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

Proposed amendments to SRFI-13



Hello everyone.  A few comments and suggestions on SRFI-13.

string-take
string-drop
string-take-right
string-drop-right
substring-move-left
substring-move-right

These (if needed) can all be defined as special cases of a more general
function, string-replace.

string-replace s1 start end s2 -> string
    replace the substring of s1 from start to end with s2

A variant, string-replace! could also be defined, which is allowed to reuse
the original storage of s1.

With a regexp package, regexp matching could used to find start and end
points, then string-replace could perform the replacement.

My sample code defines two equivalent versions of string-replace.
String-replace-simple does string replacement using substrings and append;
string-replace allocates a new string and copies characters into it.

------------

reverse-string-concatenate string-list [end] -> string
reverse-string-concatenate/shared string-list [end] -> string

Why are these included?  Without the optional parameter, the stated
equivalents are clearer.  With END these functions are odd -- why are they
needed?

------------

string-filter s char/char-set/pred [start end] -> string
string-delete s char/char-set/pred [start end] -> string

The char/char-set/pred isn't intuitive.  If char-sets are needed, perhaps
there should be a set package SRFI (is there one already?), or the
functionality could be provided by a regexp package.  Suggest dropping
string-delete and using an alternative version of string-filter which uses
a string -> string function f instead of char/char-set/pred.  (String
deletion can be done by f returning "".)  The function f should go first,
as with map and string-map.

string-filter f s [start end] -> string
    construct a string by applying f to each character of s
    function f should take a string of length one as input, and output a
string (of any length)

------------

Example code (written in MzScheme) is below.  Enjoy.

;;; dw-string.scm -- a few string manipulation functions
;;; these functions are for illustration, and lack error checking

; filter a string s using function f
(define (string-filter f s . endpoints)
  (let ((start 0)
        (end (string-length s)))
    (if (not (null? endpoints))
        (begin
          (set! start (car endpoints))
          (set! end (cadr endpoints))))
    (string-append
     (substring s 0 start)
     (apply string-append
            (map f
                 (map string
                      (string->list (substring s start end)))))
     (substring s end (string-length s)))))

; map a character-transforming function over a string
; does not take substring endpoints
(define (string-map f s)
  (list->string
         (map f
              (string->list s))))

; destructive version of string-map
; modifies the original string
; does not take substring endpoints
(define (string-map! f s)
  (do ((i (- (string-length s) 1) (- i 1)))
    ((< i 0) 'ok)
    (string-set! s i (f (string-ref s i)))))

; replace substring of s1 with s2
(define (string-replace-simple s1 start end s2)
  (string-append (substring s1 0 start)
                 s2
                 (substring s1 end (string-length s1))))

; string-copy!, needed for efficient version of replacement
; substring endpoints should be optional
(define (string-copy! target tstart s start end)
  (do ((i tstart (+ i 1))
       (j start (+ j 1)))
    ((= j end) 'ok)
    (string-set! target i (string-ref s j))))

; replacement, efficient version
; equivalent to string-replace-simple
(define (string-replace s1 start end s2)
  (let ((result
         (make-string
          (+ (string-length s1)
             (- start end)              ; this should be nonpositive
             (string-length s2)))))
    ; copy first part of s1 into result
    (string-copy! result 0 s1 0 start)
    ; copy s2 into result
    (string-copy! result start s2 0 (string-length s2))
    ; copy second part of s1 into result
    (string-copy! result (+ start (string-length s2))
                  s1 end (string-length s1))
    ; return result
    result))

;; the first two functions below demonstrate uses of string-filter
;; which could not be done by string-map

; test string-filter:  remove the vowels from a string
(define (remove-vowels s)
  (string-filter
   (lambda (s1)
     (if (member (string-ref s1 0)
                 (string->list "AEIOUaeiou"))
         ""
         (substring s1 0 1)))
   s))

; another example of string-filter
(define (local-echo s)
  (string-filter
   (lambda (s1)
     (string-append s1 s1))
   s))

;; example of using string-map

; rot13 of a character
(define (rot13 c)
  (cond ((char-upper-case? c)
         (integer->char
          (if (char<? c #\N)
              (+ (char->integer c) 13)
              (- (char->integer c) 13))))
         ((char-lower-case? c)
         (integer->char
          (if (char<? c #\n)
              (+ (char->integer c) 13)
              (- (char->integer c) 13))))
         (else c)))

; rot13 of a string
(define (string-rot13 s)
  (string-map rot13 s))