allow more first name searches when alternate identities exist but ar…

…e all stanford or blank institutio

app/models/author.rb

@@ -38,11 +38,15 @@ def institution
     Settings.HARVESTER.INSTITUTION.name
   end
 
-  # indicates if the LastName, FirstInitial form for this user is ambiguous within our author database and there are no alternate identities
+  # indicates if the LastName, FirstInitial form for this user is ambiguous within our author database (including any alternate identities that include Stanford as an insitution)
+  #  also checks to see if there are alternate identities with institutions other than Stanford, which is problematic, and should be considered ambiguous
   def ambiguous_first_initial?
     return true unless first_name && last_name
-    first_initial_not_unique = self.class.where('preferred_first_name like ? and preferred_last_name = ?', "#{first_name[0]}%", last_name).where(active_in_cap: true, cap_import_enabled: true).size > 1
-    (first_initial_not_unique || !author_identities.empty?)
+    first_initial_not_unique = self.class.where('preferred_first_name like ? and preferred_last_name = ?', "#{first_name[0]}%", last_name).size > 1
+    author_identities_not_unique = author_identities.map do |author_identity|
+      (!author_identity.institution.blank? && !author_identity.institution.include?('Stanford')) || self.class.where('preferred_first_name like ? and preferred_last_name = ? and id != ?', "#{author_identity.first_name[0]}%", author_identity.last_name, author_identity.author_id).size > 1
+    end
+    (first_initial_not_unique || author_identities_not_unique.include?(true))
   end
 
   # @return [Array<Integer>] ScienceWireIds for approved publications

spec/factories/author.rb

@@ -54,6 +54,26 @@
     end
   end
 
+  factory :author_duped_last_name, parent: :author do
+    sunetid { FactoryBot.generate(:random_id) }
+    cap_profile_id { FactoryBot.generate(:random_id) }
+    university_id { FactoryBot.generate(:random_id) }
+    california_physician_license { FactoryBot.generate(:random_string) }
+    active_in_cap { true }
+    email { '[email protected]' }
+    official_first_name { 'Albert' }
+    official_last_name { 'Edler' }
+    official_middle_name { '' }
+    preferred_first_name { 'Albert' }
+    preferred_last_name { 'Edler' }
+    preferred_middle_name { '' }
+    emails_for_harvest { '[email protected]' }
+  end
+
+  factory :inactive_author, parent: :author do
+    active_in_cap { false }
+  end
+
   factory :author_with_alternate_identities, parent: :author do
     transient do
       alt_count { 1 } # default number of alternate identities to create

spec/lib/web_of_science/query_author_spec.rb

@@ -124,8 +124,8 @@
 
       it 'ignores the bad alternate identity data' do
         expect(author_one_identity.author_identities.first.first_name).to eq '.' # bad first name
-        # we get three name variants out (we would have more if we allowed the bad name variant)
-        expect(described_class.new(author_one_identity).send(:names)).to eq %w[Edler,Alice Edler,Alice,Jim Edler,Alice,J]
+        # we do not get the name variant with the period for a first name (we would have more if we allowed the bad name variant)
+        expect(described_class.new(author_one_identity).send(:names)).to eq %w[Edler,Alice Edler,A Edler,Alice,Jim Edler,Alice,J Edler,AJ Edler,A,J]
       end
     end
   end

spec/models/author_spec.rb

@@ -30,16 +30,28 @@
   end
 
   describe '#ambiguous_first_initial?' do
-    it 'confirms ambiguous first initial' do
-      subject.update_from_cap_authorship_profile_hash(auth_hash)
-      expect(subject.author_identities.size).to eq(2) # has alternate identities
-      expect(subject.ambiguous_first_initial?).to eq(true) # thus cannot search with first initial
-    end
-    it 'confirms non-ambiguous first initial' do
+    it 'confirms non-ambiguous first initial within stanford with no alternate identities' do
       odd_name = create :odd_name
       expect(odd_name.author_identities.size).to eq(0) # has no alternate identities
       expect(odd_name.ambiguous_first_initial?).to eq(false) # and no other odd names likes this at stanford, so ok to search with first initial
     end
+    it 'confirms non-ambiguous first initial within stanford with stanford only alternate identities' do
+      subject.update_from_cap_authorship_profile_hash(auth_hash)
+      expect(subject.author_identities.size).to eq(2) # has alternate identities
+      expect(subject.ambiguous_first_initial?).to eq(false) # ok, because all of the alternate identities are stanford or no institution, and no other first name ambiguity
+    end
+    it 'confirms ambiguous first initial within stanford with no alternate identities' do
+      create :author_duped_last_name
+      expect(subject.author_identities.size).to eq(0) # no alternate identities
+      expect(subject.ambiguous_first_initial?).to eq(true) # yes, because we now have another stanford author with the same last name and same first initial
+    end
+    it 'confirms ambiguous first initial even when non ambiguous within Stanford due to a non-Stanford alternate identity existing' do
+      author_with_alternate_identities = create :author_with_alternate_identities
+      expect(author_with_alternate_identities.author_identities.size).to eq(1) # alternate identities for primary author
+      expect(author_with_alternate_identities.author_identities.first.institution).not_to be blank? # alternate institution is not empty
+      expect(author_with_alternate_identities.author_identities.first.institution.include?('Stanford')).to be false # alternate institution is not Stanford
+      expect(author_with_alternate_identities.ambiguous_first_initial?).to eq(true) # yes, because even though there are no other stanford authors with similar names, they have a non-Stanford alternate identity
+    end
   end
 
   describe '#first_name' do