Encode/Decode numeric HTML entities correctly.

utf8_tohtml handles all codepoints, and the inverse
function, utf8_unhtml, is added.

darcs-hash:20070202070509-6942e-09ed9dc37f1469055a7c04d44044768e160d60e6.gz

1 files changed, 72 insertions, 0 deletions

diff --git a/_test/cases/inc/utf8_html.test.php b/_test/cases/inc/utf8_html.test.php
new file mode 100644
index 000000000..57c9df259
--- /dev/null
+++ b/_test/cases/inc/utf8_html.test.php
@@ -0,0 +1,72 @@
+<?php
+
+require_once DOKU_INC.'inc/utf8.php';
+
+// use no mbstring help here
+if(!defined('UTF8_NOMBSTRING')) define('UTF8_NOMBSTRING',1);
+
+class utf8_html_test extends UnitTestCase {
+
+    function test_from_1byte(){
+        $in  = 'a';
+        $out = 'a';
+        $this->assertEqual(utf8_tohtml($in),$out);
+    }
+
+    function test_from_2byte(){
+        $in  = "\xc3\xbc";
+        $out = '&#252;';
+        $this->assertEqual(utf8_tohtml($in),$out);
+    }
+
+    function test_from_3byte(){
+        $in  = "\xe2\x99\x8a";
+        $out = '&#x264a;';
+        $this->assertEqual(utf8_tohtml($in),$out);
+    }
+
+    function test_from_4byte(){
+        $in  = "\xf4\x80\x80\x81";
+        $out = '&#x100001;';
+        $this->assertEqual(utf8_tohtml($in),$out);
+    }
+
+    function test_to_1byte(){
+        $out  = 'a';
+        $in = 'a';
+        $this->assertEqual(utf8_unhtml($in),$out);
+    }
+
+    function test_to_2byte(){
+        $out  = "\xc3\xbc";
+        $in = '&#252;';
+        $this->assertEqual(utf8_unhtml($in),$out);
+    }
+
+    function test_to_3byte(){
+        $out  = "\xe2\x99\x8a";
+        $in = '&#x264a;';
+        $this->assertEqual(utf8_unhtml($in),$out);
+    }
+
+    function test_to_4byte(){
+        $out  = "\xf4\x80\x80\x81";
+        $in = '&#x100001;';
+        $this->assertEqual(utf8_unhtml($in),$out);
+    }
+
+    function test_without_entities(){
+        $out  = '&amp;#38;&amp;#38;';
+        $in = '&amp;#38;&#38;amp;#38;';
+        $this->assertEqual(utf8_unhtml($in),$out);
+    }
+
+    function test_with_entities(){
+        $out  = '&#38;&amp;#38;';
+        $in = '&amp;#38;&#38;amp;#38;';
+        $this->assertEqual(utf8_unhtml($in,HTML_ENTITIES),$out);
+    }
+
+}
+
+//Setup VIM: ex: et ts=4 enc=utf-8 :