Refactored to use mailparse instead of Mail_mimeDecode. This solves the multipart/alternative bug (Mail_mimeDecode produced no output for such parts!).

git-svn-id: https://vassalengine.svn.sourceforge.net/svnroot/vassalengine/site-src/trunk@7441 67b53d14-2c14-4ace-a08f-0dab2b34000c

3 files changed, 136 insertions, 84 deletions

diff --git a/src/EmailMessage.php b/src/EmailMessage.php
index 69b7ab0..22f012e 100644
--- a/src/EmailMessage.php
+++ b/src/EmailMessage.php
@@ -20,137 +20,189 @@
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 #
 
-# FIXME: maybe use Mailparse instead of Mail_mimeDecode
-
-require_once('Mail/mimeDecode.php');
-require_once('Mail/RFC822.php');
-
 require_once(__DIR__ . '/Message.php');
 
 abstract class EmailMessage implements Message {
 
+  protected $data;
   protected $msg;
-
+  protected $parts;
+  
   public function __construct($input) {
-    $this->msg = self::decode_raw_message($input);
+    $this->data = &$input;
+
+    # build the message structure
+    $this->msg = mailparse_msg_create();
+    mailparse_msg_parse($this->msg, $this->data);
+
+    # get the part data
+    $this->parts = array();
+    foreach (mailparse_msg_get_structure($this->msg) as $part_id) {
+      $part = mailparse_msg_get_part($this->msg, $part_id);
+      $this->parts[$part_id] = mailparse_msg_get_part_data($part);
+    }
+  }
+
+  public function __destruct() {
+    mailparse_msg_free($this->msg);
   }
 
   public function getPostId() {
     return null;
   }
 
+  protected function getHeader($name) {
+    return $this->getPartHeader($this->parts[1], $name);
+  }
+
+  protected function getPartHeaders(&$part_data) {
+    $headers = &$part_data['headers'];
+    return $headers;
+  }
+
+  protected function getPartHeader(&$part_data, $name) {
+    $headers = $this->getPartHeaders($part_data);
+    return array_key_exists($name, $headers) ? $headers[$name] : false;
+  }
+
   public function getFrom() {
-    return self::parse_addr($this->msg->headers['from']);
+    $from = mailparse_rfc822_parse_addresses($this->getHeader('from'));
+    return $from[0]['address'];
   }
 
   public function getSubject() {
-    return $this->msg->headers['subject'];
+    return $this->getHeader('subject');
   }
   
   public function getMessageId() {
-    return $this->msg->headers['message-id'];
+    return $this->getHeader('message-id');
   }
 
   public function getInReplyTo() {
-    return $this->msg->headers['in-reply-to'];
+    return $this->getHeader('in-reply-to');
   }
 
   public function getReferences() {
-    return $this->msg->headers['references'];
+    return $this->getHeader('references');
   }
 
   public function getParts() {
-    return $this->msg;
+  }
+
+  protected function getPartBody(&$part_data) {
+    $beg = $part_data['starting-pos-body'];
+    $end = $part_data['ending-pos-body'];
+    return substr($this->data, $beg, $end-$beg);
+  }
+
+  protected function decode($str, $encoding) {
+    if ($encoding == 'base64') {
+      return base64_decode($str);
+    }
+    else if ($encoding == 'quoted-printable') {
+      return quoted_printable_decode($str);
+    }
+    else {
+      return $str;
+    }
   }
 
   public function getFlattenedParts() {
     $text = '';
     $attachments = array();
-    self::flatten_parts($this->msg, $text, $attachments);
+    $this->flatten_parts('1', $this->parts[1], $text, $attachments);
     return array($text, $attachments);
   }
 
-  protected static function decode_raw_message($input) {
-    $params['include_bodies'] = true;
-    $params['decode_bodies']  = true;
-    $params['decode_headers'] = true;
-    $params['input']          = $input;
-    $params['crlf']           = "\r\n";
-
-    $msg = Mail_mimeDecode::decode($params);
-
-    if (count($msg->headers) == 1 && array_key_exists(null, $msg->headers)) {
-      # An empty message has one null header.
-      throw new Exception('No message');
+  protected function flatten_subparts($part_id, &$text, &$attachments) {
+    for ($i = 1, $child_id = "$part_id.$i";
+         array_key_exists($child_id, $this->parts);
+         ++$i, $child_id = "$part_id.$i")
+    {
+      $child = $this->parts[$child_id];
+      $this->flatten_parts($child_id, $child, $text, $attachments);
     }
-
-    return $msg;
   }
 
-  protected static function parse_addr($s) {
-    $addr = Mail_RFC822::parseAddressList($s);
-    return strtolower($addr[0]->mailbox . '@' . $addr[0]->host);
-  }
+  protected function flatten_parts($part_id, &$part_data,
+                                   &$text, &$attachments) {
+    $type = $part_data['content-type'];
+    list($major, $minor) = split('/', $type, 2);
 
-  protected static function flatten_parts($part, &$text, &$attachments) {
-    switch ($part->ctype_primary) {
+    switch ($major) {
     case 'multipart':
-      if (!isset($part->parts)) {
-        throw new Exception('multipart without parts!');
-      }
+      switch ($minor) {
+      case 'alternative':
+        # check alternatives for text/plain
+        $plain = false;
+
+        for ($i = 1, $child_id = "$part_id.$i";
+             array_key_exists($child_id, $this->parts);
+             ++$i, $child_id = "$part_id.$i")
+        {
+          $child = $this->parts[$child_id];
+
+          $ctype = $child['content-type'];
+          if ($ctype == 'text/plain') {
+            # keep text/plain, chuck the rest
+            $this->flatten_parts($child_id, $child, $text, $attachments);
+            $plain = true;
+            break;
+          }
+        }
 
-      foreach ($part->parts as $subpart) {
-        self::flatten_parts($subpart, $text, $attachments);
-      }
+        if (!$plain) {
+          # no text/plain, handle the subparts as attachments
+          $this->flatten_subparts($part_id, $text, $attachments);
+        }
+        break;
+
+      case 'mixed':
+      default:
+        # handle all subparts
+        $this->flatten_subparts($part_id, $text, $attachments);
+        break;
+      }   
       break;
 
     case 'text':
-      # text/* parts go into the message body.
-      if (!isset($part->body)) {
-        throw new Exception('text without body!');
+      # NB: We don't worry about text/html here because Mailman will have
+      # already stripped it.
+
+      # Text is appended to the main text.
+      $enc = $this->getPartHeader($part_data, 'content-transfer-encoding');
+      $body = $this->getPartBody($part_data);
+      $body = $this->decode($body, $enc);
+
+      $charset = $part_data['content-charset'];
+      if (strtoupper($charset) != 'UTF-8') {
+        if (mb_check_encoding($body, $charset)) {
+          $body = mb_convert_encoding($body, 'UTF-8', $charset);
+        }
+        else {
+          $body = mb_convert_encoding($body, 'UTF-8');
+        }
       }
 
-      $text .= $part->body; 
+      $text .= "$body\n";
       break;
 
     default:
       # Everything else goes into phpBB as an attachment.
-      if (!isset($part->body)) {
-        throw new Exception('attachment without body!');
-      }
-
-      # try to find a filename
-      $filename = '';
-      if (isset($part->d_parameters)) {
-        if (array_key_exists('filename', $part->d_parameters)) {
-          $filename = $part->d_parameters['filename'];
-        }
-        else if (array_key_exists('name', $part->d_parameters)) {
-          $filename = $part->d_parameters['name'];
-        }
+      $enc = $this->getPartHeader($part_data, 'content-transfer-encoding');
+      $data = $this->getPartBody($part_data);
+      $data = $this->decode($data, $enc);    
+
+echo "$major\n";
+      $disp = $part_data['content-disposition'];
+      if ($disp == 'attachment' || $disp == 'inline') {
+        $attachments[] = array( 
+          'filename' => $part_data['disposition-filename'],
+          'mimetype' => $part_data['content-type'],
+          'comment'  => $part_data['content-description'],
+          'data'     => $data
+        );
       }
-
-      if ($filename == '') {
-        if (isset($part->ctype_parameters)) {
-          if (array_key_exists('name', $part->ctype_parameters)) {
-            $filename = $part->d_parameters['name'];
-          }
-        }
-      }
-
-      $mimetype = $part->ctype_primary . '/' . $part->ctype_secondary;
-
-      $comment = array_key_exists('content-description', $part->headers) ?
-        $part->headers['content-description'] : '';
-
-      $params = array(
-        'filename' => $filename,
-        'mimetype' => $mimetype,
-        'comment'  => $comment,
-        'data'     => $part->body
-      );
-
-      $attachments[] = $params;
     }
   }
 }
diff --git a/src/MailmanMessage.php b/src/MailmanMessage.php
index dfd100e..6ec35c6 100644
--- a/src/MailmanMessage.php
+++ b/src/MailmanMessage.php
@@ -28,9 +28,11 @@ class MailmanMessage extends EmailMessage {
   }
 
   public function getSource() {
-    return self::parse_addr(
-      substr_replace($this->msg->headers['list-post'], '', 1, 7)
-    );
+    # remove 'mailto:'
+    $lp = substr_replace($this->getHeader('list-post'), '', 1, 7);
+
+    $src = mailparse_rfc822_parse_addresses($lp);
+    return $src[0]['address'];
   }
 }
 
diff --git a/src/Message.php b/src/Message.php
index 9bb082d..00e956a 100644
--- a/src/Message.php
+++ b/src/Message.php
@@ -34,8 +34,6 @@ interface Message {
   public function getInReplyTo();
 
   public function getReferences();
-
-  public function getParts();
 }
 
 ?>