|
From the v9 documentation:
-------------------------------------------------------------
Altering the Record Format
Using the RECFM= option in the FILENAME, FILE, %INCLUDE, and INFILE
statements enables you to specify the record format of your external files.
The following example shows you how to use this option.
Usually, SAS reads a line of data until a carriage return and line feed
combination ('0D0A'x) are encountered or until just a line feed ('0A'x) is
encountered. However, sometimes data do not contain these carriage control
characters but do have fixed-length records. In this case, you can specify
RECFM=F to read your data.
To read such a file, you need to use the LRECL= option to specify the record
length and the RECFM= option to tell SAS that the records have fixed-length
record format. Here are the required statements:
data test;
infile "test.dat" lrecl=60 recfm=f;
input x y z;
run;
In this example, SAS expects fixed-length records that are 60 bytes long,
and it reads in the three numeric variables X, Y, and Z.
You can also specify RECFM=F when your data do contain carriage returns and
line feeds, but you want to read these values as part of your data instead
of treating them as carriage control characters. When you specify RECFM=F,
SAS ignores any carriage controls and line feeds and simply reads the record
length you specify.
------------------------------------------------------------
I'd suggest reading through the data with recfm=f, and replacing any
occurrence of CR with the CR-LF or just LF.
I know in Unix there's a DOS2Unix utility or something like that. I'd bet
there's similar MAC to PC utilities.
Paul Choate
DDS Data Extraction
(916) 654-2160
-----Original Message-----
From: Francis Harvey [mailto:HARVEYF1@WESTAT.COM]
Sent: Thursday, September 25, 2003 1:36 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: SAS and line breaks
Greetings,
I recently had an interesting file put before a program I wrote for
automatic processing that left me wondering what the best approach to
handling it is. This document was apparently written on a MAC as it
has lines that are broken only by carriage returns (CR) and then
reformatted on a PC with normal line breaks of carriage returns (CR)
followed by a line feed (LF).
When reading this file using SAS on a PC (which is where the
processing program runs), I initially PAD the source file to 100
characters (no line can be longer than that) and then load the data.
However, I am finding that SAS is not treating the CR without an LF as
a line break. My first inclination is to force the user to clean the
file to remove the problem before using my program, but I may be
overreacting. Another interesting point is that our Linux server may
become more widely used in the future or we may obtain UNIX generated
documents to reprocess which could cause there to be lines with an LF
without a matching CR.
So, would you force SAS to deal directly with such files, and if so,
how would you alter the data step to get them to process properly? I
am just going to assume that I will never see a file that has been
processed on both a MAC and a UNIX box as that would just seem to be
something I would have to talk to the individual about.
Francis R. Harvey III
WB 303, x3952
harveyf1@westat.com
VB programmers know the wisdom of Nothing
| 
