[+]: add some more tests v2

voku · Oct 24, 2016 · 9dba3d1 · 9dba3d1
1 parent ae6d423
commit 9dba3d1
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 20 deletions.
diff --git a/README.md b/README.md
@@ -823,9 +823,9 @@ Remove invisible characters from a string.
 UTF8::remove_invisible_characters("κόσ\0με"); // 'κόσμε'
 ```
 
-##### replace_diamond_question_mark(string $str, string $unknown = '?') : string
+##### replace_diamond_question_mark(string $str, string $replacementChar = '', bool $processInvalidUtf8 = true) : string
 
-Replace the diamond question mark (�) with the replacement.
+Replace the diamond question mark (�) and invalid-UTF8 chars with the replacement.
 
 ```php
 UTF8::replace_diamond_question_mark('中文空白�', ''); // '中文空白'

diff --git a/src/voku/helper/UTF8.php b/src/voku/helper/UTF8.php
@@ -3786,44 +3786,47 @@ public static function remove_invisible_characters($str, $url_encoded = true, $r
   }
 
   /**
-   * Replace the diamond question mark (�) with the replacement.
+   * Replace the diamond question mark (�) and invalid-UTF8 chars with the replacement.
    *
-   * @param string $str <p>The input string</p>
-   * @param string $unknown <p>The replacement character.</p>
+   * @param string $str                <p>The input string</p>
+   * @param string $replacementChar    <p>The replacement character.</p>
+   * @param bool   $processInvalidUtf8 <p>Convert invalid UTF-8 chars </p>
    *
    * @return string
    */
-  public static function replace_diamond_question_mark($str, $unknown = '?')
+  public static function replace_diamond_question_mark($str, $replacementChar = '', $processInvalidUtf8 = true)
   {
     $str = (string)$str;
 
     if (!isset($str[0])) {
       return '';
     }
 
-    $unknownHelper = $unknown;
-    if ($unknown === '') {
-      $unknownHelper = 'none';
-    }
+    if ($processInvalidUtf8 === true) {
+      $replacementCharHelper = $replacementChar;
+      if ($replacementChar === '') {
+        $replacementCharHelper = 'none';
+      }
 
-    if (self::$support['mbstring'] === false) {
-      trigger_error('UTF8::replace_diamond_question_mark() without mbstring cannot handle all chars correctly', E_USER_WARNING);
-    }
+      if (self::$support['mbstring'] === false) {
+        trigger_error('UTF8::replace_diamond_question_mark() without mbstring cannot handle all chars correctly', E_USER_WARNING);
+      }
 
-    $save = \mb_substitute_character();
-    \mb_substitute_character($unknownHelper);
-    /** @noinspection CallableParameterUseCaseInTypeContextInspection */
-    $str = \mb_convert_encoding($str, 'UTF-8', 'UTF-8');
-    \mb_substitute_character($save);
+      $save = \mb_substitute_character();
+      \mb_substitute_character($replacementCharHelper);
+      /** @noinspection CallableParameterUseCaseInTypeContextInspection */
+      $str = \mb_convert_encoding($str, 'UTF-8', 'UTF-8');
+      \mb_substitute_character($save);
+    }
 
     return str_replace(
         array(
             "\xEF\xBF\xBD",
             '�',
         ),
         array(
-            $unknown,
-            $unknown,
+            $replacementChar,
+            $replacementChar,
         ),
         $str
     );

diff --git a/tests/Utf8GlobalTest.php b/tests/Utf8GlobalTest.php
@@ -2091,6 +2091,14 @@ public function testReplaceDiamondQuestionMark()
       self::assertSame($after, UTF8::replace_diamond_question_mark($before, ''), 'tested: ' . $before . ' | counter: ' . $counter);
       ++$counter;
     }
+
+    // ---
+
+    self::assertSame('Iñtërnâtiônàlizætiøn??Iñtërnâtiônàlizætiøn', UTF8::replace_diamond_question_mark("Iñtërnâtiônàlizætiøn\xa0\xa1Iñtërnâtiônàlizætiøn", '?', true));
+
+    // ---
+
+    self::assertSame("Iñtërnâtiônàlizætiøn\xa0\xa1Iñtërnâtiônàlizætiøn", UTF8::replace_diamond_question_mark("Iñtërnâtiônàlizætiøn\xa0\xa1Iñtërnâtiônàlizætiøn", '?', false));
   }
 
   public function testRtrim()