Add U+0085, U+2028, and U+2029 to \s under Unicode.

p4raw-id: //depot/perl@14052

diff --git a/lib/unicore/Is/SpacePerl.pl b/lib/unicore/Is/SpacePerl.pl
index 3866aee..cc8f50a 100644 (file)
--- a/lib/unicore/Is/SpacePerl.pl
+++ b/lib/unicore/Is/SpacePerl.pl
@@ -5,6 +5,7 @@ return <<'END';
 0009   000A    
 000C   000D    
 0020           
+0085           
 00A0           
 1680           
 2000   200B    

diff --git a/lib/unicore/mktables b/lib/unicore/mktables
index 34d1388..067af24 100644 (file)
--- a/lib/unicore/mktables
+++ b/lib/unicore/mktables
@@ -134,7 +134,7 @@ sub gencat {
                            ||= [],  $code);
     # 005F: SPACING UNDERSCORE
     $op->($Cat->{Word}       ||= [], $code)
-       if $cat =~ /^[LMN]/ or $code eq "005F";
+       if $cat =~ /^[LMN]/ || $code eq "005F";
     $op->($Cat->{Alnum}      ||= [], $code)
        if $cat =~ /^[LMN]/;
     $op->($Cat->{Alpha}      ||= [], $code)
@@ -149,7 +149,7 @@ sub gencat {
        if $cat  =~ /^Z/ ||
            $code =~ /^(0009|000A|000B|000C|000D)$/;
     $op->($Cat->{SpacePerl}  ||= [], $code)
-       if $cat  =~ /^Z/ ||
+       if $cat  =~ /^Z/ || $code =~ /^(?:0085|2028|2029)$/ ||
            $code =~ /^(0009|000A|000C|000D)$/;
     $op->($Cat->{Blank}      ||= [], $code)
        if $code =~ /^(0020|0009)$/ ||

diff --git a/t/op/pat.t b/t/op/pat.t
index 969e2cd..bc2ed37 100755 (executable)
--- a/t/op/pat.t
+++ b/t/op/pat.t
@@ -6,7 +6,7 @@
 
 $| = 1;
 
-print "1..840\n";
+print "1..843\n";
 
 BEGIN {
     chdir 't' if -d 't';
@@ -2609,3 +2609,13 @@ print "# some Unicode properties\n";
 #    print "SS" =~
 #      /[\N{LATIN SMALL LETTER SHARP S}]/i ? "ok 842\n" : "not ok 842\n";
 }
+
+{
+    print "# more whitespace: U+0085, U+2028, U+2029\n";
+
+    # U+0085 needs to be forced to be Unicode, the \x{100} does that.
+    print "<\x{100}\x{0085}>" =~ /<\x{100}\s>/ ? "ok 841\n" : "not ok 841\n";
+    print "<\x{2028}>" =~ /<\s>/ ? "ok 842\n" : "not ok 842\n";
+    print "<\x{2029}>" =~ /<\s>/ ? "ok 843\n" : "not ok 843\n";
+}
+