return remaining chars in last part of UTF16 messages

instasent · Jun 5, 2019 · c5d7489 · c5d7489
1 parent c475537
commit c5d7489
Show file tree

Hide file tree

Showing 3 changed files with 71 additions and 5 deletions.
diff --git a/README.md b/README.md
@@ -27,6 +27,12 @@ stdClass Object
 )
 ```
 
+##### UTF16 notice
+
+When using unicode chars over U+10000 (mainly emoticons 😎) on messages larger than 70 chars the _remaining_ value will actually be the **remaining chars in last message part only**, this is due to how those chars are encoded using two 16bit chars and max part length being an odd number (67)
+
+#### Sanitization
+
 You can sanitize your text to be a valid strict GSM 03.38 charset
 
 ```php

diff --git a/SMSCounter.php b/SMSCounter.php
@@ -259,7 +259,23 @@ function ($carry, $char) {
         }
 
         $messages = (int) ceil($length / $perMessage);
-        $remaining = ($perMessage * $messages) - $length;
+
+        if ($encoding === self::UTF16 && $length > $perMessage) {
+            $count = 0;
+            foreach ($unicodeArray as $char) {
+                if ($count === $perMessage) {
+                    $count = 0;
+                } elseif ($count > $perMessage) {
+                    $count = 2;
+                }
+
+                $count += $char >= 65536 ? 2 : 1;
+            }
+
+            $remaining = $perMessage - ($count > $perMessage ? 2 : $count);
+        } else {
+            $remaining = ($perMessage * $messages) - $length;
+        }
 
         $returnset = new \stdClass();
 

diff --git a/Tests/SMSCounterTest.php b/Tests/SMSCounterTest.php
@@ -217,18 +217,62 @@ public function testUnicode()
         $this->assertEquals($expected, $count);
     }
 
-    public function testUnicodeEmoji()
+    public function testUnicodeEmojiSingleMessage()
     {
-        $text = '😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎';
+        $text = '😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎';
+        $smsCounter = new SMSCounter();
+        $count = $smsCounter->count($text);
+
+        $expected = new \stdClass();
+        $expected->encoding = SMSCounter::UTF16;
+        $expected->length = 70;
+        $expected->per_message = 70;
+        $expected->remaining = 0;
+        $expected->messages = 1;
+
+        $this->assertEquals($expected, $count);
+    }
+
+    public function testUnicodeEmojiMultiPartMessage()
+    {
+        // A char is lost at the end of first part
+        $text = '😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎';
+        $smsCounter = new SMSCounter();
+        $count = $smsCounter->count($text);
+
+        $expected = new \stdClass();
+        $expected->encoding = SMSCounter::UTF16;
+        $expected->length = 72;
+        $expected->per_message = 67;
+        $expected->remaining = 61;
+        $expected->messages = 2;
+
+        $this->assertEquals($expected, $count);
+
+        // First part is completed with a dash char (-)
+        $text = '😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎-😎😎😎';
+        $smsCounter = new SMSCounter();
+        $count = $smsCounter->count($text);
+
+        $expected = new \stdClass();
+        $expected->encoding = SMSCounter::UTF16;
+        $expected->length = 73;
+        $expected->per_message = 67;
+        $expected->remaining = 61;
+        $expected->messages = 2;
+
+        $this->assertEquals($expected, $count);
 
+        // Both parts are completed with dash chars (-)
+        $text = '😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎-😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎😎-';
         $smsCounter = new SMSCounter();
         $count = $smsCounter->count($text);
 
         $expected = new \stdClass();
         $expected->encoding = SMSCounter::UTF16;
-        $expected->length = 132;
+        $expected->length = 134;
         $expected->per_message = 67;
-        $expected->remaining = 2;
+        $expected->remaining = 0;
         $expected->messages = 2;
 
         $this->assertEquals($expected, $count);