Rewrote address parsing and now it works a lot better.
[squirrelmail.git] / class / mime / Rfc822Header.class.php
index f4db5a0294d3dd4e6c03dce8228928de7f80db60..0855604dd930da0c29ae289797813287f3e32e86 100644 (file)
@@ -3,7 +3,7 @@
 /**
  * Rfc822Header.class.php
  *
- * Copyright (c) 2002 The SquirrelMail Project Team
+ * Copyright (c) 2003 The SquirrelMail Project Team
  * Licensed under the GNU GPL. For full terms see the file COPYING.
  *
  * This contains functions needed to handle mime messages.
@@ -26,13 +26,14 @@ class Rfc822Header {
         $bcc = array(),
         $in_reply_to = '',
         $message_id = '',
-       $references = '',
+        $references = '',
         $mime = false,
         $content_type = '',
         $disposition = '',
         $xmailer = '',
         $priority = 3,
         $dnt = '',
+        $encoding = '',
         $mlist = array(),
         $more_headers = array(); /* only needed for constructing headers
                                     in smtp.php */
@@ -40,7 +41,6 @@ class Rfc822Header {
         if (is_array($hdr)) {
             $hdr = implode('', $hdr);
         }
-
         /* First we unfold the header */
         $hdr = trim(str_replace(array("\r\n\t", "\r\n "),array('', ''), $hdr));
 
@@ -51,14 +51,10 @@ class Rfc822Header {
             $pos = strpos($line, ':');
             if ($pos > 0) {
                 $field = substr($line, 0, $pos);
-               if (!strstr($field,' ')) { /* valid field */
-                   $value = trim(substr($line, $pos+1));
-                   if(!preg_match('/^X.*/i', $field) &&
-                       !preg_match('/^Subject/i', $field)) {
-                       $value = $this->stripComments($value);
-                    }
-                   $this->parseField($field, $value);
-               }
+                if (!strstr($field,' ')) { /* valid field */
+                        $value = trim(substr($line, $pos+1));
+                        $this->parseField($field, $value);
+                }
             }
         }
         if ($this->content_type == '') {
@@ -68,7 +64,6 @@ class Rfc822Header {
 
     function stripComments($value) {
         $result = '';
-
         $cnt = strlen($value);
         for ($i = 0; $i < $cnt; ++$i) {
             switch ($value{$i}) {
@@ -113,6 +108,7 @@ class Rfc822Header {
         $field = strtolower($field);
         switch($field) {
             case 'date':
+                $value = $this->stripComments($value);
                 $d = strtr($value, array('  ' => ' '));
                 $d = explode(' ', $d);
                 $this->date = getTimeStamp($d);
@@ -142,22 +138,30 @@ class Rfc822Header {
                 $this->in_reply_to = $value;
                 break;
             case 'message-id':
+                $value = $this->stripComments($value);
                 $this->message_id = $value;
                 break;
-           case 'references':
-               $this->references = $value;
-               break;
+            case 'references':
+                $value = $this->stripComments($value);
+                $this->references = $value;
+                break;
+            case 'x-confirm-reading-to':
+            case 'return-receipt-to':
             case 'disposition-notification-to':
+                $value = $this->stripComments($value);
                 $this->dnt = $this->parseAddress($value);
                 break;
             case 'mime-version':
+                $value = $this->stripComments($value);
                 $value = str_replace(' ', '', $value);
                 $this->mime = ($value == '1.0' ? true : $this->mime);
                 break;
             case 'content-type':
+                $value = $this->stripComments($value);
                 $this->parseContentType($value);
                 break;
             case 'content-disposition':
+                $value = $this->stripComments($value);
                 $this->parseDisposition($value);
                 break;
             case 'user-agent':
@@ -168,41 +172,156 @@ class Rfc822Header {
                 $this->priority = $value;
                 break;
             case 'list-post':
+                $value = $this->stripComments($value);
                 $this->mlist('post', $value);
                 break;
             case 'list-reply':
+                $value = $this->stripComments($value);            
                 $this->mlist('reply', $value);
                 break;
             case 'list-subscribe':
+                $value = $this->stripComments($value);            
                 $this->mlist('subscribe', $value);
                 break;
             case 'list-unsubscribe':
+                $value = $this->stripComments($value);
                 $this->mlist('unsubscribe', $value);
                 break;
             case 'list-archive':
+                $value = $this->stripComments($value);
                 $this->mlist('archive', $value);
                 break;
             case 'list-owner':
+                $value = $this->stripComments($value);
                 $this->mlist('owner', $value);
                 break;
             case 'list-help':
+                $value = $this->stripComments($value);
                 $this->mlist('help', $value);
                 break;
             case 'list-id':
+                $value = $this->stripComments($value);
                 $this->mlist('id', $value);
                 break;
             default:
                 break;
         }
     }
+
+    function getAddressTokens($address) {
+        $aTokens = array();
+        $aAddress = array();
+        $iCnt = strlen($address);
+        $aSpecials = array('(' ,'<' ,',' ,';' ,':');
+        $aReplace =  array(' (',' <',' ,',' ;',' :');
+        $address = str_replace($aSpecials,$aReplace,$address);
+        $i = 0;
+        while ($i < $iCnt) {
+            $cChar = $address{$i};
+            switch($cChar)
+            {
+            case '<':
+                $iEnd = strpos($address,'>',$i+1);
+                if (!$iEnd) {
+                   $sToken = substr($address,$i);
+                   $i = $iCnt;
+                } else {
+                   $sToken = substr($address,$i,$iEnd - $i +1);
+                   $i = $iEnd;
+                }
+                $sToken = str_replace($aReplace, $aSpecials,$sToken);
+                $aTokens[] = $sToken;
+                break;
+            case '"':
+                $iEnd = strpos($address,$cChar,$i+1);
+                if (!$iEnd) {
+                    $sToken = substr($address,$i);
+                    $i = $iCnt;
+                } else {
+                    // also remove the surrounding quotes
+                    $sToken = substr($address,$i+1,$iEnd - $i -1);
+                    $i = $iEnd;
+                }
+                $sToken = str_replace($aReplace, $aSpecials,$sToken);
+                $aTokens[] = $sToken;
+                break;
+            case '(':
+                $iEnd = strpos($address,')',$i);
+                if (!$iEnd) {
+                    $sToken = substr($address,$i);
+                    $i = $iCnt;
+                } else {
+                    $sToken = substr($address,$i,$iEnd - $i + 1);
+                    $i = $iEnd;
+                }
+                $sToken = str_replace($aReplace, $aSpecials,$sToken);
+                $aTokens[] = $sToken;
+                break;
+            case ',':
+            case ';':
+            case ';':
+            case ' ':
+                $aTokens[] = $cChar;
+                break;
+            default:
+                $iEnd = strpos($address,' ',$i+1);
+                if ($iEnd) {
+                    $sToken = trim(substr($address,$i,$iEnd - $i));
+                    $i = $iEnd-1;
+                } else {
+                    $sToken = trim(substr($address,$i));
+                    $i = $iCnt;
+                }
+                if ($sToken) $aTokens[] = $sToken;
+            }
+            ++$i;
+        }
+        return $aTokens;
+    }
+    function createAddressObject(&$aStack,&$aComment,&$sEmail,$sGroup='') {
+        if (!$sEmail) {
+            while (count($aStack) && !$sEmail) {
+                $sEmail = trim(array_pop($aStack));
+            }
+        }
+        if (count($aStack)) {
+            $sPersonal = trim(implode('',$aStack));
+        } else { 
+            $sPersonal = '';
+        }
+        if (!$sPersonal && count($aComment)) {
+            $sComment = trim(implode(' ',$aComment));
+            $sPersonal .= $sComment;
+        }
+        $oAddr =& new AddressStructure();
+        if ($sPersonal && substr($sPersonal,0,2) == '=?') {
+            $oAddr->personal = encodeHeader($sPersonal);
+        } else {
+            $oAddr->personal = $sPersonal;
+        }
+        $oAddr->group = $sGroup;
+        $iPosAt = strpos($sEmail,'@');
+        if ($iPosAt) {
+           $oAddr->mailbox = substr($sEmail, 0, $iPosAt);
+           $oAddr->host = substr($sEmail, $iPosAt+1);
+        } else {
+           $oAddr->mailbox = $sEmail;
+           $oAddr->host = false;
+        }
+        $oAddr->group = $sGroup;
+        $sEmail = '';
+        $aStack = $aComment = array();
+        return $oAddr;
+    }
+
     /*
      * 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
-     *                      $group   = string
+     *                      $addr_ar = array with parsed addresses // obsolete
+     *                      $group   = string // obsolete
      *                      $host    = string (default domainname in case of 
      *                                 addresses without a domainname)
      *                      $lookup  = callback function (for lookup address
@@ -210,168 +329,112 @@ class Rfc822Header {
      *                                 (without @ ) ) 
      *               output: array with addressstructure objects or only one
      *                       address_structure object.
+     *  personal name: encoded: =?charset?Q|B?string?=
+     *                 quoted:  "string"
+     *                 normal:  string
+     *  email        : <mailbox@host>
+     *               : mailbox@host
+     *  This function is also used for validating addresses returned from compose
+     *  That's also the reason that the function became a little bit huge
      */
-    function parseAddress
-    ($address, $ar=false, $addr_ar = array(), $group = '', $host='',$lookup=false) {
-        $pos = 0;
-        $j = strlen($address);
-        $name = '';
-        $addr = '';
-        while ($pos < $j) {
-            switch ($address{$pos}) {
-                case '"': /* get the personal name */
-                    if ($address{++$pos} == '"') {
-                        ++$pos;
-                    } else {
-                        while ($pos < $j && $address{$pos} != '"') {
-                            if ((substr($address, $pos, 2) == '\\"') ||
-                                (substr($address, $pos, 2) == '\\\\')) {
-                                $name .= $address{$pos++};
-                            }
-                            $name .= $address{$pos++};
-                        }
-                    }
-                    ++$pos;
-                    break;
-                case '<':  /* get email address */
-                    $addr_start = $pos++;
-                    while ($pos < $j && $address{$pos} != '>') {
-                        $addr .= $address{$pos++};
-                    }
-                    ++$pos;
-                    break;
-                case '(':  /* rip off comments */
-                    $addr_start = $pos;
-                    for (++$pos; ($pos < $j) && ($address{$pos} != ')'); ++$pos) {
-                        $addr .= $address{$pos};
-                    }
-                    $address_start = substr($address, 0, $addr_start);
-                    $address_end   = substr($address, $pos + 1);
-                    $address       = $address_start . $address_end;
-                    $j = strlen($address);
-                    $pos = $addr_start + 1;
-                    break;
-                case ',':  /* we reached a delimiter */
-                    if ($addr == '') {
-                        $addr = substr($address, 0, $pos);
-                    } else if ($name == '') {
-                        $name = trim(substr($address, 0, $addr_start));
-                    }
 
-                    $at = strpos($addr, '@');
-                    $addr_structure = new AddressStructure();
-                    $addr_structure->personal = $name;
-                    $addr_structure->group = $group;
-                    if ($at) {
-                        $addr_structure->mailbox = substr($addr, 0, $at);
-                        $addr_structure->host = substr($addr, $at+1);
+    function parseAddress($address,$ar=false,$aAddress=array(),$sGroup='',$sHost='',$lookup=false) {
+        $aTokens = $this->getAddressTokens($address);
+        $sPersonal = $sEmail = $sComment = $sGroup = '';
+        $aStack = $aComment = array();
+        foreach ($aTokens as $sToken) {
+            $cChar = $sToken{0};
+            switch ($cChar)
+            {
+            case '=':
+            case '"':
+            case ' ':
+                $aStack[] = $sToken; 
+                break;
+            case '(':
+                $aComment[] = substr($sToken,1,-1);
+                break;
+            case ';':
+                if ($sGroup) {
+                    $oAddr = end($aAddress);
+                    if ($oAddr && $oAddr->group == $sGroup) {
+                        $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
                     } else {
-                       /* if lookup function */
-                       if ($lookup) {
-                           $aAddr = call_user_func_array($lookup,array($addr));
-                           if (isset($aAddr['email'])) {
-                               $at = strpos($aAddr['email'], '@');
-                               $addr_structure->mailbox = substr($aAddr['email'], 0, $at);
-                               $addr_structure->host = substr($aAddr['email'], $at+1);
-                               if (isset($aAddr['name'])) {
-                                   $addr_structure->personal = $aAddr['name'];
-                               }
-                           }
-                       }
-                       if (!isset($addr_structure->mailbox)) {
-                           $addr_structure->mailbox = trim($addr);
-                           if ($host) {
-                               $addr_structure->host = $host;
-                           }
-                       }
-                    }
-                    $address = trim(substr($address, $pos+1));
-                    $j = strlen($address);
-                    $pos = 0;
-                    $name = '';
-                    $addr = '';
-                    $addr_ar[] = $addr_structure;
-                    break;
-                case ':':  /* process the group addresses */
-                    /* group marker */
-                    $group = substr($address, 0, $pos);
-                    $address = substr($address, $pos+1);
-                    $result = $this->parseAddress($address, $ar, $addr_ar, $group);
-                    $addr_ar = $result[0];
-                    $pos = $result[1];
-                    $address = substr($address, $pos++);
-                    $j = strlen($address);
-                    $group = '';
-                    break;
-                case ';':
-                    if ($group) {
-                        $address = substr($address, 0, $pos - 1);
+                        /* group is empty */
+                        $aAddress[] = $this->createAddressObject(array(),array(),$sGroup,'');
                     }
-                    ++$pos;
-                    break;
-                default:
-                    ++$pos;
+                    $sGroup = '';
+                    $aStack = $aComment = array();
                     break;
+                }
+            case ',':
+                $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail,$sGroup);
+                break;
+            case ':': 
+                $sGroup = implode(' ',$aStack); break;
+                $aStack = array();
+                break;
+            case '<':
+               $sEmail = trim(substr($sToken,1,-1));
+               break;
+            case '>':
+               /* skip */
+               break; 
+            default: $aStack[] = $sToken; break;
             }
         }
-        if ($addr == '') {
-            $addr = substr($address, 0, $pos);
-        } else if ($name == '') {
-            $name = trim(substr($address, 0, $addr_start));
-        }
-        $at = strpos($addr, '@');
-        $addr_structure = new AddressStructure();
-        $addr_structure->group = $group;
-        if ($at) {
-            $addr_structure->mailbox = trim(substr($addr, 0, $at));
-            $addr_structure->host = trim(substr($addr, $at+1));
-        } else {
-           /* if lookup function */
-           if ($lookup) {
-               $aAddr = call_user_func_array($lookup,array($addr));
-               if (isset($aAddr['email'])) {
-                   $at = strpos($aAddr['email'], '@');
-                    $addr_structure->mailbox = substr($aAddr['email'], 0, $at);
-                    $addr_structure->host = substr($aAddr['email'], $at+1);
-                   if (isset($aAddr['name'])) {
-                       $addr_structure->personal = $aAddr['name'];
-                   }
-               }
-           }
-           if (!isset($addr_structure->mailbox)) {
-                $addr_structure->mailbox = trim($addr);
-               if ($host) {
-                   $addr_structure->host = $host;
-               }
-           }
-        }
-        if ($group && $addr == '') { /* no addresses found in group */
-            $name = "$group";
-            $addr_structure->personal = $name;
-            $addr_ar[] = $addr_structure;
-            return (array($addr_ar,$pos+1 ));
-       } elseif ($group) {
-            $addr_structure->personal = $name;
-            $addr_ar[] = $addr_structure;
-           return (array($addr_ar,$pos+1 ));
-        } else {
-            $addr_structure->personal = $name;
-            if ($name || $addr) {
-                $addr_ar[] = $addr_structure;
+        /* now do the action again for the last address */
+        $aAddress[] = $this->createAddressObject($aStack,$aComment,$sEmail);
+        /* try to lookup the addresses in case of invalid email addresses */
+        $aProcessedAddress = array();
+        foreach ($aAddress as $oAddr) {
+          $aAddrBookAddress = array();
+          if (!$oAddr->host) {
+            $grouplookup = false;
+            if ($lookup) {
+                 $aAddr = call_user_func_array($lookup,array($oAddr->mailbox));
+                 if (isset($aAddr['email'])) {
+                     if (strpos($aAddr['email'],',')) {
+                         $grouplookup = true;
+                         $aAddrBookAddress = $this->parseAddress($aAddr['email'],true);
+                     } else {
+                         $iPosAt = strpos($aAddr['email'], '@');
+                         $oAddr->mailbox = substr($aAddr['email'], 0, $iPosAt);
+                         $oAddr->host = substr($aAddr['email'], $iPosAt+1);
+                         if (isset($aAddr['name'])) {
+                             $oAddr->personal = $aAddr['name'];
+                         } else {
+                             $oAddr->personal = encodeHeader($sPersonal);
+                         }
+                     }
+                 }
             }
+            if (!$grouplookup && !$oAddr->mailbox) {
+                $oAddr->mailbox = trim($sEmail);
+                if ($sHost && $oAddr->mailbox) {
+                    $oAddr->host = $sHost;
+                }
+            }
+          }
+          if (!$aAddrBookAddress && $oAddr->mailbox) {
+              $aProcessedAddress[] = $oAddr;
+          } else {
+              $aProcessedAddress = array_merge($aProcessedAddress,$aAddrBookAddress); 
+          }
         }
-        if ($ar) {
-            return ($addr_ar);
+        if ($ar) { 
+            return $aProcessedAddress;
+        } else {
+            return $aProcessedAddress[0];
         }
-        return ($addr_ar[0]);
-    }
+    } 
 
     function parseContentType($value) {
         $pos = strpos($value, ';');
         $props = '';
         if ($pos > 0) {
            $type = trim(substr($value, 0, $pos));
-           $props = trim(substr($type, $pos+1));
+           $props = trim(substr($value, $pos+1));
         } else {
            $type = $value;
         }
@@ -442,28 +505,36 @@ class Rfc822Header {
      * example1: header->getAddr_s('to').
      * example2: header->getAddr_s(array('to', 'cc', 'bcc'))
      */
-    function getAddr_s($arr, $separator = ',') {
+    function getAddr_s($arr, $separator = ',',$encoded=false) {
         $s = '';
 
         if (is_array($arr)) {
             foreach($arr as $arg) {
-                if ($this->getAddr_s($arg)) {
+                if ($this->getAddr_s($arg, $separator, $encoded)) {
                     $s .= $separator . $result;
                 }
             }
             $s = ($s ? substr($s, 2) : $s);
         } else {
-            eval('$addr = $this->' . $arr . ';') ;
+            $addr = $this->{$arr};
             if (is_array($addr)) {
                 foreach ($addr as $addr_o) {
                     if (is_object($addr_o)) {
-                        $s .= $addr_o->getAddress() . $separator;
+                        if ($encoded) {
+                            $s .= $addr_o->getEncodedAddress() . $separator;
+                        } else {
+                            $s .= $addr_o->getAddress() . $separator;
+                        }
                     }
                 }
                 $s = substr($s, 0, -strlen($separator));
             } else {
                 if (is_object($addr)) {
-                    $s .= $addr->getAddress();
+                    if ($encoded) {
+                        $s .= $addr->getEncodedAddress();
+                    } else {
+                        $s .= $addr->getAddress();
+                    }
                 }
             }
         }
@@ -476,7 +547,7 @@ class Rfc822Header {
                 $arr = $this->getAddr_a($argument, $excl_arr, $arr);
             }
         } else {
-            eval('$addr = $this->' . $arg . ';') ;
+            $addr = $this->{$arg};
             if (is_array($addr)) {
                 foreach ($addr as $next_addr) {
                     if (is_object($next_addr)) {
@@ -506,62 +577,63 @@ class Rfc822Header {
     }
     
     function findAddress($address, $recurs = false) {
-       $result = false;
+        $result = false;
         if (is_array($address)) {
-           $i=0;
+            $i=0;
             foreach($address as $argument) {
                 $match = $this->findAddress($argument, true);
-               $last = end($match);
-               if ($match[1]) {
-                   return $i;
-               } else {
-                   if (count($match[0]) && !$result) {
-                       $result = $i;
-                   }
-               }
-               ++$i;   
+                $last = end($match);
+                if ($match[1]) {
+                    return $i;
+                } else {
+                    if (count($match[0]) && !$result) {
+                        $result = $i;
+                    }
+                }
+                ++$i;        
             }
-       } else {
-           if (!is_array($this->cc)) $this->cc = array();
-           $srch_addr = $this->parseAddress($address);
-           $results = array();
-           foreach ($this->to as $to) {
-               if ($to->host == $srch_addr->host) {
-                   if ($to->mailbox == $srch_addr->mailbox) {
-                       $results[] = $srch_addr;
-                       if ($to->personal == $srch_addr->personal) {
-                           if ($recurs) {
-                               return array($results, true);
-                           } else {
-                               return true;
-                           }
-                       }
-                   }
-               }
-           }
-           foreach ($this->cc as $cc) {
-               if ($cc->host == $srch_addr->host) {
-                   if ($cc->mailbox == $srch_addr->mailbox) {
-                       $results[] = $srch_addr;
-                       if ($cc->personal == $srch_addr->personal) {
-                           if ($recurs) {
-                               return array($results, true);
-                           } else {
-                               return true;
-                           }
-                       }
-                   }
-               }
-           }
-           if ($recurs) {
-               return array($results, false);
-           } elseif (count($result)) {
-               return true;
-           } else {
-               return false;
-           }   
-       }
-       return $result;
+        } else {
+            if (!is_array($this->cc)) $this->cc = array();
+            $srch_addr = $this->parseAddress($address);
+            $results = array();
+            foreach ($this->to as $to) {
+                if ($to->host == $srch_addr->host) {
+                    if ($to->mailbox == $srch_addr->mailbox) {
+                        $results[] = $srch_addr;
+                        if ($to->personal == $srch_addr->personal) {
+                            if ($recurs) {
+                                return array($results, true);
+                            } else {
+                                return true;
+                            }
+                        }
+                    }
+                }
+            }
+             foreach ($this->cc as $cc) {
+                if ($cc->host == $srch_addr->host) {
+                    if ($cc->mailbox == $srch_addr->mailbox) {
+                        $results[] = $srch_addr;
+                        if ($cc->personal == $srch_addr->personal) {
+                            if ($recurs) {
+                                return array($results, true);
+                            } else {
+                                return true;
+                            }
+                        }
+                    }
+                }
+            }
+            if ($recurs) {
+                return array($results, false);
+            } elseif (count($result)) {
+                return true;
+            } else {
+                return false;
+            }        
+        }
+        //exit;
+        return $result;
     }
 
     function getContentType($type0, $type1) {