Practical Prolog - tokenizing, mapping, and transforming text
Terrence Brannon
I am aiming to make a complete Prolog version of a Perl module that
generates latin-looking text:
http://search.cpan.org/~adeola/Text-Lorem-0.3/lib/Text/Lorem.pm
My first goal is to take a string of latin-looking text and
1. split the string into a list of strings by considering whitespace
as a delimiter.
2. take the list of strings and make each string lowercase
3. remove any character from any string which is not a POSIX "word"
character, the POSIX word characters being: A-Za-z0-9 and underscore.
Practically, let's say we have the initial string of latin-looking
text like this:
lorem_text("ipsum? lorem! lingua romana perligata.").
To satisfy goal #1 above, we need a predicate
tokenize(String, ListOfStrings).
To satisfy goal #2, we need some mapping predicate which applies a
predicate that lowercases text.
Goal #3 requires a mapping predicate which applies a
predicate applies regular expression substition (or something similar)
to a string
Concretely, the generate_wordlist Perl function does all 3 goals. The
source code, for it is here:
http://gitorious.org/text-lorem/text-lorem/blobs/master/lib/Text/Lorem.pm#line17
The full latin-looking text, in Prolog is here:
http://gitorious.org/text-lorem/text-lorem/blobs/master/prolog/lorem.prolog
I request help on the Prolog version of generate_wordlist, per the
spec above. You are welcome to post here or join the gitorious project
as a contributor.
YauHsienHuang
> My first goal is to take a string of latin-looking text and
>
> 1. split the string into a list of strings by considering whitespace
> as a delimiter.
>
> 2. take the list of strings and make each string lowercase
>
> 3. remove any character from any string which is not a POSIX "word"
> character, the POSIX word characters being: A-Za-z0-9 and underscore.
>
The goal #1 is similar with the solution of Problem #9 of P-99,
https://prof.ti.bfh.ch/hew1/informatik3/prolog/p-99/ .
The goal #2 and #3 are simple recursive predicates.
The Quiet Center
wrote:
> > 1. split the string into a list of strings by considering whitespace
> > as a delimiter.
>
> > 2. take the list of strings and make each string lowercase
>
> > 3. remove any character from any string which is not a POSIX "word"
> > character, the POSIX word characters being: A-Za-z0-9 and underscore.
>
> The goal #1 is similar with the solution of Problem #9 of P-99,https://prof.ti.bfh.ch/hew1/informatik3/prolog/p-99/.
It took me a long long time to do it, but I did it:
http://gitorious.org/text-lorem/text-lorem/blobs/master/prolog/split_on_char.prolog
Now, I'd like to make it a module to import into my main program
>
> The goal #2 and #3 are simple recursive predicates.
Easy for you to say :)