Fix mbstring shim for html-entities

tchwork · Jan 12, 2015 · ffa0821 · ffa0821
1 parent f26cc98
commit ffa0821
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 3 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,3 +1,7 @@
+## v1.1.28 (2015-01-12)
+
+- fix mbstring shim for html-entities
+
 ## v1.1.27 (2015-01-11)
 
 - update to Unicode 7.0

diff --git a/class/Patchwork/PHP/Shim/Mbstring.php b/class/Patchwork/PHP/Shim/Mbstring.php
@@ -420,7 +420,27 @@ protected static function getSubpart($pos, $part, $haystack, $encoding)
 
     protected static function html_encoding_callback($m)
     {
-        return htmlentities($m[0], ENT_COMPAT, 'UTF-8');
+        $i = 1;
+        $entities = '';
+        $m = unpack('C*', htmlentities($m[0], ENT_COMPAT, 'UTF-8'));
+
+        while (isset($m[$i])) {
+            if (0x80 > $m[$i]) {
+                $entities .= chr($m[$i++]);
+                continue;
+            }
+            if (0xF0 <= $m[$i]) {
+                $c = (($m[$i++] - 0xF0) << 18) + (($m[$i++] - 0x80) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++] - 0x80;
+            } elseif (0xE0 <= $m[$i]) {
+                $c = (($m[$i++] - 0xE0) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++]  - 0x80;
+            } else {
+                $c = (($m[$i++] - 0xC0) << 6) + $m[$i++] - 0x80;
+            }
+
+            $entities .= '&#'.$c.';';
+        }
+
+        return $entities;
     }
 
     protected static function title_case_lower($s)

diff --git a/tests/Patchwork/Tests/PHP/Shim/MbstringTest.php b/tests/Patchwork/Tests/PHP/Shim/MbstringTest.php
@@ -38,9 +38,9 @@ function testmb_convert_encoding()
     {
         $this->assertSame( utf8_decode('déjà'), p::mb_convert_encoding('déjà', 'Windows-1252') );
         $this->assertSame( base64_encode('déjà'), p::mb_convert_encoding('déjà', 'Base64') );
-        $this->assertSame( 'd&eacute;j&agrave;', p::mb_convert_encoding('déjà', 'Html-entities') );
+        $this->assertSame( '&#23455;<&>d&eacute;j&agrave;', p::mb_convert_encoding('実<&>déjà', 'Html-entities') );
         $this->assertSame( 'déjà', p::mb_convert_encoding(base64_encode('déjà'), 'Utf-8', 'Base64') );
-        $this->assertSame( 'déjà', p::mb_convert_encoding('d&eacute;j&agrave;', 'Utf-8', 'Html-entities') );
+        $this->assertSame( 'déjà', p::mb_convert_encoding('d&eacute;j&#224;', 'Utf-8', 'Html-entities') );
     }
 
     /**