Re: letter occurence in a text

[gregory benison]

On Tue, May 22, 2012 at 7:26 AM, nrichard <address@hidden> wrote:
>
> hello my problem is to count occurence of letter in a text and come out with
> an assoc-list like
> '((a.16) (b.10) ... (z.5))

Why store the alist keys as one-character strings, rather than just as
characters?  Storing as characters would be simpler:

;; Given an alist 'lst' containing character counts and character 'c',
;; return an alist with the count of 'c' incremented (set to 1 if it
doesn't exist).
(define (lettre-test c lst)
  (let ((current (assoc-ref lst c)))
     (assoc-set! lst c (+ 1 (or current 0)))))

Character frequency analysis can be performed with a "fold" operation:

> (fold lettre-test '() (string->list "hello, world!"))
((#\! . 1)
 (#\d . 1)
 (#\r . 1)
 (#\w . 1)
 (#\space . 1)
 (#\, . 1)
 (#\o . 2)
 (#\l . 3)
 (#\e . 1)
 (#\h . 1))

I think it would be best to separate the filtering for alphabetic
chars from the "lettre-test" function; they're separate ideas:

> (fold lettre-test '() (filter char-alphabetic? (string->list "hello, 
> world!")))
((#\d . 1)
 (#\r . 1)
 (#\w . 1)
 (#\o . 2)
 (#\l . 3)
 (#\e . 1)
 (#\h . 1))

The drawback of this solution, as currently written, is that it can't
lazily read a file; you'd have to read the entire file into a string
first.  It should be possible to modify this to use streams rather
than lists, though.


-- 
Greg Benison <address@hidden>
[blog] http://gcbenison.wordpress.com
[twitter] @gcbenison