From f69793184e0bb02fd44925113238937124f66a0e Mon Sep 17 00:00:00 2001
From: epriestley <git@epriestley.com>
Date: Fri, 6 Dec 2013 15:47:37 -0800
Subject: [PATCH] Fix over-matching of quoted text for message bodies beginning
 with "On..."

Summary:
A user sent a message to Phabricator which looked like:

  On blah blah blah ?

  On <date>, <user> wrote:
  > blah blah blah

The current algorithm is too aggressive and thinks lines 1-3 are //all// the "On ... wrote:" string. Instead, patch only the most recent "On".

Test Plan: Added a failing test and made it pass.

Reviewers: btrahan, zeeg

Reviewed By: zeeg

CC: aran

Differential Revision: https://secure.phabricator.com/D7732
---
 .../PhabricatorMetaMTAEmailBodyParser.php     | 27 ++++++++++++++++---
 ...bricatorMetaMTAEmailBodyParserTestCase.php | 14 ++++++++++
 2 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/src/applications/metamta/parser/PhabricatorMetaMTAEmailBodyParser.php b/src/applications/metamta/parser/PhabricatorMetaMTAEmailBodyParser.php
index 3218695cc5..41a237aca1 100644
--- a/src/applications/metamta/parser/PhabricatorMetaMTAEmailBodyParser.php
+++ b/src/applications/metamta/parser/PhabricatorMetaMTAEmailBodyParser.php
@@ -50,10 +50,29 @@ final class PhabricatorMetaMTAEmailBodyParser {
   }
 
   private function stripQuotedText($body) {
-    $body = preg_replace(
-      '/^\s*>?\s*On\b.*\bwrote:.*?/msU',
-      '',
-      $body);
+
+    // Look for "On <date>, <user> wrote:". This may be split across multiple
+    // lines. We need to be careful not to remove all of a message like this:
+    //
+    //   On which day do you want to meet?
+    //
+    //   On <date>, <user> wrote:
+    //   > Let's set up a meeting.
+
+    $start = null;
+    $lines = phutil_split_lines($body);
+    foreach ($lines as $key => $line) {
+      if (preg_match('/^\s*>?\s*On\b/', $line)) {
+        $start = $key;
+      }
+      if ($start !== null) {
+        if (preg_match('/\bwrote:/', $line)) {
+          $lines = array_slice($lines, 0, $start);
+          $body = implode('', $lines);
+          break;
+        }
+      }
+    }
 
     // Outlook english
     $body = preg_replace(
diff --git a/src/applications/metamta/parser/__tests__/PhabricatorMetaMTAEmailBodyParserTestCase.php b/src/applications/metamta/parser/__tests__/PhabricatorMetaMTAEmailBodyParserTestCase.php
index d4361c6e16..86cc547b4e 100644
--- a/src/applications/metamta/parser/__tests__/PhabricatorMetaMTAEmailBodyParserTestCase.php
+++ b/src/applications/metamta/parser/__tests__/PhabricatorMetaMTAEmailBodyParserTestCase.php
@@ -31,6 +31,20 @@ final class PhabricatorMetaMTAEmailBodyParserTestCase
     }
   }
 
+  public function testFalsePositiveForOnWrote() {
+    $body = <<<EOEMAIL
+On which horse shall you ride?
+
+On Sep 23, alincoln wrote:
+
+> Hey bro do you want to go ride horses tomorrow?
+EOEMAIL;
+
+    $parser = new PhabricatorMetaMTAEmailBodyParser();
+    $stripped = $parser->stripTextBody($body);
+    $this->assertEqual("On which horse shall you ride?", $stripped);
+  }
+
   private function getEmailBodiesWithFullCommands() {
     $bodies = $this->getEmailBodies();
     $with_commands = array();