]>
Commit | Line | Data |
---|---|---|
01f1c879 | 1 | /* |
d8379ab1 | 2 | * Copyright (c) 2002 - 2009 Tony Finch <[email protected]> |
01f1c879 SR |
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 | ||
d8379ab1 TF |
26 | /* |
27 | * This code was derived from software contributed to Berkeley by Dave Yost. | |
28 | * It was rewritten to support ANSI C by Tony Finch. The original version | |
29 | * of unifdef carried the 4-clause BSD copyright licence. None of its code | |
30 | * remains in this version (though some of the names remain) so it now | |
31 | * carries a more liberal licence. | |
32 | * | |
33 | * The latest version is available from http://dotat.at/prog/unifdef | |
34 | */ | |
35 | ||
36 | static const char * const copyright[] = { | |
37 | "@(#) Copyright (c) 2002 - 2009 Tony Finch <[email protected]>\n", | |
38 | "$dotat: unifdef/unifdef.c,v 1.190 2009/11/27 17:21:26 fanf2 Exp $", | |
39 | }; | |
01f1c879 SR |
40 | |
41 | /* | |
42 | * unifdef - remove ifdef'ed lines | |
43 | * | |
44 | * Wishlist: | |
45 | * provide an option which will append the name of the | |
46 | * appropriate symbol after #else's and #endif's | |
47 | * provide an option which will check symbols after | |
48 | * #else's and #endif's to see that they match their | |
49 | * corresponding #ifdef or #ifndef | |
50 | * | |
51 | * The first two items above require better buffer handling, which would | |
52 | * also make it possible to handle all "dodgy" directives correctly. | |
53 | */ | |
54 | ||
55 | #include <ctype.h> | |
56 | #include <err.h> | |
57 | #include <stdarg.h> | |
58 | #include <stdbool.h> | |
59 | #include <stdio.h> | |
60 | #include <stdlib.h> | |
61 | #include <string.h> | |
62 | #include <unistd.h> | |
63 | ||
01f1c879 SR |
64 | /* types of input lines: */ |
65 | typedef enum { | |
66 | LT_TRUEI, /* a true #if with ignore flag */ | |
67 | LT_FALSEI, /* a false #if with ignore flag */ | |
68 | LT_IF, /* an unknown #if */ | |
69 | LT_TRUE, /* a true #if */ | |
70 | LT_FALSE, /* a false #if */ | |
71 | LT_ELIF, /* an unknown #elif */ | |
72 | LT_ELTRUE, /* a true #elif */ | |
73 | LT_ELFALSE, /* a false #elif */ | |
74 | LT_ELSE, /* #else */ | |
75 | LT_ENDIF, /* #endif */ | |
76 | LT_DODGY, /* flag: directive is not on one line */ | |
77 | LT_DODGY_LAST = LT_DODGY + LT_ENDIF, | |
78 | LT_PLAIN, /* ordinary line */ | |
79 | LT_EOF, /* end of file */ | |
d8379ab1 | 80 | LT_ERROR, /* unevaluable #if */ |
01f1c879 SR |
81 | LT_COUNT |
82 | } Linetype; | |
83 | ||
84 | static char const * const linetype_name[] = { | |
85 | "TRUEI", "FALSEI", "IF", "TRUE", "FALSE", | |
86 | "ELIF", "ELTRUE", "ELFALSE", "ELSE", "ENDIF", | |
87 | "DODGY TRUEI", "DODGY FALSEI", | |
88 | "DODGY IF", "DODGY TRUE", "DODGY FALSE", | |
89 | "DODGY ELIF", "DODGY ELTRUE", "DODGY ELFALSE", | |
90 | "DODGY ELSE", "DODGY ENDIF", | |
d8379ab1 | 91 | "PLAIN", "EOF", "ERROR" |
01f1c879 SR |
92 | }; |
93 | ||
94 | /* state of #if processing */ | |
95 | typedef enum { | |
96 | IS_OUTSIDE, | |
97 | IS_FALSE_PREFIX, /* false #if followed by false #elifs */ | |
98 | IS_TRUE_PREFIX, /* first non-false #(el)if is true */ | |
99 | IS_PASS_MIDDLE, /* first non-false #(el)if is unknown */ | |
100 | IS_FALSE_MIDDLE, /* a false #elif after a pass state */ | |
101 | IS_TRUE_MIDDLE, /* a true #elif after a pass state */ | |
102 | IS_PASS_ELSE, /* an else after a pass state */ | |
103 | IS_FALSE_ELSE, /* an else after a true state */ | |
104 | IS_TRUE_ELSE, /* an else after only false states */ | |
105 | IS_FALSE_TRAILER, /* #elifs after a true are false */ | |
106 | IS_COUNT | |
107 | } Ifstate; | |
108 | ||
109 | static char const * const ifstate_name[] = { | |
110 | "OUTSIDE", "FALSE_PREFIX", "TRUE_PREFIX", | |
111 | "PASS_MIDDLE", "FALSE_MIDDLE", "TRUE_MIDDLE", | |
112 | "PASS_ELSE", "FALSE_ELSE", "TRUE_ELSE", | |
113 | "FALSE_TRAILER" | |
114 | }; | |
115 | ||
116 | /* state of comment parser */ | |
117 | typedef enum { | |
118 | NO_COMMENT = false, /* outside a comment */ | |
119 | C_COMMENT, /* in a comment like this one */ | |
120 | CXX_COMMENT, /* between // and end of line */ | |
121 | STARTING_COMMENT, /* just after slash-backslash-newline */ | |
122 | FINISHING_COMMENT, /* star-backslash-newline in a C comment */ | |
123 | CHAR_LITERAL, /* inside '' */ | |
124 | STRING_LITERAL /* inside "" */ | |
125 | } Comment_state; | |
126 | ||
127 | static char const * const comment_name[] = { | |
128 | "NO", "C", "CXX", "STARTING", "FINISHING", "CHAR", "STRING" | |
129 | }; | |
130 | ||
131 | /* state of preprocessor line parser */ | |
132 | typedef enum { | |
133 | LS_START, /* only space and comments on this line */ | |
134 | LS_HASH, /* only space, comments, and a hash */ | |
135 | LS_DIRTY /* this line can't be a preprocessor line */ | |
136 | } Line_state; | |
137 | ||
138 | static char const * const linestate_name[] = { | |
139 | "START", "HASH", "DIRTY" | |
140 | }; | |
141 | ||
142 | /* | |
143 | * Minimum translation limits from ISO/IEC 9899:1999 5.2.4.1 | |
144 | */ | |
145 | #define MAXDEPTH 64 /* maximum #if nesting */ | |
146 | #define MAXLINE 4096 /* maximum length of line */ | |
147 | #define MAXSYMS 4096 /* maximum number of symbols */ | |
148 | ||
149 | /* | |
150 | * Sometimes when editing a keyword the replacement text is longer, so | |
151 | * we leave some space at the end of the tline buffer to accommodate this. | |
152 | */ | |
153 | #define EDITSLOP 10 | |
154 | ||
155 | /* | |
156 | * Globals. | |
157 | */ | |
158 | ||
d8379ab1 TF |
159 | static bool compblank; /* -B: compress blank lines */ |
160 | static bool lnblank; /* -b: blank deleted lines */ | |
01f1c879 SR |
161 | static bool complement; /* -c: do the complement */ |
162 | static bool debugging; /* -d: debugging reports */ | |
163 | static bool iocccok; /* -e: fewer IOCCC errors */ | |
d8379ab1 | 164 | static bool strictlogic; /* -K: keep ambiguous #ifs */ |
01f1c879 | 165 | static bool killconsts; /* -k: eval constant #ifs */ |
01f1c879 SR |
166 | static bool lnnum; /* -n: add #line directives */ |
167 | static bool symlist; /* -s: output symbol list */ | |
168 | static bool text; /* -t: this is a text file */ | |
169 | ||
170 | static const char *symname[MAXSYMS]; /* symbol name */ | |
171 | static const char *value[MAXSYMS]; /* -Dsym=value */ | |
172 | static bool ignore[MAXSYMS]; /* -iDsym or -iUsym */ | |
173 | static int nsyms; /* number of symbols */ | |
174 | ||
175 | static FILE *input; /* input file pointer */ | |
176 | static const char *filename; /* input file name */ | |
177 | static int linenum; /* current line number */ | |
178 | ||
179 | static char tline[MAXLINE+EDITSLOP];/* input buffer plus space */ | |
180 | static char *keyword; /* used for editing #elif's */ | |
181 | ||
182 | static Comment_state incomment; /* comment parser state */ | |
183 | static Line_state linestate; /* #if line parser state */ | |
184 | static Ifstate ifstate[MAXDEPTH]; /* #if processor state */ | |
185 | static bool ignoring[MAXDEPTH]; /* ignore comments state */ | |
186 | static int stifline[MAXDEPTH]; /* start of current #if */ | |
187 | static int depth; /* current #if nesting */ | |
188 | static int delcount; /* count of deleted lines */ | |
d8379ab1 TF |
189 | static unsigned blankcount; /* count of blank lines */ |
190 | static unsigned blankmax; /* maximum recent blankcount */ | |
191 | static bool constexpr; /* constant #if expression */ | |
01f1c879 SR |
192 | |
193 | static int exitstat; /* program exit status */ | |
194 | ||
195 | static void addsym(bool, bool, char *); | |
196 | static void debug(const char *, ...); | |
197 | static void done(void); | |
198 | static void error(const char *); | |
199 | static int findsym(const char *); | |
200 | static void flushline(bool); | |
d8379ab1 | 201 | static Linetype parseline(void); |
01f1c879 SR |
202 | static Linetype ifeval(const char **); |
203 | static void ignoreoff(void); | |
204 | static void ignoreon(void); | |
205 | static void keywordedit(const char *); | |
206 | static void nest(void); | |
207 | static void process(void); | |
d8379ab1 | 208 | static const char *skipargs(const char *); |
01f1c879 SR |
209 | static const char *skipcomment(const char *); |
210 | static const char *skipsym(const char *); | |
211 | static void state(Ifstate); | |
212 | static int strlcmp(const char *, const char *, size_t); | |
213 | static void unnest(void); | |
214 | static void usage(void); | |
215 | ||
d8379ab1 | 216 | #define endsym(c) (!isalnum((unsigned char)c) && c != '_') |
01f1c879 SR |
217 | |
218 | /* | |
219 | * The main program. | |
220 | */ | |
221 | int | |
222 | main(int argc, char *argv[]) | |
223 | { | |
224 | int opt; | |
225 | ||
d8379ab1 | 226 | while ((opt = getopt(argc, argv, "i:D:U:I:BbcdeKklnst")) != -1) |
01f1c879 SR |
227 | switch (opt) { |
228 | case 'i': /* treat stuff controlled by these symbols as text */ | |
229 | /* | |
230 | * For strict backwards-compatibility the U or D | |
231 | * should be immediately after the -i but it doesn't | |
232 | * matter much if we relax that requirement. | |
233 | */ | |
234 | opt = *optarg++; | |
235 | if (opt == 'D') | |
236 | addsym(true, true, optarg); | |
237 | else if (opt == 'U') | |
238 | addsym(true, false, optarg); | |
239 | else | |
240 | usage(); | |
241 | break; | |
242 | case 'D': /* define a symbol */ | |
243 | addsym(false, true, optarg); | |
244 | break; | |
245 | case 'U': /* undef a symbol */ | |
246 | addsym(false, false, optarg); | |
247 | break; | |
248 | case 'I': | |
249 | /* no-op for compatibility with cpp */ | |
250 | break; | |
d8379ab1 TF |
251 | case 'B': /* compress blank lines around removed section */ |
252 | compblank = true; | |
253 | break; | |
254 | case 'b': /* blank deleted lines instead of omitting them */ | |
255 | case 'l': /* backwards compatibility */ | |
256 | lnblank = true; | |
257 | break; | |
01f1c879 SR |
258 | case 'c': /* treat -D as -U and vice versa */ |
259 | complement = true; | |
260 | break; | |
261 | case 'd': | |
262 | debugging = true; | |
263 | break; | |
264 | case 'e': /* fewer errors from dodgy lines */ | |
265 | iocccok = true; | |
266 | break; | |
d8379ab1 TF |
267 | case 'K': /* keep ambiguous #ifs */ |
268 | strictlogic = true; | |
269 | break; | |
01f1c879 SR |
270 | case 'k': /* process constant #ifs */ |
271 | killconsts = true; | |
272 | break; | |
01f1c879 SR |
273 | case 'n': /* add #line directive after deleted lines */ |
274 | lnnum = true; | |
275 | break; | |
276 | case 's': /* only output list of symbols that control #ifs */ | |
277 | symlist = true; | |
278 | break; | |
279 | case 't': /* don't parse C comments */ | |
280 | text = true; | |
281 | break; | |
282 | default: | |
283 | usage(); | |
284 | } | |
285 | argc -= optind; | |
286 | argv += optind; | |
d8379ab1 TF |
287 | if (compblank && lnblank) |
288 | errx(2, "-B and -b are mutually exclusive"); | |
01f1c879 SR |
289 | if (argc > 1) { |
290 | errx(2, "can only do one file"); | |
291 | } else if (argc == 1 && strcmp(*argv, "-") != 0) { | |
292 | filename = *argv; | |
293 | input = fopen(filename, "r"); | |
294 | if (input == NULL) | |
295 | err(2, "can't open %s", filename); | |
296 | } else { | |
297 | filename = "[stdin]"; | |
298 | input = stdin; | |
299 | } | |
300 | process(); | |
301 | abort(); /* bug */ | |
302 | } | |
303 | ||
304 | static void | |
305 | usage(void) | |
306 | { | |
d8379ab1 | 307 | fprintf(stderr, "usage: unifdef [-BbcdeKknst] [-Ipath]" |
01f1c879 SR |
308 | " [-Dsym[=val]] [-Usym] [-iDsym[=val]] [-iUsym] ... [file]\n"); |
309 | exit(2); | |
310 | } | |
311 | ||
312 | /* | |
313 | * A state transition function alters the global #if processing state | |
314 | * in a particular way. The table below is indexed by the current | |
315 | * processing state and the type of the current line. | |
316 | * | |
317 | * Nesting is handled by keeping a stack of states; some transition | |
318 | * functions increase or decrease the depth. They also maintain the | |
319 | * ignore state on a stack. In some complicated cases they have to | |
320 | * alter the preprocessor directive, as follows. | |
321 | * | |
322 | * When we have processed a group that starts off with a known-false | |
323 | * #if/#elif sequence (which has therefore been deleted) followed by a | |
324 | * #elif that we don't understand and therefore must keep, we edit the | |
325 | * latter into a #if to keep the nesting correct. | |
326 | * | |
327 | * When we find a true #elif in a group, the following block will | |
328 | * always be kept and the rest of the sequence after the next #elif or | |
329 | * #else will be discarded. We edit the #elif into a #else and the | |
330 | * following directive to #endif since this has the desired behaviour. | |
331 | * | |
332 | * "Dodgy" directives are split across multiple lines, the most common | |
333 | * example being a multi-line comment hanging off the right of the | |
334 | * directive. We can handle them correctly only if there is no change | |
335 | * from printing to dropping (or vice versa) caused by that directive. | |
336 | * If the directive is the first of a group we have a choice between | |
337 | * failing with an error, or passing it through unchanged instead of | |
338 | * evaluating it. The latter is not the default to avoid questions from | |
339 | * users about unifdef unexpectedly leaving behind preprocessor directives. | |
340 | */ | |
341 | typedef void state_fn(void); | |
342 | ||
343 | /* report an error */ | |
344 | static void Eelif (void) { error("Inappropriate #elif"); } | |
345 | static void Eelse (void) { error("Inappropriate #else"); } | |
346 | static void Eendif(void) { error("Inappropriate #endif"); } | |
347 | static void Eeof (void) { error("Premature EOF"); } | |
348 | static void Eioccc(void) { error("Obfuscated preprocessor control line"); } | |
349 | /* plain line handling */ | |
350 | static void print (void) { flushline(true); } | |
351 | static void drop (void) { flushline(false); } | |
352 | /* output lacks group's start line */ | |
353 | static void Strue (void) { drop(); ignoreoff(); state(IS_TRUE_PREFIX); } | |
354 | static void Sfalse(void) { drop(); ignoreoff(); state(IS_FALSE_PREFIX); } | |
355 | static void Selse (void) { drop(); state(IS_TRUE_ELSE); } | |
356 | /* print/pass this block */ | |
357 | static void Pelif (void) { print(); ignoreoff(); state(IS_PASS_MIDDLE); } | |
358 | static void Pelse (void) { print(); state(IS_PASS_ELSE); } | |
359 | static void Pendif(void) { print(); unnest(); } | |
360 | /* discard this block */ | |
361 | static void Dfalse(void) { drop(); ignoreoff(); state(IS_FALSE_TRAILER); } | |
362 | static void Delif (void) { drop(); ignoreoff(); state(IS_FALSE_MIDDLE); } | |
363 | static void Delse (void) { drop(); state(IS_FALSE_ELSE); } | |
364 | static void Dendif(void) { drop(); unnest(); } | |
365 | /* first line of group */ | |
366 | static void Fdrop (void) { nest(); Dfalse(); } | |
367 | static void Fpass (void) { nest(); Pelif(); } | |
368 | static void Ftrue (void) { nest(); Strue(); } | |
369 | static void Ffalse(void) { nest(); Sfalse(); } | |
370 | /* variable pedantry for obfuscated lines */ | |
371 | static void Oiffy (void) { if (!iocccok) Eioccc(); Fpass(); ignoreon(); } | |
372 | static void Oif (void) { if (!iocccok) Eioccc(); Fpass(); } | |
373 | static void Oelif (void) { if (!iocccok) Eioccc(); Pelif(); } | |
374 | /* ignore comments in this block */ | |
375 | static void Idrop (void) { Fdrop(); ignoreon(); } | |
376 | static void Itrue (void) { Ftrue(); ignoreon(); } | |
377 | static void Ifalse(void) { Ffalse(); ignoreon(); } | |
378 | /* edit this line */ | |
379 | static void Mpass (void) { strncpy(keyword, "if ", 4); Pelif(); } | |
380 | static void Mtrue (void) { keywordedit("else\n"); state(IS_TRUE_MIDDLE); } | |
381 | static void Melif (void) { keywordedit("endif\n"); state(IS_FALSE_TRAILER); } | |
382 | static void Melse (void) { keywordedit("endif\n"); state(IS_FALSE_ELSE); } | |
383 | ||
384 | static state_fn * const trans_table[IS_COUNT][LT_COUNT] = { | |
385 | /* IS_OUTSIDE */ | |
386 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Eendif, | |
387 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eendif, | |
d8379ab1 | 388 | print, done, abort }, |
01f1c879 SR |
389 | /* IS_FALSE_PREFIX */ |
390 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Strue, Sfalse,Selse, Dendif, | |
391 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Mpass, Eioccc,Eioccc,Eioccc,Eioccc, | |
d8379ab1 | 392 | drop, Eeof, abort }, |
01f1c879 SR |
393 | /* IS_TRUE_PREFIX */ |
394 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Dfalse,Dfalse,Dfalse,Delse, Dendif, | |
395 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc, | |
d8379ab1 | 396 | print, Eeof, abort }, |
01f1c879 SR |
397 | /* IS_PASS_MIDDLE */ |
398 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Pelif, Mtrue, Delif, Pelse, Pendif, | |
399 | Oiffy, Oiffy, Fpass, Oif, Oif, Pelif, Oelif, Oelif, Pelse, Pendif, | |
d8379ab1 | 400 | print, Eeof, abort }, |
01f1c879 SR |
401 | /* IS_FALSE_MIDDLE */ |
402 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Pelif, Mtrue, Delif, Pelse, Pendif, | |
403 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eioccc,Eioccc,Eioccc,Eioccc,Eioccc, | |
d8379ab1 | 404 | drop, Eeof, abort }, |
01f1c879 SR |
405 | /* IS_TRUE_MIDDLE */ |
406 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Melif, Melif, Melif, Melse, Pendif, | |
407 | Oiffy, Oiffy, Fpass, Oif, Oif, Eioccc,Eioccc,Eioccc,Eioccc,Pendif, | |
d8379ab1 | 408 | print, Eeof, abort }, |
01f1c879 SR |
409 | /* IS_PASS_ELSE */ |
410 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Pendif, | |
411 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Pendif, | |
d8379ab1 | 412 | print, Eeof, abort }, |
01f1c879 SR |
413 | /* IS_FALSE_ELSE */ |
414 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Dendif, | |
415 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Eelif, Eelif, Eelif, Eelse, Eioccc, | |
d8379ab1 | 416 | drop, Eeof, abort }, |
01f1c879 SR |
417 | /* IS_TRUE_ELSE */ |
418 | { Itrue, Ifalse,Fpass, Ftrue, Ffalse,Eelif, Eelif, Eelif, Eelse, Dendif, | |
419 | Oiffy, Oiffy, Fpass, Oif, Oif, Eelif, Eelif, Eelif, Eelse, Eioccc, | |
d8379ab1 | 420 | print, Eeof, abort }, |
01f1c879 SR |
421 | /* IS_FALSE_TRAILER */ |
422 | { Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Dendif, | |
423 | Idrop, Idrop, Fdrop, Fdrop, Fdrop, Dfalse,Dfalse,Dfalse,Delse, Eioccc, | |
d8379ab1 | 424 | drop, Eeof, abort } |
01f1c879 SR |
425 | /*TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF |
426 | TRUEI FALSEI IF TRUE FALSE ELIF ELTRUE ELFALSE ELSE ENDIF (DODGY) | |
d8379ab1 | 427 | PLAIN EOF ERROR */ |
01f1c879 SR |
428 | }; |
429 | ||
430 | /* | |
431 | * State machine utility functions | |
432 | */ | |
433 | static void | |
434 | done(void) | |
435 | { | |
436 | if (incomment) | |
437 | error("EOF in comment"); | |
438 | exit(exitstat); | |
439 | } | |
440 | static void | |
441 | ignoreoff(void) | |
442 | { | |
443 | if (depth == 0) | |
444 | abort(); /* bug */ | |
445 | ignoring[depth] = ignoring[depth-1]; | |
446 | } | |
447 | static void | |
448 | ignoreon(void) | |
449 | { | |
450 | ignoring[depth] = true; | |
451 | } | |
452 | static void | |
453 | keywordedit(const char *replacement) | |
454 | { | |
14a036d2 SR |
455 | size_t size = tline + sizeof(tline) - keyword; |
456 | char *dst = keyword; | |
457 | const char *src = replacement; | |
458 | if (size != 0) { | |
459 | while ((--size != 0) && (*src != '\0')) | |
460 | *dst++ = *src++; | |
461 | *dst = '\0'; | |
462 | } | |
01f1c879 SR |
463 | print(); |
464 | } | |
465 | static void | |
466 | nest(void) | |
467 | { | |
d8379ab1 TF |
468 | if (depth > MAXDEPTH-1) |
469 | abort(); /* bug */ | |
470 | if (depth == MAXDEPTH-1) | |
01f1c879 | 471 | error("Too many levels of nesting"); |
d8379ab1 | 472 | depth += 1; |
01f1c879 SR |
473 | stifline[depth] = linenum; |
474 | } | |
475 | static void | |
476 | unnest(void) | |
477 | { | |
478 | if (depth == 0) | |
479 | abort(); /* bug */ | |
480 | depth -= 1; | |
481 | } | |
482 | static void | |
483 | state(Ifstate is) | |
484 | { | |
485 | ifstate[depth] = is; | |
486 | } | |
487 | ||
488 | /* | |
489 | * Write a line to the output or not, according to command line options. | |
490 | */ | |
491 | static void | |
492 | flushline(bool keep) | |
493 | { | |
494 | if (symlist) | |
495 | return; | |
496 | if (keep ^ complement) { | |
d8379ab1 TF |
497 | bool blankline = tline[strspn(tline, " \t\n")] == '\0'; |
498 | if (blankline && compblank && blankcount != blankmax) { | |
499 | delcount += 1; | |
500 | blankcount += 1; | |
501 | } else { | |
502 | if (lnnum && delcount > 0) | |
503 | printf("#line %d\n", linenum); | |
504 | fputs(tline, stdout); | |
505 | delcount = 0; | |
506 | blankmax = blankcount = blankline ? blankcount + 1 : 0; | |
507 | } | |
01f1c879 SR |
508 | } else { |
509 | if (lnblank) | |
510 | putc('\n', stdout); | |
511 | exitstat = 1; | |
512 | delcount += 1; | |
d8379ab1 | 513 | blankcount = 0; |
01f1c879 SR |
514 | } |
515 | } | |
516 | ||
517 | /* | |
518 | * The driver for the state machine. | |
519 | */ | |
520 | static void | |
521 | process(void) | |
522 | { | |
523 | Linetype lineval; | |
524 | ||
d8379ab1 TF |
525 | /* When compressing blank lines, act as if the file |
526 | is preceded by a large number of blank lines. */ | |
527 | blankmax = blankcount = 1000; | |
01f1c879 SR |
528 | for (;;) { |
529 | linenum++; | |
d8379ab1 | 530 | lineval = parseline(); |
01f1c879 SR |
531 | trans_table[ifstate[depth]][lineval](); |
532 | debug("process %s -> %s depth %d", | |
533 | linetype_name[lineval], | |
534 | ifstate_name[ifstate[depth]], depth); | |
535 | } | |
536 | } | |
537 | ||
538 | /* | |
539 | * Parse a line and determine its type. We keep the preprocessor line | |
540 | * parser state between calls in the global variable linestate, with | |
541 | * help from skipcomment(). | |
542 | */ | |
543 | static Linetype | |
d8379ab1 | 544 | parseline(void) |
01f1c879 SR |
545 | { |
546 | const char *cp; | |
547 | int cursym; | |
548 | int kwlen; | |
549 | Linetype retval; | |
550 | Comment_state wascomment; | |
551 | ||
552 | if (fgets(tline, MAXLINE, input) == NULL) | |
553 | return (LT_EOF); | |
554 | retval = LT_PLAIN; | |
555 | wascomment = incomment; | |
556 | cp = skipcomment(tline); | |
557 | if (linestate == LS_START) { | |
558 | if (*cp == '#') { | |
559 | linestate = LS_HASH; | |
560 | cp = skipcomment(cp + 1); | |
561 | } else if (*cp != '\0') | |
562 | linestate = LS_DIRTY; | |
563 | } | |
564 | if (!incomment && linestate == LS_HASH) { | |
565 | keyword = tline + (cp - tline); | |
566 | cp = skipsym(cp); | |
567 | kwlen = cp - keyword; | |
568 | /* no way can we deal with a continuation inside a keyword */ | |
569 | if (strncmp(cp, "\\\n", 2) == 0) | |
570 | Eioccc(); | |
571 | if (strlcmp("ifdef", keyword, kwlen) == 0 || | |
572 | strlcmp("ifndef", keyword, kwlen) == 0) { | |
573 | cp = skipcomment(cp); | |
574 | if ((cursym = findsym(cp)) < 0) | |
575 | retval = LT_IF; | |
576 | else { | |
577 | retval = (keyword[2] == 'n') | |
578 | ? LT_FALSE : LT_TRUE; | |
579 | if (value[cursym] == NULL) | |
580 | retval = (retval == LT_TRUE) | |
581 | ? LT_FALSE : LT_TRUE; | |
582 | if (ignore[cursym]) | |
583 | retval = (retval == LT_TRUE) | |
584 | ? LT_TRUEI : LT_FALSEI; | |
585 | } | |
586 | cp = skipsym(cp); | |
587 | } else if (strlcmp("if", keyword, kwlen) == 0) | |
588 | retval = ifeval(&cp); | |
589 | else if (strlcmp("elif", keyword, kwlen) == 0) | |
590 | retval = ifeval(&cp) - LT_IF + LT_ELIF; | |
591 | else if (strlcmp("else", keyword, kwlen) == 0) | |
592 | retval = LT_ELSE; | |
593 | else if (strlcmp("endif", keyword, kwlen) == 0) | |
594 | retval = LT_ENDIF; | |
595 | else { | |
596 | linestate = LS_DIRTY; | |
597 | retval = LT_PLAIN; | |
598 | } | |
599 | cp = skipcomment(cp); | |
600 | if (*cp != '\0') { | |
601 | linestate = LS_DIRTY; | |
602 | if (retval == LT_TRUE || retval == LT_FALSE || | |
603 | retval == LT_TRUEI || retval == LT_FALSEI) | |
604 | retval = LT_IF; | |
605 | if (retval == LT_ELTRUE || retval == LT_ELFALSE) | |
606 | retval = LT_ELIF; | |
607 | } | |
608 | if (retval != LT_PLAIN && (wascomment || incomment)) { | |
609 | retval += LT_DODGY; | |
610 | if (incomment) | |
611 | linestate = LS_DIRTY; | |
612 | } | |
d8379ab1 TF |
613 | /* skipcomment normally changes the state, except |
614 | if the last line of the file lacks a newline, or | |
615 | if there is too much whitespace in a directive */ | |
616 | if (linestate == LS_HASH) { | |
617 | size_t len = cp - tline; | |
618 | if (fgets(tline + len, MAXLINE - len, input) == NULL) { | |
619 | /* append the missing newline */ | |
620 | tline[len+0] = '\n'; | |
621 | tline[len+1] = '\0'; | |
622 | cp++; | |
623 | linestate = LS_START; | |
624 | } else { | |
625 | linestate = LS_DIRTY; | |
626 | } | |
627 | } | |
01f1c879 SR |
628 | } |
629 | if (linestate == LS_DIRTY) { | |
630 | while (*cp != '\0') | |
631 | cp = skipcomment(cp + 1); | |
632 | } | |
633 | debug("parser %s comment %s line", | |
634 | comment_name[incomment], linestate_name[linestate]); | |
635 | return (retval); | |
636 | } | |
637 | ||
638 | /* | |
639 | * These are the binary operators that are supported by the expression | |
d8379ab1 | 640 | * evaluator. |
01f1c879 | 641 | */ |
d8379ab1 TF |
642 | static Linetype op_strict(int *p, int v, Linetype at, Linetype bt) { |
643 | if(at == LT_IF || bt == LT_IF) return (LT_IF); | |
644 | return (*p = v, v ? LT_TRUE : LT_FALSE); | |
645 | } | |
646 | static Linetype op_lt(int *p, Linetype at, int a, Linetype bt, int b) { | |
647 | return op_strict(p, a < b, at, bt); | |
648 | } | |
649 | static Linetype op_gt(int *p, Linetype at, int a, Linetype bt, int b) { | |
650 | return op_strict(p, a > b, at, bt); | |
651 | } | |
652 | static Linetype op_le(int *p, Linetype at, int a, Linetype bt, int b) { | |
653 | return op_strict(p, a <= b, at, bt); | |
654 | } | |
655 | static Linetype op_ge(int *p, Linetype at, int a, Linetype bt, int b) { | |
656 | return op_strict(p, a >= b, at, bt); | |
657 | } | |
658 | static Linetype op_eq(int *p, Linetype at, int a, Linetype bt, int b) { | |
659 | return op_strict(p, a == b, at, bt); | |
660 | } | |
661 | static Linetype op_ne(int *p, Linetype at, int a, Linetype bt, int b) { | |
662 | return op_strict(p, a != b, at, bt); | |
663 | } | |
664 | static Linetype op_or(int *p, Linetype at, int a, Linetype bt, int b) { | |
665 | if (!strictlogic && (at == LT_TRUE || bt == LT_TRUE)) | |
666 | return (*p = 1, LT_TRUE); | |
667 | return op_strict(p, a || b, at, bt); | |
668 | } | |
669 | static Linetype op_and(int *p, Linetype at, int a, Linetype bt, int b) { | |
670 | if (!strictlogic && (at == LT_FALSE || bt == LT_FALSE)) | |
671 | return (*p = 0, LT_FALSE); | |
672 | return op_strict(p, a && b, at, bt); | |
673 | } | |
01f1c879 SR |
674 | |
675 | /* | |
676 | * An evaluation function takes three arguments, as follows: (1) a pointer to | |
677 | * an element of the precedence table which lists the operators at the current | |
678 | * level of precedence; (2) a pointer to an integer which will receive the | |
679 | * value of the expression; and (3) a pointer to a char* that points to the | |
680 | * expression to be evaluated and that is updated to the end of the expression | |
681 | * when evaluation is complete. The function returns LT_FALSE if the value of | |
d8379ab1 TF |
682 | * the expression is zero, LT_TRUE if it is non-zero, LT_IF if the expression |
683 | * depends on an unknown symbol, or LT_ERROR if there is a parse failure. | |
01f1c879 SR |
684 | */ |
685 | struct ops; | |
686 | ||
687 | typedef Linetype eval_fn(const struct ops *, int *, const char **); | |
688 | ||
689 | static eval_fn eval_table, eval_unary; | |
690 | ||
691 | /* | |
692 | * The precedence table. Expressions involving binary operators are evaluated | |
693 | * in a table-driven way by eval_table. When it evaluates a subexpression it | |
694 | * calls the inner function with its first argument pointing to the next | |
695 | * element of the table. Innermost expressions have special non-table-driven | |
696 | * handling. | |
697 | */ | |
698 | static const struct ops { | |
699 | eval_fn *inner; | |
700 | struct op { | |
701 | const char *str; | |
d8379ab1 | 702 | Linetype (*fn)(int *, Linetype, int, Linetype, int); |
01f1c879 SR |
703 | } op[5]; |
704 | } eval_ops[] = { | |
705 | { eval_table, { { "||", op_or } } }, | |
706 | { eval_table, { { "&&", op_and } } }, | |
707 | { eval_table, { { "==", op_eq }, | |
708 | { "!=", op_ne } } }, | |
709 | { eval_unary, { { "<=", op_le }, | |
710 | { ">=", op_ge }, | |
711 | { "<", op_lt }, | |
712 | { ">", op_gt } } } | |
713 | }; | |
714 | ||
715 | /* | |
716 | * Function for evaluating the innermost parts of expressions, | |
d8379ab1 TF |
717 | * viz. !expr (expr) number defined(symbol) symbol |
718 | * We reset the constexpr flag in the last two cases. | |
01f1c879 SR |
719 | */ |
720 | static Linetype | |
721 | eval_unary(const struct ops *ops, int *valp, const char **cpp) | |
722 | { | |
723 | const char *cp; | |
724 | char *ep; | |
725 | int sym; | |
d8379ab1 TF |
726 | bool defparen; |
727 | Linetype lt; | |
01f1c879 SR |
728 | |
729 | cp = skipcomment(*cpp); | |
730 | if (*cp == '!') { | |
731 | debug("eval%d !", ops - eval_ops); | |
732 | cp++; | |
d8379ab1 TF |
733 | lt = eval_unary(ops, valp, &cp); |
734 | if (lt == LT_ERROR) | |
735 | return (LT_ERROR); | |
736 | if (lt != LT_IF) { | |
737 | *valp = !*valp; | |
738 | lt = *valp ? LT_TRUE : LT_FALSE; | |
eedc9d83 | 739 | } |
01f1c879 SR |
740 | } else if (*cp == '(') { |
741 | cp++; | |
742 | debug("eval%d (", ops - eval_ops); | |
d8379ab1 TF |
743 | lt = eval_table(eval_ops, valp, &cp); |
744 | if (lt == LT_ERROR) | |
745 | return (LT_ERROR); | |
01f1c879 SR |
746 | cp = skipcomment(cp); |
747 | if (*cp++ != ')') | |
d8379ab1 | 748 | return (LT_ERROR); |
01f1c879 SR |
749 | } else if (isdigit((unsigned char)*cp)) { |
750 | debug("eval%d number", ops - eval_ops); | |
751 | *valp = strtol(cp, &ep, 0); | |
d8379ab1 TF |
752 | if (ep == cp) |
753 | return (LT_ERROR); | |
754 | lt = *valp ? LT_TRUE : LT_FALSE; | |
01f1c879 SR |
755 | cp = skipsym(cp); |
756 | } else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) { | |
757 | cp = skipcomment(cp+7); | |
758 | debug("eval%d defined", ops - eval_ops); | |
d8379ab1 TF |
759 | if (*cp == '(') { |
760 | cp = skipcomment(cp+1); | |
761 | defparen = true; | |
762 | } else { | |
763 | defparen = false; | |
764 | } | |
01f1c879 | 765 | sym = findsym(cp); |
d8379ab1 TF |
766 | if (sym < 0) { |
767 | lt = LT_IF; | |
768 | } else { | |
eedc9d83 | 769 | *valp = (value[sym] != NULL); |
d8379ab1 | 770 | lt = *valp ? LT_TRUE : LT_FALSE; |
eedc9d83 | 771 | } |
d8379ab1 TF |
772 | cp = skipsym(cp); |
773 | cp = skipcomment(cp); | |
774 | if (defparen && *cp++ != ')') | |
775 | return (LT_ERROR); | |
776 | constexpr = false; | |
01f1c879 SR |
777 | } else if (!endsym(*cp)) { |
778 | debug("eval%d symbol", ops - eval_ops); | |
779 | sym = findsym(cp); | |
d8379ab1 TF |
780 | cp = skipsym(cp); |
781 | if (sym < 0) { | |
782 | lt = LT_IF; | |
783 | cp = skipargs(cp); | |
784 | } else if (value[sym] == NULL) { | |
01f1c879 | 785 | *valp = 0; |
d8379ab1 TF |
786 | lt = LT_FALSE; |
787 | } else { | |
01f1c879 SR |
788 | *valp = strtol(value[sym], &ep, 0); |
789 | if (*ep != '\0' || ep == value[sym]) | |
d8379ab1 TF |
790 | return (LT_ERROR); |
791 | lt = *valp ? LT_TRUE : LT_FALSE; | |
792 | cp = skipargs(cp); | |
01f1c879 | 793 | } |
d8379ab1 | 794 | constexpr = false; |
01f1c879 SR |
795 | } else { |
796 | debug("eval%d bad expr", ops - eval_ops); | |
d8379ab1 | 797 | return (LT_ERROR); |
01f1c879 SR |
798 | } |
799 | ||
800 | *cpp = cp; | |
801 | debug("eval%d = %d", ops - eval_ops, *valp); | |
d8379ab1 | 802 | return (lt); |
01f1c879 SR |
803 | } |
804 | ||
805 | /* | |
806 | * Table-driven evaluation of binary operators. | |
807 | */ | |
808 | static Linetype | |
809 | eval_table(const struct ops *ops, int *valp, const char **cpp) | |
810 | { | |
811 | const struct op *op; | |
812 | const char *cp; | |
813 | int val; | |
d8379ab1 | 814 | Linetype lt, rt; |
01f1c879 SR |
815 | |
816 | debug("eval%d", ops - eval_ops); | |
817 | cp = *cpp; | |
d8379ab1 TF |
818 | lt = ops->inner(ops+1, valp, &cp); |
819 | if (lt == LT_ERROR) | |
820 | return (LT_ERROR); | |
01f1c879 SR |
821 | for (;;) { |
822 | cp = skipcomment(cp); | |
823 | for (op = ops->op; op->str != NULL; op++) | |
824 | if (strncmp(cp, op->str, strlen(op->str)) == 0) | |
825 | break; | |
826 | if (op->str == NULL) | |
827 | break; | |
828 | cp += strlen(op->str); | |
829 | debug("eval%d %s", ops - eval_ops, op->str); | |
d8379ab1 TF |
830 | rt = ops->inner(ops+1, &val, &cp); |
831 | if (rt == LT_ERROR) | |
832 | return (LT_ERROR); | |
833 | lt = op->fn(valp, lt, *valp, rt, val); | |
01f1c879 SR |
834 | } |
835 | ||
836 | *cpp = cp; | |
837 | debug("eval%d = %d", ops - eval_ops, *valp); | |
d8379ab1 TF |
838 | debug("eval%d lt = %s", ops - eval_ops, linetype_name[lt]); |
839 | return (lt); | |
01f1c879 SR |
840 | } |
841 | ||
842 | /* | |
843 | * Evaluate the expression on a #if or #elif line. If we can work out | |
844 | * the result we return LT_TRUE or LT_FALSE accordingly, otherwise we | |
845 | * return just a generic LT_IF. | |
846 | */ | |
847 | static Linetype | |
848 | ifeval(const char **cpp) | |
849 | { | |
850 | int ret; | |
d8379ab1 | 851 | int val = 0; |
01f1c879 SR |
852 | |
853 | debug("eval %s", *cpp); | |
d8379ab1 TF |
854 | constexpr = killconsts ? false : true; |
855 | ret = eval_table(eval_ops, &val, cpp); | |
01f1c879 | 856 | debug("eval = %d", val); |
d8379ab1 | 857 | return (constexpr ? LT_IF : ret == LT_ERROR ? LT_IF : ret); |
01f1c879 SR |
858 | } |
859 | ||
860 | /* | |
861 | * Skip over comments, strings, and character literals and stop at the | |
862 | * next character position that is not whitespace. Between calls we keep | |
863 | * the comment state in the global variable incomment, and we also adjust | |
864 | * the global variable linestate when we see a newline. | |
865 | * XXX: doesn't cope with the buffer splitting inside a state transition. | |
866 | */ | |
867 | static const char * | |
868 | skipcomment(const char *cp) | |
869 | { | |
870 | if (text || ignoring[depth]) { | |
871 | for (; isspace((unsigned char)*cp); cp++) | |
872 | if (*cp == '\n') | |
873 | linestate = LS_START; | |
874 | return (cp); | |
875 | } | |
876 | while (*cp != '\0') | |
877 | /* don't reset to LS_START after a line continuation */ | |
878 | if (strncmp(cp, "\\\n", 2) == 0) | |
879 | cp += 2; | |
880 | else switch (incomment) { | |
881 | case NO_COMMENT: | |
882 | if (strncmp(cp, "/\\\n", 3) == 0) { | |
883 | incomment = STARTING_COMMENT; | |
884 | cp += 3; | |
885 | } else if (strncmp(cp, "/*", 2) == 0) { | |
886 | incomment = C_COMMENT; | |
887 | cp += 2; | |
888 | } else if (strncmp(cp, "//", 2) == 0) { | |
889 | incomment = CXX_COMMENT; | |
890 | cp += 2; | |
891 | } else if (strncmp(cp, "\'", 1) == 0) { | |
892 | incomment = CHAR_LITERAL; | |
893 | linestate = LS_DIRTY; | |
894 | cp += 1; | |
895 | } else if (strncmp(cp, "\"", 1) == 0) { | |
896 | incomment = STRING_LITERAL; | |
897 | linestate = LS_DIRTY; | |
898 | cp += 1; | |
899 | } else if (strncmp(cp, "\n", 1) == 0) { | |
900 | linestate = LS_START; | |
901 | cp += 1; | |
902 | } else if (strchr(" \t", *cp) != NULL) { | |
903 | cp += 1; | |
904 | } else | |
905 | return (cp); | |
906 | continue; | |
907 | case CXX_COMMENT: | |
908 | if (strncmp(cp, "\n", 1) == 0) { | |
909 | incomment = NO_COMMENT; | |
910 | linestate = LS_START; | |
911 | } | |
912 | cp += 1; | |
913 | continue; | |
914 | case CHAR_LITERAL: | |
915 | case STRING_LITERAL: | |
916 | if ((incomment == CHAR_LITERAL && cp[0] == '\'') || | |
917 | (incomment == STRING_LITERAL && cp[0] == '\"')) { | |
918 | incomment = NO_COMMENT; | |
919 | cp += 1; | |
920 | } else if (cp[0] == '\\') { | |
921 | if (cp[1] == '\0') | |
922 | cp += 1; | |
923 | else | |
924 | cp += 2; | |
925 | } else if (strncmp(cp, "\n", 1) == 0) { | |
926 | if (incomment == CHAR_LITERAL) | |
927 | error("unterminated char literal"); | |
928 | else | |
929 | error("unterminated string literal"); | |
930 | } else | |
931 | cp += 1; | |
932 | continue; | |
933 | case C_COMMENT: | |
934 | if (strncmp(cp, "*\\\n", 3) == 0) { | |
935 | incomment = FINISHING_COMMENT; | |
936 | cp += 3; | |
937 | } else if (strncmp(cp, "*/", 2) == 0) { | |
938 | incomment = NO_COMMENT; | |
939 | cp += 2; | |
940 | } else | |
941 | cp += 1; | |
942 | continue; | |
943 | case STARTING_COMMENT: | |
944 | if (*cp == '*') { | |
945 | incomment = C_COMMENT; | |
946 | cp += 1; | |
947 | } else if (*cp == '/') { | |
948 | incomment = CXX_COMMENT; | |
949 | cp += 1; | |
950 | } else { | |
951 | incomment = NO_COMMENT; | |
952 | linestate = LS_DIRTY; | |
953 | } | |
954 | continue; | |
955 | case FINISHING_COMMENT: | |
956 | if (*cp == '/') { | |
957 | incomment = NO_COMMENT; | |
958 | cp += 1; | |
959 | } else | |
960 | incomment = C_COMMENT; | |
961 | continue; | |
962 | default: | |
963 | abort(); /* bug */ | |
964 | } | |
965 | return (cp); | |
966 | } | |
967 | ||
d8379ab1 TF |
968 | /* |
969 | * Skip macro arguments. | |
970 | */ | |
971 | static const char * | |
972 | skipargs(const char *cp) | |
973 | { | |
974 | const char *ocp = cp; | |
975 | int level = 0; | |
976 | cp = skipcomment(cp); | |
977 | if (*cp != '(') | |
978 | return (cp); | |
979 | do { | |
980 | if (*cp == '(') | |
981 | level++; | |
982 | if (*cp == ')') | |
983 | level--; | |
984 | cp = skipcomment(cp+1); | |
985 | } while (level != 0 && *cp != '\0'); | |
986 | if (level == 0) | |
987 | return (cp); | |
988 | else | |
989 | /* Rewind and re-detect the syntax error later. */ | |
990 | return (ocp); | |
991 | } | |
992 | ||
01f1c879 SR |
993 | /* |
994 | * Skip over an identifier. | |
995 | */ | |
996 | static const char * | |
997 | skipsym(const char *cp) | |
998 | { | |
999 | while (!endsym(*cp)) | |
1000 | ++cp; | |
1001 | return (cp); | |
1002 | } | |
1003 | ||
1004 | /* | |
d8379ab1 | 1005 | * Look for the symbol in the symbol table. If it is found, we return |
01f1c879 SR |
1006 | * the symbol table index, else we return -1. |
1007 | */ | |
1008 | static int | |
1009 | findsym(const char *str) | |
1010 | { | |
1011 | const char *cp; | |
1012 | int symind; | |
1013 | ||
1014 | cp = skipsym(str); | |
1015 | if (cp == str) | |
1016 | return (-1); | |
1017 | if (symlist) { | |
1018 | printf("%.*s\n", (int)(cp-str), str); | |
1019 | /* we don't care about the value of the symbol */ | |
1020 | return (0); | |
1021 | } | |
1022 | for (symind = 0; symind < nsyms; ++symind) { | |
1023 | if (strlcmp(symname[symind], str, cp-str) == 0) { | |
1024 | debug("findsym %s %s", symname[symind], | |
1025 | value[symind] ? value[symind] : ""); | |
1026 | return (symind); | |
1027 | } | |
1028 | } | |
1029 | return (-1); | |
1030 | } | |
1031 | ||
1032 | /* | |
1033 | * Add a symbol to the symbol table. | |
1034 | */ | |
1035 | static void | |
1036 | addsym(bool ignorethis, bool definethis, char *sym) | |
1037 | { | |
1038 | int symind; | |
1039 | char *val; | |
1040 | ||
1041 | symind = findsym(sym); | |
1042 | if (symind < 0) { | |
1043 | if (nsyms >= MAXSYMS) | |
1044 | errx(2, "too many symbols"); | |
1045 | symind = nsyms++; | |
1046 | } | |
1047 | symname[symind] = sym; | |
1048 | ignore[symind] = ignorethis; | |
1049 | val = sym + (skipsym(sym) - sym); | |
1050 | if (definethis) { | |
1051 | if (*val == '=') { | |
1052 | value[symind] = val+1; | |
1053 | *val = '\0'; | |
1054 | } else if (*val == '\0') | |
1055 | value[symind] = ""; | |
1056 | else | |
1057 | usage(); | |
1058 | } else { | |
1059 | if (*val != '\0') | |
1060 | usage(); | |
1061 | value[symind] = NULL; | |
1062 | } | |
1063 | } | |
1064 | ||
1065 | /* | |
1066 | * Compare s with n characters of t. | |
1067 | * The same as strncmp() except that it checks that s[n] == '\0'. | |
1068 | */ | |
1069 | static int | |
1070 | strlcmp(const char *s, const char *t, size_t n) | |
1071 | { | |
1072 | while (n-- && *t != '\0') | |
1073 | if (*s != *t) | |
1074 | return ((unsigned char)*s - (unsigned char)*t); | |
1075 | else | |
1076 | ++s, ++t; | |
1077 | return ((unsigned char)*s); | |
1078 | } | |
1079 | ||
1080 | /* | |
1081 | * Diagnostics. | |
1082 | */ | |
1083 | static void | |
1084 | debug(const char *msg, ...) | |
1085 | { | |
1086 | va_list ap; | |
1087 | ||
1088 | if (debugging) { | |
1089 | va_start(ap, msg); | |
1090 | vwarnx(msg, ap); | |
1091 | va_end(ap); | |
1092 | } | |
1093 | } | |
1094 | ||
1095 | static void | |
1096 | error(const char *msg) | |
1097 | { | |
1098 | if (depth == 0) | |
1099 | warnx("%s: %d: %s", filename, linenum, msg); | |
1100 | else | |
1101 | warnx("%s: %d: %s (#if line %d depth %d)", | |
1102 | filename, linenum, msg, stifline[depth], depth); | |
1103 | errx(2, "output may be truncated"); | |
1104 | } |