Merge pull request #3918 from techAPJ/better-email-parsing

FEATURE: better email reply parsing

Gemfile

@@ -47,7 +47,7 @@ gem 'aws-sdk', require: false
 gem 'excon', require: false
 gem 'unf', require: false
 
-gem 'email_reply_parser'
+gem 'discourse_email_parser'
 
 # note: for image_optim to correctly work you need to follow
 # https://github.com/toy/image_optim

Gemfile.lock

@@ -113,9 +113,9 @@ GEM
     diff-lcs (1.2.5)
     discourse-qunit-rails (0.0.8)
       railties
+    discourse_email_parser (0.6.1)
     docile (1.1.5)
     dotenv (2.0.2)
-    email_reply_parser (0.5.8)
     ember-data-source (1.0.0.beta.16.1)
       ember-source (~> 1.8)
     ember-handlebars-template (0.1.5)
@@ -451,7 +451,7 @@ DEPENDENCIES
   byebug
   certified
   discourse-qunit-rails
-  email_reply_parser
+  discourse_email_parser
   ember-rails
   ember-source (= 1.12.1)
   excon

lib/email/receiver.rb

@@ -140,7 +140,7 @@ module Email
       body = discourse_email_trimmer body
       raise EmptyEmailError if body.strip.blank?
 
-      body = EmailReplyParser.parse_reply body
+      body = DiscourseEmailParser.parse_reply body
       raise EmptyEmailError if body.strip.blank?
 
       body.force_encoding(encoding).encode("UTF-8")

spec/components/email/receiver_spec.rb

@@ -173,6 +173,14 @@ the lazy dog. The quick brown fox jumps over the lazy dog. The quick brown"
       expect(test_parse_body(fixture_file("emails/iphone_signature.eml"))).not_to match(/Sent from my iPhone/)
     end
 
+    it "strips regular signature" do
+      expect(test_parse_body(fixture_file("emails/signature.eml"))).not_to match(/Arpit/)
+    end
+
+    it "strips 'original message' context" do
+      expect(test_parse_body(fixture_file("emails/original_message_context.eml"))).not_to match(/Context/)
+    end
+
     it "properly renders email reply from gmail web client" do
       expect(test_parse_body(fixture_file("emails/gmail_web.eml"))).
           to eq(

spec/fixtures/emails/original_message_context.eml

@@ -0,0 +1,30 @@
+Delivered-To: test@mail.com
+Return-Path: <walter.white@googlemail.com>
+From: Walter White <walter.white@googlemail.com>
+Content-Type: multipart/alternative;
+  boundary=Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975
+Content-Transfer-Encoding: 7bit
+Mime-Version: 1.0 (1.0)
+Subject: Re: Signature in email replies!
+Date: Thu, 23 Oct 2014 14:43:49 +0530
+References: <1234@mail.gmail.com>
+In-Reply-To: <1234@mail.gmail.com>
+To: Arpit Jalan <test@mail.com>
+X-Mailer: iPhone Mail (12A405)
+
+
+--Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975
+Content-Type: text/plain;
+  charset=us-ascii
+Content-Transfer-Encoding: 7bit
+
+This post should not include signature.
+----Original Message----
+
+Context here.
+
+> On 23-Oct-2014, at 9:45 am, Arpit Jalan <test@mail.com> wrote:
+>
+> Signature in email replies!
+
+--Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975

spec/fixtures/emails/signature.eml

@@ -0,0 +1,29 @@
+Delivered-To: test@mail.com
+Return-Path: <walter.white@googlemail.com>
+From: Walter White <walter.white@googlemail.com>
+Content-Type: multipart/alternative;
+  boundary=Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975
+Content-Transfer-Encoding: 7bit
+Mime-Version: 1.0 (1.0)
+Subject: Re: Signature in email replies!
+Date: Thu, 23 Oct 2014 14:43:49 +0530
+References: <1234@mail.gmail.com>
+In-Reply-To: <1234@mail.gmail.com>
+To: Arpit Jalan <test@mail.com>
+X-Mailer: iPhone Mail (12A405)
+
+
+--Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975
+Content-Type: text/plain;
+  charset=us-ascii
+Content-Transfer-Encoding: 7bit
+
+This post should not include signature.
+
+----Arpit
+
+> On 23-Oct-2014, at 9:45 am, Arpit Jalan <test@mail.com> wrote:
+>
+> Signature in email replies!
+
+--Apple-Mail-8E182EEF-9DBC-41DE-A593-DF2E5EBD3975