Prev: competing risks proportional
Next: Latent Class Analysis - Question
From: Joe Matise on 30 Nov 2009 13:12
That is easy to deal with:
1. Change all quotation marks to a character not present in your data.
2. Change all pipes to quotation marks.
3. Import through normal DSD rules.
4. Revert the quotation marks.

I wonder if you could do it a bit easier by skipping the input/output and
actually changing the character map table (PROC TRANTAB, I believe) to make
SAS think | is " and " is =BD or something. Then import, and then revert t=
o
the normal translation table...

-Joe

On Mon, Nov 30, 2009 at 12:07 PM, Dan T Keating <keatingd(a)washpost.com>wrot=
e:

> A dataset I need to regularly import is comma-delimited and then uses
> pipes as the text qualifier rather than quotes (because the data itself i=
s
> laden with both quotes and commas). I lobbied for just pipe-delimited bu=
t
> the providers noted that both MS Access and mysql have for many years mad=
e
> it very easy to designate the text qualifier as any character when
> importing delimited data. The data imports easily into both of them.
>
> If DSD would recognize pipes rather than quotes, it would work perfectly.
> To replicate the funtionality of DSD, I have to program my own version of
> scan that ignores commas enclosed within pipes. The bigger and better sca=
n
> (9.2) does not include this -- its q operator only envisions quotes as th=
e
> text qualifer.
>
> The SAS solution I think I have to shoot for is regular expression patter=
n
> matching.
>
> But a good SAS solution would be some flexibility in what DSD uses as its
> text qualifier. Am I wrong in concluding that the flexibility is not ther=
e
> now?
>
> Thanks.
>
> Dan
> _________________________________
> Dan Keating
> Graphics Editor/Data, The Washington Post
> (202) 334-5047, keatingd(a)washpost.com
>
 | 
Pages: 1
Prev: competing risks proportional
Next: Latent Class Analysis - Question