Date:         Fri, 26 Jan 2007 10:47:25 -0500
Reply-To:     Arthur Tabachneck <art297@NETSCAPE.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Arthur Tabachneck <art297@NETSCAPE.NET>
Subject:      Re: PRX compared with older functions
Comments: To: Alan Churchill <SASL001@SAVIAN.NET>

Alan,

Thanks for taking the time to come up with a working pure prx solution.
However, as is, it takes much longer than Toby's mixed solution.

To level the playing field, I eliminated your macro calls for year and
day, and increased the number of records to 6 million.

The results were:

Pure PRX: real time: 2:17.64           cpu time: 36.95 seconds
Mixed:    real time  13.78 seconds     cpu time: 13.78 seconds

Art
---------
On Thu, 25 Jan 2007 22:16:05 -0700, Alan Churchill <SASL001@SAVIAN.NET>
wrote:

>%macro parse (abbr,unit) ;
>  re = prxparse("/(\d{1,2})(?:&abbr)/o") ;
>  if prxmatch(re,var) then
>     &unit = prxposn(re,1,var);
>%mend ;
>
>data test (drop=re);
>  input var $11. ;
>  %parse(y,year) ;
>  %parse(d,day) ;
>  %parse(h,hour) ;
>  %parse(m,min) ;
>  %parse(s,sec) ;
>  datalines;
>w2y
>w34y4h
>w38h45m23s
>p5d23m
>p61h
>w56s
>;
>run;
>
>
>Alan Churchill
>Savian "Bridging SAS and Microsoft Technologies"
>www.savian.net
>
>
>
>-----Original Message-----
>From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of
Arthur
>Tabachneck
>Sent: Thursday, January 25, 2007 8:56 PM
>To: SAS-L@LISTSERV.UGA.EDU
>Subject: PRX compared with older functions
>
>Last night Toby offered a partial prx solution to a scanning need that was
>raised, namely to find the number(s) that followed the characters h, m and
>s in a given set of text.
>
>data test;
>  input var $11. ;
>  datalines;
>w2y
>w34y4h
>w38h45m23s
>p5d23m
>p61h
>w56s
>;
>run;
>
>Data Need ( Keep = Var Sec Min Hour ) ;
>  Set Test ;
>  Retain Pattern ;
>  If _N_ = 1 Then Pattern = PRXParse( "/\d+[smh]/" ) ;
>  Start = 1 ;
>  Stop = Length( Var ) ;
>  Call PrxNext( Pattern , Start , Stop , Var , Match , Length ) ;
>  Do I = 1 By 1 While( ( Match > 0 ) and ( I < Stop ) ) ;
>    Temp = Substr( Var , Match , Length ) ;
>    If Index( Temp , 's' ) Then
>     Sec = Input( Compress( Temp , 's' ) , 8. ) ;
>    Else If Index( Temp , 'm' ) Then
>     Min = Input( Compress( Temp , 'm' ) , 8. ) ;
>    Else If Index( Temp , 'h' ) Then
>     Hour = Input( Compress( Temp , 'h' ) , 8. ) ;
>    Call PrxNext( Pattern , Start , Stop , Var , Match , Length ) ;
>  End ;
>Run ;
>
>A pure prx solution and possible refinement were then offered, but one
>didn't work and the other took twice as much time to complete then Toby's
>mixed solution.
>
>Can anyone offer a pure prx solution that runs faster than Toby's mixed
>offering?
>
>Art