From 3448781c40f0e45a4e7e19ab9f1e601df8485b13 Mon Sep 17 00:00:00 2001
From: Bob Trahan <bob.trahan@gmail.com>
Date: Fri, 04 Jan 2013 01:04:30 +0000
Subject: [PATCH] de-duplicate emails received by phabricator multiple times

Summary: this can happen if you have Phabricator and email lists co-mingling such that Phabricator receives an email multiple times. we can prevent this from then spamming everyone or otherwise taking the action multiple times by storing a message id hash and dropping the message if we have more than one message that matches.

Test Plan: simulated sending the same email multiple times on the command line. noted only the first one made it through.

Reviewers: epriestley

Reviewed By: epriestley

CC: aran, Korvin

Maniphest Tasks: T1726

Differential Revision: https://secure.phabricator.com/D4328
---
 scripts/mail/mail_handler.php | 3 +++
 resources/sql/patches/20130102.metamtareceivedmailmessageidhash.sql | 3 +++
 src/applications/metamta/storage/PhabricatorMetaMTAReceivedMail.php | 22 ++++++++++++++++++++++
 src/infrastructure/storage/patch/PhabricatorBuiltinPatchList.php | 5 +++++
 4 file(s) changed, 33 insertion(s)(+), 0 deletion(s)(-)

diff --git a/scripts/mail/mail_handler.php b/scripts/mail/mail_handler.php
--- a/scripts/mail/mail_handler.php
+++ b/scripts/mail/mail_handler.php
@@ -34,6 +34,9 @@
   'text' => $text_body,
   'html' => $parser->getMessageBody('html'),
 ));
+$received->setMessageIDHash(
+  PhabricatorHash::digestForIndex($received->getMessageID())
+);
 
 $attachments = array();
 foreach ($parser->getAttachments() as $attachment) {
diff --git a/resources/sql/patches/20130102.metamtareceivedmailmessageidhash.sql b/resources/sql/patches/20130102.metamtareceivedmailmessageidhash.sql
new file mode 100644
--- /dev/null
+++ b/resources/sql/patches/20130102.metamtareceivedmailmessageidhash.sql
@@ -0,0 +1,3 @@
+ALTER TABLE `{$NAMESPACE}_metamta`.`metamta_receivedmail`
+  ADD `messageIDHash` CHAR(12) BINARY NOT NULL,
+  ADD KEY `key_messageIDHash` (`messageIDHash`);
diff --git a/src/applications/metamta/storage/PhabricatorMetaMTAReceivedMail.php b/src/applications/metamta/storage/PhabricatorMetaMTAReceivedMail.php
--- a/src/applications/metamta/storage/PhabricatorMetaMTAReceivedMail.php
+++ b/src/applications/metamta/storage/PhabricatorMetaMTAReceivedMail.php
@@ -9,6 +9,7 @@
   protected $relatedPHID;
   protected $authorPHID;
   protected $message;
+  protected $messageIDHash;
 
   public function getConfiguration() {
     return array(
@@ -143,6 +144,27 @@
       $message = "Ignoring email with 'X-Phabricator-Sent-This-Message' ".
                  "header to avoid loops.";
       return $this->setMessage($message)->save();
+    }
+
+    $message_id_hash = $this->getMessageIDHash();
+    if ($message_id_hash) {
+      $messages = $this->loadAllWhere(
+        'messageIDHash = %s',
+        $message_id_hash
+      );
+      $messages_count = count($messages);
+      if ($messages_count > 1) {
+        $first_message = reset($messages);
+        if ($first_message->getID() != $this->getID()) {
+          $message = sprintf(
+            'Ignoring email with message id hash "%s" that has been seen %d '.
+            'times, including this message.',
+            $message_id_hash,
+            $messages_count
+          );
+          return $this->setMessage($message)->save();
+        }
+      }
     }
 
     list($to,
diff --git a/src/infrastructure/storage/patch/PhabricatorBuiltinPatchList.php b/src/infrastructure/storage/patch/PhabricatorBuiltinPatchList.php
--- a/src/infrastructure/storage/patch/PhabricatorBuiltinPatchList.php
+++ b/src/infrastructure/storage/patch/PhabricatorBuiltinPatchList.php
@@ -1076,6 +1076,11 @@
         'type'    => 'sql',
         'name'    => $this->getPatchPath('20130101.confxaction.sql'),
       ),
+      '20130102.metamtareceivedmailmessageidhash.sql' => array(
+        'type'    => 'sql',
+        'name'    =>
+          $this->getPatchPath('20130102.metamtareceivedmailmessageidhash.sql'),
+      ),
     );
   }
 

--
tangled.sh