#!/usr/local/bin/php -q

<?

# Mirphealy - a script to archive emails from mailing lists.
# It is based on hypermail.pl (http://web.dar.net/~northrup/hypermail.html), 
# a perl script written by Dylan Northrup, which itself was derived from the 
# original C-based hypermail (http://www.hypermail.org).  Mirphealy adds
# improved regexes and true messaging threading to hypermail.pl.

function show_usage() {
print <<<EOF
usage: mirphearly.php -d directory -f file
options: -d : Target directory for saving HTML files
         -f : Read in mail archive from file

EOF;
}

#	Ugly hack to get commandline options.  There must be a better way.

	$fp = fopen("php://stdin","r");
	fclose($fp);
	for ($c = 1; $c < sizeof($argv); $c++) {
		if ($c % 2) {
			$key = $argv[$c];
			$val = $argv[$c+1];
			$cl_opts[$key] = $val;
		}
			
	}

	$dir = $cl_opts['-d'];
	$file = $cl_opts['-f'];
	if(!$dir) {
		show_usage();
		exit;
	}
	if(!$file ) {
		show_usage();
		exit;
	}


#	Functions to use for sorting messages.

#	sort_by_date() sorts the $index[date] array, first by date then time.
#	What I have to do is parse $a and $b (the initial and next element of the
#	array) into $a_something and $b_something, compare the respective values
#	and return which comes first chronologically.  Should be interesting :-)

	function sort_by_date($a, $b) {
	
	#	Have to remove double spaces in date string because they mess with explode().

		$aa = $a[date];
		$aa = eregi_replace("[[:space:]]+", " ",$aa);
		$bb = $b[date];
		$bb = eregi_replace("[[:space:]]+", " ",$bb);
		
	#	Parse the $index[date] string and grab some vars.

		$array_date = explode(" ",$aa);
		$aday = $array_date[0]; 
		$amonth = $array_date[1]; 

		$adate = $array_date[2];
		$atime = $array_date[3];
		$ayear = $array_date[4];

	#	Grab hour/min/sec for extra "precision".	

		$array_time = explode(":",$atime);
		$ahour = $array_time[0];
		$amin = $array_time[1];
		$asec = $array_time[2];

	#	Repeat for $b.

		$brray_date = explode(" ",$bb);
		$bday = $brray_date[0]; 
		$bmonth = $brray_date[1]; 

		$bdate = $brray_date[2];
		$btime = $brray_date[3];
		$byear = $brray_date[4];

		$brray_time = explode(":",$btime);
		$bhour = $brray_time[0];
		$bmin = $brray_time[1];
		$bsec = $brray_time[2];

	if (($adate == $bdate) && ($ahour == $bhour) && ($amin == $bmin)) {
					return ($asec > $bsec) ? 1 : -1;
		} elseif (($adate == $bdate) && ($ahour == $bhour)) {
			return ($amin > $bmin) ? 1 : -1;
		} elseif (($adate == $bdate)) {
			return ($ahour > $bhour) ? 1 : -1;
		}
		return ($adate > $bdate) ? 1 : -1;
	}	

	function sort_by_thread($a, $b) {
		$asub = strtolower($a[subject]);
		$bsub = strtolower($b[subject]);
		$asub = eregi_replace("re: ","",$asub);
		$bsub = eregi_replace("re: ","",$bsub);

		if (strnatcmp($asub,$bsub) == 0) {
			return 0;
		}
		if (strnatcmp($asub,$bsub) > 0) {
			return 1;
		}
		if (strnatcmp($asub,$bsub) < 0) {
			return -1;
		}

	}

	function sort_by_author($a, $b) {
		$afrom = strtolower($a[from]);
		$bfrom = strtolower($b[from]);
		if (strnatcmp($afrom,$bfrom) == 0 ){
			return 0;
		} elseif (strnatcmp($afrom,$bfrom) > 0 ){
			return 1;
		} elseif (strnatcmp($afrom,$bfrom) < 0 ){
			return -1;
		}

	}

	$archive = $mailbox ? $mailbox : $file;

#	Grab the date from $archive and parse it to get year and month.
/*
	$arch_date = substr(strchr($archive,"/"),1);
	$arch_year = substr($arch_date, 0,4);
	$arch_month = substr($arch_date, 4,6);
*/

	$fp=fopen($archive,'r');
	$contents=fread($fp,filesize($archive));

	$messages=split("\nFrom ", $contents);
	fclose($fp);
	$i=0;
	foreach($messages as $value) {

			$index[$i][filename] = "$i.html";

#	Grab received date from header.

			if($i == 0) {
				if(preg_match("/^\S+\s+\S+\s+(.+)/",$value,$match)) {
					$index[$i][date] = $match[1];
					$start_date = $index[$i][date];
				}
			} else {
				if(preg_match("/^[^\s]+\s+(.+)/",$value,$match)) {
					$index[$i][date] = $match[1];
					$end_date = $index[$i][date];
				}
			}
			
#	Grab subject from header.

			if(preg_match("/^Subject: (.*)$/m",$value,$match)){
				$index[$i][subject] = $match[1];
			} else {
					$index[$i][subject] = "NO SUBJECT";
			}

#	Grab "From: " from header.

			if(preg_match("/^From: (.+)\S?(<(.+@.+\.?)>)/m",$value,$match)){
				$index[$i][from]= eregi_replace("\"","",$match[1]);
			}
		 		elseif(preg_match("/^From: <?((.+)@.+)>?/m",$value,$match)){
					$index[$i][from] = eregi_replace("\"","",$match[2]);
				}	elseif(preg_match("/^From: <?(.+@(.+\>)+(.+))>?/m",$value,$match)){
						$index[$i][from] = "NONE";
				} elseif(preg_match("/^From: (.*)$/m",$value,$match)){
						$index[$i][from] = eregi_replace("\"","",$match[1]);
				}

# Not doing refs or sent_date for now. 2001-04-28
/*
#	Grap any references from header.

	if(preg_match("/^References: ((<\S*>) .*)/m",$value,$match)){
		$index[$i][ref_line]=$match[0];
		$index[$i][refs]=$match[1];
		$index[$i][first_ref]=$match[2];
		
		} elseif(preg_match("/^References: (<.*>)/m",$value,$match)){
			$index[$i][ref_line]=$match[0];
			$index[$i][first_ref]=$match[1];
			$index[$i][refs]=$match[1];
		  }

#	Grab "Date: " from header which is different from $index[$i]['date'].

	if(preg_match("/^Date: (.*)$/m",$value,$match)){
		$index[$i][sent_date]=$match[0];
	}
*/

	$body=split("\n",$value);
	$blank= array_shift($body);
	while(!preg_match("/^$/",$blank)){
		$blank = array_shift($body);
	}
	$index[$i][body] = join("\n",$body);
	$i++;
}
		$size = sizeof($index);

#	Here we write each message out into a html page.

	usort($index,"sort_by_thread");


	if(!is_dir($dir)) {
		if(!(mkdir($dir,0601))) {
			echo "Can't open $dir!";
			exit;
		}
	echo "Making $dir directory.\n";
	}
	echo "Writing messages.\n";
	for ($i = 0; $i <= sizeof($index); $i++) {
		$morf = eregi_replace("re: ","",$index[$i][subject]);
		$next_morf = eregi_replace("re: ","",$index[$i+1][subject]);
		if ($next_morf == $morf) {
			$new_index[] = $index[$i];
		} else {
			$new_index[] = $index[$i];
			usort($new_index,"sort_by_date");
			for ($k = 0; $k < sizeof($new_index); $k++) {
				$from = $new_index[$k][from];
				$link = $new_index[$k][filename];
				$when = $new_index[$k][date];
				$sub = $new_index[$k][subject];
				$body = $new_index[$k][body];
				if (preg_match("/(.*).html/m",$link,$my_match)) $anchor = $my_match[1];

				if(!($fp = fopen("$dir/$link","w"))) {
					echo "Can't open $link!";
					exit;
				} else {
					fwrite($fp,"<html>\n<head>\n<title>$sub</title>\n</head>\n<body>\n");
					fwrite($fp,"<h3>$sub</h3>");
					fwrite($fp,"<b>From:</b> $from<br><b>Date</b>: <i>$when</i><p>");
					if ($k + 1 < sizeof($new_index)) {
						$flag = 1;
						$next = $new_index[$k+1][from];
						$date1 = $new_index[$k+1][date];
						$link1 = $new_index[$k+1][filename];
						fwrite($fp,"<b>Next in Thread :</b> ");
						fwrite($fp,"<a href=\"$link1\">$next</a>: <i>$date1</i><br>\n");
					} else {
						$flag = -1;
						fwrite($fp,"<b>Next in Thread :</b> <i>End of Thread</i><br>\n");
					} 
					if ($k > 0 && $k < sizeof($new_index)) {
						$glaf = 1;
						$prev = $new_index[$k-1][from];
						$date2 = $new_index[$k-1][date];
						$link2 = $new_index[$k-1][filename];
						fwrite($fp,"<b>Previous in Thread :</b> ");
						fwrite($fp,"<a href=\"$link2\">$prev</a>: <i>$date2</i>\n<br>\n");
					} else {
						$glaf = -1;
						fwrite($fp,"<b>Previous in Thread :</b> <i>Beginning of Thread</i><br>\n");	
					}
					fwrite($fp,"<b>Messages sorted by:</b> <a href=\"author.html#$anchor\">[ author ]</a>");
					fwrite($fp," <a href=\"date.html#$anchor\">[ date ]</a> <a href=\"thread.html#$anchor\">[ thread ]</a><br>\n");
					fwrite($fp,"<hr><pre>$body</pre><hr>\n");
					if ($flag > 0) {
						fwrite($fp,"<b>Next in Thread :</b> ");
						fwrite($fp,"<a href=\"$link1\">$next</a>: <i>$date1</i><br>\n");
					} elseif ($flag < 0) {
						fwrite($fp,"<b>Next in Thread :</b> <i>End of Thread</i><br>\n");
					  }
					if ($glaf > 0) {
						fwrite($fp,"<b>Previous in Thread :</b> ");
						fwrite($fp,"<a href=\"$link2\">$prev</a>: <i>$date2</i>\n<br>\n");
					} elseif ($glaf < 0) {
						fwrite($fp,"<b>Previous in Thread :</b> <i>Beginning of Thread</i><br>\n");	
					  }
					fwrite($fp,"<b>Messages sorted by:</b> <a href=\"author.html#$anchor\">[ author ]</a>");
					fwrite($fp," <a href=\"date.html#$anchor\">[ date ]</a> <a href=\"thread.html#$anchor\">[ thread ]</a><br>\n");
					fwrite($fp,"<p><font size=\"2\">This archive generated by <i><a href=\"http://www.cerritoslug.org/downloads/mirphealy.php\">Mirphealy 0.8</a>.</i></font>\n");
					fwrite($fp,"</body>\n</html>");
					fclose($fp);
				}	
			}	
			unset($new_index);
		  }
	}

#	Here we write out the date.html index page.


 	usort($index,sort_by_date);
	if(!($fp = fopen("$dir/date.html","w"))){
		echo "Can't open date.html.";
		exit;
	}
	echo "Creating date.html\n";
	fwrite($fp,"<html>\n<head><title>Archives by Date</title></head>\n");
    fwrite($fp,"<body>\n"); 
	fwrite($fp,"<a name=\"start\"><h1>Archive by Date</h1></a>\n<ul>\n");
	fwrite($fp," <li> <b><a href=\"#end\">Most recent messages</a></b>\n");
	fwrite($fp," <li> <b>Messages sorted by:</b> <a href=\"thread.html#start\">[ thread ]</a>");
	fwrite($fp,"<a href=\"author.html#start\">[ author ]</a></b>\n");
	fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
	fwrite($fp,"<b>Starting:</b> <i>$start_date</i><br>\n");
	fwrite($fp,"<b>Ending:</b> <i>$end_date</i><br>\n");
	fwrite($fp,"<b>Messages: $size</b>\n<p>\n<ul>\n");

	foreach ($index as $da) {
		if (!(isset($da[subject]))) next($da); 
	$link = $da[filename];
	$from = $da[from];
	$sub = $da[subject];
	fwrite($fp," <li><a href=\"$link\"><b>$sub</b></a> ");
	if (preg_match("/(\d+).html/m",$link,$my_match)) $anchor = $my_match[1];
	fwrite($fp,"<a name=\"$anchor\"><i>$from</i></a>\n");
	}
	fwrite($fp,"</ul>\n");
	fwrite($fp,"<a name=\"end\"><b>Last message date:</b></a> <i>$end_date</i><br>\n");
	fwrite($fp,"<ul>\n <li> <b>Messages sorted by:</b> ");
	fwrite($fp,"<a href=\"thread.html#start\">[ thread ]</a> <a href=\"author.html#start\">[ author ]</a>\n");
	fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
	fwrite($fp,"<p><font size=\"2\">This archive generated by <i><a href=\"http://www.cerritoslug.org/downloads/mirphealy.php\">Mirphealy 0.8</a>.</i></font>\n");
	fwrite($fp,"</body>\n</html>");
	fclose($fp);



#	This is for thread.html.

 		usort($index,sort_by_thread);
		if (!($fp = fopen("$dir/thread.html","w"))){
			echo "Can't open thread.html!";
			exit;
		}
		echo "Creating thread.html\n";
		fwrite($fp,"<html>\n<head><title>Archive By Subject</title></head>\n<body>\n");
		fwrite($fp,"<a name=\"start\"><h1>Archive By Subject</h1></a>\n<ul>\n");
		fwrite($fp," <li> <b>Messages sorted by:</b> ");
		fwrite($fp,"<a href=\"date.html#start\">[ date ]</a> <a href=\"author.html#start\">[ author ]</a></b>\n");
		fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
		fwrite($fp,"<b>Starting:</b> <i>$start_date</i><br>\n");
		fwrite($fp,"<b>Ending:</b> <i>$end_date</i><br>\n");
		fwrite($fp,"<b>No. of Messages: $size</b>\n<p>\n<ul>\n");
		
		foreach ($index as $ta) {
			$sub  = eregi_replace("re: ","",$ta['subject']);
			$from = $ta[from];
			$link = $ta[filename];
			$when = $ta[date];
			if(preg_match("/(.*).html/m",$link,$file_match)) $num = $file_match[1];
			if($cur_sub != $sub) {
				$what = $ta[subject];
				if(isset($cur_sub)) fwrite($fp,"  </ul>\n");
				fwrite($fp," <li><b>$what</b>\n  <ul>\n");
				$cur_sub = $sub;
			}
			fwrite($fp,"   <li><a href=\"$link\"><b>$from</b></a>");
			fwrite($fp," <a name=\"$num\"><i>$when</i></a>\n");
		}	
		fwrite($fp,"  </ul>\n");
		fwrite($fp,"</ul>\n");
		fwrite($fp,"<a name=\"end\"><b>Last message date:</b></a> <i>$end_date</i><br>\n");
		fwrite($fp,"<ul>\n <li> <b>Messages sorted by:</b> ");
		fwrite($fp,"<a href=\"date.html#start\">[ date ]</a> <a href=\"author.html#start\">[ author ]</a>\n");
		fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
		fwrite($fp,"<p><font size=\"2\">This archive generated by <i><a href=\"http://www.cerritoslug.org/downloads/mirphealy.php\">Mirphealy 0.8</a>.</i></font>\n");
		fwrite($fp,"</body>\n</html>");
		fclose($fp);


#	Let's do it for authors!

	usort($index,"sort_by_author");

	if (!($fp = fopen("$dir/author.html","w"))){
		echo "Can't open author.html!";
		exit;
	}
	echo "Creating author.html\n";
	fwrite($fp,"<html><head><title>Archive by Author</title></head><body>\n");
	fwrite($fp,"<a name=\"start\"><h1>Archive By Author</h1></a>\n<ul>\n");
	fwrite($fp," <li> <b>Messages sorted by:</b> ");
	fwrite($fp,"<a href=\"date.html#start\">[ date ]</a> <a href=\"thread.html#start\">[ thread ]</a></b>\n");
	fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
	fwrite($fp,"<b>Starting:</b> <i>$start_date</i><br>\n");
	fwrite($fp,"<b>Ending:</b> <i>$end_date</i><br>\n");
	fwrite($fp,"<b>No. of Messages: $size</b>\n<p>\n<ul>\n");

	for ($i = 0; $i <= sizeof($index); $i++) {
		$morf = $index[$i][from];
		$next_morf = $index[$i+1][from];
		if ($next_morf == $morf) {
			$author_index[] = $index[$i];
		} else {
			$author_index[] = $index[$i];
			usort($author_index,"sort_by_date");
			fwrite($fp," <li><b>$morf</b>\n  <ul>\n");
			for ($k = 0; $k < sizeof($author_index); $k++) {
				$from = $author_index[$k][from];
				$link = $author_index[$k][filename];
				$when = $author_index[$k][date];
				$sub = $author_index[$k][subject];
				if (preg_match("/(.*).html/m",$link,$my_match)) $anchor = $my_match[1];
				fwrite($fp,"   <li><a href=\"$link\">$sub</a> <a name=\"$anchor\"> <i>$when</i></a>\n");
			}
			fwrite($fp," </ul>\n");
			unset($author_index);
		}
	}
	fwrite($fp,"</ul><a name=\"end\"><b>Last message date:</b></a> <i>$end_date</i><br>\n");
	fwrite($fp,"<ul>\n <li> <b>Messages sorted by:</b> ");
	fwrite($fp,"<a href=\"date.html#start\">[ date ]</a> <a href=\"author.html#start\">[ author ]</a>\n");
	fwrite($fp," <li> <b><a href=\"..\">Other mail archives</a></b>\n</ul>\n");
	fwrite($fp,"<p><font size=\"2\">This archive generated by <i><a href=\"http://www.cerritoslug.org/downloads/mirphealy.php\">Mirphealy 0.8</a>.</i></font>\n");
	fwrite($fp,"</body></html>");
	fclose($fp);
	
	if(!file_exists("$dir/index.html")){
		echo "Copying thread.html to index.html\n";
		copy("$dir/thread.html","$dir/index.html");
	}

?>