From 4b1e543b5b624ea674c59f7081c45a47819b9db0 Mon Sep 17 00:00:00 2001
From: Jason <jason@pemulis>
Date: Wed, 9 Dec 2015 13:01:29 -0600
Subject: [PATCH] Recognize multi-line reply headers when there are quoted ones
 below

---
 lib/email_reply_parser.rb           | 28 ++++++++++++++++++++++------
 test/email_reply_parser_test.rb     |  7 +++++++
 test/emails/broken_reply_header.txt | 29 +++++++++++++++++++++++++++++
 3 files changed, 58 insertions(+), 6 deletions(-)
 create mode 100644 test/emails/broken_reply_header.txt

diff --git a/lib/email_reply_parser.rb b/lib/email_reply_parser.rb
index 982eaec..eb5880a 100644
--- a/lib/email_reply_parser.rb
+++ b/lib/email_reply_parser.rb
@@ -82,12 +82,7 @@ def read(text)
       # Normalize line endings.
       text.gsub!("\r\n", "\n")
 
-      # Check for multi-line reply headers. Some clients break up
-      # the "On DATE, NAME <EMAIL> wrote:" line into multiple lines.
-      if text =~ /^(?!On.*On\s.+?wrote:)(On\s(.+?)wrote:)$/nm
-        # Remove all new lines from the reply header.
-        text.gsub! $1, $1.gsub("\n", " ")
-      end
+      text = repair_broken_reply_header(text)
 
       # Some users may reply directly above a line of underscores.
       # In order to ensure that these fragments are split correctly,
@@ -229,6 +224,27 @@ def finish_fragment
       end
       @fragment = nil
     end
+
+    # Stitch together multi-line reply headers. Some clients break up
+    # the "On DATE, NAME <EMAIL> wrote:" line into multiple lines.
+    #
+    # If the latest reply has a line beginning with 'On', then the match
+    # will start too early and contain another match.
+    # So we repair the inner match via recursion
+    def repair_broken_reply_header(text)
+      pattern = /^(On\s(.+?)wrote:)$/nm
+      if text =~ pattern
+        match = $1
+        tail = match[1..-1]
+        if tail =~ pattern
+	      text.gsub tail, repair_broken_reply_header(tail)
+        else
+          text.gsub match, match.gsub("\n", " ")
+        end
+      else
+        text
+      end
+    end
   end
 
   ### Fragments
diff --git a/test/email_reply_parser_test.rb b/test/email_reply_parser_test.rb
index 6710838..54948b3 100644
--- a/test/email_reply_parser_test.rb
+++ b/test/email_reply_parser_test.rb
@@ -115,6 +115,13 @@ def test_deals_with_multiline_reply_headers
     assert_match /Was this/, reply.fragments[1].to_s
   end
 
+  def test_multiline_and_quoted_reply_headers
+    reply = email :broken_reply_header
+
+    assert_match /^On the other hand/,   reply.fragments[0].to_s
+    assert_match /^On Wed, Dec 9/,   reply.fragments[1].to_s
+  end
+
   def test_deals_with_windows_line_endings
     reply = email :email_1_7
 
diff --git a/test/emails/broken_reply_header.txt b/test/emails/broken_reply_header.txt
new file mode 100644
index 0000000..fadbd30
--- /dev/null
+++ b/test/emails/broken_reply_header.txt
@@ -0,0 +1,29 @@
+This is the latest reply.
+
+The thread contains two previous replies, and the most recent reply header has a line break in it.
+
+If we don't patch the line break, the reply header ends up in THIS fragment, instead of in the quoted fragment.
+
+On the one hand, there could be a line in the reply starting with the word 'On.'
+
+On the other hand, it could happen TWICE!
+
+Stranger
+
+On Wed, Dec 9, Mister Example
+<mrmister@example.com>
+wrote:
+> Hey, Stranger,
+> It's mrmister again, and I hope you reply to me. I also hope this fragment has my reply header in it!
+>
+> -mrmister
+>
+> On Tue, 2011-03-01 at 18:02 +0530, Stranger Jones wrote:
+>> Hey, mrmister,
+>>
+>> I'm getting tired of writing fake email text.
+>>
+>> Stranger
+
+--
+My name is Stranger