Re: The Inaba patch for tr/// vs. use encoding

Message-Id: <CAC896FE-E7CE-11D6-9228-0003939A104C@dan.co.jp>

p4raw-id: //depot/perl@18125

4 files changed, 241 insertions, 0 deletions

diff --git a/t/uni/tr_7jis.t b/t/uni/tr_7jis.t
new file mode 100644
index 0000000000..292a01a798
--- /dev/null
+++ b/t/uni/tr_7jis.t
@@ -0,0 +1,61 @@
+#
+# $Id$
+#
+# This script is written intentionally in EUC-JP
+# -- dankogai
+
+BEGIN {
+    if ($ENV{'PERL_CORE'}){
+        chdir 't';
+        unshift @INC, '../lib';
+    }
+    require Config; import Config;
+    if ($Config{'extensions'} !~ /\bEncode\b/) {
+      print "1..0 # Skip: Encode was not built\n";
+      exit 0;
+    }
+    if (ord("A") == 193) {
+        print "1..0 # Skip: EBCDIC\n";
+        exit 0;
+    }
+    unless (PerlIO::Layer->find('perlio')){
+        print "1..0 # Skip: PerlIO required\n";
+        exit 0;
+    }
+    $| = 1;
+    print "1..0 # does not work with iso-2022-jp yet\n";
+    exit 0;
+}
+
+use strict;
+#use Test::More qw(no_plan);
+use Test::More tests => 6;
+use Encode;
+use encoding 'iso-2022-jp';
+
+my @hiragana =  map {chr} ord("$B$!(B")..ord("$B$s(B");
+my @katakana =  map {chr} ord("$B%!(B")..ord("$B%s(B");
+my $hiragana = join('' => @hiragana);
+my $katakana = join('' => @katakana);
+my %h2k; @h2k{@hiragana} = @katakana;
+my %k2h; @k2h{@katakana} = @hiragana;
+
+# print @hiragana, "\n";
+
+my $str;
+
+$str = $hiragana; $str =~ tr/$B$!(B-$B$s(B/$B%!(B-$B%s(B/;
+is($str, $katakana, "tr// # hiragana -> katakana");
+$str = $katakana; $str =~ tr/$B%!(B-$B%s(B/$B$!(B-$B$s(B/;
+is($str, $hiragana, "tr// # hiragana -> katakana");
+
+$str = $hiragana; eval qq(\$str =~ tr/$B$!(B-$B$s(B/$B%!(B-$B%s(B/);
+is($str, $katakana, "eval qq(tr//) # hiragana -> katakana");
+$str = $katakana; eval qq(\$str =~ tr/$B%!(B-$B%s(B/$B$!(B-$B$s(B/);
+is($str, $hiragana, "eval qq(tr//) # hiragana -> katakana");
+
+$str = $hiragana; $str =~ s/([$B$!(B-$B$s(B])/$h2k{$1}/go;
+is($str, $katakana, "s/// # hiragana -> katakana");
+$str = $katakana; $str =~ s/([$B%!(B-$B%s(B])/$k2h{$1}/go;
+is($str, $hiragana, "s/// # hiragana -> katakana");
+__END__
diff --git a/t/uni/tr_eucjp.t b/t/uni/tr_eucjp.t
new file mode 100644
index 0000000000..e13acab0ed
--- /dev/null
+++ b/t/uni/tr_eucjp.t
@@ -0,0 +1,59 @@
+#
+# $Id$
+#
+# This script is written intentionally in EUC-JP
+# -- dankogai
+
+BEGIN {
+    if ($ENV{'PERL_CORE'}){
+        chdir 't';
+        unshift @INC, '../lib';
+    }
+    require Config; import Config;
+    if ($Config{'extensions'} !~ /\bEncode\b/) {
+      print "1..0 # Skip: Encode was not built\n";
+      exit 0;
+    }
+    if (ord("A") == 193) {
+        print "1..0 # Skip: EBCDIC\n";
+        exit 0;
+    }
+    unless (PerlIO::Layer->find('perlio')){
+        print "1..0 # Skip: PerlIO required\n";
+        exit 0;
+    }
+    $| = 1;
+}
+
+use strict;
+#use Test::More qw(no_plan);
+use Test::More tests => 6;
+use Encode;
+use encoding 'euc-jp';
+
+my @hiragana =  map {chr} ord("ぁ")..ord("ん");
+my @katakana =  map {chr} ord("ァ")..ord("ン");
+my $hiragana = join('' => @hiragana);
+my $katakana = join('' => @katakana);
+my %h2k; @h2k{@hiragana} = @katakana;
+my %k2h; @k2h{@katakana} = @hiragana;
+
+# print @hiragana, "\n";
+
+my $str;
+
+$str = $hiragana; $str =~ tr/ぁ-ん/ァ-ン/;
+is($str, $katakana, "tr// # hiragana -> katakana");
+$str = $katakana; $str =~ tr/ァ-ン/ぁ-ん/;
+is($str, $hiragana, "tr// # hiragana -> katakana");
+
+$str = $hiragana; eval qq(\$str =~ tr/ぁ-ん/ァ-ン/);
+is($str, $katakana, "eval qq(tr//) # hiragana -> katakana");
+$str = $katakana; eval qq(\$str =~ tr/ァ-ン/ぁ-ん/);
+is($str, $hiragana, "eval qq(tr//) # hiragana -> katakana");
+
+$str = $hiragana; $str =~ s/([ぁ-ん])/$h2k{$1}/go;
+is($str, $katakana, "s/// # hiragana -> katakana");
+$str = $katakana; $str =~ s/([ァ-ン])/$k2h{$1}/go;
+is($str, $hiragana, "s/// # hiragana -> katakana");
+__END__
diff --git a/t/uni/tr_sjis.t b/t/uni/tr_sjis.t
new file mode 100644
index 0000000000..f5ad045f0e
--- /dev/null
+++ b/t/uni/tr_sjis.t
@@ -0,0 +1,59 @@
+#
+# $Id$
+#
+# This script is written intentionally in EUC-JP
+# -- dankogai
+
+BEGIN {
+    if ($ENV{'PERL_CORE'}){
+        chdir 't';
+        unshift @INC, '../lib';
+    }
+    require Config; import Config;
+    if ($Config{'extensions'} !~ /\bEncode\b/) {
+      print "1..0 # Skip: Encode was not built\n";
+      exit 0;
+    }
+    if (ord("A") == 193) {
+        print "1..0 # Skip: EBCDIC\n";
+        exit 0;
+    }
+    unless (PerlIO::Layer->find('perlio')){
+        print "1..0 # Skip: PerlIO required\n";
+        exit 0;
+    }
+    $| = 1;
+}
+
+use strict;
+#use Test::More qw(no_plan);
+use Test::More tests => 6;
+use Encode;
+use encoding 'shiftjis';
+
+my @hiragana =  map {chr} ord("��")..ord("��");
+my @katakana =  map {chr} ord("�@")..ord("��");
+my $hiragana = join('' => @hiragana);
+my $katakana = join('' => @katakana);
+my %h2k; @h2k{@hiragana} = @katakana;
+my %k2h; @k2h{@katakana} = @hiragana;
+
+# print @hiragana, "\n";
+
+my $str;
+
+$str = $hiragana; $str =~ tr/��-��/�@-��/;
+is($str, $katakana, "tr// # hiragana -> katakana");
+$str = $katakana; $str =~ tr/�@-��/��-��/;
+is($str, $hiragana, "tr// # hiragana -> katakana");
+
+$str = $hiragana; eval qq(\$str =~ tr/��-��/�@-��/);
+is($str, $katakana, "eval qq(tr//) # hiragana -> katakana");
+$str = $katakana; eval qq(\$str =~ tr/�@-��/��-��/);
+is($str, $hiragana, "eval qq(tr//) # hiragana -> katakana");
+
+$str = $hiragana; $str =~ s/([��-��])/$h2k{$1}/go;
+is($str, $katakana, "s/// # hiragana -> katakana");
+$str = $katakana; $str =~ s/([�@-��])/$k2h{$1}/go;
+is($str, $hiragana, "s/// # hiragana -> katakana");
+__END__
diff --git a/t/uni/tr_utf8.t b/t/uni/tr_utf8.t
new file mode 100644
index 0000000000..54b9b4ff37
--- /dev/null
+++ b/t/uni/tr_utf8.t
@@ -0,0 +1,62 @@
+#
+# $Id$
+#
+# This script is written intentionally in EUC-JP
+# -- dankogai
+
+BEGIN {
+    if ($ENV{'PERL_CORE'}){
+        chdir 't';
+        unshift @INC, '../lib';
+    }
+    require Config; import Config;
+    if ($Config{'extensions'} !~ /\bEncode\b/) {
+      print "1..0 # Skip: Encode was not built\n";
+      exit 0;
+    }
+    if (ord("A") == 193) {
+        print "1..0 # Skip: EBCDIC\n";
+        exit 0;
+    }
+    unless (PerlIO::Layer->find('perlio')){
+        print "1..0 # Skip: PerlIO required\n";
+        exit 0;
+    }
+    $| = 1;
+}
+
+use strict;
+#use Test::More qw(no_plan);
+use Test::More tests => 6;
+
+# use encoding 'utf8'; # you can't uncomment this!
+# if you uncomment above, you'll get the following (as of Encode 1.80)
+#  Assertion ((dst)->sv_flags & 0xff) >= SVt_PV failed:
+#  file "Encode.xs", line 255 at t/uni/tr_utf8.t line 35.
+
+my @hiragana =  map {chr} ord("���")..ord("���");
+my @katakana =  map {chr} ord("���")..ord("���");
+my $hiragana = join('' => @hiragana);
+my $katakana = join('' => @katakana);
+my %h2k; @h2k{@hiragana} = @katakana;
+my %k2h; @k2h{@katakana} = @hiragana;
+
+# print @hiragana, "\n";
+
+my $str;
+
+$str = $hiragana; $str =~ tr/���-���/���-���/;
+is($str, $katakana, "tr// # hiragana -> katakana");
+$str = $katakana; $str =~ tr/���-���/���-���/;
+is($str, $hiragana, "tr// # hiragana -> katakana");
+
+$str = $hiragana; eval qq(\$str =~ tr/���-���/���-���/);
+is($str, $katakana, "eval qq(tr//) # hiragana -> katakana");
+$str = $katakana; eval qq(\$str =~ tr/���-���/���-���/);
+is($str, $hiragana, "eval qq(tr//) # hiragana -> katakana");
+
+$str = $hiragana; $str =~ s/([���-���])/$h2k{$1}/go;
+is($str, $katakana, "s/// # hiragana -> katakana");
+$str = $katakana; $str =~ s/([���-���])/$k2h{$1}/go;
+is($str, $hiragana, "s/// # hiragana -> katakana");
+__END__