To avoid inserting blanks into non-null fields, and also handle runs of three or more consecutive tabs, operate only on adjacent delimiters and use two nested invocations of TRANWRD.

Something like (not tested):

_infile_ = tranwrd(tranwrd(_infile_,'0909'x,'092009'x),'0909'x,'092009'x);

where 09 is the ASCII hex code for the tab and 20 is the ASCII hex code for the blank.

See http://www.listserv.uga.edu/cgi-bin/wa?A2=ind0205A&L=sas-l&P=R1837

On Tue, 30 Jul 2002 20:18:23 -0500, Kevin Myers <KevinMyers@AUSTIN.RR.COM> wrote:

>One possibility: Use the tranwrd function to translate each delimiter >character into a delimiter character either preceded or followed by some >non-delimiter character (such as a space) that wouldn't interfere with your >actual data. That would allow you to use the scan function without >interference from consecutive delimiters. For example (non-tested code!): > >myvar=scan(tranwrd(_infile_,'09'x,'2009'x),fieldnum,'09'x); > >HTH. > >s/KAM > > >----- Original Message ----- >From: "Walter Shafron" <waltshaf@ABARE.GOV.AU> >Newsgroups: bit.listserv.sas-l >To: <SAS-L@LISTSERV.UGA.EDU> >Sent: Tuesday, July 30, 2002 7:20 PM >Subject: scan function with dsd option > > >> I am reading a series of tab-delimited files, about 1000 characters >> wide and want to read/extract certain fields out of this. >> >> One way would be to jump columns using data step pointer controls, but >> even with tab-delimited files the columns work in spaces (I need field >> skips, not column skips) >> . >> Another idea was to read the wholeline (using the _infile_ variable) >> and scan for the fields that I wanted. Unfortunately the scan function >> treats contiguous delimiters as one. >> >> Any ideas would be appreciated. >> >> Walter Shafron