Description: ------------ Filenames in multibyte string should be encoded as described in RFC2231. Current head version of Mail_Mime does not support it, so I'd like to request add the feature. Reproduce code: --------------- --TEST-- Tests for RFC2231 support in Content-Disposition. --SKIPIF-- --FILE-- <?php error_reporting(E_ALL); $testEncoded="GyRCRnxLXDhsGyhCLnR4dA=="; $test = base64_decode($testEncoded); // Japanese filename in ISO-2022-JP charset. require_once('Mail/mime.php'); $Mime=new Mail_Mime(); $Mime->setTXTBody(''); $Mime->addAttachment('Japanese file',"text/plain", $test, FALSE); $body = $Mime->get(); $hdrs = ''; foreach ($Mime->headers() AS $name => $val) { $hdrs .= "$name: $val\n"; } $hdrs .= "To: Receiver <receiver@example.com>\n"; $hdrs .= "From: Sender <sender@example.com>\n"; $hdrs .= "Subject: PEAR::Mail_Mime test mail\n"; require_once('Mail/mimeDecode.php'); $mime_message = "$hdrs\n$body"; $Decoder = new Mail_mimeDecode($mime_message); $params = array( 'include_bodies' => TRUE, 'decode_bodies' => TRUE, 'decode_headers' => TRUE ); $Decoded = $Decoder->decode($params); print_r($Decoded->parts[0]->headers['content-disposition']); ?> --EXPECT-- attachment; filename*=iso-2022-jp''%1B%24BF%7CK%5C8l%1B%28B.txt

I should have make the phpt test code focused on encoding phase. the code I submitted depended on decoding code. Here's a replacement code. --TEST-- Tests for RFC2231 support in Content-Disposition. --SKIPIF-- --FILE-- <?php mb_internal_encoding('ISO-2022-JP'); error_reporting(E_ALL); $testEncoded="GyRCRnxLXDhsGyhCLnR4dA=="; $test = base64_decode($testEncoded); require_once('Mail/mime.php'); $Mime=new Mail_Mime(); $Mime->setTXTBody(''); $Mime->addAttachment('Japanese file',"text/plain", $test, FALSE); $body = $Mime->get(); $bodyarr=explode("\r\n",$body); print_r($bodyarr[3]."\r\n"); print_r($bodyarr[4]."\r\n"); ?> --EXPECT-- Content-Disposition: attachment; filename*="iso-2022-jp''%1B%24BF%7CK%5C8l%1B%28B.txt"

And I created an patch to support RFC2231 encoding feature to the current CVS head version. Index: pear/Mail_Mime/mimePart.php =================================================================== RCS file: /repository/pear/Mail_Mime/mimePart.php,v retrieving revision 1.13 diff -u -r1.13 mimePart.php --- pear/Mail_Mime/mimePart.php 10 Dec 2004 23:08:26 -0000 1.13 +++ pear/Mail_Mime/mimePart.php 18 Feb 2005 10:48:34 -0000 @@ -166,9 +166,12 @@ case 'dfilename': if (isset($headers['Content-Disposition'])) { - $headers['Content-Disposition'] .= '; filename="' . $value . '"'; + $dfilename = $this->buildDispositionParam("filename", $value); + $headers['Content-Disposition'] .= ';' . MAIL_MIMEPART_CRLF . $dfilename; + } else { - $dfilename = $value; + $dfilename = $this->buildDispositionParam("filename", $value); + } break; @@ -347,5 +350,88 @@ $output = substr($output, 0, -1 * strlen($eol)); // Don't want last crlf return $output; } + function buildDispositionParam($key, $val, $line_max = 76, $charset = NULL) + { + if(!$charset){ + if(!extension_loaded('mbstring')){ + $charset = "US-ASCII"; + }else{ + $charset = mb_internal_encoding(); + } + } + //if(extension_loaded('mbstring')){ + $internalcharset=mb_internal_encoding(); + if($internalcharset==FALSE){ + $internalcharset="US-ASCII"; + } + + if($internalcharset == "eucJP-win"){ + $internalcharset="Windows-31J"; + $val=mb_convert_encoding($val,"SJIS-win"); + } + + // Japanese mailer prefer ISO-2022-JP as described in RFC2237 + if(stristr(mb_language(),'ja')){ + $charset="ISO-2022-JP"; + } + + // Avoid 2231 encoding if possible. + if(stristr(mb_detect_encoding($val, "ASCII, ".mb_internal_encoding()),"ASCII")==false){ + $internalcharset="US-ASCII"; + }else{ + if(strcasecmp($internalcharset,$charset)){ + $val=mb_convert_encoding($val, $charset, $internalcharset); + } + } + // Fix internalcharset name for IANA registry. + $charset=mb_preferred_mime_name($charset); + //} + + // XXX The sequence below might not be safe for RFC2822, 2231 encoding may hide the problem. + # $need_escape = FALSE; + # if(mb_strpos($val, "\x09") !== FALSE) $need_escape = TRUE; + # if(mb_strpos($val, "\r") !== FALSE) $need_escape = TRUE; + # if(mb_strpos($val, "\n") !== FALSE) $need_escape = TRUE; + + $param_pair=" ".$key."=\"".$val."\""; + if($charset == "US-ASCII" && strlen($param_pair) <= $line_max) { // && $need_escape might be here. + return $param_pair; + } + + // RFC2231 encoding. + $encval = strtolower($charset) . "''" . str_replace('_', '%5F', rawurlencode($val)); + $param_pair=" ".$key."*=\"".$encval."\""; + if(strlen($param_pair) <= $line_max) { // we don't need folding + return $param_pair; + } + + + // folding + // 4 bytes for "*0*=", 1 byte for heading ' ', 1 byte for ';', => total 6 bytes + $rhs_space = $line_max - strlen($key) - 6; + + // folding + $dest = ""; + for($i = 0 ; strlen($encval) > $rhs_space ; $i++) { + $dest .= " " . $key . "*" . $i . "*="; + $pos = strrpos(substr($encval, 0, $rhs_space), '%'); + if( $rhs_space - 2 <= $pos) { // '%' ‚ªŒã‚ë‚©‚ç2byteˆÈ“à‚É‚ ‚é + $dest .= substr($encval, 0, $pos); + $encval = substr($encval, $pos); + } else { + $dest .= substr($encval, 0, $rhs_space); + $encval = substr($encval, $rhs_space); + } + $dest .= ";\r\n"; + // bytes allowed in right hand space + // 3 bytes for "**=", 1 byte for heading ' ', 1 byte for ';' => total 5 bytes + $rhs_space = $line_max - strlen(strval($i)) - strlen($key) - 5; + } + $dest .= " " . $key . "*" . $i . "*="; + $dest .= $encval; + + return $dest; + } + } // End of class ?>

Oops, sorry. the patch should be modified. Here is a diff of diff. mbstring extension treatment will be enabled. --- mimePart.patch 2005-02-18 22:04:28.123182400 +0900 +++ mimePart.patch.2 2005-02-18 23:06:29.253905600 +0900 @@ -4,7 +4,7 @@ retrieving revision 1.13 diff -u -r1.13 mimePart.php --- pear/Mail_Mime/mimePart.php 10 Dec 2004 23:08:26 -0000 1.13 -+++ pear/Mail_Mime/mimePart.php 18 Feb 2005 10:48:34 -0000 ++++ pear/Mail_Mime/mimePart.php 18 Feb 2005 14:06:29 -0000 @@ -166,9 +166,12 @@ case 'dfilename': @@ -33,7 +33,7 @@ + $charset = mb_internal_encoding(); + } + } -+ //if(extension_loaded('mbstring')){ ++ if(extension_loaded('mbstring')){ + $internalcharset=mb_internal_encoding(); + if($internalcharset==FALSE){ + $internalcharset="US-ASCII"; @@ -59,7 +59,7 @@ + } + // Fix internalcharset name for IANA registry. + $charset=mb_preferred_mime_name($charset); -+ //} ++ } + + // XXX The sequence below might not be safe for RFC2822, 2231 encoding may hide the problem. + # $need_escape = FALSE;