELSE;
|
| 578 | else if (strlcmp("endif", keyword, kwlen) == 0)
|
| 579 | retval = LT_ENDIF;
|
| 580 | else {
|
| 581 | linestate = LS_DIRTY;
|
| 582 | retval = LT_PLAIN;
|
| 583 | }
|
| 584 | cp = skipcomment(cp);
|
| 585 | if (*cp != '\0') {
|
| 586 | linestate = LS_DIRTY;
|
| 587 | if (retval == LT_TRUE || retval == LT_FALSE ||
|
| 588 | retval == LT_TRUEI || retval == LT_FALSEI)
|
| 589 | retval = LT_IF;
|
| 590 | if (retval == LT_ELTRUE || retval == LT_ELFALSE)
|
| 591 | retval = LT_ELIF;
|
| 592 | }
|
| 593 | if (retval != LT_PLAIN && (wascomment || incomment)) {
|
| 594 | retval += LT_DODGY;
|
| 595 | if (incomment)
|
| 596 | linestate = LS_DIRTY;
|
| 597 | }
|
| 598 |
|
| 599 | if (linestate == LS_HASH)
|
| 600 | abort();
|
| 601 | }
|
| 602 | if (linestate == LS_DIRTY) {
|
| 603 | while (*cp != '\0')
|
| 604 | cp = skipcomment(cp + 1);
|
| 605 | }
|
| 606 | debug("parser %s comment %s line",
|
| 607 | comment_name[incomment], linestate_name[linestate]);
|
| 608 | return (retval);
|
| 609 | }
|
| 610 |
|
| 611 | |
| 612 | * These are the binary operators that are supported by the expression
|
| 613 | * evaluator. Note that if support for division is added then we also
|
| 614 | * need short-circuiting booleans because of divide-by-zero.
|
| 615 |
|
| 616 | static int op_lt(int a, int b) { return (a < b); }
|
| 617 | static int op_gt(int a, int b) { return (a > b); }
|
| 618 | static int op_le(int a, int b) { return (a <= b); }
|
| 619 | static int op_ge(int a, int b) { return (a >= b); }
|
| 620 | static int op_eq(int a, int b) { return (a == b); }
|
| 621 | static int op_ne(int a, int b) { return (a != b); }
|
| 622 | static int op_or(int a, int b) { return (a || b); }
|
| 623 | static int op_and(int a, int b) { return (a && b); }
|
| 624 |
|
| 625 | |
| 626 | * An evaluation function takes three arguments, as follows: (1) a pointer to
|
| 627 | * an element of the precedence table which lists the operators at the current
|
| 628 | * level of precedence; (2) a pointer to an integer which will receive the
|
| 629 | * value of the expression; and (3) a pointer to a char* that points to the
|
| 630 | * expression to be evaluated and that is updated to the end of the expression
|
| 631 | * when evaluation is complete. The function returns LT_FALSE if the value of
|
| 632 | * the expression is zero, LT_TRUE if it is non-zero, or LT_IF if the
|
| 633 | * expression could not be evaluated.
|
| 634 |
|
| 635 | struct ops;
|
| 636 |
|
| 637 | typedef Linetype eval_fn(const struct ops *, int *, const char **);
|
| 638 |
|
| 639 | static eval_fn eval_table, eval_unary;
|
| 640 |
|
| 641 | |
| 642 | * The precedence table. Expressions involving binary operators are evaluated
|
| 643 | * in a table-driven way by eval_table. When it evaluates a subexpression it
|
| 644 | * calls the inner function with its first argument pointing to the next
|
| 645 | * element of the table. Innermost expressions have special non-table-driven
|
| 646 | * handling.
|
| 647 |
|
| 648 | static const struct ops {
|
| 649 | eval_fn *inner;
|
| 650 | struct op {
|
| 651 | const char *str;
|
| 652 | int (*fn)(int, int);
|
| 653 | } op[5];
|
| 654 | } eval_ops[] = {
|
| 655 | { eval_table, { { "||", op_or } } },
|
| 656 | { eval_table, { { "&&", op_and } } },
|
| 657 | { eval_table, { { "==", op_eq },
|
| 658 | { "!=", op_ne } } },
|
| 659 | { eval_unary, { { "<=", op_le },
|
| 660 | { ">=", op_ge },
|
| 661 | { "<", op_lt },
|
| 662 | { ">", op_gt } } }
|
| 663 | };
|
| 664 |
|
| 665 | |
| 666 | * Function for evaluating the innermost parts of expressions,
|
| 667 | * viz. !expr (expr) defined(symbol) symbol number
|
| 668 | * We reset the keepthis flag when we find a non-constant subexpression.
|
| 669 |
|
| 670 | static Linetype
|
| 671 | eval_unary(const struct ops *ops, int *valp, const char **cpp)
|
| 672 | {
|
| 673 | const char *cp;
|
| 674 | char *ep;
|
| 675 | int sym;
|
| 676 |
|
| 677 | cp = skipcomment(*cpp);
|
| 678 | if (*cp == '!') {
|
| 679 | debug("eval%d !", ops - eval_ops);
|
| 680 | cp++;
|
| 681 | if (eval_unary(ops, valp, &cp) == LT_IF)
|
| 682 | return (LT_IF);
|
| 683 | *valp = !*valp;
|
| 684 | } else if (*cp == '(') {
|
| 685 | cp++;
|
| 686 | debug("eval%d (", ops - eval_ops);
|
| 687 | if (eval_table(eval_ops, valp, &cp) == LT_IF)
|
| 688 | return (LT_IF);
|
| 689 | cp = skipcomment(cp);
|
| 690 | if (*cp++ != ')')
|
| 691 | return (LT_IF);
|
| 692 | } else if (isdigit((unsigned char)*cp)) {
|
| 693 | debug("eval%d number", ops - eval_ops);
|
| 694 | *valp = strtol(cp, &ep, 0);
|
| 695 | cp = skipsym(cp);
|
| 696 | } else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) {
|
| 697 | cp = skipcomment(cp+7);
|
| 698 | debug("eval%d defined", ops - eval_ops);
|
| 699 | if (*cp++ != '(')
|
| 700 | return (LT_IF);
|
| 701 | cp = skipcomment(cp);
|
| 702 | sym = findsym(cp);
|
| 703 | if (sym < 0)
|
| 704 | return (LT_IF);
|
| 705 | *valp = (value[sym] != NULL);
|
| 706 | cp = skipsym(cp);
|
| 707 | cp = skipcomment(cp);
|
| 708 | if (*cp++ != ')')
|
| 709 | return (LT_IF);
|
| 710 | keepthis = false;
|
| 711 | } else if (!endsym(*cp)) {
|
| 712 | debug("eval%d symbol", ops - eval_ops);
|
| 713 | sym = findsym(cp);
|
| 714 | if (sym < 0)
|
| 715 | return (LT_IF);
|
| 716 | if (value[sym] == NULL)
|
| 717 | *valp = 0;
|
| 718 | else {
|
| 719 | *valp = strtol(value[sym], &ep, 0);
|
| 720 | if (*ep != '\0' || ep == value[sym])
|
| 721 | return (LT_IF);
|
| 722 | }
|
| 723 | cp = skipsym(cp);
|
| 724 | keepthis = false;
|
| 725 | } else {
|
| 726 | debug("eval%d bad expr", ops - eval_ops);
|
| 727 | return (LT_IF);
|
| 728 | }
|
| 729 |
|
| 730 | *cpp = cp;
|
| 731 | debug("eval%d = %d", ops - eval_ops, *valp);
|
| 732 | return (*valp ? LT_TRUE : LT_FALSE);
|
| 733 | }
|
| 734 |
|
| 735 | |
| 736 | * Table-driven evaluation of binary operators.
|
| 737 |
|
| 738 | static Linetype
|
| 739 | eval_table(const struct ops *ops, int *valp, const char **cpp)
|
| 740 | {
|
| 741 | const struct op *op;
|
| 742 | const char *cp;
|
| 743 | int val;
|
| 744 |
|
| 745 | debug("eval%d", ops - eval_ops);
|
| 746 | cp = *cpp;
|
| 747 | if (ops->inner(ops+1, valp, &cp) == LT_IF)
|
| 748 | return (LT_IF);
|
| 749 | for (;;) {
|
| 750 | cp = skipcomment(cp);
|
| 751 | for (op = ops->op; op->str != NULL; op++)
|
| 752 | if (strncmp(cp, op->str, strlen(op->str)) == 0)
|
| 753 | break;
|
| 754 | if (op->str == NULL)
|
| 755 | break;
|
| 756 | cp += strlen(op->str);
|
| 757 | debug("eval%d %s", ops - eval_ops, op->str);
|
| 758 | if (ops->inner(ops+1, &val, &cp) == LT_IF)
|
| 759 | return (LT_IF);
|
| 760 | *valp = op->fn(*valp, val);
|
| 761 | }
|
| 762 |
|
| 763 | *cpp = cp;
|
| 764 | debug("eval%d = %d", ops - eval_ops, *valp);
|
| 765 | return (*valp ? LT_TRUE : LT_FALSE);
|
| 766 | }
|
| 767 |
|
| 768 | |
| 769 | * Evaluate the expression on a #if or #elif line. If we can work out
|
| 770 | * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we
|
| 771 | * return just a generic LT_IF.
|
| 772 |
|
| 773 | static Linetype
|
| 774 | ifeval(const char **cpp)
|
| 775 | {
|
| 776 | int ret;
|
| 777 | int val;
|
| 778 |
|
| 779 | debug("eval %s", *cpp);
|
| 780 | keepthis = killconsts ? false : true;
|
| 781 | ret = eval_table(eval_ops, &val, cpp);
|
| 782 | debug("eval = %d", val);
|
| 783 | return (keepthis ? LT_IF : ret);
|
| 784 | }
|
| 785 |
|
| 786 | |
| 787 | * Skip over comments, strings, and character literals and stop at the
|
| 788 | * next character position that is not whitespace. Between calls we keep
|
| 789 | * the comment state in the global variable incomment, and we also adjust
|
| 790 | * the global variable linestate when we see a newline.
|
| 791 | * XXX: doesn't cope with the buffer splitting inside a state transition.
|
| 792 |
|
| 793 | static const char *
|
| 794 | skipcomment(const char *cp)
|
| 795 | {
|
| 796 | if (text || ignoring[depth]) {
|
| 797 | for (; isspace((unsigned char)*cp); cp++)
|
| 798 | if (*cp == '\n')
|
| 799 | linestate = LS_START;
|
| 800 | return (cp);
|
| 801 | }
|
| 802 | while (*cp != '\0')
|
| 803 |
|
| 804 | if (strncmp(cp, "\\\n", 2) == 0)
|
| 805 | cp += 2;
|
| 806 | else switch (incomment) {
|
| 807 | case NO_COMMENT:
|
| 808 | if (strncmp(cp, "/\\\n", 3) == 0) {
|
| 809 | incomment = STARTING_COMMENT;
|
| 810 | cp += 3;
|
| 811 | } else if (strncmp(cp, "/*", 2) == 0) {
|
| 812 | incomment = C_COMMENT;
|
| 813 | cp += 2;
|
| 814 | } else if (strncmp(cp, "//", 2) == 0) {
|
| 815 | incomment = CXX_COMMENT;
|
| 816 | cp += 2;
|
| 817 | } else if (strncmp(cp, "\'", 1) == 0) {
|
| 818 | incomment = CHAR_LITERAL;
|
| 819 | linestate = LS_DIRTY;
|
| 820 | cp += 1;
|
| 821 | } else if (strncmp(cp, "\"", 1) == 0) {
|
| 822 | incomment = STRING_LITERAL;
|
| 823 | linestate = LS_DIRTY;
|
| 824 | cp += 1;
|
| 825 | } else if (strncmp(cp, "\n", 1) == 0) {
|
| 826 | linestate = LS_START;
|
| 827 | cp += 1;
|
| 828 | } else if (strchr(" \t", *cp) != NULL) {
|
| 829 | cp += 1;
|
| 830 | } else
|
| 831 | return (cp);
|
| 832 | continue;
|
| 833 | case CXX_COMMENT:
|
| 834 | if (strncmp(cp, "\n", 1) == 0) {
|
| 835 | incomment = NO_COMMENT;
|
| 836 | linestate = LS_START;
|
| 837 | }
|
| 838 | cp += 1;
|
| 839 | continue;
|
| 840 | case CHAR_LITERAL:
|
| 841 | case STRING_LITERAL:
|
| 842 | if ((incomment == CHAR_LITERAL && cp[0] == '\'') ||
|
| 843 | (incomment == STRING_LITERAL && cp[0] == '\"')) {
|
| 844 | incomment = NO_COMMENT;
|
| 845 | cp += 1;
|
| 846 | } else if (cp[0] == '\\') {
|
| 847 | if (cp[1] == '\0')
|
| 848 | cp += 1;
|
| 849 | else
|
| 850 | cp += 2;
|
| 851 | } else if (strncmp(cp, "\n", 1) == 0) {
|
| 852 | if (incomment == CHAR_LITERAL)
|
| 853 | error("unterminated char literal");
|
| 854 | else
|
| 855 | error("unterminated string literal");
|
| 856 | } else
|
| 857 | cp += 1;
|
| 858 | continue;
|
| 859 | case C_COMMENT:
|
| 860 | if (strncmp(cp, "*\\\n", 3) == 0) {
|
| 861 | incomment = FINISHING_COMMENT;
|
| 862 | cp += 3;
|
| 863 | } else if (strncmp(cp, "*/", 2) == 0) {
|
| 864 | incomment = NO_COMMENT;
|
| 865 | cp += 2;
|
| 866 | } else
|
| 867 | cp += 1;
|
| 868 | continue;
|
| 869 | case STARTING_COMMENT:
|
| 870 | if (*cp == '*') {
|
| 871 | incomment = C_COMMENT;
|
| 872 | cp += 1;
|
| 873 | } else if (*cp == '/') {
|
| 874 | incomment = CXX_COMMENT;
|
| 875 | cp += 1;
|
| 876 | } else {
|
| 877 | incomment = NO_COMMENT;
|
| 878 | linestate = LS_DIRTY;
|
| 879 | }
|
| 880 | continue;
|
| 881 | case FINISHING_COMMENT:
|
| 882 | if (*cp == '/') {
|
| 883 | incomment = NO_COMMENT;
|
| 884 | cp += 1;
|
| 885 | } else
|
| 886 | incomment = C_COMMENT;
|
| 887 | continue;
|
| 888 | default:
|
| 889 | abort();
|
| 890 | }
|
| 891 | return (cp);
|
| 892 | }
|
| 893 |
|
| 894 | |
| 895 | * Skip over an identifier.
|
| 896 |
|
| 897 | static const char *
|
| 898 | skipsym(const char *cp)
|
| 899 | {
|
| 900 | while (!endsym(*cp))
|
| 901 | ++cp;
|
| 902 | return (cp);
|
| 903 | }
|
| 904 |
|
| 905 | |
| 906 | * Look for the symbol in the symbol table. If is is found, we return
|
| 907 | * the symbol table index, else we return -1.
|
| 908 |
|
| 909 | static int
|
| 910 | findsym(const char *str)
|
| 911 | {
|
| 912 | const char *cp;
|
| 913 | int symind;
|
| 914 |
|
| 915 | cp = skipsym(str);
|
| 916 | if (cp == str)
|
| 917 | return (-1);
|
| 918 | if (symlist) {
|
| 919 | printf("%.*s\n", (int)(cp-str), str);
|
| 920 |
|
| 921 | return (0);
|
| 922 | }
|
| 923 | for (symind = 0; symind < nsyms; ++symind) {
|
| 924 | if (strlcmp(symname[symind], str, cp-str) == 0) {
|
| 925 | debug("findsym %s %s", symname[symind],
|
| 926 | value[symind] ? value[symind] : "");
|
| 927 | return (symind);
|
| 928 | }
|
| 929 | }
|
| 930 | return (-1);
|
| 931 | }
|
| 932 |
|
| 933 | |
| 934 | * Add a symbol to the symbol table.
|
| 935 |
|
| 936 | static void
|
| 937 | addsym(bool ignorethis, bool definethis, char *sym)
|
| 938 | {
|
| 939 | int symind;
|
| 940 | char *val;
|
| 941 |
|
| 942 | symind = findsym(sym);
|
| 943 | if (symind < 0) {
|
| 944 | if (nsyms >= MAXSYMS)
|
| 945 | errx(2, "too many symbols");
|
| 946 | symind = nsyms++;
|
| 947 | }
|
| 948 | symname[symind] = sym;
|
| 949 | ignore[symind] = ignorethis;
|
| 950 | val = sym + (skipsym(sym) - sym);
|
| 951 | if (definethis) {
|
| 952 | if (*val == '=') {
|
| 953 | value[symind] = val+1;
|
| 954 | *val = '\0';
|
| 955 | } else if (*val == '\0')
|
| 956 | value[symind] = "";
|
| 957 | else
|
| 958 | usage();
|
| 959 | } else {
|
| 960 | if (*val != '\0')
|
| 961 | usage();
|
| 962 | value[symind] = NULL;
|
| 963 | }
|
| 964 | }
|
| 965 |
|
| 966 | |
| 967 | * Compare s with n characters of t.
|
| 968 | * The same as strncmp() except that it checks that s[n] == '\0'.
|
| 969 |
|
| 970 | static int
|
| 971 | strlcmp(const char *s, const char *t, size_t n)
|
| 972 | {
|
| 973 | while (n-- && *t != '\0')
|
| 974 | if (*s != *t)
|
| 975 | return ((unsigned char)*s - (unsigned char)*t);
|
| 976 | else
|
| 977 | ++s, ++t;
|
| 978 | return ((unsigned char)*s);
|
| 979 | }
|
| 980 |
|
| 981 | |
| 982 | * Diagnostics.
|
| 983 |
|
| 984 | static void
|
| 985 | debug(const char *msg, ...)
|
| 986 | {
|
| 987 | va_list ap;
|
| 988 |
|
| 989 | if (debugging) {
|
| 990 | va_start(ap, msg);
|
| 991 | vwarnx(msg, ap);
|
| 992 | va_end(ap);
|
| 993 | }
|
| 994 | }
|
| 995 |
|
| 996 | static void
|
| 997 | error(const char *msg)
|
| 998 | {
|
| 999 | if (depth == 0)
|
| 1000 | warnx("%s: %d: %s", filename, linenum, msg);
|
| 1001 | else
|
| 1002 | warnx("%s: %d: %s (#if line %d depth %d)",
|
| 1003 | filename, linenum, msg, stifline[depth], depth);
|
| 1004 | errx(2, "output may be truncated");
|
| 1005 | }
|