issue-2 - exception while sorting wide characters is fixed

MANIFEST

@@ -6,6 +6,7 @@ t/Sort-Naturally-XS.t
 t/Sort-Naturally-XS-ncmp.t
 t/Sort-Naturally-XS-nsort.t
 t/Sort-Naturally-XS-sorted.t
+t/Sort-Naturally-XS-wchar.t
 Changes
 const-c.inc
 const-xs.inc

README

@@ -22,7 +22,7 @@ Sergey Yurzin, [jurzin.s@gmail.com](mailto:jurzin.s@gmail.com)
 
 COPYRIGHT AND LICENSE
 
-Copyright (C) 2017 by Sergey Yurzin
+Copyright (C) 2018 by Sergey Yurzin
 
 This library is free software; you can redistribute it and/or modify
 it under the same terms as Perl itself, either Perl version 5.18.2 or,

README.md

@@ -160,6 +160,19 @@ argument:
   # $result_ca contains A, a, B, b, C, c
 ```
 
+Also, make sure your list does not contain "wide characters", otherwise "Wide character in subroutine entry" exception
+will be thrown. Be vigilant if `use utf8` is in effect or your source code contains multibyte characters. It's a
+developer's responsibility to explicitly encode characters in a target encoding:
+
+```perl
+  use utf8;
+  use Encode;
+  use Sort::Naturally::XS qw/sorted/;
+
+  my $fruits = [qw/яблоко банан манго киви груша/];
+  my $result = sorted([map {Encode::encode('utf8', $_)} @{$fruits}], locale => 'ru_RU.utf8');
+``` 
+
 Note: due to the complexity of a cross-platform support, a locale aware sorting is guaranteed on Unix-like operating
 systems only.
 
@@ -228,7 +241,7 @@ Sergey Yurzin, [jurzin.s@gmail.com](mailto:jurzin.s@gmail.com)
 
 ## COPYRIGHT AND LICENSE
 
-Copyright (C) 2017 by Sergey Yurzin
+Copyright (C) 2018 by Sergey Yurzin
 
 This library is free software; you can redistribute it and/or modify
 it under the same terms as Perl itself, either Perl version 5.18.2 or,

XS.xs

@@ -13,16 +13,16 @@
 static I32
 S_sv_ncmp(pTHX_ SV *a, SV *b)
 {
-    const char *ia = (const char *) SvPVbyte_nolen(a);
-    const char *ib = (const char *) SvPVbyte_nolen(b);
+    const char *ia = (const char *) SvPVutf8_nolen(a);
+    const char *ib = (const char *) SvPVutf8_nolen(b);
     return _ncmp(ia, ib, 0, 0);
 }
 
 static I32
 S_sv_ncmp_reverse(pTHX_ SV *a, SV *b)
 {
-    const char *ia = (const char *) SvPVbyte_nolen(a);
-    const char *ib = (const char *) SvPVbyte_nolen(b);
+    const char *ia = (const char *) SvPVutf8_nolen(a);
+    const char *ib = (const char *) SvPVutf8_nolen(b);
     return _ncmp(ia, ib, 1, 0);
 }
 

lib/Sort/Naturally/XS.pm

@@ -163,6 +163,17 @@ keyword argument:
   my $result_ca = sorted($list, locale => 'en_CA.utf8');
   # $result_ca contains A, a, B, b, C, c
 
+Also, make sure your list does not contain "wide characters", otherwise "Wide character in subroutine entry" exception
+will be thrown. Be vigilant if C<use utf8> is in effect or your source code contains multibyte characters. It's a
+developer's responsibility to explicitly encode characters in a target encoding:
+
+  use utf8;
+  use Encode;
+  use Sort::Naturally::XS qw/sorted/;
+
+  my $fruits = [qw/яблоко банан манго киви груша/];
+  my $result = sorted([map {Encode::encode('utf8', $_)} @{$fruits}], locale => 'ru_RU.utf8');
+
 Note: due to the complexity of a cross-platform support, a locale aware sorting is guaranteed on Unix-like operating
 systems only.
 
@@ -237,7 +248,7 @@ Sergey Yurzin, L<jurzin.s@gmail.com|mailto:jurzin.s@gmail.com>
 
 =head1 COPYRIGHT AND LICENSE
 
-Copyright (C) 2017 by Sergey Yurzin
+Copyright (C) 2018 by Sergey Yurzin
 
 This library is free software; you can redistribute it and/or modify
 it under the same terms as Perl itself, either Perl version 5.18.2 or,

t/Sort-Naturally-XS-sorted.t

@@ -109,7 +109,6 @@ SKIP: {
         my $ar_ca_local__actual = sorted($ar_en_local, locale => $locale);
         ok(eq_array($ar_ca_local__expected, $ar_ca_local__actual), 'Locale CA test');
     }
-
 }
 
 done_testing();

t/Sort-Naturally-XS-wchar.t

@@ -0,0 +1,37 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+use Test::More;
+use Sort::Naturally::XS qw/nsort ncmp sorted/;
+use List::Util qw/first/;
+use utf8;
+use Config;
+
+my $ar_mixed_utf8 = [qw/Як-100 Ка-8 Ми-20 Ка-10 Ка-26 Ка-15 Ка-25 Ми-4 Ми-6 Ми-8 Ка-31 Ми-14 Ми-24 Ка-18 Ка-22 Ми-26
+    Ми-30 Ми-171 Як-24 Як-60 Ка-27 Ка-29 Ка-32 Ка-126 Ми-10 Ми-1/];
+my $ar_mixed_utf8__expected = [qw/Ка-8 Ка-10 Ка-15 Ка-18 Ка-22 Ка-25 Ка-26 Ка-27 Ка-29 Ка-31 Ка-32 Ка-126 Ми-1 Ми-4 Ми-6
+    Ми-8 Ми-10 Ми-14 Ми-20 Ми-24 Ми-26 Ми-30 Ми-171 Як-24 Як-60 Як-100/];
+ok(eq_array($ar_mixed_utf8__expected, [nsort(@{$ar_mixed_utf8})]), "Wide characters in input of 'nsort'");
+
+ok(eq_array($ar_mixed_utf8__expected, [sort {ncmp($a, $b)} @{$ar_mixed_utf8}]), "Wide characters in input of 'ncmp'");
+
+ok(eq_array($ar_mixed_utf8__expected, sorted($ar_mixed_utf8)), "Wide characters in input of 'sorted'");
+
+# issue-2 example
+{
+    no utf8;
+
+    my @issue_2_list = ( qq(\x{2603}), q(abc) );
+    my @issue_2_list__expected = ( q(abc), qq(\x{2603}) );
+    my @issue_2_list__actual = nsort(@issue_2_list);
+
+    ok(eq_array(\@issue_2_list__expected, \@issue_2_list__actual), "Wide character (not letter) in input of 'nsort'");
+
+    @issue_2_list__actual = sort {ncmp($a, $b)} @issue_2_list;
+    ok(eq_array(\@issue_2_list__expected, \@issue_2_list__actual), "Wide character (not letter) in input of 'ncmp'");
+
+    ok(eq_array(\@issue_2_list__expected, sorted(\@issue_2_list)), "Wide character (not letter) in input of 'sorted'");
+}
+
+done_testing();