From 96cc0d9fcd80c4bdc4bd122ccca5a3ce2dbe47c8 Mon Sep 17 00:00:00 2001
From: Nick Budak <budak@stanford.edu>
Date: Thu, 29 Feb 2024 15:36:39 -0800
Subject: [PATCH 1/2] Update to GeoCombine v0.9

---
 Gemfile      | 1 +
 Gemfile.lock | 6 +++++-
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/Gemfile b/Gemfile
index 76f4cf5a..19f50d67 100644
--- a/Gemfile
+++ b/Gemfile
@@ -78,6 +78,7 @@ gem 'twitter-typeahead-rails'
 gem 'blacklight_range_limit', '~> 7.0'
 gem 'redis', '~> 5.0'
 gem 'geo_monitor', '~> 0.7', github: 'geoblacklight/geo_monitor'
+gem 'geo_combine', '>= 0.9' # For OpenGeoMetadata indexing
 gem 'sidekiq', '~> 7.0'
 gem 'whenever', require: false
 gem 'bootstrap', '~> 4.0'
diff --git a/Gemfile.lock b/Gemfile.lock
index e452e981..34fb1ab4 100644
--- a/Gemfile.lock
+++ b/Gemfile.lock
@@ -227,13 +227,16 @@ GEM
     faraday-net_http_persistent (2.1.0)
       faraday (~> 2.5)
       net-http-persistent (~> 4.0)
+    faraday-retry (2.2.0)
+      faraday (~> 2.0)
     ffi (1.16.3)
     ffi-compiler (1.0.1)
       ffi (>= 1.0.0)
       rake
-    geo_combine (0.8.0)
+    geo_combine (0.9.0)
       activesupport
       faraday-net_http_persistent (~> 2.0)
+      faraday-retry (~> 2.2)
       git
       json-schema
       nokogiri
@@ -573,6 +576,7 @@ DEPENDENCIES
   dlss-capistrano
   factory_bot_rails (~> 6.2.0)
   faraday (~> 2.0)
+  geo_combine (>= 0.9)
   geo_monitor (~> 0.7)!
   geoblacklight (~> 3.8)
   honeybadger

From 00032fd42cbd01c466d4a42bbd1306ad14f21e82 Mon Sep 17 00:00:00 2001
From: Nick Budak <budak@stanford.edu>
Date: Tue, 13 Feb 2024 15:37:34 -0800
Subject: [PATCH 2/2] Add scheduled tasks to pull and index OpenGeoMetadata
 records

---
 config/schedule.rb        |  5 +++++
 lib/tasks/earthworks.rake | 21 +++------------------
 2 files changed, 8 insertions(+), 18 deletions(-)

diff --git a/config/schedule.rb b/config/schedule.rb
index cfc94021..34f2f41c 100644
--- a/config/schedule.rb
+++ b/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
diff --git a/lib/tasks/earthworks.rake b/lib/tasks/earthworks.rake
index 0f0a0407..3d920a8b 100644
--- a/lib/tasks/earthworks.rake
+++ b/lib/tasks/earthworks.rake
@@ -114,35 +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
       harvester = Earthworks::Harvester.new(ogm_repos: Settings.OGM_REPOS)
       indexer = GeoCombine::Indexer.new
-      puts "Indexing #{harvester.ogm_path} into #{indexer.solr_url}"
-      total = indexer.index(harvester.docs_to_index)
-      puts "Indexed #{total} documents"
-    end
-
-    desc 'Index content from GeoBlacklight sites'
-    task :harvest_geo_blacklight do
-      GeoCombine::GeoBlacklightHarvester.configure do
-        Settings.GEO_BLACKLIGHT_HARVEST_SITES.to_h
-      end
-
-      Settings.GEO_BLACKLIGHT_HARVEST_SITES.to_h.keys.each do |site_key|
-        Rake::Task['geocombine:geoblacklight_harvester:index'].invoke(site_key)
-      end
+      indexer.index(harvester.docs_to_index)
     end
   end