From 5a16c4cb8d283903e09cb6d6dff65ecc11879add Mon Sep 17 00:00:00 2001
From: Luca Spiller <luca@blissfulsystems.com>
Date: Mon, 11 Jul 2016 15:52:31 +0200
Subject: [PATCH] Strip attachments from emails when converting

---
 lib/heathen/processor_methods/rfc822totext.rb | 12 +++++++
 lib/heathen/task.rb                           |  6 ++++
 spec/fixtures/heathen/email.eml               | 34 +++++++++++++++++++
 .../processor_methods/rfc822totext_spec.rb    | 23 +++++++++++++
 4 files changed, 75 insertions(+)
 create mode 100644 lib/heathen/processor_methods/rfc822totext.rb
 create mode 100644 spec/fixtures/heathen/email.eml
 create mode 100644 spec/heathen/processor_methods/rfc822totext_spec.rb

diff --git a/lib/heathen/processor_methods/rfc822totext.rb b/lib/heathen/processor_methods/rfc822totext.rb
new file mode 100644
index 0000000..af35c05
--- /dev/null
+++ b/lib/heathen/processor_methods/rfc822totext.rb
@@ -0,0 +1,12 @@
+require 'mail'
+
+module Heathen
+  class Processor
+    def rfc822totext
+      expect_mime_type 'message/rfc822'
+
+      mail = Mail.read(job.content_file).without_attachments!
+      job.content = mail.to_s
+    end
+  end
+end
diff --git a/lib/heathen/task.rb b/lib/heathen/task.rb
index 872d013..da47a05 100644
--- a/lib/heathen/task.rb
+++ b/lib/heathen/task.rb
@@ -71,6 +71,10 @@ def task_key action, mime_type
       perform_task 'ocr'
     when %r[text/html]
       wkhtmltopdf '-d 100 --encoding UTF-8'
+    when %r[message/rfc822]
+      rfc822totext
+      job.reset_content_file!
+      libreoffice format: 'pdf'
     else
       libreoffice format: 'pdf'
   end
@@ -92,6 +96,8 @@ def task_key action, mime_type
       htmltotext
     when %r[application/pdf]
       pdftotext
+    when %r[message/rfc822]
+      rfc822totext
     else
       libreoffice format: 'txt'
   end
diff --git a/spec/fixtures/heathen/email.eml b/spec/fixtures/heathen/email.eml
new file mode 100644
index 0000000..6531f03
--- /dev/null
+++ b/spec/fixtures/heathen/email.eml
@@ -0,0 +1,34 @@
+                                                                                                                                                                                                                                                               
+MIME-Version: 1.0
+Received: by 10.79.89.194 with HTTP; Mon, 11 Jul 2016 06:41:56 -0700 (PDT)
+Date: Mon, 11 Jul 2016 16:41:56 +0300
+Delivered-To: employee@example.com
+Message-ID: <CABZS92tgEpMBnoPybJbsreEFJXbv0H1aP30NVUJ6F2Q9b5_BFw@mail.gmail.com>
+Subject: Important Reports
+From: Boss <boss@example.com>
+To: Employee <employee@example.com>
+Content-Type: multipart/mixed; boundary=001a114a86e6d9bf9605375c50db
+
+--001a114a86e6d9bf9605375c50db
+Content-Type: multipart/alternative; boundary=001a114a86e6d9bf7a05375c50d9
+
+--001a114a86e6d9bf7a05375c50d9
+Content-Type: text/plain; charset=UTF-8
+
+Attached are the important reports.
+
+--001a114a86e6d9bf7a05375c50d9
+Content-Type: text/html; charset=UTF-8
+
+<div dir="ltr">Attached are the important reports.
+</div>
+
+--001a114a86e6d9bf7a05375c50d9--
+--001a114a86e6d9bf9605375c50db
+Content-Type: text/csv; charset=US-ASCII; name="reports.csv"
+Content-Disposition: attachment; filename="reports.csv"
+Content-Transfer-Encoding: base64
+X-Attachment-Id: f_iqi2vbfj0
+
+b25lLHR3byx0aHJlZQoxLDIsMwo=
+--001a114a86e6d9bf9605375c50db--
diff --git a/spec/heathen/processor_methods/rfc822totext_spec.rb b/spec/heathen/processor_methods/rfc822totext_spec.rb
new file mode 100644
index 0000000..918634a
--- /dev/null
+++ b/spec/heathen/processor_methods/rfc822totext_spec.rb
@@ -0,0 +1,23 @@
+require 'spec_helper'
+
+describe Heathen::Processor do
+  let(:content) { File.read(fixture('heathen/email.eml')) }
+  let(:job) { Heathen::Job.new 'foo', content, 'en' }
+  let(:processor) { described_class.new job: job, logger: Logger.new($stderr) }
+
+  before do
+    allow(content).to receive(:mime_type).and_return('message/rfc822')
+  end
+
+  after do
+    processor.clean_up
+  end
+
+  context '#rfc822totext' do
+    it 'strips attachments' do
+      processor.rfc822totext
+      expect(job.content.mime_type).to eq 'message/rfc822; charset=us-ascii'
+      expect(job.content).to_not include('reports.csv')
+    end
+  end
+end