PATCH: Large omnibus patch to clean up the JRRT quotes

[p5sagit/p5-mst-13.2.git] / ext / Encode / t / guess.t

diff --git a/ext/Encode/t/guess.t b/ext/Encode/t/guess.t
index 563bc6f..5bfbf4e 100644 (file)
--- a/ext/Encode/t/guess.t
+++ b/ext/Encode/t/guess.t
@@ -9,8 +9,8 @@ BEGIN {
       exit 0;
     }
     if (ord("A") == 193) {
-       print "1..0 # Skip: EBCDIC\n";
-       exit 0;
+    print "1..0 # Skip: EBCDIC\n";
+    exit 0;
     }
     $| = 1;
 }
@@ -21,7 +21,7 @@ use File::Spec;
 use Encode qw(decode encode find_encoding _utf8_off);
 
 #use Test::More qw(no_plan);
-use Test::More tests => 17;
+use Test::More tests => 29;
 use_ok("Encode::Guess");
 {
     no warnings;
@@ -48,6 +48,7 @@ my $jisx0208 = File::Spec->catfile(dirname(__FILE__), 'jisx0208.utf');
 my $jisx0212 = File::Spec->catfile(dirname(__FILE__), 'jisx0212.utf');
 
 open my $fh, $jisx0208 or die "$jisx0208: $!";
+binmode($fh);
 $utf8off = join('' => <$fh>);
 close $fh;
 $utf8on = decode('utf8', $utf8off);
@@ -77,6 +78,7 @@ Encode::Guess->set_suspects(keys %CJKT);
 
 for my $name (keys %CJKT){
     open my $fh, $CJKT{$name} or die "$CJKT{$name}: $!";
+    binmode($fh);
     $utf8off = join('' => <$fh>);
     close $fh;
 
@@ -84,4 +86,32 @@ for my $name (keys %CJKT){
     is(guess_encoding($test)->name, $name, "CJKT:$name");
 }
 
+my $ambiguous =  "\x{5c0f}\x{98fc}\x{5f3e}";
+my $english   = "The quick brown fox jumps over the black lazy dog.";
+for my $utf (qw/UTF-16 UTF-32/){
+    for my $bl (qw/BE LE/){
+    my $test = encode("$utf$bl" => $english);
+    is(guess_encoding($test)->name, "$utf$bl", "$utf$bl");
+    }
+}
+for my $bl (qw/BE LE/){
+    my $test = encode("UTF-16$bl" => $ambiguous);
+    my $result = guess_encoding($test);
+    ok(! ref($result), "UTF-16$bl:$result");
+}
+
+
+
+Encode::Guess->set_suspects();
+for my $jp (@jp){
+    # intentionally set $1 a priori -- see Changes
+    my $test = "English";
+    '$1' =~ m/^(.*)/o;
+    is(guess_encoding($test, ($jp))->name, 'ascii', 
+       "ascii vs $jp (\$1 messed)");
+    $test = encode($jp, $test . "\n\x{65e5}\x{672c}\x{8a9e}");
+    is(guess_encoding($test, ($jp))->name, 
+       $jp, "$jp vs ascii (\$1 messed)");
+}
+
 __END__;