Add some basic signature stripping

Summary: See discussion in T789. Covered the obvious cases, at least. We can refine this as we get a larger sample size.

Test Plan: Unit test coverage.

Reviewers: btrahan, vrana, jungejason

Reviewed By: btrahan

CC: aran

Maniphest Tasks: T789

Differential Revision: https://secure.phabricator.com/D2154

Author: epriestley

src/applications/metamta/parser/PhabricatorMetaMTAEmailBodyParser.php

@@ -18,13 +18,11 @@
 
 final class PhabricatorMetaMTAEmailBodyParser {
 
-  public function __construct($corpus) {
-    $this->corpus = $corpus;
+  public function stripTextBody($body) {
+    return $this->stripSignature($this->stripQuotedText($body));
   }
 
-  public function stripQuotedText() {
-    $body = $this->corpus;
-
+  private function stripQuotedText($body) {
     $body = preg_replace(
       '/^\s*On\b.*\bwrote:.*?/msU',
       '',
@@ -42,9 +40,26 @@ public function stripQuotedText() {
       '',
       $body);
 
+    return rtrim($body);
+  }
+
+  private function stripSignature($body) {
+    // Quasi-"standard" delimiter, for lols see:
+    //   https://bugzilla.mozilla.org/show_bug.cgi?id=58406
+    $body = preg_replace(
+      '/^-- +$.*/sm',
+      '',
+      $body);
+
     // HTC Mail application (mobile)
     $body = preg_replace(
-      '/^\s*Sent from my HTC smartphone.*?/msU',
+      '/^\s*^Sent from my HTC smartphone.*/sm',
+      '',
+      $body);
+
+    // Apple iPhone
+    $body = preg_replace(
+      '/^\s*^Sent from my iPhone\s*$.*/sm',
       '',
       $body);
 

src/applications/metamta/parser/__tests__/PhabricatorMetaMTAEmailBodyParserTestCase.php

@@ -22,13 +22,15 @@ final class PhabricatorMetaMTAEmailBodyParserTestCase
   public function testQuotedTextStripping() {
     $bodies = $this->getEmailBodies();
     foreach ($bodies as $body) {
-      $parser = new PhabricatorMetaMTAEmailBodyParser($body);
-      $stripped = $parser->stripQuotedText();
+      $parser = new PhabricatorMetaMTAEmailBodyParser();
+      $stripped = $parser->stripTextBody($body);
       $this->assertEqual("OKAY", $stripped);
     }
   }
 
   private function getEmailBodies() {
+    $trailing_space = ' ';
+
     return array(
 <<<EOEMAIL
 OKAY
@@ -86,6 +88,21 @@ private function getEmailBodies() {
 Subject: Some Text Date: Mon, Apr 2, 2012 1:42 pm
 > ...
 EOEMAIL
+,
+<<<EOEMAIL
+OKAY
+
+--{$trailing_space}
+Abraham Lincoln
+Supreme Galactic Emperor
+EOEMAIL
+,
+<<<EOEMAIL
+OKAY
+
+Sent from my iPhone
+EOEMAIL
+,
     );
   }
 

src/applications/metamta/storage/receivedmail/PhabricatorMetaMTAReceivedMail.php

@@ -205,8 +205,8 @@ public function processReceivedMail() {
   public function getCleanTextBody() {
     $body = idx($this->bodies, 'text');
 
-    $parser = new PhabricatorMetaMTAEmailBodyParser($body);
-    return $parser->stripQuotedText();
+    $parser = new PhabricatorMetaMTAEmailBodyParser();
+    return $parser->stripTextBody($body);
   }
 
   public static function loadReceiverObject($receiver_name) {