updated scripts and added new rake tasks

app/models/stash_engine/ror_org.rb

@@ -22,19 +22,19 @@ class RorOrg < ApplicationRecord
 
     # Search the RorOrgs for the given string. This will search name, acronyms, aliases, etc.
     # @return an Array of Hashes { id: 'https://ror.org/12345', name: 'Sample University' }
-    def self.find_by_ror_name(query, max_results = ROR_MAX_RESULTS)
+    def self.find_by_ror_name(query)
       return [] unless query.present?
 
       query = query.downcase
       # First, find matches at the beginning of the name string, and exact matches in the acronyms/aliases
       resp = where("LOWER(name) LIKE ? OR JSON_SEARCH(LOWER(acronyms), 'all', ?) or JSON_SEARCH(LOWER(aliases), 'all', ?)",
-                   "#{query}%", query.to_s, query.to_s).limit(max_results)
+                   "#{query}%", query.to_s, query.to_s).limit(ROR_MAX_RESULTS)
       results = resp.map do |r|
         { id: r.ror_id, name: r.name, country: r.country, acronyms: r.acronyms, aliases: r.aliases }
       end
 
       # If we don't have enough results, find matches at the beginning of the acronyms/aliases
-      if results.size < max_results
+      if results.size < ROR_MAX_RESULTS
         resp = where("JSON_SEARCH(LOWER(acronyms), 'all', ?) or JSON_SEARCH(LOWER(aliases), 'all', ?)",
                      "#{query}%", "#{query}%").limit(ROR_MAX_RESULTS - results.size)
         resp.each do |r|
@@ -43,8 +43,8 @@ def self.find_by_ror_name(query, max_results = ROR_MAX_RESULTS)
       end
 
       # If we don't have enough results, find matches elsewhere in the name string
-      if results.size < max_results
-        resp = where('LOWER(name) LIKE ?', "%#{query}%").limit(max_results - results.size)
+      if results.size < ROR_MAX_RESULTS
+        resp = where('LOWER(name) LIKE ?', "%#{query}%").limit(ROR_MAX_RESULTS - results.size)
         resp.each do |r|
           results << { id: r.ror_id, name: r.name, country: r.country, acronyms: r.acronyms, aliases: r.aliases }
         end
@@ -53,6 +53,31 @@ def self.find_by_ror_name(query, max_results = ROR_MAX_RESULTS)
       results.flatten.uniq
     end
 
+    # Search the RorOrgs for the given string. This will search name, acronyms, aliases, etc.
+    # @return an Array of Hashes { id: 'https://ror.org/12345', name: 'Sample University' }
+    # This method is used for auto-matching scripts, where no human has to confirm the match.
+    def self.find_by_name_for_auto_matching(query)
+      max_results = 10
+      return [] unless query.present?
+
+      query = query.downcase
+      # First, find matches at the beginning of the name string, and exact matches in the acronyms/aliases
+      resp = where("LOWER(name) LIKE ? OR JSON_SEARCH(LOWER(acronyms), 'all', ?) or JSON_SEARCH(LOWER(aliases), 'all', ?)",
+                   "#{query}%", query.to_s, query.to_s).limit(max_results)
+      results = resp.map do |r|
+        { id: r.ror_id, name: r.name, country: r.country, acronyms: r.acronyms, aliases: r.aliases }
+      end
+
+      return results if results.any?
+
+      # If we don't have enough results, find matches at the beginning of the acronyms/aliases
+      resp = where("JSON_SEARCH(LOWER(acronyms), 'all', ?) or JSON_SEARCH(LOWER(aliases), 'all', ?)",
+                   "#{query}%", "#{query}%").limit(max_results)
+      resp.map do |r|
+        { id: r.ror_id, name: r.name, country: r.country, acronyms: r.acronyms, aliases: r.aliases }
+      end
+    end
+
     # Return the first match for the given name
     # @return a StashEngine::RorOrg or nil
     def self.find_first_by_ror_name(ror_name)

cron/monthly.sh

@@ -23,5 +23,9 @@ bundle exec rails link_out:seed_genbank_ids >> /home/ec2-user/deploy/shared/log/
 bundle exec rails link_out:publish >> /home/ec2-user/deploy/shared/log/link_out_publish.log 2>&1
 
 # Update ROR organizations
-bundle exec rails affiliation_import:update_ror_orgs >>/home/ec2-user/deploy/shared/log/ror_update.log 2>&1
-bundle exec rails affiliation_import:update_affiliations_names >>/home/ec2-user/deploy/shared/log/affiliations_name_updates.log 2>&1
+bundle exec rails affiliation_import:update_ror_orgs >> /home/ec2-user/deploy/shared/log/ror_update.log 2>&1
+bundle exec rails affiliation_import:update_affiliations_names >> /home/ec2-user/deploy/shared/log/affiliations_name_updates.log 2>&1
+
+# Cleanup affiliation/contributor records
+bundle exec rails cleanup:affiliations_wo_ror >> /home/ec2-user/deploy/shared/log/affiliations_wo_ror_cleanup.log 2>&1
+bundle exec rails cleanup:contributors_wo_ror >> /home/ec2-user/deploy/shared/log/contributors_wo_ror_cleanup.log 2>&1

lib/stash/organization/affiliation_ror_matcher.rb

@@ -16,7 +16,7 @@ def connect_to_ror(item, ror)
         rep ||= StashDatacite::Affiliation.from_ror_id(ror_id: ror_id)
         to_fix = StashDatacite::Affiliation.where(ror_id: nil, long_name: item.long_name)
 
-        update_affiliation_name(rep, ror) if ror[:name] != rep.long_name
+        update_affiliation_name(rep, ror)
 
         message = 'Replacing affiliations with'
         puts " - #{message} name \"#{item.long_name}\" (ids: #{to_fix.ids}) with \"#{ror[:name]}\" (id: #{rep.id || 'new'})"
@@ -25,11 +25,6 @@ def connect_to_ror(item, ror)
         return unless perform_updates
 
         to_fix.each do |aff|
-          # updating contributors affiliation with new affiliation
-          aff.contributors.each do |contributor|
-            contributor.affiliations << rep unless contributor.affiliations.include?(rep)
-          end
-
           # updating authors affiliation with new affiliation
           aff.authors.each do |author|
             author.affiliation = rep
@@ -39,8 +34,10 @@ def connect_to_ror(item, ror)
       end
 
       def update_affiliation_name(rep, ror)
+        return if ror[:name] == rep.long_name
+
         rep.update(long_name: ror[:name]) if perform_updates
-        message = 'Updating existing affiliation name with'
+        message = 'Updating existing affiliation name'
         puts " - #{message} \"#{rep.long_name}\" (id: #{rep.id}) with \"#{ror[:name]}\""
         @csv_rows << [rep.id, rep.long_name, rep.authors.count, message, ror[:name]]
       end

lib/stash/organization/base_ror_matcher.rb

@@ -8,7 +8,7 @@ class BaseRorMatcher
 
       attr_reader :perform_updates
 
-      def initialize(perform_updates: false, start_id: nil, end_id: nil, start_created_at: nil, end_created_at: nil)
+      def initialize(perform_updates: true, start_id: nil, end_id: nil, start_created_at: nil, end_created_at: nil)
         @perform_updates = perform_updates
         @end_id = end_id
         @start_id = start_id
@@ -58,8 +58,8 @@ def end_report
           [],
           [@text.gsub('Processing', 'From')],
           [" - Updated: #{@updates_count} records."],
-          [" - No ROR found: #{@no_ror_found_count} records."],
-          [" - Multiple RORs found: #{@multiple_ror_found_count} records."]
+          [" - Multiple RORs found: #{@multiple_ror_found_count} records."],
+          [" - No ROR found: #{@no_ror_found_count} records."]
         ]
         update_csv_report(messages)
 
@@ -112,13 +112,14 @@ def handle_item(item, item_name, index)
           return
         end
 
-        rors = StashEngine::RorOrg.find_by_ror_name(item_name, 1)
+        rors = StashEngine::RorOrg.find_by_name_for_auto_matching(item_name)
         case rors.count
         when 0
           @no_ror_found_count += 1
-          message = 'Could not find ROR'
-          @csv_rows << [item.id, item_name, message]
-          puts " - #{message} for \"#{item_name}\""
+          # Do not add to CSV report, nor log file, as it will increase the file size too much
+          # message = 'Could not find ROR'
+          # @csv_rows << [item.id, item_name, message]
+          # puts " - #{message} for \"#{item_name}\""
         when 1
           connect_to_ror(item, rors.first)
         else

lib/tasks/cleanup.rake

@@ -0,0 +1,18 @@
+# :nocov:
+namespace :cleanup do
+
+  # example usage: RAILS_ENV=development bundle exec rake cleanup:affiliations_wo_ror
+  # https://github.com/datadryad/dryad-app/blob/main/documentation/technical_notes/affiliations.md#cleaning-affiliation-names
+  desc 'Match Affiliations with ROR organizations'
+  task affiliations_wo_ror: :environment do
+    Stash::Organization::AffiliationRorMatcher.new.perform
+  end
+
+  # example usage: RAILS_ENV=development bundle exec rake cleanup:contributors_wo_ror
+  # https://github.com/datadryad/dryad-app/blob/main/documentation/technical_notes/contributors.md#cleaning-contributor-names
+  desc 'Match Contributors with ROR organizations'
+  task contributors_wo_ror: :environment do
+    Stash::Organization::ContributorRorMatcher.new.perform
+  end
+end
+# :nocov: