Re: a question about reading in alphabet letters

> Hi,
>
> I am practicing reading English sentences from a file, using get0/2. My
> codes are as follows:
>
> readInSentences :-
> open('sentences.txt', read, Stream, [encoding(utf8)]),
> readIn(Stream, List),
> close(Stream),
> write(List),nl,
> put_list(List).
>
> readIn(Stream, []) :-
> get0(Stream, -1).

This calls get0/2 -- these days you should probably
be using the ISO predicates get_char/1 or get_code/1 --
which reads a character, and then it tries to unify
that character with -1.  If you are not at the end of
the stream, that unification will fail.  At this
point, the character you just read is GONE FOREVER.

I can't think of *any* time that it makes sense to
call get0([Stream, ]Code) or get_code([Stream, ]Code)
or get_char([Stream, ]Char) with Code or Char anything
but a variable.  There should probably be a style
warning for it.  The thing is that when you learn that
the character *wasn't* what you were expecting, you
get *no* information about what it is.  The next call
to a character input command will read the *next*
character.

You should certainly use

read_in(Stream, List) :-
    get0(Stream, Code),
    (   Code = -1 ->
        List = []
    ; ....

or possibly

read_in(Stream, List) :-
    get0(Stream, Code),
    read_in(Code, Stream, List).

(This is a style I like.  Think of the predicates without
a Code argument as representing states in a deterministic
finite state automaton and the clauses of the predicate
_with_ a Code argument as representing the transitions.
Why deterministic?  Because the get... predicates don't
backtrack.)

> readIn(Stream, [Word|Tail]) :-
> get0(Stream, Word),

You *write* Word, but what you have is a character (code),
not a word.

> readIn(Stream, Tail).

You will find sentence reading commands in Clocksin & Mellish,
The Art of Prolog, The Craft of Prolog, the DEC-10 Prolog
library, and several other places.
>
> Could someone tell me why some of the alphabetic letters are not read? I
> use utf-8 because eventually I want to try my codes on Chinese texts.

This has nothing to do with UTF-8.