From 9705c249ac9da3aeb9f5bd86aeaf3a98a7ea4f2d Mon Sep 17 00:00:00 2001
From: Matt-Yorkley <9029026+Matt-Yorkley@users.noreply.github.com>
Date: Fri, 15 Feb 2019 14:59:15 +0000
Subject: [PATCH] Sanitize no-parsable BOM characters inserted by Excel

---
 .../admin/product_import_controller.rb        |  3 ++-
 app/services/upload_sanitizer.rb              | 18 +++++++++++++++++
 spec/services/upload_sanitizer_spec.rb        | 20 +++++++++++++++++++
 3 files changed, 40 insertions(+), 1 deletion(-)
 create mode 100644 app/services/upload_sanitizer.rb
 create mode 100644 spec/services/upload_sanitizer_spec.rb

diff --git a/app/controllers/admin/product_import_controller.rb b/app/controllers/admin/product_import_controller.rb
index 201d57033f..72a5a63f30 100644
--- a/app/controllers/admin/product_import_controller.rb
+++ b/app/controllers/admin/product_import_controller.rb
@@ -84,7 +84,8 @@ module Admin
       directory = 'tmp/product_import'
       Dir.mkdir(directory) unless File.exist?(directory)
       File.open(Rails.root.join(directory, filename + extension), 'wb') do |f|
-        f.write(upload.read)
+        data = UploadSanitizer.new(upload.read).call
+        f.write(data)
         f.path
       end
     end
diff --git a/app/services/upload_sanitizer.rb b/app/services/upload_sanitizer.rb
new file mode 100644
index 0000000000..132669c884
--- /dev/null
+++ b/app/services/upload_sanitizer.rb
@@ -0,0 +1,18 @@
+# Formats uploaded files to UTF-8 encoding and strips unexpected BOM characters.
+# Takes an open File object as input
+class UploadSanitizer
+  def initialize(upload)
+    @data = upload
+  end
+
+  def call
+    @data.force_encoding('UTF-8')
+    strip_bom_character
+  end
+
+  private
+
+  def strip_bom_character
+    @data.gsub("\xEF\xBB\xBF".force_encoding("UTF-8"), '')
+  end
+end
diff --git a/spec/services/upload_sanitizer_spec.rb b/spec/services/upload_sanitizer_spec.rb
new file mode 100644
index 0000000000..6d8e6f029c
--- /dev/null
+++ b/spec/services/upload_sanitizer_spec.rb
@@ -0,0 +1,20 @@
+require 'spec_helper'
+
+describe UploadSanitizer do
+  describe "#call" do
+    let(:upload) do
+      File.open("/tmp/unsanitized.csv", 'wb:ascii-8bit') do |f|
+        f << "\xEF\xBB\xBF"
+        f << "Test"
+      end
+    end
+    let(:service) { UploadSanitizer.new(File.open(upload).read) }
+
+    it "sanitizes the uploaded file" do
+      sanitized_upload = service.call
+
+      expect(sanitized_upload.encoding.name).to eq "UTF-8"
+      expect(sanitized_upload.to_s).to eq "Test"
+    end
+  end
+end