Unicode characters are not supported

[Nevon]

/ # ./root/bin/pokemonsay фывафва
 ________________ 
< ÑÑваÑва >
 ---------------- 
           \
            \
             \
              \
        ▄▄▄▄
       ▄▄  ▄▄▄ 
   ▄▄▄▄▄▄▄ ▄   
 ▄▄▄▄ ▄▄ ▄ ▄▄▄  ▄▄▄▄▄
 ▀ ▄  ▄▄▄ ▄▄ ▄▄▄▄▄   
  ▄▄▄▄▄▄▄ ▄▄▄ ▄▄▄   ▄▄
 ▄▄▄ ▄ ▄▄ ▄  ▄▄▄ ▄▄ ▄▄▄ 
  ▄▄▄▄▄ ▄ ▄ ▄▄ ▄▄ ▄ ▄ ▄▀
  ▀▀▀    ▄  ▀▄▄▄  ▄▄▄▀
       ▀▄▄▄▀    ▀▀▀

Rhyhorn

[xaviervia]

We tested this in docker with Alpine linux. echo and cowsay print the unicode characters just fine, so the issue seems to be coming from pokemonsay.

Great project btw, love it ❤️

[Nevon]

Cross-posted this issue upstream, since it looks like the issue is in rossy/img2xterm#9

[possatti]

Thanks for the support guys.

Funny thing, I am Brazilian but I have only tested with English text so far. But it's just like @Nevon said, it looks like an issue with img2xterm.

[possatti]

I've been thinking about this and I think there's a way around the problem. It's not the most elegant solution, but I could remove binmode STDOUT, ":utf8"; from all cowfiles and pipe the output of cowsay to a sed process which will remove that warning line.

I went on to experiment with that. I removed the binmode thing and it worked. I've put the thing back again expecting it to not work... and it also worked. Suddenly I don't understand anything anymore, but mine is working with a clean repo. It's like I have done nothing, but it is working.

I wish I've noticed what happened and I would come up with a solution, but I don't.

[xaviervia]

sed sounds like a beautiful solution to me. No need for more if it works.

[possatti]

Guys, could you try the branch unicode. I think there is no need to install, just clone the repo and do ./pokemonsay.sh "こんにちは世界" a few times. I might have fixed the problem.

If you can, please test the master branch as well. Nothing changed on master, but I'm puzzled that I can't reproduce the problem anymore.

I'm sorry it took so long for me to take a look at this. It's my last year in college and things are really tough. xD

[Nevon]

At first I figured maybe it was an issue with your expression, so I made one that might be a little more robust, but it's also failing. However, it is matching the text normally, so there's something fishy.

However, it works with a plain non-ansi nonsense cowfile:

Then I thought that maybe the warning was in stderr or something, so I tried combining the streams before doing the sed, but that didn't work either. Super weird.

Having never written Perl in my life, I have no idea if the following is correct, but I think the binmode(STDOUT, ":utf8"); should stay. The files are encoded as us-ascii, which is a subset of utf8, so we are expecting the output to be utf8. So I went the other direction and modified cowsay to basically tell it that anything and everything is utf8 and should be treated as such:

use utf8;
binmode STDOUT, ":utf8";
binmode STDERR, ":utf8";
binmode STDIN,  ":utf8";

But nope...

I'm sorry it took so long for me to take a look at this. It's my last year in college and things are really tough. xD

Don't kill yourself over it, man. It's a silly way to have pokemon in your terminal, not a cure for cancer.

[lookoutldz]

6 years gone. Did you solve the problem?

[JasonLamv-t]

6 years gone. Did you solve the problem?

8 years now. :)