X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=t%2Funi%2Fcase.pl;h=b6df5a8089bc42cb5fde6bac9e9341ef407db79f;hb=1858f5c364ee8a037ee41664a9d5964814b7e038;hp=25f8f4c97738203603200f20056e78899c2efe5c;hpb=23cd9172ab4c1e71bdfc3b5ce019766a0c5f966c;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/t/uni/case.pl b/t/uni/case.pl
index 25f8f4c..b6df5a8 100644
--- a/t/uni/case.pl
+++ b/t/uni/case.pl
@@ -2,6 +2,10 @@ use File::Spec;
 
 require "test.pl";
 
+sub unidump {
+    join " ", map { sprintf "%04X", $_ } unpack "U*", $_[0];
+}
+
 sub casetest {
     my ($base, $spec, $func) = @_;
     my $file = File::Spec->catfile(File::Spec->catdir(File::Spec->updir,
@@ -23,10 +27,15 @@ sub casetest {
     my $both;
 
     for my $i (sort keys %$spec) {
-	$both++ if ++$seen{hex $i} == 2;
+	if (++$seen{hex $i} == 2) {
+	    warn "$base: $i seen twice\n";
+	    $both++;
+	}
     }
     print "# ", scalar keys %$spec, " special mappings\n";
 
+    exit(1) if $both;
+
     my %none;
     for my $i (map { ord } split //,
 	       "\e !\"#\$%&'()+,-./0123456789:;<=>?\@[\\]^_{|}~\b") {
@@ -38,37 +47,86 @@ sub casetest {
     my $tests = 
 	(scalar keys %simple) +
 	(scalar keys %$spec) +
-	(scalar keys %none) - $both;
+	(scalar keys %none);
     print "1..$tests\n";
 
     my $test = 1;
 
     for my $i (sort { hex $a <=> hex $b } keys %simple) {
-	my $w = "$i -> $simple{$i}";
+	my $w = $simple{$i};
 	my $c = pack "U0U", hex $i;
 	my $d = $func->($c);
+	my $e = unidump($d);
 	print $d eq pack("U0U", hex $simple{$i}) ?
-	    "ok $test # $w\n" : "not ok $test # $w\n";
+	    "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
 	$test++;
     }
 
     for my $i (sort { hex $a <=> hex $b } keys %$spec) {
-	next if $seen{hex $i} == 2;
-	my $w = qq[$i -> "] . display($spec->{$i}) . qq["];
+	my $w = unidump($spec->{$i});
 	my $c = pack "U0U", hex $i;
 	my $d = $func->($c);
-	print $d eq $spec->{$i} ?
-	    "ok $test # $w\n" : "not ok $test # $w\n";
+	my $e = unidump($d);
+	if (ord "A" == 193) { # EBCDIC
+	    # We need to a little bit of remapping.
+	    #
+	    # For example, in titlecase (ucfirst) mapping
+	    # of U+0149 the Unicode mapping is U+02BC U+004E.
+	    # The 4E is N, which in EBCDIC is 2B--
+	    # and the ucfirst() does that right.
+	    # The problem is that our reference
+	    # data is in Unicode code points.
+	    #
+	    # The Right Way here would be to use, say,
+	    # Encode, to remap the less-than 0x100 code points,
+	    # but let's try to be Encode-independent here. 
+	    #
+	    # These are the titlecase exceptions:
+	    #
+	    #         Unicode   Unicode+EBCDIC  
+	    #
+	    # 0149 -> 02BC 004E (02BC 002B)
+	    # 01F0 -> 004A 030C (00A2 030C)
+	    # 1E96 -> 0048 0331 (00E7 0331)
+	    # 1E97 -> 0054 0308 (00E8 0308)
+	    # 1E98 -> 0057 030A (00EF 030A)
+	    # 1E99 -> 0059 030A (00DF 030A)
+	    # 1E9A -> 0041 02BE (00A0 02BE)
+	    #
+	    # The uppercase exceptions are identical.
+	    #
+	    # The lowercase has one more:
+	    #
+	    #         Unicode   Unicode+EBCDIC  
+	    #
+	    # 0130 -> 0069 0307 (00D1 0307)
+	    #
+	    if ($i =~ /^(0130|0149|01F0|1E96|1E97|1E98|1E99|1E9A)$/) {
+		$e =~ s/004E/002B/; # N
+		$e =~ s/004A/00A2/; # J
+		$e =~ s/0048/00E7/; # H
+		$e =~ s/0054/00E8/; # T
+		$e =~ s/0057/00EF/; # W
+		$e =~ s/0059/00DF/; # Y
+		$e =~ s/0041/00A0/; # A
+		$e =~ s/0069/00D1/; # i
+	    }
+	    # We have to map the output, not the input, because
+	    # pack/unpack U has been EBCDICified, too, it would
+	    # just undo our remapping.
+	}
+	print $w eq $e ?
+	    "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
 	$test++;
     }
 
-
     for my $i (sort { $a <=> $b } keys %none) {
-	my $w = sprintf "%04X -> %04X", $i, $i;
-	my $c = pack "U0U", $i;
+	my $w = $i = sprintf "%04X", $i;
+	my $c = pack "U0U", hex $i;
 	my $d = $func->($c);
+	my $e = unidump($d);
 	print $d eq $c ?
-	    "ok $test # $w\n" : "not ok $test # $w\n";
+	    "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
 	$test++;
     }
 }