Add a scheduled task to pull and index OGM data

Closes #639

config/schedule.rb

@@ -44,3 +44,8 @@
 every 1.day, at: '4:04 am', roles: %i[whenevs] do
   rake 'rake earthworks:prune_old_search_data[14]'
 end
+
+every :tuesday, at: '5:04 am', roles: %i[whenevs] do
+  rake 'rake earthworks:opengeometadata:pull'
+  rake 'rake earthworks:opengeometadata:index'
+end

lib/tasks/earthworks.rake

@@ -114,25 +114,20 @@ namespace :earthworks do
     desc 'Initialize OpenGeoMetadata repositories'
     task :clone do
       harvester = Earthworks::Harvester.new(ogm_repos: Settings.OGM_REPOS)
-      total = harvester.clone_all
-      puts "Cloned #{total} repositories"
+      harvester.clone_all
     end
 
     desc 'Fetch updated OpenGeoMetadata records for indexing'
     task :pull do
       harvester = Earthworks::Harvester.new(ogm_repos: Settings.OGM_REPOS)
-      total = harvester.pull_all
-      puts "Updated #{total} repositories"
+      harvester.pull_all
     end
 
     desc 'Index OpenGeoMetadata repositories'
-    task :index do
+    task index: :environment do
       harvester = Earthworks::Harvester.new(ogm_repos: Settings.OGM_REPOS)
-      solr_url = Blacklight.connection_config[:url]
-      indexer = GeoCombine::Indexer.new(solr: RSolr.connect(url: solr_url, adapter: :net_http_persistent))
-      puts "Indexing #{harvester.ogm_path} into #{indexer.solr_url}"
-      total = indexer.index(harvester.docs_to_index)
-      puts "Indexed #{total} documents"
+      indexer = GeoCombine::Indexer.new
+      indexer.index(harvester.docs_to_index)
     end
   end