From a0c486ca5f2cc82519d2f04344b9f1c23917544d Mon Sep 17 00:00:00 2001
From: jrhoads <jrhoads@datacite.org>
Date: Mon, 2 Dec 2024 10:14:53 +0100
Subject: [PATCH 1/4] Standardize; always use 'terms' for filters, not 'term'

---
 app/models/doi/graphql_query.rb               | 30 ++++++++--------
 .../doi/graphql_query_builder_filters_spec.rb | 36 +++++++++++++------
 2 files changed, 40 insertions(+), 26 deletions(-)

diff --git a/app/models/doi/graphql_query.rb b/app/models/doi/graphql_query.rb
index 46611b534..391dba466 100644
--- a/app/models/doi/graphql_query.rb
+++ b/app/models/doi/graphql_query.rb
@@ -125,22 +125,22 @@ def filters
 
       filter = []
       filter << { terms: { doi: options[:ids].map(&:upcase) } } if options[:ids].present?
-      filter << { term: { resource_type_id: options[:resource_type_id].underscore.dasherize } } if options[:resource_type_id].present?
+      filter << { terms: { resource_type_id: [options[:resource_type_id].underscore.dasherize] } } if options[:resource_type_id].present?
       filter << { terms: { "types.resourceType": options[:resource_type].split(",") } } if options[:resource_type].present?
       filter << { terms: { agency: options[:agency].split(",").map(&:downcase) } } if options[:agency].present?
       filter << { terms: { prefix: options[:prefix].to_s.split(",") } } if options[:prefix].present?
       filter << { terms: { language: options[:language].to_s.split(",").map(&:downcase) } } if options[:language].present?
-      filter << { term: { uid: options[:uid] } } if options[:uid].present?
+      filter << { terms: { uid: [options[:uid]] } } if options[:uid].present?
       filter << { range: { created: { gte: "#{options[:created].split(',').min}||/y", lte: "#{options[:created].split(',').max}||/y", format: "yyyy" } } } if options[:created].present?
       filter << { range: { publication_year: { gte: "#{options[:published].split(',').min}||/y", lte: "#{options[:published].split(',').max}||/y", format: "yyyy" } } } if options[:published].present?
-      filter << { term: { schema_version: "http://datacite.org/schema/kernel-#{options[:schema_version]}" } } if options[:schema_version].present?
+      filter << { terms: { schema_version: ["http://datacite.org/schema/kernel-#{options[:schema_version]}"] } } if options[:schema_version].present?
       filter << { terms: { "subjects.subject": options[:subject].split(",") } } if options[:subject].present?
       if options[:pid_entity].present?
-        filter << { term: { "subjects.subjectScheme": "PidEntity" } }
+        filter << { terms: { "subjects.subjectScheme": ["PidEntity"] } }
         filter << { terms: { "subjects.subject": options[:pid_entity].split(",").map(&:humanize) } }
       end
       if options[:field_of_science].present?
-        filter << { term: { "subjects.subjectScheme": "Fields of Science and Technology (FOS)" } }
+        filter << { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } }
         filter << { terms: { "subjects.subject": options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
       end
       if options[:field_of_science_repository].present?
@@ -150,7 +150,7 @@ def filters
         filter << { terms: { "fields_of_science_combined": options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
       end
       filter << { terms: { "rights_list.rightsIdentifier" => options[:license].split(",") } } if options[:license].present?
-      filter << { term: { source: options[:source] } } if options[:source].present?
+      filter << { terms: { source: [options[:source]] } } if options[:source].present?
       filter << { range: { reference_count: { "gte": options[:has_references].to_i } } } if options[:has_references].present?
       filter << { range: { citation_count: { "gte": options[:has_citations].to_i } } } if options[:has_citations].present?
       filter << { range: { part_count: { "gte": options[:has_parts].to_i } } } if options[:has_parts].present?
@@ -159,25 +159,25 @@ def filters
       filter << { range: { version_of_count: { "gte": options[:has_version_of].to_i } } } if options[:has_version_of].present?
       filter << { range: { view_count: { "gte": options[:has_views].to_i } } } if options[:has_views].present?
       filter << { range: { download_count: { "gte": options[:has_downloads].to_i } } } if options[:has_downloads].present?
-      filter << { term: { "landing_page.status": options[:link_check_status] } } if options[:link_check_status].present?
+      filter << { terms: { "landing_page.status": [options[:link_check_status]] } } if options[:link_check_status].present?
       filter << { exists: { field: "landing_page.checked" } } if options[:link_checked].present?
-      filter << { term: { "landing_page.hasSchemaOrg": options[:link_check_has_schema_org] } } if options[:link_check_has_schema_org].present?
-      filter << { term: { "landing_page.bodyHasPid": options[:link_check_body_has_pid] } } if options[:link_check_body_has_pid].present?
+      filter << { terms: { "landing_page.hasSchemaOrg": [options[:link_check_has_schema_org]] } } if options[:link_check_has_schema_org].present?
+      filter << { terms: { "landing_page.bodyHasPid": [options[:link_check_body_has_pid]] } } if options[:link_check_body_has_pid].present?
       filter << { exists: { field: "landing_page.schemaOrgId" } } if options[:link_check_found_schema_org_id].present?
       filter << { exists: { field: "landing_page.dcIdentifier" } } if options[:link_check_found_dc_identifier].present?
       filter << { exists: { field: "landing_page.citationDoi" } } if options[:link_check_found_citation_doi].present?
       filter << { range: { "landing_page.redirectCount": { "gte": options[:link_check_redirect_count_gte] } } } if options[:link_check_redirect_count_gte].present?
       filter << { terms: { aasm_state: options[:state].to_s.split(",") } } if options[:state].present?
       filter << { range: { registered: { gte: "#{options[:registered].split(',').min}||/y", lte: "#{options[:registered].split(',').max}||/y", format: "yyyy" } } } if options[:registered].present?
-      filter << { term: { consortium_id: { value: options[:consortium_id], case_insensitive: true } } } if options[:consortium_id].present?
+      filter << { terms: { consortium_id: [options[:consortium_id].downcase] } } if options[:consortium_id].present?
       # TODO align PID parsing
-      filter << { term: { "client.re3data_id" => doi_from_url(options[:re3data_id]) } } if options[:re3data_id].present?
-      filter << { term: { "client.opendoar_id" => options[:opendoar_id] } } if options[:opendoar_id].present?
+      filter << { terms: { "client.re3data_id": [doi_from_url(options[:re3data_id])] } } if options[:re3data_id].present?
+      filter << { terms: { "client.opendoar_id": [options[:opendoar_id]] } } if options[:opendoar_id].present?
       filter << { terms: { "client.certificate" => options[:certificate].split(",") } } if options[:certificate].present?
       filter << { terms: { "creators.nameIdentifiers.nameIdentifier" => options[:user_id].split(",").collect { |id| "https://orcid.org/#{orcid_from_url(id)}" } } } if options[:user_id].present?
-      filter << { term: { "creators.nameIdentifiers.nameIdentifierScheme" => "ORCID" } } if options[:has_person].present?
-      filter << { term: { "client.client_type" =>  options[:client_type] } } if options[:client_type]
-      filter << { term: { "types.resourceTypeGeneral" => "PhysicalObject" } } if options[:client_type] == "igsnCatalog"
+      filter << { terms: { "creators.nameIdentifiers.nameIdentifierScheme": ["ORCID"] } } if options[:has_person].present?
+      filter << { terms: { "client.client_type": [options[:client_type]] } } if options[:client_type]
+      filter << { terms: { "types.resourceTypeGeneral": ["PhysicalObject"] } } if options[:client_type] == "igsnCatalog"
 
       filter
     end
diff --git a/spec/models/doi/graphql_query_builder_filters_spec.rb b/spec/models/doi/graphql_query_builder_filters_spec.rb
index 261ef7af2..5953846a0 100644
--- a/spec/models/doi/graphql_query_builder_filters_spec.rb
+++ b/spec/models/doi/graphql_query_builder_filters_spec.rb
@@ -1,5 +1,3 @@
-
-
 # frozen_string_literal: true
 
 require "rails_helper"
@@ -22,7 +20,7 @@
         options = { resource_type_id: "Journal_Article" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { resource_type_id: "journal-article" } }
+          { terms: { resource_type_id: ["journal-article"] } }
         )
       end
 
@@ -63,7 +61,7 @@
         options = { uid: "10.5438/0012" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { uid: "10.5438/0012" } }
+          { terms: { uid: ["10.5438/0012"] } }
         )
       end
 
@@ -79,7 +77,15 @@
         options = { consortium_id: "dc" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { consortium_id: { case_insensitive: true, value: "dc" } } }
+          { terms: { consortium_id: ["dc"] } }
+        )
+      end
+
+      it "handles registered" do
+        options = { registered: "2021,2023" }
+        builder = described_class.new(query, options)
+        expect(builder.filters).to include(
+          { range: { registered: { gte: "2021||/y", lte: "2023||/y", format: "yyyy" } } }
         )
       end
 
@@ -97,7 +103,7 @@
         options = { re3data_id: "10.17616/r31njmjx" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "client.re3data_id" => "10.17616/r31njmjx" } }
+          { terms: { "client.re3data_id": ["10.17616/r31njmjx"] } }
         )
       end
 
@@ -105,7 +111,15 @@
         options = { opendoar_id: "123456" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "client.opendoar_id" => "123456" } }
+          { terms: { "client.opendoar_id": ["123456"] } }
+        )
+      end
+
+      it "handles certificates" do
+        options = { certificate: "CoreTrustSeal,WDS" }
+        builder = described_class.new(query, options)
+        expect(builder.filters).to include(
+          { terms: { "client.certificate" => ["CoreTrustSeal", "WDS"] } }
         )
       end
 
@@ -207,7 +221,7 @@
         options = { pid_entity: "dataset,software" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "subjects.subjectScheme": "PidEntity" } },
+          { terms: { "subjects.subjectScheme": ["PidEntity"] } },
           { terms: { "subjects.subject": ["Dataset", "Software"] } }
         )
       end
@@ -216,7 +230,7 @@
         options = { field_of_science: "computer_science,mathematics" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "subjects.subjectScheme": "Fields of Science and Technology (FOS)" } },
+          { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } },
           { terms: { "subjects.subject": ["FOS: Computer science", "FOS: Mathematics"] } }
         )
       end
@@ -227,7 +241,7 @@
         options = { link_check_status: "200" }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "landing_page.status": "200" } }
+          { terms: { "landing_page.status": ["200"] } }
         )
       end
 
@@ -235,7 +249,7 @@
         options = { link_check_has_schema_org: true }
         builder = described_class.new(query, options)
         expect(builder.filters).to include(
-          { term: { "landing_page.hasSchemaOrg": true } }
+          { terms: { "landing_page.hasSchemaOrg": [true] } }
         )
       end
     end

From 10c444cd3a138be1924bc0ab39f1b7b6c2237854 Mon Sep 17 00:00:00 2001
From: jrhoads <jrhoads@datacite.org>
Date: Mon, 9 Dec 2024 22:06:46 +0100
Subject: [PATCH 2/4] Reorder filter terms.  Group by type of field

---
 app/models/doi/graphql_query.rb | 33 ++++++++++++++++-----------------
 1 file changed, 16 insertions(+), 17 deletions(-)

diff --git a/app/models/doi/graphql_query.rb b/app/models/doi/graphql_query.rb
index 391dba466..3780e0f02 100644
--- a/app/models/doi/graphql_query.rb
+++ b/app/models/doi/graphql_query.rb
@@ -125,30 +125,16 @@ def filters
 
       filter = []
       filter << { terms: { doi: options[:ids].map(&:upcase) } } if options[:ids].present?
+      filter << { terms: { uid: [options[:uid]] } } if options[:uid].present?
       filter << { terms: { resource_type_id: [options[:resource_type_id].underscore.dasherize] } } if options[:resource_type_id].present?
       filter << { terms: { "types.resourceType": options[:resource_type].split(",") } } if options[:resource_type].present?
       filter << { terms: { agency: options[:agency].split(",").map(&:downcase) } } if options[:agency].present?
       filter << { terms: { prefix: options[:prefix].to_s.split(",") } } if options[:prefix].present?
       filter << { terms: { language: options[:language].to_s.split(",").map(&:downcase) } } if options[:language].present?
-      filter << { terms: { uid: [options[:uid]] } } if options[:uid].present?
       filter << { range: { created: { gte: "#{options[:created].split(',').min}||/y", lte: "#{options[:created].split(',').max}||/y", format: "yyyy" } } } if options[:created].present?
       filter << { range: { publication_year: { gte: "#{options[:published].split(',').min}||/y", lte: "#{options[:published].split(',').max}||/y", format: "yyyy" } } } if options[:published].present?
       filter << { terms: { schema_version: ["http://datacite.org/schema/kernel-#{options[:schema_version]}"] } } if options[:schema_version].present?
       filter << { terms: { "subjects.subject": options[:subject].split(",") } } if options[:subject].present?
-      if options[:pid_entity].present?
-        filter << { terms: { "subjects.subjectScheme": ["PidEntity"] } }
-        filter << { terms: { "subjects.subject": options[:pid_entity].split(",").map(&:humanize) } }
-      end
-      if options[:field_of_science].present?
-        filter << { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } }
-        filter << { terms: { "subjects.subject": options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
-      end
-      if options[:field_of_science_repository].present?
-        filter << { terms: { "fields_of_science_repository": options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
-      end
-      if options[:field_of_science_combined].present?
-        filter << { terms: { "fields_of_science_combined": options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
-      end
       filter << { terms: { "rights_list.rightsIdentifier" => options[:license].split(",") } } if options[:license].present?
       filter << { terms: { source: [options[:source]] } } if options[:source].present?
       filter << { range: { reference_count: { "gte": options[:has_references].to_i } } } if options[:has_references].present?
@@ -170,14 +156,27 @@ def filters
       filter << { terms: { aasm_state: options[:state].to_s.split(",") } } if options[:state].present?
       filter << { range: { registered: { gte: "#{options[:registered].split(',').min}||/y", lte: "#{options[:registered].split(',').max}||/y", format: "yyyy" } } } if options[:registered].present?
       filter << { terms: { consortium_id: [options[:consortium_id].downcase] } } if options[:consortium_id].present?
-      # TODO align PID parsing
-      filter << { terms: { "client.re3data_id": [doi_from_url(options[:re3data_id])] } } if options[:re3data_id].present?
+      filter << { terms: { "client.re3data_id": [doi_from_url(options[:re3data_id])] } } if options[:re3data_id].present? # TODO align PID parsing
       filter << { terms: { "client.opendoar_id": [options[:opendoar_id]] } } if options[:opendoar_id].present?
       filter << { terms: { "client.certificate" => options[:certificate].split(",") } } if options[:certificate].present?
       filter << { terms: { "creators.nameIdentifiers.nameIdentifier" => options[:user_id].split(",").collect { |id| "https://orcid.org/#{orcid_from_url(id)}" } } } if options[:user_id].present?
       filter << { terms: { "creators.nameIdentifiers.nameIdentifierScheme": ["ORCID"] } } if options[:has_person].present?
       filter << { terms: { "client.client_type": [options[:client_type]] } } if options[:client_type]
       filter << { terms: { "types.resourceTypeGeneral": ["PhysicalObject"] } } if options[:client_type] == "igsnCatalog"
+      if options[:pid_entity].present?
+        filter << { terms: { "subjects.subjectScheme": ["PidEntity"] } }
+        filter << { terms: { "subjects.subject": options[:pid_entity].split(",").map(&:humanize) } }
+      end
+      if options[:field_of_science].present?
+        filter << { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } }
+        filter << { terms: { "subjects.subject": options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
+      end
+      if options[:field_of_science_repository].present?
+        filter << { terms: { "fields_of_science_repository": options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
+      end
+      if options[:field_of_science_combined].present?
+        filter << { terms: { "fields_of_science_combined": options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
+      end
 
       filter
     end

From 948ea00164117c8a0bc38933a94b7e4820d39d3f Mon Sep 17 00:00:00 2001
From: jrhoads <jrhoads@datacite.org>
Date: Thu, 12 Dec 2024 15:44:48 +0100
Subject: [PATCH 3/4] Refactor filter builder into its own class.  Call that
 from the graphql query builder

---
 app/models/doi/graphql_query.rb         | 62 +----------------
 app/models/doi/search/filter_builder.rb | 91 +++++++++++++++++++++++++
 2 files changed, 92 insertions(+), 61 deletions(-)
 create mode 100644 app/models/doi/search/filter_builder.rb

diff --git a/app/models/doi/graphql_query.rb b/app/models/doi/graphql_query.rb
index 3780e0f02..571fce420 100644
--- a/app/models/doi/graphql_query.rb
+++ b/app/models/doi/graphql_query.rb
@@ -118,67 +118,7 @@ def must
     end
 
     def filters
-      options = @options
-
-      # turn ids into an array if provided as comma-separated string
-      options[:ids] = options[:ids].split(",") if options[:ids].is_a?(String)
-
-      filter = []
-      filter << { terms: { doi: options[:ids].map(&:upcase) } } if options[:ids].present?
-      filter << { terms: { uid: [options[:uid]] } } if options[:uid].present?
-      filter << { terms: { resource_type_id: [options[:resource_type_id].underscore.dasherize] } } if options[:resource_type_id].present?
-      filter << { terms: { "types.resourceType": options[:resource_type].split(",") } } if options[:resource_type].present?
-      filter << { terms: { agency: options[:agency].split(",").map(&:downcase) } } if options[:agency].present?
-      filter << { terms: { prefix: options[:prefix].to_s.split(",") } } if options[:prefix].present?
-      filter << { terms: { language: options[:language].to_s.split(",").map(&:downcase) } } if options[:language].present?
-      filter << { range: { created: { gte: "#{options[:created].split(',').min}||/y", lte: "#{options[:created].split(',').max}||/y", format: "yyyy" } } } if options[:created].present?
-      filter << { range: { publication_year: { gte: "#{options[:published].split(',').min}||/y", lte: "#{options[:published].split(',').max}||/y", format: "yyyy" } } } if options[:published].present?
-      filter << { terms: { schema_version: ["http://datacite.org/schema/kernel-#{options[:schema_version]}"] } } if options[:schema_version].present?
-      filter << { terms: { "subjects.subject": options[:subject].split(",") } } if options[:subject].present?
-      filter << { terms: { "rights_list.rightsIdentifier" => options[:license].split(",") } } if options[:license].present?
-      filter << { terms: { source: [options[:source]] } } if options[:source].present?
-      filter << { range: { reference_count: { "gte": options[:has_references].to_i } } } if options[:has_references].present?
-      filter << { range: { citation_count: { "gte": options[:has_citations].to_i } } } if options[:has_citations].present?
-      filter << { range: { part_count: { "gte": options[:has_parts].to_i } } } if options[:has_parts].present?
-      filter << { range: { part_of_count: { "gte": options[:has_part_of].to_i } } } if options[:has_part_of].present?
-      filter << { range: { version_count: { "gte": options[:has_versions].to_i } } } if options[:has_versions].present?
-      filter << { range: { version_of_count: { "gte": options[:has_version_of].to_i } } } if options[:has_version_of].present?
-      filter << { range: { view_count: { "gte": options[:has_views].to_i } } } if options[:has_views].present?
-      filter << { range: { download_count: { "gte": options[:has_downloads].to_i } } } if options[:has_downloads].present?
-      filter << { terms: { "landing_page.status": [options[:link_check_status]] } } if options[:link_check_status].present?
-      filter << { exists: { field: "landing_page.checked" } } if options[:link_checked].present?
-      filter << { terms: { "landing_page.hasSchemaOrg": [options[:link_check_has_schema_org]] } } if options[:link_check_has_schema_org].present?
-      filter << { terms: { "landing_page.bodyHasPid": [options[:link_check_body_has_pid]] } } if options[:link_check_body_has_pid].present?
-      filter << { exists: { field: "landing_page.schemaOrgId" } } if options[:link_check_found_schema_org_id].present?
-      filter << { exists: { field: "landing_page.dcIdentifier" } } if options[:link_check_found_dc_identifier].present?
-      filter << { exists: { field: "landing_page.citationDoi" } } if options[:link_check_found_citation_doi].present?
-      filter << { range: { "landing_page.redirectCount": { "gte": options[:link_check_redirect_count_gte] } } } if options[:link_check_redirect_count_gte].present?
-      filter << { terms: { aasm_state: options[:state].to_s.split(",") } } if options[:state].present?
-      filter << { range: { registered: { gte: "#{options[:registered].split(',').min}||/y", lte: "#{options[:registered].split(',').max}||/y", format: "yyyy" } } } if options[:registered].present?
-      filter << { terms: { consortium_id: [options[:consortium_id].downcase] } } if options[:consortium_id].present?
-      filter << { terms: { "client.re3data_id": [doi_from_url(options[:re3data_id])] } } if options[:re3data_id].present? # TODO align PID parsing
-      filter << { terms: { "client.opendoar_id": [options[:opendoar_id]] } } if options[:opendoar_id].present?
-      filter << { terms: { "client.certificate" => options[:certificate].split(",") } } if options[:certificate].present?
-      filter << { terms: { "creators.nameIdentifiers.nameIdentifier" => options[:user_id].split(",").collect { |id| "https://orcid.org/#{orcid_from_url(id)}" } } } if options[:user_id].present?
-      filter << { terms: { "creators.nameIdentifiers.nameIdentifierScheme": ["ORCID"] } } if options[:has_person].present?
-      filter << { terms: { "client.client_type": [options[:client_type]] } } if options[:client_type]
-      filter << { terms: { "types.resourceTypeGeneral": ["PhysicalObject"] } } if options[:client_type] == "igsnCatalog"
-      if options[:pid_entity].present?
-        filter << { terms: { "subjects.subjectScheme": ["PidEntity"] } }
-        filter << { terms: { "subjects.subject": options[:pid_entity].split(",").map(&:humanize) } }
-      end
-      if options[:field_of_science].present?
-        filter << { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } }
-        filter << { terms: { "subjects.subject": options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
-      end
-      if options[:field_of_science_repository].present?
-        filter << { terms: { "fields_of_science_repository": options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
-      end
-      if options[:field_of_science_combined].present?
-        filter << { terms: { "fields_of_science_combined": options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
-      end
-
-      filter
+      Doi::Search::FilterBuilder.new(@options).build
     end
 
     def get_should_clause
diff --git a/app/models/doi/search/filter_builder.rb b/app/models/doi/search/filter_builder.rb
new file mode 100644
index 000000000..f1e0955f1
--- /dev/null
+++ b/app/models/doi/search/filter_builder.rb
@@ -0,0 +1,91 @@
+# frozen_string_literal: true
+
+class Doi
+  module Search
+    class FilterBuilder
+      include Modelable
+
+      def initialize(options)
+        @options = options
+      end
+
+      def build
+        options = @options
+
+        # turn ids into an array if provided as comma-separated string
+        options[:ids] = options[:ids].split(",") if options[:ids].is_a?(String)
+
+        filter = []
+        filter << { terms: { doi: options[:ids].map(&:upcase) } } if options[:ids].present?
+        filter << { terms: { uid: [options[:uid]] } } if options[:uid].present?
+        filter << { terms: { resource_type_id: [options[:resource_type_id].underscore.dasherize] } } if options[:resource_type_id].present?
+        filter << { terms: { "types.resourceType": options[:resource_type].split(",") } } if options[:resource_type].present?
+        filter << { terms: { agency: options[:agency].split(",").map(&:downcase) } } if options[:agency].present?
+        filter << { terms: { prefix: options[:prefix].to_s.split(",") } } if options[:prefix].present?
+        filter << { terms: { language: options[:language].to_s.split(",").map(&:downcase) } } if options[:language].present?
+        filter << { range: { created: { gte: "#{options[:created].split(',').min}||/y", lte: "#{options[:created].split(',').max}||/y", format: "yyyy" } } } if options[:created].present?
+        filter << { range: { publication_year: { gte: "#{options[:published].split(',').min}||/y", lte: "#{options[:published].split(',').max}||/y", format: "yyyy" } } } if options[:published].present?
+        filter << { terms: { schema_version: ["http://datacite.org/schema/kernel-#{options[:schema_version]}"] } } if options[:schema_version].present?
+        filter << { terms: { "subjects.subject": options[:subject].split(",") } } if options[:subject].present?
+        filter << { terms: { "rights_list.rightsIdentifier" => options[:license].split(",") } } if options[:license].present?
+        filter << { terms: { source: [options[:source]] } } if options[:source].present?
+        filter << { range: { reference_count: { "gte": options[:has_references].to_i } } } if options[:has_references].present?
+        filter << { range: { citation_count: { "gte": options[:has_citations].to_i } } } if options[:has_citations].present?
+        filter << { range: { part_count: { "gte": options[:has_parts].to_i } } } if options[:has_parts].present?
+        filter << { range: { part_of_count: { "gte": options[:has_part_of].to_i } } } if options[:has_part_of].present?
+        filter << { range: { version_count: { "gte": options[:has_versions].to_i } } } if options[:has_versions].present?
+        filter << { range: { version_of_count: { "gte": options[:has_version_of].to_i } } } if options[:has_version_of].present?
+        filter << { range: { view_count: { "gte": options[:has_views].to_i } } } if options[:has_views].present?
+        filter << { range: { download_count: { "gte": options[:has_downloads].to_i } } } if options[:has_downloads].present?
+        filter << { terms: { "landing_page.status": [options[:link_check_status]] } } if options[:link_check_status].present?
+        filter << { exists: { field: "landing_page.checked" } } if options[:link_checked].present?
+        filter << { terms: { "landing_page.hasSchemaOrg": [options[:link_check_has_schema_org]] } } if options[:link_check_has_schema_org].present?
+        filter << { terms: { "landing_page.bodyHasPid": [options[:link_check_body_has_pid]] } } if options[:link_check_body_has_pid].present?
+        filter << { exists: { field: "landing_page.schemaOrgId" } } if options[:link_check_found_schema_org_id].present?
+        filter << { exists: { field: "landing_page.dcIdentifier" } } if options[:link_check_found_dc_identifier].present?
+        filter << { exists: { field: "landing_page.citationDoi" } } if options[:link_check_found_citation_doi].present?
+        filter << { range: { "landing_page.redirectCount": { "gte": options[:link_check_redirect_count_gte] } } } if options[:link_check_redirect_count_gte].present?
+        filter << { terms: { aasm_state: options[:state].to_s.split(",") } } if options[:state].present?
+        filter << { range: { registered: { gte: "#{options[:registered].split(',').min}||/y", lte: "#{options[:registered].split(',').max}||/y", format: "yyyy" } } } if options[:registered].present?
+        filter << { terms: { consortium_id: [options[:consortium_id].downcase] } } if options[:consortium_id].present?
+        filter << { terms: { "client.re3data_id": [doi_from_url(options[:re3data_id])] } } if options[:re3data_id].present? # TODO align PID parsing
+        filter << { terms: { "client.opendoar_id": [options[:opendoar_id]] } } if options[:opendoar_id].present?
+        filter << { terms: { "client.certificate" => options[:certificate].split(",") } } if options[:certificate].present?
+        filter << { terms: { "creators.nameIdentifiers.nameIdentifier" => options[:user_id].split(",").collect { |id| "https://orcid.org/#{orcid_from_url(id)}" } } } if options[:user_id].present?
+        filter << { terms: { "creators.nameIdentifiers.nameIdentifierScheme": ["ORCID"] } } if options[:has_person].present?
+        filter << { terms: { "client.client_type": [options[:client_type]] } } if options[:client_type]
+        filter << { terms: { "types.resourceTypeGeneral": ["PhysicalObject"] } } if options[:client_type] == "igsnCatalog"
+        filter.push(*build_pid_entity_filter) if options[:pid_entity].present?
+        filter.push(*build_field_of_science_filter) if options[:field_of_science].present?
+        filter << build_field_of_science_repository_filter if options[:field_of_science_repository].present?
+        filter << build_field_of_science_combined_filter if options[:field_of_science_combined].present?
+
+        filter
+      end
+
+      private
+
+      def build_pid_entity_filter
+        [
+          { terms: { "subjects.subjectScheme": ["PidEntity"] } },
+          { terms: { "subjects.subject": @options[:pid_entity].split(",").map(&:humanize) } }
+        ]
+      end
+
+      def build_field_of_science_filter
+        [
+          { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } },
+          { terms: { "subjects.subject": @options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
+        ]
+      end
+
+      def build_field_of_science_repository_filter
+        { terms: { "fields_of_science_repository": @options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
+      end
+
+      def build_field_of_science_combined_filter
+        { terms: { "fields_of_science_combined": @options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
+      end
+    end
+  end
+end

From 24537887a543589a57aa729b484ab61d46a4f30b Mon Sep 17 00:00:00 2001
From: jrhoads <jrhoads@datacite.org>
Date: Thu, 12 Dec 2024 15:46:33 +0100
Subject: [PATCH 4/4] Appease Rubocop

---
 app/models/doi/search/filter_builder.rb | 37 ++++++++++++-------------
 1 file changed, 18 insertions(+), 19 deletions(-)

diff --git a/app/models/doi/search/filter_builder.rb b/app/models/doi/search/filter_builder.rb
index f1e0955f1..76c9ab3f4 100644
--- a/app/models/doi/search/filter_builder.rb
+++ b/app/models/doi/search/filter_builder.rb
@@ -64,28 +64,27 @@ def build
       end
 
       private
+        def build_pid_entity_filter
+          [
+            { terms: { "subjects.subjectScheme": ["PidEntity"] } },
+            { terms: { "subjects.subject": @options[:pid_entity].split(",").map(&:humanize) } }
+          ]
+        end
 
-      def build_pid_entity_filter
-        [
-          { terms: { "subjects.subjectScheme": ["PidEntity"] } },
-          { terms: { "subjects.subject": @options[:pid_entity].split(",").map(&:humanize) } }
-        ]
-      end
+        def build_field_of_science_filter
+          [
+            { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } },
+            { terms: { "subjects.subject": @options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
+          ]
+        end
 
-      def build_field_of_science_filter
-        [
-          { terms: { "subjects.subjectScheme": ["Fields of Science and Technology (FOS)"] } },
-          { terms: { "subjects.subject": @options[:field_of_science].split(",").map { |s| "FOS: " + s.humanize } } }
-        ]
-      end
+        def build_field_of_science_repository_filter
+          { terms: { "fields_of_science_repository": @options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
+        end
 
-      def build_field_of_science_repository_filter
-        { terms: { "fields_of_science_repository": @options[:field_of_science_repository].split(",").map { |s| s.humanize } } }
-      end
-
-      def build_field_of_science_combined_filter
-        { terms: { "fields_of_science_combined": @options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
-      end
+        def build_field_of_science_combined_filter
+          { terms: { "fields_of_science_combined": @options[:field_of_science_combined].split(",").map { |s| s.humanize } } }
+        end
     end
   end
 end