webarchive to mhtml converter

webarchive2mhtml.bat

@echo off
setlocal
set SCRIPTDIR=%~dp0

:exec
if "%~1"=="" goto end
echo "%~1"
php "%SCRIPTDIR%webarchive2mhtml.php" "%~1"
shift
goto exec

:end

webarchive2mhtml.php

<?php

require_once(dirname(__FILE__).'/cfpropertylist/CFPropertyList.php');

if(!(isset($argv[1]) && is_file($argv[1]))){
	exit;
}

$fn = realpath($argv[1]);
$fninf = pathinfo($fn);

try{
	$plist = new CFPropertyList($fn);
	$d = $plist->toArray();
}catch(DOMException $e){
	print $fninf['basename']." is not valid webarchive(plist) file.".PHP_EOL;
	exit;
}

if(!isset($d['WebMainResource'])){
	print $fninf['basename']." is not valid webarchive file.".PHP_EOL;
	exit;
}

print 'Converting "'.$fninf['basename'].'" to "'.$fninf['filename'].'.mht" ... ';

$dm = &$d['WebMainResource'];

$boundary = '----=_NextPart_'.date("YmdHis").'.'.mt_rand(10000000, 99999999);

$m = array();
$m[] = 'Date: '.date(DATE_RFC2822);
$m[] = 'MIME-Version: 1.0';
$m[] = 'Content-Type: multipart/related;';
$m[] = '	type="'.$dm['WebResourceMIMEType'].'";';
$m[] = '	boundary="'.$boundary.'"';
$m[] = '';
$m[] = 'This is a multi-part message in MIME format.';
$m[] = '';

addPart($m, $dm['WebResourceMIMEType'].(isset($dm['WebResourceTextEncodingName'])?'; charset="'.$dm['WebResourceTextEncodingName'].'"':''), $dm['WebResourceURL'], $dm['WebResourceData']);

foreach($d['WebSubresources'] as $ds){
	addPart($m, $ds['WebResourceMIMEType'], $ds['WebResourceURL'], $ds['WebResourceData']);
}

$m[] = '--'.$boundary.'--';

file_put_contents($fninf['dirname'].'/'.$fninf['filename'].'.mht', implode(PHP_EOL, $m));

print "Done.".PHP_EOL;

exit;

function addPart(&$target, $cType, $cLocation, $cData){
	$target[] = '--'.$GLOBALS['boundary'];
	if(preg_match("|^text/|", $cType)){
		$cTransEnc = 'quoted-printable';
		$cData = mb_convert_encoding($cData, 'Quoted-Printable', 'auto');
	}else{
		$cTransEnc = 'base64';
		$cData = chunk_split(base64_encode($cData));
	}
	$target[] = 'Content-Type: '.$cType;
	$target[] = 'Content-Transfer-Encoding: '.$cTransEnc;
	$target[] = 'Content-Location: '.$cLocation;
	$target[] = '';
	$target[] = $cData;
	$target[] = '';
}

?>