Added function to convert ISO-8559-1 to UTF-8

Author: evert

lib/Sabre/DAV/StringUtil.php

@@ -64,6 +64,27 @@ static public function textMatch($haystack, $needle, $collation, $matchType = 'c
 
         }
 
+    }
+
+    /**
+     * This method takes an input string, checks if it's not valid UTF-8 and 
+     * attempts to convert it to UTF-8 if it's not.
+     *
+     * Note that currently this can only convert ISO-8559-1 to UTF-8 (latin-1), 
+     * anything else will likely fail.
+     *
+     * @param string $input 
+     * @return string 
+     */
+    static public function ensureUTF8($input) {
+
+        $encoding = mb_detect_encoding($input , array('UTF-8','ISO-8859-1'), true);
+
+        if ($encoding === 'ISO-8859-1') {
+            return utf8_encode($input);
+        } else {
+            return $input;
+        } 
 
     }
 

lib/Sabre/DAV/URLUtil.php

@@ -86,6 +86,7 @@ static function decodePathSegment($path) {
 
             case 'ISO-8859-1' : 
                 $path = utf8_encode($path);
+
         }
 
         return $path;

tests/Sabre/DAV/StringUtilTest.php

@@ -5,7 +5,7 @@ class Sabre_DAV_StringUtilTest extends PHPUnit_Framework_TestCase {
     /**
      * @dataProvider dataset
      */ 
-    function testEverything($haystack, $needle, $collation, $matchType, $result) {
+    function testTextMatch($haystack, $needle, $collation, $matchType, $result) {
 
         $this->assertEquals($result, Sabre_DAV_StringUtil::textMatch($haystack, $needle, $collation, $matchType));
 
@@ -81,4 +81,40 @@ public function testBadMatchType() {
 
     }
 
+    public function testEnsureUTF8_ascii() {
+
+        $inputString = "harkema";
+        $outputString = "harkema";
+
+        $this->assertEquals(
+            $outputString,
+            Sabre_DAV_StringUtil::ensureUTF8($inputString)
+        );
+
+    }
+
+    public function testEnsureUTF8_latin1() {
+
+        $inputString = "m\xfcnster";
+        $outputString = "münster";
+
+        $this->assertEquals(
+            $outputString,
+            Sabre_DAV_StringUtil::ensureUTF8($inputString)
+        );
+
+    }
+
+    public function testEnsureUTF8_utf8() {
+
+        $inputString = "m\xc3\xbcnster";
+        $outputString = "münster";
+
+        $this->assertEquals(
+            $outputString,
+            Sabre_DAV_StringUtil::ensureUTF8($inputString)
+        );
+
+    }
+
 }