1 files changed, 33 insertions, 4 deletions

diff --git a/t/pagetitle.t b/t/pagetitle.t
index d9aa62063..d00d9d297 100755
--- a/t/pagetitle.t
+++ b/t/pagetitle.t
@@ -1,13 +1,42 @@
 #!/usr/bin/perl
 use warnings;
 use strict;
-use Test::More tests => 7;
+use Test::More;
 
 BEGIN { use_ok("IkiWiki"); }
 
+# pagetitle(x) => XML-escaped form of page title
 is(pagetitle("foo_bar"), "foo bar");
 is(pagetitle("foo_bar_baz"), "foo bar baz");
 is(pagetitle("foo_bar__33__baz"), "foo bar!baz");
-is(pagetitle("foo_bar__1234__baz"), "foo barӒbaz");
-is(pagetitle("foo_bar___33___baz"), "foo bar ! baz");
-is(pagetitle("foo_bar___95___baz"), "foo bar _ baz");
+# &#1234 is U+04D2 CYRILLIC CAPITAL LETTER A WITH DIAERESIS
+is(pagetitle("foo_bar__1234__baz"), "foo barӒbaz", 'Unicode in BMP');
+# &#8800 is U+2260 NOT EQUAL TO
+is(pagetitle("foo_bar__8800__baz"), "foo bar≠baz", 'Unicode in BMP');
+is(pagetitle("foo_bar___33___baz"), "foo bar ! baz", 'Exclamation mark');
+is(pagetitle("foo_bar___95___baz"), "foo bar _ baz", 'Underscore');
+# Outside basic multilingual plane: &#128169 is U+1F4A9 PILE OF POO
+is(pagetitle("foo_bar__128169__baz"), "foo bar💩baz", 'Unicode outside BMP');
+
+# pagetitle(x, false) => same
+is(pagetitle("foo_bar__33__baz", 0), "foo bar!baz");
+is(pagetitle("foo_bar__1234__baz", undef), "foo barӒbaz", 'Unicode in BMP');
+is(pagetitle("foo_bar__8800__baz", undef), "foo bar≠baz", 'Unicode in BMP');
+is(pagetitle("foo_bar___33___baz", ""), "foo bar ! baz", 'Exclamation mark');
+is(pagetitle("foo_bar___95___baz", 0), "foo bar _ baz", 'Underscore');
+is(pagetitle("foo_bar__128169__baz", 0), "foo bar💩baz", 'Unicode outside BMP');
+
+# pagetitle(x, true) => unescaped form of page title
+is(pagetitle("foo_bar", 1), "foo bar");
+is(pagetitle("foo_bar_baz", 'unescaped'), "foo bar baz");
+is(pagetitle("foo_bar__33__baz", 42), "foo bar!baz");
+is(chr(1234), "\x{04D2}");
+is(pagetitle("foo_bar__1234__baz", 1), "foo bar\x{04D2}baz", 'Unicode in BMP');
+is(chr(8800), "\x{2260}");
+is(pagetitle("foo_bar__8800__baz", 1), "foo bar\x{2260}baz", 'Unicode in BMP');
+is(pagetitle("foo_bar___33___baz", 1), "foo bar ! baz");
+is(pagetitle("foo_bar___95___baz", 1), "foo bar _ baz");
+is(chr(128169), "\x{0001F4A9}");
+is(pagetitle("foo_bar__128169__baz", 1), "foo bar\x{0001F4A9}baz", 'Unicode outside BMP');
+
+done_testing;