dor_indexing 1.5.0 → 2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ae485a8233c79356bffbc42bef8d8e9f3ef1afbf3db962a2ba43b4ce522d7cd7
-  data.tar.gz: 1a05728e58b24d5f94164881a92326c9232402e241ce589dab14231e448ce60a
+  metadata.gz: b76dd6ecf919e3653810f59c166d313f77eba4a3046bcd1598abd996bf57c74d
+  data.tar.gz: be615ccc690756cd7aff121e175d544caea9cf3b1dd70a6d03e0fd3c9ab14c43
 SHA512:
-  metadata.gz: 24cfdc3e6be2af97c092e893825af54f8e3e3e47d65b9b955ea90c4ebda86072dbd87f32e2b3778e5fbee005be00c6877efac710e4d87b178f2521cee42f8b62
-  data.tar.gz: 7a1fc488f95678830e07fb919371e3ef8f5125d8bfcbf40d62109c51f87203fb6c424bef4da4fd8b7a9ec77a87bfee3f92274325482ab58d602599d069bbafe7
+  metadata.gz: d04715c311bfd64ac62169e18d4fea188b6c2dc0e8263c38413e3eac3f985e901f3f5ba1181695f5e24ecbbdb713c937bd6c334c0e1f4ec622f293cea4a2195b
+  data.tar.gz: de51ceba624569579cbfe1c1f259ba3568158443ed849ed3d253a144ed6c25346d89916df530c7bc91891afe5f3de19e508cca9cc23818f9e61b262b7306a2a2

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    dor_indexing (1.5.0)
+    dor_indexing (2.0.0)
       activesupport
       cocina-models (~> 0.95.1)
       dor-services-client (~> 14.0)

data/README.md

@@ -26,14 +26,22 @@ If bundler is not being used to manage dependencies, install the gem by executin
 
 ## Usage
 
-DorIndexing that a configured Workflow Client, DOR Services 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:, dor_services_client:, cocina_repository:)
+doc = DorIndexing.build(
+  cocina_with_metadata:,
+  workflow_client:,
+  cocina_finder:,
+  administrative_tags_finder:,
+  release_tags_finder:
+)
 ```
 
 ## Testing
@@ -47,4 +55,3 @@ Argo is the blacklight app that uses the Solr index extensively, and it already
 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/lib/dor_indexing/builders/document_builder.rb

@@ -48,19 +48,20 @@ class DorIndexing
 
       @@parent_collections = {} # rubocop:disable Style/ClassVars
 
-      def self.for(model:, workflow_client:, dor_services_client:, cocina_repository:)
-        new(model:, workflow_client:, dor_services_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:, dor_services_client:, cocina_repository:)
+      def initialize(model:, workflow_client:, cocina_finder:, administrative_tags_finder:, release_tags_finder:)
         @model = model
         @workflow_client = workflow_client
-        @dor_services_client = dor_services_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
@@ -70,13 +71,14 @@ class DorIndexing
                                          parent_collections:,
                                          administrative_tags:,
                                          workflow_client:,
-                                         dor_services_client:,
-                                         cocina_repository:)
+                                         cocina_finder:,
+                                         administrative_tags_finder:,
+                                         release_tags_finder:)
       end
 
       private
 
-      attr_reader :model, :workflow_client, :dor_services_client, :cocina_repository
+      attr_reader :model, :workflow_client, :cocina_finder, :administrative_tags_finder, :release_tags_finder
 
       def id
         model.externalIdentifier
@@ -90,8 +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
@@ -99,8 +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

@@ -4,12 +4,12 @@ class DorIndexing
   module Indexers
     # Indexes the object's release tags
     class ReleasableIndexer
-      attr_reader :cocina, :parent_collections, :dor_services_client
+      attr_reader :cocina, :parent_collections, :release_tags_finder
 
-      def initialize(cocina:, parent_collections:, dor_services_client:, **)
+      def initialize(cocina:, parent_collections:, release_tags_finder:, **)
         @cocina = cocina
         @parent_collections = parent_collections
-        @dor_services_client = dor_services_client
+        @release_tags_finder = release_tags_finder
       end
 
       # @return [Hash] the partial solr document for releasable concerns
@@ -19,18 +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
@@ -41,10 +50,8 @@ class DorIndexing
 
       def tags_from_collection
         parent_collections.each_with_object({}) do |collection, result|
-          collection_object_client = dor_services_client.object(collection.externalIdentifier)
-          collection_object_client
-            .release_tags
-            .list
+          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)
@@ -53,10 +60,8 @@ class DorIndexing
       end
 
       def tags_from_item
-        object_client = dor_services_client.object(cocina.externalIdentifier)
-        object_client
-          .release_tags
-          .list
+        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)

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 = '1.5.0'
+  VERSION = '2.0.0'
 end

data/lib/dor_indexing.rb

@@ -17,13 +17,14 @@ require 'marc/vocab'
 # Builds solr documents for indexing.
 class DorIndexing
   # @return [Hash] the solr document
-  def self.build(cocina_with_metadata:, workflow_client:, dor_services_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:,
-      dor_services_client:,
-      cocina_repository:
+      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: 1.5.0
+  version: 2.0.0
 platform: ruby
 authors:
 - Justin Littman
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-03-27 00:00:00.000000000 Z
+date: 2024-03-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -149,7 +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
@@ -167,6 +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
@@ -194,7 +194,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.18
+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