From aab41e6670262e485be070bbc25059eabcb7722e Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Mon, 15 Jul 2019 09:22:23 +0200
Subject: [PATCH] transition to new affiliations format

---
 app/jobs/affiliation_job.rb | 24 +++++++++++++++
 app/models/doi.rb           | 61 +++++++++++++++++++++++++++++++++++--
 lib/tasks/doi.rake          |  5 +++
 3 files changed, 88 insertions(+), 2 deletions(-)
 create mode 100644 app/jobs/affiliation_job.rb

diff --git a/app/jobs/affiliation_job.rb b/app/jobs/affiliation_job.rb
new file mode 100644
index 000000000..b68cf8131
--- /dev/null
+++ b/app/jobs/affiliation_job.rb
@@ -0,0 +1,24 @@
+class AffiliationJob < ActiveJob::Base
+  queue_as :lupo_background
+
+  def perform(doi_id)
+    logger = Logger.new(STDOUT)
+    doi = Doi.where(doi: doi_id).first
+
+    if doi.present?
+      new_creators = Array.wrap(creators).map do |c|
+        c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+        c
+      end
+      new_contributors = Array.wrap(contributors).map do |c|
+        c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+        c
+      end
+      doi.update_attributes(creators: new_creators, contributors: new_contributors)
+
+      doi.__elasticsearch__.index_document
+    else
+      logger.info "[Affiliation] Error updaing DOI " + doi_id + ": not found"
+    end
+  end
+end
diff --git a/app/models/doi.rb b/app/models/doi.rb
index cbbc03fad..6ff6072f5 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -309,8 +309,8 @@ def as_indexed_json(options={})
       "doi" => doi,
       "identifier" => identifier,
       "url" => url,
-      "creators" => creators,
-      "contributors" => contributors,
+      "creators" => creators_with_affiliations,
+      "contributors" => contributors_with_affiliations,
       "creator_names" => creator_names,
       "titles" => titles,
       "descriptions" => descriptions,
@@ -544,6 +544,24 @@ def creator_names
     end
   end
 
+  # use newer index with old database following schema 4.3 changes
+  def creators_with_affiliations
+    Array.wrap(creators).map do |c|
+      c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+      c
+    end
+  end
+
+  def contributors_with_affiliations
+    Array.wrap(contributors).map do |c|
+      c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+      c
+    end
+  end
+
+  def convert_affiliation
+  end
+
   def doi=(value)
     write_attribute(:doi, value.upcase) if value.present?
   end
@@ -820,6 +838,45 @@ def set_defaults
     self.updated = Time.zone.now.utc.iso8601
   end
 
+  # convert affiliations from string to hash, following changes in schema 4.3
+  def self.convert_affiliations
+    logger = Logger.new(STDOUT)
+
+    response = Doi.query("creators.affiliation:* -creators.affiliation.name:*", page: { size: 1, cursor: [] })
+    logger.info "#{response.results.total} DOIs found that have the affiliation in the old format."
+
+    if response.results.total > 0
+      # walk through results using cursor
+      cursor = []
+
+      while response.results.results.length > 0 do
+        response = Doi.query("creators.affiliation:* -creators.affiliation.name:*", page: { size: 1000, cursor: cursor })
+        break unless response.results.results.length > 0
+
+        logger.info "[Affiliation] Updating #{response.results.results.length} DOIs starting with _id #{response.results.to_a.first[:_id]}."
+        cursor = response.results.to_a.last[:sort]
+
+        response.results.results.each do |d|
+          AffiliationJob.perform_later(d.doi)
+        end
+      end
+    end
+  end
+      
+  def creators_with_affiliations
+    Array.wrap(creators).map do |c|
+      c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+      c
+    end
+  end
+
+  def contributors_with_affiliations
+    Array.wrap(contributors).map do |c|
+      c["affiliation"] = { "name" => c["affiliation"] } if c["affiliation"].is_a?(String)
+      c
+    end
+  end
+
   def self.migrate_landing_page(options={})
     logger = Logger.new(STDOUT)
     logger.info "Starting migration"
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index 89aaf64e4..2fe71ab05 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -77,6 +77,11 @@ namespace :doi do
     Doi.set_minted
   end
 
+  desc 'Convert affiliations to new format'
+  task :convert_affiliations => :environment do
+    Doi.convert_affiliations
+  end
+
   desc 'Migrates landing page data handling camelCase changes at same time'
   task :migrate_landing_page => :environment do
     Doi.migrate_landing_page