|
I was thinking along the same lines. I suggest the following modifications
to the first DATA step, to avoid duplicate observations, to handle wrapping
beyond a second line, and to pick up the last observation:
Code END=DONE on the INFILE statement. This precludes the use of inline
(CARDS;) data. Instead the records must be in an external file.
Eliminate the first PUT statement.
Insert the statements
if done then put line;
run;
at the end.
On Fri, 12 Sep 2003 15:04:19 -0400, Ian Whitlock <WHITLOI1@WESTAT.COM>
wrote:
>jsl,
>
>I would make it a two step process, one to fix the file and one t read the
>fixed file. This works with the data given. You may need to change
lengths
>when more data is considered. If one logical record is more than two lines
>then the first step will need more changes.
>
>filename temp temp ;
>data _null_ ;
> length line $ 200 ;
> retain line ;
> infile cards firstobs = 2 ;
> input ;
> file temp ;
> if substr(_infile_,1,2) ^= " " then
> do ;
> line = trim(line) || _infile_ ;
> put line ;
> end ;
> else
> do ;
> if _n_ > 1 then put line ;
> line = left(_infile_) ;
> end ;
>cards ;
>..........
>;
>
>data w ;
> infile temp dsd dlm = "/" ;
> input name :$30. age :5. title :$70. remuneration :$20. ;
>run ;
>
>IanWhitlock@westat.com
>-----Original Message-----
>From: jsl [mailto:nospam@NOSPAM.COM]
>Sent: Friday, September 12, 2003 2:44 PM
>To: SAS-L@LISTSERV.UGA.EDU
>Subject: Data with Carriage return where I don't want it.
>
>
>I have a very large "text file" data set that has a bunch of pockets of
>information that looks like the attached text file. I have been able to
>read it just fine for the most part keying off words in the first line, but
>I can't figure out the easiest way to deal with the unwanted hard return.
>That is, given the source of the data, I have no control over the unwanted
>carriage return within a record. However, I do know that the first two
>columns are always blank unless it's a record that wrapped. Anybody have
>simple suggestion on how to keep those lines together? Thanks, Jim
|
