From 7022102a2b1214b005a3f3ae582a7abf4521f67c Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Mon, 17 Dec 2018 13:37:28 +0100
Subject: [PATCH] improve import missing dois rake test

---
 app/jobs/doi_import_by_day_missing_job.rb |  7 +++++++
 app/models/doi.rb                         | 25 ++++++++++++++++-------
 lib/tasks/doi.rake                        |  9 ++++++--
 3 files changed, 32 insertions(+), 9 deletions(-)
 create mode 100644 app/jobs/doi_import_by_day_missing_job.rb

diff --git a/app/jobs/doi_import_by_day_missing_job.rb b/app/jobs/doi_import_by_day_missing_job.rb
new file mode 100644
index 000000000..c7f8e0cc8
--- /dev/null
+++ b/app/jobs/doi_import_by_day_missing_job.rb
@@ -0,0 +1,7 @@
+class DoiImportByDayMissingJob < ActiveJob::Base
+  queue_as :lupo_background
+
+  def perform(options={})
+    Doi.import_by_day_missing(options)
+  end
+end
\ No newline at end of file
diff --git a/app/models/doi.rb b/app/models/doi.rb
index ab5a1acc6..446020a0e 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -367,6 +367,17 @@ def self.import_all(options={})
     end
   end
 
+  def self.import_missing(options={})
+    from_date = options[:from_date].present? ? Date.parse(options[:from_date]) : Date.current
+    until_date = options[:until_date].present? ? Date.parse(options[:until_date]) : Date.current
+
+    # get every day between from_date and until_date
+    (from_date..until_date).each do |d|
+      DoiImportByDayMissingJob.perform_later(from_date: d.strftime("%F"))
+      puts "Queued importing for missing DOIs created on #{d.strftime("%F")}."
+    end
+  end
+
   def self.import_by_day(options={})
     return nil unless options[:from_date].present?
     from_date = Date.parse(options[:from_date])
@@ -397,15 +408,15 @@ def self.import_by_day(options={})
     end
   end
 
-  def self.import_missing(options={})
-    from_date = options[:from_date].present? ? Date.parse(options[:from_date]) : Date.current
-    until_date = options[:until_date].present? ? Date.parse(options[:until_date]) : Date.current
+  def self.import_by_day_missing(options={})
+    return nil unless options[:from_date].present?
+    from_date = Date.parse(options[:from_date])
 
     count = 0
 
     logger = Logger.new(STDOUT)
 
-    Doi.where(schema_version: nil).where(created: from_date.midnight..until_date.end_of_day).find_each do |doi|
+    Doi.where(schema_version: nil).where(created: from_date.midnight..from_date.end_of_day).find_each do |doi|
       begin
         string = doi.current_metadata.present? ? doi.current_metadata.xml : nil
         meta = doi.read_datacite(string: string, sandbox: doi.sandbox)
@@ -413,8 +424,8 @@ def self.import_missing(options={})
           [a.to_sym, meta[a]]
         end.to_h.merge(schema_version: meta["schema_version"] || "http://datacite.org/schema/kernel-4", version_info: meta["version"], xml: string)
 
-        # update_attributes will trigger validations and Elasticsearch indexing
-        doi.update_attributes(attrs)
+        # update_columns will NOT trigger validations and Elasticsearch indexing
+        doi.update_columns(attrs)
       rescue TypeError, NoMethodError, ActiveRecord::LockWaitTimeout => error
         logger.error "[MySQL] Error importing metadata for " + doi.doi + ": " + error.message
       else
@@ -423,7 +434,7 @@ def self.import_missing(options={})
     end
 
     if count > 0
-      logger.info "[MySQL] Imported metadata for #{count} DOIs created #{options[:from_date]} - #{options[:until_date]}."
+      logger.info "[MySQL] Imported metadata for #{count} DOIs created on #{options[:from_date]}."
     end
   end
 
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index 29d609d12..be3ce1cb1 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -29,8 +29,13 @@ namespace :doi do
 
   desc 'Import missing DOIs'
   task :import_missing => :environment do
-    from_date = ENV['FROM_DATE'] || Date.current.strftime("%F")
-    until_date = ENV['UNTIL_DATE'] || Date.current.strftime("%F")
+    if ENV['YEAR'].present?
+      from_date = "#{ENV['YEAR']}-01-01"
+      until_date = "#{ENV['YEAR']}-12-31"
+    else
+      from_date = ENV['FROM_DATE'] || Date.current.strftime("%F")
+      until_date = ENV['UNTIL_DATE'] || Date.current.strftime("%F")
+    end
 
     Doi.import_missing(from_date: from_date, until_date: until_date)
   end