1 /*
2 * Copyright (c) 2002 - 2014 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/stat.h>
47
48 #include <ctype.h>
49 #include <err.h>
50 #include <stdarg.h>
51 #include <stdbool.h>
52 #include <stdio.h>
53 #include <stdlib.h>
54 #include <string.h>
55 #include <unistd.h>
56
57 static const char copyright[] =
58 #include "version.h"
59 "@(#) $Author: deraadt $\n"
60 "@(#) $URL: http://dotat.at/prog/unifdef $\n"
61 ;
62
63 /* types of input lines: */
64 typedef enum {
65 LT_TRUEI, /* a true #if with ignore flag */
66 LT_FALSEI, /* a false #if with ignore flag */
67 LT_IF, /* an unknown #if */
68 LT_TRUE, /* a true #if */
69 LT_FALSE, /* a false #if */
70 LT_ELIF, /* an unknown #elif */
71 LT_ELTRUE, /* a true #elif */
72 LT_ELFALSE, /* a false #elif */
73 LT_ELSE, /* #else */
74 LT_ENDIF, /* #endif */
75 LT_DODGY, /* flag: directive is not on one line */
76 LT_DODGY_LAST = LT_DODGY + LT_ENDIF,
77 LT_PLAIN, /* ordinary line */
78 LT_EOF, /* end of file */
79 LT_ERROR, /* unevaluable #if */
80 LT_COUNT
81 } Linetype;
82
83 static char const * const linetype_name[] = {
84 "TRUEI", "FALSEI", "IF", "TRUE", "FALSE",
85 "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF",
86 "DODGY TRUEI", "DODGY FALSEI",
87 "DODGY IF", "DODGY TRUE", "DODGY FALSE",
88 "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE",
89 "DODGY ELSE", "DODGY ENDIF",
90 "PLAIN", "EOF", "ERROR"
91 };
92
93 #define linetype_if2elif(lt) ((Linetype)(lt - LT_IF + LT_ELIF))
94 #define linetype_2dodgy(lt) ((Linetype)(lt + LT_DODGY))
95
96 /* state of #if processing */
97 typedef enum {
98 IS_OUTSIDE,
99 IS_FALSE_PREFIX, /* false #if followed by false #elifs */
100 IS_TRUE_PREFIX, /* first non-false #(el)if is true */
101 IS_PASS_MIDDLE, /* first non-false #(el)if is unknown */
102 IS_FALSE_MIDDLE, /* a false #elif after a pass state */
103 IS_TRUE_MIDDLE, /* a true #elif after a pass state */
104 IS_PASS_ELSE, /* an else after a pass state */
105 IS_FALSE_ELSE, /* an else after a true state */
106 IS_TRUE_ELSE, /* an else after only false states */
107 IS_FALSE_TRAILER, /* #elifs after a true are false */
108 IS_COUNT
109 } Ifstate;
110
111 static char const * const ifstate_name[] = {
112 "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX",
113 "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE",
114 "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE",
115 "FALSE_TRAILER"
116 };
117
118 /* state of comment parser */
119 typedef enum {
120 NO_COMMENT = false, /* outside a comment */
121 C_COMMENT, /* in a comment like this one */
122 CXX_COMMENT, /* between // and end of line */
123 STARTING_COMMENT, /* just after slash-backslash-newline */
124 FINISHING_COMMENT, /* star-backslash-newline in a C comment */
125 CHAR_LITERAL, /* inside '' */
126 STRING_LITERAL /* inside "" */
127 } Comment_state;
128
129 static char const * const comment_name[] = {
130 "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING"
131 };
132
133 /* state of preprocessor line parser */
134 typedef enum {
135 LS_START, /* only space and comments on this line */
136 LS_HASH, /* only space, comments, and a hash */
137 LS_DIRTY /* this line can't be a preprocessor line */
138 } Line_state;
139
140 static char const * const linestate_name[] = {
141 "START", "HASH", "DIRTY"
142 };
143
144 /*
145 * Minimum translation limits from ISO/IEC 9899:1999 5.2.4.1
146 */
147 #define MAXDEPTH 64 /* maximum #if nesting */
148 #define MAXLINE 4096 /* maximum length of line */
149 #define MAXSYMS 16384 /* maximum number of symbols */
150
151 /*
152 * Sometimes when editing a keyword the replacement text is longer, so
153 * we leave some space at the end of the tline buffer to accommodate this.
154 */
155 #define EDITSLOP 10
156
157 /*
158 * Globals.
159 */
160
161 static bool compblank; /* -B: compress blank lines */
162 static bool lnblank; /* -b: blank deleted lines */
163 static bool complement; /* -c: do the complement */
164 static bool debugging; /* -d: debugging reports */
165 static bool inplace; /* -m: modify in place */
166 static bool iocccok; /* -e: fewer IOCCC errors */
167 static bool strictlogic; /* -K: keep ambiguous #ifs */
168 static bool killconsts; /* -k: eval constant #ifs */
169 static bool lnnum; /* -n: add #line directives */
170 static bool symlist; /* -s: output symbol list */
171 static bool symdepth; /* -S: output symbol depth */
172 static bool text; /* -t: this is a text file */
173
174 static const char *symname[MAXSYMS]; /* symbol name */
175 static const char *value[MAXSYMS]; /* -Dsym=value */
176 static bool ignore[MAXSYMS]; /* -iDsym or -iUsym */
177 static int nsyms; /* number of symbols */
178
179 static FILE *input; /* input file pointer */
180 static const char *filename; /* input file name */
181 static int linenum; /* current line number */
182 static const char *linefile; /* file name for #line */
183 static FILE *output; /* output file pointer */
184 static const char *ofilename; /* output file name */
185 static const char *backext; /* backup extension */
186 static char *tempname; /* avoid splatting input */
187
188 static char tline[MAXLINE+EDITSLOP];/* input buffer plus space */
189 static char *keyword; /* used for editing #elif's */
190
191 /*
192 * When processing a file, the output's newline style will match the
193 * input's, and unifdef correctly handles CRLF or LF endings whatever
194 * the platform's native style. The stdio streams are opened in binary
195 * mode to accommodate platforms whose native newline style is CRLF.
196 * When the output isn't a processed input file (when it is error /
197 * debug / diagnostic messages) then unifdef uses native line endings.
198 */
199
200 static const char *newline; /* input file format */
201 static const char newline_unix[] = "\n";
202 static const char newline_crlf[] = "\r\n";
203
204 static Comment_state incomment; /* comment parser state */
205 static Line_state linestate; /* #if line parser state */
206 static Ifstate ifstate[MAXDEPTH]; /* #if processor state */
207 static bool ignoring[MAXDEPTH]; /* ignore comments state */
208 static int stifline[MAXDEPTH]; /* start of current #if */
209 static int depth; /* current #if nesting */
210 static int delcount; /* count of deleted lines */
211 static unsigned blankcount; /* count of blank lines */
212 static unsigned blankmax; /* maximum recent blankcount */
213 static bool constexpr; /* constant #if expression */
214 static bool zerosyms; /* to format symdepth output */
215 static bool firstsym; /* ditto */
216
217 static int exitmode; /* exit status mode */
218 static int exitstat; /* program exit status */
219
220 static void addsym1(bool, bool, char *);
221 static void addsym2(bool, const char *, const char *);
222 static char *astrcat(const char *, const char *);
223 static void cleantemp(void);
224 static void closeio(void);
225 static void debug(const char *, ...);
226 static void debugsym(const char *, int);
227 static bool defundef(void);
228 static void defundefile(const char *);
229 static void done(void);
230 static void error(const char *);
231 static int findsym(const char **);
232 static void flushline(bool);
233 static void hashline(void);
234 static void help(void);
235 static Linetype ifeval(const char **);
236 static void ignoreoff(void);
237 static void ignoreon(void);
238 static void indirectsym(void);
239 static void keywordedit(const char *);
240 static const char *matchsym(const char *, const char *);
241 static void nest(void);
242 static Linetype parseline(void);
243 static void process(void);
244 static void processinout(const char *, const char *);
245 static const char *skipargs(const char *);
246 static const char *skipcomment(const char *);
247 static const char *skiphash(void);
248 static const char *skipline(const char *);
249 static const char *skipsym(const char *);
250 static void state(Ifstate);
251 static void unnest(void);
252 static void usage(void);
253 static void version(void);
254 static const char *xstrdup(const char *, const char *);
255 static FILE * mktempmode(char *tmp, int mode);
256
257 #define endsym(c) (!isalnum((unsigned char)c) && c != '_')
258
259 /*
260 * The main program.
261 */
262 int
main(int argc,char * argv[])263 main(int argc, char *argv[])
264 {
265 const char *errstr;
266 int opt;
267
268 if (pledge("stdio rpath wpath cpath fattr", NULL) == -1)
269 err(1, "pledge");
270
271 while ((opt = getopt(argc, argv, "i:D:U:f:I:M:o:x:bBcdehKklmnsStV")) != -1)
272 switch (opt) {
273 case 'i': /* treat stuff controlled by these symbols as text */
274 /*
275 * For strict backwards-compatibility the U or D
276 * should be immediately after the -i but it doesn't
277 * matter much if we relax that requirement.
278 */
279 opt = *optarg++;
280 if (opt == 'D')
281 addsym1(true, true, optarg);
282 else if (opt == 'U')
283 addsym1(true, false, optarg);
284 else
285 usage();
286 break;
287 case 'D': /* define a symbol */
288 addsym1(false, true, optarg);
289 break;
290 case 'U': /* undef a symbol */
291 addsym1(false, false, optarg);
292 break;
293 case 'I': /* no-op for compatibility with cpp */
294 break;
295 case 'b': /* blank deleted lines instead of omitting them */
296 case 'l': /* backwards compatibility */
297 lnblank = true;
298 break;
299 case 'B': /* compress blank lines around removed section */
300 compblank = true;
301 break;
302 case 'c': /* treat -D as -U and vice versa */
303 complement = true;
304 break;
305 case 'd':
306 debugging = true;
307 break;
308 case 'e': /* fewer errors from dodgy lines */
309 iocccok = true;
310 break;
311 case 'f': /* definitions file */
312 defundefile(optarg);
313 break;
314 case 'h':
315 help();
316 break;
317 case 'K': /* keep ambiguous #ifs */
318 strictlogic = true;
319 break;
320 case 'k': /* process constant #ifs */
321 killconsts = true;
322 break;
323 case 'm': /* modify in place */
324 inplace = true;
325 break;
326 case 'M': /* modify in place and keep backup */
327 inplace = true;
328 backext = optarg;
329 break;
330 case 'n': /* add #line directive after deleted lines */
331 lnnum = true;
332 break;
333 case 'o': /* output to a file */
334 ofilename = optarg;
335 break;
336 case 's': /* only output list of symbols that control #ifs */
337 symlist = true;
338 break;
339 case 'S': /* list symbols with their nesting depth */
340 symlist = symdepth = true;
341 break;
342 case 't': /* don't parse C comments */
343 text = true;
344 break;
345 case 'V':
346 version();
347 break;
348 case 'x':
349 exitmode = strtonum(optarg, 0, 2, &errstr);
350 if (errstr)
351 errx(1, "-x %s: %s", optarg, errstr);
352 break;
353 default:
354 usage();
355 }
356 argc -= optind;
357 argv += optind;
358 if (compblank && lnblank)
359 errx(2, "-B and -b are mutually exclusive");
360 if (symlist && (ofilename != NULL || inplace || argc > 1))
361 errx(2, "-s only works with one input file");
362 if (argc > 1 && ofilename != NULL)
363 errx(2, "-o cannot be used with multiple input files");
364 if (argc > 1 && !inplace)
365 errx(2, "multiple input files require -m or -M");
366 if (argc == 0 && inplace)
367 errx(2, "can't edit stdin in place");
368 if (argc == 0)
369 argc = 1;
370 if (argc == 1 && !inplace && ofilename == NULL)
371 ofilename = "-";
372 indirectsym();
373
374 atexit(cleantemp);
375 if (ofilename != NULL)
376 processinout(*argv, ofilename);
377 else while (argc-- > 0) {
378 processinout(*argv, *argv);
379 argv++;
380 }
381 switch(exitmode) {
382 case(0): exit(exitstat);
383 case(1): exit(!exitstat);
384 case(2): exit(0);
385 default: abort(); /* bug */
386 }
387 }
388
389 /*
390 * File logistics.
391 */
392 static void
processinout(const char * ifn,const char * ofn)393 processinout(const char *ifn, const char *ofn)
394 {
395 struct stat st;
396
397 if (ifn == NULL || strcmp(ifn, "-") == 0) {
398 filename = "[stdin]";
399 linefile = NULL;
400 input = stdin;
401 } else {
402 filename = ifn;
403 linefile = ifn;
404 input = fopen(ifn, "rb");
405 if (input == NULL)
406 err(2, "can't open %s", ifn);
407 }
408 if (strcmp(ofn, "-") == 0) {
409 output = stdout;
410 process();
411 return;
412 }
413 if (stat(ofn, &st) == -1) {
414 output = fopen(ofn, "wb");
415 if (output == NULL)
416 err(2, "can't create %s", ofn);
417 process();
418 return;
419 }
420
421 tempname = astrcat(ofn, ".XXXXXX");
422 output = mktempmode(tempname, st.st_mode);
423 if (output == NULL)
424 err(2, "can't create %s", tempname);
425
426 process();
427
428 if (backext != NULL && *backext != '\0') {
429 char *backname = astrcat(ofn, backext);
430 if (rename(ofn, backname) == -1)
431 err(2, "can't rename \"%s\" to \"%s\"", ofn, backname);
432 free(backname);
433 }
434 if (rename(tempname, ofn) == -1)
435 err(2, "can't rename \"%s\" to \"%s\"", tempname, ofn);
436 free(tempname);
437 tempname = NULL;
438 }
439
440 /*
441 * For cleaning up if there is an error.
442 */
443 static void
cleantemp(void)444 cleantemp(void)
445 {
446 if (tempname != NULL)
447 remove(tempname);
448 }
449
450 /*
451 * Self-identification functions.
452 */
453
454 static void
version(void)455 version(void)
456 {
457 const char *c = copyright;
458 for (;;) {
459 while (*++c != '$')
460 if (*c == '\0')
461 exit(0);
462 while (*++c != '$')
463 putc(*c, stderr);
464 putc('\n', stderr);
465 }
466 }
467
468 static void
synopsis(FILE * fp)469 synopsis(FILE *fp)
470 {
471 fprintf(fp,
472 "usage: unifdef [-BbcdehKkmnSstV] [-[i]Dsym[=val]] [-[i]Usym] [-f defile]\n"
473 " [-M backext] [-o outfile] [-x 0 | 1 | 2] file ...\n");
474 }
475
476 static void
usage(void)477 usage(void)
478 {
479 synopsis(stderr);
480 exit(2);
481 }
482
483 static void
help(void)484 help(void)
485 {
486 synopsis(stdout);
487 printf(
488 " -Dsym=val define preprocessor symbol with given value\n"
489 " -Dsym define preprocessor symbol with value 1\n"
490 " -Usym preprocessor symbol is undefined\n"
491 " -iDsym=val \\ ignore C strings and comments\n"
492 " -iDsym ) in sections controlled by these\n"
493 " -iUsym / preprocessor symbols\n"
494 " -fpath file containing #define and #undef directives\n"
495 " -b blank lines instead of deleting them\n"
496 " -B compress blank lines around deleted section\n"
497 " -c complement (invert) keep vs. delete\n"
498 " -d debugging mode\n"
499 " -e ignore multiline preprocessor directives\n"
500 " -h print help\n"
501 " -Ipath extra include file path (ignored)\n"
502 " -K disable && and || short-circuiting\n"
503 " -k process constant #if expressions\n"
504 " -Mext modify in place and keep backups\n"
505 " -m modify input files in place\n"
506 " -n add #line directives to output\n"
507 " -opath output file name\n"
508 " -S list #if control symbols with nesting\n"
509 " -s list #if control symbols\n"
510 " -t ignore C strings and comments\n"
511 " -V print version\n"
512 " -x{012} exit status mode\n"
513 );
514 exit(0);
515 }
516
517 /*
518 * A state transition function alters the global #if processing state
519 * in a particular way. The table below is indexed by the current
520 * processing state and the type of the current line.
521 *
522 * Nesting is handled by keeping a stack of states; some transition
523 * functions increase or decrease the depth. They also maintain the
524 * ignore state on a stack. In some complicated cases they have to
525 * alter the preprocessor directive, as follows.
526 *
527 * When we have processed a group that starts off with a known-false
528 * #if/#elif sequence (which has therefore been deleted) followed by a
529 * #elif that we don't understand and therefore must keep, we edit the
530 * latter into a #if to keep the nesting correct. We use memcpy() to
531 * overwrite the 4 byte token "elif" with "if " without a '\0' byte.
532 *
533 * When we find a true #elif in a group, the following block will
534 * always be kept and the rest of the sequence after the next #elif or
535 * #else will be discarded. We edit the #elif into a #else and the
536 * following directive to #endif since this has the desired behaviour.
537 *
538 * "Dodgy" directives are split across multiple lines, the most common
539 * example being a multi-line comment hanging off the right of the
540 * directive. We can handle them correctly only if there is no change
541 * from printing to dropping (or vice versa) caused by that directive.
542 * If the directive is the first of a group we have a choice between
543 * failing with an error, or passing it through unchanged instead of
544 * evaluating it. The latter is not the default to avoid questions from
545 * users about unifdef unexpectedly leaving behind preprocessor directives.
546 */
547 typedef void state_fn(void);
548
549 /* report an error */
Eelif(void)550 static void Eelif (void) { error("Inappropriate #elif"); }
Eelse(void)551 static void Eelse (void) { error("Inappropriate #else"); }
Eendif(void)552 static void Eendif(void) { error("Inappropriate #endif"); }
Eeof(void)553 static void Eeof (void) { error("Premature EOF"); }
Eioccc(void)554 static void Eioccc(void) { error("Obfuscated preprocessor control line"); }
555 /* plain line handling */
print(void)556 static void print (void) { flushline(true); }
drop(void)557 static void drop (void) { flushline(false); }
558 /* output lacks group's start line */
Strue(void)559 static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); }
Sfalse(void)560 static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); }
Selse(void)561 static void Selse (void) { drop(); state(IS_TRUE_ELSE); }
562 /* print/pass this block */
Pelif(void)563 static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); }
Pelse(void)564 static void Pelse (void) { print(); state(IS_PASS_ELSE); }
Pendif(void)565 static void Pendif(void) { print(); unnest(); }
566 /* discard this block */
Dfalse(void)567 static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); }
Delif(void)568 static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); }
Delse(void)569 static void Delse (void) { drop(); state(IS_FALSE_ELSE); }
Dendif(void)570 static void Dendif(void) { drop(); unnest(); }
571 /* first line of group */
Fdrop(void)572 static void Fdrop (void) { nest(); Dfalse(); }
Fpass(void)573 static void Fpass (void) { nest(); Pelif(); }
Ftrue(void)574 static void Ftrue (void) { nest(); Strue(); }
Ffalse(void)575 static void Ffalse(void) { nest(); Sfalse(); }
576 /* variable pedantry for obfuscated lines */
Oiffy(void)577 static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); }
Oif(void)578 static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); }
Oelif(void)579 static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); }
580 /* ignore comments in this block */
Idrop(void)581 static void Idrop (void) { Fdrop(); ignoreon(); }
Itrue(void)582 static void Itrue (void) { Ftrue(); ignoreon(); }
Ifalse(void)583 static void Ifalse(void) { Ffalse(); ignoreon(); }
584 /* modify this line */
Mpass(void)585 static void Mpass (void) { memcpy(keyword, "if ", 4); Pelif(); }
Mtrue(void)586 static void Mtrue (void) { keywordedit("else"); state(IS_TRUE_MIDDLE); }
Melif(void)587 static void Melif (void) { keywordedit("endif"); state(IS_FALSE_TRAILER); }
Melse(void)588 static void Melse (void) { keywordedit("endif"); state(IS_FALSE_ELSE); }
589
590 static state_fn * const trans_table[IS_COUNT][LT_COUNT] = {
591 /* IS_OUTSIDE */
592 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif,
593 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eendif,
594 print, done, abort },
595 /* IS_FALSE_PREFIX */
596 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif,
597 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc,
598 drop, Eeof, abort },
599 /* IS_TRUE_PREFIX */
600 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif,
601 Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
602 print, Eeof, abort },
603 /* IS_PASS_MIDDLE */
604 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif,
605 Oiffy, Oiffy, Fpass, Oif, Oif, Pelif, Oelif, Oelif, Pelse, Pendif,
606 print, Eeof, abort },
607 /* IS_FALSE_MIDDLE */
608 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif,
609 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
610 drop, Eeof, abort },
611 /* IS_TRUE_MIDDLE */
612 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif,
613 Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Pendif,
614 print, Eeof, abort },
615 /* IS_PASS_ELSE */
616 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif,
617 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Pendif,
618 print, Eeof, abort },
619 /* IS_FALSE_ELSE */
620 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif,
621 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc,
622 drop, Eeof, abort },
623 /* IS_TRUE_ELSE */
624 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif,
625 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eioccc,
626 print, Eeof, abort },
627 /* IS_FALSE_TRAILER */
628 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif,
629 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc,
630 drop, Eeof, abort }
631 /*TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF
632 TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF (DODGY)
633 PLAIN EOF ERROR */
634 };
635
636 /*
637 * State machine utility functions
638 */
639 static void
ignoreoff(void)640 ignoreoff(void)
641 {
642 if (depth == 0)
643 abort(); /* bug */
644 ignoring[depth] = ignoring[depth-1];
645 }
646 static void
ignoreon(void)647 ignoreon(void)
648 {
649 ignoring[depth] = true;
650 }
651 static void
keywordedit(const char * replacement)652 keywordedit(const char *replacement)
653 {
654 snprintf(keyword, tline + sizeof(tline) - keyword,
655 "%s%s", replacement, newline);
656 print();
657 }
658 static void
nest(void)659 nest(void)
660 {
661 if (depth > MAXDEPTH-1)
662 abort(); /* bug */
663 if (depth == MAXDEPTH-1)
664 error("Too many levels of nesting");
665 depth += 1;
666 stifline[depth] = linenum;
667 }
668 static void
unnest(void)669 unnest(void)
670 {
671 if (depth == 0)
672 abort(); /* bug */
673 depth -= 1;
674 }
675 static void
state(Ifstate is)676 state(Ifstate is)
677 {
678 ifstate[depth] = is;
679 }
680
681 /*
682 * The last state transition function. When this is called,
683 * lineval == LT_EOF, so the process() loop will terminate.
684 */
685 static void
done(void)686 done(void)
687 {
688 if (incomment)
689 error("EOF in comment");
690 closeio();
691 }
692
693 /*
694 * Write a line to the output or not, according to command line options.
695 * If writing fails, closeio() will print the error and exit.
696 */
697 static void
flushline(bool keep)698 flushline(bool keep)
699 {
700 if (symlist)
701 return;
702 if (keep ^ complement) {
703 bool blankline = tline[strspn(tline, " \t\r\n")] == '\0';
704 if (blankline && compblank && blankcount != blankmax) {
705 delcount += 1;
706 blankcount += 1;
707 } else {
708 if (lnnum && delcount > 0)
709 hashline();
710 if (fputs(tline, output) == EOF)
711 closeio();
712 delcount = 0;
713 blankmax = blankcount = blankline ? blankcount + 1 : 0;
714 }
715 } else {
716 if (lnblank && fputs(newline, output) == EOF)
717 closeio();
718 exitstat = 1;
719 delcount += 1;
720 blankcount = 0;
721 }
722 if (debugging && fflush(output) == EOF)
723 closeio();
724 }
725
726 /*
727 * Format of #line directives depends on whether we know the input filename.
728 */
729 static void
hashline(void)730 hashline(void)
731 {
732 int e;
733
734 if (linefile == NULL)
735 e = fprintf(output, "#line %d%s", linenum, newline);
736 else
737 e = fprintf(output, "#line %d \"%s\"%s",
738 linenum, linefile, newline);
739 if (e < 0)
740 closeio();
741 }
742
743 /*
744 * Flush the output and handle errors.
745 */
746 static void
closeio(void)747 closeio(void)
748 {
749 /* Tidy up after findsym(). */
750 if (symdepth && !zerosyms)
751 printf("\n");
752 if (output != NULL && (ferror(output) || fclose(output) == EOF))
753 err(2, "%s: can't write to output", filename);
754 fclose(input);
755 }
756
757 /*
758 * The driver for the state machine.
759 */
760 static void
process(void)761 process(void)
762 {
763 Linetype lineval = LT_PLAIN;
764 /* When compressing blank lines, act as if the file
765 is preceded by a large number of blank lines. */
766 blankmax = blankcount = 1000;
767 zerosyms = true;
768 newline = NULL;
769 linenum = 0;
770 while (lineval != LT_EOF) {
771 lineval = parseline();
772 trans_table[ifstate[depth]][lineval]();
773 debug("process line %d %s -> %s depth %d",
774 linenum, linetype_name[lineval],
775 ifstate_name[ifstate[depth]], depth);
776 }
777 }
778
779 /*
780 * Parse a line and determine its type. We keep the preprocessor line
781 * parser state between calls in the global variable linestate, with
782 * help from skipcomment().
783 */
784 static Linetype
parseline(void)785 parseline(void)
786 {
787 const char *cp;
788 int cursym;
789 Linetype retval;
790 Comment_state wascomment;
791
792 wascomment = incomment;
793 cp = skiphash();
794 if (cp == NULL)
795 return (LT_EOF);
796 if (newline == NULL) {
797 if (strrchr(tline, '\n') == strrchr(tline, '\r') + 1)
798 newline = newline_crlf;
799 else
800 newline = newline_unix;
801 }
802 if (*cp == '\0') {
803 retval = LT_PLAIN;
804 goto done;
805 }
806 keyword = tline + (cp - tline);
807 if ((cp = matchsym("ifdef", keyword)) != NULL ||
808 (cp = matchsym("ifndef", keyword)) != NULL) {
809 cp = skipcomment(cp);
810 if ((cursym = findsym(&cp)) < 0)
811 retval = LT_IF;
812 else {
813 retval = (keyword[2] == 'n')
814 ? LT_FALSE : LT_TRUE;
815 if (value[cursym] == NULL)
816 retval = (retval == LT_TRUE)
817 ? LT_FALSE : LT_TRUE;
818 if (ignore[cursym])
819 retval = (retval == LT_TRUE)
820 ? LT_TRUEI : LT_FALSEI;
821 }
822 } else if ((cp = matchsym("if", keyword)) != NULL)
823 retval = ifeval(&cp);
824 else if ((cp = matchsym("elif", keyword)) != NULL)
825 retval = linetype_if2elif(ifeval(&cp));
826 else if ((cp = matchsym("else", keyword)) != NULL)
827 retval = LT_ELSE;
828 else if ((cp = matchsym("endif", keyword)) != NULL)
829 retval = LT_ENDIF;
830 else {
831 cp = skipsym(keyword);
832 /* no way can we deal with a continuation inside a keyword */
833 if (strncmp(cp, "\\\r\n", 3) == 0 ||
834 strncmp(cp, "\\\n", 2) == 0)
835 Eioccc();
836 cp = skipline(cp);
837 retval = LT_PLAIN;
838 goto done;
839 }
840 cp = skipcomment(cp);
841 if (*cp != '\0') {
842 cp = skipline(cp);
843 if (retval == LT_TRUE || retval == LT_FALSE ||
844 retval == LT_TRUEI || retval == LT_FALSEI)
845 retval = LT_IF;
846 if (retval == LT_ELTRUE || retval == LT_ELFALSE)
847 retval = LT_ELIF;
848 }
849 /* the following can happen if the last line of the file lacks a
850 newline or if there is too much whitespace in a directive */
851 if (linestate == LS_HASH) {
852 long len = cp - tline;
853 if (fgets(tline + len, MAXLINE - len, input) == NULL) {
854 if (ferror(input))
855 err(2, "can't read %s", filename);
856 /* append the missing newline at eof */
857 strlcpy(tline + len, newline, sizeof(tline) - len);
858 cp += strlen(newline);
859 linestate = LS_START;
860 } else {
861 linestate = LS_DIRTY;
862 }
863 }
864 if (retval != LT_PLAIN && (wascomment || linestate != LS_START)) {
865 retval = linetype_2dodgy(retval);
866 linestate = LS_DIRTY;
867 }
868 done:
869 debug("parser line %d state %s comment %s line", linenum,
870 comment_name[incomment], linestate_name[linestate]);
871 return (retval);
872 }
873
874 /*
875 * These are the binary operators that are supported by the expression
876 * evaluator.
877 */
op_strict(long * p,long v,Linetype at,Linetype bt)878 static Linetype op_strict(long *p, long v, Linetype at, Linetype bt) {
879 if(at == LT_IF || bt == LT_IF) return (LT_IF);
880 return (*p = v, v ? LT_TRUE : LT_FALSE);
881 }
op_lt(long * p,Linetype at,long a,Linetype bt,long b)882 static Linetype op_lt(long *p, Linetype at, long a, Linetype bt, long b) {
883 return op_strict(p, a < b, at, bt);
884 }
op_gt(long * p,Linetype at,long a,Linetype bt,long b)885 static Linetype op_gt(long *p, Linetype at, long a, Linetype bt, long b) {
886 return op_strict(p, a > b, at, bt);
887 }
op_le(long * p,Linetype at,long a,Linetype bt,long b)888 static Linetype op_le(long *p, Linetype at, long a, Linetype bt, long b) {
889 return op_strict(p, a <= b, at, bt);
890 }
op_ge(long * p,Linetype at,long a,Linetype bt,long b)891 static Linetype op_ge(long *p, Linetype at, long a, Linetype bt, long b) {
892 return op_strict(p, a >= b, at, bt);
893 }
op_eq(long * p,Linetype at,long a,Linetype bt,long b)894 static Linetype op_eq(long *p, Linetype at, long a, Linetype bt, long b) {
895 return op_strict(p, a == b, at, bt);
896 }
op_ne(long * p,Linetype at,long a,Linetype bt,long b)897 static Linetype op_ne(long *p, Linetype at, long a, Linetype bt, long b) {
898 return op_strict(p, a != b, at, bt);
899 }
op_or(long * p,Linetype at,long a,Linetype bt,long b)900 static Linetype op_or(long *p, Linetype at, long a, Linetype bt, long b) {
901 if (!strictlogic && (at == LT_TRUE || bt == LT_TRUE))
902 return (*p = 1, LT_TRUE);
903 return op_strict(p, a || b, at, bt);
904 }
op_and(long * p,Linetype at,long a,Linetype bt,long b)905 static Linetype op_and(long *p, Linetype at, long a, Linetype bt, long b) {
906 if (!strictlogic && (at == LT_FALSE || bt == LT_FALSE))
907 return (*p = 0, LT_FALSE);
908 return op_strict(p, a && b, at, bt);
909 }
910
911 /*
912 * An evaluation function takes three arguments, as follows: (1) a pointer to
913 * an element of the precedence table which lists the operators at the current
914 * level of precedence; (2) a pointer to an integer which will receive the
915 * value of the expression; and (3) a pointer to a char* that points to the
916 * expression to be evaluated and that is updated to the end of the expression
917 * when evaluation is complete. The function returns LT_FALSE if the value of
918 * the expression is zero, LT_TRUE if it is non-zero, LT_IF if the expression
919 * depends on an unknown symbol, or LT_ERROR if there is a parse failure.
920 */
921 struct ops;
922
923 typedef Linetype eval_fn(const struct ops *, long *, const char **);
924
925 static eval_fn eval_table, eval_unary;
926
927 /*
928 * The precedence table. Expressions involving binary operators are evaluated
929 * in a table-driven way by eval_table. When it evaluates a subexpression it
930 * calls the inner function with its first argument pointing to the next
931 * element of the table. Innermost expressions have special non-table-driven
932 * handling.
933 */
934 struct op {
935 const char *str;
936 Linetype (*fn)(long *, Linetype, long, Linetype, long);
937 };
938 struct ops {
939 eval_fn *inner;
940 struct op op[5];
941 };
942 static const struct ops eval_ops[] = {
943 { eval_table, { { "||", op_or } } },
944 { eval_table, { { "&&", op_and } } },
945 { eval_table, { { "==", op_eq },
946 { "!=", op_ne } } },
947 { eval_unary, { { "<=", op_le },
948 { ">=", op_ge },
949 { "<", op_lt },
950 { ">", op_gt } } }
951 };
952
953 /* Current operator precedence level */
prec(const struct ops * ops)954 static long prec(const struct ops *ops)
955 {
956 return (ops - eval_ops);
957 }
958
959 /*
960 * Function for evaluating the innermost parts of expressions,
961 * viz. !expr (expr) number defined(symbol) symbol
962 * We reset the constexpr flag in the last two cases.
963 */
964 static Linetype
eval_unary(const struct ops * ops,long * valp,const char ** cpp)965 eval_unary(const struct ops *ops, long *valp, const char **cpp)
966 {
967 const char *cp;
968 char *ep;
969 int sym;
970 bool defparen;
971 Linetype lt;
972
973 cp = skipcomment(*cpp);
974 if (*cp == '!') {
975 debug("eval%d !", prec(ops));
976 cp++;
977 lt = eval_unary(ops, valp, &cp);
978 if (lt == LT_ERROR)
979 return (LT_ERROR);
980 if (lt != LT_IF) {
981 *valp = !*valp;
982 lt = *valp ? LT_TRUE : LT_FALSE;
983 }
984 } else if (*cp == '(') {
985 cp++;
986 debug("eval%d (", prec(ops));
987 lt = eval_table(eval_ops, valp, &cp);
988 if (lt == LT_ERROR)
989 return (LT_ERROR);
990 cp = skipcomment(cp);
991 if (*cp++ != ')')
992 return (LT_ERROR);
993 } else if (isdigit((unsigned char)*cp)) {
994 debug("eval%d number", prec(ops));
995 *valp = strtol(cp, &ep, 0);
996 if (ep == cp)
997 return (LT_ERROR);
998 lt = *valp ? LT_TRUE : LT_FALSE;
999 cp = ep;
1000 } else if (matchsym("defined", cp) != NULL) {
1001 cp = skipcomment(cp+7);
1002 if (*cp == '(') {
1003 cp = skipcomment(cp+1);
1004 defparen = true;
1005 } else {
1006 defparen = false;
1007 }
1008 sym = findsym(&cp);
1009 cp = skipcomment(cp);
1010 if (defparen && *cp++ != ')') {
1011 debug("eval%d defined missing ')'", prec(ops));
1012 return (LT_ERROR);
1013 }
1014 if (sym < 0) {
1015 debug("eval%d defined unknown", prec(ops));
1016 lt = LT_IF;
1017 } else {
1018 debug("eval%d defined %s", prec(ops), symname[sym]);
1019 *valp = (value[sym] != NULL);
1020 lt = *valp ? LT_TRUE : LT_FALSE;
1021 }
1022 constexpr = false;
1023 } else if (!endsym(*cp)) {
1024 debug("eval%d symbol", prec(ops));
1025 sym = findsym(&cp);
1026 if (sym < 0) {
1027 lt = LT_IF;
1028 cp = skipargs(cp);
1029 } else if (value[sym] == NULL) {
1030 *valp = 0;
1031 lt = LT_FALSE;
1032 } else {
1033 *valp = strtol(value[sym], &ep, 0);
1034 if (*ep != '\0' || ep == value[sym])
1035 return (LT_ERROR);
1036 lt = *valp ? LT_TRUE : LT_FALSE;
1037 cp = skipargs(cp);
1038 }
1039 constexpr = false;
1040 } else {
1041 debug("eval%d bad expr", prec(ops));
1042 return (LT_ERROR);
1043 }
1044
1045 *cpp = cp;
1046 debug("eval%d = %d", prec(ops), *valp);
1047 return (lt);
1048 }
1049
1050 /*
1051 * Table-driven evaluation of binary operators.
1052 */
1053 static Linetype
eval_table(const struct ops * ops,long * valp,const char ** cpp)1054 eval_table(const struct ops *ops, long *valp, const char **cpp)
1055 {
1056 const struct op *op;
1057 const char *cp;
1058 long val;
1059 Linetype lt, rt;
1060
1061 debug("eval%d", prec(ops));
1062 cp = *cpp;
1063 lt = ops->inner(ops+1, valp, &cp);
1064 if (lt == LT_ERROR)
1065 return (LT_ERROR);
1066 for (;;) {
1067 cp = skipcomment(cp);
1068 for (op = ops->op; op->str != NULL; op++)
1069 if (strncmp(cp, op->str, strlen(op->str)) == 0)
1070 break;
1071 if (op->str == NULL)
1072 break;
1073 cp += strlen(op->str);
1074 debug("eval%d %s", prec(ops), op->str);
1075 rt = ops->inner(ops+1, &val, &cp);
1076 if (rt == LT_ERROR)
1077 return (LT_ERROR);
1078 lt = op->fn(valp, lt, *valp, rt, val);
1079 }
1080
1081 *cpp = cp;
1082 debug("eval%d = %d", prec(ops), *valp);
1083 debug("eval%d lt = %s", prec(ops), linetype_name[lt]);
1084 return (lt);
1085 }
1086
1087 /*
1088 * Evaluate the expression on a #if or #elif line. If we can work out
1089 * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we
1090 * return just a generic LT_IF.
1091 */
1092 static Linetype
ifeval(const char ** cpp)1093 ifeval(const char **cpp)
1094 {
1095 Linetype ret;
1096 long val = 0;
1097
1098 debug("eval %s", *cpp);
1099 constexpr = killconsts ? false : true;
1100 ret = eval_table(eval_ops, &val, cpp);
1101 debug("eval = %d", val);
1102 return (constexpr ? LT_IF : ret == LT_ERROR ? LT_IF : ret);
1103 }
1104
1105 /*
1106 * Read a line and examine its initial part to determine if it is a
1107 * preprocessor directive. Returns NULL on EOF, or a pointer to a
1108 * preprocessor directive name, or a pointer to the zero byte at the
1109 * end of the line.
1110 */
1111 static const char *
skiphash(void)1112 skiphash(void)
1113 {
1114 const char *cp;
1115
1116 linenum++;
1117 if (fgets(tline, MAXLINE, input) == NULL) {
1118 if (ferror(input))
1119 err(2, "can't read %s", filename);
1120 else
1121 return (NULL);
1122 }
1123 cp = skipcomment(tline);
1124 if (linestate == LS_START && *cp == '#') {
1125 linestate = LS_HASH;
1126 return (skipcomment(cp + 1));
1127 } else if (*cp == '\0') {
1128 return (cp);
1129 } else {
1130 return (skipline(cp));
1131 }
1132 }
1133
1134 /*
1135 * Mark a line dirty and consume the rest of it, keeping track of the
1136 * lexical state.
1137 */
1138 static const char *
skipline(const char * cp)1139 skipline(const char *cp)
1140 {
1141 const char *pcp;
1142 if (*cp != '\0')
1143 linestate = LS_DIRTY;
1144 while (*cp != '\0') {
1145 cp = skipcomment(pcp = cp);
1146 if (pcp == cp)
1147 cp++;
1148 }
1149 return (cp);
1150 }
1151
1152 /*
1153 * Skip over comments, strings, and character literals and stop at the
1154 * next character position that is not whitespace. Between calls we keep
1155 * the comment state in the global variable incomment, and we also adjust
1156 * the global variable linestate when we see a newline.
1157 * XXX: doesn't cope with the buffer splitting inside a state transition.
1158 */
1159 static const char *
skipcomment(const char * cp)1160 skipcomment(const char *cp)
1161 {
1162 if (text || ignoring[depth]) {
1163 for (; isspace((unsigned char)*cp); cp++)
1164 if (*cp == '\n')
1165 linestate = LS_START;
1166 return (cp);
1167 }
1168 while (*cp != '\0')
1169 /* don't reset to LS_START after a line continuation */
1170 if (strncmp(cp, "\\\r\n", 3) == 0)
1171 cp += 3;
1172 else if (strncmp(cp, "\\\n", 2) == 0)
1173 cp += 2;
1174 else switch (incomment) {
1175 case NO_COMMENT:
1176 if (strncmp(cp, "/\\\r\n", 4) == 0) {
1177 incomment = STARTING_COMMENT;
1178 cp += 4;
1179 } else if (strncmp(cp, "/\\\n", 3) == 0) {
1180 incomment = STARTING_COMMENT;
1181 cp += 3;
1182 } else if (strncmp(cp, "/*", 2) == 0) {
1183 incomment = C_COMMENT;
1184 cp += 2;
1185 } else if (strncmp(cp, "//", 2) == 0) {
1186 incomment = CXX_COMMENT;
1187 cp += 2;
1188 } else if (strncmp(cp, "\'", 1) == 0) {
1189 incomment = CHAR_LITERAL;
1190 linestate = LS_DIRTY;
1191 cp += 1;
1192 } else if (strncmp(cp, "\"", 1) == 0) {
1193 incomment = STRING_LITERAL;
1194 linestate = LS_DIRTY;
1195 cp += 1;
1196 } else if (strncmp(cp, "\n", 1) == 0) {
1197 linestate = LS_START;
1198 cp += 1;
1199 } else if (strchr(" \r\t", *cp) != NULL) {
1200 cp += 1;
1201 } else
1202 return (cp);
1203 continue;
1204 case CXX_COMMENT:
1205 if (strncmp(cp, "\n", 1) == 0) {
1206 incomment = NO_COMMENT;
1207 linestate = LS_START;
1208 }
1209 cp += 1;
1210 continue;
1211 case CHAR_LITERAL:
1212 case STRING_LITERAL:
1213 if ((incomment == CHAR_LITERAL && cp[0] == '\'') ||
1214 (incomment == STRING_LITERAL && cp[0] == '\"')) {
1215 incomment = NO_COMMENT;
1216 cp += 1;
1217 } else if (cp[0] == '\\') {
1218 if (cp[1] == '\0')
1219 cp += 1;
1220 else
1221 cp += 2;
1222 } else if (strncmp(cp, "\n", 1) == 0) {
1223 if (incomment == CHAR_LITERAL)
1224 error("Unterminated char literal");
1225 else
1226 error("Unterminated string literal");
1227 } else
1228 cp += 1;
1229 continue;
1230 case C_COMMENT:
1231 if (strncmp(cp, "*\\\r\n", 4) == 0) {
1232 incomment = FINISHING_COMMENT;
1233 cp += 4;
1234 } else if (strncmp(cp, "*\\\n", 3) == 0) {
1235 incomment = FINISHING_COMMENT;
1236 cp += 3;
1237 } else if (strncmp(cp, "*/", 2) == 0) {
1238 incomment = NO_COMMENT;
1239 cp += 2;
1240 } else
1241 cp += 1;
1242 continue;
1243 case STARTING_COMMENT:
1244 if (*cp == '*') {
1245 incomment = C_COMMENT;
1246 cp += 1;
1247 } else if (*cp == '/') {
1248 incomment = CXX_COMMENT;
1249 cp += 1;
1250 } else {
1251 incomment = NO_COMMENT;
1252 linestate = LS_DIRTY;
1253 }
1254 continue;
1255 case FINISHING_COMMENT:
1256 if (*cp == '/') {
1257 incomment = NO_COMMENT;
1258 cp += 1;
1259 } else
1260 incomment = C_COMMENT;
1261 continue;
1262 default:
1263 abort(); /* bug */
1264 }
1265 return (cp);
1266 }
1267
1268 /*
1269 * Skip macro arguments.
1270 */
1271 static const char *
skipargs(const char * cp)1272 skipargs(const char *cp)
1273 {
1274 const char *ocp = cp;
1275 int level = 0;
1276 cp = skipcomment(cp);
1277 if (*cp != '(')
1278 return (cp);
1279 do {
1280 if (*cp == '(')
1281 level++;
1282 if (*cp == ')')
1283 level--;
1284 cp = skipcomment(cp+1);
1285 } while (level != 0 && *cp != '\0');
1286 if (level == 0)
1287 return (cp);
1288 else
1289 /* Rewind and re-detect the syntax error later. */
1290 return (ocp);
1291 }
1292
1293 /*
1294 * Skip over an identifier.
1295 */
1296 static const char *
skipsym(const char * cp)1297 skipsym(const char *cp)
1298 {
1299 while (!endsym(*cp))
1300 ++cp;
1301 return (cp);
1302 }
1303
1304 /*
1305 * Skip whitespace and take a copy of any following identifier.
1306 */
1307 static const char *
getsym(const char ** cpp)1308 getsym(const char **cpp)
1309 {
1310 const char *cp = *cpp, *sym;
1311
1312 cp = skipcomment(cp);
1313 cp = skipsym(sym = cp);
1314 if (cp == sym)
1315 return NULL;
1316 *cpp = cp;
1317 return (xstrdup(sym, cp));
1318 }
1319
1320 /*
1321 * Check that s (a symbol) matches the start of t, and that the
1322 * following character in t is not a symbol character. Returns a
1323 * pointer to the following character in t if there is a match,
1324 * otherwise NULL.
1325 */
1326 static const char *
matchsym(const char * s,const char * t)1327 matchsym(const char *s, const char *t)
1328 {
1329 while (*s != '\0' && *t != '\0')
1330 if (*s != *t)
1331 return (NULL);
1332 else
1333 ++s, ++t;
1334 if (*s == '\0' && endsym(*t))
1335 return(t);
1336 else
1337 return(NULL);
1338 }
1339
1340 /*
1341 * Look for the symbol in the symbol table. If it is found, we return
1342 * the symbol table index, else we return -1.
1343 */
1344 static int
findsym(const char ** strp)1345 findsym(const char **strp)
1346 {
1347 const char *str;
1348 int symind;
1349
1350 str = *strp;
1351 *strp = skipsym(str);
1352 if (symlist) {
1353 if (*strp == str)
1354 return (-1);
1355 if (symdepth && firstsym)
1356 printf("%s%3d", zerosyms ? "" : "\n", depth);
1357 firstsym = zerosyms = false;
1358 printf("%s%.*s%s",
1359 symdepth ? " " : "",
1360 (int)(*strp-str), str,
1361 symdepth ? "" : "\n");
1362 /* we don't care about the value of the symbol */
1363 return (0);
1364 }
1365 for (symind = 0; symind < nsyms; ++symind) {
1366 if (matchsym(symname[symind], str) != NULL) {
1367 debugsym("findsym", symind);
1368 return (symind);
1369 }
1370 }
1371 return (-1);
1372 }
1373
1374 /*
1375 * Resolve indirect symbol values to their final definitions.
1376 */
1377 static void
indirectsym(void)1378 indirectsym(void)
1379 {
1380 const char *cp;
1381 int changed, sym, ind;
1382
1383 do {
1384 changed = 0;
1385 for (sym = 0; sym < nsyms; ++sym) {
1386 if (value[sym] == NULL)
1387 continue;
1388 cp = value[sym];
1389 ind = findsym(&cp);
1390 if (ind == -1 || ind == sym ||
1391 *cp != '\0' ||
1392 value[ind] == NULL ||
1393 value[ind] == value[sym])
1394 continue;
1395 debugsym("indir...", sym);
1396 value[sym] = value[ind];
1397 debugsym("...ectsym", sym);
1398 changed++;
1399 }
1400 } while (changed);
1401 }
1402
1403 /*
1404 * Add a symbol to the symbol table, specified with the format sym=val
1405 */
1406 static void
addsym1(bool ignorethis,bool definethis,char * symval)1407 addsym1(bool ignorethis, bool definethis, char *symval)
1408 {
1409 const char *sym, *val;
1410
1411 sym = symval;
1412 val = skipsym(sym);
1413 if (definethis && *val == '=') {
1414 symval[val - sym] = '\0';
1415 val = val + 1;
1416 } else if (*val == '\0') {
1417 val = definethis ? "1" : NULL;
1418 } else {
1419 usage();
1420 }
1421 addsym2(ignorethis, sym, val);
1422 }
1423
1424 /*
1425 * Add a symbol to the symbol table.
1426 */
1427 static void
addsym2(bool ignorethis,const char * sym,const char * val)1428 addsym2(bool ignorethis, const char *sym, const char *val)
1429 {
1430 const char *cp = sym;
1431 int symind;
1432
1433 symind = findsym(&cp);
1434 if (symind < 0) {
1435 if (nsyms >= MAXSYMS)
1436 errx(2, "too many symbols");
1437 symind = nsyms++;
1438 }
1439 ignore[symind] = ignorethis;
1440 symname[symind] = sym;
1441 value[symind] = val;
1442 debugsym("addsym", symind);
1443 }
1444
1445 static void
debugsym(const char * why,int symind)1446 debugsym(const char *why, int symind)
1447 {
1448 debug("%s %s%c%s", why, symname[symind],
1449 value[symind] ? '=' : ' ',
1450 value[symind] ? value[symind] : "undef");
1451 }
1452
1453 /*
1454 * Add symbols to the symbol table from a file containing
1455 * #define and #undef preprocessor directives.
1456 */
1457 static void
defundefile(const char * fn)1458 defundefile(const char *fn)
1459 {
1460 filename = fn;
1461 input = fopen(fn, "rb");
1462 if (input == NULL)
1463 err(2, "can't open %s", fn);
1464 linenum = 0;
1465 while (defundef())
1466 ;
1467 if (ferror(input))
1468 err(2, "can't read %s", filename);
1469 else
1470 fclose(input);
1471 if (incomment)
1472 error("EOF in comment");
1473 }
1474
1475 /*
1476 * Read and process one #define or #undef directive
1477 */
1478 static bool
defundef(void)1479 defundef(void)
1480 {
1481 const char *cp, *kw, *sym, *val, *end;
1482
1483 cp = skiphash();
1484 if (cp == NULL)
1485 return (false);
1486 if (*cp == '\0')
1487 goto done;
1488 /* strip trailing whitespace, and do a fairly rough check to
1489 avoid unsupported multi-line preprocessor directives */
1490 end = cp + strlen(cp);
1491 while (end > tline && strchr(" \t\n\r", end[-1]) != NULL)
1492 --end;
1493 if (end > tline && end[-1] == '\\')
1494 Eioccc();
1495
1496 kw = cp;
1497 if ((cp = matchsym("define", kw)) != NULL) {
1498 sym = getsym(&cp);
1499 if (sym == NULL)
1500 error("Missing macro name in #define");
1501 if (*cp == '(') {
1502 val = "1";
1503 } else {
1504 cp = skipcomment(cp);
1505 val = (cp < end) ? xstrdup(cp, end) : "";
1506 }
1507 debug("#define");
1508 addsym2(false, sym, val);
1509 } else if ((cp = matchsym("undef", kw)) != NULL) {
1510 sym = getsym(&cp);
1511 if (sym == NULL)
1512 error("Missing macro name in #undef");
1513 cp = skipcomment(cp);
1514 debug("#undef");
1515 addsym2(false, sym, NULL);
1516 } else {
1517 error("Unrecognized preprocessor directive");
1518 }
1519 skipline(cp);
1520 done:
1521 debug("parser line %d state %s comment %s line", linenum,
1522 comment_name[incomment], linestate_name[linestate]);
1523 return (true);
1524 }
1525
1526 /*
1527 * Concatenate two strings into new memory, checking for failure.
1528 */
1529 static char *
astrcat(const char * s1,const char * s2)1530 astrcat(const char *s1, const char *s2)
1531 {
1532 char *s;
1533
1534 if (asprintf(&s, "%s%s", s1, s2) == -1)
1535 err(2, "asprintf");
1536 return (s);
1537 }
1538
1539 /*
1540 * Duplicate a segment of a string, checking for failure.
1541 */
1542 static const char *
xstrdup(const char * start,const char * end)1543 xstrdup(const char *start, const char *end)
1544 {
1545 size_t n;
1546 char *s;
1547
1548 if (end < start) abort(); /* bug */
1549 n = (size_t)(end - start) + 1;
1550 s = (char *)malloc(n);
1551 if (s == NULL)
1552 err(2, "malloc");
1553 snprintf(s, n, "%s", start);
1554 return (s);
1555 }
1556
1557 /*
1558 * Diagnostics.
1559 */
1560 static void
debug(const char * msg,...)1561 debug(const char *msg, ...)
1562 {
1563 va_list ap;
1564
1565 if (debugging) {
1566 va_start(ap, msg);
1567 vwarnx(msg, ap);
1568 va_end(ap);
1569 }
1570 }
1571
1572 static void
error(const char * msg)1573 error(const char *msg)
1574 {
1575 if (depth == 0)
1576 warnx("%s: %d: %s", filename, linenum, msg);
1577 else
1578 warnx("%s: %d: %s (#if line %d depth %d)",
1579 filename, linenum, msg, stifline[depth], depth);
1580 closeio();
1581 errx(2, "Output may be truncated");
1582 }
1583
1584 static FILE *
mktempmode(char * tmp,int mode)1585 mktempmode(char *tmp, int mode)
1586 {
1587 int fd = mkstemp(tmp);
1588 if (fd == -1)
1589 return (NULL);
1590 fchmod(fd, mode & (S_IRWXU|S_IRWXG|S_IRWXO));
1591 return (fdopen(fd, "wb"));
1592 }
1593