blacklight-spotlight 3.0.0.rc3 → 3.0.1

data/app/models/spotlight/field_metadata.rb

@@ -45,18 +45,14 @@ module Spotlight
 
     # This gets the number of *documents* with a field
     def document_counts
-      @document_count ||= begin
-        solr_response.facet_queries.each_with_object({}) do |(k, v), h|
-          h[k.split(/:/).first] = v
-        end
+      @document_count ||= solr_response.facet_queries.each_with_object({}) do |(k, v), h|
+        h[k.split(/:/).first] = v
       end
     end
 
     def terms
-      @terms ||= begin
-        solr_response.aggregations.each_with_object({}) do |(facet_name, facet), h|
-          h[facet_name] = facet.items.map(&:label)
-        end
+      @terms ||= solr_response.aggregations.each_with_object({}) do |(facet_name, facet), h|
+        h[facet_name] = facet.items.map(&:label)
       end
     end
 

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

@@ -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

@@ -18,14 +18,12 @@ module Spotlight
     delegate :available_view_fields,
              :blacklight_config,
              :current_exhibit,
-             :document_show_link_field,
-             :index_fields,
-             :index_field_label,
              :spotlight,
              :t,
-             :view_label,
              to: :context
 
+    DOCUMENT_TITLE_KEY = '__spotlight_document_title___'
+
     attr_reader :context, :page
 
     def initialize(context:, page:)
@@ -35,7 +33,7 @@ module Spotlight
 
     def as_json(*)
       {
-        'blacklight-configuration-index-fields': available_index_fields,
+        'blacklight-configuration-index-fields': available_caption_fields,
         'blacklight-configuration-search-views': available_view_configs,
         'attachment-endpoint': attachment_endpoint,
         'autocomplete-exhibit-catalog-path': exhibit_autocomplete_endpoint,
@@ -48,15 +46,18 @@ module Spotlight
 
     private
 
-    def available_index_fields
-      fields = blacklight_config.index_fields.map { |k, _v| { key: k, label: index_field_label(blacklight_config.document_model.new, k) } }
-      fields.unshift(key: document_show_link_field, label: t(:'spotlight.pages.form.title_placeholder')) unless index_fields.include? document_show_link_field
+    def available_caption_fields
+      default_caption_fields + blacklight_config.index_fields.reject { |_k, v| v.caption == false }.map { |k, v| { key: k, label: v.display_label } }
+    end
 
+    def default_caption_fields
+      fields = []
+      fields << { key: DOCUMENT_TITLE_KEY, label: t(:'spotlight.pages.form.title_placeholder') } unless blacklight_config.index_fields.key? DOCUMENT_TITLE_KEY
       fields
     end
 
     def available_view_configs
-      available_view_fields.map { |k, _| { key: k, label: view_label(k) } }
+      available_view_fields.map { |k, v| { key: k, label: v.display_label(k) } }
     end
 
     def attachment_endpoint

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/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

@@ -14,7 +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

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

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

@@ -5,6 +5,18 @@ module Spotlight
     # Raw solr document uploads
     class JsonUpload < Spotlight::Resource
       store :data, accessors: :json
+
+      # The indexing pipeline for JSON uploads copies the data from the stored
+      # `#data` field directly into the indexed document.
+      def self.indexing_pipeline
+        @indexing_pipeline ||= super.dup.tap do |pipeline|
+          pipeline.sources = [Spotlight::Etl::Sources::StoredData]
+
+          pipeline.transforms = [
+            Spotlight::Etl::Transforms::IdentityTransform
+          ] + pipeline.transforms
+        end
+      end
     end
   end
 end

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

@@ -10,8 +10,6 @@ module Spotlight
       # we want to do this before reindexing
       after_create :update_document_sidecar
 
-      self.document_builder_class = UploadSolrDocumentBuilder
-
       def self.fields(exhibit)
         @fields ||= {}
         @fields[exhibit] ||= begin
@@ -25,6 +23,15 @@ module Spotlight
         end
       end
 
+      def self.indexing_pipeline
+        @indexing_pipeline ||= super.dup.tap do |pipeline|
+          pipeline.transforms = [
+            ->(data, p) { data.merge({ p.context.document_model.unique_key.to_sym => p.source.compound_id }) },
+            Spotlight::Etl::Transforms::SourceMethodTransform(:to_solr)
+          ] + pipeline.transforms
+        end
+      end
+
       def compound_id
         "#{exhibit_id}-#{id}"
       end
@@ -33,6 +40,22 @@ module Spotlight
         @sidecar ||= document_model.new(id: compound_id).sidecar(exhibit)
       end
 
+      def to_solr
+        return {} unless upload&.file_present?
+
+        spotlight_routes = Spotlight::Engine.routes.url_helpers
+        riiif = Riiif::Engine.routes.url_helpers
+
+        dimensions = Riiif::Image.new(upload_id).info
+
+        {
+          spotlight_full_image_width_ssm: dimensions.width,
+          spotlight_full_image_height_ssm: dimensions.height,
+          Spotlight::Engine.config.thumbnail_field => riiif.image_path(upload, size: '!400,400'),
+          Spotlight::Engine.config.iiif_manifest_field => spotlight_routes.manifest_exhibit_solr_document_path(exhibit, compound_id)
+        }
+      end
+
       private
 
       def configured_fields