multiple alternatives regexp question

multiple alternatives regexp question

[Joe Corneli]

I am interested in a regexp that I think should be written
like this (for inclusion in a program):

"\\([[:lower:]]\\|[[:digit:]]\\|A\\)\\([A-Z]\\)"

In other words, a lower case letter, or a digit, or an `A',
followed by a capital letter.

Unfortunately, when I search and replace like so:

(let ((case-fold-search nil))
(while (re-search-forward "\\([[:lower:]]\\|[[:digit:]]\\|A\\)\\([A-Z]\\)"
                          nil t)
  (replace-match (concat (match-string 1) " " (match-string 2)))))

on something
like 

ProofOfPropertiesOfTraceOfAMatrix 

I get back

Proof Of Properties Of Trace Of AMatrix 
                                 ^

(Note that if you run the above code on the output string, it will
make a space in the appropriate place!)

Why is the `A' not being treated the way I want the first time
around?

Re: multiple alternatives regexp question

[Jesper Harder]

Joe Corneli <jcorneli@math.utexas.edu> writes:

> Unfortunately, when I search and replace like so:
>
> (let ((case-fold-search nil))
> (while (re-search-forward "\\([[:lower:]]\\|[[:digit:]]\\|A\\)\\([A-Z]\\)"
>                           nil t)
>   (replace-match (concat (match-string 1) " " (match-string 2)))))
>
> on something like
>
> ProofOfPropertiesOfTraceOfAMatrix 
>
> I get back
>
> Proof Of Properties Of Trace Of AMatrix 
>                                  ^
> Why is the `A' not being treated the way I want the first time
> around?

It doesn't work because `replace-match' leaves point after the
replacement text.  You could use something like this instead:

   (replace-match "\\1 " t nil nil 1)

Re: multiple alternatives regexp question

[Stefan Monnier]

> I am interested in a regexp that I think should be written
> like this (for inclusion in a program):

> "\\([[:lower:]]\\|[[:digit:]]\\|A\\)\\([A-Z]\\)"

You mean "\\([[:lower:][:digit:]A]\\)\\([[:upper:]]\\)" ?

> Why is the `A' not being treated the way I want the first time
> around?

See Jesper's answer,


        Stefan

Re: multiple alternatives regexp question

[Joe Corneli]

> You mean "\\([[:lower:][:digit:]A]\\)\\([[:upper:]]\\)" ?

I suppose that's equivalent to what I wrote, though I didn't
know could do that...

Anyway, I see why the replacement doesn't work the first
time & consequently why it does work the second time, so
thanks for the help!