<?php

/*
  directory exporter for bsdos

  usage: php dir_export.php > disk.csv

  (c) 2009 ub880d
*/

// file types in sense of having header/body or not..
$headbody = array ("--","H-", "-B", "HB");

// open disk
$fp = fopen("disk","rb");

// print csv header
print "disk;dir;loc;headbody;type;name;start;len;vars;altaddr;altlen;date;time\n";

$dnum = 0;
while (!feof($fp)) {

  // drive (image) number are '1' based
  $dnum++;
  
  // free memory .. not really needed, but without this script will use more than 4MB of memory for a while and it is not necessary ;]
  if (isset($disk)) unset($disk);

  // read image if there is less than 2MB to read, finish (maybe it won't work over network.. but who cares? ;])
  $disk = fread($fp, 2*1024*1024);
  if (strlen($disk)<2*1024*1024) break;
  
  // mb02 format checks.. this is only simple check, maybe we could check also diskette ID
  if (ord($disk[3]) != 2 || ord($disk[32]) != 0 || ord($disk[37]) != 0) continue;

  // geometry values.. we don't need them for our csv, but we will check them
  $trcks = ord($disk[4]) + 256*ord($disk[5]);
  $sects = ord($disk[6]) + 256*ord($disk[7]);
  $sides = ord($disk[8]) + 256*ord($disk[9]);
  $clust = ord($disk[10]) + 256*ord($disk[11]);

  // check also some reasonable values for geometry
  if ($sides>2 || $sects>16 || $trcks>255) continue; 
  
  // this is directory map sector
  $dirmap = ord($disk[12]) + 256*ord($disk[13]);
  
  // size of 1 fat.. we don't need this, it is here just for curious people ;]
  $fatsz = ord($disk[14]) + 256*ord($disk[15]);
  
  // addresses of first and second fat, but we will use first one only
  $fat1 = ord($disk[18]) + 256*ord($disk[19]);
  $fat2 = ord($disk[20]) + 256*ord($disk[21]);

  // diskette can contain up to 256 directories
  for ($dir = 0;$dir<256;$dir++) {
    
    // is this valid bsdos directory? (is it used?)
    if (ord($disk[$dirmap*1024 + $dir*4]) != 0x80) continue;
    
    // get directory sector
    $dirsec = ord($disk[$dirmap*1024 + $dir*4 + 2]) + 256*((ord($disk[$dirmap*1024 + $dir*4 + 3])) & 0x1f);
    
    // order of directory sector
    $dirsecnum = 0;
    
    while (1) {
    
      // in 1 directory sector can lay up to 32 files
      for ($j = 0;$j<32;$j++) {
        
        // first in directory list is name of directory, so we skip that one
        if (($j + 32*$dirsecnum) == 0) continue;
        
        // skip unused/erased directory entries
        $f = ord($disk[$dirsec*1024 + $j*32]);
        if (($f & 0xcf) != 0x80) continue;

        // count file location in directory
        $loc = $j + 32*$dirsecnum;
        
        // get header/body string
        $hb = $headbody[($f/16) & 3];

        // file type.. 0-basic, 3-bytes .. and many others ;]
        $type = ord($disk[$dirsec*1024 + $j*32 + 5]);
        
        // filename
        $name = substr($disk, $dirsec*1024 + $j*32 + 6, 10);
        
        // this part is to substitute nasty characters with theirs ascii value in brackets
        $n = "";
        for ($i = 0;$i<10;$i++)
          if (ord($name[$i])<32 || ord($name[$i])>127) $n .= "[".ord($name[$i])."]";
          else $n .= $name[$i];
        $name = $n;
        
        // get various sizes and addresses of file, bsdos seems to use $start and $altlen in CAT print
        $len     = ord($disk[$dirsec*1024 + $j*32 + 16]) + 256*ord($disk[$dirsec*1024 + $j*32 + 17]);
        $start   = ord($disk[$dirsec*1024 + $j*32 + 18]) + 256*ord($disk[$dirsec*1024 + $j*32 + 19]);
        $vars    = ord($disk[$dirsec*1024 + $j*32 + 20]) + 256*ord($disk[$dirsec*1024 + $j*32 + 21]);
        $altaddr = ord($disk[$dirsec*1024 + $j*32 + 22]) + 256*ord($disk[$dirsec*1024 + $j*32 + 23]);
        $altlen  = ord($disk[$dirsec*1024 + $j*32 + 24]) + 256*ord($disk[$dirsec*1024 + $j*32 + 25])
                   + 65536*(ord($disk[$dirsec*1024 + $j*32 + 26]) + 256*ord($disk[$dirsec*1024 + $j*32 + 27]));
        
        // get date and time
        $tim = ord($disk[$dirsec*1024 + $j*32 + 1]) + 256*ord($disk[$dirsec*1024 + $j*32 + 2]);
	$dat = ord($disk[$dirsec*1024 + $j*32 + 3]) + 256*ord($disk[$dirsec*1024 + $j*32 + 4]);
	$hod = ($tim >> 11) & 0x1f;
	$min = ($tim >> 5) & 0x3f;
	$sec = ($tim << 1) & 0x3f;

	$yr  = ($dat >> 9) + 1980;
	$mon = (($dat >> 5) & 0x0f);
	$day = ($dat & 0x1f);
	
	// if $day==0 or $mon==0 .. date is not valid
	if ($day*$mon != 0) $date = sprintf("%02d.%02d.%04d", $day, $mon, $yr);
	else $date = "";
	
	// maybe we can also assume that time is wrong when $day==0 or $mon==0, but for now..
	$time = sprintf("%02d:%02d", $hod, $min);
	
	// and finaly print collected data
        print "$dnum;$dir;$loc;$hb;$type;\"$name\";$start;$len;$vars;$altaddr;$altlen;$date;$time\n";
      }
      
      // get next directory sector from fat1
      $next1 = ord($disk[$fat1*1024 + $dirsec*2]);
      $next2 = ord($disk[$fat1*1024 + $dirsec*2 + 1]);

      // if it was last directory sector then exit loop
      if (($next2 & 0xe0) != 0xc0) break;

      // advance to next directory sector
      $dirsec = $next1 + 256*($next2 & 0x0f);
      $dirsecnum++;
      
    }
  }
  

}

// we are done..
fclose($fp);

?>