Re: About COPY command (and probably file fdw too) - Mailing list pgsql-general
From | Melvin Davidson |
---|---|
Subject | Re: About COPY command (and probably file fdw too) |
Date | |
Msg-id | CANu8Fiz_MBX3pUqhXY29e+zU_1bSOKNhwroKE1ZGG0DXncJzbQ@mail.gmail.com Whole thread Raw |
In response to | Re: About COPY command (and probably file fdw too) (Nicolas Paris <niparisco@gmail.com>) |
Responses |
Re: About COPY command (and probably file fdw too)
|
List | pgsql-general |
I understand what you want with regards to skipping columns in input, but rather than wait to see if that feature is added to a future version of PostgreSQL, probably the best work around is to
1. CREATE an intermediate table with all columns in the input text file.
2. COPY into the intermediate table.
On Thu, May 21, 2015 at 4:33 PM, Nicolas Paris <niparisco@gmail.com> wrote:
Hi,To me this would be great. Why not the ability to restrict lines tooCOPY stafflist (userid, username, staffid)
FROM 'myfile.txt'
WITH (FORMAT text, DELIMITER E'\t', COLUMNS (1, 2, 7), LINES(2:1000,2000:3000), ENCODING 'windows-1250')=> subset of full data.2015-05-21 22:25 GMT+02:00 Stefan Stefanov <stefanov.sm@abv.bg>:Hi,
Maybe I need to clarify a little.
The suggested option “[SKIP] COLUMNS <columnslist>” would contain columns' positions in the file so that only some of the columns in a text file would be read into a table.
Example: copy the first, second and seventh columns form myfile.txt into table "stafflist". myfile.txt has many columns.
COPY stafflist (userid, username, staffid)
FROM 'myfile.txt'
WITH (FORMAT text, DELIMITER E'\t', COLUMNS (1, 2, 7), ENCODING 'windows-1250')
BR, Stefan
-------- Оригинално писмо --------
От: Nicolas Paris niparisco@gmail.com
Относно: Re: [GENERAL] About COPY command (and probably file fdw too)
До: Stefan Stefanov <stefanov.sm@abv.bg>
Изпратено на: 20.05.2015 23:212015-05-20 22:16 GMT+02:00 Stefan Stefanov <stefanov.sm@abv.bg>:Hi,I have been using COPY .. FROM a lot these days for reading in tabular data and it does a very good job. Still there is an inconvenience when a (large) text file contains more columns than the target table or the columns’ order differs. I can imagine three ways round and none is really nice -- mount the file as a foreign table with all the text file’s columns then insert into the target table a select from the foreign table;- create an intermediate table with all the text file’s columns, copy into it from the file then insert into the target table and finally drop the intermediate table when no more files are expected;- remove the unneeded columns from the file with a text editor prior to COPY-ing.I think that this is happening often in real life and therefore have a suggestion to add this option “[SKIP] COLUMNS <columnslist>” to the WITH clause of COPY .. FROM. It may be very useful in file fdw too.To be able to re-arrange columns’ order would come as a free bonus for users.Sincerely,Stefan StefanovHi,I guess it already does (from documentation):COPY table_name [ ( column_name [, ...] ) ] FROM { 'filename' | STDIN } [ [ WITH ] ( option [, ...] ) ]Then you can order the column_name as the source file has.
--
Melvin Davidson
I reserve the right to fantasize. Whether or not you
wish to share my fantasy is entirely up to you.
I reserve the right to fantasize. Whether or not you
wish to share my fantasy is entirely up to you.
pgsql-general by date: