How can I fix this syntax error in prolog?

Question

I am writing in Prolog and keep getting the error;

syntax error: current or previous operator needs brackets

When referring to te=he following code:

convert_to_tokens([Word|Rest], [Token|RestTokens]) :-
    atom_string(Atom, Word),
    (   Atom == '+' -> Token = pl
    ;   Atom == '-' -> Token = mi
    ;   Atom == '*' -> Token = ti
    ;   Atom == '/' -> Token = di
    ;   number_string(Number, Word) -> Token = Number
    ;   atom_chars(Word, [H|T]), char_type(H, alpha) -> Token = id(Word)
    ;   Token = Atom
    ),
    convert_to_tokens(Rest, RestTokens).

I have tried using actual brackets, curly brackets, removing the brackets, etc. But I cannot seem to find a solution. Can anyone figure out what it is asking for? Thank you!

Look at the *entire* error message - I expect it includes a line number. — brebs, May 11 '23 at 16:28

false · Answer 1 · 2023-05-11T17:35:51.847

1

The brackets the error message refers to are round brackets. In place of Atom == '+' write Atom == (+) and same for -, *, / to make it ISO syntax. That specific error message was produced by gnu-prolog.

Also note that the code was probably written for SWI, since e.g. atom_string/2 is not ISO, but is present in SWI. All these conversions are only necessary in SWI btw.

edited May 11 '23 at 17:35

answered May 11 '23 at 17:29

false

10,264
13
101
209

Nicholas Carey · Answer 2 · 2023-05-11T17:49:37.523

I don't see that error. Once I remove your singleton variable warning, it just fails: https://swish.swi-prolog.org/p/chEjJzTM.pl

But if your Prolog is whining about infix (or other) operators like that, just enclose them in parenthesis:

( + )

But wouldn't it be cleaner and more, well, Prolog-like, to do something like this?

https://swish.swi-prolog.org/p/MSlYBdWG.pl

convert_to_tokens( []    , []      ) .
convert_to_tokens( [W|Ws] , [T|Ts] ) :-
    atom_string(A, W),
    map_token( A, T) ,
    !,
    convert_to_tokens(Ws,Ts).

map_token( + , pl    ) .
map_token( - , mi    ) .
map_token( * , ti    ) .
map_token( / , di    ) .
map_token( A , T     ) :- number_atom(A,T) .
map_token( A , id(A) ) :- sub_atom(A,0,1,_,C), char_type(C,alpha).
map_token( A , A     ) .

number_atom(N,A) :- atom_number(A,N) .

where

?- convert_to_tokens(
  [ "+" , "-" , "*" , "/" , "12345", "alpha" , "!@#" ] ,
  Tokens
) .

produces the expected

Tokens = [ pl , mi , ti , di , 12345 , id(alpha) , '!@#' ]

How can I fix this syntax error in prolog?

2 Answers2