Is it possible to write a C macro that returns the number of its arguments?
I want something that does:
foo(1) -> 1
foo(cat, dog) -> 2
foo(red, green, blue) -> 3
Even better if this macro can be defined in such a way that it works with ## so that
foo(1) -> bar1(1)
foo(cat, dog) -> bar2(cat, dog)
foo(red, green, blue) -> car3(red, green, blue)
Thanks!
EDIT: I really want a macro, not a function. Suggestions to use functions will be downvoted.
It can be done - the mechanism was explained in the comp.std.c newsgroup in January 2006. There was another question about this recently on SO 2124339.
I stashed the code away, just in case...
#ifndef JLSS_ID_NARG_H
#define JLSS_ID_NARG_H
/*
** http://groups.google.com/group/comp.std.c/browse_thread/thread/77ee8c8f92e4a3fb/346fc464319b1ee5?pli=1
**
** Newsgroups: comp.std.c
** From: Laurent Deniau <laurent.deniau@cern.ch>
** Date: Mon, 16 Jan 2006 18:43:40 +0100
** Subject: __VA_NARG__
**
** A year ago, I was asking here for an equivalent of __VA_NARG__ which
** would return the number of arguments contained in __VA_ARGS__ before its
** expansion. In fact my problem at that time (detecting for a third
** argument) was solved by the solution of P. Mensonides. But I was still
** thinking that the standard should have provided such a facilities rather
** easy to compute for cpp.
**
** This morning I had to face again the same problem, that is knowing the
** number of arguments contained in __VA_ARGS__ before its expansion (after
** its expansion can always be achieved if you can do it before). I found a
** simple non-iterative solution which may be of interest here as an answer
** to who will ask in the future for a kind of __VA_NARG__ in the standard
** and I post it for archiving. May be some more elegant-efficient solution
** exists?
**
** Returns NARG, the number of arguments contained in __VA_ARGS__ before
** expansion as far as NARG is >0 and <64 (cpp limits):
**
** #define PP_NARG( ...) PP_NARG_(__VA_ARGS__,PP_RSEQ_N())
** #define PP_NARG_(...) PP_ARG_N(__VA_ARGS__)
** #define PP_ARG_N(_1,_2,_3,_4,_5,_6,_7,_8,_9,[..],_61,_62,_63,N,...) N
** #define PP_RSEQ_N() 63,62,61,60,[..],9,8,7,6,5,4,3,2,1,0
**
** [..] stands for the continuation of the sequence omitted here for
** lisibility.
**
** PP_NARG(A) -> 1
** PP_NARG(A,B) -> 2
** PP_NARG(A,B,C) -> 3
** PP_NARG(A,B,C,D) -> 4
** PP_NARG(A,B,C,D,E) -> 5
** PP_NARG(A1,A2,[..],A62,A63) -> 63
**
** ======
**
** Newsgroups: comp.std.c
** From: Roland Illig <roland.il...@gmx.de>
** Date: Fri, 20 Jan 2006 12:58:41 +0100
** Subject: Re: __VA_NARG__
**
** Laurent Deniau wrote:
** > This morning I had to face again the same problem, that is knowing the
** > number of arguments contained in __VA_ARGS__ before its expansion (after
** > its expansion can always be achieved if you can do it before). I found a
** > simple non-iterative solution which may be of interest here as an answer
** > to who will ask in the future for a kind of __VA_NARG__ in the standard
** > and I post it for archiving. May be some more elegant-efficient solution
** > exists?
**
** Thanks for this idea. I really like it.
**
** For those that only want to copy and paste it, here is the expanded version:
**
** // Some test cases
** PP_NARG(A) -> 1
** PP_NARG(A,B) -> 2
** PP_NARG(A,B,C) -> 3
** PP_NARG(A,B,C,D) -> 4
** PP_NARG(A,B,C,D,E) -> 5
** PP_NARG(1,2,3,4,5,6,7,8,9,0, // 1..10
** 1,2,3,4,5,6,7,8,9,0, // 11..20
** 1,2,3,4,5,6,7,8,9,0, // 21..30
** 1,2,3,4,5,6,7,8,9,0, // 31..40
** 1,2,3,4,5,6,7,8,9,0, // 41..50
** 1,2,3,4,5,6,7,8,9,0, // 51..60
** 1,2,3) -> 63
**
**Note: using PP_NARG() without arguments would violate 6.10.3p4 of ISO C99.
*/
/* The PP_NARG macro returns the number of arguments that have been
** passed to it.
*/
#define PP_NARG(...) \
PP_NARG_(__VA_ARGS__,PP_RSEQ_N())
#define PP_NARG_(...) \
PP_ARG_N(__VA_ARGS__)
#define PP_ARG_N( \
_1, _2, _3, _4, _5, _6, _7, _8, _9,_10, \
_11,_12,_13,_14,_15,_16,_17,_18,_19,_20, \
_21,_22,_23,_24,_25,_26,_27,_28,_29,_30, \
_31,_32,_33,_34,_35,_36,_37,_38,_39,_40, \
_41,_42,_43,_44,_45,_46,_47,_48,_49,_50, \
_51,_52,_53,_54,_55,_56,_57,_58,_59,_60, \
_61,_62,_63, N, ...) N
#define PP_RSEQ_N() \
63,62,61,60, \
59,58,57,56,55,54,53,52,51,50, \
49,48,47,46,45,44,43,42,41,40, \
39,38,37,36,35,34,33,32,31,30, \
29,28,27,26,25,24,23,22,21,20, \
19,18,17,16,15,14,13,12,11,10, \
9, 8, 7, 6, 5, 4, 3, 2, 1, 0
#endif /* JLSS_ID_NARG_H */
It works fine as long as there are no more than 64 arguments. Here's the test code I used:
#include "narg.h"
#include <stdio.h>
#define PRINT(pp_narg) printf("%2d = %s\n", pp_narg, # pp_narg)
#ifndef lint
/* Prevent over-aggressive optimizers from eliminating ID string */
extern const char jlss_id_narg_c[];
const char jlss_id_narg_c[] = "@(#)$Id: narg.c,v 1.2 2010/01/24 18:12:05 jleffler Exp $";
#endif /* lint */
int
main(void)
{
PRINT(PP_NARG(A));
PRINT(PP_NARG(A, B));
PRINT(PP_NARG(A, B, C));
PRINT(PP_NARG(A, B, C, D));
PRINT(PP_NARG(A, B, C, D, E));
PRINT(PP_NARG(1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 1..10
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 11..20
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 21..30
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 31..40
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 41..50
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 51..60
1, 2, 3));
/**
** If the number of arguments to PP_NARG() is greater than 63, the
** 64th argument is returned. This is well-defined behaviour, but
** not exactly what was intended.
*/
PRINT(PP_NARG(1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 1..10
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 11..20
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 21..30
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 31..40
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 41..50
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 51..60
1, 2, 3, -123456789));
PRINT(PP_NARG(1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 1..10
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 11..20
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 21..30
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 31..40
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 41..50
1, 2, 3, 4, 5, 6, 7, 8, 9, 0, // 51..60
1, 2, 3, -123456789, -987654321));
return(0);
}
I realize this is a pretty old question, but as no one ever answered the 'bonus' to add the number to the function names, here is an example of that part with Jonathan's answer reduced to 9 args for brevity. It assumes you have predefined the numbered functions, or will use this as a basis to define them.
#define MKFN(fn,...) MKFN_N(fn,##__VA_ARGS__,9,8,7,6,5,4,3,2,1,0)(__VA_ARGS__)
#define MKFN_N(fn,n0,n1,n2,n3,n4,n5,n6,n7,n8,n,...) fn##n
#define myfunc(...) MKFN(myfunc,##__VA_ARGS__)
myfunc();
myfunc(a,b,c,d,e,f);
//gcc -E this.c
//myfunc0();
//myfunc6(a,b,c,d,e,f);
one place where I have seen these types of functions was in the Linux kernel's syscalls, but the numbers would be off by one because the first argument is the syscall number (usually defined as __NR_something), so here is an example that accounts for that.
#define MKFN(fn,...) MKFN_N(fn,##__VA_ARGS__,9,8,7,6,5,4,3,2,1,0)(__VA_ARGS__)
#define MKFN_N(fn,NR,n0,n1,n2,n3,n4,n5,n6,n7,n8,n,...) fn##n
#define syscall(...) MKFN(syscall,##__VA_ARGS__)
syscall(__NR_fork);
syscall(77,a,b,c,d,e,f);//fake example
I use following macro:
#define NUMARGS(...) (sizeof((int[]){__VA_ARGS__})/sizeof(int))
Please note that it works only for C99, because variadic macros were not supported in C89. Although it doesn't work for zero arguments.
But if you are using GCC, you can use slightly modified macro:
#define NUMARGS(...) (sizeof((int[]){0, ##__VA_ARGS__})/sizeof(int)-1)
It works correctly even for zero arguments, because GCC's preprocessor removes extra comma when you are pasting empty __VA_ARGS__.
