Use 'utf-8' (instead of utf8) as the encoding, it seems to solve any encoding issues

This commit is contained in:
Nathan Gray 2011-09-06 18:21:39 +00:00
parent 67d7615ecc
commit d71b807b8f

View File

@ -63,7 +63,8 @@ abstract class bo_merge
public static $tidy_config = array( public static $tidy_config = array(
'output-xml' => true, // Entity encoding 'output-xml' => true, // Entity encoding
'show-body-only' => true, 'show-body-only' => true,
'output-encoding' => 'utf8', 'output-encoding' => 'utf-8',
'input-encoding' => 'utf-8',
'quote-ampersand' => false, // Prevent double encoding 'quote-ampersand' => false, // Prevent double encoding
'quote-nbsp' => true, // XSLT can handle spaces easier 'quote-nbsp' => true, // XSLT can handle spaces easier
'preserve-entities' => true, 'preserve-entities' => true,
@ -740,10 +741,7 @@ abstract class bo_merge
{ {
// Clean HTML, if it's being kept // Clean HTML, if it's being kept
if($replace_tags && extension_loaded('tidy')) { if($replace_tags && extension_loaded('tidy')) {
$value = tidy_repair_string($value, self::$tidy_config + $value = tidy_repair_string($value, self::$tidy_config);
// Need to detect encoding to get special chars right
array('input-encoding'=>mb_detect_encoding($value))
);
} }
// replace </p> and <br /> with CRLF (remove <p> and CRLF) // replace </p> and <br /> with CRLF (remove <p> and CRLF)
$value = str_replace(array("\r","\n",'<p>','</p>','<br />'),array('','','',"\r\n","\r\n"),$value); $value = str_replace(array("\r","\n",'<p>','</p>','<br />'),array('','','',"\r\n","\r\n"),$value);