Em 28/07/2012 19:41, Tom Lane escreveu:
> Edson Richter <edsonrichter@hotmail.com> writes:
>> If I do apply trim over the substring, no spaces are removed, and I
>> cannot understand why.
>> ...
>> But using the regular expression matching "^\s*" and "\s*$" works, and
>> spaces are removed:
> I think what this means is that what you say are runs of spaces are no
> such thing, but are some other whitespace character(s). Perhaps tabs,
> or non-breaking spaces?
>
> regards, tom lane
>
By all means, you are right.
I supposed they are spaces... at least in the original CVS file.
I've used an Hex editor, and they are character C2A0 (UTF-8) in the file.
Now, executing the following query surprises me:
select rslinha2,
ascii(substring(rslinha2 from 3 for position('(+)' in rslinha2)-4))
from plan_maio
limit 1
resulted in the following:
"2. TAXA VIGILANCIA (+) R$ 13,00";160
So, file_fdw converted my original spaces into character 160.
That's the reason for trim not working.
Thanks for your toughs. You led me to the right direction: when
importing data from files, not always what looks like is what it is.
Regards,
Edson Richter.
