From 82d18754bc58edbdb0ed4835bc0a34e2eb8f4eb6 Mon Sep 17 00:00:00 2001
From: Sergey Linnik <linniksa@gmail.com>
Date: Mon, 30 Jan 2017 19:59:18 +0300
Subject: [PATCH] autofix utf8 strings

---
 src/Fetch/Message.php         |  4 +++
 src/Fetch/UTF8.php            | 54 +++++++++++++++++++++++++++++++++++
 tests/Fetch/Test/MIMETest.php |  4 +++
 3 files changed, 62 insertions(+)
 create mode 100644 src/Fetch/UTF8.php

diff --git a/src/Fetch/Message.php b/src/Fetch/Message.php
index dccb0dc..dc76535 100755
--- a/src/Fetch/Message.php
+++ b/src/Fetch/Message.php
@@ -577,6 +577,10 @@ class Message
         $to   = strtolower($to);
 
         if ($from === $to) {
+            if ($to === 'utf-8') {
+                return UTF8::fix($text);
+            }
+
             return $text;
         }
 
diff --git a/src/Fetch/UTF8.php b/src/Fetch/UTF8.php
new file mode 100644
index 0000000..87b0869
--- /dev/null
+++ b/src/Fetch/UTF8.php
@@ -0,0 +1,54 @@
+<?php
+
+namespace Fetch;
+
+final class UTF8
+{
+    public static function fix($text)
+    {
+        if(!is_string($text)) {
+            return $text;
+        }
+
+        $buf = '';
+        for ($i = 0, $max = strlen($text); $i < $max; $i++) {
+            $c1 = $text{$i};
+            if ($c1 >= "\xc0") { //Should be converted to UTF8, if it's not UTF8 already
+                $c2 = $i + 1 >= $max ? "\x00" : $text[$i + 1];
+                $c3 = $i + 2 >= $max ? "\x00" : $text[$i + 2];
+                $c4 = $i + 3 >= $max ? "\x00" : $text[$i + 3];
+
+                if ($c1 >= "\xc0" & $c1 <= "\xdf") { //looks like 2 bytes UTF8
+                    if ($c2 >= "\x80" && $c2 <= "\xbf") { //yeah, almost sure it's UTF8 already
+                        $buf .= $c1 . $c2;
+                        $i++;
+                    } else { //not valid UTF8.  Convert it.
+                        $buf .= '?';
+                    }
+                } elseif ($c1 >= "\xe0" & $c1 <= "\xef") { //looks like 3 bytes UTF8
+                    if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { //yeah, almost sure it's UTF8 already
+                        $buf .= $c1 . $c2 . $c3;
+                        $i   += 2;
+                    } else { //not valid UTF8.  Convert it.
+                        $buf .= '?';
+                    }
+                } elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { //looks like 4 bytes UTF8
+                    if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { //yeah, almost sure it's UTF8 already
+                        $buf .= $c1 . $c2 . $c3 . $c4;
+                        $i   += 3;
+                    } else { //not valid UTF8.  Convert it.
+                        $buf .= '?';
+                    }
+                } else { //doesn't look like UTF8, but should be converted
+                    $buf .= '?';
+                }
+            } elseif (($c1 & "\xc0") === "\x80") { // needs conversion
+                $buf .= '?';
+            } else { // it doesn't need conversion
+                $buf .= $c1;
+            }
+        }
+
+        return $buf;
+    }
+}
diff --git a/tests/Fetch/Test/MIMETest.php b/tests/Fetch/Test/MIMETest.php
index f49a517..3fc8c26 100644
--- a/tests/Fetch/Test/MIMETest.php
+++ b/tests/Fetch/Test/MIMETest.php
@@ -34,6 +34,10 @@ class MIMETest extends \PHPUnit_Framework_TestCase
                 . PHP_EOL .
                 '=?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?='
             ),
+            array(
+                '?????? ?????.pdf',
+                '=?UTF-8?B?' .base64_encode("\xCF\xF0\xE8\xEC\xE5\xF0 \xEF\xEB\xE0\xED\xE0\x2E\x70\x64\x66") . '?=',
+            ),
         );
     }