pagetitle.t, linkpage.t, titlepage.t: Exercise Unicode more

Signed-off-by: Simon McVittie <smcv@debian.org>

diff --git a/t/linkpage.t b/t/linkpage.t
index 8085de15321240de18e334a064864db721fa84b3..db5fc7ae580d979cf60d582aee7c7486b1117966 100755 (executable)
--- a/t/linkpage.t
+++ b/t/linkpage.t
@@ -1,7 +1,7 @@
 #!/usr/bin/perl
 use warnings;
 use strict;
-use Test::More tests => 7;
+use Test::More;
 
 BEGIN { use_ok("IkiWiki"); }
 
@@ -11,3 +11,9 @@ is(linkpage("foo bar/baz"), "foo_bar/baz");
 is(linkpage("foo bar&baz"), "foo_bar__38__baz");
 is(linkpage("foo bar & baz"), "foo_bar___38___baz");
 is(linkpage("foo bar_baz"), "foo_bar_baz");
+is(linkpage("foo bar\xACbaz"), "foo_bar__172__baz", 'U+00AC is in Latin-1 range');
+is(linkpage("foo bar\x{04D2}baz"), "foo_bar\x{04D2}baz", 'U+04D2 is alphanumeric');
+is(linkpage("foo bar\x{2260}baz"), "foo_bar__8800__baz", 'U+2260 is nonalphanumeric');
+is(linkpage("foo bar\x{0001F4A9}baz"), "foo_bar__128169__baz", 'U+1F4A9 is outside BMP');
+
+done_testing;

diff --git a/t/pagetitle.t b/t/pagetitle.t
index d9aa620633d6b64e393a366c19a510b3e851542f..d00d9d29792cb517865621f864b028a5c23456ad 100755 (executable)
--- 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;

diff --git a/t/titlepage.t b/t/titlepage.t
index 5df33423e97052c4bad1ada8607fc8f44a505d4f..d6bb1b12116b052bfe925fc5e5443a2224d6aeec 100755 (executable)
--- a/t/titlepage.t
+++ b/t/titlepage.t
@@ -1,7 +1,7 @@
 #!/usr/bin/perl
 use warnings;
 use strict;
-use Test::More tests => 7;
+use Test::More;
 
 BEGIN { use_ok("IkiWiki"); }
 
@@ -11,3 +11,9 @@ is(titlepage("foo bar/baz"), "foo_bar/baz");
 is(titlepage("foo bar&baz"), "foo_bar__38__baz");
 is(titlepage("foo bar & baz"), "foo_bar___38___baz");
 is(titlepage("foo bar_baz"), "foo_bar__95__baz");
+is(titlepage("foo bar\xACbaz"), "foo_bar__172__baz", 'U+00AC is in Latin-1 range');
+is(titlepage("foo bar\x{04D2}baz"), "foo_bar\x{04D2}baz", 'U+04D2 is alphanumeric');
+is(titlepage("foo bar\x{2260}baz"), "foo_bar__8800__baz", 'U+2260 is nonalphanumeric');
+is(titlepage("foo bar\x{0001F4A9}baz"), "foo_bar__128169__baz", 'U+1F4A9 is outside BMP');
+
+done_testing;