Problem with COPY CSV

[Christopher Kings-Lynne <chriskl(at)familyhealth(dot)com(dot)au>]

Attached is a small test extract from the USDA nutrient database.

The problem is that the script won't load the COPY data correctly.  This 
is with CVS HEAD (and 8.1).

It is the 4th column in the table that gives the problem (nutr_no 
integer).  Each of the 3 COPY rows has a different way of specifying a 
number:

* As an unquoted empty value
* As a quoted number
* As a quoted empty value

Now, I'm specifying "NULL AS ''" in the COPY command but it refuses to 
recognize the quoted empty value as NULL.

Anyone have any ideas?  Is this a bug?  The manual even says that using 
"NULL AS ''" lets you make no distinction between empty and quoted empty 
values.

Chris

Attachment: test.sql.gz
Description: GNU Zip compressed data