Merge pull request #535 from datacite/update-target-doi

fix rake task

app/jobs/target_doi_by_id_job.rb

@@ -5,12 +5,11 @@ class TargetDoiByIdJob < ActiveJob::Base
     Rails.logger.error error.message
   end
 
-  def perform(id, options={})
+  def perform(id, options = nil)
     item = Event.where(uuid: id).first
     return false if item.blank?
 
     item.set_source_and_target_doi
-
     if item.save
       Rails.logger.info "Target doi for #{item.uuid} updated."
     else

app/models/concerns/indexable.rb

@@ -275,6 +275,8 @@ def query(query, options={})
         filter << { terms: { registrant_id: options[:registrant_id].split(",") }} if options[:registrant_id].present?
         filter << { terms: { registrant_id: options[:provider_id].split(",") }} if options[:provider_id].present?
         filter << { terms: { issn: options[:issn].split(",") }} if options[:issn].present?
+
+        must_not << { exists: { field: "target_doi" }} if options[:update_target_doi].present?
       elsif self.name == "Prefix"
         if query.present?
           must = [{ prefix: { prefix: query }}]

app/models/event.rb

@@ -379,32 +379,6 @@ def self.update_crossref(options = {})
     response.results.total
   end
 
-  def self.update_target_doi(options = {})
-    size = (options[:size] || 1000).to_i
-    cursor = (options[:cursor] || [])
-    target_relation_type_id = options[:target_relation_type_id]
-
-    response = Event.query(nil, target_relation_type_id: target_relation_type_id, page: { size: 1, cursor: [] })
-    Rails.logger.info "[Update] #{response.results.total} events with target_relation_type_id #{target_relation_type_id.to_s}."
-
-    # walk through results using cursor
-    if response.results.total > 0
-      while response.results.results.length > 0 do
-        response = Event.query(nil, target_relation_type_id: target_relation_type_id, page: { size: size, cursor: cursor })
-        break unless response.results.results.length.positive?
-
-        Rails.logger.info "[Update] Updating #{response.results.results.length} events with target_relation_type_id #{target_relation_type_id.to_s} starting with _id #{response.results.to_a.first[:_id]}."
-        cursor = response.results.to_a.last[:sort]
-
-        ids = response.results.results.map(&:uuid).uniq
-
-        TargetDoiJob.perform_later(ids, options)
-      end
-    end
-
-    response.results.total
-  end
-
   def self.update_datacite_crossref(options = {})
     update_datacite_ra(options.merge(ra: "crossref"))
   end
@@ -601,7 +575,7 @@ def self.label_state_event(event)
   # +job_name+:: Acive Job class name of the Job that would be executed on every matched results 
   def self.loop_through_events(options)
     size = (options[:size] || 1000).to_i
-    cursor = [options[:from_id] || Doi.minimum(:id).to_i, options[:until_id] || Doi.maximum(:id).to_i]
+    cursor = options[:cursor] || [options[:from_id] || Event.minimum(:id).to_i, options[:until_id] || Event.maximum(:id).to_i]
     filter = options[:filter] || {}
     label = options[:label] || ""
     job_name = options[:job_name] || ""
@@ -620,7 +594,7 @@ def self.loop_through_events(options)
         cursor = response.results.to_a.last[:sort]
         Rails.logger.info "#{label} Cursor: #{cursor} "
 
-        ids = response.results.results.map(&:uuid)
+        ids = response.results.results.map(&:uuid).uniq
         ids.each do |id|
           Object.const_get(job_name).perform_later(id, filter)
         end

lib/tasks/event.rake

@@ -61,11 +61,15 @@ namespace :event do
     Event.update_registrant(cursor: cursor, size: ENV['SIZE'])
   end
 
-  desc 'update target doi'
-  task :update_target_doi => :environment do
-    cursor = ENV['CURSOR'].to_s.split(",") || [Event.minimum(:id), Event.minimum(:id)]
-
-    Event.update_target_doi(cursor: cursor, target_relation_type_id: ENV['TARGET_RELATION_TYPE_ID'], size: ENV['SIZE'])
+  desc "update target doi"
+  task update_target_doi: :environment do
+    options = {
+      cursor: ENV["CURSOR"].present? ? Base64.urlsafe_decode64(ENV["CURSOR"]).split(",", 2) : [],
+      filter: { update_target_doi: true },
+      label: "[UpdateTargetDoi] Updating",
+      job_name: "TargetDoiByIdJob",
+    }
+    Event.loop_through_events(options)
   end
 end