Date:         Wed, 27 Sep 2006 16:26:38 -0400
Reply-To:     Chang Chung <chang_y_chung@HOTMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Chang Chung <chang_y_chung@HOTMAIL.COM>
Subject:      Re: How to get "YET???" to become "YET?"
Comments: To: Charles Patridge <charles_s_patridge@PRODIGY.NET>

On Wed, 27 Sep 2006 11:34:50 -0400, Charles Patridge
<charles_s_patridge@PRODIGY.NET> wrote:

>Thanks for the assistance - I am going to use AG's solution - much shorter
>and to the point - sorry <grin>
>
>The following should do it.
>
>  data one;
>  input x $;
>    cards;
>    one?
>    two??
>    three???
>    four????
>    five?????
>  ;
>  run;
>
>   data two;
>     set one;
>     justone=prxchange('s/\?+/?/',-1,x);
>   run;
...

Hi,

I wouldn't go with the data step two above, since I am not sure if I have
explicit control over the length of the new variable, justone. So, I would
put a length statement (or a retain or attrib or just an assignment or
whatever...). Also the negative one (-1) does not make sense because the
goal seems to replace the trailing conjecutive multiple quotation marks into
one. If this is what you aim for and you should not do (-1) since it would
do much more. So, it comes down to something like:

  data three;
    set one;
    length justone $8;
    justone=prxchange('s/\?+/?/',1,x);
  run;

Now given this neccessary modification, it looks so much like (sas) rx solution:

  data four;
    set one;
    length justone $8;
    call rxchange(" '?'+ to '?' ", 1, x, justone);
  run;

Or, it seems to me that the sas rx looks much easier to read than prx
solution. :-)

I totally agree with data _null_ in that the guy who wrote sas rx is not bad
at all. I rather feel that si lost an opportunity to make sas's version of
rx much more powerful by switching to perl rx. By definition, perl rx is
made to work in perl not in sas. So perl rx in sas will always be something
short of the real perl rx running inside perl. si could have extended sas rx
so that it could take full advantage of sas such as calling sas data step
functions or statements inside the rx, for example. If compatibility was an
issue, then si could have made the sas rx an open source and given it away
free, too, as perl did... I know I am dreaming :-)

Cheers,
Chang