blacklight-spotlight 3.0.0.rc3 → 3.0.0.rc4

data/app/models/concerns/spotlight/resources/open_graph.rb

@@ -1,36 +0,0 @@
-# frozen_string_literal: true
-
-module Spotlight
-  module Resources
-    ##
-    # OpenGraph metadata harvester
-    module OpenGraph
-      extend ActiveSupport::Concern
-      include Spotlight::Resources::Web
-
-      def opengraph
-        @opengraph ||= begin
-          page = {}
-
-          body.css('meta').select { |m| m.attribute('property') }.each do |m|
-            page[m.attribute('property').to_s] = m.attribute('content').to_s
-          end
-
-          page
-        end
-      end
-
-      def opengraph_properties
-        Hash[opengraph.map do |k, v|
-          ["#{opengraph_solr_field_name(k)}_tesim", v]
-        end]
-      end
-
-      private
-
-      def opengraph_solr_field_name(field)
-        field.parameterize(separator: '_')
-      end
-    end
-  end
-end

data/app/models/spotlight/reindexing_log_entry.rb

@@ -1,42 +0,0 @@
-# frozen_string_literal: true
-
-module Spotlight
-  ##
-  # a log entry representing an attempt to reindex some number of records in an exhibit
-  class ReindexingLogEntry < ActiveRecord::Base
-    enum job_status: { unstarted: 0, in_progress: 1, succeeded: 2, failed: 3 }
-
-    belongs_to :exhibit, class_name: 'Spotlight::Exhibit'
-    belongs_to :user, class_name: '::User', optional: true
-
-    # null start times sort to the top, to more easily surface pending reindexing
-    default_scope { order(Arel.sql('start_time IS NOT NULL, start_time DESC')) }
-    scope :recent, -> { limit(5) }
-    scope :started_or_completed, -> { where.not(job_status: 'unstarted') }
-
-    def duration
-      end_time - start_time if end_time
-    end
-
-    def in_progress!
-      self.start_time = Time.zone.now
-      super
-    rescue StandardError
-      Rails.logger.error "unexpected error updating log entry to :in_progress from #{caller}"
-    end
-
-    def succeeded!
-      self.end_time = Time.zone.now
-      super
-    rescue StandardError
-      Rails.logger.error "unexpected error updating log entry to :succeeded from #{caller}"
-    end
-
-    def failed!
-      self.end_time = Time.zone.now
-      super
-    rescue StandardError
-      Rails.logger.error "unexpected error updating log entry to :failed from #{caller}"
-    end
-  end
-end

data/app/services/spotlight/resources/iiif_builder.rb

@@ -1,19 +0,0 @@
-# frozen_string_literal: true
-
-module Spotlight
-  module Resources
-    # transforms a IiifHarvester into solr documents
-    class IiifBuilder < Spotlight::SolrDocumentBuilder
-      def to_solr
-        return to_enum(:to_solr) { 0 } unless block_given?
-
-        base_doc = super
-        resource.iiif_manifests.each do |manifest|
-          manifest.with_exhibit(exhibit)
-          manifest_solr = manifest.to_solr
-          yield base_doc.merge(manifest_solr) if manifest_solr.present?
-        end
-      end
-    end
-  end
-end

data/app/services/spotlight/solr_document_builder.rb

@@ -1,77 +0,0 @@
-# frozen_string_literal: true
-
-module Spotlight
-  # Creates solr documents for the documents in a resource
-  class SolrDocumentBuilder
-    def initialize(resource)
-      @resource = resource
-    end
-
-    attr_reader :resource
-
-    delegate :exhibit, :document_model, to: :resource
-
-    ##
-    # @return an enumerator of all the indexable documents for this resource
-    def documents_to_index
-      data = to_solr
-      return [] if data.blank?
-
-      data &&= [data] if data.is_a? Hash
-
-      return to_enum(:documents_to_index) { data.size } unless block_given?
-
-      data.lazy.reject(&:blank?).each do |doc|
-        yield doc.reverse_merge(exhibit_solr_doc(doc[unique_key]).to_solr)
-      end
-    end
-
-    protected
-
-    ##
-    # @abstract
-    # Convert this resource into zero-to-many new solr documents. The data here
-    # should be merged into the resource-specific {#to_solr} data.
-    #
-    # @return [Hash] a single solr document hash
-    # @return [Enumerator<Hash>] multiple solr document hashes. This can be a
-    #   simple array, or an lazy enumerator
-    def to_solr
-      spotlight_resource_metadata_for_solr
-    end
-
-    private
-
-    # Null object for SolrDocument
-    module NilSolrDocument
-      def self.to_solr
-        {}
-      end
-    end
-
-    ##
-    # Get any exhibit-specific metadata stored in e.g. sidecars, tags, etc
-    # This needs the generated solr document
-    # @returns [#to_solr] something that responds to `to_solr'
-    def exhibit_solr_doc(id)
-      return NilSolrDocument unless document_model || id.present?
-
-      document_model.build_for_exhibit(id, exhibit, resource: (resource if resource.persisted?))
-    end
-
-    def unique_key
-      if document_model
-        document_model.unique_key.to_sym
-      else
-        :id
-      end
-    end
-
-    def spotlight_resource_metadata_for_solr
-      {
-        Spotlight::Engine.config.resource_global_id_field => (resource.to_global_id.to_s if resource.persisted?),
-        document_model.resource_type_field => resource.class.to_s.tableize
-      }
-    end
-  end
-end

data/app/services/spotlight/upload_solr_document_builder.rb

@@ -1,57 +0,0 @@
-# frozen_string_literal: true
-
-module Spotlight
-  # Creates solr documents for the uploaded documents in a resource
-  class UploadSolrDocumentBuilder < SolrDocumentBuilder
-    delegate :compound_id, to: :resource
-
-    def to_solr
-      super.tap do |solr_hash|
-        add_default_solr_fields solr_hash
-        add_sidecar_fields solr_hash
-
-        if attached_file?
-          add_image_dimensions solr_hash
-          add_file_versions solr_hash
-          add_manifest_path solr_hash
-        end
-      end
-    end
-
-    private
-
-    def add_default_solr_fields(solr_hash)
-      solr_hash[exhibit.blacklight_config.document_model.unique_key.to_sym] = compound_id
-    end
-
-    def add_image_dimensions(solr_hash)
-      dimensions = Riiif::Image.new(resource.upload_id).info
-      solr_hash[:spotlight_full_image_width_ssm] = dimensions.width
-      solr_hash[:spotlight_full_image_height_ssm] = dimensions.height
-    end
-
-    def add_file_versions(solr_hash)
-      solr_hash[Spotlight::Engine.config.thumbnail_field] = riiif.image_path(resource.upload_id, size: '!400,400')
-    end
-
-    def add_sidecar_fields(solr_hash)
-      solr_hash.merge! resource.sidecar.to_solr
-    end
-
-    def add_manifest_path(solr_hash)
-      solr_hash[Spotlight::Engine.config.iiif_manifest_field] = spotlight_routes.manifest_exhibit_solr_document_path(exhibit, resource.compound_id)
-    end
-
-    def spotlight_routes
-      Spotlight::Engine.routes.url_helpers
-    end
-
-    def riiif
-      Riiif::Engine.routes.url_helpers
-    end
-
-    def attached_file?
-      resource.upload&.file_present?
-    end
-  end
-end

data/spec/factories/reindexing_log_entries.rb

@@ -1,54 +0,0 @@
-# frozen_string_literal: true
-
-FactoryBot.define do
-  factory :unstarted_reindexing_log_entry, class: 'Spotlight::ReindexingLogEntry' do
-    items_reindexed_count { 15 }
-    job_status { 'unstarted' }
-    exhibit
-    user
-  end
-
-  factory :reindexing_log_entry, class: 'Spotlight::ReindexingLogEntry' do
-    items_reindexed_count { 10 }
-    start_time { Time.zone.parse('2017-01-05 23:00:00') }
-    end_time { Time.zone.parse('2017-01-05 23:05:00') }
-    job_status { 'succeeded' }
-    exhibit
-    user
-  end
-
-  factory :reindexing_log_entry_no_user, class: 'Spotlight::ReindexingLogEntry' do
-    items_reindexed_count { 10 }
-    start_time { Time.zone.parse('2017-01-05 23:00:00') }
-    end_time { Time.zone.parse('2017-01-05 23:05:00') }
-    job_status { 'succeeded' }
-    exhibit
-  end
-
-  factory :in_progress_reindexing_log_entry, class: 'Spotlight::ReindexingLogEntry' do
-    items_reindexed_count { 100 }
-    start_time { Time.zone.now - 300 }
-    end_time { nil }
-    job_status { 'in_progress' }
-    exhibit
-    user
-  end
-
-  factory :recent_reindexing_log_entry, class: 'Spotlight::ReindexingLogEntry' do
-    sequence(:items_reindexed_count)
-    start_time { Time.zone.now - 86_400 }
-    end_time { Time.zone.now - 86_100 }
-    job_status { 'succeeded' }
-    exhibit
-    user
-  end
-
-  factory :failed_reindexing_log_entry, class: 'Spotlight::ReindexingLogEntry' do
-    items_reindexed_count { 10 }
-    start_time { Time.zone.parse('2017-01-10 23:00:00') }
-    end_time { Time.zone.parse('2017-01-10 23:05:00') }
-    job_status { 'failed' }
-    exhibit
-    user
-  end
-end

data/spec/models/spotlight/reindexing_log_entry_spec.rb

@@ -1,129 +0,0 @@
-# frozen_string_literal: true
-
-describe Spotlight::ReindexingLogEntry, type: :model do
-  subject { FactoryBot.build(:reindexing_log_entry) }
-
-  describe 'scope' do
-    before do
-      (0..10).to_a.each { FactoryBot.create(:recent_reindexing_log_entry) }
-      FactoryBot.create(:unstarted_reindexing_log_entry)
-      (0..10).to_a.each { FactoryBot.create(:recent_reindexing_log_entry) }
-    end
-
-    let(:sorted_log_entry_list) do
-      unstarted_entries = described_class.where(start_time: nil).to_a
-      started_entries = described_class.where.not(start_time: nil).to_a.sort_by(&:start_time).reverse
-      unstarted_entries + started_entries # null start times should be first
-    end
-
-    context 'default' do
-      it 'sorts by start_time in descending order' do
-        default_log_entry_list = described_class.all.to_a
-        expect(default_log_entry_list).to eq sorted_log_entry_list
-      end
-    end
-
-    context 'recent' do
-      it 'returns the most recent 5 entries (sorted by start_time descending)' do
-        recent_log_entry_list = described_class.recent.to_a
-        expect(recent_log_entry_list).to eq sorted_log_entry_list[0..4]
-      end
-    end
-  end
-
-  describe '#duration' do
-    context 'when end_time is present' do
-      it 'is calculated as difference between end_time and start_time' do
-        expect(subject.duration).to eq 300
-      end
-    end
-
-    context 'when end_time is not present' do
-      subject { FactoryBot.build(:in_progress_reindexing_log_entry) }
-
-      it 'is nil' do
-        expect(subject.duration).to be nil
-      end
-    end
-  end
-
-  describe 'state updating methods' do
-    describe '#in_progress!' do
-      subject { FactoryBot.build(:unstarted_reindexing_log_entry) }
-
-      context 'executes normally' do
-        it 'sets start_time and job_status' do
-          lower_time_bound = Time.zone.now
-          subject.in_progress!
-          upper_time_bound = Time.zone.now
-
-          expect(subject.start_time).to be_between(lower_time_bound, upper_time_bound)
-          expect(subject.job_status).to eq 'in_progress'
-        end
-      end
-
-      context 'encounters an unexpected error' do
-        it "traps the exception and logs an error so that the caller doesn't have to deal with it" do
-          expect(subject).to receive(:'start_time=').and_raise StandardError.new # try to blow up the in_progress! call
-          expect(Rails.logger).to receive(:error) do |arg|
-            expect(arg).to match(/^unexpected error updating log entry to :in_progress from \[".*reindexing_log_entry.rb/)
-          end
-
-          expect { subject.in_progress! }.not_to raise_error
-        end
-      end
-    end
-
-    describe '#succeeded!' do
-      subject { FactoryBot.build(:in_progress_reindexing_log_entry) }
-
-      context 'executes normally' do
-        it 'sets end_time and job_status' do
-          lower_time_bound = Time.zone.now
-          subject.succeeded!
-          upper_time_bound = Time.zone.now
-
-          expect(subject.end_time).to be_between(lower_time_bound, upper_time_bound)
-          expect(subject.job_status).to eq 'succeeded'
-        end
-      end
-
-      context 'encounters an unexpected error' do
-        it "traps the exception and logs an error so that the caller doesn't have to deal with it" do
-          expect(subject).to receive(:'end_time=').and_raise StandardError.new # try to blow up the succeeded! call
-          expect(Rails.logger).to receive(:error) do |arg|
-            expect(arg).to match(/^unexpected error updating log entry to :succeeded from \[".*reindexing_log_entry.rb/)
-          end
-
-          expect { subject.succeeded! }.not_to raise_error
-        end
-      end
-    end
-
-    describe '#failed!' do
-      subject { FactoryBot.build(:in_progress_reindexing_log_entry) }
-
-      context 'executes normally' do
-        it 'sets end_time and job_status' do
-          lower_time_bound = Time.zone.now
-          subject.failed!
-          upper_time_bound = Time.zone.now
-
-          expect(subject.end_time).to be_between(lower_time_bound, upper_time_bound)
-          expect(subject.job_status).to eq 'failed'
-        end
-      end
-
-      context 'encounters an unexpected error' do
-        it "traps the exception and logs an error so that the caller doesn't have to deal with it" do
-          expect(subject).to receive(:'end_time=').and_raise StandardError.new # try to blow up the failed! call
-          expect(Rails.logger).to receive(:error) do |arg|
-            expect(arg).to match(/^unexpected error updating log entry to :failed from \[".*reindexing_log_entry.rb/)
-          end
-
-          expect { subject.failed! }.not_to raise_error
-        end
-      end
-    end
-  end
-end

data/spec/models/spotlight/resources/open_graph_spec.rb

@@ -1,65 +0,0 @@
-# frozen_string_literal: true
-
-describe Spotlight::Resources::OpenGraph, type: :model do
-  class TestDocBuilder < Spotlight::SolrDocumentBuilder
-    def to_solr
-      super.merge(resource.opengraph_properties)
-    end
-  end
-
-  class TestResource < Spotlight::Resource
-    self.document_builder_class = TestDocBuilder
-    include Spotlight::Resources::OpenGraph
-  end
-
-  subject { TestResource.new url: 'info:url' }
-
-  let(:exhibit) { double(solr_data: {}, blacklight_config: Blacklight::Configuration.new) }
-
-  describe '#to_solr' do
-    before do
-      allow(subject).to receive_messages id: 15, opengraph_properties: {}, exhibit: exhibit, persisted?: true
-    end
-
-    let(:solr_doc) { subject.document_builder.to_solr }
-
-    it 'includes this record id' do
-      expect(solr_doc).to include spotlight_resource_id_ssim: subject.to_global_id.to_s
-    end
-
-    it 'includes opengraph properties' do
-      allow(subject).to receive_messages opengraph_properties: { a: 1, b: 2 }
-
-      expect(solr_doc).to include a: 1, b: 2
-    end
-  end
-
-  describe '#opengraph_properties' do
-    it 'maps opengraph properties to solr fields' do
-      allow(subject).to receive_messages opengraph: { 'og_title' => 'title', 'og_description' => 'description' }
-      expect(subject.opengraph_properties).to include 'og_title_tesim' => 'title', 'og_description_tesim' => 'description'
-    end
-  end
-
-  describe '#opengraph' do
-    let(:body) do
-      Nokogiri::HTML.parse <<-EOF
-        <html><head>
-        <meta property="og:title" content="The Ground Truth: The Human Cost of War"/>
-        <meta property="og:description" content="The Ground Truth: The Human Cost of War is our soldiers' perspective of the Iraq War"/>
-        <meta property="og:type" content="video.movie"/>
-        <meta property="og:site_name" content="Internet Archive"/>
-        <meta property="og:video" content="https://archive.org/download/Ground_Truth/GroundTruth1_bb_512kb.mp4"/>
-        <meta property="og:video:width" content="320"/>
-        <meta property="og:video:height" content="240"/>
-        </head></html>
-      EOF
-    end
-
-    it 'extracts opengraph <meta> tags' do
-      allow(subject).to receive_messages(body: body)
-      expect(subject.opengraph).to include 'og:title', 'og:description', 'og:type', 'og:type', 'og:site_name', 'og:video', 'og:video:width', 'og:video:height'
-      expect(subject.opengraph['og:title']).to eq 'The Ground Truth: The Human Cost of War'
-    end
-  end
-end