blacklight-spotlight 3.0.0.alpha.10 → 3.0.0.rc5

data/app/models/spotlight/job_tracker.rb

@@ -0,0 +1,114 @@
+# frozen_string_literal: true
+
+module Spotlight
+  # Associate background jobs with records
+  class JobTracker < ActiveRecord::Base
+    scope :recent, -> { order('updated_at DESC').limit(5) }
+    scope :in_progress, -> { where.not(status: %w[completed failed]) }
+    scope :completed, -> { where(status: %w[completed failed]) }
+
+    belongs_to :on, polymorphic: true
+    belongs_to :resource, polymorphic: true
+    belongs_to :user, optional: true, class_name: Spotlight::Engine.config.user_class # rubocop:disable Rails/ReflectionClassName
+    has_many :events, as: :resource, dependent: :delete_all
+    has_many :job_trackers, as: :on, dependent: Rails.version > '6.1' ? :destroy_async : :destroy
+    has_many :subevents, through: :job_trackers, source: :events
+
+    serialize :data
+
+    after_initialize do
+      self.data ||= {}
+    end
+
+    after_commit do
+      next unless on.is_a? Spotlight::JobTracker
+
+      UpdateJobTrackersJob.perform_later(self)
+    end
+
+    def label
+      "[#{job_class.titleize}] #{resource_label}"
+    end
+
+    def resource_label
+      return resource.filename if resource.is_a? ActiveStorage::Blob
+      return resource.name if resource.is_a? Upload
+
+      resource_id
+    end
+
+    def job_status
+      return {} unless job_id
+
+      @job_status ||= ActiveJob::Status.get(job_id)
+    end
+
+    def progress_label
+      return number_with_delimiter(progress) unless total?
+
+      "#{number_with_delimiter(progress)} / #{number_with_delimiter(total)}"
+    end
+
+    def progress
+      data[:progress] || job_status[:progress] || 0
+    end
+
+    def total(default: progress)
+      [progress, data[:total] || job_status[:total] || default].max
+    end
+
+    def total?
+      total(default: 0).positive?
+    end
+
+    def percent
+      return nil unless total?
+
+      (100.0 * progress) / total
+    end
+
+    def status
+      @status ||= super
+      @status ||= 'missing'
+      @status = 'in_progress' if @status == 'completed' && job_trackers.any? { |t| t.in_progress? || t.enqueued? }
+
+      @status
+    end
+
+    def enqueued?
+      status == 'enqueued'
+    end
+
+    def in_progress?
+      status == 'in_progress'
+    end
+
+    def completed?
+      status == 'completed'
+    end
+
+    def failed?
+      status == 'failed'
+    end
+
+    def append_log_entry(type:, exhibit: nil, **args)
+      events.create(type: type, exhibit: exhibit, data: args)
+    rescue StandardError => e
+      Rails.logger.error("Unable to create log entry for job tracker #{id}: #{e}")
+    end
+
+    def top_level_job_tracker
+      if on.is_a?(Spotlight::JobTracker)
+        on.top_level_job_tracker
+      else
+        self
+      end
+    end
+
+    private
+
+    def number_with_delimiter(*args)
+      ActiveSupport::NumberHelper.number_to_delimited(*args)
+    end
+  end
+end

data/app/models/spotlight/page.rb

@@ -9,8 +9,8 @@ module Spotlight
     extend FriendlyId
     # Note: This configuration also needs to be duplicated on the
     # STI models ({Spotlight::AboutPage}, {Spotlight::FeaturePage}, {Spotlight::HomePage})
-    friendly_id :title, use: %i[slugged scoped finders history], scope: %i[exhibit locale] do |config|
-      config.reserved_words.concat(%w[update_all])
+    friendly_id :title, use: %i[slugged scoped finders history], scope: %i[exhibit locale], treat_reserved_as_conflict: true do |config|
+      config.reserved_words&.concat(%w[update_all contacts])
     end
 
     belongs_to :exhibit, touch: true
@@ -51,9 +51,7 @@ module Spotlight
     end
 
     def content
-      @content ||= begin
-        Spotlight::PageContent.for(self, :content)
-      end
+      @content ||= Spotlight::PageContent.for(self, :content)
     end
 
     def content_type

data/app/models/spotlight/page_configurations.rb

@@ -27,6 +27,7 @@ module Spotlight
              to: :context
 
     attr_reader :context, :page
+
     def initialize(context:, page:)
       @context = context
       @page = page
@@ -39,6 +40,7 @@ module Spotlight
         'attachment-endpoint': attachment_endpoint,
         'autocomplete-exhibit-catalog-path': exhibit_autocomplete_endpoint,
         'autocomplete-exhibit-pages-path': page_autocomplete_endpoint,
+        'autocomplete-exhibit-browse-groups-path': browse_groups_autocomplete_endpoint,
         'autocomplete-exhibit-searches-path': search_autocomplete_endpoint,
         'preview-url': page_preview_url
       }.merge(downstream_parameters)
@@ -61,6 +63,10 @@ module Spotlight
       spotlight.exhibit_attachments_path(current_exhibit)
     end
 
+    def browse_groups_autocomplete_endpoint
+      spotlight.exhibit_groups_path(current_exhibit)
+    end
+
     def exhibit_autocomplete_endpoint
       spotlight.autocomplete_exhibit_catalog_path(current_exhibit, q: '%QUERY', format: 'json')
     end

data/app/models/spotlight/page_content.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require 'spotlight/page_content/sir_trevor'
+
 module Spotlight
   # Factory for picking the right page content renderer
   module PageContent

data/app/models/spotlight/resource.rb

@@ -4,10 +4,17 @@ module Spotlight
   ##
   # Exhibit resources
   class Resource < ActiveRecord::Base
-    include ActiveSupport::Benchmarkable
-
-    class_attribute :document_builder_class
-    self.document_builder_class = SolrDocumentBuilder
+    class_attribute :indexing_pipeline, default: (Spotlight::Etl::Pipeline.new do |pipeline|
+      pipeline.sources = [Spotlight::Etl::Sources::IdentitySource]
+      pipeline.transforms = [
+        reject_blank: Spotlight::Etl::Transforms::RejectBlank,
+        reject_missing: Spotlight::Etl::Transforms::RejectMissingUniqueId,
+        apply_exhibit_metadata: Spotlight::Etl::Transforms::ApplyExhibitMetadata,
+        apply_application_metadata: Spotlight::Etl::Transforms::ApplyApplicationMetadata,
+        apply_pipeline_metadata: Spotlight::Etl::Transforms::ApplyPipelineMetadata
+      ]
+      pipeline.loaders = [Spotlight::Etl::SolrLoader]
+    end)
 
     extend ActiveModel::Callbacks
     define_model_callbacks :index
@@ -16,24 +23,22 @@ module Spotlight
 
     belongs_to :exhibit
     has_many :solr_document_sidecars
+    has_many :events, as: :resource
 
     serialize :data, Hash
 
-    after_index :commit
-    after_index :touch_exhibit!
-
     ##
     # Persist the record to the database, and trigger a reindex to solr
     #
     # @param [Hash] All arguments will be passed through to ActiveRecord's #save method
-    def save_and_index(*args)
-      save(*args) && reindex_later
+    def save_and_index(reindex_options: {}, **args)
+      save(*args) && reindex_later(**reindex_options)
     end
 
     ##
     # Enqueue an asynchronous reindexing job for this resource
-    def reindex_later
-      Spotlight::ReindexJob.perform_later(self)
+    def reindex_later(**args)
+      Spotlight::ReindexJob.perform_later(self, **args)
     end
 
     def document_model
@@ -45,67 +50,28 @@ module Spotlight
       # Index the result of {#to_solr} into the index in batches of {#batch_size}
       #
       # @return [Integer] number of records indexed
-      def reindex(reindexing_log_entry = nil)
-        benchmark "Reindexing #{self} (batch size: #{batch_size})" do
-          count = 0
-
-          run_callbacks :index do
-            document_builder.documents_to_index.each_slice(batch_size) do |batch|
-              write_to_index(batch)
-              count += batch.length
-              reindexing_log_entry&.update(items_reindexed_count: count)
-            end
-
-            count
+      def reindex(touch: true, **args, &block)
+        i = 0
+        run_callbacks :index do
+          indexing_pipeline.call(Spotlight::Etl::Context.new(self, commit: true, **args)) do |data|
+            i += 1
+            block&.call(data)
           end
         end
-      end
-
-      def document_builder
-        @document_builder ||= document_builder_class.new(self)
-      end
-
-      private
-
-      def blacklight_solr
-        @solr ||= RSolr.connect(connection_config.merge(adapter: connection_config[:http_adapter]))
-      end
 
-      def connection_config
-        Blacklight.connection_config
-      end
+        touch_exhibit! if touch
 
-      def batch_size
-        Spotlight::Engine.config.solr_batch_size
+        i
       end
 
-      def write_to_index(batch)
-        documents = documents_that_have_ids(batch)
-        return unless write? && documents.present?
-
-        blacklight_solr.update params: { commitWithin: 500 },
-                               data: documents.to_json,
-                               headers: { 'Content-Type' => 'application/json' }
+      def estimated_size(**args)
+        indexing_pipeline.estimated_size(Spotlight::Etl::Context.new(self, **args))
       end
 
-      def commit
-        return unless write?
-
-        blacklight_solr.commit
-      rescue StandardError => e
-        Rails.logger.warn "Unable to commit to solr: #{e}"
-      end
+      private
 
       def touch_exhibit!
-        exhibit.touch
-      end
-
-      def write?
-        Spotlight::Engine.config.writable_index
-      end
-
-      def documents_that_have_ids(document_list)
-        document_list.reject { |d| d[document_builder.document_model.unique_key.to_sym].blank? }
+        exhibit&.touch
       end
     end
   end

data/app/models/spotlight/resources/csv_upload.rb

@@ -3,9 +3,10 @@
 module Spotlight
   module Resources
     ##
-    # Shim object for CSV Uploads. see {Spotlight::AddUploadsFromCSV}
+    # Shim object for CSV Uploads. see {Spotlight::AddUploadsFromCsv}
     class CsvUpload
       attr_reader :url
+
       include ActiveModel::Model
       extend ActiveModel::Translation
     end

data/app/models/spotlight/resources/iiif_harvester.rb

@@ -7,7 +7,6 @@ module Spotlight
     # harvest Images from IIIF Manifest and turn them into a Spotlight::Resource
     # Note: IIIF API : http://iiif.io/api/presentation/2.0
     class IiifHarvester < Spotlight::Resource
-      self.document_builder_class = Spotlight::Resources::IiifBuilder
       self.weight = -5000
 
       validate :valid_url?
@@ -16,6 +15,16 @@ module Spotlight
         @iiif_manifests ||= IiifService.parse(url)
       end
 
+      def self.indexing_pipeline
+        @indexing_pipeline ||= super.dup.tap do |pipeline|
+          pipeline.sources = [Spotlight::Etl::Sources::SourceMethodSource(:iiif_manifests)]
+
+          pipeline.transforms = [
+            ->(data, p) { data.merge(p.source.to_solr(exhibit: p.context.resource.exhibit)) }
+          ] + pipeline.transforms
+        end
+      end
+
       private
 
       def valid_url?
@@ -24,8 +33,8 @@ module Spotlight
 
       def url_is_iiif?(url)
         valid_content_types = ['application/json', 'application/ld+json']
-        req = Faraday.head(url)
-        req = Faraday.get(url) if req.status == 405
+        req = Spotlight::Resources::IiifService.http_client.head(url)
+        req = Spotlight::Resources::IiifService.http_client.get(url) if req.status == 405
         return unless req.success?
 
         valid_content_types.any? do |valid_type|

data/app/models/spotlight/resources/iiif_manifest.rb

@@ -6,6 +6,7 @@ module Spotlight
     # A PORO to construct a solr hash for a given IiifManifest
     class IiifManifest
       attr_reader :collection
+
       def initialize(attrs = {})
         @url = attrs[:url]
         @manifest = attrs[:manifest]
@@ -13,7 +14,9 @@ module Spotlight
         @solr_hash = {}
       end
 
-      def to_solr
+      def to_solr(exhibit: nil)
+        @exhibit = exhibit if exhibit
+
         add_document_id
         add_label
         add_thumbnail_url
@@ -36,6 +39,7 @@ module Spotlight
       private
 
       attr_reader :url, :manifest, :exhibit, :solr_hash
+
       delegate :blacklight_config, to: :exhibit
 
       def add_document_id
@@ -67,7 +71,7 @@ module Spotlight
       end
 
       def add_label
-        return unless title_fields.present? && manifest.try(:label)
+        return unless title_fields.present? && manifest&.label
 
         Array.wrap(title_fields).each do |field|
           solr_hash[field] = metadata_class.new(manifest).label
@@ -142,7 +146,7 @@ module Spotlight
       end
 
       def thumbnail_field
-        blacklight_config.index.try(:thumbnail_field)
+        blacklight_config.index.thumbnail_field
       end
 
       def full_image_field
@@ -150,11 +154,11 @@ module Spotlight
       end
 
       def tile_source_field
-        blacklight_config.show.try(:tile_source_field)
+        blacklight_config.show.tile_source_field
       end
 
       def title_fields
-        Spotlight::Engine.config.iiif_title_fields || blacklight_config.index.try(:title_field)
+        Spotlight::Engine.config.iiif_title_fields || blacklight_config.index&.title_field
       end
 
       def sidecar
@@ -185,7 +189,7 @@ module Spotlight
         end
 
         def label
-          return unless manifest.try(:label)
+          return unless manifest&.label
 
           Array(json_ld_value(manifest.label)).map { |v| html_sanitize(v) }.first
         end
@@ -195,7 +199,7 @@ module Spotlight
         attr_reader :manifest
 
         def metadata
-          manifest.try(:metadata) || []
+          manifest&.metadata || []
         end
 
         def metadata_hash

data/app/models/spotlight/resources/iiif_service.rb

@@ -29,6 +29,13 @@ module Spotlight
         recursive_manifests(new(url))
       end
 
+      def self.http_client
+        Faraday.new do |b|
+          b.use FaradayMiddleware::FollowRedirects
+          b.adapter Faraday.default_adapter
+        end
+      end
+
       protected
 
       def object
@@ -41,8 +48,8 @@ module Spotlight
 
       class << self
         def iiif_response(url)
-          Faraday.get(url).body
-        rescue Faraday::Error::ConnectionFailed, Faraday::TimeoutError => e
+          Spotlight::Resources::IiifService.http_client.get(url).body
+        rescue Faraday::ConnectionFailed, Faraday::TimeoutError => e
           Rails.logger.warn("HTTP GET for #{url} failed with #{e}")
           {}.to_json
         end