From: | "Andrew Dunstan" <andrew(at)dunslane(dot)net> |
---|---|
To: | <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | <pgman(at)candle(dot)pha(dot)pa(dot)us>, <chriskl(at)familyhealth(dot)com(dot)au>, <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Different length lines in COPY CSV |
Date: | 2005-12-12 09:32:22 |
Message-ID: | 3044.24.211.165.134.1134379942.squirrel@www.dunslane.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Tom Lane said:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
>> Christopher Kings-Lynne wrote:
>>> Is there any way to force COPY to accept that there will be lines of
>>> different length in a data file?
>
>> I suppose we could have a TRAILINGNULL flag to COPY but because few
>> ask for this feature, it hardly seems worth it.
>
> There is no chance that we'll ever be able to cope with every insane
> file format that some benighted program claims is CSV. The harder we
> try, the more we will lose the ability to detect data errors at all;
> not to mention the likely negative consequences for the readability and
> performance of the COPY code. I think "fix it with a perl script" is a
> very reasonable answer for cases like this one.
>
I agree.
The COPY code is probably on the edge of maintainability now.
Our CSV routines accept a wide variety of imports formats, but a fixed
number of columns is required. Maybe we need a pgfoundry project with some
general perl CSV munging utilities - this issue comes up often enough.
cheers
andrew
From | Date | Subject | |
---|---|---|---|
Next Message | Andreas Pflug | 2005-12-12 11:32:51 | pg_relation_size locking |
Previous Message | Michael Glaesemann | 2005-12-12 08:42:24 | Re: psql patch: new host/port |