From c2f1d06b98eedb47e931d87910aae14e9fbef1c0 Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Wed, 24 Oct 2018 18:51:24 +0200
Subject: [PATCH] tweak background jobs

---
 app/models/doi.rb               | 21 ++++++++++-----------
 lib/tasks/doi.rake              |  7 +++----
 spec/lib/tasks/doi_rake_spec.rb |  4 ++--
 3 files changed, 15 insertions(+), 17 deletions(-)

diff --git a/app/models/doi.rb b/app/models/doi.rb
index dcab58e08..1932d9b6f 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -235,20 +235,19 @@ def self.index(options={})
     # get every day between from_date and until_date
     (from_date..until_date).each do |d|
       DoiIndexByDayJob.perform_later(from_date: d.strftime("%F"))
-    end
-
-    "Queued indexing for DOIs created from #{from_date.strftime("%F")} until #{until_date.strftime("%F")}."
+      puts "Queued indexing for DOIs created on #{d.strftime("%F")}."
+    end    
   end
 
   def self.index_by_day(options={})
-    from_date = options[:from_date].present? ? Date.parse(options[:from_date]) : Date.current
-    until_date = from_date + 1.day
+    return nil unless options[:from_date].present?
+    
     errors = 0
     count = 0
 
     logger = Logger.new(STDOUT)
 
-    Doi.where("created >= ?", from_date.strftime("%F") + " 00:00:00").where("created < ?", until_date.strftime("%F") + " 00:00:00").where("updated > indexed").find_in_batches(batch_size: 500) do |dois|
+    Doi.where(created: [options[:from_date] + " 00:00:00", options[:from_date] + " 23:59:59"]).not_indexed.find_in_batches(batch_size: 500) do |dois|
       response = Doi.__elasticsearch__.client.bulk \
         index:   Doi.index_name,
         type:    Doi.document_type,
@@ -260,22 +259,22 @@ def self.index_by_day(options={})
     end
 
     if errors > 1
-      logger.info "[Elasticsearch] #{errors} errors indexing #{count} DOIs created on #{from_date.strftime("%F")}."
+      logger.info "[Elasticsearch] #{errors} errors indexing #{count} DOIs created on #{options[:from_date]}."
     elsif count > 1
-      logger.info "[Elasticsearch] Indexed #{count} DOIs created on #{from_date.strftime("%F")}."
+      logger.info "[Elasticsearch] Indexed #{count} DOIs created on #{options[:from_date]}."
     end
   rescue Elasticsearch::Transport::Transport::Errors::RequestEntityTooLarge, Faraday::ConnectionFailed => error
-    logger.info "[Elasticsearch] Error #{error.message} indexing DOIs created on #{from_date.strftime("%F")}."
+    logger.info "[Elasticsearch] Error #{error.message} indexing DOIs created on #{options[:from_date]}."
 
     count = 0
 
-    Doi.where("created >= ?", from_date.strftime("%F") + " 00:00:00").where("created < ?", until_date.strftime("%F") + " 00:00:00").where("updated > indexed").find_each do |doi|
+    Doi.where(created: [options[:from_date] + " 00:00:00", options[:from_date] + " 23:59:59"]).not_indexed.find_each do |doi|
       IndexJob.perform_later(doi)
       doi.update_column(:indexed, Time.zone.now)  
       count += 1
     end
   
-    logger.info "[Elasticsearch] Indexed #{count} DOIs created on #{from_date.strftime("%F")}."
+    logger.info "[Elasticsearch] Indexed #{count} DOIs created on #{options[:from_date]}."
   end
 
   def uid
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index 89ada32c6..a5748b394 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -12,12 +12,11 @@ namespace :doi do
       from_date = "#{ENV['YEAR']}-01-01"
       until_date = "#{ENV['YEAR']}-12-31"
     else
-      from_date = ENV['FROM_DATE'] || Date.current.beginning_of_month.strftime("%F")
-      until_date = ENV['UNTIL_DATE'] || Date.current.end_of_month.strftime("%F")
+      from_date = ENV['FROM_DATE'] || Date.current.strftime("%F")
+      until_date = ENV['UNTIL_DATE'] || Date.current.strftime("%F")
     end
 
-    response = Doi.index(from_date: from_date, until_date: until_date)
-    puts response
+    Doi.index(from_date: from_date, until_date: until_date)
   end
 
   desc 'Index DOIs per day'
diff --git a/spec/lib/tasks/doi_rake_spec.rb b/spec/lib/tasks/doi_rake_spec.rb
index 137ef8c8c..2e2cf0420 100644
--- a/spec/lib/tasks/doi_rake_spec.rb
+++ b/spec/lib/tasks/doi_rake_spec.rb
@@ -8,14 +8,14 @@
   ENV['UNTIL_DATE'] = "2018-08-05"
 
   let!(:doi)  { create_list(:doi, 10) }
-  let(:output) { "Queued indexing for DOIs created from 2018-01-04 until 2018-08-05.\n" }
+  let(:output) { "Queued indexing for DOIs created on 2018-01-04.\n" }
 
   it "prerequisites should include environment" do
     expect(subject.prerequisites).to include("environment")
   end
 
   it "should run the rake task" do
-    expect(capture_stdout { subject.invoke }).to eq(output)
+    expect(capture_stdout { subject.invoke }).to start_with(output)
   end
 
   it "should enqueue an DoiIndexByDayJob" do