Re^4: Tk and Non-ASCII File Names

graff,

Thanks for your input. I tried decode but I'm still error-ridden; I've included the updates below. Any other ideas or encoding tricks where I can see what's going on under the hood?

Command Line Output:

06_Protection_de_la_tÃªte.xml: No such file or directory
No such file or directory
Assuming 'require Tk::ExecuteCommand;' at ./tmp.pl line 24
[download]

Tk Output:

06_Protection_de_la_tÃªte.xml: No such file or directory
[download]

Code:

#!/usr/local/bin/perl
use warnings;
use strict;
use Tk;
use File::Find::Rule;
use Encode qw(decode);

#my $file = '06_Protection_de_la_tête.xml';
my @files = File::Find::Rule->file->name('*.xml')->in('.');
my $file = shift @files;
$file = decode('iso-8859-1', $file);

my $cmd = "ls -l $file";

### Try ls.
print qx($cmd);

### Try reading the first line.
open my $F, '<', $file;
print $! ? "$!\n" : scalar <$F> ;

### Try ls via Tk.
my $mw = MainWindow->new;
my $exec = $mw->ExecuteCommand(
    -command => $cmd,
)->pack;
$exec->execute_command;
$exec->update;

MainLoop;
[download]

Comment on Re^4: Tk and Non-ASCII File Names Select or Download Code

Replies are listed 'Best First'.
Re^5: Tk and Non-ASCII File Names by graff (Chancellor) on Oct 08, 2010 at 06:51 UTC
In case you didn't know, the strings you just posted as the "Command Line Output" and "Tk Output" both contain a valid utf8 character ("e with circumflex"); it looks bad because both your shell terminal window and your Tk window are using iso-8859-1 encoding to display the data; the two "wrong" characters that you see are actually the two bytes of the utf8-encoded character, being interpreted as separate 8859-1 characters instead. Having the string in utf8 encoding probably explains why the `open()` fails as well -- you've actually "changed" the file name (the value of $file), by changing the encoding. So, I'm a little confused about the nature of the original problem... What happens if you take this most recent instance of your code (using File::Find::Rule), and comment out this line? `$file = decode('iso-8859-1', $file );` [download]	[reply] [d/l] [select]
Re^6: Tk and Non-ASCII File Names by eff_i_g (Curate) on Oct 08, 2010 at 13:56 UTC
graff, Aye, the UTF-8 is valid, but my locale is en_US.ISO8859-1 (or at least some of it is; see my original post). If I comment out the encoding line the `ls` works as well as reading a line from the file, but Tk still reports: `06_Protection_de_la_tÃªte.xml: No such file or directory` [download] Back to the original problem: without any fancy stuff (decoding, encoding, etc.) commands with non-ASCII characters work fine except when they are ran through Tk::ExecuteCommand, which is what I'm trying to fix because it's a Tk application;the others are just examples. I posted a solution for this in an earlier reply, but it involves changing the module's code and I'm not sure how wise that is—if I'm fixing a valid bug or creating a potential problem for the future. Thanks for sticking with me on this one!	[reply] [d/l] [select]