From 52c6e7adc8a948b9a60b82a9725610ba1e486283 Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Sun, 20 Sep 2020 10:09:18 +0200
Subject: [PATCH] rake tasks to re-import metadata based on Elasticsearch query
 #638

---
 app/jobs/import_doi_job.rb       |  7 +++++++
 app/models/doi.rb                | 11 +++++------
 lib/tasks/doi.rake               | 22 +++++++++++++++++++---
 spec/jobs/import_doi_job_spec.rb | 16 ++++++++++++++++
 spec/jobs/update_doi_job_spec.rb | 16 ++++++++++++++++
 5 files changed, 63 insertions(+), 9 deletions(-)
 create mode 100644 app/jobs/import_doi_job.rb
 create mode 100644 spec/jobs/import_doi_job_spec.rb
 create mode 100644 spec/jobs/update_doi_job_spec.rb

diff --git a/app/jobs/import_doi_job.rb b/app/jobs/import_doi_job.rb
new file mode 100644
index 000000000..ada847c17
--- /dev/null
+++ b/app/jobs/import_doi_job.rb
@@ -0,0 +1,7 @@
+class ImportDoiJob < ActiveJob::Base
+  queue_as :lupo_background
+
+  def perform(doi_id, options={})
+    Doi.import_one(doi_id: doi_id)
+  end
+end
diff --git a/app/models/doi.rb b/app/models/doi.rb
index 9fa208905..d90f949f9 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -1224,8 +1224,8 @@ def self.index_one(doi_id: nil)
 
   def self.import_one(doi_id: nil)
     doi = Doi.where(doi: doi_id).first
-    if doi.nil?
-      Rails.logger.error "[MySQL] DOI #{doi_id} not found."
+    if doi.blank?
+      Rails.logger.error "[MySQL] Error importing DOI " + doi_id + ": not found"
       return nil
     end
 
@@ -1238,17 +1238,16 @@ def self.import_one(doi_id: nil)
     meta = doi.read_datacite(string: string, sandbox: doi.sandbox)
     attrs = %w(creators contributors titles publisher publication_year types descriptions container sizes formats language dates identifiers related_identifiers funding_references geo_locations rights_list subjects content_url version_info).map do |a|
       [a.to_sym, meta[a]]
-    end.to_h.merge(schema_version: meta["schema_version"] || "http://datacite.org/schema/kernel-4", xml: string)
+    end.to_h.merge(schema_version: meta["schema_version"] || "http://datacite.org/schema/kernel-4", xml: string, version: doi.version.to_i + 1)
 
     # update_attributes will trigger validations and Elasticsearch indexing
     doi.update_attributes(attrs)
-    Rails.logger.warn "[MySQL] Imported metadata for DOI " + doi.doi + "."
-    "Imported DOI #{doi.doi}."
+    Rails.logger.info "[MySQL] Imported metadata for DOI " + doi.doi + "."
   rescue TypeError, NoMethodError, RuntimeError, ActiveRecord::StatementInvalid, ActiveRecord::LockWaitTimeout => e
     if doi.present?
       Rails.logger.error "[MySQL] Error importing metadata for " + doi.doi + ": " + e.message
-      "Imported DOI #{doi.doi}."
     else
+      Rails.logger.error "[MySQL] Error importing metadata: " + e.message
       Raven.capture_exception(e)
     end
   end
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index cb7122359..e38f45348 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -154,10 +154,10 @@ namespace :doi do
     Doi.loop_through_dois(options)
   end
 
-  desc "Trigger DOI Update based on query"
+  desc "Trigger DOI update based on query"
   task update_dois_by_query: :environment do
-    # Ensure we have to specify a query of some kind.
-    if ENV['QUERY'].nil?
+    # Ensure we have specified a query of some kind.
+    if ENV['QUERY'].blank?
       puts "ENV['QUERY'] is required"
       exit
     end
@@ -171,6 +171,22 @@ namespace :doi do
     puts Doi.loop_through_dois(options)
   end
 
+  desc "Trigger DOI import based on query"
+  task import_dois_by_query: :environment do
+    # Ensure we have specified a query of some kind.
+    if ENV['QUERY'].blank?
+      puts "ENV['QUERY'] is required"
+      exit
+    end
+
+    options = {
+      query: ENV["QUERY"],
+      label: "[ImportDoiByQuery]",
+      job_name: "ImportDoiJob",
+      cursor: ENV["CURSOR"].present? ? Base64.urlsafe_decode64(ENV["CURSOR"]).split(",", 2) : [],
+    }
+    puts Doi.loop_through_dois(options)
+  end
 
   # until all Crossref DOIs are indexed as otherDoi
   desc "Refresh metadata"
diff --git a/spec/jobs/import_doi_job_spec.rb b/spec/jobs/import_doi_job_spec.rb
new file mode 100644
index 000000000..f93969fe6
--- /dev/null
+++ b/spec/jobs/import_doi_job_spec.rb
@@ -0,0 +1,16 @@
+require "rails_helper"
+
+describe ImportDoiJob, type: :job do
+  let(:doi) { create(:doi) }
+  subject(:job) { ImportDoiJob.perform_later(doi.doi) }
+
+  it "queues the job" do
+    expect { job }.to have_enqueued_job(ImportDoiJob)
+      .on_queue("test_lupo_background").at_least(1).times
+  end
+
+  after do
+    clear_enqueued_jobs
+    clear_performed_jobs
+  end
+end
diff --git a/spec/jobs/update_doi_job_spec.rb b/spec/jobs/update_doi_job_spec.rb
new file mode 100644
index 000000000..8b514fcd9
--- /dev/null
+++ b/spec/jobs/update_doi_job_spec.rb
@@ -0,0 +1,16 @@
+require "rails_helper"
+
+describe UpdateDoiJob, type: :job do
+  let(:doi) { create(:doi) }
+  subject(:job) { UpdateDoiJob.perform_later(doi.doi) }
+
+  it "queues the job" do
+    expect { job }.to have_enqueued_job(UpdateDoiJob)
+      .on_queue("test_lupo_background").at_least(1).times
+  end
+
+  after do
+    clear_enqueued_jobs
+    clear_performed_jobs
+  end
+end