Use effective_status for Page#status (#1103)

Pages have a `status` attribute that is meant to be a convenient way to figure out what the current status of a page is, skipping over short-lived, erroneous errors from a single snapshot. A while back, we added "effective" status codes to versions, which attempt to figure out if a version with a 200 status code actually should have been an error code (surprisingly common!). We should have updated Page's status calculation to use that, since the page is really about what is effectively the current state, but missed it at the time.

app/models/page.rb

@@ -174,8 +174,8 @@ def ensure_page_urls
     urls.find_or_create_by!(url:) if saved_change_to_attribute?('url')
   end
 
-  def update_status
-    new_status = calculate_status
+  def update_status(relative_to: nil)
+    new_status = calculate_status(relative_to:)
     self.update(status: new_status) unless new_status.zero?
     self.status
   end
@@ -317,9 +317,10 @@ def calculate_status(relative_to: nil)
       version_time = last_time - capture_time
       total_time += version_time
 
-      if version.status >= 400
+      version_status = version.effective_status
+      if version_status >= 400
         error_time += version_time
-        latest_error ||= version.status
+        latest_error ||= version_status
       end
       last_time = version.capture_time
     end

lib/tasks/update_page_statuses.rake

@@ -1,9 +1,14 @@
 desc 'Update the `status` field on all pages. The first parameter optionally sets what pages to update: "recent" (default, pages updated recently enough to change status), "unknown" (only pages with unknown status), "all" (all pages)'
-task :update_page_statuses, [:where] => [:environment] do |_t, args|
+task :update_page_statuses, [:where, :at_time] => [:environment] do |_t, args|
   where_options = ['recent', 'unknown', 'all']
   where = args[:where] || where_options[0]
   abort("First argument must be one of (#{where_options.join ', '})") unless where_options.include? where
 
+  at_time = args[:at_time]
+  if at_time.present? && at_time != 'latest_version'
+    at_time = Time.parse(args[:at_time])
+  end
+
   ActiveRecord::Migration.say_with_time('Updating status codes on pages...') do
     DataHelpers.with_activerecord_log_level(:error) do
       page_set = Page.all.order(created_at: :asc)
@@ -22,7 +27,13 @@ task :update_page_statuses, [:where] => [:environment] do |_t, args|
       total = page_set.size
 
       DataHelpers.iterate_each(page_set, batch_size: 500) do |page|
-        page.update_status
+        relative_to = if at_time == 'latest_version'
+                        page.latest&.capture_time
+                      else
+                        at_time
+                      end
+
+        page.update_status(relative_to:)
         completed += 1
         if Time.now - last_update > 2
           DataHelpers.log_progress(completed, total)

test/models/page_test.rb

@@ -218,6 +218,15 @@ class PageTest < ActiveSupport::TestCase
     assert_equal(page.update_status, 404, 'Status should match the latest error code')
   end
 
+  test 'pages use version#effective_status, not raw status' do
+    page = Page.create(url: 'https://example.gov/')
+    page.versions.create(capture_time: Time.now - 15.days, status: 200, title: '404 Not Found')
+    page.versions.create(capture_time: Time.now - 12.days, status: 200, title: '404 Not Found')
+    page.versions.create(capture_time: Time.now - 10.days, status: 200, title: '404 Not Found')
+    page.versions.create(capture_time: Time.now - 1.day, status: 200, title: '404 Not Found')
+    assert_equal(page.update_status, 404, 'Status should be the effective_status of the versions')
+  end
+
   test 'pages can calculate a status even when some versions have no status' do
     page = Page.create(url: 'https://example.gov/')
     page.versions.create(capture_time: Time.now - 12.days)