topic Re: Data is not a SAS data set in SAS Programming

Data is not a SAS data set

alice_cimb — Thu, 10 Sep 2020 07:22:57 GMT

Hi,

I gotten the sasdata from someone else, but not able to execute or view the sasdata via SAS EG or SAS Base. The error message that I got is as per below:

The pc. library is located in my D drive.

58
59 data Car_v2_avph;
60 set pc.Car_v2_avph;
ERROR: File PC.CAR_V2_AVPH.DATA is not a SAS data set.
61 run;

NOTE: The SAS System stopped processing this step because of errors.
WARNING: The data set WORK.CAR_V2_AVPH may be incomplete. When this step was stopped there were
0 observations and 0 variables.
NOTE: DATA statement used (Total process time):
real time 0.03 seconds
cpu time 0.01 seconds

I am able to view the data from the sas Explorer.

Below is the sasdata properties if i open from my local D drive.

Please assist if I can save this data

Re: Data is not a SAS data set

Shmuel — Thu, 10 Sep 2020 08:01:46 GMT

Was the dataset created on same OS type as the SAS EG ? (Windows / unix or linox )

Re: Data is not a SAS data set

alice_cimb — Thu, 10 Sep 2020 08:16:54 GMT

Yes, is created under same OS, which is Window. But it created in Y2014, is an old sasdata, we are retrieving it for our analysis .

Re: Data is not a SAS data set

RichardDeVen — Thu, 10 Sep 2020 10:58:23 GMT

How was the data set file (the .sas7bdat) 'gotten' or otherwise delivered to you ?

- FTP may have converted characters

- A text editor or other file editor may have been used to look at the file and accidentally overwritten it

- File could have be retrieved from an archiving application using incorrect settings or environment variables

Is it possible they used PROC CPORT to create a file with an incorrect filename extension ?

Have you looked at the file with a Hex viewer? Example sashelp.cars file in Hex Editor Neo

Re: Data is not a SAS data set

jimbarbour — Thu, 10 Sep 2020 14:31:53 GMT

I @RichardDeVen is probably on the right track. Is the original file available anywhere? If you can get back to the original file, you may be OK. If it was FTP'd to you and they used ASCII mode, then the file could be corrupted. There are other ways that the file can be corrupted such as, as @RichardDeVen suggests, opening in in an editor window and accidentally saving it. Again, if the original file is still available or can somehow be pulled back from archive, you may be OK. I think the version that you have may be unreadable even though SAS can "see" it.

Jim

Re: Data is not a SAS data set

ballardw — Thu, 10 Sep 2020 14:39:17 GMT

I would go back to the "someone else" and discuss what may have been done with this file.

What version of SAS are you currently running? And which version created the data set (probably have to get that from "someone else")? If by any chance you are using SAS 9.2, especially a 32-bit version, or earlier and the set was made with a more recent version this can happen because of changes to the data set header information.

Re: Data is not a SAS data set

Tom — Thu, 10 Sep 2020 15:35:29 GMT

In general SAS7BDAT file format has been very portable.

The one issue I remember that caused trouble was the introduction some type of larger integer values used for some type of internal record number with 64bit versions of SAS that caused the files to be unreadable on 32bit version of SAS. The fix was to re-create the file with the 64bit verison of SAS with the proper dataset option to use the older shorter values instead. Then the file could be read with the 32bit SAS.

(I have forgotten the field and/or dataset option name)

Re: Data is not a SAS data set

ballardw — Thu, 10 Sep 2020 16:19:48 GMT

@Tom wrote:

In general SAS7BDAT file format has been very portable.

The one issue I remember that caused trouble was the introduction some type of larger integer values used for some type of internal record number with 64bit versions of SAS that caused the files to be unreadable on 32bit version of SAS. The fix was to re-create the file with the 64bit verison of SAS with the proper dataset option to use the older shorter values instead. Then the file could be read with the 32bit SAS.

(I have forgotten the field and/or dataset option name)

I think this is EXTENDOBSCOUNTER which could be a system option, data set option or libname option. That is what I was thinking of when I asked about versions earlier.The default for all three is YES and would need to be set to NO for a 32-bit to be able to read the data set. IF the file has enough records (more than 2,147,483,647) then the header information may not be correct for reporting the number of observations as that is the limit for the 32-bit info.

I had run into that when SAS 9.3 came out but I was using 9.2.

Re: Data is not a SAS data set

RichardDeVen — Thu, 10 Sep 2020 19:49:46 GMT

Try using a legacy library engine and see if the data content is accessible.

LIBNAME WAYBACK V8 'D:\Data';

Re: Data is not a SAS data set

alice_cimb — Fri, 11 Sep 2020 02:23:13 GMT

Hi @RichardDeVen

Have checked with the person who passed me the data. The data was retrieved from archived folder, and should be corrupted when they extracted it. This is a monthly generate data, only those data that extracted from archived folder are impacted, while some of the original dataset which is not in keep in archived still can be used.

I have ask them to try to re-extract from the archived folder using others tools. But if this fail again, what can I do to save the file? Can I open the data using another tools and import to text or csv?

Below is the data view in Hex view, please advice what can i do about it?

Thanks

Re: Data is not a SAS data set

alice_cimb — Fri, 11 Sep 2020 02:26:15 GMT

@RichardDeVen

Below is the screen shot for the data in Hex.

Re: Data is not a SAS data set

mkeintz — Fri, 11 Sep 2020 04:07:20 GMT

Actually the popup windows shown by @alice_cimb doesn't show that it is/was, a sas dataset, corrupted or not. I can reproduce the same popups just by renaming a .TXT file to have a .sas7bdat extension and right-clicking. It's just the o/s and sas making an inference based on the extension. The failure only occurs when asking sas to look inside.

Re: Data is not a SAS data set

jimbarbour — Fri, 11 Sep 2020 04:47:05 GMT

Well, I can't say for 100% sure, but that hex display of the SAS dataset doesn't look good. I would expect to see a few things that were intelligible, but I don't see any in your hex file. For example, here's a random SAS data set from my system. Much of it looks like junk, but I can make out "SAS FILESUMMARY," "DATA," and "9.0401M6X64_SRV16."

            Âê`³Ï½’ 	Ç1Œ"" 332         "" 3323#3 >          SAS FILESUMMARY                                                         DATA        ºIüˆÌwÜAºIüˆÌwÜA     ”ÑÀ     ”ÑÀ                 9.0401M6X64_SRV16                                                       ºIüˆö“íö“íö“í              €']    ºIüˆÌwÜA

I hope I'm wrong, but the hex display of your file doesn't look promising.

Jim

Re: Data is not a SAS data set

RichardDeVen — Fri, 11 Sep 2020 05:06:57 GMT

1F 9D is a 'magic number' associated with tar zip (.z or tar.z files). Try to untar the file and then unzip that (or unzip then untar). With any luck you will get your actual data set out the end.

Re: Data is not a SAS data set

alice_cimb — Fri, 11 Sep 2020 09:24:15 GMT

Okay, will try that, thanks~!

Re: Data is not a SAS data set

Kurt_Bremser — Fri, 11 Sep 2020 10:06:35 GMT

1F9D is the magic number for the .Z file format, which is created by the UNIX compress utility.

Run uncompress and see if you can use the result; if that is not the case, run tar -tf uncompressed_file to check if it is actually a tar archive.

Re: Data is not a SAS data set

alice_cimb — Mon, 14 Sep 2020 06:56:33 GMT

Thanks @Kurt and all, based on your advice, the problem is solved. We modified the file name by adding .Z at behind (car_v2_avph.sas7bdat.Z ) and uncompress the file.

Thanks for the help~!