?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Travis CI is no more
[unifdef.git] / unifdef.c
blobdc145a24232fc50bf0341fe1e1916be933f77a1c
1 /*
2 * Copyright (c) 2002 - 2020 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 #include "version.h"
50 "@(#) $Author: Tony Finch (dot@dotat.at) $\n"
51 "@(#) $URL: http://dotat.at/prog/unifdef $\n"
52 ;
53
54 /* types of input lines: */
55 typedef enum {
56 LT_TRUEI, /* a true #if with ignore flag */
57 LT_FALSEI, /* a false #if with ignore flag */
58 LT_IF, /* an unknown #if */
59 LT_TRUE, /* a true #if */
60 LT_FALSE, /* a false #if */
61 LT_ELIF, /* an unknown #elif */
62 LT_ELTRUE, /* a true #elif */
63 LT_ELFALSE, /* a false #elif */
64 LT_ELSE, /* #else */
65 LT_ENDIF, /* #endif */
66 LT_DODGY, /* flag: directive is not on one line */
67 LT_DODGY_LAST = LT_DODGY + LT_ENDIF,
68 LT_PLAIN, /* ordinary line */
69 LT_EOF, /* end of file */
70 LT_ERROR, /* unevaluable #if */
71 LT_COUNT
72 } Linetype;
73
74 static char const * const linetype_name[] = {
75 "TRUEI", "FALSEI", "IF", "TRUE", "FALSE",
76 "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF",
77 "DODGY TRUEI", "DODGY FALSEI",
78 "DODGY IF", "DODGY TRUE", "DODGY FALSE",
79 "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE",
80 "DODGY ELSE", "DODGY ENDIF",
81 "PLAIN", "EOF", "ERROR"
82 };
83
84 #define linetype_if2elif(lt) ((Linetype)(lt - LT_IF + LT_ELIF))
85 #define linetype_2dodgy(lt) ((Linetype)(lt + LT_DODGY))
86
87 /* state of #if processing */
88 typedef enum {
89 IS_OUTSIDE,
90 IS_FALSE_PREFIX, /* false #if followed by false #elifs */
91 IS_TRUE_PREFIX, /* first non-false #(el)if is true */
92 IS_PASS_MIDDLE, /* first non-false #(el)if is unknown */
93 IS_FALSE_MIDDLE, /* a false #elif after a pass state */
94 IS_TRUE_MIDDLE, /* a true #elif after a pass state */
95 IS_PASS_ELSE, /* an else after a pass state */
96 IS_FALSE_ELSE, /* an else after a true state */
97 IS_TRUE_ELSE, /* an else after only false states */
98 IS_FALSE_TRAILER, /* #elifs after a true are false */
99 IS_COUNT
100 } Ifstate;
101
102 static char const * const ifstate_name[] = {
103 "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX",
104 "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE",
105 "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE",
106 "FALSE_TRAILER"
107 };
108
109 /* state of comment parser */
110 typedef enum {
111 NO_COMMENT = false, /* outside a comment */
112 C_COMMENT, /* in a comment like this one */
113 CXX_COMMENT, /* between // and end of line */
114 STARTING_COMMENT, /* just after slash-backslash-newline */
115 FINISHING_COMMENT, /* star-backslash-newline in a C comment */
116 CHAR_LITERAL, /* inside '' */
117 STRING_LITERAL, /* inside "" */
118 RAW_STRING_LITERAL /* inside R"()" */
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
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
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
437 cleantemp(void)
438 {
439 if (tempname != NULL)
440 remove(tempname);
441 }
442
443 /*
444 * Self-identification functions.
445 */
446
447 static 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
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
470 usage(void)
471 {
472 synopsis(stderr);
473 exit(2);
474 }
475
476 static 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 */
543 static void Eelif (void) { error("Inappropriate #elif"); }
544 static void Eelse (void) { error("Inappropriate #else"); }
545 static void Eendif(void) { error("Inappropriate #endif"); }
546 static void Eeof (void) { error("Premature EOF"); }
547 static void Eioccc(void) { error("Obfuscated preprocessor control line"); }
548 /* plain line handling */
549 static void print (void) { flushline(true); }
550 static void drop (void) { flushline(false); }
551 /* output lacks group's start line */
552 static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); }
553 static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); }
554 static void Selse (void) { drop(); state(IS_TRUE_ELSE); }
555 /* print/pass this block */
556 static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); }
557 static void Pelse (void) { print(); state(IS_PASS_ELSE); }
558 static void Pendif(void) { print(); unnest(); }
559 /* discard this block */
560 static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); }
561 static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); }
562 static void Delse (void) { drop(); state(IS_FALSE_ELSE); }
563 static void Dendif(void) { drop(); unnest(); }
564 /* first line of group */
565 static void Fdrop (void) { nest(); Dfalse(); }
566 static void Fpass (void) { nest(); Pelif(); }
567 static void Ftrue (void) { nest(); Strue(); }
568 static void Ffalse(void) { nest(); Sfalse(); }
569 /* variable pedantry for obfuscated lines */
570 static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); }
571 static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); }
572 static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); }
573 /* ignore comments in this block */
574 static void Idrop (void) { Fdrop(); ignoreon(); }
575 static void Itrue (void) { Ftrue(); ignoreon(); }
576 static void Ifalse(void) { Ffalse(); ignoreon(); }
577 /* modify this line */
578 static void Mpass (void) { memcpy(keyword, "if ", 4); Pelif(); }
579 static void Mtrue (void) { keywordedit("else"); state(IS_TRUE_MIDDLE); }
580 static void Melif (void) { keywordedit("endif"); state(IS_FALSE_TRAILER); }
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
633 ignoreoff(void)
634 {
635 if (depth == 0)
636 abort(); /* bug */
637 ignoring[depth] = ignoring[depth-1];
638 }
639 static void
640 ignoreon(void)
641 {
642 ignoring[depth] = true;
643 }
644 static void
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
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
663 unnest(void)
664 {
665 if (depth == 0)
666 abort(); /* bug */
667 depth -= 1;
668 }
669 static void
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
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
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
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
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
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
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 debug("parser insert newline at EOF", linenum);
853 strcpy(tline + len, newline);
854 cp += strlen(newline);
855 linestate = LS_START;
856 } else {
857 debug("parser concatenate dangling whitespace");
858 ++linenum;
859 cp = skipcomment(cp);
860 }
861 }
862 if (retval != LT_PLAIN && (wascomment || linestate != LS_START)) {
863 retval = linetype_2dodgy(retval);
864 linestate = LS_DIRTY;
865 }
866 done:
867 debug("parser line %d state %s comment %s line", linenum,
868 comment_name[incomment], linestate_name[linestate]);
869 return (retval);
870 }
871
872 /*
873 * These are the binary operators that are supported by the expression
874 * evaluator.
875 */
876 static Linetype op_strict(long *p, long v, Linetype at, Linetype bt) {
877 if(at == LT_IF || bt == LT_IF) return (LT_IF);
878 return (*p = v, v ? LT_TRUE : LT_FALSE);
879 }
880 static Linetype op_lt(long *p, Linetype at, long a, Linetype bt, long b) {
881 return op_strict(p, a < b, at, bt);
882 }
883 static Linetype op_gt(long *p, Linetype at, long a, Linetype bt, long b) {
884 return op_strict(p, a > b, at, bt);
885 }
886 static Linetype op_le(long *p, Linetype at, long a, Linetype bt, long b) {
887 return op_strict(p, a <= b, at, bt);
888 }
889 static Linetype op_ge(long *p, Linetype at, long a, Linetype bt, long b) {
890 return op_strict(p, a >= b, at, bt);
891 }
892 static Linetype op_eq(long *p, Linetype at, long a, Linetype bt, long b) {
893 return op_strict(p, a == b, at, bt);
894 }
895 static Linetype op_ne(long *p, Linetype at, long a, Linetype bt, long b) {
896 return op_strict(p, a != b, at, bt);
897 }
898 static Linetype op_or(long *p, Linetype at, long a, Linetype bt, long b) {
899 if (!strictlogic && (at == LT_TRUE || bt == LT_TRUE))
900 return (*p = 1, LT_TRUE);
901 return op_strict(p, a || b, at, bt);
902 }
903 static Linetype op_and(long *p, Linetype at, long a, Linetype bt, long b) {
904 if (!strictlogic && (at == LT_FALSE || bt == LT_FALSE))
905 return (*p = 0, LT_FALSE);
906 return op_strict(p, a && b, at, bt);
907 }
908 static Linetype op_blsh(long *p, Linetype at, long a, Linetype bt, long b) {
909 return op_strict(p, a << b, at, bt);
910 }
911 static Linetype op_brsh(long *p, Linetype at, long a, Linetype bt, long b) {
912 return op_strict(p, a >> b, at, bt);
913 }
914 static Linetype op_add(long *p, Linetype at, long a, Linetype bt, long b) {
915 return op_strict(p, a + b, at, bt);
916 }
917 static Linetype op_sub(long *p, Linetype at, long a, Linetype bt, long b) {
918 return op_strict(p, a - b, at, bt);
919 }
920 static Linetype op_mul(long *p, Linetype at, long a, Linetype bt, long b) {
921 return op_strict(p, a * b, at, bt);
922 }
923 static Linetype op_div(long *p, Linetype at, long a, Linetype bt, long b) {
924 if (bt != LT_TRUE) {
925 debug("eval division by zero");
926 return (LT_ERROR);
927 }
928 return op_strict(p, a / b, at, bt);
929 }
930 static Linetype op_mod(long *p, Linetype at, long a, Linetype bt, long b) {
931 return op_strict(p, a % b, at, bt);
932 }
933 static Linetype op_bor(long *p, Linetype at, long a, Linetype bt, long b) {
934 return op_strict(p, a | b, at, bt);
935 }
936 static Linetype op_bxor(long *p, Linetype at, long a, Linetype bt, long b) {
937 return op_strict(p, a ^ b, at, bt);
938 }
939 static Linetype op_band(long *p, Linetype at, long a, Linetype bt, long b) {
940 return op_strict(p, a & b, at, bt);
941 }
942
943 /*
944 * An evaluation function takes three arguments, as follows: (1) a pointer to
945 * an element of the precedence table which lists the operators at the current
946 * level of precedence; (2) a pointer to an integer which will receive the
947 * value of the expression; and (3) a pointer to a char* that points to the
948 * expression to be evaluated and that is updated to the end of the expression
949 * when evaluation is complete. The function returns LT_FALSE if the value of
950 * the expression is zero, LT_TRUE if it is non-zero, LT_IF if the expression
951 * depends on an unknown symbol, or LT_ERROR if there is a parse failure.
952 */
953 struct ops;
954
955 typedef Linetype eval_fn(const struct ops *, long *, const char **);
956
957 static eval_fn eval_table, eval_unary;
958
959 /*
960 * The precedence table. Expressions involving binary operators are evaluated
961 * in a table-driven way by eval_table. When it evaluates a subexpression it
962 * calls the inner function with its first argument pointing to the next
963 * element of the table. Innermost expressions have special non-table-driven
964 * handling.
965 *
966 * The stop characters help with lexical analysis: an operator is not
967 * recognized if it is followed by one of the stop characters because
968 * that would make it a different operator.
969 */
970 struct op {
971 const char *str;
972 Linetype (*fn)(long *, Linetype, long, Linetype, long);
973 const char *stop;
974 };
975 struct ops {
976 eval_fn *inner;
977 struct op op[5];
978 };
979 static const struct ops eval_ops[] = {
980 { eval_table, { { "||", op_or, NULL } } },
981 { eval_table, { { "&&", op_and, NULL } } },
982 { eval_table, { { "|", op_bor, "|" } } },
983 { eval_table, { { "^", op_bxor, NULL } } },
984 { eval_table, { { "&", op_band, "&" } } },
985 { eval_table, { { "==", op_eq, NULL },
986 { "!=", op_ne, NULL } } },
987 { eval_table, { { "<=", op_le, NULL },
988 { ">=", op_ge, NULL },
989 { "<", op_lt, "<=" },
990 { ">", op_gt, ">=" } } },
991 { eval_table, { { "<<", op_blsh, NULL },
992 { ">>", op_brsh, NULL } } },
993 { eval_table, { { "+", op_add, NULL },
994 { "-", op_sub, NULL } } },
995 { eval_unary, { { "*", op_mul, NULL },
996 { "/", op_div, NULL },
997 { "%", op_mod, NULL } } },
998 };
999
1000 /* Current operator precedence level */
1001 static long prec(const struct ops *ops)
1002 {
1003 return (ops - eval_ops);
1004 }
1005
1006 /*
1007 * Function for evaluating the innermost parts of expressions,
1008 * viz. !expr (expr) number defined(symbol) symbol
1009 * We reset the constexpr flag in the last two cases.
1010 */
1011 static Linetype
1012 eval_unary(const struct ops *ops, long *valp, const char **cpp)
1013 {
1014 const char *cp;
1015 char *ep;
1016 int sym;
1017 bool defparen;
1018 Linetype lt;
1019
1020 cp = skipcomment(*cpp);
1021 if (*cp == '!') {
1022 debug("eval%d !", prec(ops));
1023 cp++;
1024 lt = eval_unary(ops, valp, &cp);
1025 if (lt == LT_ERROR)
1026 return (LT_ERROR);
1027 if (lt != LT_IF) {
1028 *valp = !*valp;
1029 lt = *valp ? LT_TRUE : LT_FALSE;
1030 }
1031 } else if (*cp == '~') {
1032 debug("eval%d ~", prec(ops));
1033 cp++;
1034 lt = eval_unary(ops, valp, &cp);
1035 if (lt == LT_ERROR)
1036 return (LT_ERROR);
1037 if (lt != LT_IF) {
1038 *valp = ~(*valp);
1039 lt = *valp ? LT_TRUE : LT_FALSE;
1040 }
1041 } else if (*cp == '-') {
1042 debug("eval%d -", prec(ops));
1043 cp++;
1044 lt = eval_unary(ops, valp, &cp);
1045 if (lt == LT_ERROR)
1046 return (LT_ERROR);
1047 if (lt != LT_IF) {
1048 *valp = -(*valp);
1049 lt = *valp ? LT_TRUE : LT_FALSE;
1050 }
1051 } else if (*cp == '(') {
1052 cp++;
1053 debug("eval%d (", prec(ops));
1054 lt = eval_table(eval_ops, valp, &cp);
1055 if (lt == LT_ERROR)
1056 return (LT_ERROR);
1057 cp = skipcomment(cp);
1058 if (*cp++ != ')')
1059 return (LT_ERROR);
1060 } else if (isdigit((unsigned char)*cp)) {
1061 debug("eval%d number", prec(ops));
1062 *valp = strtol(cp, &ep, 0);
1063 if (ep == cp)
1064 return (LT_ERROR);
1065 lt = *valp ? LT_TRUE : LT_FALSE;
1066 cp = ep;
1067 } else if (matchsym("defined", cp) != NULL) {
1068 cp = skipcomment(cp+7);
1069 if (*cp == '(') {
1070 cp = skipcomment(cp+1);
1071 defparen = true;
1072 } else {
1073 defparen = false;
1074 }
1075 sym = findsym(&cp);
1076 cp = skipcomment(cp);
1077 if (defparen && *cp++ != ')') {
1078 debug("eval%d defined missing ')'", prec(ops));
1079 return (LT_ERROR);
1080 }
1081 if (sym < 0) {
1082 debug("eval%d defined unknown", prec(ops));
1083 lt = LT_IF;
1084 } else {
1085 debug("eval%d defined %s", prec(ops), symname[sym]);
1086 *valp = (value[sym] != NULL);
1087 lt = *valp ? LT_TRUE : LT_FALSE;
1088 }
1089 constexpr = false;
1090 } else if (!endsym(*cp)) {
1091 debug("eval%d symbol", prec(ops));
1092 sym = findsym(&cp);
1093 if (sym < 0) {
1094 lt = LT_IF;
1095 cp = skipargs(cp);
1096 } else if (value[sym] == NULL) {
1097 *valp = 0;
1098 lt = LT_FALSE;
1099 } else {
1100 *valp = strtol(value[sym], &ep, 0);
1101 if (*ep != '\0' || ep == value[sym])
1102 return (LT_ERROR);
1103 lt = *valp ? LT_TRUE : LT_FALSE;
1104 cp = skipargs(cp);
1105 }
1106 constexpr = false;
1107 } else {
1108 debug("eval%d bad expr", prec(ops));
1109 return (LT_ERROR);
1110 }
1111
1112 *cpp = cp;
1113 debug("eval%d = %d", prec(ops), *valp);
1114 return (lt);
1115 }
1116
1117 /*
1118 * Table-driven evaluation of binary operators.
1119 */
1120 static Linetype
1121 eval_table(const struct ops *ops, long *valp, const char **cpp)
1122 {
1123 const struct op *op;
1124 const char *cp;
1125 long val = 0;
1126 Linetype lt, rt;
1127
1128 debug("eval%d", prec(ops));
1129 cp = *cpp;
1130 lt = ops->inner(ops+1, valp, &cp);
1131 if (lt == LT_ERROR)
1132 return (LT_ERROR);
1133 for (;;) {
1134 cp = skipcomment(cp);
1135 for (op = ops->op; op->str != NULL; op++) {
1136 if (strncmp(cp, op->str, strlen(op->str)) == 0) {
1137 /* assume only one-char operators have stop chars */
1138 if (op->stop != NULL && cp[1] != '\0' &&
1139 strchr(op->stop, cp[1]) != NULL)
1140 continue;
1141 else
1142 break;
1143 }
1144 }
1145 if (op->str == NULL)
1146 break;
1147 cp += strlen(op->str);
1148 debug("eval%d %s", prec(ops), op->str);
1149 rt = ops->inner(ops+1, &val, &cp);
1150 if (rt == LT_ERROR)
1151 return (LT_ERROR);
1152 lt = op->fn(valp, lt, *valp, rt, val);
1153 }
1154
1155 *cpp = cp;
1156 debug("eval%d = %d", prec(ops), *valp);
1157 debug("eval%d lt = %s", prec(ops), linetype_name[lt]);
1158 return (lt);
1159 }
1160
1161 /*
1162 * Evaluate the expression on a #if or #elif line. If we can work out
1163 * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we
1164 * return just a generic LT_IF.
1165 */
1166 static Linetype
1167 ifeval(const char **cpp)
1168 {
1169 Linetype ret;
1170 long val = 0;
1171
1172 debug("eval %s", *cpp);
1173 constexpr = killconsts ? false : true;
1174 ret = eval_table(eval_ops, &val, cpp);
1175 debug("eval = %d", val);
1176 return (constexpr ? LT_IF : ret == LT_ERROR ? LT_IF : ret);
1177 }
1178
1179 /*
1180 * Read a line and examine its initial part to determine if it is a
1181 * preprocessor directive. Returns NULL on EOF, or a pointer to a
1182 * preprocessor directive name, or a pointer to the zero byte at the
1183 * end of the line.
1184 */
1185 static const char *
1186 skiphash(void)
1187 {
1188 const char *cp;
1189
1190 linenum++;
1191 if (fgets(tline, MAXLINE, input) == NULL) {
1192 if (ferror(input))
1193 err(2, "can't read %s", filename);
1194 else
1195 return (NULL);
1196 }
1197 cp = skipcomment(tline);
1198 if (linestate == LS_START && *cp == '#') {
1199 linestate = LS_HASH;
1200 return (skipcomment(cp + 1));
1201 } else if (*cp == '\0') {
1202 return (cp);
1203 } else {
1204 return (skipline(cp));
1205 }
1206 }
1207
1208 /*
1209 * Mark a line dirty and consume the rest of it, keeping track of the
1210 * lexical state.
1211 */
1212 static const char *
1213 skipline(const char *cp)
1214 {
1215 const char *pcp;
1216 if (*cp != '\0')
1217 linestate = LS_DIRTY;
1218 while (*cp != '\0') {
1219 cp = skipcomment(pcp = cp);
1220 if (pcp == cp)
1221 cp++;
1222 }
1223 return (cp);
1224 }
1225
1226 /*
1227 * Skip over comments, strings, and character literals and stop at the
1228 * next character position that is not whitespace. Between calls we keep
1229 * the comment state in the global variable incomment, and we also adjust
1230 * the global variable linestate when we see a newline.
1231 * XXX: doesn't cope with the buffer splitting inside a state transition.
1232 */
1233 static const char *
1234 skipcomment(const char *cp)
1235 {
1236 if (text || ignoring[depth]) {
1237 for (; isspace((unsigned char)*cp); cp++)
1238 if (*cp == '\n')
1239 linestate = LS_START;
1240 return (cp);
1241 }
1242 while (*cp != '\0')
1243 /* don't reset to LS_START after a line continuation */
1244 if (strncmp(cp, "\\\r\n", 3) == 0)
1245 cp += 3;
1246 else if (strncmp(cp, "\\\n", 2) == 0)
1247 cp += 2;
1248 else switch (incomment) {
1249 case NO_COMMENT:
1250 if (strncmp(cp, "/\\\r\n", 4) == 0) {
1251 incomment = STARTING_COMMENT;
1252 cp += 4;
1253 } else if (strncmp(cp, "/\\\n", 3) == 0) {
1254 incomment = STARTING_COMMENT;
1255 cp += 3;
1256 } else if (strncmp(cp, "/*", 2) == 0) {
1257 incomment = C_COMMENT;
1258 cp += 2;
1259 } else if (strncmp(cp, "//", 2) == 0) {
1260 incomment = CXX_COMMENT;
1261 cp += 2;
1262 } else if (strncmp(cp, "\'", 1) == 0) {
1263 incomment = CHAR_LITERAL;
1264 linestate = LS_DIRTY;
1265 cp += 1;
1266 } else if (strncmp(cp, "\"", 1) == 0) {
1267 incomment = STRING_LITERAL;
1268 linestate = LS_DIRTY;
1269 cp += 1;
1270 } else if (strncmp(cp, "R\"(", 3) == 0) {
1271 incomment = RAW_STRING_LITERAL;
1272 linestate = LS_DIRTY;
1273 cp += 3;
1274 } else if (strncmp(cp, "\n", 1) == 0) {
1275 linestate = LS_START;
1276 cp += 1;
1277 } else if (strchr(" \r\t", *cp) != NULL) {
1278 cp += 1;
1279 } else
1280 return (cp);
1281 continue;
1282 case CXX_COMMENT:
1283 if (strncmp(cp, "\n", 1) == 0) {
1284 incomment = NO_COMMENT;
1285 linestate = LS_START;
1286 }
1287 cp += 1;
1288 continue;
1289 case CHAR_LITERAL:
1290 case STRING_LITERAL:
1291 if ((incomment == CHAR_LITERAL && cp[0] == '\'') ||
1292 (incomment == STRING_LITERAL && cp[0] == '\"')) {
1293 incomment = NO_COMMENT;
1294 cp += 1;
1295 } else if (cp[0] == '\\') {
1296 if (cp[1] == '\0')
1297 cp += 1;
1298 else
1299 cp += 2;
1300 } else if (strncmp(cp, "\n", 1) == 0) {
1301 if (incomment == CHAR_LITERAL)
1302 error("Unterminated char literal");
1303 else
1304 error("Unterminated string literal");
1305 } else
1306 cp += 1;
1307 continue;
1308 case RAW_STRING_LITERAL:
1309 if (strncmp(cp, ")\"", 2) == 0) {
1310 incomment = NO_COMMENT;
1311 cp += 2;
1312 } else
1313 cp += 1;
1314 continue;
1315 case C_COMMENT:
1316 if (strncmp(cp, "*\\\r\n", 4) == 0) {
1317 incomment = FINISHING_COMMENT;
1318 cp += 4;
1319 } else if (strncmp(cp, "*\\\n", 3) == 0) {
1320 incomment = FINISHING_COMMENT;
1321 cp += 3;
1322 } else if (strncmp(cp, "*/", 2) == 0) {
1323 incomment = NO_COMMENT;
1324 cp += 2;
1325 } else
1326 cp += 1;
1327 continue;
1328 case STARTING_COMMENT:
1329 if (*cp == '*') {
1330 incomment = C_COMMENT;
1331 cp += 1;
1332 } else if (*cp == '/') {
1333 incomment = CXX_COMMENT;
1334 cp += 1;
1335 } else {
1336 incomment = NO_COMMENT;
1337 linestate = LS_DIRTY;
1338 }
1339 continue;
1340 case FINISHING_COMMENT:
1341 if (*cp == '/') {
1342 incomment = NO_COMMENT;
1343 cp += 1;
1344 } else
1345 incomment = C_COMMENT;
1346 continue;
1347 default:
1348 abort(); /* bug */
1349 }
1350 return (cp);
1351 }
1352
1353 /*
1354 * Skip macro arguments.
1355 */
1356 static const char *
1357 skipargs(const char *cp)
1358 {
1359 const char *ocp = cp;
1360 int level = 0;
1361 cp = skipcomment(cp);
1362 if (*cp != '(')
1363 return (cp);
1364 do {
1365 if (*cp == '(')
1366 level++;
1367 if (*cp == ')')
1368 level--;
1369 cp = skipcomment(cp+1);
1370 } while (level != 0 && *cp != '\0');
1371 if (level == 0)
1372 return (cp);
1373 else
1374 /* Rewind and re-detect the syntax error later. */
1375 return (ocp);
1376 }
1377
1378 /*
1379 * Skip over an identifier.
1380 */
1381 static const char *
1382 skipsym(const char *cp)
1383 {
1384 while (!endsym(*cp))
1385 ++cp;
1386 return (cp);
1387 }
1388
1389 /*
1390 * Skip whitespace and take a copy of any following identifier.
1391 */
1392 static const char *
1393 getsym(const char **cpp)
1394 {
1395 const char *cp = *cpp, *sym;
1396
1397 cp = skipcomment(cp);
1398 cp = skipsym(sym = cp);
1399 if (cp == sym)
1400 return NULL;
1401 *cpp = cp;
1402 return (xstrdup(sym, cp));
1403 }
1404
1405 /*
1406 * Check that s (a symbol) matches the start of t, and that the
1407 * following character in t is not a symbol character. Returns a
1408 * pointer to the following character in t if there is a match,
1409 * otherwise NULL.
1410 */
1411 static const char *
1412 matchsym(const char *s, const char *t)
1413 {
1414 while (*s != '\0' && *t != '\0')
1415 if (*s != *t)
1416 return (NULL);
1417 else
1418 ++s, ++t;
1419 if (*s == '\0' && endsym(*t))
1420 return(t);
1421 else
1422 return(NULL);
1423 }
1424
1425 /*
1426 * Look for the symbol in the symbol table. If it is found, we return
1427 * the symbol table index, else we return -1.
1428 */
1429 static int
1430 findsym(const char **strp)
1431 {
1432 const char *str;
1433 int symind;
1434
1435 str = *strp;
1436 *strp = skipsym(str);
1437 if (symlist) {
1438 if (*strp == str)
1439 return (-1);
1440 if (symdepth && firstsym)
1441 printf("%s%3d", zerosyms ? "" : "\n", depth);
1442 firstsym = zerosyms = false;
1443 printf("%s%.*s%s",
1444 symdepth ? " " : "",
1445 (int)(*strp-str), str,
1446 symdepth ? "" : "\n");
1447 /* we don't care about the value of the symbol */
1448 return (0);
1449 }
1450 for (symind = 0; symind < nsyms; ++symind) {
1451 if (matchsym(symname[symind], str) != NULL) {
1452 debugsym("findsym", symind);
1453 return (symind);
1454 }
1455 }
1456 return (-1);
1457 }
1458
1459 /*
1460 * Resolve indirect symbol values to their final definitions.
1461 */
1462 static void
1463 indirectsym(void)
1464 {
1465 const char *cp;
1466 int changed, sym, ind;
1467
1468 do {
1469 changed = 0;
1470 for (sym = 0; sym < nsyms; ++sym) {
1471 if (value[sym] == NULL)
1472 continue;
1473 cp = value[sym];
1474 ind = findsym(&cp);
1475 if (ind == -1 || ind == sym ||
1476 *cp != '\0' ||
1477 value[ind] == NULL ||
1478 value[ind] == value[sym])
1479 continue;
1480 debugsym("indir...", sym);
1481 value[sym] = value[ind];
1482 debugsym("...ectsym", sym);
1483 changed++;
1484 }
1485 } while (changed);
1486 }
1487
1488 /*
1489 * Add a symbol to the symbol table, specified with the format sym=val
1490 */
1491 static void
1492 addsym1(bool ignorethis, bool definethis, char *symval)
1493 {
1494 const char *sym, *val;
1495
1496 sym = symval;
1497 val = skipsym(sym);
1498 if (definethis && *val == '=') {
1499 symval[val - sym] = '\0';
1500 val = val + 1;
1501 } else if (*val == '\0') {
1502 val = definethis ? "1" : NULL;
1503 } else {
1504 usage();
1505 }
1506 addsym2(ignorethis, sym, val);
1507 }
1508
1509 /*
1510 * Add a symbol to the symbol table.
1511 */
1512 static void
1513 addsym2(bool ignorethis, const char *sym, const char *val)
1514 {
1515 const char *cp = sym;
1516 int symind;
1517
1518 symind = findsym(&cp);
1519 if (symind < 0) {
1520 if (nsyms >= MAXSYMS)
1521 errx(2, "too many symbols");
1522 symind = nsyms++;
1523 }
1524 ignore[symind] = ignorethis;
1525 symname[symind] = sym;
1526 value[symind] = val;
1527 debugsym("addsym", symind);
1528 }
1529
1530 static void
1531 debugsym(const char *why, int symind)
1532 {
1533 debug("%s %s%c%s", why, symname[symind],
1534 value[symind] ? '=' : ' ',
1535 value[symind] ? value[symind] : "undef");
1536 }
1537
1538 /*
1539 * Add symbols to the symbol table from a file containing
1540 * #define and #undef preprocessor directives.
1541 */
1542 static void
1543 defundefile(const char *fn)
1544 {
1545 filename = fn;
1546 input = fopen(fn, "rb");
1547 if (input == NULL)
1548 err(2, "can't open %s", fn);
1549 linenum = 0;
1550 while (defundef())
1551 ;
1552 if (ferror(input))
1553 err(2, "can't read %s", filename);
1554 else
1555 fclose(input);
1556 if (incomment)
1557 error("EOF in comment");
1558 }
1559
1560 /*
1561 * Read and process one #define or #undef directive
1562 */
1563 static bool
1564 defundef(void)
1565 {
1566 const char *cp, *kw, *sym, *val, *end;
1567
1568 cp = skiphash();
1569 if (cp == NULL)
1570 return (false);
1571 if (*cp == '\0')
1572 goto done;
1573 /* strip trailing whitespace, and do a fairly rough check to
1574 avoid unsupported multi-line preprocessor directives */
1575 end = cp + strlen(cp);
1576 while (end > tline && strchr(" \t\n\r", end[-1]) != NULL)
1577 --end;
1578 if (end > tline && end[-1] == '\\')
1579 Eioccc();
1580
1581 kw = cp;
1582 if ((cp = matchsym("define", kw)) != NULL) {
1583 sym = getsym(&cp);
1584 if (sym == NULL)
1585 error("Missing macro name in #define");
1586 if (*cp == '(') {
1587 val = "1";
1588 } else {
1589 cp = skipcomment(cp);
1590 val = (cp < end) ? xstrdup(cp, end) : "";
1591 }
1592 debug("#define");
1593 addsym2(false, sym, val);
1594 } else if ((cp = matchsym("undef", kw)) != NULL) {
1595 sym = getsym(&cp);
1596 if (sym == NULL)
1597 error("Missing macro name in #undef");
1598 cp = skipcomment(cp);
1599 debug("#undef");
1600 addsym2(false, sym, NULL);
1601 } else {
1602 error("Unrecognized preprocessor directive");
1603 }
1604 skipline(cp);
1605 done:
1606 debug("parser line %d state %s comment %s line", linenum,
1607 comment_name[incomment], linestate_name[linestate]);
1608 return (true);
1609 }
1610
1611 /*
1612 * Concatenate two strings into new memory, checking for failure.
1613 */
1614 static char *
1615 astrcat(const char *s1, const char *s2)
1616 {
1617 char *s;
1618 int len;
1619 size_t size;
1620
1621 len = snprintf(NULL, 0, "%s%s", s1, s2);
1622 if (len < 0)
1623 err(2, "snprintf");
1624 size = (size_t)len + 1;
1625 s = (char *)malloc(size);
1626 if (s == NULL)
1627 err(2, "malloc");
1628 snprintf(s, size, "%s%s", s1, s2);
1629 return (s);
1630 }
1631
1632 /*
1633 * Duplicate a segment of a string, checking for failure.
1634 */
1635 static const char *
1636 xstrdup(const char *start, const char *end)
1637 {
1638 size_t n;
1639 char *s;
1640
1641 if (end < start) abort(); /* bug */
1642 n = (size_t)(end - start) + 1;
1643 s = (char *)malloc(n);
1644 if (s == NULL)
1645 err(2, "malloc");
1646 snprintf(s, n, "%s", start);
1647 return (s);
1648 }
1649
1650 /*
1651 * Diagnostics.
1652 */
1653 static void
1654 debug(const char *msg, ...)
1655 {
1656 va_list ap;
1657
1658 if (debugging) {
1659 va_start(ap, msg);
1660 vwarnx(msg, ap);
1661 va_end(ap);
1662 }
1663 }
1664
1665 static void
1666 error(const char *msg)
1667 {
1668 if (depth == 0)
1669 warnx("%s: %d: %s", filename, linenum, msg);
1670 else
1671 warnx("%s: %d: %s (#if line %d depth %d)",
1672 filename, linenum, msg, stifline[depth], depth);
1673 closeio();
1674 errx(2, "Output may be truncated");
1675 }
lightweight C preprocessor