Merge branch 'bugfix/html-email-brackets' into 'master'

unwrap links without an href

Closes #27645

See merge request !9045

changelogs/unreleased/27645-html-email-brackets-bug.yml

@@ -0,0 +1,4 @@
+---
+title: Fix an email parsing bug where brackets would be inserted in emails from some Outlook clients
+merge_request: 9045
+author: jneen

lib/gitlab/email/html_parser.rb

@@ -17,6 +17,13 @@ module Gitlab
       def filter_replies!
         document.xpath('//blockquote').each(&:remove)
         document.xpath('//table').each(&:remove)
+
+        # bogus links with no href are sometimes added by outlook,
+        # and can result in Html2Text adding extra square brackets
+        # to the text, so we unwrap them here.
+        document.xpath('//a[not(@href)]').each do |link|
+          link.replace(link.children)
+        end
       end
 
       def filtered_html

spec/fixtures/emails/html_empty_link.eml

@@ -0,0 +1,26 @@
+
+MIME-Version: 1.0
+Received: by 10.25.161.144 with HTTP; Tue, 7 Oct 2014 22:17:17 -0700 (PDT)
+X-Originating-IP: [117.207.85.84]
+In-Reply-To: <5434c8b52bb3a_623ff09fec70f049749@discourse-app.mail>
+References: <topic/35@discourse.techapj.com>
+  <5434c8b52bb3a_623ff09fec70f049749@discourse-app.mail>
+Date: Wed, 8 Oct 2014 10:47:17 +0530
+Delivered-To: arpit@techapj.com
+Message-ID: <CAOJeqne=SJ_LwN4sb-0Y95ejc2OpreVhdmcPn0TnmwSvTCYzzQ@mail.gmail.com>
+Subject: Re: [Discourse] [Meta] Welcome to techAPJ's Discourse!
+From: Arpit Jalan <arpit@techapj.com>
+To: Discourse <mail+e1c7f2a380e33840aeb654f075490bad@arpitjalan.com>Accept-Language: en-US
+Content-Language: en-US
+X-MS-Has-Attach:
+X-MS-TNEF-Correlator:
+x-originating-ip: [134.68.31.227]
+Content-Type: multipart/alternative;
+        boundary="_000_B0DFE1BEB3739743BC9B639D0E6BC8FF217A6341IUMSSGMBX104ads_"
+MIME-Version: 1.0
+
+--_000_B0DFE1BEB3739743BC9B639D0E6BC8FF217A6341IUMSSGMBX104ads_
+Content-Type: text/html; charset="utf-8"
+
+<a name="_MailEndCompose">no brackets!</a>
+--_000_B0DFE1BEB3739743BC9B639D0E6BC8FF217A6341IUMSSGMBX104ads_--

spec/lib/gitlab/email/reply_parser_spec.rb

@@ -208,5 +208,9 @@ describe Gitlab::Email::ReplyParser, lib: true do
     it "properly renders html-only email from MS Outlook" do
       expect(test_parse_body(fixture_file("emails/outlook_html.eml"))).to eq("Microsoft Outlook 2010")
     end
+
+    it "does not wrap links with no href in unnecessary brackets" do
+      expect(test_parse_body(fixture_file("emails/html_empty_link.eml"))).to eq("no brackets!")
+    end
   end
 end