Postgres Pro
Downloads
Home   >   mailing lists

Thread: Utility that creates table schema from csv data?

Utility that creates table schema from csv data?

From
CSN
Date:
Probably wishful thinking, but who knows - maybe
there's something in contrib! I have a bunch of csv
data with the field names specified on the first line
of the various files. Is there any such utility that
will create a table schema using the field names AND
look through the data and determine what data types
each field should be?

Thanks,
CSN



__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com

Re: Utility that creates table schema from csv data?

From
Richard Huxton
Date:
CSN wrote:
> Probably wishful thinking, but who knows - maybe
> there's something in contrib! I have a bunch of csv
> data with the field names specified on the first line
> of the various files. Is there any such utility that
> will create a table schema using the field names AND
> look through the data and determine what data types
> each field should be?

I'm not sure that's possible in a general sense. Is 123 text or a
number? How do I prove it one way or another?

However, if you were happy to end up with "good enough" types (i.e.
might end up using text instead of a more specific type), and know that
data was representative then it shouldn't be too tricky to write some
pattern-matching Perl to do the job.

--
   Richard Huxton
   Archonet Ltd

Re: Utility that creates table schema from csv data?

From
"Reid Thompson"
Date:
Richard Huxton wrote:
> CSN wrote:
>> Probably wishful thinking, but who knows - maybe
>> there's something in contrib! I have a bunch of csv
>> data with the field names specified on the first line
>> of the various files. Is there any such utility that
>> will create a table schema using the field names AND
>> look through the data and determine what data types
>> each field should be?
>
> I'm not sure that's possible in a general sense. Is 123 text or a
> number? How do I prove it one way or another?
>
> However, if you were happy to end up with "good enough" types (i.e.
> might end up using text instead of a more specific type), and
> know that
> data was representative then it shouldn't be too tricky to write some
> pattern-matching Perl to do the job.

ruby's auto-typing might do the same, w/o having to write
pattern-matching.

reid