[Prev][Next][Index][Thread]

Re: Strtok/split?

In article <3A3F90D8.69545A8A@h2g2.com>,
  robm@h2g2.com wrote:
>
> What's the best way of reproducing the functionality of C's strtok()
> and/or Perl's split() in Dylan?
> I need to tokenize a string.

There is a function tokenize-environment-variable in
"Functional Developer\Sources\lib\operating-system\tokenize-
variable.dylan"
which does nearly the same as strtok, you only need to copy it and give
more universal names to the function, the parameter, and the delimiter.

It's working, but I don't like it very much, because it is implemented
in a pure imperative style. Here is a different version in a pure
functional style:


define method tokenize-string
  (string :: <string>, #key delimiter :: <character> = ' ')
  => (strings :: <sequence>);
  let len = size (string);
  for (i from len to -1 by -1,
       match? :: <boolean> = #f then i < 1 | string[i - 1] = delimiter,
       collect :: <list> = #() then
           if (match? & i + 1 < bound)
             let token = copy-sequence (string, start: i+1, end: bound);
             pair (token, collect)
           else
             collect
           end,
       bound :: <integer> = len then if (match?) i else bound end)
    finally collect
  end
end method;


Surely someone with more experience in functional programming can do it
better. If anybody has any ideas for improvements, please post them, so
that I can learn from it.

---
Helmut


Sent via Deja.com
http://www.deja.com/
References: