Re: bind-string-match: comments please

[weber]

On 30 set, 15:30, Andreas Politz <poli...@fh-trier.de> wrote:
> weber wrote:
> > I've written this helper function:
>
> > (defun bind-string-match (regexp string &optional start)
> >   (let ((ret (string-match regexp string start)))
> >    (when ret
> >      (catch 'end
> >            (dolist (i (number-sequence 1 10))
> >              (let ((match (match-string i string)))
> >                    (if match
> >                            (eval `(setq ,(intern (concat "$" 
> > (number-to-string i))) ,match))
> >                            (throw 'end t))))))
> >    ret))
>
> > so that I can access the match groups with $1.. instead of (match-
> > string 1 string).
> > Is there any recommendation against this type of functions? Other than
> > polluting the global namespace?
>
> I think sooner or later you get into trouble, if you don't clear
> the match variables. They could still hold values from 10 matches
> ago, while the current regexp doesn't match at all.
>
> > Also, can one rewrite it without the `eval'?
>
> The 'q' in setq stands for quoted. So, just using set should
> work.
>
>
>
> > TIA,
> > weber
>
> > PS: I'm not a Perl programmer!
>
> Maybe what you really want is some kind of macro ?
>
> (defmacro with-string-matches (string &rest body)
>    `(let ,(mapcar (lambda (submatch)
>                      (list (intern (format "$%d" submatch))
>                            (match-string submatch (eval string))))
>                 (number-sequence 0 10))
>       ,@body))
>
> (let ((str "abbbbc") idx)
>    (setq idx (string-match "a\\(b+\\)c" str))
>    (with-string-matches str
>      (message "%s %s" $0 $1)))
>
> -ap

Yeah, I was drifting that way after recalling pg's anaphoric
macros....
Thanks a lot :)
-weber