misc other fixes for loading openalex snapshot version

rails/Dockerfile

@@ -9,6 +9,7 @@ WORKDIR /rails
 
 # Set production environment
 ENV RAILS_ENV="production" \
+    RAILS_MASTER_KEY=$RAILS_MASTER_KEY \ 
     BUNDLE_DEPLOYMENT="1" \
     BUNDLE_PATH="/usr/local/bundle" \
     BUNDLE_WITHOUT="development"

rails/app/graphql/types/author_type.rb

@@ -12,11 +12,24 @@ class AuthorType < Types::BaseObject
     field :last_known_institution, String
     field :works_api_url, String
 
-    field :authors_ids, Types::AuthorsIdsType
     field :works, [Types::WorkType]
+    def works
+      object.works
+    end
 
+    field :authors_ids, Types::AuthorsIdsType
     def authors_ids
       object.authors_ids
     end
+
+    field :institutions, [Types::InstitutionType]
+    def institutions
+      object.institutions.uniq # move this to the Author model probably
+    end
+
+    field :authors_counts_by_year, [Types::AuthorsCountsByYearType]
+    def authors_counts_by_year
+      object.authors_counts_by_year
+    end
   end
 end

rails/app/graphql/types/query_type.rb

@@ -24,23 +24,24 @@ def nodes(ids:)
       ids.map { |id| context.schema.object_from_id(id, context) }
     end
 
-    field :institution_by_openalex_id,
-          Types::InstitutionType,
-          null: true,
-          description: 'Fetches an institution by institution_openalex_id' do
-      argument :institution_openalex_id, String, required: true
-    end
-    def institution(institution_openalex_id:)
-      Institution.find_by(institution_openalex_id: institution_openalex_id)
-    end
+    # need to index institutions by openalex id
+    # field :institution_by_openalex_id,
+    #       Types::InstitutionType,
+    #       null: true,
+    #       description: 'Fetches an institution by institution_openalex_id' do
+    #   argument :institution_openalex_id, String, required: true
+    # end
+    # def institution_by_openalex_id(institution_openalex_id:)
+    #   Institution.find_by(institution_openalex_id: institution_openalex_id)
+    # end
 
     field :institution_by_ror,
           Types::InstitutionType,
           null: true,
           description: 'Fetches an institution by ROR' do
       argument :ror, String, required: true
     end
-    def institution(institution_openalex_id:)
+    def institution_by_ror(ror:)
       Institution.find_by(ror: ror)
     end
 
@@ -63,7 +64,12 @@ def author_by_openalex_id(author_openalex_id:)
       argument :orcid, String, required: true
     end
     def author_by_orcid(orcid:)
-      Author.find_by(orcid: orcid)
+      # do some error handling for invalid entries
+      if "https://orcid.org/".in? orcid
+        Author.find_by(orcid: orcid)
+      else
+        Author.find_by(orcid: "https://orcid.org/" + orcid)
+      end
     end
 
     # work entry points

rails/app/graphql/types/topic_type.rb

@@ -12,12 +12,16 @@ class TopicType < Types::BaseObject
     field :openalex_domain_id, Integer
     field :openalex_domain_display_name, String
     field :description, String
-    field :keywords, String
     field :works_api_url, String
     field :wikipedia_id, String
     field :works_count, Integer
     field :cited_by_count, Integer
     field :created_at, GraphQL::Types::ISO8601DateTime, null: false
     field :updated_at, GraphQL::Types::ISO8601DateTime, null: false
+
+    field :keywords, [String]
+    def keywords
+      object.keywords
+    end
   end
 end

rails/app/graphql/types/work_type.rb

@@ -53,5 +53,10 @@ def works_biblio
     def topics
       object.topics
     end
+
+    field :works_ids, Types::WorksIdsType
+    def works_ids
+      object.works_ids
+    end
   end
 end

rails/app/models/author.rb

@@ -4,12 +4,15 @@ class Author < ApplicationRecord
   has_one :authors_ids,
           primary_key: :author_openalex_id,
           foreign_key: :author_openalex_id
+
   has_many :authors_counts_by_year,
            primary_key: :author_openalex_id,
            foreign_key: :author_openalex_id
+
   has_many :works_authorships,
            primary_key: :author_openalex_id,
            foreign_key: :author_openalex_id
+
   has_many :institutions, through: :works_authorships
   has_many :works, through: :works_authorships
 end

rails/config/database.yml

@@ -3,17 +3,17 @@ default: &default
   encoding: unicode
   username: bookworm
   # password: <%= ENV["POSTGRES_PASSWORD"] %>
-  password: password
   pool: 5
 
 development:
-  host: localhost
   <<: *default
+  host: localhost
+  password: password
   database: bookworm_development
 
 test:
-  host: localhost
   <<: *default
+  host: localhost
   database: bookworm_test
   password: password
 

rails/config/environments/development.rb

@@ -3,6 +3,8 @@
 require 'active_support/core_ext/integer/time'
 
 Rails.application.configure do
+  config.hosts << 'ec2-34-215-107-32.us-west-2.compute.amazonaws.com:3000'
+  config.hosts << 'ec2-35-94-47-108.us-west-2.compute.amazonaws.com:3000'
   config.log_level = :debug
   # Settings specified here will take precedence over those in config/application.rb.
 

rails/db/seeds.rb

@@ -98,7 +98,7 @@
 # end
 
 # # create initial work node
-# BookWormApiSchema.execute(
+# BookWormSchema.execute(
 #   sign_in_mutation,
 #   variables: {
 #     email: user_1.email,
@@ -107,7 +107,7 @@
 # )
 
 # work_node_response =
-#   BookWormApiSchema.execute(
+#   BookWormSchema.execute(
 #     create_work_node_mutation,
 #     variables: {
 #       investigationId: investigation_1.id.to_s,
@@ -122,7 +122,7 @@
 #   work_node_response['data']['createWorkNode']['work']['openalexId']
 
 # response =
-#   BookWormApiSchema.execute(
+#   BookWormSchema.execute(
 #     add_references_mutation,
 #     variables: {
 #       openalexId: work_node_openalex_id,
@@ -134,7 +134,7 @@
 #   ).to_h
 
 # investigation_response =
-#   BookWormApiSchema.execute(
+#   BookWormSchema.execute(
 #     investigation_query,
 #     variables: {
 #       id: investigation_1.id

rails/lib/tasks/load_works.rake

@@ -30,7 +30,7 @@ namespace :data_import do
         language: row[12]
       }
 
-      if works.count >= 5000
+      if works.count >= 10000
         Work.insert_all(works)
         works = []
       end