410 file analytics

app/models/file_download_stat_decorator.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax hyrax-v3.5.0 to require Hyrax::Download so the method below doesn't fail
+
+Hyrax::Download # rubocop:disable Lint/Void
+
+module FileDownloadStatClass
+  # Hyrax::Download is sent to Hyrax::Analytics.profile as #hyrax__download
+  # see Legato::ProfileMethods.method_name_from_klass
+  def ga_statistics(start_date, file)
+    profile = Hyrax::Analytics.profile
+    unless profile
+      Hyrax.logger.error("Google Analytics profile has not been established. Unable to fetch statistics.")
+      return []
+    end
+    # OVERRIDE Hyrax hyrax-v3.5.0
+    profile.hyrax__download(sort: 'date',
+                            start_date: start_date,
+                            end_date: Date.yesterday,
+                            limit: 10_000)
+           .for_file(file.id)
+  end
+end
+
+FileDownloadStat.singleton_class.send(:prepend, FileDownloadStatClass)

app/models/hyrax/statistic_decorator.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax hyrax-v3.5.0 to require Hyrax::Pageview so the method below doesn't fail
+
+Hyrax::Pageview # rubocop:disable Lint/Void
+
+module Hyrax
+  module StatisticClassDecorator
+    # Hyrax::Pageview is sent to Hyrax::Analytics.profile as #hyrax__pageview
+    # see Legato::ProfileMethods.method_name_from_klass
+    def ga_statistics(start_date, object)
+      path = polymorphic_path(object)
+      profile = Hyrax::Analytics.profile
+      unless profile
+        Rails.logger.error("Google Analytics profile has not been established. Unable to fetch statistics.")
+        return []
+      end
+      # OVERRIDE Hyrax hyrax-v3.5.0
+      profile.hyrax__pageview(sort: 'date',
+                              start_date: start_date,
+                              end_date: Date.yesterday,
+                              limit: 10_000)
+             .for_path(path)
+    end
+  end
+end
+
+Hyrax::Statistic.singleton_class.send(:prepend, Hyrax::StatisticClassDecorator)

app/models/work_view_stat_spec.rb

@@ -0,0 +1,114 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax hyrax-v3.5.0 to reference hyrax__pageviews as listed in app/models/hyrax/statistic_decorator.rb
+
+# rubocop:disable Metrics/BlockLength
+RSpec.describe WorkViewStat, type: :model do
+  let(:work_id) { work.id }
+  let(:user_id) { 123 }
+  let(:date) { Time.new.in_time_zone }
+  let(:work_stat) { described_class.create(work_views: "25", date: date, work_id: work_id, user_id: user_id) }
+  let(:work) { mock_model(GenericWork, id: 199) }
+
+  it "has attributes" do
+    expect(work_stat).to respond_to(:work_views)
+    expect(work_stat).to respond_to(:date)
+    expect(work_stat).to respond_to(:work_id)
+    expect(work_stat.work_id).to eq("199")
+    expect(work_stat.date).to eq(date)
+    expect(work_stat.work_views).to eq(25)
+    expect(work_stat.user_id).to eq(user_id)
+  end
+
+  describe ".ga_statistic" do
+    let(:start_date) { 2.days.ago }
+    let(:expected_path) { Rails.application.routes.url_helpers.hyrax_generic_work_path(work) }
+
+    before do
+      allow(Hyrax::Analytics).to receive(:profile).and_return(profile)
+    end
+    context "when a profile is available" do
+      let(:views) { double }
+      # OVERRIDE Hyrax hyrax-v3.5.0
+      let(:profile) { double(hyrax__pageviews: views) }
+
+      it "calls the Legato method with the correct path" do
+        expect(views).to receive(:for_path).with(expected_path)
+        described_class.ga_statistics(start_date, work)
+      end
+    end
+
+    context "when a profile not available" do
+      let(:profile) { nil }
+
+      it "calls the Legato method with the correct path" do
+        expect(described_class.ga_statistics(start_date, work)).to be_empty
+      end
+    end
+  end
+
+  describe "#statistics" do
+    let(:dates) do
+      ldates = []
+      4.downto(0) { |idx| ldates << (Time.zone.today - idx.day) }
+      ldates
+    end
+    let(:date_strs) do
+      dates.map { |date| date.strftime("%Y%m%d") }
+    end
+
+    let(:view_output) do
+      [[statistic_date(dates[0]), 4], [statistic_date(dates[1]), 8], [statistic_date(dates[2]), 6], [statistic_date(dates[3]), 10]]
+    end
+
+    # This is what the data looks like that's returned from Google Analytics (GA) via the Legato gem
+    # Due to the nature of querying GA, testing this data in an automated fashion is problematc.
+    # Sample data structures were created by sending real events to GA from a test instance of
+    # Scholarsphere.  The data below are essentially a "cut and paste" from the output of query
+    # results from the Legato gem.
+    let(:sample_work_pageview_statistics) do
+      [
+        SpecStatistic.new(date: date_strs[0], pageviews: 4),
+        SpecStatistic.new(date: date_strs[1], pageviews: 8),
+        SpecStatistic.new(date: date_strs[2], pageviews: 6),
+        SpecStatistic.new(date: date_strs[3], pageviews: 10)
+      ]
+    end
+
+    describe "cache empty" do
+      let(:stats) do
+        expect(described_class).to receive(:ga_statistics).and_return(sample_work_pageview_statistics)
+        described_class.statistics(work, Time.zone.today - 4.days, user_id)
+      end
+
+      it "includes cached ga data" do
+        expect(stats.map(&:to_flot)).to include(*view_output)
+      end
+
+      it "caches data" do
+        expect(stats.map(&:to_flot)).to include(*view_output)
+        expect(stats.first.user_id).to eq user_id
+
+        # at this point all data should be cached
+        allow(described_class).to receive(:ga_statistics).with(Time.zone.today, work).and_raise("We should not call Google Analytics All data should be cached!")
+
+        stats2 = described_class.statistics(work, Time.zone.today - 5.days)
+        expect(stats2.map(&:to_flot)).to include(*view_output)
+      end
+    end
+
+    describe "cache loaded" do
+      let!(:work_view_stat) { described_class.create(date: (Time.zone.today - 5.days).to_datetime, work_id: work_id, work_views: "25") }
+
+      let(:stats) do
+        expect(described_class).to receive(:ga_statistics).and_return(sample_work_pageview_statistics)
+        described_class.statistics(work, Time.zone.today - 5.days)
+      end
+
+      it "includes cached data" do
+        expect(stats.map(&:to_flot)).to include([work_view_stat.date.to_i * 1000, work_view_stat.work_views], *view_output)
+      end
+    end
+  end
+end
+# rubocop:enable Metrics/BlockLength

spec/models/file_download_stat_spec.rb

@@ -0,0 +1,109 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax hyrax-v3.5.0 to reference hyrax__pageviews as listed in app/models/file_download_stat_decorator.rb
+
+RSpec.describe FileDownloadStat, type: :model do
+  let(:file_id) { file.id }
+  let(:date) { Time.current }
+  let(:file_stat) { described_class.new(downloads: "2", date: date, file_id: file_id) }
+  let(:file) { mock_model(FileSet, id: 99) }
+
+  it "has attributes" do
+    expect(file_stat).to respond_to(:downloads)
+    expect(file_stat).to respond_to(:date)
+    expect(file_stat).to respond_to(:file_id)
+    expect(file_stat.file_id).to eq("99")
+    expect(file_stat.date).to eq(date)
+    expect(file_stat.downloads).to eq(2)
+  end
+
+  describe ".ga_statistic" do
+    let(:start_date) { 2.days.ago }
+    let(:expected_path) { Rails.application.routes.url_helpers.hyrax_file_set_path(file) }
+
+    before do
+      allow(Hyrax::Analytics).to receive(:profile).and_return(profile)
+    end
+    context "when a profile is available" do
+      let(:views) { double }
+      # OVERRIDE Hyrax hyrax-v3.5.0
+      let(:profile) { double(hyrax__download: views) }
+
+      it "calls the Legato method with the correct path" do
+        expect(views).to receive(:for_file).with(99)
+        described_class.ga_statistics(start_date, file)
+      end
+    end
+
+    context "when a profile not available" do
+      let(:profile) { nil }
+
+      it "calls the Legato method with the correct path" do
+        expect(described_class.ga_statistics(start_date, file)).to be_empty
+      end
+    end
+  end
+
+  describe "#statistics" do
+    let(:dates) do
+      ldates = []
+      4.downto(0) { |idx| ldates << (Time.zone.today - idx.day) }
+      ldates
+    end
+    let(:date_strs) do
+      dates.map { |date| date.strftime("%Y%m%d") }
+    end
+
+    let(:download_output) do
+      [[statistic_date(dates[0]), 1], [statistic_date(dates[1]), 1], [statistic_date(dates[2]), 2], [statistic_date(dates[3]), 3]]
+    end
+
+    # This is what the data looks like that's returned from Google Analytics (GA) via the Legato gem
+    # Due to the nature of querying GA, testing this data in an automated fashion is problematc.
+    # Sample data structures were created by sending real events to GA from a test instance of
+    # Scholarsphere.  The data below are essentially a "cut and paste" from the output of query
+    # results from the Legato gem.
+    let(:sample_download_statistics) do
+      [
+        SpecStatistic.new(eventCategory: "Files", eventAction: "Downloaded", eventLabel: "hyrax:x920fw85p", date: date_strs[0], totalEvents: "1"),
+        SpecStatistic.new(eventCategory: "Files", eventAction: "Downloaded", eventLabel: "hyrax:x920fw85p", date: date_strs[1], totalEvents: "1"),
+        SpecStatistic.new(eventCategory: "Files", eventAction: "Downloaded", eventLabel: "hyrax:x920fw85p", date: date_strs[2], totalEvents: "2"),
+        SpecStatistic.new(eventCategory: "Files", eventAction: "Downloaded", eventLabel: "hyrax:x920fw85p", date: date_strs[3], totalEvents: "3")
+      ]
+    end
+
+    describe "cache empty" do
+      let(:stats) do
+        expect(described_class).to receive(:ga_statistics).and_return(sample_download_statistics)
+        described_class.statistics(file, Time.zone.today - 4.days)
+      end
+
+      it "includes cached ga data" do
+        expect(stats.map(&:to_flot)).to include(*download_output)
+      end
+
+      it "caches data" do
+        expect(stats.map(&:to_flot)).to include(*download_output)
+
+        # at this point all data should be cached
+        allow(described_class).to receive(:ga_statistics).with(Time.zone.today, file).and_raise("We should not call Google Analytics All data should be cached!")
+
+        stats2 = described_class.statistics(file, Time.zone.today - 4.days)
+        expect(stats2.map(&:to_flot)).to include(*download_output)
+      end
+    end
+
+    describe "cache loaded" do
+      let!(:file_download_stat) { described_class.create(date: (Time.zone.today - 5.days).to_datetime, file_id: file_id, downloads: "25") }
+
+      let(:stats) do
+        expect(described_class).to receive(:ga_statistics).and_return(sample_download_statistics)
+        described_class.statistics(file, Time.zone.today - 5.days)
+      end
+
+      it "includes cached data" do
+        expect(stats.map(&:to_flot)).to include([file_download_stat.date.to_i * 1000, file_download_stat.downloads], *download_output)
+      end
+    end
+  end
+end