3 files changed, 96 insertions, 4 deletions

diff --git a/app/helpers/social_helper.rb b/app/helpers/social_helper.rb
index d6932a2f..e43f5a5c 100644
--- a/app/helpers/social_helper.rb
+++ b/app/helpers/social_helper.rb
@@ -92,11 +92,14 @@ module SocialHelper
     @page&.published_at&.iso8601
   end
 
-  # request.path rather than a routing helper: it is already the locale's
-  # own canonical form -- unprefixed for German, /en/ for English -- and
-  # dropping the query string is what makes it canonical.
+  # The address of the version actually being served, not of the URL that
+  # was requested. So /de/updates/foo points at /updates/foo, and a page
+  # with no English translation requested under /en/ points at the German
+  # URL rather than claiming to be an English page. The query string is
+  # dropped, which is what makes it canonical.
   def og_canonical_url
-    og_absolute_url(request.path)
+    locale = @page&.persisted? ? og_locale_key : I18n.locale
+    og_locale_url(locale)
   end
 
   def og_locale
@@ -104,6 +107,48 @@ module SocialHelper
                      OG_LOCALES[I18n.default_locale.to_sym])
   end
 
+  # Locales that can appear in a URL. Cccms::LOCALES will replace this when
+  # the set_locale allowlist lands; :root is in available_locales but is not
+  # a language.
+  def og_url_locales
+    I18n.available_locales - [:root]
+  end
+
+  # The request path with any locale prefix stripped, so a locale-specific
+  # URL can be rebuilt from it. /en/updates/foo, /de/updates/foo and
+  # /updates/foo all reduce to /updates/foo.
+  def og_bare_path
+    pattern = og_url_locales.join("|")
+    request.path.sub(%r{\A/(?:#{pattern})(?=/|\z)}, "").presence || "/"
+  end
+
+  # Absolute URL for this page in one locale. The default locale is
+  # unprefixed, matching default_url_options, so /updates/foo is the
+  # canonical German address and /de/updates/foo is a duplicate of it.
+  def og_locale_url(locale)
+    locale = locale.to_sym
+    locale = I18n.default_locale unless og_url_locales.include?(locale)
+
+    path = og_bare_path
+    return og_absolute_url(path) if locale == I18n.default_locale
+
+    og_absolute_url(path == "/" ? "/#{locale}" : "/#{locale}#{path}")
+  end
+
+  # One entry per locale in which the page genuinely has a translation,
+  # including a self-reference, which Google requires. A locale with no
+  # translation is omitted: /en/ would render German content through the
+  # fallback chain, which is not an English version of the page. With only
+  # one translation there is nothing to declare.
+  def og_hreflang_alternates
+    return [] unless @page&.persisted?
+
+    locales = @page.translated_locales.map(&:to_sym) & og_url_locales
+    return [] if locales.size < 2
+
+    locales.sort.map { |locale| [locale.to_s, og_locale_url(locale)] }
+  end
+
   # Only locales in which this page genuinely has a translation, so a
   # crawler is not told about a variant that would fall back.
   def og_locale_alternates
diff --git a/app/views/layouts/_social_meta.html.erb b/app/views/layouts/_social_meta.html.erb
index 9a86d26d..36547d26 100644
--- a/app/views/layouts/_social_meta.html.erb
+++ b/app/views/layouts/_social_meta.html.erb
@@ -23,6 +23,13 @@
     <meta property="og:locale:alternate" content="<%= alternate %>"/>
   <% end %>
 
+  <% if (hreflang_alternates = og_hreflang_alternates).any? %>
+    <% hreflang_alternates.each do |code, url| %>
+      <link rel="alternate" hreflang="<%= code %>" href="<%= url %>"/>
+    <% end %>
+    <link rel="alternate" hreflang="x-default" href="<%= og_locale_url(I18n.default_locale) %>"/>
+  <% end %>
+
   <% if og_published_time %>
     <meta property="article:published_time" content="<%= og_published_time %>"/>
   <% end %>
diff --git a/test/controllers/content_controller_test.rb b/test/controllers/content_controller_test.rb
index 304fd921..5bb02f3d 100644
--- a/test/controllers/content_controller_test.rb
+++ b/test/controllers/content_controller_test.rb
@@ -190,6 +190,46 @@ class ContentControllerTest < ActionController::TestCase
       FileUtils.rm_rf(Rails.root.join("tmp", "test_uploads", asset.id.to_s))
     end
   end
+
+  test "a translated page declares hreflang alternates and canonicalises per locale" do
+    node  = create_node_under_root "og_hreflang_test"
+    draft = find_or_create_draft(node, @user1)
+    draft.title = "Zweisprachig"
+    draft.save
+    node.publish_draft!
+    node.reload
+    Globalize.with_locale(:en) { node.head.update!(:title => "Bilingual") }
+
+    get :render_page, params: { :locale => "de", :page_path => ["og_hreflang_test"] }
+
+    assert_response :success
+    assert_select "link[rel=alternate][hreflang=de][href=?]",
+                  "http://test.host/og_hreflang_test"
+    assert_select "link[rel=alternate][hreflang=en][href=?]",
+                  "http://test.host/en/og_hreflang_test"
+    assert_select "link[rel=alternate][hreflang='x-default'][href=?]",
+                  "http://test.host/og_hreflang_test"
+    assert_select "link[rel=canonical][href=?]",
+                  "http://test.host/og_hreflang_test"
+  end
+
+  test "an untranslated page declares no alternates and canonicalises to the default locale" do
+    node  = create_node_under_root "og_single_locale_test"
+    draft = find_or_create_draft(node, @user1)
+    draft.title = "Nur Deutsch"
+    draft.save
+    node.publish_draft!
+
+    get :render_page, params: { :locale => "en", :page_path => ["og_single_locale_test"] }
+
+    assert_response :success
+    assert_select "link[rel=alternate][hreflang]", false,
+                  "one translation is nothing to declare"
+    # Served German through the fallback chain, so the German URL is
+    # canonical rather than the /en/ address that was requested.
+    assert_select "link[rel=canonical][href=?]",
+                  "http://test.host/og_single_locale_test"
+  end
   
   protected