From 0537365cac67f8b9d8b9f8d10459de1488569372 Mon Sep 17 00:00:00 2001
From: rxu <rxu@mail.ru>
Date: Tue, 23 Nov 2021 15:12:43 +0700
Subject: [PATCH] [ticket/16912] Rewrite email headers encoding method

Based on symfony/polyfill-iconv implementations.

PHPBB3-16912
---
 phpBB/includes/functions_messenger.php | 98 +++++++++++++++-----------
 1 file changed, 58 insertions(+), 40 deletions(-)

diff --git a/phpBB/includes/functions_messenger.php b/phpBB/includes/functions_messenger.php
index b7f3eeda36..52c205ceb5 100644
--- a/phpBB/includes/functions_messenger.php
+++ b/phpBB/includes/functions_messenger.php
@@ -1840,66 +1840,84 @@ class smtp_class
 }
 
 /**
-* Encodes the given string for proper display in UTF-8.
-*
-* This version is using base64 encoded data. The downside of this
-* is if the mail client does not understand this encoding the user
-* is basically doomed with an unreadable subject.
-*
-* Please note that this version fully supports RFC 2045 section 6.8.
-*
-* @param string $str
-* @param string $eol End of line we are using (optional to be backwards compatible)
-*/
+ * Encodes the given string for proper display in UTF-8 or US-ASCII.
+ *
+ * This version is based on iconv_mime_encode() implementation
+ * from symfomy/polyfill-iconv
+ * https://github.com/symfony/polyfill-iconv/blob/fd324208ec59a39ebe776e6e9ec5540ad4f40aaa/Iconv.php#L355
+ *
+ * @param string $str
+ * @param string $eol End of line we are using (optional to be backwards compatible)
+ *
+ * @return string
+ */
 function mail_encode($str, $eol = "\r\n")
 {
 	// Check if string contains ASCII only characters
 	$is_ascii = strlen($str) === utf8_strlen($str);
 
-	// Define start delimimter, end delimiter and spacer
+	$scheme = $is_ascii ? "Q" : "B";
+
+	// Define start delimimter, end delimiter
 	// Use the Quoted-Printable encoding for ASCII strings to avoid unnecessary encoding in Base64
-	$start = $is_ascii ? "=?US-ASCII?Q?" : "=?UTF-8?B?";
+	$start = $is_ascii ? "=?US-ASCII?$scheme?" : "=?UTF-8?$scheme?";
 	$end = "?=";
-	$delimiter = "$eol ";
 
 	// Maximum encoded-word length is 75 as per RFC 2047 section 2.
 	// $split_length *must* be a multiple of 4, but <= 75 - strlen($start . $delimiter . $end)!!!
-	$split_length = 75 - strlen($start . $delimiter . $end);
+	$split_length = 75 - strlen($start . $eol . $end);
 	$split_length = $split_length - $split_length % 4;
 
-	// Use the Quoted-Printable encoding for ASCII strings to avoid unnecessary encoding in Base64
-	// quoted_printable_encode() splits lines at length of 75 characters with =\r\n delimiter, amend this feature
-	$encoded_str = $is_ascii ? str_replace("=\r\n", '', quoted_printable_encode($str)) : base64_encode($str);
+	$line_length = \strlen($start) + \strlen($end);
+	$line_offset = \strlen($start) + 1;
+	$line_data = '';
 
-	// If encoded string meets the limits, we just return with the correct data.
-	if (strlen($encoded_str) <= $split_length)
+	$is_quoted_printable = 'Q' === $scheme;
+
+	preg_match_all('/./us', $str, $chars);
+	$chars = $chars[0] ?? [];
+
+	$str = [];
+	foreach ($chars as $char)
 	{
-		return $start . $encoded_str . $end;
-	}
+		$encoded_char = $is_quoted_printable
+			? $char = preg_replace_callback(
+				'/[=_\?\x20\x00-\x1F\x80-\xFF]/',
+				function ($matches)
+				{
+					$hex = dechex(\ord($matches[0]));
+					$hex = strlen($hex) == 1 ? "0$hex" : $hex;
+					return '=' . strtoupper($hex);
+				},
+				$char
+			)
+			: base64_encode($line_data . $char);
 
-	// If there is only ASCII data, we just return what we want, correctly splitting the lines.
-	if ($is_ascii)
-	{
-		return $start . implode($end . $delimiter . $start, str_split($encoded_str, $split_length)) . $end;
-	}
-
-	// UTF-8 data, compose encoded lines
-	$array = utf8_str_split($str);
-	$str = '';
-
-	while (count($array))
-	{
-		$text = '';
-
-		while (count($array) && intval((strlen($text . $array[0]) + 2) / 3) << 2 <= $split_length)
+		if (isset($encoded_char[$split_length - $line_length]))
 		{
-			$text .= array_shift($array);
+			if (!$is_quoted_printable)
+			{
+				$line_data = base64_encode($line_data);
+			}
+			$str[] = $start . $line_data . $end;
+			$line_length = $line_offset;
+			$line_data = '';
 		}
 
-		$str .= $start . base64_encode($text) . $end . $delimiter;
+		$line_data .= $char;
+		$is_quoted_printable && $line_length += \strlen($char);
 	}
 
-	return substr($str, 0, -strlen($delimiter));
+	if ('' !== $line_data)
+	{
+		if (!$is_quoted_printable)
+		{
+			$line_data = base64_encode($line_data);
+		}
+		$str[] = $start . $line_data . $end;
+	}
+
+	return implode($eol . ' ', $str);
 }
 
 /**