1 /*
2 * Copyright (c) 2002 - 2015 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 "unifdef.h"
47
48 static const char copyright[] =
49 "@(#) $Version: unifdef-2.11 $\n"
50 "@(#) $FreeBSD$\n"
51 "@(#) $Author: Tony Finch (dot@dotat.at) $\n"
52 "@(#) $URL: http://dotat.at/prog/unifdef $\n"
53 ;
54
55 /* types of input lines: */
56 typedef enum {
57 LT_TRUEI, /* a true #if with ignore flag */
58 LT_FALSEI, /* a false #if with ignore flag */
59 LT_IF, /* an unknown #if */
60 LT_TRUE, /* a true #if */
61 LT_FALSE, /* a false #if */
62 LT_ELIF, /* an unknown #elif */
63 LT_ELTRUE, /* a true #elif */
64 LT_ELFALSE, /* a false #elif */
65 LT_ELSE, /* #else */
66 LT_ENDIF, /* #endif */
67 LT_DODGY, /* flag: directive is not on one line */
68 LT_DODGY_LAST = LT_DODGY + LT_ENDIF,
69 LT_PLAIN, /* ordinary line */
70 LT_EOF, /* end of file */
71 LT_ERROR, /* unevaluable #if */
72 LT_COUNT
73 } Linetype;
74
75 static char const * const linetype_name[] = {
76 "TRUEI", "FALSEI", "IF", "TRUE", "FALSE",
77 "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF",
78 "DODGY TRUEI", "DODGY FALSEI",
79 "DODGY IF", "DODGY TRUE", "DODGY FALSE",
80 "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE",
81 "DODGY ELSE", "DODGY ENDIF",
82 "PLAIN", "EOF", "ERROR"
83 };
84
85 #define linetype_if2elif(lt) ((Linetype)(lt - LT_IF + LT_ELIF))
86 #define linetype_2dodgy(lt) ((Linetype)(lt + LT_DODGY))
87
88 /* state of #if processing */
89 typedef enum {
90 IS_OUTSIDE,
91 IS_FALSE_PREFIX, /* false #if followed by false #elifs */
92 IS_TRUE_PREFIX, /* first non-false #(el)if is true */
93 IS_PASS_MIDDLE, /* first non-false #(el)if is unknown */
94 IS_FALSE_MIDDLE, /* a false #elif after a pass state */
95 IS_TRUE_MIDDLE, /* a true #elif after a pass state */
96 IS_PASS_ELSE, /* an else after a pass state */
97 IS_FALSE_ELSE, /* an else after a true state */
98 IS_TRUE_ELSE, /* an else after only false states */
99 IS_FALSE_TRAILER, /* #elifs after a true are false */
100 IS_COUNT
101 } Ifstate;
102
103 static char const * const ifstate_name[] = {
104 "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX",
105 "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE",
106 "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE",
107 "FALSE_TRAILER"
108 };
109
110 /* state of comment parser */
111 typedef enum {
112 NO_COMMENT = false, /* outside a comment */
113 C_COMMENT, /* in a comment like this one */
114 CXX_COMMENT, /* between // and end of line */
115 STARTING_COMMENT, /* just after slash-backslash-newline */
116 FINISHING_COMMENT, /* star-backslash-newline in a C comment */
117 CHAR_LITERAL, /* inside '' */
118 STRING_LITERAL /* inside "" */
119 } Comment_state;
120
121 static char const * const comment_name[] = {
122 "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING"
123 };
124
125 /* state of preprocessor line parser */
126 typedef enum {
127 LS_START, /* only space and comments on this line */
128 LS_HASH, /* only space, comments, and a hash */
129 LS_DIRTY /* this line can't be a preprocessor line */
130 } Line_state;
131
132 static char const * const linestate_name[] = {
133 "START", "HASH", "DIRTY"
134 };
135
136 /*
137 * Minimum translation limits from ISO/IEC 9899:1999 5.2.4.1
138 */
139 #define MAXDEPTH 64 /* maximum #if nesting */
140 #define MAXLINE 4096 /* maximum length of line */
141 #define MAXSYMS 16384 /* maximum number of symbols */
142
143 /*
144 * Sometimes when editing a keyword the replacement text is longer, so
145 * we leave some space at the end of the tline buffer to accommodate this.
146 */
147 #define EDITSLOP 10
148
149 /*
150 * Globals.
151 */
152
153 static bool compblank; /* -B: compress blank lines */
154 static bool lnblank; /* -b: blank deleted lines */
155 static bool complement; /* -c: do the complement */
156 static bool debugging; /* -d: debugging reports */
157 static bool inplace; /* -m: modify in place */
158 static bool iocccok; /* -e: fewer IOCCC errors */
159 static bool strictlogic; /* -K: keep ambiguous #ifs */
160 static bool killconsts; /* -k: eval constant #ifs */
161 static bool lnnum; /* -n: add #line directives */
162 static bool symlist; /* -s: output symbol list */
163 static bool symdepth; /* -S: output symbol depth */
164 static bool text; /* -t: this is a text file */
165
166 static const char *symname[MAXSYMS]; /* symbol name */
167 static const char *value[MAXSYMS]; /* -Dsym=value */
168 static bool ignore[MAXSYMS]; /* -iDsym or -iUsym */
169 static int nsyms; /* number of symbols */
170
171 static FILE *input; /* input file pointer */
172 static const char *filename; /* input file name */
173 static int linenum; /* current line number */
174 static const char *linefile; /* file name for #line */
175 static FILE *output; /* output file pointer */
176 static const char *ofilename; /* output file name */
177 static const char *backext; /* backup extension */
178 static char *tempname; /* avoid splatting input */
179
180 static char tline[MAXLINE+EDITSLOP];/* input buffer plus space */
181 static char *keyword; /* used for editing #elif's */
182
183 /*
184 * When processing a file, the output's newline style will match the
185 * input's, and unifdef correctly handles CRLF or LF endings whatever
186 * the platform's native style. The stdio streams are opened in binary
187 * mode to accommodate platforms whose native newline style is CRLF.
188 * When the output isn't a processed input file (when it is error /
189 * debug / diagnostic messages) then unifdef uses native line endings.
190 */
191
192 static const char *newline; /* input file format */
193 static const char newline_unix[] = "\n";
194 static const char newline_crlf[] = "\r\n";
195
196 static Comment_state incomment; /* comment parser state */
197 static Line_state linestate; /* #if line parser state */
198 static Ifstate ifstate[MAXDEPTH]; /* #if processor state */
199 static bool ignoring[MAXDEPTH]; /* ignore comments state */
200 static int stifline[MAXDEPTH]; /* start of current #if */
201 static int depth; /* current #if nesting */
202 static int delcount; /* count of deleted lines */
203 static unsigned blankcount; /* count of blank lines */
204 static unsigned blankmax; /* maximum recent blankcount */
205 static bool constexpr; /* constant #if expression */
206 static bool zerosyms; /* to format symdepth output */
207 static bool firstsym; /* ditto */
208
209 static int exitmode; /* exit status mode */
210 static int exitstat; /* program exit status */
211 static bool altered; /* was this file modified? */
212
213 static void addsym1(bool, bool, char *);
214 static void addsym2(bool, const char *, const char *);
215 static char *astrcat(const char *, const char *);
216 static void cleantemp(void);
217 static void closeio(void);
218 static void debug(const char *, ...);
219 static void debugsym(const char *, int);
220 static bool defundef(void);
221 static void defundefile(const char *);
222 static void done(void);
223 static void error(const char *);
224 static int findsym(const char **);
225 static void flushline(bool);
226 static void hashline(void);
227 static void help(void);
228 static Linetype ifeval(const char **);
229 static void ignoreoff(void);
230 static void ignoreon(void);
231 static void indirectsym(void);
232 static void keywordedit(const char *);
233 static const char *matchsym(const char *, const char *);
234 static void nest(void);
235 static Linetype parseline(void);
236 static void process(void);
237 static void processinout(const char *, const char *);
238 static const char *skipargs(const char *);
239 static const char *skipcomment(const char *);
240 static const char *skiphash(void);
241 static const char *skipline(const char *);
242 static const char *skipsym(const char *);
243 static void state(Ifstate);
244 static void unnest(void);
245 static void usage(void);
246 static void version(void);
247 static const char *xstrdup(const char *, const char *);
248
249 #define endsym(c) (!isalnum((unsigned char)c) && c != '_')
250
251 /*
252 * The main program.
253 */
254 int
main(int argc,char * argv[])255 main(int argc, char *argv[])
256 {
257 int opt;
258
259 while ((opt = getopt(argc, argv, "i:D:U:f:I:M:o:x:bBcdehKklmnsStV")) != -1)
260 switch (opt) {
261 case 'i': /* treat stuff controlled by these symbols as text */
262 /*
263 * For strict backwards-compatibility the U or D
264 * should be immediately after the -i but it doesn't
265 * matter much if we relax that requirement.
266 */
267 opt = *optarg++;
268 if (opt == 'D')
269 addsym1(true, true, optarg);
270 else if (opt == 'U')
271 addsym1(true, false, optarg);
272 else
273 usage();
274 break;
275 case 'D': /* define a symbol */
276 addsym1(false, true, optarg);
277 break;
278 case 'U': /* undef a symbol */
279 addsym1(false, false, optarg);
280 break;
281 case 'I': /* no-op for compatibility with cpp */
282 break;
283 case 'b': /* blank deleted lines instead of omitting them */
284 case 'l': /* backwards compatibility */
285 lnblank = true;
286 break;
287 case 'B': /* compress blank lines around removed section */
288 compblank = true;
289 break;
290 case 'c': /* treat -D as -U and vice versa */
291 complement = true;
292 break;
293 case 'd':
294 debugging = true;
295 break;
296 case 'e': /* fewer errors from dodgy lines */
297 iocccok = true;
298 break;
299 case 'f': /* definitions file */
300 defundefile(optarg);
301 break;
302 case 'h':
303 help();
304 break;
305 case 'K': /* keep ambiguous #ifs */
306 strictlogic = true;
307 break;
308 case 'k': /* process constant #ifs */
309 killconsts = true;
310 break;
311 case 'm': /* modify in place */
312 inplace = true;
313 break;
314 case 'M': /* modify in place and keep backup */
315 inplace = true;
316 if (strlen(optarg) > 0)
317 backext = optarg;
318 break;
319 case 'n': /* add #line directive after deleted lines */
320 lnnum = true;
321 break;
322 case 'o': /* output to a file */
323 ofilename = optarg;
324 break;
325 case 's': /* only output list of symbols that control #ifs */
326 symlist = true;
327 break;
328 case 'S': /* list symbols with their nesting depth */
329 symlist = symdepth = true;
330 break;
331 case 't': /* don't parse C comments */
332 text = true;
333 break;
334 case 'V':
335 version();
336 break;
337 case 'x':
338 exitmode = atoi(optarg);
339 if(exitmode < 0 || exitmode > 2)
340 usage();
341 break;
342 default:
343 usage();
344 }
345 argc -= optind;
346 argv += optind;
347 if (compblank && lnblank)
348 errx(2, "-B and -b are mutually exclusive");
349 if (symlist && (ofilename != NULL || inplace || argc > 1))
350 errx(2, "-s only works with one input file");
351 if (argc > 1 && ofilename != NULL)
352 errx(2, "-o cannot be used with multiple input files");
353 if (argc > 1 && !inplace)
354 errx(2, "multiple input files require -m or -M");
355 if (argc == 0 && inplace)
356 errx(2, "-m requires an input file");
357 if (argc == 0)
358 argc = 1;
359 if (argc == 1 && !inplace && ofilename == NULL)
360 ofilename = "-";
361 indirectsym();
362
363 atexit(cleantemp);
364 if (ofilename != NULL)
365 processinout(*argv, ofilename);
366 else while (argc-- > 0) {
367 processinout(*argv, *argv);
368 argv++;
369 }
370 switch(exitmode) {
371 case(0): exit(exitstat);
372 case(1): exit(!exitstat);
373 case(2): exit(0);
374 default: abort(); /* bug */
375 }
376 }
377
378 /*
379 * File logistics.
380 */
381 static void
processinout(const char * ifn,const char * ofn)382 processinout(const char *ifn, const char *ofn)
383 {
384 struct stat st;
385
386 if (ifn == NULL || strcmp(ifn, "-") == 0) {
387 filename = "[stdin]";
388 linefile = NULL;
389 input = fbinmode(stdin);
390 } else {
391 filename = ifn;
392 linefile = ifn;
393 input = fopen(ifn, "rb");
394 if (input == NULL)
395 err(2, "can't open %s", ifn);
396 }
397 if (strcmp(ofn, "-") == 0) {
398 output = fbinmode(stdout);
399 process();
400 return;
401 }
402 if (stat(ofn, &st) < 0) {
403 output = fopen(ofn, "wb");
404 if (output == NULL)
405 err(2, "can't create %s", ofn);
406 process();
407 return;
408 }
409
410 tempname = astrcat(ofn, ".XXXXXX");
411 output = mktempmode(tempname, st.st_mode);
412 if (output == NULL)
413 err(2, "can't create %s", tempname);
414
415 process();
416
417 if (backext != NULL) {
418 char *backname = astrcat(ofn, backext);
419 if (rename(ofn, backname) < 0)
420 err(2, "can't rename \"%s\" to \"%s\"", ofn, backname);
421 free(backname);
422 }
423 /* leave file unmodified if unifdef made no changes */
424 if (!altered && backext == NULL) {
425 if (remove(tempname) < 0)
426 warn("can't remove \"%s\"", tempname);
427 } else if (replace(tempname, ofn) < 0)
428 err(2, "can't rename \"%s\" to \"%s\"", tempname, ofn);
429 free(tempname);
430 tempname = NULL;
431 }
432
433 /*
434 * For cleaning up if there is an error.
435 */
436 static void
cleantemp(void)437 cleantemp(void)
438 {
439 if (tempname != NULL)
440 remove(tempname);
441 }
442
443 /*
444 * Self-identification functions.
445 */
446
447 static void
version(void)448 version(void)
449 {
450 const char *c = copyright;
451 for (;;) {
452 while (*++c != '$')
453 if (*c == '\0')
454 exit(0);
455 while (*++c != '$')
456 putc(*c, stderr);
457 putc('\n', stderr);
458 }
459 }
460
461 static void
synopsis(FILE * fp)462 synopsis(FILE *fp)
463 {
464 fprintf(fp,
465 "usage: unifdef [-bBcdehKkmnsStV] [-x{012}] [-Mext] [-opath] \\\n"
466 " [-[i]Dsym[=val]] [-[i]Usym] [-fpath] ... [file] ...\n");
467 }
468
469 static void
usage(void)470 usage(void)
471 {
472 synopsis(stderr);
473 exit(2);
474 }
475
476 static void
help(void)477 help(void)
478 {
479 synopsis(stdout);
480 printf(
481 " -Dsym=val define preprocessor symbol with given value\n"
482 " -Dsym define preprocessor symbol with value 1\n"
483 " -Usym preprocessor symbol is undefined\n"
484 " -iDsym=val \\ ignore C strings and comments\n"
485 " -iDsym ) in sections controlled by these\n"
486 " -iUsym / preprocessor symbols\n"
487 " -fpath file containing #define and #undef directives\n"
488 " -b blank lines instead of deleting them\n"
489 " -B compress blank lines around deleted section\n"
490 " -c complement (invert) keep vs. delete\n"
491 " -d debugging mode\n"
492 " -e ignore multiline preprocessor directives\n"
493 " -h print help\n"
494 " -Ipath extra include file path (ignored)\n"
495 " -K disable && and || short-circuiting\n"
496 " -k process constant #if expressions\n"
497 " -Mext modify in place and keep backups\n"
498 " -m modify input files in place\n"
499 " -n add #line directives to output\n"
500 " -opath output file name\n"
501 " -S list #if control symbols with nesting\n"
502 " -s list #if control symbols\n"
503 " -t ignore C strings and comments\n"
504 " -V print version\n"
505 " -x{012} exit status mode\n"
506 );
507 exit(0);
508 }
509
510 /*
511 * A state transition function alters the global #if processing state
512 * in a particular way. The table below is indexed by the current
513 * processing state and the type of the current line.
514 *
515 * Nesting is handled by keeping a stack of states; some transition
516 * functions increase or decrease the depth. They also maintain the
517 * ignore state on a stack. In some complicated cases they have to
518 * alter the preprocessor directive, as follows.
519 *
520 * When we have processed a group that starts off with a known-false
521 * #if/#elif sequence (which has therefore been deleted) followed by a
522 * #elif that we don't understand and therefore must keep, we edit the
523 * latter into a #if to keep the nesting correct. We use memcpy() to
524 * overwrite the 4 byte token "elif" with "if " without a '\0' byte.
525 *
526 * When we find a true #elif in a group, the following block will
527 * always be kept and the rest of the sequence after the next #elif or
528 * #else will be discarded. We edit the #elif into a #else and the
529 * following directive to #endif since this has the desired behaviour.
530 *
531 * "Dodgy" directives are split across multiple lines, the most common
532 * example being a multi-line comment hanging off the right of the
533 * directive. We can handle them correctly only if there is no change
534 * from printing to dropping (or vice versa) caused by that directive.
535 * If the directive is the first of a group we have a choice between
536 * failing with an error, or passing it through unchanged instead of
537 * evaluating it. The latter is not the default to avoid questions from
538 * users about unifdef unexpectedly leaving behind preprocessor directives.
539 */
540 typedef void state_fn(void);
541
542 /* report an error */
Eelif(void)543 static void Eelif (void) { error("Inappropriate #elif"); }
Eelse(void)544 static void Eelse (void) { error("Inappropriate #else"); }
Eendif(void)545 static void Eendif(void) { error("Inappropriate #endif"); }
Eeof(void)546 static void Eeof (void) { error("Premature EOF"); }
Eioccc(void)547 static void Eioccc(void) { error("Obfuscated preprocessor control line"); }
548 /* plain line handling */
print(void)549 static void print (void) { flushline(true); }
drop(void)550 static void drop (void) { flushline(false); }
551 /* output lacks group's start line */
Strue(void)552 static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); }
Sfalse(void)553 static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); }
Selse(void)554 static void Selse (void) { drop(); state(IS_TRUE_ELSE); }
555 /* print/pass this block */
Pelif(void)556 static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); }
Pelse(void)557 static void Pelse (void) { print(); state(IS_PASS_ELSE); }
Pendif(void)558 static void Pendif(void) { print(); unnest(); }
559 /* discard this block */
Dfalse(void)560 static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); }
Delif(void)561 static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); }
Delse(void)562 static void Delse (void) { drop(); state(IS_FALSE_ELSE); }
Dendif(void)563 static void Dendif(void) { drop(); unnest(); }
564 /* first line of group */
Fdrop(void)565 static void Fdrop (void) { nest(); Dfalse(); }
Fpass(void)566 static void Fpass (void) { nest(); Pelif(); }
Ftrue(void)567 static void Ftrue (void) { nest(); Strue(); }
Ffalse(void)568 static void Ffalse(void) { nest(); Sfalse(); }
569 /* variable pedantry for obfuscated lines */
Oiffy(void)570 static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); }
Oif(void)571 static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); }
Oelif(void)572 static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); }
573 /* ignore comments in this block */
Idrop(void)574 static void Idrop (void) { Fdrop(); ignoreon(); }
Itrue(void)575 static void Itrue (void) { Ftrue(); ignoreon(); }
Ifalse(void)576 static void Ifalse(void) { Ffalse(); ignoreon(); }
577 /* modify this line */
Mpass(void)578 static void Mpass (void) { memcpy(keyword, "if ", 4); Pelif(); }
Mtrue(void)579 static void Mtrue (void) { keywordedit("else"); state(IS_TRUE_MIDDLE); }
Melif(void)580 static void Melif (void) { keywordedit("endif"); state(IS_FALSE_TRAILER); }
Melse(void)581 static void Melse (void) { keywordedit("endif"); state(IS_FALSE_ELSE); }
582
583 static state_fn * const trans_table[IS_COUNT][LT_COUNT] = {
584 /* IS_OUTSIDE */
585 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif,
586 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eendif,
587 print, done, abort },
588 /* IS_FALSE_PREFIX */
589 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif,
590 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc,
591 drop, Eeof, abort },
592 /* IS_TRUE_PREFIX */
593 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif,
594 Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
595 print, Eeof, abort },
596 /* IS_PASS_MIDDLE */
597 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif,
598 Oiffy, Oiffy, Fpass, Oif, Oif, Pelif, Oelif, Oelif, Pelse, Pendif,
599 print, Eeof, abort },
600 /* IS_FALSE_MIDDLE */
601 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif,
602 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc,
603 drop, Eeof, abort },
604 /* IS_TRUE_MIDDLE */
605 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif,
606 Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Pendif,
607 print, Eeof, abort },
608 /* IS_PASS_ELSE */
609 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif,
610 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Pendif,
611 print, Eeof, abort },
612 /* IS_FALSE_ELSE */
613 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif,
614 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc,
615 drop, Eeof, abort },
616 /* IS_TRUE_ELSE */
617 { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif,
618 Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eioccc,
619 print, Eeof, abort },
620 /* IS_FALSE_TRAILER */
621 { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif,
622 Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc,
623 drop, Eeof, abort }
624 /*TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF
625 TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF (DODGY)
626 PLAIN EOF ERROR */
627 };
628
629 /*
630 * State machine utility functions
631 */
632 static void
ignoreoff(void)633 ignoreoff(void)
634 {
635 if (depth == 0)
636 abort(); /* bug */
637 ignoring[depth] = ignoring[depth-1];
638 }
639 static void
ignoreon(void)640 ignoreon(void)
641 {
642 ignoring[depth] = true;
643 }
644 static void
keywordedit(const char * replacement)645 keywordedit(const char *replacement)
646 {
647 snprintf(keyword, tline + sizeof(tline) - keyword,
648 "%s%s", replacement, newline);
649 altered = true;
650 print();
651 }
652 static void
nest(void)653 nest(void)
654 {
655 if (depth > MAXDEPTH-1)
656 abort(); /* bug */
657 if (depth == MAXDEPTH-1)
658 error("Too many levels of nesting");
659 depth += 1;
660 stifline[depth] = linenum;
661 }
662 static void
unnest(void)663 unnest(void)
664 {
665 if (depth == 0)
666 abort(); /* bug */
667 depth -= 1;
668 }
669 static void
state(Ifstate is)670 state(Ifstate is)
671 {
672 ifstate[depth] = is;
673 }
674
675 /*
676 * The last state transition function. When this is called,
677 * lineval == LT_EOF, so the process() loop will terminate.
678 */
679 static void
done(void)680 done(void)
681 {
682 if (incomment)
683 error("EOF in comment");
684 closeio();
685 }
686
687 /*
688 * Write a line to the output or not, according to command line options.
689 * If writing fails, closeio() will print the error and exit.
690 */
691 static void
flushline(bool keep)692 flushline(bool keep)
693 {
694 if (symlist)
695 return;
696 if (keep ^ complement) {
697 bool blankline = tline[strspn(tline, " \t\r\n")] == '\0';
698 if (blankline && compblank && blankcount != blankmax) {
699 delcount += 1;
700 blankcount += 1;
701 } else {
702 if (lnnum && delcount > 0)
703 hashline();
704 if (fputs(tline, output) == EOF)
705 closeio();
706 delcount = 0;
707 blankmax = blankcount = blankline ? blankcount + 1 : 0;
708 }
709 } else {
710 if (lnblank && fputs(newline, output) == EOF)
711 closeio();
712 altered = true;
713 delcount += 1;
714 blankcount = 0;
715 }
716 if (debugging && fflush(output) == EOF)
717 closeio();
718 }
719
720 /*
721 * Format of #line directives depends on whether we know the input filename.
722 */
723 static void
hashline(void)724 hashline(void)
725 {
726 int e;
727
728 if (linefile == NULL)
729 e = fprintf(output, "#line %d%s", linenum, newline);
730 else
731 e = fprintf(output, "#line %d \"%s\"%s",
732 linenum, linefile, newline);
733 if (e < 0)
734 closeio();
735 }
736
737 /*
738 * Flush the output and handle errors.
739 */
740 static void
closeio(void)741 closeio(void)
742 {
743 /* Tidy up after findsym(). */
744 if (symdepth && !zerosyms)
745 printf("\n");
746 if (output != NULL && (ferror(output) || fclose(output) == EOF))
747 err(2, "%s: can't write to output", filename);
748 fclose(input);
749 }
750
751 /*
752 * The driver for the state machine.
753 */
754 static void
process(void)755 process(void)
756 {
757 Linetype lineval = LT_PLAIN;
758 /* When compressing blank lines, act as if the file
759 is preceded by a large number of blank lines. */
760 blankmax = blankcount = 1000;
761 zerosyms = true;
762 newline = NULL;
763 linenum = 0;
764 altered = false;
765 while (lineval != LT_EOF) {
766 lineval = parseline();
767 trans_table[ifstate[depth]][lineval]();
768 debug("process line %d %s -> %s depth %d",
769 linenum, linetype_name[lineval],
770 ifstate_name[ifstate[depth]], depth);
771 }
772 exitstat |= altered;
773 }
774
775 /*
776 * Parse a line and determine its type. We keep the preprocessor line
777 * parser state between calls in the global variable linestate, with
778 * help from skipcomment().
779 */
780 static Linetype
parseline(void)781 parseline(void)
782 {
783 const char *cp;
784 int cursym;
785 Linetype retval;
786 Comment_state wascomment;
787
788 wascomment = incomment;
789 cp = skiphash();
790 if (cp == NULL)
791 return (LT_EOF);
792 if (newline == NULL) {
793 if (strrchr(tline, '\n') == strrchr(tline, '\r') + 1)
794 newline = newline_crlf;
795 else
796 newline = newline_unix;
797 }
798 if (*cp == '\0') {
799 retval = LT_PLAIN;
800 goto done;
801 }
802 keyword = tline + (cp - tline);
803 if ((cp = matchsym("ifdef", keyword)) != NULL ||
804 (cp = matchsym("ifndef", keyword)) != NULL) {
805 cp = skipcomment(cp);
806 if ((cursym = findsym(&cp)) < 0)
807 retval = LT_IF;
808 else {
809 retval = (keyword[2] == 'n')
810 ? LT_FALSE : LT_TRUE;
811 if (value[cursym] == NULL)
812 retval = (retval == LT_TRUE)
813 ? LT_FALSE : LT_TRUE;
814 if (ignore[cursym])
815 retval = (retval == LT_TRUE)
816 ? LT_TRUEI : LT_FALSEI;
817 }
818 } else if ((cp = matchsym("if", keyword)) != NULL)
819 retval = ifeval(&cp);
820 else if ((cp = matchsym("elif", keyword)) != NULL)
821 retval = linetype_if2elif(ifeval(&cp));
822 else if ((cp = matchsym("else", keyword)) != NULL)
823 retval = LT_ELSE;
824 else if ((cp = matchsym("endif", keyword)) != NULL)
825 retval = LT_ENDIF;
826 else {
827 cp = skipsym(keyword);
828 /* no way can we deal with a continuation inside a keyword */
829 if (strncmp(cp, "\\\r\n", 3) == 0 ||
830 strncmp(cp, "\\\n", 2) == 0)
831 Eioccc();
832 cp = skipline(cp);
833 retval = LT_PLAIN;
834 goto done;
835 }
836 cp = skipcomment(cp);
837 if (*cp != '\0') {
838 cp = skipline(cp);
839 if (retval == LT_TRUE || retval == LT_FALSE ||
840 retval == LT_TRUEI || retval == LT_FALSEI)
841 retval = LT_IF;
842 if (retval == LT_ELTRUE || retval == LT_ELFALSE)
843 retval = LT_ELIF;
844 }
845 /* the following can happen if the last line of the file lacks a
846 newline or if there is too much whitespace in a directive */
847 if (linestate == LS_HASH) {
848 long len = cp - tline;
849 if (fgets(tline + len, MAXLINE - len, input) == NULL) {
850 if (ferror(input))
851 err(2, "can't read %s", filename);
852 /* append the missing newline at eof */
853 strcpy(tline + len, newline);
854 cp += strlen(newline);
855 linestate = LS_START;
856 } else {
857 linestate = LS_DIRTY;
858 }
859 }
860 if (retval != LT_PLAIN && (wascomment || linestate != LS_START)) {
861 retval = linetype_2dodgy(retval);
862 linestate = LS_DIRTY;
863 }
864 done:
865 debug("parser line %d state %s comment %s line", linenum,
866 comment_name[incomment], linestate_name[linestate]);
867 return (retval);
868 }
869
870 /*
871 * These are the binary operators that are supported by the expression
872 * evaluator.
873 */
op_strict(long * p,long v,Linetype at,Linetype bt)874 static Linetype op_strict(long *p, long v, Linetype at, Linetype bt) {
875 if(at == LT_IF || bt == LT_IF) return (LT_IF);
876 return (*p = v, v ? LT_TRUE : LT_FALSE);
877 }
op_lt(long * p,Linetype at,long a,Linetype bt,long b)878 static Linetype op_lt(long *p, Linetype at, long a, Linetype bt, long b) {
879 return op_strict(p, a < b, at, bt);
880 }
op_gt(long * p,Linetype at,long a,Linetype bt,long b)881 static Linetype op_gt(long *p, Linetype at, long a, Linetype bt, long b) {
882 return op_strict(p, a > b, at, bt);
883 }
op_le(long * p,Linetype at,long a,Linetype bt,long b)884 static Linetype op_le(long *p, Linetype at, long a, Linetype bt, long b) {
885 return op_strict(p, a <= b, at, bt);
886 }
op_ge(long * p,Linetype at,long a,Linetype bt,long b)887 static Linetype op_ge(long *p, Linetype at, long a, Linetype bt, long b) {
888 return op_strict(p, a >= b, at, bt);
889 }
op_eq(long * p,Linetype at,long a,Linetype bt,long b)890 static Linetype op_eq(long *p, Linetype at, long a, Linetype bt, long b) {
891 return op_strict(p, a == b, at, bt);
892 }
op_ne(long * p,Linetype at,long a,Linetype bt,long b)893 static Linetype op_ne(long *p, Linetype at, long a, Linetype bt, long b) {
894 return op_strict(p, a != b, at, bt);
895 }
op_or(long * p,Linetype at,long a,Linetype bt,long b)896 static Linetype op_or(long *p, Linetype at, long a, Linetype bt, long b) {
897 if (!strictlogic && (at == LT_TRUE || bt == LT_TRUE))
898 return (*p = 1, LT_TRUE);
899 return op_strict(p, a || b, at, bt);
900 }
op_and(long * p,Linetype at,long a,Linetype bt,long b)901 static Linetype op_and(long *p, Linetype at, long a, Linetype bt, long b) {
902 if (!strictlogic && (at == LT_FALSE || bt == LT_FALSE))
903 return (*p = 0, LT_FALSE);
904 return op_strict(p, a && b, at, bt);
905 }
op_blsh(long * p,Linetype at,long a,Linetype bt,long b)906 static Linetype op_blsh(long *p, Linetype at, long a, Linetype bt, long b) {
907 return op_strict(p, a << b, at, bt);
908 }
op_brsh(long * p,Linetype at,long a,Linetype bt,long b)909 static Linetype op_brsh(long *p, Linetype at, long a, Linetype bt, long b) {
910 return op_strict(p, a >> b, at, bt);
911 }
op_add(long * p,Linetype at,long a,Linetype bt,long b)912 static Linetype op_add(long *p, Linetype at, long a, Linetype bt, long b) {
913 return op_strict(p, a + b, at, bt);
914 }
op_sub(long * p,Linetype at,long a,Linetype bt,long b)915 static Linetype op_sub(long *p, Linetype at, long a, Linetype bt, long b) {
916 return op_strict(p, a - b, at, bt);
917 }
op_mul(long * p,Linetype at,long a,Linetype bt,long b)918 static Linetype op_mul(long *p, Linetype at, long a, Linetype bt, long b) {
919 return op_strict(p, a * b, at, bt);
920 }
op_div(long * p,Linetype at,long a,Linetype bt,long b)921 static Linetype op_div(long *p, Linetype at, long a, Linetype bt, long b) {
922 if (bt != LT_TRUE) {
923 debug("eval division by zero");
924 return (LT_ERROR);
925 }
926 return op_strict(p, a / b, at, bt);
927 }
op_mod(long * p,Linetype at,long a,Linetype bt,long b)928 static Linetype op_mod(long *p, Linetype at, long a, Linetype bt, long b) {
929 return op_strict(p, a % b, at, bt);
930 }
op_bor(long * p,Linetype at,long a,Linetype bt,long b)931 static Linetype op_bor(long *p, Linetype at, long a, Linetype bt, long b) {
932 return op_strict(p, a | b, at, bt);
933 }
op_bxor(long * p,Linetype at,long a,Linetype bt,long b)934 static Linetype op_bxor(long *p, Linetype at, long a, Linetype bt, long b) {
935 return op_strict(p, a ^ b, at, bt);
936 }
op_band(long * p,Linetype at,long a,Linetype bt,long b)937 static Linetype op_band(long *p, Linetype at, long a, Linetype bt, long b) {
938 return op_strict(p, a & b, at, bt);
939 }
940
941 /*
942 * An evaluation function takes three arguments, as follows: (1) a pointer to
943 * an element of the precedence table which lists the operators at the current
944 * level of precedence; (2) a pointer to an integer which will receive the
945 * value of the expression; and (3) a pointer to a char* that points to the
946 * expression to be evaluated and that is updated to the end of the expression
947 * when evaluation is complete. The function returns LT_FALSE if the value of
948 * the expression is zero, LT_TRUE if it is non-zero, LT_IF if the expression
949 * depends on an unknown symbol, or LT_ERROR if there is a parse failure.
950 */
951 struct ops;
952
953 typedef Linetype eval_fn(const struct ops *, long *, const char **);
954
955 static eval_fn eval_table, eval_unary;
956
957 /*
958 * The precedence table. Expressions involving binary operators are evaluated
959 * in a table-driven way by eval_table. When it evaluates a subexpression it
960 * calls the inner function with its first argument pointing to the next
961 * element of the table. Innermost expressions have special non-table-driven
962 * handling.
963 *
964 * The stop characters help with lexical analysis: an operator is not
965 * recognized if it is followed by one of the stop characters because
966 * that would make it a different operator.
967 */
968 struct op {
969 const char *str;
970 Linetype (*fn)(long *, Linetype, long, Linetype, long);
971 const char *stop;
972 };
973 struct ops {
974 eval_fn *inner;
975 struct op op[5];
976 };
977 static const struct ops eval_ops[] = {
978 { eval_table, { { "||", op_or, NULL } } },
979 { eval_table, { { "&&", op_and, NULL } } },
980 { eval_table, { { "|", op_bor, "|" } } },
981 { eval_table, { { "^", op_bxor, NULL } } },
982 { eval_table, { { "&", op_band, "&" } } },
983 { eval_table, { { "==", op_eq, NULL },
984 { "!=", op_ne, NULL } } },
985 { eval_table, { { "<=", op_le, NULL },
986 { ">=", op_ge, NULL },
987 { "<", op_lt, "<=" },
988 { ">", op_gt, ">=" } } },
989 { eval_table, { { "<<", op_blsh, NULL },
990 { ">>", op_brsh, NULL } } },
991 { eval_table, { { "+", op_add, NULL },
992 { "-", op_sub, NULL } } },
993 { eval_unary, { { "*", op_mul, NULL },
994 { "/", op_div, NULL },
995 { "%", op_mod, NULL } } },
996 };
997
998 /* Current operator precedence level */
prec(const struct ops * ops)999 static long prec(const struct ops *ops)
1000 {
1001 return (ops - eval_ops);
1002 }
1003
1004 /*
1005 * Function for evaluating the innermost parts of expressions,
1006 * viz. !expr (expr) number defined(symbol) symbol
1007 * We reset the constexpr flag in the last two cases.
1008 */
1009 static Linetype
eval_unary(const struct ops * ops,long * valp,const char ** cpp)1010 eval_unary(const struct ops *ops, long *valp, const char **cpp)
1011 {
1012 const char *cp;
1013 char *ep;
1014 int sym;
1015 bool defparen;
1016 Linetype lt;
1017
1018 cp = skipcomment(*cpp);
1019 if (*cp == '!') {
1020 debug("eval%d !", prec(ops));
1021 cp++;
1022 lt = eval_unary(ops, valp, &cp);
1023 if (lt == LT_ERROR)
1024 return (LT_ERROR);
1025 if (lt != LT_IF) {
1026 *valp = !*valp;
1027 lt = *valp ? LT_TRUE : LT_FALSE;
1028 }
1029 } else if (*cp == '~') {
1030 debug("eval%d ~", prec(ops));
1031 cp++;
1032 lt = eval_unary(ops, valp, &cp);
1033 if (lt == LT_ERROR)
1034 return (LT_ERROR);
1035 if (lt != LT_IF) {
1036 *valp = ~(*valp);
1037 lt = *valp ? LT_TRUE : LT_FALSE;
1038 }
1039 } else if (*cp == '-') {
1040 debug("eval%d -", prec(ops));
1041 cp++;
1042 lt = eval_unary(ops, valp, &cp);
1043 if (lt == LT_ERROR)
1044 return (LT_ERROR);
1045 if (lt != LT_IF) {
1046 *valp = -(*valp);
1047 lt = *valp ? LT_TRUE : LT_FALSE;
1048 }
1049 } else if (*cp == '(') {
1050 cp++;
1051 debug("eval%d (", prec(ops));
1052 lt = eval_table(eval_ops, valp, &cp);
1053 if (lt == LT_ERROR)
1054 return (LT_ERROR);
1055 cp = skipcomment(cp);
1056 if (*cp++ != ')')
1057 return (LT_ERROR);
1058 } else if (isdigit((unsigned char)*cp)) {
1059 debug("eval%d number", prec(ops));
1060 *valp = strtol(cp, &ep, 0);
1061 if (ep == cp)
1062 return (LT_ERROR);
1063 lt = *valp ? LT_TRUE : LT_FALSE;
1064 cp = ep;
1065 } else if (matchsym("defined", cp) != NULL) {
1066 cp = skipcomment(cp+7);
1067 if (*cp == '(') {
1068 cp = skipcomment(cp+1);
1069 defparen = true;
1070 } else {
1071 defparen = false;
1072 }
1073 sym = findsym(&cp);
1074 cp = skipcomment(cp);
1075 if (defparen && *cp++ != ')') {
1076 debug("eval%d defined missing ')'", prec(ops));
1077 return (LT_ERROR);
1078 }
1079 if (sym < 0) {
1080 debug("eval%d defined unknown", prec(ops));
1081 lt = LT_IF;
1082 } else {
1083 debug("eval%d defined %s", prec(ops), symname[sym]);
1084 *valp = (value[sym] != NULL);
1085 lt = *valp ? LT_TRUE : LT_FALSE;
1086 }
1087 constexpr = false;
1088 } else if (!endsym(*cp)) {
1089 debug("eval%d symbol", prec(ops));
1090 sym = findsym(&cp);
1091 if (sym < 0) {
1092 lt = LT_IF;
1093 cp = skipargs(cp);
1094 } else if (value[sym] == NULL) {
1095 *valp = 0;
1096 lt = LT_FALSE;
1097 } else {
1098 *valp = strtol(value[sym], &ep, 0);
1099 if (*ep != '\0' || ep == value[sym])
1100 return (LT_ERROR);
1101 lt = *valp ? LT_TRUE : LT_FALSE;
1102 cp = skipargs(cp);
1103 }
1104 constexpr = false;
1105 } else {
1106 debug("eval%d bad expr", prec(ops));
1107 return (LT_ERROR);
1108 }
1109
1110 *cpp = cp;
1111 debug("eval%d = %d", prec(ops), *valp);
1112 return (lt);
1113 }
1114
1115 /*
1116 * Table-driven evaluation of binary operators.
1117 */
1118 static Linetype
eval_table(const struct ops * ops,long * valp,const char ** cpp)1119 eval_table(const struct ops *ops, long *valp, const char **cpp)
1120 {
1121 const struct op *op;
1122 const char *cp;
1123 long val = 0;
1124 Linetype lt, rt;
1125
1126 debug("eval%d", prec(ops));
1127 cp = *cpp;
1128 lt = ops->inner(ops+1, valp, &cp);
1129 if (lt == LT_ERROR)
1130 return (LT_ERROR);
1131 for (;;) {
1132 cp = skipcomment(cp);
1133 for (op = ops->op; op->str != NULL; op++) {
1134 if (strncmp(cp, op->str, strlen(op->str)) == 0) {
1135 /* assume only one-char operators have stop chars */
1136 if (op->stop != NULL && cp[1] != '\0' &&
1137 strchr(op->stop, cp[1]) != NULL)
1138 continue;
1139 else
1140 break;
1141 }
1142 }
1143 if (op->str == NULL)
1144 break;
1145 cp += strlen(op->str);
1146 debug("eval%d %s", prec(ops), op->str);
1147 rt = ops->inner(ops+1, &val, &cp);
1148 if (rt == LT_ERROR)
1149 return (LT_ERROR);
1150 lt = op->fn(valp, lt, *valp, rt, val);
1151 }
1152
1153 *cpp = cp;
1154 debug("eval%d = %d", prec(ops), *valp);
1155 debug("eval%d lt = %s", prec(ops), linetype_name[lt]);
1156 return (lt);
1157 }
1158
1159 /*
1160 * Evaluate the expression on a #if or #elif line. If we can work out
1161 * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we
1162 * return just a generic LT_IF.
1163 */
1164 static Linetype
ifeval(const char ** cpp)1165 ifeval(const char **cpp)
1166 {
1167 Linetype ret;
1168 long val = 0;
1169
1170 debug("eval %s", *cpp);
1171 constexpr = killconsts ? false : true;
1172 ret = eval_table(eval_ops, &val, cpp);
1173 debug("eval = %d", val);
1174 return (constexpr ? LT_IF : ret == LT_ERROR ? LT_IF : ret);
1175 }
1176
1177 /*
1178 * Read a line and examine its initial part to determine if it is a
1179 * preprocessor directive. Returns NULL on EOF, or a pointer to a
1180 * preprocessor directive name, or a pointer to the zero byte at the
1181 * end of the line.
1182 */
1183 static const char *
skiphash(void)1184 skiphash(void)
1185 {
1186 const char *cp;
1187
1188 linenum++;
1189 if (fgets(tline, MAXLINE, input) == NULL) {
1190 if (ferror(input))
1191 err(2, "can't read %s", filename);
1192 else
1193 return (NULL);
1194 }
1195 cp = skipcomment(tline);
1196 if (linestate == LS_START && *cp == '#') {
1197 linestate = LS_HASH;
1198 return (skipcomment(cp + 1));
1199 } else if (*cp == '\0') {
1200 return (cp);
1201 } else {
1202 return (skipline(cp));
1203 }
1204 }
1205
1206 /*
1207 * Mark a line dirty and consume the rest of it, keeping track of the
1208 * lexical state.
1209 */
1210 static const char *
skipline(const char * cp)1211 skipline(const char *cp)
1212 {
1213 const char *pcp;
1214 if (*cp != '\0')
1215 linestate = LS_DIRTY;
1216 while (*cp != '\0') {
1217 cp = skipcomment(pcp = cp);
1218 if (pcp == cp)
1219 cp++;
1220 }
1221 return (cp);
1222 }
1223
1224 /*
1225 * Skip over comments, strings, and character literals and stop at the
1226 * next character position that is not whitespace. Between calls we keep
1227 * the comment state in the global variable incomment, and we also adjust
1228 * the global variable linestate when we see a newline.
1229 * XXX: doesn't cope with the buffer splitting inside a state transition.
1230 */
1231 static const char *
skipcomment(const char * cp)1232 skipcomment(const char *cp)
1233 {
1234 if (text || ignoring[depth]) {
1235 for (; isspace((unsigned char)*cp); cp++)
1236 if (*cp == '\n')
1237 linestate = LS_START;
1238 return (cp);
1239 }
1240 while (*cp != '\0')
1241 /* don't reset to LS_START after a line continuation */
1242 if (strncmp(cp, "\\\r\n", 3) == 0)
1243 cp += 3;
1244 else if (strncmp(cp, "\\\n", 2) == 0)
1245 cp += 2;
1246 else switch (incomment) {
1247 case NO_COMMENT:
1248 if (strncmp(cp, "/\\\r\n", 4) == 0) {
1249 incomment = STARTING_COMMENT;
1250 cp += 4;
1251 } else if (strncmp(cp, "/\\\n", 3) == 0) {
1252 incomment = STARTING_COMMENT;
1253 cp += 3;
1254 } else if (strncmp(cp, "/*", 2) == 0) {
1255 incomment = C_COMMENT;
1256 cp += 2;
1257 } else if (strncmp(cp, "//", 2) == 0) {
1258 incomment = CXX_COMMENT;
1259 cp += 2;
1260 } else if (strncmp(cp, "\'", 1) == 0) {
1261 incomment = CHAR_LITERAL;
1262 linestate = LS_DIRTY;
1263 cp += 1;
1264 } else if (strncmp(cp, "\"", 1) == 0) {
1265 incomment = STRING_LITERAL;
1266 linestate = LS_DIRTY;
1267 cp += 1;
1268 } else if (strncmp(cp, "\n", 1) == 0) {
1269 linestate = LS_START;
1270 cp += 1;
1271 } else if (strchr(" \r\t", *cp) != NULL) {
1272 cp += 1;
1273 } else
1274 return (cp);
1275 continue;
1276 case CXX_COMMENT:
1277 if (strncmp(cp, "\n", 1) == 0) {
1278 incomment = NO_COMMENT;
1279 linestate = LS_START;
1280 }
1281 cp += 1;
1282 continue;
1283 case CHAR_LITERAL:
1284 case STRING_LITERAL:
1285 if ((incomment == CHAR_LITERAL && cp[0] == '\'') ||
1286 (incomment == STRING_LITERAL && cp[0] == '\"')) {
1287 incomment = NO_COMMENT;
1288 cp += 1;
1289 } else if (cp[0] == '\\') {
1290 if (cp[1] == '\0')
1291 cp += 1;
1292 else
1293 cp += 2;
1294 } else if (strncmp(cp, "\n", 1) == 0) {
1295 if (incomment == CHAR_LITERAL)
1296 error("Unterminated char literal");
1297 else
1298 error("Unterminated string literal");
1299 } else
1300 cp += 1;
1301 continue;
1302 case C_COMMENT:
1303 if (strncmp(cp, "*\\\r\n", 4) == 0) {
1304 incomment = FINISHING_COMMENT;
1305 cp += 4;
1306 } else if (strncmp(cp, "*\\\n", 3) == 0) {
1307 incomment = FINISHING_COMMENT;
1308 cp += 3;
1309 } else if (strncmp(cp, "*/", 2) == 0) {
1310 incomment = NO_COMMENT;
1311 cp += 2;
1312 } else
1313 cp += 1;
1314 continue;
1315 case STARTING_COMMENT:
1316 if (*cp == '*') {
1317 incomment = C_COMMENT;
1318 cp += 1;
1319 } else if (*cp == '/') {
1320 incomment = CXX_COMMENT;
1321 cp += 1;
1322 } else {
1323 incomment = NO_COMMENT;
1324 linestate = LS_DIRTY;
1325 }
1326 continue;
1327 case FINISHING_COMMENT:
1328 if (*cp == '/') {
1329 incomment = NO_COMMENT;
1330 cp += 1;
1331 } else
1332 incomment = C_COMMENT;
1333 continue;
1334 default:
1335 abort(); /* bug */
1336 }
1337 return (cp);
1338 }
1339
1340 /*
1341 * Skip macro arguments.
1342 */
1343 static const char *
skipargs(const char * cp)1344 skipargs(const char *cp)
1345 {
1346 const char *ocp = cp;
1347 int level = 0;
1348 cp = skipcomment(cp);
1349 if (*cp != '(')
1350 return (cp);
1351 do {
1352 if (*cp == '(')
1353 level++;
1354 if (*cp == ')')
1355 level--;
1356 cp = skipcomment(cp+1);
1357 } while (level != 0 && *cp != '\0');
1358 if (level == 0)
1359 return (cp);
1360 else
1361 /* Rewind and re-detect the syntax error later. */
1362 return (ocp);
1363 }
1364
1365 /*
1366 * Skip over an identifier.
1367 */
1368 static const char *
skipsym(const char * cp)1369 skipsym(const char *cp)
1370 {
1371 while (!endsym(*cp))
1372 ++cp;
1373 return (cp);
1374 }
1375
1376 /*
1377 * Skip whitespace and take a copy of any following identifier.
1378 */
1379 static const char *
getsym(const char ** cpp)1380 getsym(const char **cpp)
1381 {
1382 const char *cp = *cpp, *sym;
1383
1384 cp = skipcomment(cp);
1385 cp = skipsym(sym = cp);
1386 if (cp == sym)
1387 return NULL;
1388 *cpp = cp;
1389 return (xstrdup(sym, cp));
1390 }
1391
1392 /*
1393 * Check that s (a symbol) matches the start of t, and that the
1394 * following character in t is not a symbol character. Returns a
1395 * pointer to the following character in t if there is a match,
1396 * otherwise NULL.
1397 */
1398 static const char *
matchsym(const char * s,const char * t)1399 matchsym(const char *s, const char *t)
1400 {
1401 while (*s != '\0' && *t != '\0')
1402 if (*s != *t)
1403 return (NULL);
1404 else
1405 ++s, ++t;
1406 if (*s == '\0' && endsym(*t))
1407 return(t);
1408 else
1409 return(NULL);
1410 }
1411
1412 /*
1413 * Look for the symbol in the symbol table. If it is found, we return
1414 * the symbol table index, else we return -1.
1415 */
1416 static int
findsym(const char ** strp)1417 findsym(const char **strp)
1418 {
1419 const char *str;
1420 int symind;
1421
1422 str = *strp;
1423 *strp = skipsym(str);
1424 if (symlist) {
1425 if (*strp == str)
1426 return (-1);
1427 if (symdepth && firstsym)
1428 printf("%s%3d", zerosyms ? "" : "\n", depth);
1429 firstsym = zerosyms = false;
1430 printf("%s%.*s%s",
1431 symdepth ? " " : "",
1432 (int)(*strp-str), str,
1433 symdepth ? "" : "\n");
1434 /* we don't care about the value of the symbol */
1435 return (0);
1436 }
1437 for (symind = 0; symind < nsyms; ++symind) {
1438 if (matchsym(symname[symind], str) != NULL) {
1439 debugsym("findsym", symind);
1440 return (symind);
1441 }
1442 }
1443 return (-1);
1444 }
1445
1446 /*
1447 * Resolve indirect symbol values to their final definitions.
1448 */
1449 static void
indirectsym(void)1450 indirectsym(void)
1451 {
1452 const char *cp;
1453 int changed, sym, ind;
1454
1455 do {
1456 changed = 0;
1457 for (sym = 0; sym < nsyms; ++sym) {
1458 if (value[sym] == NULL)
1459 continue;
1460 cp = value[sym];
1461 ind = findsym(&cp);
1462 if (ind == -1 || ind == sym ||
1463 *cp != '\0' ||
1464 value[ind] == NULL ||
1465 value[ind] == value[sym])
1466 continue;
1467 debugsym("indir...", sym);
1468 value[sym] = value[ind];
1469 debugsym("...ectsym", sym);
1470 changed++;
1471 }
1472 } while (changed);
1473 }
1474
1475 /*
1476 * Add a symbol to the symbol table, specified with the format sym=val
1477 */
1478 static void
addsym1(bool ignorethis,bool definethis,char * symval)1479 addsym1(bool ignorethis, bool definethis, char *symval)
1480 {
1481 const char *sym, *val;
1482
1483 sym = symval;
1484 val = skipsym(sym);
1485 if (definethis && *val == '=') {
1486 symval[val - sym] = '\0';
1487 val = val + 1;
1488 } else if (*val == '\0') {
1489 val = definethis ? "1" : NULL;
1490 } else {
1491 usage();
1492 }
1493 addsym2(ignorethis, sym, val);
1494 }
1495
1496 /*
1497 * Add a symbol to the symbol table.
1498 */
1499 static void
addsym2(bool ignorethis,const char * sym,const char * val)1500 addsym2(bool ignorethis, const char *sym, const char *val)
1501 {
1502 const char *cp = sym;
1503 int symind;
1504
1505 symind = findsym(&cp);
1506 if (symind < 0) {
1507 if (nsyms >= MAXSYMS)
1508 errx(2, "too many symbols");
1509 symind = nsyms++;
1510 }
1511 ignore[symind] = ignorethis;
1512 symname[symind] = sym;
1513 value[symind] = val;
1514 debugsym("addsym", symind);
1515 }
1516
1517 static void
debugsym(const char * why,int symind)1518 debugsym(const char *why, int symind)
1519 {
1520 debug("%s %s%c%s", why, symname[symind],
1521 value[symind] ? '=' : ' ',
1522 value[symind] ? value[symind] : "undef");
1523 }
1524
1525 /*
1526 * Add symbols to the symbol table from a file containing
1527 * #define and #undef preprocessor directives.
1528 */
1529 static void
defundefile(const char * fn)1530 defundefile(const char *fn)
1531 {
1532 filename = fn;
1533 input = fopen(fn, "rb");
1534 if (input == NULL)
1535 err(2, "can't open %s", fn);
1536 linenum = 0;
1537 while (defundef())
1538 ;
1539 if (ferror(input))
1540 err(2, "can't read %s", filename);
1541 else
1542 fclose(input);
1543 if (incomment)
1544 error("EOF in comment");
1545 }
1546
1547 /*
1548 * Read and process one #define or #undef directive
1549 */
1550 static bool
defundef(void)1551 defundef(void)
1552 {
1553 const char *cp, *kw, *sym, *val, *end;
1554
1555 cp = skiphash();
1556 if (cp == NULL)
1557 return (false);
1558 if (*cp == '\0')
1559 goto done;
1560 /* strip trailing whitespace, and do a fairly rough check to
1561 avoid unsupported multi-line preprocessor directives */
1562 end = cp + strlen(cp);
1563 while (end > tline && strchr(" \t\n\r", end[-1]) != NULL)
1564 --end;
1565 if (end > tline && end[-1] == '\\')
1566 Eioccc();
1567
1568 kw = cp;
1569 if ((cp = matchsym("define", kw)) != NULL) {
1570 sym = getsym(&cp);
1571 if (sym == NULL)
1572 error("Missing macro name in #define");
1573 if (*cp == '(') {
1574 val = "1";
1575 } else {
1576 cp = skipcomment(cp);
1577 val = (cp < end) ? xstrdup(cp, end) : "";
1578 }
1579 debug("#define");
1580 addsym2(false, sym, val);
1581 } else if ((cp = matchsym("undef", kw)) != NULL) {
1582 sym = getsym(&cp);
1583 if (sym == NULL)
1584 error("Missing macro name in #undef");
1585 cp = skipcomment(cp);
1586 debug("#undef");
1587 addsym2(false, sym, NULL);
1588 } else {
1589 error("Unrecognized preprocessor directive");
1590 }
1591 skipline(cp);
1592 done:
1593 debug("parser line %d state %s comment %s line", linenum,
1594 comment_name[incomment], linestate_name[linestate]);
1595 return (true);
1596 }
1597
1598 /*
1599 * Concatenate two strings into new memory, checking for failure.
1600 */
1601 static char *
astrcat(const char * s1,const char * s2)1602 astrcat(const char *s1, const char *s2)
1603 {
1604 char *s;
1605 int len;
1606 size_t size;
1607
1608 len = snprintf(NULL, 0, "%s%s", s1, s2);
1609 if (len < 0)
1610 err(2, "snprintf");
1611 size = (size_t)len + 1;
1612 s = (char *)malloc(size);
1613 if (s == NULL)
1614 err(2, "malloc");
1615 snprintf(s, size, "%s%s", s1, s2);
1616 return (s);
1617 }
1618
1619 /*
1620 * Duplicate a segment of a string, checking for failure.
1621 */
1622 static const char *
xstrdup(const char * start,const char * end)1623 xstrdup(const char *start, const char *end)
1624 {
1625 size_t n;
1626 char *s;
1627
1628 if (end < start) abort(); /* bug */
1629 n = (size_t)(end - start) + 1;
1630 s = malloc(n);
1631 if (s == NULL)
1632 err(2, "malloc");
1633 snprintf(s, n, "%s", start);
1634 return (s);
1635 }
1636
1637 /*
1638 * Diagnostics.
1639 */
1640 static void
debug(const char * msg,...)1641 debug(const char *msg, ...)
1642 {
1643 va_list ap;
1644
1645 if (debugging) {
1646 va_start(ap, msg);
1647 vwarnx(msg, ap);
1648 va_end(ap);
1649 }
1650 }
1651
1652 static void
error(const char * msg)1653 error(const char *msg)
1654 {
1655 if (depth == 0)
1656 warnx("%s: %d: %s", filename, linenum, msg);
1657 else
1658 warnx("%s: %d: %s (#if line %d depth %d)",
1659 filename, linenum, msg, stifline[depth], depth);
1660 closeio();
1661 errx(2, "Output may be truncated");
1662 }
1663