Nicholas Flintham | 1e3d311 | 2013-04-10 10:48:38 +0100 | [diff] [blame^] | 1 | /* |
| 2 | * Copyright (c) 2002 - 2011 Tony Finch <dot@dotat.at> |
| 3 | * |
| 4 | * Redistribution and use in source and binary forms, with or without |
| 5 | * modification, are permitted provided that the following conditions |
| 6 | * are met: |
| 7 | * 1. Redistributions of source code must retain the above copyright |
| 8 | * notice, this list of conditions and the following disclaimer. |
| 9 | * 2. Redistributions in binary form must reproduce the above copyright |
| 10 | * notice, this list of conditions and the following disclaimer in the |
| 11 | * documentation and/or other materials provided with the distribution. |
| 12 | * |
| 13 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND |
| 14 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 15 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 16 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE |
| 17 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 18 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 19 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 20 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 21 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 22 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 23 | * SUCH DAMAGE. |
| 24 | */ |
| 25 | |
| 26 | /* |
| 27 | * unifdef - remove ifdef'ed lines |
| 28 | * |
| 29 | * This code was derived from software contributed to Berkeley by Dave Yost. |
| 30 | * It was rewritten to support ANSI C by Tony Finch. The original version |
| 31 | * of unifdef carried the 4-clause BSD copyright licence. None of its code |
| 32 | * remains in this version (though some of the names remain) so it now |
| 33 | * carries a more liberal licence. |
| 34 | * |
| 35 | * Wishlist: |
| 36 | * provide an option which will append the name of the |
| 37 | * appropriate symbol after #else's and #endif's |
| 38 | * provide an option which will check symbols after |
| 39 | * #else's and #endif's to see that they match their |
| 40 | * corresponding #ifdef or #ifndef |
| 41 | * |
| 42 | * These require better buffer handling, which would also make |
| 43 | * it possible to handle all "dodgy" directives correctly. |
| 44 | */ |
| 45 | |
| 46 | #include <sys/types.h> |
| 47 | #include <sys/stat.h> |
| 48 | |
| 49 | #include <ctype.h> |
| 50 | #include <err.h> |
| 51 | #include <errno.h> |
| 52 | #include <stdarg.h> |
| 53 | #include <stdbool.h> |
| 54 | #include <stdio.h> |
| 55 | #include <stdlib.h> |
| 56 | #include <string.h> |
| 57 | #include <unistd.h> |
| 58 | |
| 59 | const char copyright[] = |
| 60 | "@(#) $Version: unifdef-2.5 $\n" |
| 61 | "@(#) $Author: Tony Finch (dot@dotat.at) $\n" |
| 62 | "@(#) $URL: http://dotat.at/prog/unifdef $\n" |
| 63 | ; |
| 64 | |
| 65 | typedef enum { |
| 66 | LT_TRUEI, |
| 67 | LT_FALSEI, |
| 68 | LT_IF, |
| 69 | LT_TRUE, |
| 70 | LT_FALSE, |
| 71 | LT_ELIF, |
| 72 | LT_ELTRUE, |
| 73 | LT_ELFALSE, |
| 74 | LT_ELSE, |
| 75 | LT_ENDIF, |
| 76 | LT_DODGY, |
| 77 | LT_DODGY_LAST = LT_DODGY + LT_ENDIF, |
| 78 | LT_PLAIN, |
| 79 | LT_EOF, |
| 80 | LT_ERROR, |
| 81 | LT_COUNT |
| 82 | } Linetype; |
| 83 | |
| 84 | static char const * const linetype_name[] = { |
| 85 | "TRUEI", "FALSEI", "IF", "TRUE", "FALSE", |
| 86 | "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF", |
| 87 | "DODGY TRUEI", "DODGY FALSEI", |
| 88 | "DODGY IF", "DODGY TRUE", "DODGY FALSE", |
| 89 | "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE", |
| 90 | "DODGY ELSE", "DODGY ENDIF", |
| 91 | "PLAIN", "EOF", "ERROR" |
| 92 | }; |
| 93 | |
| 94 | typedef enum { |
| 95 | IS_OUTSIDE, |
| 96 | IS_FALSE_PREFIX, |
| 97 | IS_TRUE_PREFIX, |
| 98 | IS_PASS_MIDDLE, |
| 99 | IS_FALSE_MIDDLE, |
| 100 | IS_TRUE_MIDDLE, |
| 101 | IS_PASS_ELSE, |
| 102 | IS_FALSE_ELSE, |
| 103 | IS_TRUE_ELSE, |
| 104 | IS_FALSE_TRAILER, |
| 105 | IS_COUNT |
| 106 | } Ifstate; |
| 107 | |
| 108 | static char const * const ifstate_name[] = { |
| 109 | "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX", |
| 110 | "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE", |
| 111 | "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE", |
| 112 | "FALSE_TRAILER" |
| 113 | }; |
| 114 | |
| 115 | typedef enum { |
| 116 | NO_COMMENT = false, |
| 117 | C_COMMENT, |
| 118 | CXX_COMMENT, |
| 119 | STARTING_COMMENT, |
| 120 | FINISHING_COMMENT, |
| 121 | CHAR_LITERAL, |
| 122 | STRING_LITERAL |
| 123 | } Comment_state; |
| 124 | |
| 125 | static char const * const comment_name[] = { |
| 126 | "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING" |
| 127 | }; |
| 128 | |
| 129 | typedef enum { |
| 130 | LS_START, |
| 131 | LS_HASH, |
| 132 | LS_DIRTY |
| 133 | } Line_state; |
| 134 | |
| 135 | static char const * const linestate_name[] = { |
| 136 | "START", "HASH", "DIRTY" |
| 137 | }; |
| 138 | |
| 139 | #define MAXDEPTH 64 |
| 140 | #define MAXLINE 4096 |
| 141 | #define MAXSYMS 4096 |
| 142 | |
| 143 | #define EDITSLOP 10 |
| 144 | |
| 145 | #define TEMPLATE "unifdef.XXXXXX" |
| 146 | |
| 147 | |
| 148 | static bool compblank; |
| 149 | static bool lnblank; |
| 150 | static bool complement; |
| 151 | static bool debugging; |
| 152 | static bool iocccok; |
| 153 | static bool strictlogic; |
| 154 | static bool killconsts; |
| 155 | static bool lnnum; |
| 156 | static bool symlist; |
| 157 | static bool symdepth; |
| 158 | static bool text; |
| 159 | |
| 160 | static const char *symname[MAXSYMS]; |
| 161 | static const char *value[MAXSYMS]; |
| 162 | static bool ignore[MAXSYMS]; |
| 163 | static int nsyms; |
| 164 | |
| 165 | static FILE *input; |
| 166 | static const char *filename; |
| 167 | static int linenum; |
| 168 | static FILE *output; |
| 169 | static const char *ofilename; |
| 170 | static bool overwriting; |
| 171 | static char tempname[FILENAME_MAX]; |
| 172 | |
| 173 | static char tline[MAXLINE+EDITSLOP]; |
| 174 | static char *keyword; |
| 175 | |
| 176 | static const char *newline; |
| 177 | static const char newline_unix[] = "\n"; |
| 178 | static const char newline_crlf[] = "\r\n"; |
| 179 | |
| 180 | static Comment_state incomment; |
| 181 | static Line_state linestate; |
| 182 | static Ifstate ifstate[MAXDEPTH]; |
| 183 | static bool ignoring[MAXDEPTH]; |
| 184 | static int stifline[MAXDEPTH]; |
| 185 | static int depth; |
| 186 | static int delcount; |
| 187 | static unsigned blankcount; |
| 188 | static unsigned blankmax; |
| 189 | static bool constexpr; |
| 190 | static bool zerosyms = true; |
| 191 | static bool firstsym; |
| 192 | |
| 193 | static int exitstat; |
| 194 | |
| 195 | static void addsym(bool, bool, char *); |
| 196 | static void closeout(void); |
| 197 | static void debug(const char *, ...); |
| 198 | static void done(void); |
| 199 | static void error(const char *); |
| 200 | static int findsym(const char *); |
| 201 | static void flushline(bool); |
| 202 | static Linetype parseline(void); |
| 203 | static Linetype ifeval(const char **); |
| 204 | static void ignoreoff(void); |
| 205 | static void ignoreon(void); |
| 206 | static void keywordedit(const char *); |
| 207 | static void nest(void); |
| 208 | static void process(void); |
| 209 | static const char *skipargs(const char *); |
| 210 | static const char *skipcomment(const char *); |
| 211 | static const char *skipsym(const char *); |
| 212 | static void state(Ifstate); |
| 213 | static int strlcmp(const char *, const char *, size_t); |
| 214 | static void unnest(void); |
| 215 | static void usage(void); |
| 216 | static void version(void); |
| 217 | |
| 218 | #define endsym(c) (!isalnum((unsigned char)c) && c != '_') |
| 219 | |
| 220 | int |
| 221 | main(int argc, char *argv[]) |
| 222 | { |
| 223 | int opt; |
| 224 | |
| 225 | while ((opt = getopt(argc, argv, "i:D:U:I:o:bBcdeKklnsStV")) != -1) |
| 226 | switch (opt) { |
| 227 | case 'i': |
| 228 | opt = *optarg++; |
| 229 | if (opt == 'D') |
| 230 | addsym(true, true, optarg); |
| 231 | else if (opt == 'U') |
| 232 | addsym(true, false, optarg); |
| 233 | else |
| 234 | usage(); |
| 235 | break; |
| 236 | case 'D': |
| 237 | addsym(false, true, optarg); |
| 238 | break; |
| 239 | case 'U': |
| 240 | addsym(false, false, optarg); |
| 241 | break; |
| 242 | case 'I': |
| 243 | break; |
| 244 | case 'b': |
| 245 | case 'l': |
| 246 | lnblank = true; |
| 247 | break; |
| 248 | case 'B': |
| 249 | compblank = true; |
| 250 | break; |
| 251 | case 'c': |
| 252 | complement = true; |
| 253 | break; |
| 254 | case 'd': |
| 255 | debugging = true; |
| 256 | break; |
| 257 | case 'e': |
| 258 | iocccok = true; |
| 259 | break; |
| 260 | case 'K': |
| 261 | strictlogic = true; |
| 262 | break; |
| 263 | case 'k': |
| 264 | killconsts = true; |
| 265 | break; |
| 266 | case 'n': |
| 267 | lnnum = true; |
| 268 | break; |
| 269 | case 'o': |
| 270 | ofilename = optarg; |
| 271 | break; |
| 272 | case 's': |
| 273 | symlist = true; |
| 274 | break; |
| 275 | case 'S': |
| 276 | symlist = symdepth = true; |
| 277 | break; |
| 278 | case 't': |
| 279 | text = true; |
| 280 | break; |
| 281 | case 'V': |
| 282 | version(); |
| 283 | default: |
| 284 | usage(); |
| 285 | } |
| 286 | argc -= optind; |
| 287 | argv += optind; |
| 288 | if (compblank && lnblank) |
| 289 | errx(2, "-B and -b are mutually exclusive"); |
| 290 | if (argc > 1) { |
| 291 | errx(2, "can only do one file"); |
| 292 | } else if (argc == 1 && strcmp(*argv, "-") != 0) { |
| 293 | filename = *argv; |
| 294 | input = fopen(filename, "rb"); |
| 295 | if (input == NULL) |
| 296 | err(2, "can't open %s", filename); |
| 297 | } else { |
| 298 | filename = "[stdin]"; |
| 299 | input = stdin; |
| 300 | } |
| 301 | if (ofilename == NULL) { |
| 302 | ofilename = "[stdout]"; |
| 303 | output = stdout; |
| 304 | } else { |
| 305 | struct stat ist, ost; |
| 306 | if (stat(ofilename, &ost) == 0 && |
| 307 | fstat(fileno(input), &ist) == 0) |
| 308 | overwriting = (ist.st_dev == ost.st_dev |
| 309 | && ist.st_ino == ost.st_ino); |
| 310 | if (overwriting) { |
| 311 | const char *dirsep; |
| 312 | int ofd; |
| 313 | |
| 314 | dirsep = strrchr(ofilename, '/'); |
| 315 | if (dirsep != NULL) |
| 316 | snprintf(tempname, sizeof(tempname), |
| 317 | "%.*s/" TEMPLATE, |
| 318 | (int)(dirsep - ofilename), ofilename); |
| 319 | else |
| 320 | snprintf(tempname, sizeof(tempname), |
| 321 | TEMPLATE); |
| 322 | ofd = mkstemp(tempname); |
| 323 | if (ofd != -1) |
| 324 | output = fdopen(ofd, "wb+"); |
| 325 | if (output == NULL) |
| 326 | err(2, "can't create temporary file"); |
| 327 | fchmod(ofd, ist.st_mode & (S_IRWXU|S_IRWXG|S_IRWXO)); |
| 328 | } else { |
| 329 | output = fopen(ofilename, "wb"); |
| 330 | if (output == NULL) |
| 331 | err(2, "can't open %s", ofilename); |
| 332 | } |
| 333 | } |
| 334 | process(); |
| 335 | abort(); |
| 336 | } |
| 337 | |
| 338 | static void |
| 339 | version(void) |
| 340 | { |
| 341 | const char *c = copyright; |
| 342 | for (;;) { |
| 343 | while (*++c != '$') |
| 344 | if (*c == '\0') |
| 345 | exit(0); |
| 346 | while (*++c != '$') |
| 347 | putc(*c, stderr); |
| 348 | putc('\n', stderr); |
| 349 | } |
| 350 | } |
| 351 | |
| 352 | static void |
| 353 | usage(void) |
| 354 | { |
| 355 | fprintf(stderr, "usage: unifdef [-bBcdeKknsStV] [-Ipath]" |
| 356 | " [-Dsym[=val]] [-Usym] [-iDsym[=val]] [-iUsym] ... [file]\n"); |
| 357 | exit(2); |
| 358 | } |
| 359 | |
| 360 | typedef void state_fn(void); |
| 361 | |
| 362 | static void Eelif (void) { error("Inappropriate #elif"); } |
| 363 | static void Eelse (void) { error("Inappropriate #else"); } |
| 364 | static void Eendif(void) { error("Inappropriate #endif"); } |
| 365 | static void Eeof (void) { error("Premature EOF"); } |
| 366 | static void Eioccc(void) { error("Obfuscated preprocessor control line"); } |
| 367 | static void print (void) { flushline(true); } |
| 368 | static void drop (void) { flushline(false); } |
| 369 | static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); } |
| 370 | static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); } |
| 371 | static void Selse (void) { drop(); state(IS_TRUE_ELSE); } |
| 372 | static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); } |
| 373 | static void Pelse (void) { print(); state(IS_PASS_ELSE); } |
| 374 | static void Pendif(void) { print(); unnest(); } |
| 375 | static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); } |
| 376 | static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); } |
| 377 | static void Delse (void) { drop(); state(IS_FALSE_ELSE); } |
| 378 | static void Dendif(void) { drop(); unnest(); } |
| 379 | static void Fdrop (void) { nest(); Dfalse(); } |
| 380 | static void Fpass (void) { nest(); Pelif(); } |
| 381 | static void Ftrue (void) { nest(); Strue(); } |
| 382 | static void Ffalse(void) { nest(); Sfalse(); } |
| 383 | static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); } |
| 384 | static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); } |
| 385 | static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); } |
| 386 | static void Idrop (void) { Fdrop(); ignoreon(); } |
| 387 | static void Itrue (void) { Ftrue(); ignoreon(); } |
| 388 | static void Ifalse(void) { Ffalse(); ignoreon(); } |
| 389 | static void Mpass (void) { strncpy(keyword, "if ", 4); Pelif(); } |
| 390 | static void Mtrue (void) { keywordedit("else"); state(IS_TRUE_MIDDLE); } |
| 391 | static void Melif (void) { keywordedit("endif"); state(IS_FALSE_TRAILER); } |
| 392 | static void Melse (void) { keywordedit("endif"); state(IS_FALSE_ELSE); } |
| 393 | |
| 394 | static state_fn * const trans_table[IS_COUNT][LT_COUNT] = { |
| 395 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif, |
| 396 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eendif, |
| 397 | print, done, abort }, |
| 398 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif, |
| 399 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc, |
| 400 | drop, Eeof, abort }, |
| 401 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif, |
| 402 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc, |
| 403 | print, Eeof, abort }, |
| 404 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif, |
| 405 | Oiffy, Oiffy, Fpass, Oif, Oif, Pelif, Oelif, Oelif, Pelse, Pendif, |
| 406 | print, Eeof, abort }, |
| 407 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif, |
| 408 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc, |
| 409 | drop, Eeof, abort }, |
| 410 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif, |
| 411 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Pendif, |
| 412 | print, Eeof, abort }, |
| 413 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif, |
| 414 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Pendif, |
| 415 | print, Eeof, abort }, |
| 416 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif, |
| 417 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc, |
| 418 | drop, Eeof, abort }, |
| 419 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif, |
| 420 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eioccc, |
| 421 | print, Eeof, abort }, |
| 422 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif, |
| 423 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc, |
| 424 | drop, Eeof, abort } |
| 425 | }; |
| 426 | |
| 427 | static void |
| 428 | ignoreoff(void) |
| 429 | { |
| 430 | if (depth == 0) |
| 431 | abort(); |
| 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 | snprintf(keyword, tline + sizeof(tline) - keyword, |
| 443 | "%s%s", replacement, newline); |
| 444 | print(); |
| 445 | } |
| 446 | static void |
| 447 | nest(void) |
| 448 | { |
| 449 | if (depth > MAXDEPTH-1) |
| 450 | abort(); |
| 451 | if (depth == MAXDEPTH-1) |
| 452 | error("Too many levels of nesting"); |
| 453 | depth += 1; |
| 454 | stifline[depth] = linenum; |
| 455 | } |
| 456 | static void |
| 457 | unnest(void) |
| 458 | { |
| 459 | if (depth == 0) |
| 460 | abort(); |
| 461 | depth -= 1; |
| 462 | } |
| 463 | static void |
| 464 | state(Ifstate is) |
| 465 | { |
| 466 | ifstate[depth] = is; |
| 467 | } |
| 468 | |
| 469 | static void |
| 470 | flushline(bool keep) |
| 471 | { |
| 472 | if (symlist) |
| 473 | return; |
| 474 | if (keep ^ complement) { |
| 475 | bool blankline = tline[strspn(tline, " \t\r\n")] == '\0'; |
| 476 | if (blankline && compblank && blankcount != blankmax) { |
| 477 | delcount += 1; |
| 478 | blankcount += 1; |
| 479 | } else { |
| 480 | if (lnnum && delcount > 0) |
| 481 | printf("#line %d%s", linenum, newline); |
| 482 | fputs(tline, output); |
| 483 | delcount = 0; |
| 484 | blankmax = blankcount = blankline ? blankcount + 1 : 0; |
| 485 | } |
| 486 | } else { |
| 487 | if (lnblank) |
| 488 | fputs(newline, output); |
| 489 | exitstat = 1; |
| 490 | delcount += 1; |
| 491 | blankcount = 0; |
| 492 | } |
| 493 | if (debugging) |
| 494 | fflush(output); |
| 495 | } |
| 496 | |
| 497 | static void |
| 498 | process(void) |
| 499 | { |
| 500 | blankmax = blankcount = 1000; |
| 501 | for (;;) { |
| 502 | Linetype lineval = parseline(); |
| 503 | trans_table[ifstate[depth]][lineval](); |
| 504 | debug("process line %d %s -> %s depth %d", |
| 505 | linenum, linetype_name[lineval], |
| 506 | ifstate_name[ifstate[depth]], depth); |
| 507 | } |
| 508 | } |
| 509 | |
| 510 | static void |
| 511 | closeout(void) |
| 512 | { |
| 513 | if (symdepth && !zerosyms) |
| 514 | printf("\n"); |
| 515 | if (fclose(output) == EOF) { |
| 516 | warn("couldn't write to %s", ofilename); |
| 517 | if (overwriting) { |
| 518 | unlink(tempname); |
| 519 | errx(2, "%s unchanged", filename); |
| 520 | } else { |
| 521 | exit(2); |
| 522 | } |
| 523 | } |
| 524 | } |
| 525 | |
| 526 | static void |
| 527 | done(void) |
| 528 | { |
| 529 | if (incomment) |
| 530 | error("EOF in comment"); |
| 531 | closeout(); |
| 532 | if (overwriting && rename(tempname, ofilename) == -1) { |
| 533 | warn("couldn't rename temporary file"); |
| 534 | unlink(tempname); |
| 535 | errx(2, "%s unchanged", ofilename); |
| 536 | } |
| 537 | exit(exitstat); |
| 538 | } |
| 539 | |
| 540 | static Linetype |
| 541 | parseline(void) |
| 542 | { |
| 543 | const char *cp; |
| 544 | int cursym; |
| 545 | int kwlen; |
| 546 | Linetype retval; |
| 547 | Comment_state wascomment; |
| 548 | |
| 549 | linenum++; |
| 550 | if (fgets(tline, MAXLINE, input) == NULL) |
| 551 | return (LT_EOF); |
| 552 | if (newline == NULL) { |
| 553 | if (strrchr(tline, '\n') == strrchr(tline, '\r') + 1) |
| 554 | newline = newline_crlf; |
| 555 | else |
| 556 | newline = newline_unix; |
| 557 | } |
| 558 | retval = LT_PLAIN; |
| 559 | wascomment = incomment; |
| 560 | cp = skipcomment(tline); |
| 561 | if (linestate == LS_START) { |
| 562 | if (*cp == '#') { |
| 563 | linestate = LS_HASH; |
| 564 | firstsym = true; |
| 565 | cp = skipcomment(cp + 1); |
| 566 | } else if (*cp != '\0') |
| 567 | linestate = LS_DIRTY; |
| 568 | } |
| 569 | if (!incomment && linestate == LS_HASH) { |
| 570 | keyword = tline + (cp - tline); |
| 571 | cp = skipsym(cp); |
| 572 | kwlen = cp - keyword; |
| 573 | |
| 574 | if (strncmp(cp, "\\\r\n", 3) == 0 || |
| 575 | strncmp(cp, "\\\n", 2) == 0) |
| 576 | Eioccc(); |
| 577 | if (strlcmp("ifdef", keyword, kwlen) == 0 || |
| 578 | strlcmp("ifndef", keyword, kwlen) == 0) { |
| 579 | cp = skipcomment(cp); |
| 580 | if ((cursym = findsym(cp)) < 0) |
| 581 | retval = LT_IF; |
| 582 | else { |
| 583 | retval = (keyword[2] == 'n') |
| 584 | ? LT_FALSE : LT_TRUE; |
| 585 | if (value[cursym] == NULL) |
| 586 | retval = (retval == LT_TRUE) |
| 587 | ? LT_FALSE : LT_TRUE; |
| 588 | if (ignore[cursym]) |
| 589 | retval = (retval == LT_TRUE) |
| 590 | ? LT_TRUEI : LT_FALSEI; |
| 591 | } |
| 592 | cp = skipsym(cp); |
| 593 | } else if (strlcmp("if", keyword, kwlen) == 0) |
| 594 | retval = ifeval(&cp); |
| 595 | else if (strlcmp("elif", keyword, kwlen) == 0) |
| 596 | retval = ifeval(&cp) - LT_IF + LT_ELIF; |
| 597 | else if (strlcmp("else", keyword, kwlen) == 0) |
| 598 | retval = LT_ELSE; |
| 599 | else if (strlcmp("endif", keyword, kwlen) == 0) |
| 600 | retval = LT_ENDIF; |
| 601 | else { |
| 602 | linestate = LS_DIRTY; |
| 603 | retval = LT_PLAIN; |
| 604 | } |
| 605 | cp = skipcomment(cp); |
| 606 | if (*cp != '\0') { |
| 607 | linestate = LS_DIRTY; |
| 608 | if (retval == LT_TRUE || retval == LT_FALSE || |
| 609 | retval == LT_TRUEI || retval == LT_FALSEI) |
| 610 | retval = LT_IF; |
| 611 | if (retval == LT_ELTRUE || retval == LT_ELFALSE) |
| 612 | retval = LT_ELIF; |
| 613 | } |
| 614 | if (retval != LT_PLAIN && (wascomment || incomment)) { |
| 615 | retval += LT_DODGY; |
| 616 | if (incomment) |
| 617 | linestate = LS_DIRTY; |
| 618 | } |
| 619 | if (linestate == LS_HASH) { |
| 620 | size_t len = cp - tline; |
| 621 | if (fgets(tline + len, MAXLINE - len, input) == NULL) { |
| 622 | |
| 623 | strcpy(tline + len, newline); |
| 624 | cp += strlen(newline); |
| 625 | linestate = LS_START; |
| 626 | } else { |
| 627 | linestate = LS_DIRTY; |
| 628 | } |
| 629 | } |
| 630 | } |
| 631 | if (linestate == LS_DIRTY) { |
| 632 | while (*cp != '\0') |
| 633 | cp = skipcomment(cp + 1); |
| 634 | } |
| 635 | debug("parser line %d state %s comment %s line", linenum, |
| 636 | comment_name[incomment], linestate_name[linestate]); |
| 637 | return (retval); |
| 638 | } |
| 639 | |
| 640 | static Linetype op_strict(int *p, int v, Linetype at, Linetype bt) { |
| 641 | if(at == LT_IF || bt == LT_IF) return (LT_IF); |
| 642 | return (*p = v, v ? LT_TRUE : LT_FALSE); |
| 643 | } |
| 644 | static Linetype op_lt(int *p, Linetype at, int a, Linetype bt, int b) { |
| 645 | return op_strict(p, a < b, at, bt); |
| 646 | } |
| 647 | static Linetype op_gt(int *p, Linetype at, int a, Linetype bt, int b) { |
| 648 | return op_strict(p, a > b, at, bt); |
| 649 | } |
| 650 | static Linetype op_le(int *p, Linetype at, int a, Linetype bt, int b) { |
| 651 | return op_strict(p, a <= b, at, bt); |
| 652 | } |
| 653 | static Linetype op_ge(int *p, Linetype at, int a, Linetype bt, int b) { |
| 654 | return op_strict(p, a >= b, at, bt); |
| 655 | } |
| 656 | static Linetype op_eq(int *p, Linetype at, int a, Linetype bt, int b) { |
| 657 | return op_strict(p, a == b, at, bt); |
| 658 | } |
| 659 | static Linetype op_ne(int *p, Linetype at, int a, Linetype bt, int b) { |
| 660 | return op_strict(p, a != b, at, bt); |
| 661 | } |
| 662 | static Linetype op_or(int *p, Linetype at, int a, Linetype bt, int b) { |
| 663 | if (!strictlogic && (at == LT_TRUE || bt == LT_TRUE)) |
| 664 | return (*p = 1, LT_TRUE); |
| 665 | return op_strict(p, a || b, at, bt); |
| 666 | } |
| 667 | static Linetype op_and(int *p, Linetype at, int a, Linetype bt, int b) { |
| 668 | if (!strictlogic && (at == LT_FALSE || bt == LT_FALSE)) |
| 669 | return (*p = 0, LT_FALSE); |
| 670 | return op_strict(p, a && b, at, bt); |
| 671 | } |
| 672 | |
| 673 | struct ops; |
| 674 | |
| 675 | typedef Linetype eval_fn(const struct ops *, int *, const char **); |
| 676 | |
| 677 | static eval_fn eval_table, eval_unary; |
| 678 | |
| 679 | static const struct ops { |
| 680 | eval_fn *inner; |
| 681 | struct op { |
| 682 | const char *str; |
| 683 | Linetype (*fn)(int *, Linetype, int, Linetype, int); |
| 684 | } op[5]; |
| 685 | } eval_ops[] = { |
| 686 | { eval_table, { { "||", op_or } } }, |
| 687 | { eval_table, { { "&&", op_and } } }, |
| 688 | { eval_table, { { "==", op_eq }, |
| 689 | { "!=", op_ne } } }, |
| 690 | { eval_unary, { { "<=", op_le }, |
| 691 | { ">=", op_ge }, |
| 692 | { "<", op_lt }, |
| 693 | { ">", op_gt } } } |
| 694 | }; |
| 695 | |
| 696 | static Linetype |
| 697 | eval_unary(const struct ops *ops, int *valp, const char **cpp) |
| 698 | { |
| 699 | const char *cp; |
| 700 | char *ep; |
| 701 | int sym; |
| 702 | bool defparen; |
| 703 | Linetype lt; |
| 704 | |
| 705 | cp = skipcomment(*cpp); |
| 706 | if (*cp == '!') { |
| 707 | debug("eval%d !", ops - eval_ops); |
| 708 | cp++; |
| 709 | lt = eval_unary(ops, valp, &cp); |
| 710 | if (lt == LT_ERROR) |
| 711 | return (LT_ERROR); |
| 712 | if (lt != LT_IF) { |
| 713 | *valp = !*valp; |
| 714 | lt = *valp ? LT_TRUE : LT_FALSE; |
| 715 | } |
| 716 | } else if (*cp == '(') { |
| 717 | cp++; |
| 718 | debug("eval%d (", ops - eval_ops); |
| 719 | lt = eval_table(eval_ops, valp, &cp); |
| 720 | if (lt == LT_ERROR) |
| 721 | return (LT_ERROR); |
| 722 | cp = skipcomment(cp); |
| 723 | if (*cp++ != ')') |
| 724 | return (LT_ERROR); |
| 725 | } else if (isdigit((unsigned char)*cp)) { |
| 726 | debug("eval%d number", ops - eval_ops); |
| 727 | *valp = strtol(cp, &ep, 0); |
| 728 | if (ep == cp) |
| 729 | return (LT_ERROR); |
| 730 | lt = *valp ? LT_TRUE : LT_FALSE; |
| 731 | cp = skipsym(cp); |
| 732 | } else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) { |
| 733 | cp = skipcomment(cp+7); |
| 734 | debug("eval%d defined", ops - eval_ops); |
| 735 | if (*cp == '(') { |
| 736 | cp = skipcomment(cp+1); |
| 737 | defparen = true; |
| 738 | } else { |
| 739 | defparen = false; |
| 740 | } |
| 741 | sym = findsym(cp); |
| 742 | if (sym < 0) { |
| 743 | lt = LT_IF; |
| 744 | } else { |
| 745 | *valp = (value[sym] != NULL); |
| 746 | lt = *valp ? LT_TRUE : LT_FALSE; |
| 747 | } |
| 748 | cp = skipsym(cp); |
| 749 | cp = skipcomment(cp); |
| 750 | if (defparen && *cp++ != ')') |
| 751 | return (LT_ERROR); |
| 752 | constexpr = false; |
| 753 | } else if (!endsym(*cp)) { |
| 754 | debug("eval%d symbol", ops - eval_ops); |
| 755 | sym = findsym(cp); |
| 756 | cp = skipsym(cp); |
| 757 | if (sym < 0) { |
| 758 | lt = LT_IF; |
| 759 | cp = skipargs(cp); |
| 760 | } else if (value[sym] == NULL) { |
| 761 | *valp = 0; |
| 762 | lt = LT_FALSE; |
| 763 | } else { |
| 764 | *valp = strtol(value[sym], &ep, 0); |
| 765 | if (*ep != '\0' || ep == value[sym]) |
| 766 | return (LT_ERROR); |
| 767 | lt = *valp ? LT_TRUE : LT_FALSE; |
| 768 | cp = skipargs(cp); |
| 769 | } |
| 770 | constexpr = false; |
| 771 | } else { |
| 772 | debug("eval%d bad expr", ops - eval_ops); |
| 773 | return (LT_ERROR); |
| 774 | } |
| 775 | |
| 776 | *cpp = cp; |
| 777 | debug("eval%d = %d", ops - eval_ops, *valp); |
| 778 | return (lt); |
| 779 | } |
| 780 | |
| 781 | static Linetype |
| 782 | eval_table(const struct ops *ops, int *valp, const char **cpp) |
| 783 | { |
| 784 | const struct op *op; |
| 785 | const char *cp; |
| 786 | int val; |
| 787 | Linetype lt, rt; |
| 788 | |
| 789 | debug("eval%d", ops - eval_ops); |
| 790 | cp = *cpp; |
| 791 | lt = ops->inner(ops+1, valp, &cp); |
| 792 | if (lt == LT_ERROR) |
| 793 | return (LT_ERROR); |
| 794 | for (;;) { |
| 795 | cp = skipcomment(cp); |
| 796 | for (op = ops->op; op->str != NULL; op++) |
| 797 | if (strncmp(cp, op->str, strlen(op->str)) == 0) |
| 798 | break; |
| 799 | if (op->str == NULL) |
| 800 | break; |
| 801 | cp += strlen(op->str); |
| 802 | debug("eval%d %s", ops - eval_ops, op->str); |
| 803 | rt = ops->inner(ops+1, &val, &cp); |
| 804 | if (rt == LT_ERROR) |
| 805 | return (LT_ERROR); |
| 806 | lt = op->fn(valp, lt, *valp, rt, val); |
| 807 | } |
| 808 | |
| 809 | *cpp = cp; |
| 810 | debug("eval%d = %d", ops - eval_ops, *valp); |
| 811 | debug("eval%d lt = %s", ops - eval_ops, linetype_name[lt]); |
| 812 | return (lt); |
| 813 | } |
| 814 | |
| 815 | static Linetype |
| 816 | ifeval(const char **cpp) |
| 817 | { |
| 818 | int ret; |
| 819 | int val = 0; |
| 820 | |
| 821 | debug("eval %s", *cpp); |
| 822 | constexpr = killconsts ? false : true; |
| 823 | ret = eval_table(eval_ops, &val, cpp); |
| 824 | debug("eval = %d", val); |
| 825 | return (constexpr ? LT_IF : ret == LT_ERROR ? LT_IF : ret); |
| 826 | } |
| 827 | |
| 828 | static const char * |
| 829 | skipcomment(const char *cp) |
| 830 | { |
| 831 | if (text || ignoring[depth]) { |
| 832 | for (; isspace((unsigned char)*cp); cp++) |
| 833 | if (*cp == '\n') |
| 834 | linestate = LS_START; |
| 835 | return (cp); |
| 836 | } |
| 837 | while (*cp != '\0') |
| 838 | |
| 839 | if (strncmp(cp, "\\\r\n", 3) == 0) |
| 840 | cp += 3; |
| 841 | else if (strncmp(cp, "\\\n", 2) == 0) |
| 842 | cp += 2; |
| 843 | else switch (incomment) { |
| 844 | case NO_COMMENT: |
| 845 | if (strncmp(cp, "/\\\r\n", 4) == 0) { |
| 846 | incomment = STARTING_COMMENT; |
| 847 | cp += 4; |
| 848 | } else if (strncmp(cp, "/\\\n", 3) == 0) { |
| 849 | incomment = STARTING_COMMENT; |
| 850 | cp += 3; |
| 851 | } else if (strncmp(cp, "/*", 2) == 0) { |
| 852 | incomment = C_COMMENT; |
| 853 | cp += 2; |
| 854 | } else if (strncmp(cp, "//", 2) == 0) { |
| 855 | incomment = CXX_COMMENT; |
| 856 | cp += 2; |
| 857 | } else if (strncmp(cp, "\'", 1) == 0) { |
| 858 | incomment = CHAR_LITERAL; |
| 859 | linestate = LS_DIRTY; |
| 860 | cp += 1; |
| 861 | } else if (strncmp(cp, "\"", 1) == 0) { |
| 862 | incomment = STRING_LITERAL; |
| 863 | linestate = LS_DIRTY; |
| 864 | cp += 1; |
| 865 | } else if (strncmp(cp, "\n", 1) == 0) { |
| 866 | linestate = LS_START; |
| 867 | cp += 1; |
| 868 | } else if (strchr(" \r\t", *cp) != NULL) { |
| 869 | cp += 1; |
| 870 | } else |
| 871 | return (cp); |
| 872 | continue; |
| 873 | case CXX_COMMENT: |
| 874 | if (strncmp(cp, "\n", 1) == 0) { |
| 875 | incomment = NO_COMMENT; |
| 876 | linestate = LS_START; |
| 877 | } |
| 878 | cp += 1; |
| 879 | continue; |
| 880 | case CHAR_LITERAL: |
| 881 | case STRING_LITERAL: |
| 882 | if ((incomment == CHAR_LITERAL && cp[0] == '\'') || |
| 883 | (incomment == STRING_LITERAL && cp[0] == '\"')) { |
| 884 | incomment = NO_COMMENT; |
| 885 | cp += 1; |
| 886 | } else if (cp[0] == '\\') { |
| 887 | if (cp[1] == '\0') |
| 888 | cp += 1; |
| 889 | else |
| 890 | cp += 2; |
| 891 | } else if (strncmp(cp, "\n", 1) == 0) { |
| 892 | if (incomment == CHAR_LITERAL) |
| 893 | error("unterminated char literal"); |
| 894 | else |
| 895 | error("unterminated string literal"); |
| 896 | } else |
| 897 | cp += 1; |
| 898 | continue; |
| 899 | case C_COMMENT: |
| 900 | if (strncmp(cp, "*\\\r\n", 4) == 0) { |
| 901 | incomment = FINISHING_COMMENT; |
| 902 | cp += 4; |
| 903 | } else if (strncmp(cp, "*\\\n", 3) == 0) { |
| 904 | incomment = FINISHING_COMMENT; |
| 905 | cp += 3; |
| 906 | } else if (strncmp(cp, "*/", 2) == 0) { |
| 907 | incomment = NO_COMMENT; |
| 908 | cp += 2; |
| 909 | } else |
| 910 | cp += 1; |
| 911 | continue; |
| 912 | case STARTING_COMMENT: |
| 913 | if (*cp == '*') { |
| 914 | incomment = C_COMMENT; |
| 915 | cp += 1; |
| 916 | } else if (*cp == '/') { |
| 917 | incomment = CXX_COMMENT; |
| 918 | cp += 1; |
| 919 | } else { |
| 920 | incomment = NO_COMMENT; |
| 921 | linestate = LS_DIRTY; |
| 922 | } |
| 923 | continue; |
| 924 | case FINISHING_COMMENT: |
| 925 | if (*cp == '/') { |
| 926 | incomment = NO_COMMENT; |
| 927 | cp += 1; |
| 928 | } else |
| 929 | incomment = C_COMMENT; |
| 930 | continue; |
| 931 | default: |
| 932 | abort(); |
| 933 | } |
| 934 | return (cp); |
| 935 | } |
| 936 | |
| 937 | static const char * |
| 938 | skipargs(const char *cp) |
| 939 | { |
| 940 | const char *ocp = cp; |
| 941 | int level = 0; |
| 942 | cp = skipcomment(cp); |
| 943 | if (*cp != '(') |
| 944 | return (cp); |
| 945 | do { |
| 946 | if (*cp == '(') |
| 947 | level++; |
| 948 | if (*cp == ')') |
| 949 | level--; |
| 950 | cp = skipcomment(cp+1); |
| 951 | } while (level != 0 && *cp != '\0'); |
| 952 | if (level == 0) |
| 953 | return (cp); |
| 954 | else |
| 955 | |
| 956 | return (ocp); |
| 957 | } |
| 958 | |
| 959 | static const char * |
| 960 | skipsym(const char *cp) |
| 961 | { |
| 962 | while (!endsym(*cp)) |
| 963 | ++cp; |
| 964 | return (cp); |
| 965 | } |
| 966 | |
| 967 | static int |
| 968 | findsym(const char *str) |
| 969 | { |
| 970 | const char *cp; |
| 971 | int symind; |
| 972 | |
| 973 | cp = skipsym(str); |
| 974 | if (cp == str) |
| 975 | return (-1); |
| 976 | if (symlist) { |
| 977 | if (symdepth && firstsym) |
| 978 | printf("%s%3d", zerosyms ? "" : "\n", depth); |
| 979 | firstsym = zerosyms = false; |
| 980 | printf("%s%.*s%s", |
| 981 | symdepth ? " " : "", |
| 982 | (int)(cp-str), str, |
| 983 | symdepth ? "" : "\n"); |
| 984 | |
| 985 | return (0); |
| 986 | } |
| 987 | for (symind = 0; symind < nsyms; ++symind) { |
| 988 | if (strlcmp(symname[symind], str, cp-str) == 0) { |
| 989 | debug("findsym %s %s", symname[symind], |
| 990 | value[symind] ? value[symind] : ""); |
| 991 | return (symind); |
| 992 | } |
| 993 | } |
| 994 | return (-1); |
| 995 | } |
| 996 | |
| 997 | static void |
| 998 | addsym(bool ignorethis, bool definethis, char *sym) |
| 999 | { |
| 1000 | int symind; |
| 1001 | char *val; |
| 1002 | |
| 1003 | symind = findsym(sym); |
| 1004 | if (symind < 0) { |
| 1005 | if (nsyms >= MAXSYMS) |
| 1006 | errx(2, "too many symbols"); |
| 1007 | symind = nsyms++; |
| 1008 | } |
| 1009 | symname[symind] = sym; |
| 1010 | ignore[symind] = ignorethis; |
| 1011 | val = sym + (skipsym(sym) - sym); |
| 1012 | if (definethis) { |
| 1013 | if (*val == '=') { |
| 1014 | value[symind] = val+1; |
| 1015 | *val = '\0'; |
| 1016 | } else if (*val == '\0') |
| 1017 | value[symind] = "1"; |
| 1018 | else |
| 1019 | usage(); |
| 1020 | } else { |
| 1021 | if (*val != '\0') |
| 1022 | usage(); |
| 1023 | value[symind] = NULL; |
| 1024 | } |
| 1025 | debug("addsym %s=%s", symname[symind], |
| 1026 | value[symind] ? value[symind] : "undef"); |
| 1027 | } |
| 1028 | |
| 1029 | static int |
| 1030 | strlcmp(const char *s, const char *t, size_t n) |
| 1031 | { |
| 1032 | while (n-- && *t != '\0') |
| 1033 | if (*s != *t) |
| 1034 | return ((unsigned char)*s - (unsigned char)*t); |
| 1035 | else |
| 1036 | ++s, ++t; |
| 1037 | return ((unsigned char)*s); |
| 1038 | } |
| 1039 | |
| 1040 | static void |
| 1041 | debug(const char *msg, ...) |
| 1042 | { |
| 1043 | va_list ap; |
| 1044 | |
| 1045 | if (debugging) { |
| 1046 | va_start(ap, msg); |
| 1047 | vwarnx(msg, ap); |
| 1048 | va_end(ap); |
| 1049 | } |
| 1050 | } |
| 1051 | |
| 1052 | static void |
| 1053 | error(const char *msg) |
| 1054 | { |
| 1055 | if (depth == 0) |
| 1056 | warnx("%s: %d: %s", filename, linenum, msg); |
| 1057 | else |
| 1058 | warnx("%s: %d: %s (#if line %d depth %d)", |
| 1059 | filename, linenum, msg, stifline[depth], depth); |
| 1060 | closeout(); |
| 1061 | errx(2, "output may be truncated"); |
| 1062 | } |