encoding.patch (7,276 bytes)
--- class.phpmailer.php.orig 2005-06-29 04:23:45.000000000 +0200
+++ class.phpmailer.php 2007-11-13 19:14:18.000000000 +0100
@@ -656,6 +656,9 @@
*/
function WrapText($message, $length, $qp_mode = false) {
$soft_break = ($qp_mode) ? sprintf(" =%s", $this->LE) : $this->LE;
+ // If utf-8 encoding is used, we will need to make sure we don't
+ // split multibyte characters when we wrap
+ $is_utf8 = (strtolower($this->CharSet) == "utf-8");
$message = $this->FixEOL($message);
if (substr($message, -1) == $this->LE)
@@ -678,9 +681,11 @@
if ($space_left > 20)
{
$len = $space_left;
- if (substr($word, $len - 1, 1) == "=")
+ if ($is_utf8) {
+ $len = $this->UTF8CharBoundary($word, $len);
+ } elseif (substr($word, $len - 1, 1) == "=") {
$len--;
- elseif (substr($word, $len - 2, 1) == "=")
+ } elseif (substr($word, $len - 2, 1) == "=")
$len -= 2;
$part = substr($word, 0, $len);
$word = substr($word, $len);
@@ -696,9 +701,11 @@
while (strlen($word) > 0)
{
$len = $length;
- if (substr($word, $len - 1, 1) == "=")
+ if ($is_utf8) {
+ $len = $this->UTF8CharBoundary($word, $len);
+ } elseif (substr($word, $len - 1, 1) == "=") {
$len--;
- elseif (substr($word, $len - 2, 1) == "=")
+ } elseif (substr($word, $len - 2, 1) == "=")
$len -= 2;
$part = substr($word, 0, $len);
$word = substr($word, $len);
@@ -727,6 +734,65 @@
return $message;
}
+ /**
+ * Finds last character boundary prior to maxLength in a utf-8
+ * quoted (printable) encoded string.
+ * Original written by Colin Brown.
+ *
+ * @access private
+ *
+ * @param string $encodedText utf-8 QP text
+ * @param int $maxLength find last character boundary prior to this length
+ *
+ * @return int
+ */
+ function UTF8CharBoundary($encodedText, $maxLength)
+ {
+ $foundSplitPos = false;
+ $lookBack = 3;
+
+ while (!$foundSplitPos)
+ {
+ $lastChunk = substr($encodedText, $maxLength - $lookBack, $lookBack);
+ $encodedCharPos = strpos($lastChunk, "=");
+
+ if ($encodedCharPos !== false) {
+ // Found start of encoded character byte within $lookBack block.
+ // Check the encoded byte value (the 2 chars after the '=')
+ $hex = substr($encodedText, $maxLength - $lookBack + $encodedCharPos + 1, 2);
+ $dec = hexdec($hex);
+ if ($dec < 128)
+ {
+ // Single byte character.
+
+ // If the encoded char was found at pos 0, it will fit
+ // otherwise reduce maxLength to start of the encoded char
+ $maxLength = ($encodedCharPos == 0) ? $maxLength :
+ $maxLength - ($lookBack - $encodedCharPos);
+ $foundSplitPos = true;
+ }
+ elseif ($dec >= 192)
+ {
+ // First byte of a multi byte character
+
+ // Reduce maxLength to split at start of character
+ $maxLength = $maxLength - ($lookBack - $encodedCharPos);
+ $foundSplitPos = true;
+ }
+ elseif ($dec < 192)
+ {
+ // Middle byte of a multi byte character, look further back
+ $lookBack += 3;
+ }
+ } else {
+ // No encoded character found
+ $foundSplitPos = true;
+ }
+ }
+
+ return $maxLength;
+ }
+
/**
* Set the body wrapping.
* @access private
@@ -1166,9 +1232,15 @@
// Try to select the encoding which should produce the shortest output
if (strlen($str)/3 < $x) {
$encoding = 'B';
- $encoded = base64_encode($str);
- $maxlen -= $maxlen % 4;
- $encoded = trim(chunk_split($encoded, $maxlen, "\n"));
+ if (function_exists('mb_strlen') && $this->HasMultiBytes($str)) {
+ // Use a custom function which correctly encodes and wraps long
+ // multibyte strings without breaking lines within a character
+ $encoded = $this->Base64EncodeWrapMB($str);
+ } else {
+ $encoded = base64_encode($str);
+ $maxlen -= $maxlen % 4;
+ $encoded = trim(chunk_split($encoded, $maxlen, "\n"));
+ }
} else {
$encoding = 'Q';
$encoded = $this->EncodeQ($str, $position);
@@ -1182,6 +1254,74 @@
return $encoded;
}
+
+ /**
+ * Checks if a string contains multibyte characters.
+ *
+ * @access private
+ *
+ * @param string $str multi-byte text to wrap encode
+ *
+ * @return bool
+ */
+ function HasMultiBytes($str)
+ {
+ if (function_exists('mb_strlen')) {
+ return (strlen($str) > mb_strlen($str, $this->CharSet));
+ } else {
+ // Assume no multibytes (we can't handle without mbstring functions anyway)
+ return False;
+ }
+ }
+
+ /**
+ * Correctly encodes and wraps long multibyte strings for mail headers
+ * without breaking lines within a character.
+ *
+ * Adapted from a function by paravoid at http://uk.php.net/manual/en/function.mb-encode-mimeheader.php
+ *
+ * @access private
+ *
+ * @param string $str multi-byte text to wrap encode
+ *
+ * @return string
+ */
+ function Base64EncodeWrapMB($str)
+ {
+ $start = "=?".$this->CharSet."?B?";
+ $end = "?=";
+ $encoded = "";
+
+ $mb_length = mb_strlen($str, $this->CharSet);
+ // Each line must have length <= 75, including $start and $end
+ $length = 75 - strlen($start) - strlen($end);
+ // Average multi-byte ratio
+ $ratio = $mb_length / strlen($str);
+ // Base64 has a 4:3 ratio
+ $offset = $avgLength = floor($length * $ratio * .75);
+
+ for ($i = 0; $i < $mb_length; $i += $offset)
+ {
+ $lookBack = 0;
+
+ do {
+ $offset = $avgLength - $lookBack;
+ $chunk = mb_substr($str, $i, $offset, $this->CharSet);
+ $chunk = base64_encode($chunk);
+ $lookBack++;
+ }
+ while (strlen($chunk) > $length);
+
+ $encoded .= $chunk . $this->LE;
+ }
+
+ // Chomp the last linefeed
+ $encoded = substr($encoded, 0, -strlen($this->LE));
+
+ return $encoded;
+ }
+
+
/**
* Encode string to quoted-printable.
* @access private
