From cba2dd400e6f021f2c2e8822718e78eef1a99509 Mon Sep 17 00:00:00 2001
From: Eric Larson <ewlarson@gmail.com>
Date: Mon, 6 Nov 2023 09:30:24 -0600
Subject: [PATCH] Image Harvest: skip :failed and :placeheld states

We don't need to attempt image harvesting for placeheld or failed items.

Addresses #465
---
 config/schedule.rb                        |  2 +-
 lib/tasks/geoportal/geoportal_images.rake | 20 +++++++++++++++++++-
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/config/schedule.rb b/config/schedule.rb
index 059ad6789..491ff5492 100644
--- a/config/schedule.rb
+++ b/config/schedule.rb
@@ -11,7 +11,7 @@
 end
 # Harvest thumbnail images for search results
 every :day, at: '12:05am', roles: [:app] do
-  rake 'gblsci:images:harvest_retry'
+  rake 'geoportal:queue_incomplete_states'
 end
 every :day, at: '12:30am', roles: [:app] do
   rake 'sitemap:refresh'
diff --git a/lib/tasks/geoportal/geoportal_images.rake b/lib/tasks/geoportal/geoportal_images.rake
index fc9843aa9..ea823b40f 100644
--- a/lib/tasks/geoportal/geoportal_images.rake
+++ b/lib/tasks/geoportal/geoportal_images.rake
@@ -27,7 +27,25 @@ namespace :geoportal do
 
   desc 'Harvest images - Queue incomplete states for reprocessing'
   task queue_incomplete_states: :environment do
-    puts "Deprecated / Instead try: bundle exec rake gblsci:images:harvest_retry"
+    # Skipping :failed, :placeheld
+    states = [
+      :initialized,
+      :queued,
+      :processing
+    ]
+
+    states.each do |state|
+      sidecars = SolrDocumentSidecar.in_state(state)
+
+      puts "#{state} - #{sidecars.size}"
+
+      sidecars.each do |sc|
+        document = Geoblacklight::SolrDocument.find(sc.document_id)
+        GeoblacklightSidecarImages::StoreImageJob.perform_later(document.id)
+      rescue
+        puts "orphaned / #{sc.document_id}"
+      end
+    end
   end
 
   desc 'Failed State - Inspect metadata'