Added TNEF support to decode MS Outlook (winmail.dat) attachments

1 files changed, 352 insertions, 0 deletions

diff --git a/program/lib/tnef_decoder.inc b/program/lib/tnef_decoder.inc
new file mode 100644
index 000000000..f9b7c3684
--- /dev/null
+++ b/program/lib/tnef_decoder.inc
@@ -0,0 +1,352 @@
+<?php
+/*
+ * tnef_decoder.php
+ *  Graham Norbury <gnorbury@bondcar.com>
+ *  (c) 2002 (GNU GPL - see ../../COPYING)
+ *
+ *  Functions for decoding TNEF attachments in native PHP
+ *
+ *  Adapted from original designs by:
+ *    Thomas Boll <tb@boll.ch>             [tnef.c]
+ *    Mark Simpson <damned@world.std.com>  [tnef-1.1.1]
+ *
+ */
+
+define("TNEF_SIGNATURE",      0x223e9f78);
+define("TNEF_LVL_MESSAGE",    0x01);
+define("TNEF_LVL_ATTACHMENT", 0x02);
+
+define("TNEF_STRING", 0x00010000);
+define("TNEF_TEXT",   0x00020000);
+define("TNEF_BYTE",   0x00060000);
+define("TNEF_WORD",   0x00070000);
+define("TNEF_DWORD",  0x00080000);
+
+define("TNEF_ASUBJECT",   TNEF_DWORD  | 0x8004);
+define("TNEF_AMCLASS",    TNEF_WORD   | 0x8008);
+define("TNEF_BODYTEXT",   TNEF_TEXT   | 0x800c);
+define("TNEF_ATTACHDATA", TNEF_BYTE   | 0x800f);
+define("TNEF_AFILENAME",  TNEF_STRING | 0x8010);
+define("TNEF_ARENDDATA",  TNEF_BYTE   | 0x9002);
+define("TNEF_AMAPIATTRS", TNEF_BYTE   | 0x9005);
+define("TNEF_AVERSION",   TNEF_DWORD  | 0x9006);
+
+define("TNEF_MAPI_NULL",           0x0001);
+define("TNEF_MAPI_SHORT",          0x0002);
+define("TNEF_MAPI_INT",            0x0003);
+define("TNEF_MAPI_FLOAT",          0x0004);
+define("TNEF_MAPI_DOUBLE",         0x0005);
+define("TNEF_MAPI_CURRENCY",       0x0006);
+define("TNEF_MAPI_APPTIME",        0x0007);
+define("TNEF_MAPI_ERROR",          0x000a);
+define("TNEF_MAPI_BOOLEAN",        0x000b);
+define("TNEF_MAPI_OBJECT",         0x000d);
+define("TNEF_MAPI_INT8BYTE",       0x0014);
+define("TNEF_MAPI_STRING",         0x001e);
+define("TNEF_MAPI_UNICODE_STRING", 0x001f);
+define("TNEF_MAPI_SYSTIME",        0x0040);
+define("TNEF_MAPI_CLSID",          0x0048);
+define("TNEF_MAPI_BINARY",         0x0102);
+
+define("TNEF_MAPI_ATTACH_MIME_TAG",      0x370E);
+define("TNEF_MAPI_ATTACH_LONG_FILENAME", 0x3707);
+define("TNEF_MAPI_ATTACH_DATA",          0x3701);
+
+function tnef_getx($size, &$buf)
+{
+   $value = null;
+   if (strlen($buf) >= $size)
+   {
+      $value = substr($buf, 0, $size);
+      $buf = substr_replace($buf, '', 0, $size);
+   }
+   return $value;
+}
+
+function tnef_geti8(&$buf)
+{
+   $value = null;
+   if (strlen($buf) >= 1)
+   {
+      $value = ord($buf{0});
+      $buf = substr_replace($buf, '', 0, 1);
+   }
+   return $value;
+}
+
+function tnef_geti16(&$buf)
+{
+   $value = null;
+   if (strlen($buf) >= 2)
+   {
+      $value = ord($buf{0}) +
+               (ord($buf{1}) << 8);
+      $buf = substr_replace($buf, '', 0, 2);
+   }
+   return $value;
+}
+
+function tnef_geti32(&$buf)
+{
+   $value = null;
+   if (strlen($buf) >= 4)
+   {
+      $value = ord($buf{0}) +
+               (ord($buf{1}) << 8) +
+               (ord($buf{2}) << 16) +
+               (ord($buf{3}) << 24);
+      $buf = substr_replace($buf, '', 0, 4);
+   }
+   return $value;
+}
+
+function tnef_decode_attribute($attribute, &$buf)
+{
+   global $debug, $download;
+
+   $length = tnef_geti32($buf);
+   $value = tnef_getx($length, $buf); //data
+   tnef_geti16($buf); //checksum
+
+   if ($debug)
+   {
+      printf("ATTRIBUTE[%08x] %d bytes\n", $attribute, $length);
+   }
+
+   switch($attribute)
+   {
+      case TNEF_BODYTEXT:
+         if (!$download)
+         {
+            printf("<b>Embedded message:</b><pre>%s</pre>",$value);
+         }
+         break;
+
+      default:
+   }
+}
+
+function extract_mapi_attrs($buf, &$attachment_data)
+{
+   global $debug;
+
+   tnef_geti32($buf); // number of attributes
+   while(strlen($buf) > 0)
+   {
+      $value = null;
+      $length = 0;
+      $attr_type = tnef_geti16($buf);
+      $attr_name = tnef_geti16($buf);
+      if ($debug)
+      {
+         printf("mapi attribute: %04x:%04x\n", $attr_type, $attr_name);
+      }
+      switch($attr_type)
+      {
+         case TNEF_MAPI_SHORT:
+            $value = tnef_geti16($buf);
+            break;
+
+         case TNEF_MAPI_INT:
+         case TNEF_MAPI_BOOLEAN:
+            $value = tnef_geti32($buf);
+            break;
+
+         case TNEF_MAPI_FLOAT:
+            $value = tnef_getx(4, $buf);
+            break;
+
+         case TNEF_MAPI_DOUBLE:
+         case TNEF_MAPI_SYSTIME:
+            $value = tnef_getx(8, $buf);
+            break;
+
+         case TNEF_MAPI_STRING:
+         case TNEF_MAPI_UNICODE_STRING:
+         case TNEF_MAPI_BINARY:
+         case TNEF_MAPI_OBJECT:
+            $num_vals = tnef_geti32($buf);
+            for ($i = 0; $i < $num_vals; $i++) // usually just 1
+            {
+               $length = tnef_geti32($buf);
+               $buflen = $length + ((4 - ($length % 4)) % 4); // pad to next 4 byte boundary
+               $value = substr(tnef_getx($buflen, $buf), 0, $length); // read and truncate to length
+            }
+            break;
+
+         default:
+            if ($debug)
+            {
+               echo("Unknown mapi attribute!\n");
+            }
+      }
+
+      // store any interesting attributes
+      switch($attr_name)
+      {
+         case TNEF_MAPI_ATTACH_LONG_FILENAME: // used in preference to AFILENAME value
+            $attachment_data[0]['name'] = ereg_replace('.*[\/](.*)$', '\1', $value); // strip path
+            break;
+
+         case TNEF_MAPI_ATTACH_MIME_TAG: // Is this ever set, and what is format?
+            $attachment_data[0]['type0'] = ereg_replace('^(.*)/.*', '\1', $value);
+            $attachment_data[0]['type1'] = ereg_replace('.*/(.*)$', '\1', $value);
+            break;
+
+         case TNEF_MAPI_ATTACH_DATA:
+            tnef_getx(16, $value); // skip the next 16 bytes (unknown data)
+            array_shift($attachment_data); // eliminate the current (bogus) attachment
+            do_tnef_decode($value, $attachment_data); // recursively process the attached message
+            break;
+
+         default:
+      }
+   }
+}
+
+function tnef_decode_message(&$buf)
+{
+   global $debug;
+
+   if ($debug)
+   {
+      echo("MESSAGE ");
+   }
+
+   $attribute = tnef_geti32($buf);
+   tnef_decode_attribute($attribute, $buf);
+}
+
+function tnef_decode_attachment(&$buf, &$attachment_data)
+{
+   global $debug;
+
+   if ($debug)
+   {
+      echo("ATTACHMENT ");
+   }
+
+   $attribute = tnef_geti32($buf);
+   switch($attribute)
+   {    
+      case TNEF_ARENDDATA: // marks start of new attachment
+         $length = tnef_geti32($buf);
+         tnef_getx($length, $buf);
+         tnef_geti16($buf); //checksum
+         if ($debug)
+         {
+            printf("ARENDDATA[%08x]: %d bytes\n", $attribute, $length);
+         }
+         // add a new default data block to hold details of this attachment
+         // reverse order is easier to handle later!
+         array_unshift($attachment_data, array('type0'  => 'application',
+                                               'type1'  => 'octet-stream',
+                                               'name'   => 'unknown',
+                                               'stream' => ''));
+         break;
+
+      case TNEF_AFILENAME: // filename
+         $length = tnef_geti32($buf);
+         $attachment_data[0]['name'] = ereg_replace('.*[\/](.*)$',
+                                                    '\1',
+                                                    tnef_getx($length, $buf)); // strip path
+         tnef_geti16($buf); //checksum
+         if ($debug)
+         {
+            printf("AFILENAME[%08x]: %s\n", $attribute, $attachment_data[0]['name']);
+         }
+         break;
+
+      case TNEF_ATTACHDATA: // the attachment itself
+         $length = tnef_geti32($buf);
+         $attachment_data[0]['size'] = $length;
+         $attachment_data[0]['stream'] = tnef_getx($length, $buf);
+         tnef_geti16($buf); //checksum
+         if ($debug)
+         {
+            printf("ATTACHDATA[%08x]: %d bytes\n", $attribute, $length);
+         }
+         break;
+
+      case TNEF_AMAPIATTRS:
+         $length = tnef_geti32($buf);
+         $value = tnef_getx($length, $buf);
+         tnef_geti16($buf); //checksum
+         if ($debug)
+         {
+            printf("AMAPIATTRS[%08x]: %d bytes\n", $attribute, $length);
+         }
+         extract_mapi_attrs($value, $attachment_data);
+         break;
+
+      default:
+         tnef_decode_attribute($attribute, $buf);
+   }
+}
+
+function do_tnef_decode(&$buf, &$attachment_data)
+{
+   global $debug;
+
+   $tnef_signature = tnef_geti32($buf);
+   if ($tnef_signature == TNEF_SIGNATURE)
+   {
+      $tnef_key = tnef_geti16($buf);
+      if ($debug)
+      {
+         printf("Signature: 0x%08x\nKey: 0x%04x\n", $tnef_signature, $tnef_key);
+      }
+
+      while (strlen($buf) > 0)
+      {
+         $lvl_type = tnef_geti8($buf);
+         switch($lvl_type)
+         {
+            case TNEF_LVL_MESSAGE:
+               tnef_decode_message($buf);
+               break;
+
+            case TNEF_LVL_ATTACHMENT:
+               tnef_decode_attachment($buf, $attachment_data);
+               break;
+
+            default:
+               if ($debug)
+               {
+                  echo("Invalid file format!");
+               }
+               break 2;
+         }
+      }
+   }
+   else
+   {
+      if ($debug)
+      {
+         echo("Invalid file format!");
+      }
+   }
+}
+
+function tnef_decode($buf)
+{
+   global $debug;
+
+   $attachment_data = array();
+
+   if ($debug)
+   {
+      echo("<pre>");
+   }
+
+   do_tnef_decode($buf, $attachment_data);
+
+   if ($debug)
+   {
+      echo("</pre>");
+   }
+   return array_reverse($attachment_data);
+
+}
+
+?>
\ No newline at end of file