[~]: optimize "UTF8::is_binary()" + more tests

voku · Dec 24, 2016 · cd7fbc3 · cd7fbc3
1 parent d98de59
commit cd7fbc3
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 10 deletions.
diff --git a/src/voku/helper/UTF8.php b/src/voku/helper/UTF8.php
@@ -2737,20 +2737,16 @@ public static function is_base64($str)
    */
   public static function is_binary($input)
   {
+    if (preg_match('~^[01]+$~', $input)) {
+      return true;
+    }
 
     $testLength = strlen($input);
-
-    if (
-        preg_match('~^[01]+$~', $input)
-        ||
-        substr_count($input, "\x00") > 0
-        ||
-        ($testLength ? substr_count($input, '^ -~') / $testLength > 0.3 : 1 === 0)
-    ) {
+    if (substr_count($input, "\x0") / $testLength > 0.3) {
       return true;
-    } else {
-      return false;
     }
+
+    return false;
   }
 
   /**

diff --git a/tests/Utf8GlobalTest.php b/tests/Utf8GlobalTest.php
@@ -1365,8 +1365,20 @@ public function testIsBinary()
         'öäü'          => false,
         ''             => false,
         '1'            => false,
+        '01010101'     => true,
         decbin(324546) => true,
         01             => true,
+        1020304        => false,
+        01020304       => false,
+        11020304       => false,
+        '1010101'      => true,
+        11111111       => true,
+        00000000       => true,
+        "\x00\x01"     => true,
+        "\x01\x00"     => true,
+        "\x01\x02"     => false,
+        "\x01\x00ab"   => false, // <= 30% binary
+        "\x01\x00a"    => true, // >= 30% binary
     );
 
     foreach ($tests as $before => $after) {