I have CSV files blowing up because they have double quotes around numeric
values: " 1.34" yields ERROR: invalid input syntax for type numeric: "
1.34"

And occasionally these are empty, and you see invalid syntax for type
numeric: " "

Are there flags I can pass to COPY to avoid this? There are properly quoted
string values elsewhere in the final, so I'm reticent to strip out double
quote characters or anything.

Thank you.

--
Wells Oliver
wellsoliver (AT) gmail (DOT) com

Wells Oliver asked:



There's no easy command line method that I know of. Perhaps if you can detect patterns some sed / awk might help.

I'd probably load the CSV file into a temporary work table with the columnsthat are giving you issues defined as type "text." Then use SQL to remove the offending quotation marks from the afflicted columns. Create new columns of type numeric in the table, and copy the values from the text variant to the numeric ones with the proper cast one at a time; if it blows up you go back to resolving issues. Then copy the table using the numeric columns to the desired destination.

Hopefully others might have better ideas ... best o' luck!


Greg Williamson


--
Sent via pgsql-admin mailing list (pgsql-admin (AT) postgresql (DOT) org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-admin

Wells Oliver <wellsoliver (AT) gmail (DOT) com> writes:
Is that an exact copy of the error message? Because there are double
quotes in that error message's text; what you are showing does not
suggest that they are part of the data value. I'm wondering about
non-visible characters myself.

Well, that you're going to have to do something else about. Possibly
you could treat the empty string as null? (See the 'null string'
option to COPY.)

regards, tom lane

--
Sent via pgsql-admin mailing list (pgsql-admin (AT) postgresql (DOT) org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-admin