Date:         Tue, 31 May 2005 07:12:03 -0700
Reply-To:     chris <fast_rabbit@GMX.CH>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         chris <fast_rabbit@GMX.CH>
Organization: http://groups.google.com
Subject:      Re: documentation of Perl regex directly in the code
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

Richard A. DeVenezia wrote:



> Rune:
>
> One perlish way to comment uses trailing #mycomment and the /x option.  This
> does not work in SAS.  Your best bet is probably to use CATS function with
> SAS comments.  Note: CATS trims spaces, so blending in a space only can be
> frustrating.  Use \x20 to match a space when build a regex pattern using
> CATS.
>
> In this sample I cheat a little.  I use a capture group (\x20) to store a
> space and then use it in the replacement.  This was only done because CATS
> trims out leading and trailing spaces and the original replacement part
> would have been reduced to $2$1 without a space.
>
> data namechange;
>     set staff;
>     re = prxparse (
> CATS
> ('s'      /* Perform a substitution */
> ,'/'      /* Start regular expression*/
> ,'('      /* Start capture buffer # 1 to store the last name*/
> ,'[^,]+'  /* match one or more non-comma characters */
> ,')'      /* end capture buffer # 1 */
> ,','      /* match a comma*/
> ,'(\x20)' /* match a space*/
> ,'('      /* start capture buffer # 2 to store the first name*/
> ,'\w+'    /* match a word character one or more times */
> ,'('      /* Start capture buffer # 3. It is part of buffer # 2*/
> ,'\s+'    /* match a white space*/
> ,'\w+'    /* match a word character one or more times*/
> ,')'      /* end capture buffer # 3, hold first name and middle name*/
> ,'?'      /* match zero or one time*/
> ,')'      /* end capture buffer # 2*/
> ,'/'      /* end regular expression and start replacement text*/
> ,'$3'     /* insert captured buffer # 2*/
> ,'$2'     /* insert a space*/
> ,'$1'     /* insert capture buffer # 1 */
> ,'/'      /* end replacement text*/
> )
> );
>     NewName = prxchange(re,1,Name);
> run;
>
> --
> Richard A. DeVenezia
> http://www.devenezia.com/

This is a good way to do this, but does it really help readability?
I would stick with the initial regex and document is seperately on a
new line:

re = prxparse( 's/([^,]+), (\w+(\s+\w+)?) /$2 $1/');
/*  regex to reverse a string up to a comma with one or two words
following
    the comma
   ([^,]+)         capture one or more non-comma character into $1
    ,              match a blank and a comma
    (\w+(\s+\w+)?) capture one or two blocks of characters
                   separated by blanks into $2
*/

chris