topic Re: Transcoding problem in SAS Programming

Transcoding problem

apolitical — Tue, 22 Aug 2017 20:59:23 GMT

I have a csv file with some non-English characters in it. Using data infile in regular SAS, once the foreign character is encountered, the system stops reading in more data and ignores the remainder of the dataset, resulting in a large chunk of the file not imported. I then tried the same code under SAS 9.4 with unicode support, which seems to read in the correct number of rows. The problem is when I try to save it as a dataset on a pre-defined library:

"ERROR: Some character data was lost during transcoding in the dataset lib1.dat1. Either the data
contains characters that are not representable in the new encoding or truncation occurred during transcoding."

and

"WARNING: The data set lib1.data1 may be incomplete. When this step was stopped there were x observations and y variables." x is way less than the number of records in the dataset in the work library.

Is this because dataset is currently in unicode and needs to be converted to regular coding somehow before it can be saved and used? Thanks.

Re: Transcoding problem

kiranv_ — Tue, 22 Aug 2017 21:51:28 GMT

this is an encoding issue. Encoding setting in sas config might defaulted to latin1. we had problem like this before and we tried to fix it

with changing a session option, it did not worked out. What we were suggested at that point of time to change the encoding setting to UTF8 in sas config file.

do a

proc options option=encoding;    
run;

I think you will see latin1 as your encoding.

There might be another solution, but I am just sharing my experience.

Re: Transcoding problem

apolitical — Tue, 22 Aug 2017 22:11:30 GMT

Thanks. Actually I see " ENCODING=UTF-8" because I'm under SAS with unicode support I suppose. I am guessing that the problem of incampatibility arises when I try to save a dataset that's in unicode into a dataset in another coding, although I don't know why SAS would automatically choose to save it in another coding format without asking, maybe that's the default.

Re: Transcoding problem

SASKiwi — Tue, 22 Aug 2017 22:29:00 GMT

I had a similar encoding problem recently and solved it this way using the OUTENCODING option:

libname inlib 'MyInputLibrary'; 
libname outlib 'MyOutputLibrary' outencoding=asciiany;
proc copy noclone in=inlib out=outlib;
   select MyDataset;
run;

Re: Transcoding problem

kiranv_ — Wed, 23 Aug 2017 00:01:05 GMT

please check this below link

http://support.sas.com/kb/41/925.html

Re: Transcoding problem

apolitical — Wed, 23 Aug 2017 15:54:32 GMT

Thanks everyone for the advice. I also found adding an encoding option in the data step can work:
data lib1.data1 (encoding=any);
set data1;
run;