2  * Copyright (c) 2002 - 2020 Tony Finch <dot@dotat.at>
27  * unifdef - remove ifdef'ed lines
31  * of unifdef carried the 4-clause BSD copyright licence. None of its code
54 /* types of input lines: */
84 #define linetype_if2elif(lt) ((Linetype)(lt - LT_IF + LT_ELIF))
91 	IS_TRUE_PREFIX,		/* first non-false #(el)if is true */
92 	IS_PASS_MIDDLE,		/* first non-false #(el)if is unknown */
114 	STARTING_COMMENT,	/* just after slash-backslash-newline */
115 	FINISHING_COMMENT,	/* star-backslash-newline in a C comment */
156 	bool		ignore;		/* -iDsym or -iUsym */
162 	return (strcmp(a->name, b->name));  in macro_cmp()
172 static bool             compblank;		/* -B: compress blank lines */
173 static bool             lnblank;		/* -b: blank deleted lines */
174 static bool             complement;		/* -c: do the complement */
175 static bool             debugging;		/* -d: debugging reports */
176 static bool             inplace;		/* -m: modify in place */
177 static bool             iocccok;		/* -e: fewer IOCCC errors */
178 static bool             strictlogic;		/* -K: keep ambiguous #ifs */
179 static bool             killconsts;		/* -k: eval constant #ifs */
180 static bool             lnnum;			/* -n: add #line directives */
181 static bool             symlist;		/* -s: output symbol list */
182 static bool             symdepth;		/* -S: output symbol depth */
183 static bool             text;			/* -t: this is a text file */
185 static FILE            *input;			/* input file pointer */  variable
186 static const char      *filename;		/* input file name */
192 static char            *tempname;		/* avoid splatting input */
194 static char             tline[MAXLINE+EDITSLOP];/* input buffer plus space */
199  * input's, and unifdef correctly handles CRLF or LF endings whatever
202  * When the output isn't a processed input file (when it is error /
206 static const char      *newline;		/* input file format */
215 static int              depth;			/* current #if nesting */  variable
273 	while ((opt = getopt(argc, argv, "i:D:U:f:I:M:o:x:bBcdehKklmnsStV")) != -1)  in main()
277 			 * For strict backwards-compatibility the U or D  in main()
278 			 * should be immediately after the -i but it doesn't  in main()
295 		case 'I': /* no-op for compatibility with cpp */  in main()
304 		case 'c': /* treat -D as -U and vice versa */  in main()
342 		case 'S': /* list symbols with their nesting depth */  in main()
359 	argc -= optind;  in main()
362 		errx(2, "-B and -b are mutually exclusive");  in main()
364 		errx(2, "-s only works with one input file");  in main()
366 		errx(2, "-o cannot be used with multiple input files");  in main()
368 		errx(2, "multiple input files require -m or -M");  in main()
370 		errx(2, "-m requires an input file");  in main()
374 		ofilename = "-";  in main()
380 	else while (argc-- > 0) {  in main()
400 	if (ifn == NULL || strcmp(ifn, "-") == 0) {  in processinout()
403 		input = fbinmode(stdin);  in processinout()
407 		input = fopen(ifn, "rb");  in processinout()
408 		if (input == NULL)  in processinout()
411 	if (strcmp(ofn, "-") == 0) {  in processinout()
458  * Self-identification functions.
479 	    "usage:	unifdef [-bBcdehKkmnsStV] [-x{012}] [-Mext] [-opath] \\\n"  in synopsis()
480 	    "		[-[i]Dsym[=val]] [-[i]Usym] [-fpath] ... [file] ...\n");  in synopsis()
495 	    "	-Dsym=val  define preprocessor symbol with given value\n"  in help()
496 	    "	-Dsym      define preprocessor symbol with value 1\n"  in help()
497 	    "	-Usym	   preprocessor symbol is undefined\n"  in help()
498 	    "	-iDsym=val \\  ignore C strings and comments\n"  in help()
499 	    "	-iDsym      ) in sections controlled by these\n"  in help()
500 	    "	-iUsym	   /  preprocessor symbols\n"  in help()
501 	    "	-fpath	file containing #define and #undef directives\n"  in help()
502 	    "	-b	blank lines instead of deleting them\n"  in help()
503 	    "	-B	compress blank lines around deleted section\n"  in help()
504 	    "	-c	complement (invert) keep vs. delete\n"  in help()
505 	    "	-d	debugging mode\n"  in help()
506 	    "	-e	ignore multiline preprocessor directives\n"  in help()
507 	    "	-h	print help\n"  in help()
508 	    "	-Ipath	extra include file path (ignored)\n"  in help()
509 	    "	-K	disable && and || short-circuiting\n"  in help()
510 	    "	-k	process constant #if expressions\n"  in help()
511 	    "	-Mext	modify in place and keep backups\n"  in help()
512 	    "	-m	modify input files in place\n"  in help()
513 	    "	-n	add #line directives to output\n"  in help()
514 	    "	-opath	output file name\n"  in help()
515 	    "	-S	list #if control symbols with nesting\n"  in help()
516 	    "	-s	list #if control symbols\n"  in help()
517 	    "	-t	ignore C strings and comments\n"  in help()
518 	    "	-V	print version\n"  in help()
519 	    "	-x{012}	exit status mode\n"  in help()
530  * functions increase or decrease the depth. They also maintain the
534  * When we have processed a group that starts off with a known-false
546  * example being a multi-line comment hanging off the right of the
649 	if (depth == 0)  in ignoreoff()
651 	ignoring[depth] = ignoring[depth-1];  in ignoreoff()
656 	ignoring[depth] = true;  in ignoreon()
661 	snprintf(keyword, tline + sizeof(tline) - keyword,  in keywordedit()
669 	if (depth > MAXDEPTH-1)  in nest()
671 	if (depth == MAXDEPTH-1)  in nest()
673 	depth += 1;  in nest()
674 	stifline[depth] = linenum;  in nest()
679 	if (depth == 0)  in unnest()
681 	depth -= 1;  in unnest()
686 	ifstate[depth] = is;  in state()
735  * Format of #line directives depends on whether we know the input filename.
762 	fclose(input);  in closeio()
781 		trans_table[ifstate[depth]][lineval]();  in process()
782 		debug("process line %d %s -> %s depth %d",  in process()
784 		    ifstate_name[ifstate[depth]], depth);  in process()
816 	keyword = tline + (cp - tline);  in parseline()
825 			if (cursym->value == NULL)  in parseline()
828 			if (cursym->ignore)  in parseline()
862 		long len = cp - tline;  in parseline()
863 		if (fgets(tline + len, MAXLINE - len, input) == NULL) {  in parseline()
864 			if (ferror(input))  in parseline()
932 	return op_strict(p, a - b, at, bt);  in op_sub()
964  * the expression is zero, LT_TRUE if it is non-zero, LT_IF if the expression
975  * in a table-driven way by eval_table. When it evaluates a subexpression it
977  * element of the table. Innermost expressions have special non-table-driven
1008 			{ "-",  op_sub,  NULL } } },
1017 	return (ops - eval_ops);  in prec()
1055 	} else if (*cp == '-') {  in eval_unary()
1056 		debug("eval%d -", prec(ops));  in eval_unary()
1062 			*valp = -(*valp);  in eval_unary()
1099 			debug("eval%d defined %s", prec(ops), sym->name);  in eval_unary()
1100 			*valp = (sym->value != NULL);  in eval_unary()
1110 		} else if (sym->value == NULL) {  in eval_unary()
1114 			*valp = strtol(sym->value, &ep, 0);  in eval_unary()
1115 			if (*ep != '\0' || ep == sym->value)  in eval_unary()
1132  * Table-driven evaluation of binary operators.
1144 	lt = ops->inner(ops+1, valp, &cp);  in eval_table()
1149 		for (op = ops->op; op->str != NULL; op++) {  in eval_table()
1150 			if (strncmp(cp, op->str, strlen(op->str)) == 0) {  in eval_table()
1151 				/* assume only one-char operators have stop chars */  in eval_table()
1152 				if (op->stop != NULL && cp[1] != '\0' &&  in eval_table()
1153 				    strchr(op->stop, cp[1]) != NULL)  in eval_table()
1159 		if (op->str == NULL)  in eval_table()
1161 		cp += strlen(op->str);  in eval_table()
1162 		debug("eval%d %s", prec(ops), op->str);  in eval_table()
1163 		rt = ops->inner(ops+1, &val, &cp);  in eval_table()
1166 		lt = op->fn(valp, lt, *valp, rt, val);  in eval_table()
1205 	if (fgets(tline, MAXLINE, input) == NULL) {  in skiphash()
1206 		if (ferror(input))  in skiphash()
1250 	if (text || ignoring[depth]) {  in skipcomment()
1382 			level--;  in skipargs()
1388 	/* Rewind and re-detect the syntax error later. */  in skipargs()
1441  * the symbol table index, else we return -1.
1456 			printf("%s%3d", zerosyms ? "" : "\n", depth);  in findsym()
1460 		       (int)(*strp-str), str,  in findsym()
1467 	 * 'str' just points into the current mid-parse input and is not  in findsym()
1468 	 * nul-terminated.  We know the length of the symbol, *strp - str, but  in findsym()
1469 	 * need to provide a nul-terminated lookup key for RB_FIND's comparison  in findsym()
1472 	strkey = malloc(*strp - str + 1);  in findsym()
1473 	memcpy(strkey, str, *strp - str);  in findsym()
1474 	strkey[*strp - str] = 0;  in findsym()
1498 			if (sym->value == NULL)  in indirectsym()
1500 			key.name = sym->value;  in indirectsym()
1503 			    ind->value == NULL ||  in indirectsym()
1504 			    ind->value == sym->value)  in indirectsym()
1507 			sym->value = ind->value;  in indirectsym()
1525 		symval[val - sym] = '\0';  in addsym1()
1547 		sym->ignore = ignorethis;  in addsym2()
1548 		sym->name = symname;  in addsym2()
1549 		sym->value = val;  in addsym2()
1554 		sym->ignore = ignorethis;  in addsym2()
1555 		sym->value = val;  in addsym2()
1563 	debug("%s %s%c%s", why, sym->name,  in debugsym()
1564 	    sym->value ? '=' : ' ',  in debugsym()
1565 	    sym->value ? sym->value : "undef");  in debugsym()
1576 	input = fopen(fn, "rb");  in defundefile()
1577 	if (input == NULL)  in defundefile()
1582 	if (ferror(input))  in defundefile()
1585 		fclose(input);  in defundefile()
1604 	   avoid unsupported multi-line preprocessor directives */  in defundef()
1606 	while (end > tline && strchr(" \t\n\r", end[-1]) != NULL)  in defundef()
1607 		--end;  in defundef()
1608 	if (end > tline && end[-1] == '\\')  in defundef()
1672 	n = (size_t)(end - start) + 1;  in xstrdup()
1698 	if (depth == 0)  in error()
1701 		warnx("%s: %d: %s (#if line %d depth %d)",  in error()
1702 		    filename, linenum, msg, stifline[depth], depth);  in error()