Date:         Wed, 18 Apr 2007 08:29:41 -0700
Reply-To:     aleph <donald.owen@CHOICEPOINT.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         aleph <donald.owen@CHOICEPOINT.COM>
Organization: http://groups.google.com
Subject:      Reading delimited data on multiple lines
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

I am trying to input a delimited file in which some (not all)
observations
are on multiple lines; the number of lines is unpredictable. [SAS
9.1.3 UNIX]
The following example code produces the desired output:

DATA OK;
INFILE DATALINES DLM="#" DSD;
INPUT W X A :$13. Y Z;
DATALINES;
1#2#HELLO KIDDO##8
3#4#GOODBYE KIDDO#9#7
5#6#MAMA WINS###
;
RUN;

CORRECT OUTPUT
   Obs  W   X                    A           Y    Z

    1     1    2    HELLO KIDDO         .     8
    2     3    4    GOODBYE KIDDO   9    7
    3     5    6    MAMA WINS           .     .



This code reads sample data that closely resembles the actual data I'm
working with:

DATA NOTOK;
INFILE DATALINES DLM="#" DSD;
INPUT W X A :$13. Y Z;
DATALINES;
1#2#HELLO KIDDO##8
3#4#GOODBYE
KIDDO#9#7
5#6#MAMA WINS###
;
RUN;

INCORRECT OUTPUT
    Obs  W   X                  A        Y    Z

     1     1    2    HELLO KIDDO    .    8
     2     3    4    GOODBYE         .    9
     3     5    6    MAMA WINS      .    .

Note that the value for A in obs 2 has 2 errors; all other values are
correct.
The actual raw data set includes 37 vars and 32 obs - I can fix the
problem
in this case by manually deleting LFs until each obs is on a single
line, but
I need to produce code that will handle potentially much larger files.

I have tried a number of tricks but nothing even comes close to
resolving
the problem.

I would be most grateful for any insight.

Thanks,
Don