From 3cd5d53e6f4c71977062bdbf0f59d5df653b25a5 Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Sat, 19 Dec 2020 19:42:52 +0100
Subject: [PATCH] better messaging during import. datacite/datacite#965

---
 app/jobs/doi_not_indexed_job.rb |  4 ++--
 app/models/client.rb            |  4 ++--
 app/models/doi.rb               | 10 +++-------
 lib/tasks/client.rake           |  2 +-
 4 files changed, 8 insertions(+), 12 deletions(-)

diff --git a/app/jobs/doi_not_indexed_job.rb b/app/jobs/doi_not_indexed_job.rb
index 80b6d19fc..c9fa75152 100644
--- a/app/jobs/doi_not_indexed_job.rb
+++ b/app/jobs/doi_not_indexed_job.rb
@@ -3,7 +3,7 @@
 class DoiNotIndexedJob < ApplicationJob
   queue_as :lupo_background
 
-  def perform(client_id, _options = {})
-    Doi.import_by_client(client_id: client_id)
+  def perform(client_id, options = {})
+    Doi.import_by_client(client_id, options)
   end
 end
diff --git a/app/models/client.rb b/app/models/client.rb
index 2c880d702..ef7cf6224 100644
--- a/app/models/client.rb
+++ b/app/models/client.rb
@@ -718,8 +718,8 @@ def self.import_dois_not_indexed(query: nil)
 
     # loop through repositories that have DOIs not indexed in Elasticsearch
     table.each do |row|
-      Rails.logger.info "Indexing #{row["DOIs in Database"]} DOIs (#{row["DOIs missing"]} missing) for repository #{row["Repository ID"]}."
-      DoiNotIndexedJob.perform_later(row["Repository ID"])
+      Rails.logger.info "Started to import #{row["DOIs in Database"]} DOIs (#{row["DOIs missing"]} missing) for repository #{row["Repository ID"]}."
+      DoiNotIndexedJob.perform_later(row["Repository ID"], doi_count: row["DOIs in Database"], missing_count: row["DOIs missing"])
     end
   end
 
diff --git a/app/models/doi.rb b/app/models/doi.rb
index 655988a54..790c2264e 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -1243,7 +1243,7 @@ def self.import_one(doi_id: nil)
     message
   end
 
-  def self.import_by_client(client_id: nil)
+  def self.import_by_client(client_id, options={})
     client = ::Client.where(symbol: client_id).first
     return nil if client.blank?
 
@@ -1271,14 +1271,10 @@ def self.import_by_client(client_id: nil)
       end
 
       count += dois.length
-      Rails.logger.info "[Elasticsearch] Imported #{count} DOIs for repository #{client_id}."
+      Rails.logger.info "[Elasticsearch] So far imported #{count} DOIs for repository #{client_id}."
     end
 
-    if errors > 1
-      Rails.logger.error "[Elasticsearch] #{errors} errors importing #{count} DOIs for repository #{client_id}."
-    elsif count > 0
-      Rails.logger.info "[Elasticsearch] Imported a total of #{count} DOIs for repository #{client_id}."
-    end
+    Rails.logger.info "[Elasticsearch] Imported a total of #{count} DOIs (#{options[:missing_count].to_i - errors} out of #{options[:missing_count].to_i} errors resolved) for repository #{client_id}."
 
     count
   rescue Elasticsearch::Transport::Transport::Errors::RequestEntityTooLarge, Faraday::ConnectionFailed, ActiveRecord::LockWaitTimeout => e
diff --git a/lib/tasks/client.rake b/lib/tasks/client.rake
index 7a9bada66..e9e4e2e4a 100644
--- a/lib/tasks/client.rake
+++ b/lib/tasks/client.rake
@@ -99,7 +99,7 @@ namespace :client do
 
     # import DOIs for client
     puts "#{client.dois.length} DOIs will be imported."
-    Doi.import_by_client(client_id: ENV["CLIENT_ID"])
+    Doi.import_by_client(ENV["CLIENT_ID"])
   end
 
   desc "Import dois not indexed"