RE: Number of arguments, a. la $# in gnu make?

[Rakesh Sharma]

This is a template for a make-based "tr" macro which transliterates from
listA to listB in the text.

Before running "tr" I intend to run some validations on the arguments to "tr".
The macro "-tr-validate" does that task and it is here that the number of 
arguments to a function's need was sorely felt. 

Also in this macro is it possible to factor in the $0 in the exit-code macro 
itself, so that we dont need to specify it in the invocation of exit-code macro 
everytime.? I tried various alternatives using $(eval....
but nothing worked.

################# gnu make code for ##############
#  %   make --version
# GNU Make 4.0
# Built for i686-pc-cygwin
###############################################

SHELL := /bin/sh
COMMA := ,

[a-z] := a b c d e f g h i j k l m n o p q r s t u v w x y z #
[A-Z] := A B C D E F G H I J K L M N O P Q R S T U V W X Y Z #

.PHONY: t

# message macro for the wrong argc
define tr-usage-1
BAD ARGS
*** incorrect number of arguments to function 'tr'.
*** valid number of arguments is 3 to function 'tr'.
*** Usage: \$$(call tr,source_list,target_list,string)
*** Ex: \$$(call tr,a b c d,A B C D,string_of_text)
endef

# message macro for an empty argv
define tr-usage-2
INVALID ARGS
*** invalid argument to function 'tr'.
*** cannot pass an empty argument to function 'tr'.
*** Usage: \$$(call tr,source_list,target_list,string)
*** Ex: \$$(call tr,a b c d,A B C D,text)
endef

# message macro for mismatched source & target argv
define tr-usage-3
SIZE MISMATCHED ARGS
*** mismatch sized arguments to function 'tr'.
*** first & second arguments size need to contain same number of elements
endef

# exit codes
E_BAD_ARGS        := 1
E_INVALID_ARGS    := 2
E_MISMATCHED_ARGS := 3

#$(call exit-code,$0,E_BAD_ARGS)
exit-code = $(error $($(join $(word 1,$(subst -, ,$1)),-usage-$($2))))

define -tr-validate
$(if $1$2$3,$(if $2$3,$(if $3$1,$(if $1$2,$(if $3,$(if $1,$(if $2,$(if 
$(filter-out $(words $1),$(words $2)),$(call 
exit-code,$0,E_MISMATCHED_ARGS)),$(call exit-code,$0,E_INVALID_ARGS)),$(call 
exit-code,$0,E_INVALID_ARGS)),$(call exit-code,$0,E_BAD_ARGS)),$(call 
exit-code,$0,E_INVALID_ARGS)),$(call exit-code,$0,E_INVALID_ARGS)),$(call 
exit-code,$0,E_BAD_ARGS)),$(call exit-code,$0,E_BAD_ARGS))
endef

define -tr-body
$(if $1,$$(subst $(firstword $1),$(call -tr-body,$(wordlist 2,$(words 
$1),$1),$2)),$2)
endef

#$(call tr,listA,listB,text)
define tr
$(strip \
    $(call -tr-validate,$1,$2,$3)\
    $(eval result := $(call -tr-body,$(join $(addsuffix $(COMMA),$1),$2),$3))\
    $(result))
endef

lc = $(call tr,$([A-Z]),$([a-z]),$1)
uc = $(call tr,$([a-z]),$([A-Z]),$1)

L := $([a-z])
M := $([A-Z])
S := MiXed_CaSe

## Test-suite for the "tr" macro
#
#                            AAA
#                            rrr
#                            ggg
#                            123
# x := $(call tr)          # xxx <--- 0 args --> E_BAD_ARGS
# x := $(call tr,$L)       # 1xx <--- 1 arg  --> E_BAD_ARGS
# x := $(call tr,,$M)      # x1x <--- 1 arg  --> E_INVALID_ARGS
# x := $(call tr,,,$S)     # xx1 <--- 1 arg  --> E_INVALID_ARGS
# x := $(call tr,$L,$M)    # 11x <--- 2 args --> E_BAD_ARGS
# x := $(call tr,,$M,$S)   # x11 <--- 2 args --> E_INVALID_ARGS
# x := $(call tr,$L,,$S)   # 1x1 <--- 2 args --> E_INVALID_ARGS
  x := $(call tr,$L,$M,$S) # 111 <--- 3 args --> E_MISMATCHED_ARGS or success

t:
    @echo "b4 transliteration=$(S)";
    @echo "af transliteration=$(x)";
#######################################################################


> Subject: Re: Number of arguments, a. la $# in gnu make?
> From: address@hidden
> To: address@hidden
> CC: address@hidden
> Date: Sat, 15 Feb 2014 09:46:58 -0500
> 
> On Sat, 2014-02-15 at 02:44 -0800, Rakesh Sharma wrote:
> > Is there a function/command for determining the number of arguments
> > passed to a user-defined function (== macro) in GNU make?
> 
> Unfortunately nothing like that exists today :-(
> 
> It's not exactly clear how one would use such a thing since the looping
> constructs in GNU make are quite anemic.  But we'd be interested to hear
> about your needs.
>