Better support for malformed/absent dates and headerlines ending in only "\n".
[squirrelmail.git] / class / mime / Rfc822Header.class.php
index 4c3015bc31fde354036589466082148246cb16c2..9f106d86b62d334ede869a4062f9666dc1e1a23f 100644 (file)
@@ -3,24 +3,26 @@
 /**
  * Rfc822Header.class.php
  *
- * Copyright (c) 2003 The SquirrelMail Project Team
+ * Copyright (c) 2003-2004 The SquirrelMail Project Team
  * Licensed under the GNU GPL. For full terms see the file COPYING.
  *
  * This contains functions needed to handle mime messages.
  *
  * $Id$
+ * @package squirrelmail
  */
 
-/*
- * rdc822_header class
+/**
  * input: header_string or array
+ * @package squirrelmail
  */
 class Rfc822Header {
-    var $date = '',
+    var $date = -1,
         $subject = '',
         $from = array(),
         $sender = '',
         $reply_to = array(),
+        $mail_followup_to = array(),
         $to = array(),
         $cc = array(),
         $bcc = array(),
@@ -34,6 +36,8 @@ class Rfc822Header {
         $priority = 3,
         $dnt = '',
         $encoding = '',
+        $content_id = '',
+        $content_desc = '',
         $mlist = array(),
         $more_headers = array(); /* only needed for constructing headers
                                     in smtp.php */
@@ -41,12 +45,12 @@ class Rfc822Header {
         if (is_array($hdr)) {
             $hdr = implode('', $hdr);
         }
-        /* First we unfold the header */
-        $hdr = trim(str_replace(array("\r\n\t", "\r\n "),array('', ''), $hdr));
+        /* First we replace \r\n by \n and unfold the header */
+        $hdr = trim(str_replace(array("\r\n", "\n\t", "\n "),array("\n", ' ', ' '), $hdr));
 
         /* Now we can make a new header array with */
         /* each element representing a headerline  */
-        $hdr = explode("\r\n" , $hdr);
+        $hdr = explode("\n" , $hdr);
         foreach ($hdr as $line) {
             $pos = strpos($line, ':');
             if ($pos > 0) {
@@ -125,6 +129,9 @@ class Rfc822Header {
             case 'reply-to':
                 $this->reply_to = $this->parseAddress($value, true);
                 break;
+            case 'mail-followup-to':
+                $this->mail_followup_to = $this->parseAddress($value, true);
+                break;
             case 'to':
                 $this->to = $this->parseAddress($value, true);
                 break;
@@ -164,6 +171,16 @@ class Rfc822Header {
                 $value = $this->stripComments($value);
                 $this->parseDisposition($value);
                 break;
+            case 'content-transfer-encoding':
+                $this->encoding = $value;
+                break;
+            case 'content-description':
+                $this->content_desc = $value;
+                break;
+            case 'content-id':
+                $value = $this->stripComments($value);
+                $this->content_id = $value;
+                break;
             case 'user-agent':
             case 'x-mailer':
                 $this->xmailer = $value;
@@ -176,11 +193,11 @@ class Rfc822Header {
                 $this->mlist('post', $value);
                 break;
             case 'list-reply':
-                $value = $this->stripComments($value);            
+                $value = $this->stripComments($value);
                 $this->mlist('reply', $value);
                 break;
             case 'list-subscribe':
-                $value = $this->stripComments($value);            
+                $value = $this->stripComments($value);
                 $this->mlist('subscribe', $value);
                 break;
             case 'list-unsubscribe':
@@ -230,10 +247,22 @@ class Rfc822Header {
                    $i = $iEnd;
                 }
                 $sToken = str_replace($aReplace, $aSpecials,$sToken);
-                $aTokens[] = $sToken;
+                if ($sToken) $aTokens[] = $sToken;
                 break;
             case '"':
                 $iEnd = strpos($address,$cChar,$i+1);
+                if ($iEnd) {
+                   // skip escaped quotes
+                   $prev_char = $address{$iEnd-1};
+                   while ($prev_char === '\\' && substr($address,$iEnd-2,2) !== '\\\\') {
+                       $iEnd = strpos($address,$cChar,$iEnd+1);
+                       if ($iEnd) {
+                          $prev_char = $address{$iEnd-1};
+                       } else {
+                          $prev_char = false;
+                       }
+                   }
+                }
                 if (!$iEnd) {
                     $sToken = substr($address,$i);
                     $i = $iCnt;
@@ -246,6 +275,7 @@ class Rfc822Header {
                 if ($sToken) $aTokens[] = $sToken;
                 break;
             case '(':
+                array_pop($aTokens); //remove inserted space
                 $iEnd = strpos($address,')',$i);
                 if (!$iEnd) {
                     $sToken = substr($address,$i);
@@ -277,11 +307,30 @@ class Rfc822Header {
                         $i = $iEnd;
                     }
                 }
+                // check the next token in case comments appear in the middle of email addresses
+                $prevToken = end($aTokens);
+                if (!in_array($prevToken,$aSpecials,true)) {
+                    if ($i+1<strlen($address) && !in_array($address{$i+1},$aSpecials,true)) {
+                        $iEnd = strpos($address,' ',$i+1);
+                        if ($iEnd) {
+                            $sNextToken = trim(substr($address,$i+1,$iEnd - $i -1));
+                            $i = $iEnd-1;
+                        } else {
+                            $sNextToken = trim(substr($address,$i+1));
+                            $i = $iCnt;
+                        }
+                        // remove the token
+                        array_pop($aTokens);
+                        // create token and add it again
+                        $sNewToken = $prevToken . $sNextToken;
+                        if($sNewToken) $aTokens[] = $sNewToken;
+                    }
+                }
                 $sToken = str_replace($aReplace, $aSpecials,$sToken);
-                $aTokens[] = $sToken;
+                if ($sToken) $aTokens[] = $sToken;
                 break;
             case ',':
-            case ';':
+            case ':':
             case ';':
             case ' ':
                 $aTokens[] = $cChar;
@@ -302,6 +351,7 @@ class Rfc822Header {
         return $aTokens;
     }
     function createAddressObject(&$aStack,&$aComment,&$sEmail,$sGroup='') {
+        //$aStack=explode(' ',implode('',$aStack));
         if (!$sEmail) {
             while (count($aStack) && !$sEmail) {
                 $sEmail = trim(array_pop($aStack));
@@ -309,7 +359,7 @@ class Rfc822Header {
         }
         if (count($aStack)) {
             $sPersonal = trim(implode('',$aStack));
-        } else { 
+        } else {
             $sPersonal = '';
         }
         if (!$sPersonal && count($aComment)) {
@@ -322,7 +372,7 @@ class Rfc822Header {
         } else {
             $oAddr->personal = $sPersonal;
         }
-        $oAddr->group = $sGroup;
//       $oAddr->group = $sGroup;
         $iPosAt = strpos($sEmail,'@');
         if ($iPosAt) {
            $oAddr->mailbox = substr($sEmail, 0, $iPosAt);
@@ -331,25 +381,24 @@ class Rfc822Header {
            $oAddr->mailbox = $sEmail;
            $oAddr->host = false;
         }
-        $oAddr->group = $sGroup;
         $sEmail = '';
         $aStack = $aComment = array();
         return $oAddr;
     }
 
     /*
-     * parseAddress: recursive function for parsing address strings and store 
+     * parseAddress: recursive function for parsing address strings and store
      *               them in an address stucture object.
      *               input: $address = string
      *                      $ar      = boolean (return array instead of only the
      *                                 first element)
      *                      $addr_ar = array with parsed addresses // obsolete
      *                      $group   = string // obsolete
-     *                      $host    = string (default domainname in case of 
+     *                      $host    = string (default domainname in case of
      *                                 addresses without a domainname)
      *                      $lookup  = callback function (for lookup address
      *                                 strings which are probably nicks
-     *                                 (without @ ) ) 
+     *                                 (without @ ) )
      *               output: array with addressstructure objects or only one
      *                       address_structure object.
      *  personal name: encoded: =?charset?Q|B?string?=
@@ -372,20 +421,19 @@ class Rfc822Header {
             case '=':
             case '"':
             case ' ':
-                $aStack[] = $sToken; 
+                $aStack[] = $sToken;
                 break;
             case '(':
                 $aComment[] = substr($sToken,1,-1);
                 break;
             case ';':
                 if ($sGroup) {
+                    $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
                     $oAddr = end($aAddress);
-                    if ($oAddr && $oAddr->group == $sGroup) {
-                        $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
-                    } else {
-                        /* group is empty */
-                        $aAddress[] = $this->createAddressObject(array(),array(),$sGroup,'');
+                    if(!$oAddr || ((isset($oAddr)) && !$oAddr->mailbox && !$oAddr->personal)) {
+                        $sEmail = $sGroup . ':;';
                     }
+                    $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
                     $sGroup = '';
                     $aStack = $aComment = array();
                     break;
@@ -393,8 +441,9 @@ class Rfc822Header {
             case ',':
                 $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
                 break;
-            case ':': 
-                $sGroup = implode(' ',$aStack); break;
+            case ':':
+                $sGroup = trim(implode(' ',$aStack));
+                $sGroup = preg_replace('/\s+/',' ',$sGroup);
                 $aStack = array();
                 break;
             case '<':
@@ -402,7 +451,7 @@ class Rfc822Header {
                break;
             case '>':
                /* skip */
-               break; 
+               break;
             default: $aStack[] = $sToken; break;
             }
         }
@@ -437,20 +486,24 @@ class Rfc822Header {
                 if ($sHost && $oAddr->mailbox) {
                     $oAddr->host = $sHost;
                 }
+            } else if (!$grouplookup && !$oAddr->host) {
+                if ($sHost && $oAddr->mailbox) {
+                    $oAddr->host = $sHost;
+                }
             }
           }
           if (!$aAddrBookAddress && $oAddr->mailbox) {
               $aProcessedAddress[] = $oAddr;
           } else {
-              $aProcessedAddress = array_merge($aProcessedAddress,$aAddrBookAddress); 
+              $aProcessedAddress = array_merge($aProcessedAddress,$aAddrBookAddress);
           }
         }
-        if ($ar) { 
+        if ($ar) {
             return $aProcessedAddress;
         } else {
             return $aProcessedAddress[0];
         }
-    } 
+    }
 
     function parseContentType($value) {
         $pos = strpos($value, ';');
@@ -472,6 +525,40 @@ class Rfc822Header {
         $this->content_type = $content_type;
     }
 
+    /* RFC2184 */
+    function processParameters($aParameters) {
+        $aResults = array();
+        $aCharset = array();
+        // handle multiline parameters
+        foreach($aParameters as $key => $value) {
+            if ($iPos = strpos($key,'*')) {
+                $sKey = substr($key,0,$iPos);
+                if (!isset($aResults[$sKey])) {
+                    $aResults[$sKey] = $value;
+                    if (substr($key,-1) == '*') { // parameter contains language/charset info
+                        $aCharset[] = $sKey;
+                    }
+                } else {
+                    $aResults[$sKey] .= $value;
+                }
+            } else {
+                $aResults[$key] = $value;
+            }
+        }
+        foreach ($aCharset as $key) {
+            $value = $aResults[$key];
+            // extract the charset & language
+            $charset = substr($value,0,strpos($value,"'"));
+            $value = substr($value,strlen($charset)+1);
+            $language = substr($value,0,strpos($value,"'"));
+            $value = substr($value,strlen($charset)+1);
+            // FIX ME What's the status of charset decode with language information ????
+            $value = charset_decode($charset,$value);
+            $aResults[$key] = $value;
+        }
+        return $aResults;
+    }
+
     function parseProperties($value) {
         $propArray = explode(';', $value);
         $propResultArray = array();
@@ -487,7 +574,7 @@ class Rfc822Header {
                 $propResultArray[$key] = $val;
             }
         }
-        return $propResultArray;
+        return $this->processParameters($propResultArray);
     }
 
     function parseDisposition($value) {
@@ -598,7 +685,7 @@ class Rfc822Header {
         }
         return $arr;
     }
-    
+
     function findAddress($address, $recurs = false) {
         $result = false;
         if (is_array($address)) {
@@ -613,7 +700,7 @@ class Rfc822Header {
                         $result = $i;
                     }
                 }
-                ++$i;        
+                ++$i;
             }
         } else {
             if (!is_array($this->cc)) $this->cc = array();
@@ -653,7 +740,7 @@ class Rfc822Header {
                 return true;
             } else {
                 return false;
-            }        
+            }
         }
         //exit;
         return $result;