problem with CHAR compressed SAS datasets about readstat

[reikoch]

Excellent - it works for my test files!
Thanks lots!

[evanmiller]

Notes for myself:

It appears that a subheader pointer in bar.sas7bdat is pointing to compressed data, but the pointer itself indicates that the data is not compressed. (The byte at offset 5809096 is 0x00, whereas all of the other pointers in the block have the value 0x04.) In a way the issue is the converse of #37. I'm not sure if the compression bit is stored elsewhere or if a workaround is required here.

[reikoch]

If it helps I can produce a really char compressed dataset, not just one labelled so. I see problems with char compressed datasets so often that I do not believe it to be restricted to "wrong" labelling.

[evanmiller]

The data you provided is indeed compressed. The "mislabeling" occurs for a short segment of data deep in the file. The trouble occurs because a compressed file can contain both compressed and uncompressed chunks, and these individual chunks are sometimes mislabeled (apparently).

[reikoch]

Most char compressed SAS datasets can now be read by development version 0.2.0.9000 of haven. But the simple example ietest2.sas7bdat in [https://github.com/reikoch/testfiles/blob/master/ietest2.sas7bdat] returns for read_sas("ietest2.sas7bdat"):

ReadStat: Error parsing page 0, bytes 8192-16383

Error: Failed to parse /opt/BIOSTAT/home/kochr4/sas7bdat/ietest2.sas7bdat: Invalid file, or file has unsupported features.

[reikoch]

It looks like Python module sas7bdat from [https://bitbucket.org/jaredhobbs/sas7bdat] has solved decoding of SAS compressed datasets...

[evanmiller]

Fixed in c92f697

Please re-open the issue if you are still experiencing problems.