Question

UTF-8 characters are destroyed when processed with the JSON library (maybe this is similar to Problem with decoding unicode JSON in perl, however setting binmode only creates another problem).

I have reduced the problem down to the following example:

(hlovdal) localhost:/tmp/my_test>cat my_test.pl
#!/usr/bin/perl -w
use strict;
use warnings;
use JSON;
use File::Slurp;
use Getopt::Long;
use Encode;

my $set_binmode = 0;
GetOptions("set-binmode" => \$set_binmode);

if ($set_binmode) {
        binmode(STDIN,  ":encoding(UTF-8)");
        binmode(STDOUT, ":encoding(UTF-8)");
        binmode(STDERR, ":encoding(UTF-8)");
}

sub check {
        my $text = shift;
        return "is_utf8(): " . (Encode::is_utf8($text) ? "1" : "0") . ", is_utf8(1): " . (Encode::is_utf8($text, 1) ? "1" : "0"). ". ";
}

my $my_test = "hei på deg";
my $json_text = read_file('my_test.json');
my $hash_ref = JSON->new->utf8->decode($json_text);

print check($my_test), "\$my_test = $my_test\n";
print check($json_text), "\$json_text = $json_text";
print check($$hash_ref{'my_test'}), "\$\$hash_ref{'my_test'} = " . $$hash_ref{'my_test'} .  "\n";

(hlovdal) localhost:/tmp/my_test>

When running testing the text is for some reason crippeled into iso-8859-1. Setting binmode sort of solves it but then causes double encoding of other strings.

(hlovdal) localhost:/tmp/my_test>cat my_test.json 
{ "my_test" : "hei på deg" }
(hlovdal) localhost:/tmp/my_test>file my_test.json 
my_test.json: UTF-8 Unicode text
(hlovdal) localhost:/tmp/my_test>hexdump -c my_test.json 
0000000   {       "   m   y   _   t   e   s   t   "       :       "   h
0000010   e   i       p 303 245       d   e   g   "       }  \n        
000001e
(hlovdal) localhost:/tmp/my_test>
(hlovdal) localhost:/tmp/my_test>perl my_test.pl
is_utf8(): 0, is_utf8(1): 0. $my_test = hei på deg
is_utf8(): 0, is_utf8(1): 0. $json_text = { "my_test" : "hei på deg" }
is_utf8(): 1, is_utf8(1): 1. $$hash_ref{'my_test'} = hei p� deg
(hlovdal) localhost:/tmp/my_test>perl my_test.pl --set-binmode
is_utf8(): 0, is_utf8(1): 0. $my_test = hei på deg
is_utf8(): 0, is_utf8(1): 0. $json_text = { "my_test" : "hei på deg" }
is_utf8(): 1, is_utf8(1): 1. $$hash_ref{'my_test'} = hei på deg
(hlovdal) localhost:/tmp/my_test>

What is causing this and how to solve?

---

This is on a newly installed and up to date Fedora 15 system.

(hlovdal) localhost:/tmp/my_test>perl --version | grep version
This is perl 5, version 12, subversion 4 (v5.12.4) built for x86_64-linux-thread-multi
(hlovdal) localhost:/tmp/my_test>rpm -q perl-JSON
perl-JSON-2.51-1.fc15.noarch
(hlovdal) localhost:/tmp/my_test>locale
LANG=en_US.UTF-8
LC_CTYPE="en_US.UTF-8"
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_COLLATE="en_US.UTF-8"
LC_MONETARY="en_US.UTF-8"
LC_MESSAGES="en_US.UTF-8"
LC_PAPER="en_US.UTF-8"
LC_NAME="en_US.UTF-8"
LC_ADDRESS="en_US.UTF-8"
LC_TELEPHONE="en_US.UTF-8"
LC_MEASUREMENT="en_US.UTF-8"
LC_IDENTIFICATION="en_US.UTF-8"
LC_ALL=
(hlovdal) localhost:/tmp/my_test>

---

Update: Adding use utf8 does not solve it, characters are still not processed right (although slightly different from before):

(hlovdal) localhost:/tmp/my_test>perl  my_test.pl
is_utf8(): 1, is_utf8(1): 1. $my_test = hei p� deg
is_utf8(): 0, is_utf8(1): 0. $json_text = { "my_test" : "hei på deg" }
is_utf8(): 1, is_utf8(1): 1. $$hash_ref{'my_test'} = hei p� deg
(hlovdal) localhost:/tmp/my_test>perl  my_test.pl --set-binmode
is_utf8(): 1, is_utf8(1): 1. $my_test = hei på deg
is_utf8(): 0, is_utf8(1): 0. $json_text = { "my_test" : "hei på deg" }
is_utf8(): 1, is_utf8(1): 1. $$hash_ref{'my_test'} = hei på deg
(hlovdal) localhost:/tmp/my_test>

As noted by perlunifaq

Can I use Unicode in my Perl sources?

Yes, you can! If your sources are
UTF-8 encoded, you can indicate that
with the use utf8 pragma.

use utf8;

This doesn’t do anything to your
input, or to your output. It only
influences the way your sources are
read. You can use Unicode in string
literals, in identifiers (but they
still have to be “word characters”
according to \w ), and even in custom
delimiters.

Answer 1

The core of the problem was JSON’s expectation of an octet array instead of character string (solved in this question). However I was also missing several things related to unicode, like “use utf8”. Here is the diff required to make the code in the example work fully:

--- my_test.pl.orig 2011-08-03 15:44:44.217868886 +0200
+++ my_test.pl  2011-08-03 15:55:30.152379269 +0200
@@ -1,19 +1,14 @@
-#!/usr/bin/perl -w
+#!/usr/bin/perl -CSAD
 use strict;
 use warnings;
 use JSON;
 use File::Slurp;
 use Getopt::Long;
 use Encode;
-
-my $set_binmode = 0;
-GetOptions("set-binmode" => \$set_binmode);
-
-if ($set_binmode) {
-        binmode(STDIN,  ":encoding(UTF-8)");
-        binmode(STDOUT, ":encoding(UTF-8)");
-        binmode(STDERR, ":encoding(UTF-8)");
-}
+use utf8;
+use warnings qw< FATAL utf8 >;
+use open qw( :encoding(UTF-8) :std );
+use feature qw< unicode_strings >;

 sub check {
         my $text = shift;
@@ -21,8 +16,9 @@
 }

 my $my_test = "hei på deg";
-my $json_text = read_file('my_test.json');
-my $hash_ref = JSON->new->utf8->decode($json_text);
+my $json_text = read_file('my_test.json', binmode => ':encoding(UTF-8)');
+my $json_bytes = encode('UTF-8', $json_text);
+my $hash_ref = JSON->new->utf8->decode($json_bytes);

 print check($my_test), "\$my_test = $my_test\n";
 print check($json_text), "\$json_text = $json_text";