I believe in 6.12 the RXPARSE argument must start with a grave aka backquote (`) 6.12 Prelim RX docs are at ftp://ftp.sas.com/pub/neural/rx612.txt

-- Richard DeVenezia - SAS Macros and AF Tools http://www.devenezia.com

"Chakravarthy, Venky" <Venky.Chakravarthy@PFIZER.COM> wrote in message news:2195003E3380D411965D00508B60994803909BE7@aascooby.research.aa.wl.com... > Mike, > > I ran your code (earlier version and the current) in v6.12. I expected the > code to work, since the RX functions were already experimental in v6.11 and > v6.12. Although it ran without ERROR messages, it generated NOTES with > "Invalid first argument" to the CALL RXCHANGE portion and thus did not > produce the expected results. Pasted, below my sig, is the relevant portion > of the log from v6.12. > > I could not find any documentation on the RX functions specific to V6.12. I > later learnt, from SAS tech support, that the preliminary documentation > provided around experimental status time, has since been incorporated into > the v7/v8 production docs. This is, of course, with updates for the bugs > fixed. So, if v6.12 has the RX bugs (that were later fixed), it could > explain the log. > > There appears to be one outstanding bug associated with the RX (purported to > be fixed in V9). If the source and target variables are the same, the > existing value is completely replaced by the TO string in the RXPARSE. This > is documented (see > http://www.sas.com/service/techsup/unotes/SN/005/005186.html). > > Thanks for introducing the RX functions. These and other v8 functions should > come in handy when this shop moves 8.2 to production. This should happen > soon. > > Kind Regards, > > Venky > #****************************************# > # E-mail: venky.chakravarthy@pfizer.com # > # swovcc@hotmail.com # > # Phone: (734) 622-1963 # > #****************************************# > > 1 data _null_; > 2 length old new $ 32; > 3 if _n_ = 1 then rx = rxparse > 4 (" <: ^' '> <' ' ' '*> <^' ' :> TO =1 ',' =3 "); > 5 retain rx; > 6 input old $char32.; > 7 call rxchange (rx,1,trim(left(old)),new); > 8 put / 'OLD: ' old $char32. / 'NEW: ' new $char32.; > 9 cards; > > NOTE: Invalid first argument to function RXCHANGE at line 7 column 8. > > OLD: one blank > NEW: > RULE:----+----1----+----2----+----3----+----4----+----5----+----6----+----7- > ---+----8----+----9-- > 10 one blank > OLD=one blank NEW= RX=1 _ERROR_=1 _N_=1 > NOTE: Invalid first argument to function RXCHANGE at line 7 column 8. > > <rest of log snipped> > > -----Original Message----- > From: Mike Rhoads [mailto:RHOADSM1@WESTAT.COM] > Sent: Monday, December 03, 2001 8:50 AM > To: SAS-L@LISTSERV.UGA.EDU > Subject: Re: Parsing a string > > > I wound up putting together a slightly different version of the program that > handles multiple replacements. (Recall that the original question was about > replacing strings of 2 or more blanks with a comma, while leaving single > blanks alone.) The revised program is below. A few notes on the pattern: > < > tags a part of the matching pattern, so that the part of the string that > is matched by that part of the pattern can be used in the replacement > expression (after the keyword TO) > =1 refers to the part of the string matched by the first tagged part of the > pattern > * 0 to many occurrences of what immediately precedes it > + 1 to many occurrences of what immediately precedes it > ( ) grouping > ^ any character EXCEPT those in the immediately following string > > Mike Rhoads > Westat > RhoadsM1@Westat.com > > > data _null_; > length old new $ 32; > if _n_ = 1 then rx = rxparse ( > /* Optional leading blanks followed by 1+ nonblanks */ > "< ' '* (^' ')+ >" || > /* Followed by 2 or more blanks */ > "< ' ' ' '* >" || > /* Replace with the first part unchanged, then a comma */ > "TO =1 ',' " > ); > retain rx; > input old $char32.; > call rxchange (rx,999,trim(old),new); > put / 'OLD: ' old $char32. / 'NEW: ' new $char32.; > cards; > 1 2 3 4 5 > 1 1 1 1 1 > 2 2 2 2 2 > 22 22 22 22 22 > 3 3 33 3 333 > 3 3 33 3 333 > one two three endofline > three two one zero > one blank > one blank > two blanks > three blanks > four blanks > two blanks > another, string. > ; > run;