Update Serializer to account for UTF-8 chars (#553)

* Update Serializer.php * Update SerializerTest.php * Add test for clipping UTF-8 characters * Fix long string test not counting non-ascii characters correctly * Specify encoding and make sure the mbstring extension is loaded * Only use printable ascii in long string serialize test * Fix testLongString tests (last time)
getsentry · May 2, 2018 · d16c6cb · d16c6cb
1 parent 7553267
commit d16c6cb
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 10 deletions.
diff --git a/lib/Raven/Serializer.php b/lib/Raven/Serializer.php
@@ -94,19 +94,23 @@ public function serialize($value, $max_depth = 3, $_depth = 0)
     protected function serializeString($value)
     {
         $value = (string) $value;
-        if (function_exists('mb_detect_encoding')
-            && function_exists('mb_convert_encoding')
-        ) {
+
+        // Check if mbstring extension is loaded
+        if (extension_loaded('mbstring')) {
             // we always guarantee this is coerced, even if we can't detect encoding
             if ($currentEncoding = mb_detect_encoding($value, $this->mb_detect_order)) {
                 $value = mb_convert_encoding($value, 'UTF-8', $currentEncoding);
             } else {
                 $value = mb_convert_encoding($value, 'UTF-8');
             }
-        }
 
-        if (strlen($value) > $this->message_limit) {
-            $value = substr($value, 0, $this->message_limit - 10) . ' {clipped}';
+            if (mb_strlen($value) > $this->message_limit) {
+                $value = mb_substr($value, 0, $this->message_limit - 10, 'UTF-8') . ' {clipped}';
+            }
+        } else {
+            if (strlen($value) > $this->message_limit) {
+                $value = substr($value, 0, $this->message_limit - 10) . ' {clipped}';
+            }
         }
 
         return $value;

diff --git a/test/Raven/Tests/SerializerTest.php b/test/Raven/Tests/SerializerTest.php
@@ -121,8 +121,8 @@ public function testLongString()
         for ($i = 0; $i < 100; $i++) {
             foreach (array(100, 1000, 1010, 1024, 1050, 1100, 10000) as $length) {
                 $input = '';
-                for ($i = 0; $i < $length; $i++) {
-                    $input .= chr(mt_rand(0, 255));
+                for ($j = 0; $j < $length; $j++) {
+                    $input .= chr(mt_rand(ord('a'), ord('z')));
                 }
                 $result = $serializer->serialize($input);
                 $this->assertInternalType('string', $result);
@@ -141,8 +141,8 @@ public function testLongStringWithOverwrittenMessageLength()
         for ($i = 0; $i < 100; $i++) {
             foreach (array(100, 490, 499, 500, 501, 1000, 10000) as $length) {
                 $input = '';
-                for ($i = 0; $i < $length; $i++) {
-                    $input .= chr(mt_rand(0, 255));
+                for ($j = 0; $j < $length; $j++) {
+                    $input .= chr(mt_rand(ord('a'), ord('z')));
                 }
                 $result = $serializer->serialize($input);
                 $this->assertInternalType('string', $result);
@@ -164,4 +164,21 @@ public function testSerializeValueResource()
         $this->assertInternalType('string', $result);
         $this->assertEquals('Resource stream', $result);
     }
+
+    public function testClippingUTF8Characters()
+    {
+        if (!extension_loaded('mbstring')) {
+            $this->markTestSkipped('mbstring extension is not enabled.');
+        }
+
+        $teststring = 'Прекратите надеяться, что ваши пользователи будут сообщать об ошибках';
+        $serializer = new Raven_Serializer(null, 19); // Length of 19 will clip character in half if no mb_* string functions are used for the teststring
+
+        $clipped = $serializer->serialize($teststring);
+        $this->assertEquals('Прекратит {clipped}', $clipped);
+
+        Raven_Compat::json_encode($clipped);
+
+        $this->assertEquals(JSON_ERROR_NONE, json_last_error());
+    }
 }