Mime type detection fails with "Can't read SAT" about messytables HOT 6 CLOSED

Probably a bug/limitation in the libmagic library. But if you drop the period from the extension argument (i.e. extension="xls") it should fall back to using that and hopefully will work.

from messytables.

Should be fixed in b63480e

If you don't provide an extension:

ValueError: Unrecognized MIME type: application/CDFV2-corrupt

If you provide .xls (which should really be xls):

ValueError: Could not determine MIME type and unrecognized extension: .xls

If you provide a correct extension xls:

<messytables.excel.XLSTableSet object at 0x10e97d250>

from messytables.

Refiled over at scraperwiki#1, as I want to have a closer look to make it simpler.

from messytables.

If you use any_tableset with a CDF file, no mimetype and no extension with the code in b63480e for me it returns application/CDFV2-corrupt, but if I use magic.fromfile(fileobj.name) I get application/vnd.ms-excel

Presumably there is data later in the file that definitively identifies it as application/vnd.ms-excel (as CDF could be any office file type).

from messytables.

Messytables only looks at the first 1k, I believe [citation needed] that magic checks the first 4k.

There's a fix for this in my pull request; not sure whether this fixes that explicit example.

from messytables.

Nope. It requires at least 64K (of a 95K file)

Tested with:
dd if=Food_price_indices_data_deflated.xls bs=1 count=65536 of=Food_64K.xls

and similar.
(There's a specific C module in the file libary for CDF files.)

from messytables.