dor_indexing 1.4.1 → 2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9c1e2280668e9085122b921fd927b75e9982fa5fda9067a9245899ef93db77ae
-  data.tar.gz: 3e8d89e77994cb62ffcc1c598e47716c3a9c9f8aa8e37f226159e8fef947082d
+  metadata.gz: b76dd6ecf919e3653810f59c166d313f77eba4a3046bcd1598abd996bf57c74d
+  data.tar.gz: be615ccc690756cd7aff121e175d544caea9cf3b1dd70a6d03e0fd3c9ab14c43
 SHA512:
-  metadata.gz: 24e3bf95ad1c541d3b403b2233018ab1da89be80cdcd3fa9acb2793e2b57673c8364202a4a548f8fe1c9ff60444585b5cf2d020968cb5b50614ad80b118d7bae
-  data.tar.gz: 6c015737932b1f01819d89f5f91948ab43d09e1537d6bf41ba8df85778c9ed0f18dc85b408a7fdc6ceb48ce6cfc4f7127b74da0c415aac6819d7dd460cf8a4eb
+  metadata.gz: d04715c311bfd64ac62169e18d4fea188b6c2dc0e8263c38413e3eac3f985e901f3f5ba1181695f5e24ecbbdb713c937bd6c334c0e1f4ec622f293cea4a2195b
+  data.tar.gz: de51ceba624569579cbfe1c1f259ba3568158443ed849ed3d253a144ed6c25346d89916df530c7bc91891afe5f3de19e508cca9cc23818f9e61b262b7306a2a2

data/Gemfile.lock

@@ -1,9 +1,10 @@
 PATH
   remote: .
   specs:
-    dor_indexing (1.4.1)
+    dor_indexing (2.0.0)
       activesupport
       cocina-models (~> 0.95.1)
+      dor-services-client (~> 14.0)
       dor-workflow-client (~> 7.0)
       honeybadger
       marc-vocab (~> 0.3.0)
@@ -26,7 +27,7 @@ GEM
     ast (2.4.2)
     attr_extras (7.1.0)
     base64 (0.2.0)
-    bigdecimal (3.1.6)
+    bigdecimal (3.1.7)
     byebug (11.1.3)
     cocina-models (0.95.1)
       activesupport
@@ -51,6 +52,13 @@ GEM
       activesupport
     diff-lcs (1.5.1)
     docile (1.4.0)
+    dor-services-client (14.4.0)
+      activesupport (>= 4.2, < 8)
+      cocina-models (~> 0.95.1)
+      deprecation
+      faraday (~> 2.0)
+      faraday-retry
+      zeitwerk (~> 2.1)
     dor-workflow-client (7.0.2)
       activesupport (>= 3.2.1, < 8)
       deprecation (>= 0.99.0)
@@ -89,7 +97,7 @@ GEM
       net-http
     faraday-retry (2.2.0)
       faraday (~> 2.0)
-    honeybadger (5.6.0)
+    honeybadger (5.8.0)
     i18n (1.14.4)
       concurrent-ruby (~> 1.0)
     ice_nine (0.11.2)
@@ -99,7 +107,7 @@ GEM
       multi_json
     language_server-protocol (3.17.0.3)
     marc-vocab (0.3.0)
-    minitest (5.22.2)
+    minitest (5.22.3)
     mods (3.0.4)
       edtf (~> 3.0)
       iso-639
@@ -109,9 +117,9 @@ GEM
     mutex_m (0.2.0)
     net-http (0.4.1)
       uri
-    nokogiri (1.16.2-x86_64-darwin)
+    nokogiri (1.16.3-x86_64-darwin)
       racc (~> 1.4)
-    nokogiri (1.16.2-x86_64-linux)
+    nokogiri (1.16.3-x86_64-linux)
       racc (~> 1.4)
     nom-xml (1.2.0)
       i18n
@@ -146,7 +154,7 @@ GEM
     rspec-support (3.13.1)
     rss (0.3.0)
       rexml
-    rubocop (1.62.0)
+    rubocop (1.62.1)
       json (~> 2.3)
       language_server-protocol (>= 3.17.0)
       parallel (~> 1.10)

data/README.md

@@ -26,12 +26,32 @@ If bundler is not being used to manage dependencies, install the gem by executin
 
 ## Usage
 
-DorIndexing that a configured Workflow Client and a Cocina Repository be injected.
+DorIndexing requires interaction with the SDR workflow API and also needs the following:
 
-The Cocina Repository provides methods for finding Cocina objects and administrative tags. One possible implementation of a Cocina Repository would be to use DOR Services Client.
+* a callable that takes a single argument (a druid) and returns the Cocina for the corresponding object
+* a callable that takes a single argument (a druid) and returns the list of administrative tags for the corresponding object
+* a callable that takes a single argument (a druid) and returns the list of release tags for the corresponding object
 
 ```ruby
 require 'dor_indexing'
 
-doc = DorIndexing.build(cocina_with_metadata:, workflow_client:, cocina_repository:)
+doc = DorIndexing.build(
+  cocina_with_metadata:,
+  workflow_client:,
+  cocina_finder:,
+  administrative_tags_finder:,
+  release_tags_finder:
+)
 ```
+
+## Testing
+
+### Integration Testing with Solr
+
+We build and update the Solr index via dor-indexing-app amd dor-services-app, both of which use this gem for indexing logic.
+
+Argo is the blacklight app that uses the Solr index extensively, and it already has the docker containers to create new test objects in dor-services-app and index them (via dor_indexing_app to Solr).  And Argo is the app built on top of the Solr index, so a good place to check results.
+
+To ensure our indexing behavior produces the desired results, it was easiest to put
+the full stack integration tests in the argo repository -- they can be found in
+https://github.com/sul-dlss/argo/tree/main/spec/features/indexing_xxx_spec.rb

data/dor_indexing.gemspec

@@ -33,6 +33,7 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency 'activesupport'
   spec.add_dependency 'cocina-models', '~> 0.95.1'
+  spec.add_dependency 'dor-services-client', '~> 14.0'
   spec.add_dependency 'dor-workflow-client', '~> 7.0'
   spec.add_dependency 'honeybadger'
   spec.add_dependency 'marc-vocab', '~> 0.3.0'

data/lib/dor_indexing/builders/document_builder.rb

@@ -48,18 +48,20 @@ class DorIndexing
 
       @@parent_collections = {} # rubocop:disable Style/ClassVars
 
-      def self.for(model:, workflow_client:, cocina_repository:)
-        new(model:, workflow_client:, cocina_repository:).for
+      def self.for(...)
+        new(...).for
       end
 
       def self.reset_parent_collections
         @@parent_collections = {} # rubocop:disable Style/ClassVars
       end
 
-      def initialize(model:, workflow_client:, cocina_repository:)
+      def initialize(model:, workflow_client:, cocina_finder:, administrative_tags_finder:, release_tags_finder:)
         @model = model
         @workflow_client = workflow_client
-        @cocina_repository = cocina_repository
+        @cocina_finder = cocina_finder
+        @administrative_tags_finder = administrative_tags_finder
+        @release_tags_finder = release_tags_finder
       end
 
       # @param [Cocina::Models::DROWithMetadata,Cocina::Models::CollectionWithMetadata,Cocina::Model::AdminPolicyWithMetadata] model
@@ -69,12 +71,14 @@ class DorIndexing
                                          parent_collections:,
                                          administrative_tags:,
                                          workflow_client:,
-                                         cocina_repository:)
+                                         cocina_finder:,
+                                         administrative_tags_finder:,
+                                         release_tags_finder:)
       end
 
       private
 
-      attr_reader :model, :workflow_client, :cocina_repository
+      attr_reader :model, :workflow_client, :cocina_finder, :administrative_tags_finder, :release_tags_finder
 
       def id
         model.externalIdentifier
@@ -88,8 +92,8 @@ class DorIndexing
         return [] unless model.dro?
 
         Array(model.structural&.isMemberOf).filter_map do |rel_druid|
-          @@parent_collections[rel_druid] ||= cocina_repository.find(rel_druid)
-        rescue DorIndexing::CocinaRepository::RepositoryError
+          @@parent_collections[rel_druid] ||= cocina_finder.call(rel_druid)
+        rescue DorIndexing::RepositoryError
           Honeybadger.notify("Bad association found on #{model.externalIdentifier}. #{rel_druid} could not be found")
           # This may happen if the referenced Collection does not exist (bad data)
           nil
@@ -97,8 +101,8 @@ class DorIndexing
       end
 
       def administrative_tags
-        cocina_repository.administrative_tags(id)
-      rescue DorIndexing::CocinaRepository::RepositoryError
+        administrative_tags_finder.call(id)
+      rescue DorIndexing::RepositoryError
         []
       end
     end

data/lib/dor_indexing/indexers/identifiable_indexer.rb

@@ -4,13 +4,14 @@ class DorIndexing
   module Indexers
     # Indexes the druid, metadata sources, and the apo titles
     class IdentifiableIndexer
-      attr_reader :cocina, :cocina_repository
+      attr_reader :cocina, :cocina_finder, :administrative_tags_finder
 
       CURRENT_CATALOG_TYPE = 'folio'
 
-      def initialize(cocina:, cocina_repository:, **)
+      def initialize(cocina:, cocina_finder:, administrative_tags_finder:, **)
         @cocina = cocina
-        @cocina_repository = cocina_repository
+        @cocina_finder = cocina_finder
+        @administrative_tags_finder = administrative_tags_finder
       end
 
       ## Module-level variable, shared between ALL mixin includers (and ALL *their* includers/extenders)!
@@ -72,13 +73,13 @@ class DorIndexing
       # populate cache if necessary
       def populate_cache(rel_druid)
         @@apo_hash[rel_druid] ||= begin
-          related_obj = cocina_repository.find(rel_druid)
+          related_obj = cocina_finder.call(rel_druid)
           # APOs don't have projects, and since Hydrus is set to be retired, I don't want to
           # add the cocina property. Just check the tags service instead.
           is_from_hydrus = hydrus_tag?(rel_druid)
           title = Cocina::Models::Builders::TitleBuilder.build(related_obj.description.title)
           { 'related_obj_title' => title, 'is_from_hydrus' => is_from_hydrus }
-        rescue CocinaRepository::RepositoryError
+        rescue RepositoryError
           Honeybadger.notify("Bad association found on #{cocina.externalIdentifier}. #{rel_druid} could not be found")
           # This may happen if the given APO or Collection does not exist (bad data)
           { 'related_obj_title' => rel_druid, 'is_from_hydrus' => false }
@@ -86,7 +87,7 @@ class DorIndexing
       end
 
       def hydrus_tag?(id)
-        cocina_repository.administrative_tags(id).include?('Project : Hydrus')
+        administrative_tags_finder.call(id).include?('Project : Hydrus')
       end
     end
   end

data/lib/dor_indexing/indexers/releasable_indexer.rb

@@ -4,11 +4,12 @@ class DorIndexing
   module Indexers
     # Indexes the object's release tags
     class ReleasableIndexer
-      attr_reader :cocina, :parent_collections
+      attr_reader :cocina, :parent_collections, :release_tags_finder
 
-      def initialize(cocina:, parent_collections:, **)
+      def initialize(cocina:, parent_collections:, release_tags_finder:, **)
         @cocina = cocina
         @parent_collections = parent_collections
+        @release_tags_finder = release_tags_finder
       end
 
       # @return [Hash] the partial solr document for releasable concerns
@@ -18,18 +19,27 @@ class DorIndexing
         {
           'released_to_ssim' => tags.map(&:to).uniq,
           'released_to_searchworks_dttsi' => searchworks_release_date,
-          'released_to_earthworks_dttsi' => earthworks_release_date
+          'released_to_earthworks_dttsi' => earthworks_release_date,
+          'released_to_purl_sitemap_dttsi' => purl_sitemap_release_date
         }.compact
       end
 
       private
 
+      def purl_sitemap_release_date
+        date_for_tag 'PURL sitemap'
+      end
+
       def earthworks_release_date
-        tags.find { |tag| tag.to == 'Earthworks' }&.date&.utc&.iso8601
+        date_for_tag 'Earthworks'
       end
 
       def searchworks_release_date
-        tags.find { |tag| tag.to == 'Searchworks' }&.date&.utc&.iso8601
+        date_for_tag 'Searchworks'
+      end
+
+      def date_for_tag(project)
+        tags.find { |tag| tag.to == project }&.date&.utc&.iso8601
       end
 
       # Item tags have precidence over collection tags, so if the collection is release=true
@@ -40,8 +50,9 @@ class DorIndexing
 
       def tags_from_collection
         parent_collections.each_with_object({}) do |collection, result|
-          Array(collection.administrative.releaseTags)
-            .select { |tag| tag.what == 'collection' }
+          release_tags_finder
+            .call(collection.externalIdentifier)
+            .select { |tag| tag.what == 'self' }
             .group_by(&:to).map do |project, releases_for_project|
               result[project] = releases_for_project.max_by(&:date)
             end
@@ -49,13 +60,12 @@ class DorIndexing
       end
 
       def tags_from_item
-        released_for.group_by(&:to).transform_values do |releases_for_project|
-          releases_for_project.max_by(&:date)
-        end
-      end
-
-      def released_for
-        Array(cocina.administrative.releaseTags)
+        release_tags_finder
+          .call(cocina.externalIdentifier)
+          .select { |tag| tag.what == 'self' }
+          .group_by(&:to).transform_values do |releases_for_project|
+            releases_for_project.max_by(&:date)
+          end
       end
     end
   end

data/lib/dor_indexing/repository_error.rb

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+
+class DorIndexing
+  # Error raised retrieving Cocina objects, administrative tags, or release tags
+  # In DSA, the concrete implementation backs this with CocinaObjectStore.
+  # In DIA, the concrete implementation backs this with Dor Services Client.
+  class RepositoryError < StandardError; end
+end

data/lib/dor_indexing/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 class DorIndexing
-  VERSION = '1.4.1'
+  VERSION = '2.0.0'
 end

data/lib/dor_indexing.rb

@@ -10,14 +10,21 @@ require 'active_support/core_ext/object/blank'
 require 'active_support/core_ext/enumerable'
 require 'active_support/core_ext/string'
 require 'cocina/models'
+require 'dor/services/client'
 require 'honeybadger'
 require 'marc/vocab'
 
 # Builds solr documents for indexing.
 class DorIndexing
   # @return [Hash] the solr document
-  def self.build(cocina_with_metadata:, workflow_client:, cocina_repository:)
+  def self.build(cocina_with_metadata:, workflow_client:, cocina_finder:, administrative_tags_finder:, release_tags_finder:)
     Honeybadger.context({ identifier: cocina_with_metadata.externalIdentifier })
-    DorIndexing::Builders::DocumentBuilder.for(model: cocina_with_metadata, workflow_client:, cocina_repository:).to_solr
+    DorIndexing::Builders::DocumentBuilder.for(
+      model: cocina_with_metadata,
+      workflow_client:,
+      cocina_finder:,
+      administrative_tags_finder:,
+      release_tags_finder:
+    ).to_solr
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dor_indexing
 version: !ruby/object:Gem::Version
-  version: 1.4.1
+  version: 2.0.0
 platform: ruby
 authors:
 - Justin Littman
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-03-11 00:00:00.000000000 Z
+date: 2024-03-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -38,6 +38,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 0.95.1
+- !ruby/object:Gem::Dependency
+  name: dor-services-client
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '14.0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '14.0'
 - !ruby/object:Gem::Dependency
   name: dor-workflow-client
   requirement: !ruby/object:Gem::Requirement
@@ -135,7 +149,6 @@ files:
 - lib/dor_indexing/builders/publisher_name_builder.rb
 - lib/dor_indexing/builders/temporal_builder.rb
 - lib/dor_indexing/builders/topic_builder.rb
-- lib/dor_indexing/cocina_repository.rb
 - lib/dor_indexing/indexers/administrative_tag_indexer.rb
 - lib/dor_indexing/indexers/basic_indexer.rb
 - lib/dor_indexing/indexers/collection_title_indexer.rb
@@ -153,6 +166,7 @@ files:
 - lib/dor_indexing/indexers/workflow_process_indexer.rb
 - lib/dor_indexing/indexers/workflows_indexer.rb
 - lib/dor_indexing/marc_country.rb
+- lib/dor_indexing/repository_error.rb
 - lib/dor_indexing/selectors/event_selector.rb
 - lib/dor_indexing/selectors/pub_year_selector.rb
 - lib/dor_indexing/version.rb
@@ -180,7 +194,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.10
+rubygems_version: 3.5.6
 signing_key:
 specification_version: 4
 summary: Library for creating Solr documents for SDR indexing.

data/lib/dor_indexing/cocina_repository.rb

@@ -1,24 +0,0 @@
-# frozen_string_literal: true
-
-class DorIndexing
-  # Interface for retrieving Cocina objects.
-  # In DSA, the concrete implementation backs this with CocinaObjectStore.
-  # In DIA, the concrete implementation backs this with Dor Services Client.
-  class CocinaRepository
-    class RepositoryError < StandardError; end
-
-    # @param [String] druid
-    # @return [Cocina::Models::DROWithMetadata,Cocina::Models::CollectionWithMetadata,Cocina::Models::AdminPolicyWithMetadata]
-    # @raise [RepositoryError] if the object is not found or other error occurs
-    def find(druid)
-      raise NotImplementedError
-    end
-
-    # @param [String] druid
-    # @return [Array<String>] administrative tags
-    # @raise [RepositoryError] if the object is not found or other error occurs
-    def administrative_tags(druid)
-      raise NotImplementedError
-    end
-  end
-end