sed strict pattern

[antistes]

Hi forum,

i have a small sed script witch strips all special chars from stdin
sed -e "s/[^Aa0-Zz9]/_/g"

My enviroment var LANG is pt_PT.UTF-8 and when i try to strip something like this

echo "Página de teste" | sed -e "s/[^Aa0-Zz9]/_/g"

it outputs 'Página_de_teste' instead of 'P_gina_de_teste' or even better 'Pagina_de_teste'

some help please...

[vsemaska]

Did you '# export LANG' after setting it?

[antistes]

Thats my native lang, i've installed the box (CentOS 4.5) with this lang, and i want to keep it.

The thing is sed thinks that 'á' is included in my search pattern and that's not good.
I just want sed to distinguish one from another.

[Franklin52]

Try:

echo "Página de teste" | sed 's/[^A-Z ^a-z ^0-9]/_/g'

Regards

[antistes]

Try:

echo "Página de teste" | sed 's/[^A-Z ^a-z ^0-9]/_/g'

Regards

Thanks, but still not working, the output is
'Página de teste'

[Franklin52]

You can try something like:

echo "Página de teste" | sed 's/[^A-Z^a-z^0-9]/_/g' | sed 's/á/a/g'

Regards

[antistes]

Cool thats it
it worked fine

just one tiny little problem

i've got to predict all portuguese special chars, and they're quite a bit,
but i can live with that

Thanks