topic Re: Problem With Bad Data in SAS Procedures

Problem With Bad Data

OS2Rules — Thu, 27 Sep 2012 15:49:48 GMT

Hi All:

Not sure whether to post in this forum, but it seems the best fit.

I am running the following code reading a not too large file:

data work01;

infile "my\file\name.txt" lrecl=400 pad missover end=eof;

input @001 var01 $char400.;

There is a lot of "bad" data in the file - commas, special characters, etc.

My problem is that when a single quote appears in the data (as in the name: O'Leary) the read seems to stop

at that point and I get no more data. This is only about half way through the data.

There is no indication in the log that any problem was encountered.

Any ideas how I can get past this? I need to keep the data as it is (bad characters and all) for the output.

Thanks in advance.

Re: Problem With Bad Data

data_null__ — Thu, 27 Sep 2012 16:04:46 GMT

Maybe you need INFILE statement option IGNOREDOSEOF.

Re: Problem With Bad Data

OS2Rules — Thu, 27 Sep 2012 16:14:50 GMT

Correct on the first try!

Wonder why this option has anything to do with what I assume is quoted text?

Re: Problem With Bad Data

data_null__ — Thu, 27 Sep 2012 16:32:32 GMT

It is not a quote just look like it. Use LIST statement and if it is unprintable control character it will show the line in HEX and you can see what the character actually is.

Re: Problem With Bad Data

TomKari — Fri, 28 Sep 2012 17:18:41 GMT

I have often hit problems where someone has used a document editor like Microsoft Word to prepare data. Word will automatically change many characters, for instance the double quote " character to what it calls smartquotes, which have different open and close characters. SAS frequently coughs up a hairball when it sees these.

Tom

Re: Problem With Bad Data

Doc_Duke — Sat, 29 Sep 2012 18:55:59 GMT

The control character was probably a (ctrl-Z). Back in the days of DOS, and version 6 of PC SAS, that was still in use and has been maintained in SAS as an EOF character of for PC SAS since. SAS will read most other control characters into a character variable.

Doc Muhlbaier

Duke