Merge pull request #1448 from sul-dlss/geo-aardvark

Add Aardvark indexing config for Earthworks

Gemfile

@@ -42,5 +42,7 @@ end
 gem 'activesupport', '~> 7.0'
 gem 'slop'
 
+gem 'cocina-models'
 gem 'dor-event-client'
 gem 'factory_bot', '~> 6.2'
+gem 'stanford-geo', '0.2.0'

Gemfile.lock

@@ -33,6 +33,7 @@ GEM
       sshkit (>= 1.6.1, != 1.7.0)
     amq-protocol (2.3.2)
     ast (2.4.2)
+    attr_extras (7.1.0)
     base64 (0.2.0)
     bcrypt_pbkdf (1.1.1)
     bcrypt_pbkdf (1.1.1-arm64-darwin)
@@ -63,6 +64,24 @@ GEM
       sshkit (~> 1.2)
     capistrano-shared_configs (0.2.2)
     chronic (0.10.2)
+    cocina-models (0.97.0)
+      activesupport
+      deprecation
+      dry-struct (~> 1.0)
+      dry-types (~> 1.1)
+      edtf
+      equivalent-xml
+      i18n
+      jsonpath
+      nokogiri
+      openapi3_parser
+      openapi_parser (~> 1.0)
+      super_diff
+      thor
+      zeitwerk (~> 2.1)
+    commonmarker (1.1.4-arm64-darwin)
+    commonmarker (1.1.4-x86_64-darwin)
+    commonmarker (1.1.4-x86_64-linux)
     concurrent-ruby (1.3.3)
     config (5.5.1)
       deep_merge (~> 1.2, >= 1.2.1)
@@ -77,6 +96,8 @@ GEM
       irb (~> 1.10)
       reline (>= 0.3.8)
     deep_merge (1.2.2)
+    deprecation (1.1.0)
+      activesupport
     diff-lcs (1.5.1)
     digest-crc (0.6.5)
       rake (>= 12.0.0, < 14.0.0)
@@ -98,9 +119,31 @@ GEM
     dot-properties (0.1.4)
       bundler (>= 2.2.33)
     drb (2.2.1)
+    dry-core (1.0.1)
+      concurrent-ruby (~> 1.0)
+      zeitwerk (~> 2.6)
+    dry-inflector (1.0.0)
+    dry-logic (1.5.0)
+      concurrent-ruby (~> 1.0)
+      dry-core (~> 1.0, < 2)
+      zeitwerk (~> 2.6)
+    dry-struct (1.6.0)
+      dry-core (~> 1.0, < 2)
+      dry-types (>= 1.7, < 2)
+      ice_nine (~> 0.11)
+      zeitwerk (~> 2.6)
+    dry-types (1.7.2)
+      bigdecimal (~> 3.0)
+      concurrent-ruby (~> 1.0)
+      dry-core (~> 1.0)
+      dry-inflector (~> 1.0)
+      dry-logic (~> 1.4)
+      zeitwerk (~> 2.6)
     ed25519 (1.3.0)
     edtf (3.1.1)
       activesupport (>= 3.0, < 8.0)
+    equivalent-xml (0.6.0)
+      nokogiri (>= 1.4.3)
     erubi (1.13.0)
     factory_bot (6.4.6)
       activesupport (>= 5.0.0)
@@ -125,12 +168,15 @@ GEM
     httpclient (2.8.3)
     i18n (1.14.5)
       concurrent-ruby (~> 1.0)
+    ice_nine (0.11.2)
     io-console (0.7.2)
     irb (1.13.2)
       rdoc (>= 4.0.0)
       reline (>= 0.4.2)
     iso-639 (0.3.6)
     json (2.7.2)
+    jsonpath (1.1.5)
+      multi_json
     language_server-protocol (3.17.0.3)
     llhttp-ffi (0.5.0)
       ffi-compiler (~> 1.0)
@@ -156,6 +202,7 @@ GEM
       rails_autolink
       stanford-mods (~> 3.3, >= 3.3.9)
       view_component
+    multi_json (1.15.0)
     mutex_m (0.2.0)
     net-scp (4.0.0)
       net-ssh (>= 2.6.5, < 8.0.0)
@@ -171,10 +218,16 @@ GEM
     nom-xml (1.2.0)
       i18n
       nokogiri
+    openapi3_parser (0.10.0)
+      commonmarker (>= 1.0)
+    openapi_parser (1.0.0)
+    optimist (3.1.0)
     parallel (1.25.1)
     parser (3.3.3.0)
       ast (~> 2.4.1)
       racc
+    patience_diff (1.2.0)
+      optimist (~> 3.0)
     pg (1.5.6)
     psych (5.1.2)
       stringio
@@ -264,12 +317,17 @@ GEM
       net-scp (>= 1.1.2)
       net-sftp (>= 2.1.2)
       net-ssh (>= 2.8.0)
+    stanford-geo (0.2.0)
     stanford-mods (3.3.9)
       activesupport
       mods (~> 3.0, >= 3.0.4)
     statsd-ruby (1.5.0)
     stringio (3.1.1)
     strscan (3.1.0)
+    super_diff (0.12.1)
+      attr_extras (>= 6.2.4)
+      diff-lcs
+      patience_diff
     thor (1.3.1)
     traject (3.8.2)
       concurrent-ruby (>= 0.8.0)
@@ -315,6 +373,7 @@ DEPENDENCIES
   capistrano-bundler
   capistrano-rvm
   capistrano-shared_configs
+  cocina-models
   config
   csv
   debouncer
@@ -337,6 +396,7 @@ DEPENDENCIES
   ruby-kafka
   simplecov
   slop
+  stanford-geo (= 0.2.0)
   stanford-mods (~> 3.0)
   statsd-ruby
   traject (~> 3.0)

config/deploy/stage.rb

@@ -22,12 +22,12 @@
   {
     key: 'earthworks_stage_indexer',
     count: 1,
-    config: './lib/traject/config/geo_config.rb',
+    config: './lib/traject/config/geo_aardvark_config.rb',
     settings: {
       'log.file' => 'log/traject_earthworks-stage-indexer.log',
       'kafka.topic' => 'purl_fetcher_stage',
       'kafka.consumer_group_id' => 'earthworks-stage-indexer',
-      'solr.url' => 'http://sul-solr.stanford.edu/solr/earthworks-stage',
+      'solr.url' => 'http://sul-solr.stanford.edu/solr/earthworks-aardvark-stage',
       'purl.url' => 'https://sul-purl-stage.stanford.edu',
       'stacks.url' => 'https://sul-stacks-stage.stanford.edu',
       'geoserver.pub_url' => 'https://earthworks-geoserver-stage-b.stanford.edu/geoserver',

lib/public_cocina_record.rb

@@ -3,27 +3,74 @@
 require 'http'
 
 class PublicCocinaRecord
-  attr_reader :druid, :purl_url
+  attr_reader :public_cocina_doc, :druid, :purl_url
 
-  def self.fetch(url)
-    response = HTTP.get(url)
-    response.body if response.status.ok?
+  def self.fetch(druid, purl_url: 'https://purl.stanford.edu')
+    response = HTTP.get("#{purl_url}/#{druid}.json")
+    new(druid, response.body, purl_url:) if response.status.ok?
   end
 
-  def initialize(druid, purl_url: 'https://purl.stanford.edu')
+  def initialize(druid, public_cocina, purl_url: 'https://purl.stanford.edu')
     @druid = druid
     @purl_url = purl_url
+    @public_cocina_doc = JSON.parse(public_cocina)
+  end
+
+  def cocina_access
+    @cocina_access ||= public_cocina_doc['access']
+  end
+
+  def cocina_structural
+    @cocina_structural ||= public_cocina_doc['structural']
+  end
+
+  def cocina_description
+    @cocina_description ||= public_cocina_doc['description']
+  end
+
+  def cocina_titles(type: :main)
+    titles = cocina_description['title'].map { |title| Cocina::Models::Title.new(title) }
+    case type
+    when :main
+      Cocina::Models::Builders::TitleBuilder.main_title(titles)
+    when :full
+      Cocina::Models::Builders::TitleBuilder.full_title(titles)
+    when :additional
+      Cocina::Models::Builders::TitleBuilder.additional_titles(titles)
+    else
+      raise ArgumentError, "Invalid title type: #{type}"
+    end
+  end
+
+  def created
+    Time.parse(public_cocina_doc['created'])
+  end
+
+  def modified
+    Time.parse(public_cocina_doc['modified'])
+  end
+
+  def content_type
+    public_cocina_doc['type'].split('/').last
+  end
+
+  def files
+    cocina_structural&.fetch('contains', [])&.flat_map { |fileset| fileset.dig('structural', 'contains') } || []
   end
 
   def public_cocina?
-    !!public_cocina
+    public_cocina.present?
+  end
+
+  def collection?
+    content_type == 'collection'
   end
 
-  def public_cocina
-    @public_cocina ||= self.class.fetch("#{purl_url}/#{druid}.json")
+  def public?
+    [cocina_access['view'], cocina_access['download']].include? 'world'
   end
 
-  def public_cocina_doc
-    @public_cocina_doc ||= JSON.parse(public_cocina)
+  def stanford_only?
+    [cocina_access['view'], cocina_access['download']].include? 'stanford'
   end
 end

lib/public_xml_record.rb

@@ -1,22 +1,20 @@
 # frozen_string_literal: true
 
 require 'http'
-require 'active_support' # some transitive dependencies don't require active_support this first, as they must in Rails 7
-require 'active_support/core_ext/module/delegation'
 require 'mods_display'
 require 'dor/rights_auth'
 
 class PublicXmlRecord
   attr_reader :public_xml_doc, :druid, :purl_url
 
-  def self.fetch(purl_url, druid)
+  def self.fetch(druid, purl_url: 'https://purl.stanford.edu')
     response = HTTP.get("#{purl_url}/#{druid}.xml")
-    new(purl_url, druid, response.body) if response.status.ok?
+    new(druid, response.body, purl_url:) if response.status.ok?
   end
 
-  def initialize(purl_url, druid, public_xml)
-    @purl_url = purl_url
+  def initialize(druid, public_xml, purl_url: 'https://purl.stanford.edu')
     @druid = druid
+    @purl_url = purl_url
     @public_xml_doc = Nokogiri::XML(public_xml)
   end
 

lib/purl_record.rb

@@ -18,21 +18,32 @@ def searchworks_id
     catkey.nil? ? druid : catkey
   end
 
-  def public_xml?
-    !!public_xml
+  def druid_tree
+    druid.match(/(..)(...)(..)(....)/).captures.join('/')
   end
 
   def public_xml
-    @public_xml ||= PublicXmlRecord.fetch(purl_url, druid)
+    @public_xml ||= PublicXmlRecord.fetch(druid, purl_url:)
   end
 
-  delegate :mods, :rights, :public?, :stanford_only?, :rights_xml, :collection?,
+  def public_cocina
+    @public_cocina ||= PublicCocinaRecord.fetch(druid, purl_url:)
+  end
+
+  def public_xml?
+    public_xml.present?
+  end
+
+  def public_cocina?
+    public_cocina.present?
+  end
+
+  delegate :mods, :rights, :rights_xml, :collection?, :public?, :stanford_only?,
            :thumb, :dor_content_type, :dor_resource_content_type, :dor_file_mimetype,
            :dor_resource_count, :dor_read_rights, :collections, :constituents,
            :catkey, :label, :stanford_mods, :mods_display,
            :public_xml_doc, to: :public_xml
 
-  def druid_tree
-    druid.match(/(..)(...)(..)(....)/).captures.join('/')
-  end
+  delegate :cocina_access, :cocina_structural, :cocina_description, :cocina_titles,
+           :created, :modified, :public_cocina_doc, :content_type, :files, to: :public_cocina
 end