1/*	$NetBSD$	*/
2
3/* Copyright (C) 1989, 1997, 1998, 1999 Aladdin Enterprises.  All rights reserved. */
4
5/*Id: ansi2knr.c,v 1.1 2000/01/21 00:54:15 alainm Exp*/
6/* Convert ANSI C function definitions to K&R ("traditional C") syntax */
7
8/*
9ansi2knr is distributed in the hope that it will be useful, but WITHOUT ANY
10WARRANTY.  No author or distributor accepts responsibility to anyone for the
11consequences of using it or for whether it serves any particular purpose or
12works at all, unless he says so in writing.  Refer to the GNU General Public
13License (the "GPL") for full details.
14
15Everyone is granted permission to copy, modify and redistribute ansi2knr,
16but only under the conditions described in the GPL.  A copy of this license
17is supposed to have been given to you along with ansi2knr so you can know
18your rights and responsibilities.  It should be in a file named COPYLEFT,
19or, if there is no file named COPYLEFT, a file named COPYING.  Among other
20things, the copyright notice and this notice must be preserved on all
21copies.
22
23We explicitly state here what we believe is already implied by the GPL: if
24the ansi2knr program is distributed as a separate set of sources and a
25separate executable file which are aggregated on a storage medium together
26with another program, this in itself does not bring the other program under
27the GPL, nor does the mere fact that such a program or the procedures for
28constructing it invoke the ansi2knr executable bring any other part of the
29program under the GPL.
30*/
31
32/*
33 * Usage:
34	ansi2knr [--filename FILENAME] [INPUT_FILE [OUTPUT_FILE]]
35 * --filename provides the file name for the #line directive in the output,
36 * overriding input_file (if present).
37 * If no input_file is supplied, input is read from stdin.
38 * If no output_file is supplied, output goes to stdout.
39 * There are no error messages.
40 *
41 * ansi2knr recognizes function definitions by seeing a non-keyword
42 * identifier at the left margin, followed by a left parenthesis,
43 * with a right parenthesis as the last character on the line,
44 * and with a left brace as the first token on the following line
45 * (ignoring possible intervening comments), except that a line
46 * consisting of only
47 *	identifier1(identifier2)
48 * will not be considered a function definition unless identifier2 is
49 * the word "void", and a line consisting of
50 *	identifier1(identifier2, <<arbitrary>>)
51 * will not be considered a function definition.
52 * ansi2knr will recognize a multi-line header provided
53 * that no intervening line ends with a left or right brace or a semicolon.
54 * These algorithms ignore whitespace and comments, except that
55 * the function name must be the first thing on the line.
56 * The following constructs will confuse it:
57 *	- Any other construct that starts at the left margin and
58 *	    follows the above syntax (such as a macro or function call).
59 *	- Some macros that tinker with the syntax of function headers.
60 */
61
62/*
63 * The original and principal author of ansi2knr is L. Peter Deutsch
64 * <ghost@aladdin.com>.  Other authors are noted in the change history
65 * that follows (in reverse chronological order):
66	lpd 1999-04-12 added minor fixes from Pavel Roskin
67		<pavel_roskin@geocities.com> for clean compilation with
68		gcc -W -Wall
69	lpd 1999-03-22 added hack to recognize lines consisting of
70		identifier1(identifier2, xxx) as *not* being procedures
71	lpd 1999-02-03 made indentation of preprocessor commands consistent
72	lpd 1999-01-28 fixed two bugs: a '/' in an argument list caused an
73		endless loop; quoted strings within an argument list
74		confused the parser
75	lpd 1999-01-24 added a check for write errors on the output,
76		suggested by Jim Meyering <meyering@ascend.com>
77	lpd 1998-11-09 added further hack to recognize identifier(void)
78		as being a procedure
79	lpd 1998-10-23 added hack to recognize lines consisting of
80		identifier1(identifier2) as *not* being procedures
81	lpd 1997-12-08 made input_file optional; only closes input and/or
82		output file if not stdin or stdout respectively; prints
83		usage message on stderr rather than stdout; adds
84		--filename switch (changes suggested by
85		<ceder@lysator.liu.se>)
86	lpd 1996-01-21 added code to cope with not HAVE_CONFIG_H and with
87		compilers that don't understand void, as suggested by
88		Tom Lane
89	lpd 1996-01-15 changed to require that the first non-comment token
90		on the line following a function header be a left brace,
91		to reduce sensitivity to macros, as suggested by Tom Lane
92		<tgl@sss.pgh.pa.us>
93	lpd 1995-06-22 removed #ifndefs whose sole purpose was to define
94		undefined preprocessor symbols as 0; changed all #ifdefs
95		for configuration symbols to #ifs
96	lpd 1995-04-05 changed copyright notice to make it clear that
97		including ansi2knr in a program does not bring the entire
98		program under the GPL
99	lpd 1994-12-18 added conditionals for systems where ctype macros
100		don't handle 8-bit characters properly, suggested by
101		Francois Pinard <pinard@iro.umontreal.ca>;
102		removed --varargs switch (this is now the default)
103	lpd 1994-10-10 removed CONFIG_BROKETS conditional
104	lpd 1994-07-16 added some conditionals to help GNU `configure',
105		suggested by Francois Pinard <pinard@iro.umontreal.ca>;
106		properly erase prototype args in function parameters,
107		contributed by Jim Avera <jima@netcom.com>;
108		correct error in writeblanks (it shouldn't erase EOLs)
109	lpd 1989-xx-xx original version
110 */
111
112/* Most of the conditionals here are to make ansi2knr work with */
113/* or without the GNU configure machinery. */
114
115#if HAVE_CONFIG_H
116# include <config.h>
117#endif
118
119#include <stdio.h>
120#include <ctype.h>
121
122#if HAVE_CONFIG_H
123
124/*
125   For properly autoconfiguring ansi2knr, use AC_CONFIG_HEADER(config.h).
126   This will define HAVE_CONFIG_H and so, activate the following lines.
127 */
128
129# if STDC_HEADERS || HAVE_STRING_H
130#  include <string.h>
131# else
132#  include <strings.h>
133# endif
134
135#else /* not HAVE_CONFIG_H */
136
137/* Otherwise do it the hard way */
138
139# ifdef BSD
140#  include <strings.h>
141# else
142#  ifdef VMS
143    extern int strlen(), strncmp();
144#  else
145#   include <string.h>
146#  endif
147# endif
148
149#endif /* not HAVE_CONFIG_H */
150
151#if STDC_HEADERS
152# include <stdlib.h>
153#else
154/*
155   malloc and free should be declared in stdlib.h,
156   but if you've got a K&R compiler, they probably aren't.
157 */
158# ifdef MSDOS
159#  include <malloc.h>
160# else
161#  ifdef VMS
162     extern char *malloc();
163     extern void free();
164#  else
165     extern char *malloc();
166     extern int free();
167#  endif
168# endif
169
170#endif
171
172/* Define NULL (for *very* old compilers). */
173#ifndef NULL
174# define NULL (0)
175#endif
176
177/*
178 * The ctype macros don't always handle 8-bit characters correctly.
179 * Compensate for this here.
180 */
181#ifdef isascii
182# undef HAVE_ISASCII		/* just in case */
183# define HAVE_ISASCII 1
184#else
185#endif
186#if STDC_HEADERS || !HAVE_ISASCII
187# define is_ascii(c) 1
188#else
189# define is_ascii(c) isascii(c)
190#endif
191
192#define is_space(c) (is_ascii(c) && isspace(c))
193#define is_alpha(c) (is_ascii(c) && isalpha(c))
194#define is_alnum(c) (is_ascii(c) && isalnum(c))
195
196/* Scanning macros */
197#define isidchar(ch) (is_alnum(ch) || (ch) == '_')
198#define isidfirstchar(ch) (is_alpha(ch) || (ch) == '_')
199
200/* Forward references */
201char *skipspace();
202char *scanstring();
203int writeblanks();
204int test1();
205int convert1();
206
207/* The main program */
208int
209main(argc, argv)
210    int argc;
211    char *argv[];
212{	FILE *in = stdin;
213	FILE *out = stdout;
214	char *filename = 0;
215	char *program_name = argv[0];
216	char *output_name = 0;
217#define bufsize 5000			/* arbitrary size */
218	char *buf;
219	char *line;
220	char *more;
221	char *usage =
222	  "Usage: ansi2knr [--filename FILENAME] [INPUT_FILE [OUTPUT_FILE]]\n";
223	/*
224	 * In previous versions, ansi2knr recognized a --varargs switch.
225	 * If this switch was supplied, ansi2knr would attempt to convert
226	 * a ... argument to va_alist and va_dcl; if this switch was not
227	 * supplied, ansi2knr would simply drop any such arguments.
228	 * Now, ansi2knr always does this conversion, and we only
229	 * check for this switch for backward compatibility.
230	 */
231	int convert_varargs = 1;
232	int output_error;
233
234	while ( argc > 1 && argv[1][0] == '-' ) {
235	  if ( !strcmp(argv[1], "--varargs") ) {
236	    convert_varargs = 1;
237	    argc--;
238	    argv++;
239	    continue;
240	  }
241	  if ( !strcmp(argv[1], "--filename") && argc > 2 ) {
242	    filename = argv[2];
243	    argc -= 2;
244	    argv += 2;
245	    continue;
246	  }
247	  fprintf(stderr, "%s: Unrecognized switch: %s\n", program_name,
248		  argv[1]);
249	  fprintf(stderr, usage);
250	  exit(1);
251	}
252	switch ( argc )
253	   {
254	default:
255		fprintf(stderr, usage);
256		exit(0);
257	case 3:
258		output_name = argv[2];
259		out = fopen(output_name, "w");
260		if ( out == NULL ) {
261		  fprintf(stderr, "%s: Cannot open output file %s\n",
262			  program_name, output_name);
263		  exit(1);
264		}
265		/* falls through */
266	case 2:
267		in = fopen(argv[1], "r");
268		if ( in == NULL ) {
269		  fprintf(stderr, "%s: Cannot open input file %s\n",
270			  program_name, argv[1]);
271		  exit(1);
272		}
273		if ( filename == 0 )
274		  filename = argv[1];
275		/* falls through */
276	case 1:
277		break;
278	   }
279	if ( filename )
280	  fprintf(out, "#line 1 \"%s\"\n", filename);
281	buf = malloc(bufsize);
282	if ( buf == NULL )
283	   {
284		fprintf(stderr, "Unable to allocate read buffer!\n");
285		exit(1);
286	   }
287	line = buf;
288	while ( fgets(line, (unsigned)(buf + bufsize - line), in) != NULL )
289	   {
290test:		line += strlen(line);
291		switch ( test1(buf) )
292		   {
293		case 2:			/* a function header */
294			convert1(buf, out, 1, convert_varargs);
295			break;
296		case 1:			/* a function */
297			/* Check for a { at the start of the next line. */
298			more = ++line;
299f:			if ( line >= buf + (bufsize - 1) ) /* overflow check */
300			  goto wl;
301			if ( fgets(line, (unsigned)(buf + bufsize - line), in) == NULL )
302			  goto wl;
303			switch ( *skipspace(more, 1) )
304			  {
305			  case '{':
306			    /* Definitely a function header. */
307			    convert1(buf, out, 0, convert_varargs);
308			    fputs(more, out);
309			    break;
310			  case 0:
311			    /* The next line was blank or a comment: */
312			    /* keep scanning for a non-comment. */
313			    line += strlen(line);
314			    goto f;
315			  default:
316			    /* buf isn't a function header, but */
317			    /* more might be. */
318			    fputs(buf, out);
319			    strcpy(buf, more);
320			    line = buf;
321			    goto test;
322			  }
323			break;
324		case -1:		/* maybe the start of a function */
325			if ( line != buf + (bufsize - 1) ) /* overflow check */
326			  continue;
327			/* falls through */
328		default:		/* not a function */
329wl:			fputs(buf, out);
330			break;
331		   }
332		line = buf;
333	   }
334	if ( line != buf )
335	  fputs(buf, out);
336	free(buf);
337	if ( output_name ) {
338	  output_error = ferror(out);
339	  output_error |= fclose(out);
340	} else {		/* out == stdout */
341	  fflush(out);
342	  output_error = ferror(out);
343	}
344	if ( output_error ) {
345	  fprintf(stderr, "%s: error writing to %s\n", program_name,
346		  (output_name ? output_name : "stdout"));
347	  exit(1);
348	}
349	if ( in != stdin )
350	  fclose(in);
351	return 0;
352}
353
354/* Skip over whitespace and comments, in either direction. */
355char *
356skipspace(p, dir)
357    register char *p;
358    register int dir;			/* 1 for forward, -1 for backward */
359{	for ( ; ; )
360	   {	while ( is_space(*p) )
361		  p += dir;
362		if ( !(*p == '/' && p[dir] == '*') )
363		  break;
364		p += dir;  p += dir;
365		while ( !(*p == '*' && p[dir] == '/') )
366		   {	if ( *p == 0 )
367			  return p;	/* multi-line comment?? */
368			p += dir;
369		   }
370		p += dir;  p += dir;
371	   }
372	return p;
373}
374
375/* Scan over a quoted string, in either direction. */
376char *
377scanstring(p, dir)
378    register char *p;
379    register int dir;
380{
381    for (p += dir; ; p += dir)
382	if (*p == '"' && p[-dir] != '\\')
383	    return p + dir;
384}
385
386/*
387 * Write blanks over part of a string.
388 * Don't overwrite end-of-line characters.
389 */
390int
391writeblanks(start, end)
392    char *start;
393    char *end;
394{	char *p;
395	for ( p = start; p < end; p++ )
396	  if ( *p != '\r' && *p != '\n' )
397	    *p = ' ';
398	return 0;
399}
400
401/*
402 * Test whether the string in buf is a function definition.
403 * The string may contain and/or end with a newline.
404 * Return as follows:
405 *	0 - definitely not a function definition;
406 *	1 - definitely a function definition;
407 *	2 - definitely a function prototype (NOT USED);
408 *	-1 - may be the beginning of a function definition,
409 *		append another line and look again.
410 * The reason we don't attempt to convert function prototypes is that
411 * Ghostscript's declaration-generating macros look too much like
412 * prototypes, and confuse the algorithms.
413 */
414int
415test1(buf)
416    char *buf;
417{	register char *p = buf;
418	char *bend;
419	char *endfn;
420	int contin;
421
422	if ( !isidfirstchar(*p) )
423	  return 0;		/* no name at left margin */
424	bend = skipspace(buf + strlen(buf) - 1, -1);
425	switch ( *bend )
426	   {
427	   case ';': contin = 0 /*2*/; break;
428	   case ')': contin = 1; break;
429	   case '{': return 0;		/* not a function */
430	   case '}': return 0;		/* not a function */
431	   default: contin = -1;
432	   }
433	while ( isidchar(*p) )
434	  p++;
435	endfn = p;
436	p = skipspace(p, 1);
437	if ( *p++ != '(' )
438	  return 0;		/* not a function */
439	p = skipspace(p, 1);
440	if ( *p == ')' )
441	  return 0;		/* no parameters */
442	/* Check that the apparent function name isn't a keyword. */
443	/* We only need to check for keywords that could be followed */
444	/* by a left parenthesis (which, unfortunately, is most of them). */
445	   {	static char *words[] =
446		   {	"asm", "auto", "case", "char", "const", "double",
447			"extern", "float", "for", "if", "int", "long",
448			"register", "return", "short", "signed", "sizeof",
449			"static", "switch", "typedef", "unsigned",
450			"void", "volatile", "while", 0
451		   };
452		char **key = words;
453		char *kp;
454		unsigned len = endfn - buf;
455
456		while ( (kp = *key) != 0 )
457		   {	if ( strlen(kp) == len && !strncmp(kp, buf, len) )
458			  return 0;	/* name is a keyword */
459			key++;
460		   }
461	   }
462	   {
463	       char *id = p;
464	       ptrdiff_t len;
465	       /*
466		* Check for identifier1(identifier2) and not
467		* identifier1(void), or identifier1(identifier2, xxxx).
468		*/
469
470	       while ( isidchar(*p) )
471		   p++;
472	       len = p - id;
473	       p = skipspace(p, 1);
474	       if (*p == ',' ||
475		   (*p == ')' && (len != 4 || strncmp(id, "void", 4)))
476		   )
477		   return 0;	/* not a function */
478	   }
479	/*
480	 * If the last significant character was a ), we need to count
481	 * parentheses, because it might be part of a formal parameter
482	 * that is a procedure.
483	 */
484	if (contin > 0) {
485	    int level = 0;
486
487	    for (p = skipspace(buf, 1); *p; p = skipspace(p + 1, 1))
488		level += (*p == '(' ? 1 : *p == ')' ? -1 : 0);
489	    if (level > 0)
490		contin = -1;
491	}
492	return contin;
493}
494
495/* Convert a recognized function definition or header to K&R syntax. */
496int
497convert1(buf, out, header, convert_varargs)
498    char *buf;
499    FILE *out;
500    int header;			/* Boolean */
501    int convert_varargs;	/* Boolean */
502{	char *endfn;
503	register char *p;
504	/*
505	 * The breaks table contains pointers to the beginning and end
506	 * of each argument.
507	 */
508	char **breaks;
509	unsigned num_breaks = 2;	/* for testing */
510	char **btop;
511	char **bp;
512	char **ap;
513	char *vararg = 0;
514
515	/* Pre-ANSI implementations don't agree on whether strchr */
516	/* is called strchr or index, so we open-code it here. */
517	for ( endfn = buf; *(endfn++) != '('; )
518	  ;
519top:	p = endfn;
520	breaks = (char **)malloc(sizeof(char *) * num_breaks * 2);
521	if ( breaks == NULL )
522	   {	/* Couldn't allocate break table, give up */
523		fprintf(stderr, "Unable to allocate break table!\n");
524		fputs(buf, out);
525		return -1;
526	   }
527	btop = breaks + num_breaks * 2 - 2;
528	bp = breaks;
529	/* Parse the argument list */
530	do
531	   {	int level = 0;
532		char *lp = NULL;
533		char *rp = NULL;
534		char *end = NULL;
535
536		if ( bp >= btop )
537		   {	/* Filled up break table. */
538			/* Allocate a bigger one and start over. */
539			free((char *)breaks);
540			num_breaks <<= 1;
541			goto top;
542		   }
543		*bp++ = p;
544		/* Find the end of the argument */
545		for ( ; end == NULL; p++ )
546		   {	switch(*p)
547			   {
548			   case ',':
549				if ( !level ) end = p;
550				break;
551			   case '(':
552				if ( !level ) lp = p;
553				level++;
554				break;
555			   case ')':
556				if ( --level < 0 ) end = p;
557				else rp = p;
558				break;
559			   case '/':
560				if (p[1] == '*')
561				    p = skipspace(p, 1) - 1;
562				break;
563			   case '"':
564			       p = scanstring(p, 1) - 1;
565			       break;
566			   default:
567				;
568			   }
569		   }
570		/* Erase any embedded prototype parameters. */
571		if ( lp && rp )
572		  writeblanks(lp + 1, rp);
573		p--;			/* back up over terminator */
574		/* Find the name being declared. */
575		/* This is complicated because of procedure and */
576		/* array modifiers. */
577		for ( ; ; )
578		   {	p = skipspace(p - 1, -1);
579			switch ( *p )
580			   {
581			   case ']':	/* skip array dimension(s) */
582			   case ')':	/* skip procedure args OR name */
583			   {	int level = 1;
584				while ( level )
585				 switch ( *--p )
586				   {
587				   case ']': case ')':
588				       level++;
589				       break;
590				   case '[': case '(':
591				       level--;
592				       break;
593				   case '/':
594				       if (p > buf && p[-1] == '*')
595					   p = skipspace(p, -1) + 1;
596				       break;
597				   case '"':
598				       p = scanstring(p, -1) + 1;
599				       break;
600				   default: ;
601				   }
602			   }
603				if ( *p == '(' && *skipspace(p + 1, 1) == '*' )
604				   {	/* We found the name being declared */
605					while ( !isidfirstchar(*p) )
606					  p = skipspace(p, 1) + 1;
607					goto found;
608				   }
609				break;
610			   default:
611				goto found;
612			   }
613		   }
614found:		if ( *p == '.' && p[-1] == '.' && p[-2] == '.' )
615		  {	if ( convert_varargs )
616			  {	*bp++ = "va_alist";
617				vararg = p-2;
618			  }
619			else
620			  {	p++;
621				if ( bp == breaks + 1 )	/* sole argument */
622				  writeblanks(breaks[0], p);
623				else
624				  writeblanks(bp[-1] - 1, p);
625				bp--;
626			  }
627		   }
628		else
629		   {	while ( isidchar(*p) ) p--;
630			*bp++ = p+1;
631		   }
632		p = end;
633	   }
634	while ( *p++ == ',' );
635	*bp = p;
636	/* Make a special check for 'void' arglist */
637	if ( bp == breaks+2 )
638	   {	p = skipspace(breaks[0], 1);
639		if ( !strncmp(p, "void", 4) )
640		   {	p = skipspace(p+4, 1);
641			if ( p == breaks[2] - 1 )
642			   {	bp = breaks;	/* yup, pretend arglist is empty */
643				writeblanks(breaks[0], p + 1);
644			   }
645		   }
646	   }
647	/* Put out the function name and left parenthesis. */
648	p = buf;
649	while ( p != endfn ) putc(*p, out), p++;
650	/* Put out the declaration. */
651	if ( header )
652	  {	fputs(");", out);
653		for ( p = breaks[0]; *p; p++ )
654		  if ( *p == '\r' || *p == '\n' )
655		    putc(*p, out);
656	  }
657	else
658	  {	for ( ap = breaks+1; ap < bp; ap += 2 )
659		  {	p = *ap;
660			while ( isidchar(*p) )
661			  putc(*p, out), p++;
662			if ( ap < bp - 1 )
663			  fputs(", ", out);
664		  }
665		fputs(")  ", out);
666		/* Put out the argument declarations */
667		for ( ap = breaks+2; ap <= bp; ap += 2 )
668		  (*ap)[-1] = ';';
669		if ( vararg != 0 )
670		  {	*vararg = 0;
671			fputs(breaks[0], out);		/* any prior args */
672			fputs("va_dcl", out);		/* the final arg */
673			fputs(bp[0], out);
674		  }
675		else
676		  fputs(breaks[0], out);
677	  }
678	free((char *)breaks);
679	return 0;
680}
681