Shell won't recognise foreign characters

BlueberryPickles · 07-12-2008, 10:49 AM

Hi everyone,

I'm having a nightmare trying to write a shell script that, among other things, changes to a directory that has foreign characters in its name.

From Ubuntu's terminal, I can type:

$ cd Ontology/Top/World/Français/

and it moves me to my Français directory without a problem.

In my shell script, however, when I have:

#!/bin/sh

LC_ALL="fr_FR.UTF-8"

dir="Ontology/Top/World/Français/"

# Move to the top level of the ontology
cd $dir

# Find all of the categories at the top level of the ontology
ls > directories.tmp

I get the error message:

cd: 8: can't cd to Ontology/Top/World/Fran�ais/

Why won't my script let me use the ç character when I can do it without a problem from the terminal?

Help please!

b0uncer · 07-12-2008, 01:18 PM

I'm not a shell script master when it comes to using locales, but would you need to export the LC_ALL value - did you check that it's actually set?

EDIT: but if LC_ALL is set to that value when you run the script, it should be in use anyway if I'm not mistaken. Just as a guess, does the charset/keymap of your console (which sh uses; not necessarily the same that for example gnome-terminal uses) contain the glyph you're using?

ghostdog74 · 07-12-2008, 02:07 PM

or you can use a wildcard

Mr. C. · 07-13-2008, 03:23 AM

I had no trouble with the script (I added a mkdir to the script to create the directories for me):

Code:

$ cat doit.sh
#!/bin/sh

LC_ALL="fr_FR.UTF-8"

dir="Ontology/Top/World/Français/"
mkdir -p $dir

# Move to the top level of the ontology
cd $dir

# Find all of the categories at the top level of the ontology
ls > directories.tmp

$ ./doit.sh

$ LC_ALL="fr_FR.UTF-8" ls -R
Ontology/ doit.sh*

./Ontology:
Top/

./Ontology/Top:
World/

./Ontology/Top/World:
Fran\çais/

./Ontology/Top/World/Français:
directories.tmp

My sh is bash-based, perhaps yours is Bourne-shell based, or behaves in compatibility mode with called by /bin/sh. Try changing #!/bin/sh to #!/bin/bash

BlueberryPickles · 07-14-2008, 03:52 AM

Tried that, but no success...What did help was to change the foreign characters to character symbols! Not so readable but it worked...