Optionally support cocina lookup

.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2020-10-06 20:10:00 UTC using RuboCop version 0.92.0.
+# on 2020-10-08 19:34:35 UTC using RuboCop version 0.92.0.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
@@ -14,7 +14,13 @@ Layout/ArgumentAlignment:
   Exclude:
     - 'spec/indexers/processable_indexer_spec.rb'
 
-# Offense count: 9
+# Offense count: 18
+# Cop supports --auto-correct.
+# Configuration parameters: AllowUnusedKeywordArguments, IgnoreEmptyMethods, IgnoreNotImplementedMethods.
+Lint/UnusedMethodArgument:
+  Enabled: false
+
+# Offense count: 8
 # Configuration parameters: IgnoredMethods.
 Metrics/AbcSize:
   Max: 35
@@ -24,7 +30,7 @@ Metrics/AbcSize:
 Metrics/CyclomaticComplexity:
   Max: 10
 
-# Offense count: 5
+# Offense count: 4
 # Configuration parameters: CountComments, CountAsOne, ExcludedMethods.
 Metrics/MethodLength:
   Max: 26
@@ -45,10 +51,10 @@ Performance/CollectionLiteralInLoop:
 RSpec/MultipleExpectations:
   Max: 10
 
-# Offense count: 25
+# Offense count: 30
 # Configuration parameters: AllowSubject.
 RSpec/MultipleMemoizedHelpers:
-  Max: 11
+  Max: 12
 
 # Offense count: 2
 # Configuration parameters: IgnoreSharedExamples.
@@ -96,4 +102,4 @@ Style/FormatStringToken:
 # Configuration parameters: AutoCorrect, AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, IgnoredPatterns.
 # URISchemes: http, https
 Layout/LineLength:
-  Max: 189
+  Max: 185

Gemfile

@@ -17,11 +17,11 @@ gem 'rack-timeout', '~> 0.5.1'
 
 # Reduces boot times through caching; required in config/boot.rb
 gem 'bootsnap', '>= 1.1.0', require: false
-
 gem 'dor-rights-auth', '~>1.5' # 1.5 or better is needed to ensure we index controlled digital lending rights object correctly
 gem 'dor-services', '~> 9.0', '>= 9.2.1' # must be 9.2.1 to get fix from https://github.com/sul-dlss/dor-services/pull/695
 gem 'dor-services-client', '~> 6.0'
 gem 'dor-workflow-client', '~> 3.20'
+gem 'dry-monads', '~> 1.3'
 gem 'okcomputer' # for monitoring
 gem 'rsolr', '~> 2.0'
 

Gemfile.lock

@@ -177,6 +177,10 @@ GEM
       concurrent-ruby (~> 1.0)
       dry-core (~> 0.2)
       dry-equalizer (~> 0.2)
+    dry-monads (1.3.5)
+      concurrent-ruby (~> 1.0)
+      dry-core (~> 0.4, >= 0.4.4)
+      dry-equalizer
     dry-schema (1.5.4)
       concurrent-ruby (~> 1.0)
       dry-configurable (~> 0.8, >= 0.8.3)
@@ -479,6 +483,7 @@ DEPENDENCIES
   dor-services (~> 9.0, >= 9.2.1)
   dor-services-client (~> 6.0)
   dor-workflow-client (~> 3.20)
+  dry-monads (~> 1.3)
   erubis
   faraday
   honeybadger (~> 4.5)

app/controllers/dor_controller.rb

@@ -2,6 +2,8 @@
 
 # Main controller of application
 class DorController < ApplicationController
+  include Dry::Monads[:result]
+
   def reindex
     @solr_doc = reindex_pid params[:pid], logger: generate_index_logger(request.uuid), add_attributes: { commitWithin: params.fetch(:commitWithin, 1000).to_i }
     solr.commit unless params[:commitWithin] # reindex_pid doesn't commit, but callers of this method may expect the update to be committed immediately
@@ -39,15 +41,22 @@ def generate_index_logger(entry_id)
   def reindex_pid(pid, logger:, add_attributes:)
     obj = nil
     solr_doc = nil
+    cocina = nil
 
     # benchmark how long it takes to load the object
     load_stats = Benchmark.measure('load_instance') do
       obj = Dor.find pid
+      cocina = begin
+        Success(Dor::Services::Client.object(pid).find)
+      rescue StandardError => e
+        Honeybadger.notify("Unable to convert #{pid} to a cocina model. #{e.message}")
+        Failure(:conversion_error)
+      end
     end.format('%n realtime %rs total CPU %ts').gsub(/[()]/, '')
 
     # benchmark how long it takes to convert the object to a Solr document
     to_solr_stats = Benchmark.measure('to_solr') do
-      indexer = Indexer.for(obj)
+      indexer = Indexer.for(obj, cocina: cocina)
       solr_doc = indexer.to_solr
       solr.add(solr_doc, add_attributes: add_attributes)
     end.format('%n realtime %rs total CPU %ts').gsub(/[()]/, '')

app/indexers/administrative_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class AdministrativeMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/administrative_tag_indexer.rb

@@ -9,7 +9,7 @@ class AdministrativeTagIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/composite_indexer.rb

@@ -8,16 +8,16 @@ def initialize(*indexers)
     @indexers = indexers
   end
 
-  def new(resource:)
-    Instance.new(indexers, resource: resource)
+  def new(resource:, cocina:)
+    Instance.new(indexers, resource: resource, cocina: cocina)
   end
 
   class Instance
     attr_reader :indexers, :resource
 
-    def initialize(indexers, resource:)
+    def initialize(indexers, resource:, cocina:)
       @resource = resource
-      @indexers = indexers.map { |i| i.new(resource: resource) }
+      @indexers = indexers.map { |i| i.new(resource: resource, cocina: cocina) }
     end
 
     # @return [Hash] the merged solr document for all the sub-indexers

app/indexers/content_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class ContentMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/data_indexer.rb

@@ -6,7 +6,7 @@ class DataIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/data_quality_indexer.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+class DataQualityIndexer
+  attr_reader :resource, :cocina
+
+  def initialize(resource:, cocina:)
+    @resource = resource
+    @cocina = cocina
+  end
+
+  # @return [Hash] the partial solr document for identityMetadata
+  def to_solr
+    { 'data_quality_ssim' => messages }
+  end
+
+  private
+
+  def messages
+    [source_id_message].compact.tap do |messages|
+      messages << 'Cocina conversion failed' if cocina.failure?
+    end
+  end
+
+  def source_id_message
+    if source_id.present?
+      'non-comformant sourceId' unless valid_source_id?
+    else
+      'missing sourceId'
+    end
+  end
+
+  def source_id
+    @source_id ||= resource.identityMetadata.sourceId
+  end
+
+  def valid_source_id?
+    /^.+:.+$/.match?(source_id)
+  end
+end

app/indexers/default_object_rights_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class DefaultObjectRightsDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/describable_indexer.rb

@@ -3,7 +3,7 @@
 class DescribableIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/descriptive_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class DescriptiveMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/embargo_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class EmbargoMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/identifiable_indexer.rb

@@ -22,7 +22,7 @@ class IdentifiableIndexer
   }.freeze
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/identity_metadata_datastream_indexer.rb

@@ -5,7 +5,7 @@ class IdentityMetadataDatastreamIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 
@@ -21,7 +21,6 @@ def to_solr
       plain_identifiers << id
       ns_identifiers << source_id
       solr_doc['source_id_ssim'] = [source_id]
-      solr_doc['source_id_errors_ssim'] = 'non-comformant' unless /^.+:.+$/.match?(source_id)
     end
 
     resource.identityMetadata.otherId.compact.each do |qid|

app/indexers/object_profile_indexer.rb

@@ -5,7 +5,7 @@ class ObjectProfileIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/processable_indexer.rb

@@ -5,7 +5,7 @@ class ProcessableIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/provenance_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class ProvenanceMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/releasable_indexer.rb

@@ -5,7 +5,7 @@ class ReleasableIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/rights_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class RightsMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/role_metadata_datastream_indexer.rb

@@ -5,7 +5,7 @@ class RoleMetadataDatastreamIndexer
 
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/version_metadata_datastream_indexer.rb

@@ -3,7 +3,7 @@
 class VersionMetadataDatastreamIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/indexers/workflows_indexer.rb

@@ -4,7 +4,7 @@
 class WorkflowsIndexer
   attr_reader :resource
 
-  def initialize(resource:)
+  def initialize(resource:, cocina:)
     @resource = resource
   end
 

app/services/indexer.rb

@@ -2,6 +2,7 @@
 
 class Indexer
   ADMIN_POLICY_INDEXER = CompositeIndexer.new(
+    DataQualityIndexer,
     AdministrativeTagIndexer,
     DataIndexer,
     RoleMetadataDatastreamIndexer,
@@ -20,6 +21,7 @@ class Indexer
   )
 
   COLLECTION_INDEXER = CompositeIndexer.new(
+    DataQualityIndexer,
     AdministrativeTagIndexer,
     DataIndexer,
     ProvenanceMetadataDatastreamIndexer,
@@ -36,6 +38,7 @@ class Indexer
   )
 
   ITEM_INDEXER = CompositeIndexer.new(
+    DataQualityIndexer,
     AdministrativeTagIndexer,
     DataIndexer,
     ProvenanceMetadataDatastreamIndexer,
@@ -54,6 +57,7 @@ class Indexer
   )
 
   SET_INDEXER = CompositeIndexer.new(
+    DataQualityIndexer,
     AdministrativeTagIndexer,
     DataIndexer,
     ProvenanceMetadataDatastreamIndexer,
@@ -78,7 +82,9 @@ class Indexer
     Dor::Set => SET_INDEXER
   }.freeze
 
-  def self.for(obj)
-    INDEXERS.fetch(obj.class).new(resource: obj)
+  # @param [Dor::Abstract] obj
+  # @param [Dry::Monads::Result] cocina
+  def self.for(obj, cocina:)
+    INDEXERS.fetch(obj.class).new(resource: obj, cocina: cocina)
   end
 end