From 04821b1701a420185cbd672e4f4727eb402dbbce Mon Sep 17 00:00:00 2001
From: Gareth Rees <gareth@garethrees.co.uk>
Date: Mon, 24 Jul 2023 10:06:13 +0100
Subject: [PATCH] Exclude email providers from body home pages

Many smaller authorities use a request email address from a general
purpose public email provider. We don't want e.g. gmail.com being set as
the home page for the authority.

This is configurable in the theme by setting the class attribute:

    # Add to the defaults
    PublicBody.excluded_calculated_home_page_domains << %w[
      example.net
    ]

    # Clear the defaults and set your own list
    PublicBody.excluded_calculated_home_page_domains = %w[
      example.org
      example.net
    ]

Fixes https://github.com/mysociety/alaveteli/issues/6434
---
 .../public_body/calculated_home_page.rb       | 29 +++++++++++++++++++
 spec/models/public_body_spec.rb               | 16 ++++++++++
 2 files changed, 45 insertions(+)

diff --git a/app/models/public_body/calculated_home_page.rb b/app/models/public_body/calculated_home_page.rb
index 87495b1ab6..7efbc73b29 100644
--- a/app/models/public_body/calculated_home_page.rb
+++ b/app/models/public_body/calculated_home_page.rb
@@ -1,5 +1,27 @@
 # Guess the home page based on the request email domain.
 module PublicBody::CalculatedHomePage
+  extend ActiveSupport::Concern
+
+  included do
+    cattr_accessor :excluded_calculated_home_page_domains, default: %w[
+      aol.com
+      btinternet.com
+      gmail.com
+      gmx.com
+      hotmail.com
+      icloud.com
+      mail.com
+      mail.ru
+      outlook.com
+      protonmail.com
+      qq.com
+      yahoo.com
+      yandex.com
+      ymail.com
+      zoho.com
+    ]
+  end
+
   # Guess home page from the request email, or use explicit override, or nil
   # if not known.
   #
@@ -9,7 +31,14 @@ def calculated_home_page
     if home_page && !home_page.empty?
       home_page[URI.regexp(%w(http https))] ? home_page : "http://#{home_page}"
     elsif request_email_domain
+      return if excluded_calculated_home_page_domain?(request_email_domain)
       "http://www.#{request_email_domain}"
     end
   end
+
+  private
+
+  def excluded_calculated_home_page_domain?(domain)
+    excluded_calculated_home_page_domains.include?(domain)
+  end
 end
diff --git a/spec/models/public_body_spec.rb b/spec/models/public_body_spec.rb
index 5c6a0f8fd6..39190aac61 100644
--- a/spec/models/public_body_spec.rb
+++ b/spec/models/public_body_spec.rb
@@ -1905,6 +1905,12 @@ def set_default_attributes(public_body)
 end
 
 RSpec.describe PublicBody do
+  around do |example|
+    previous = PublicBody.excluded_calculated_home_page_domains
+    PublicBody.excluded_calculated_home_page_domains = %w[example.net]
+    example.run
+    PublicBody.excluded_calculated_home_page_domains = previous
+  end
 
   describe 'calculated home page' do
     it "returns the home page verbatim if it's present" do
@@ -1943,6 +1949,16 @@ def set_default_attributes(public_body)
       public_body = PublicBody.new(home_page: 'https://example.com')
       expect(public_body.calculated_home_page).to eq('https://example.com')
     end
+
+    it 'does not calculate the homepage for excluded domains' do
+      public_body = PublicBody.new(request_email: 'x@example.net')
+      expect(public_body.calculated_home_page).to be_nil
+    end
+
+    it 'ignores case sensitivity for excluded domains' do
+      public_body = PublicBody.new(request_email: 'x@EXAMPLE.net')
+      expect(public_body.calculated_home_page).to be_nil
+    end
   end
 
   describe '#site_administration?' do