Prolog - Appending to an existing file term list - list

Question

I have a term contained in a file:
fruit(apple, []).
I'm trying to write input to it, so that each set of information will be appending to the empty list as a list to compose a list of lists, in a manner like:
fruit(apple, [[30, 'fresh'], [10, 'old'], ... ]).
So far I only know how to write a separate term:
add_fruit(File, Type, Price, State) :-
open(File, append, Stream),
writeq(Stream, fruit(Type, [Price, State])),
write(Stream, '.'),
nl(Stream), nl(Stream),
close(Stream).
This will produce a term such as:
fruit(apple, [20, 'some state']).
However I don't want to have to create a new term every time I use the add_fruit predicate, rather I want to append a list containing Price and State to the appropriate fruit type term.
Would it be possible to extend the predicate so that it will write to the existing term, rather than create a new one?

Answer 1

You are mixing up concepts here a bit. First off, if you have a Prolog file, say "food.pl" with the contents:
fruit(apple, []).
then fruit/2 is not exactly a term, but rather the fact fruit/2 (predicate without a body, only with a head) which maps the atom apple (first argument) to the empty list (the second argument).
What you can do now is consult the file, which will put the fact in the database.
Another thing: are you sure you need to keep appending to the list in the second argument? Why not normalize your database, and have, for example:
fruit(apple, price_state(30, fresh)).
fruit(apple, price_state(10, old)).
% etc
Here, you now have a table, fruit, with two columns. The first is the name of the fruit, the second is a term price_state/2 with the price as the first argument and the state as the second. You could have just as well said:
fruit_price_state(apple, 30, fresh).
fruit_price_state(apple, 10, old).
% etc
if you indeed want your table to map the fruit to a state and a price. You can now add and remove rows from the table using assertz and retract. In other words, design your database as you would design a relational database, using facts as tables. This approach translates directly to Prolog. You can then query the database, for example, to give you the combination of price and state for apples:
?- bagof(price_state(P, S), fruit(apple, P, S), PSs).
This will put a list of price_state/2 in PSs, or fail, if you don't have apples in your database (you could use findall/3 if you prefer to get an empty list instead, but then you would have to deal with that empty list in any predicate that takes PSs).
One way to deal with persistence, that is, load facts to the database, add/remove rows, and then save the updated database back to the same file is using Edinburgh-style I/O with see, tell, told etc.
If you are using SWI-Prolog, you also have the option of using library(persistency). See the link for a working minimal example.