aggregations for views, downloads and citations. #390

datacite · Feb 2, 2020 · ee4a10f · ee4a10f
1 parent fd66e25
commit ee4a10f
Show file tree

Hide file tree

Showing 9 changed files with 277 additions and 36 deletions.
diff --git a/app/controllers/concerns/countable.rb b/app/controllers/concerns/countable.rb
@@ -14,7 +14,7 @@ def doi_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id: nil
       else
         response = Doi.query(nil, page: { number: 1, size: 0 })
       end
-      
+
       response.results.total.positive? ? facet_by_year(response.response.aggregations.created.buckets) : []
     end
 
@@ -30,8 +30,8 @@ def view_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id: ni
       else
         response = Doi.query(nil, page: { number: 1, size: 0 })
       end
-      
-      response.results.total.positive? ? facet_by_year(response.response.aggregations.views.buckets) : 0
+
+      response.results.total.positive? ? metric_facet_by_year(response.response.aggregations.views.buckets) : []
     end
 
     def download_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id: nil, state: nil)
@@ -46,8 +46,8 @@ def download_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id
       else
         response = Doi.query(nil, page: { number: 1, size: 0 })
       end
-      
-      response.results.total.positive? ? facet_by_year(response.response.aggregations.downloads.buckets) : 0
+
+      response.results.total.positive? ? metric_facet_by_year(response.response.aggregations.downloads.buckets) : []
     end
 
     def citation_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id: nil, state: nil)
@@ -62,8 +62,8 @@ def citation_count(client_id: nil, provider_id: nil, consortium_id: nil, user_id
       else
         response = Doi.query(nil, page: { number: 1, size: 0 })
       end
-      
-      response.results.total.positive? ? facet_by_year(response.response.aggregations.citations.buckets) : 0
+
+      response.results.total.positive? ? metric_facet_by_year(response.response.aggregations.citations.buckets) : []
     end
 
     # cumulative count clients by year

diff --git a/app/controllers/concerns/facetable.rb b/app/controllers/concerns/facetable.rb
@@ -33,6 +33,14 @@ def facet_by_year(arr)
       end
     end
 
+    def metric_facet_by_year(arr)
+      arr.map do |hsh|
+        { "id" => hsh["key_as_string"][0..3],
+          "title" => hsh["key_as_string"][0..3],
+          "count" => hsh.dig("metric_count", "value").to_i }
+      end
+    end
+
     def facet_annual(arr)
       arr.map do |hsh|
         { "id" => hsh["key"][0..3],

diff --git a/app/controllers/dois_controller.rb b/app/controllers/dois_controller.rb
@@ -170,43 +170,50 @@ def index
         certificates = nil
 
         bm = Benchmark.ms {
-          states = total > 0 ? facet_by_key(response.aggregations.states.buckets) : nil
-          resource_types = total > 0 ? facet_by_resource_type(response.aggregations.resource_types.buckets) : nil
-          years = total > 0 ? facet_by_year(response.aggregations.years.buckets) : nil
-          created = total > 0 ? facet_by_year(response.aggregations.created.buckets) : nil
-          registered = total > 0 ? facet_by_year(response.aggregations.registered.buckets) : nil
-          providers = total > 0 ? facet_by_provider(response.aggregations.providers.buckets) : nil
-          clients = total > 0 ? facet_by_client(response.aggregations.clients.buckets) : nil
-          prefixes = total > 0 ? facet_by_key(response.aggregations.prefixes.buckets) : nil
-          schema_versions = total > 0 ? facet_by_schema(response.aggregations.schema_versions.buckets) : nil
+          states = total.positive? ? facet_by_key(response.aggregations.states.buckets) : nil
+          resource_types = total.positive? ? facet_by_resource_type(response.aggregations.resource_types.buckets) : nil
+          years = total.positive? ? facet_by_year(response.aggregations.years.buckets) : nil
+          created = total.positive? ? facet_by_year(response.aggregations.created.buckets) : nil
+          registered = total.positive? ? facet_by_year(response.aggregations.registered.buckets) : nil
+          providers = total.positive? ? facet_by_provider(response.aggregations.providers.buckets) : nil
+          clients = total.positive? ? facet_by_client(response.aggregations.clients.buckets) : nil
+          prefixes = total.positive? ? facet_by_key(response.aggregations.prefixes.buckets) : nil
+          schema_versions = total.positive? ? facet_by_schema(response.aggregations.schema_versions.buckets) : nil
         }
         logger.warn method: "GET", path: "/dois", message: "AggregationsBasic /dois", duration: bm
 
         bm = Benchmark.ms {
-          affiliations = total > 0 ? facet_by_affiliation(response.aggregations.affiliations.buckets) : nil
-          sources = total > 0 ? facet_by_key(response.aggregations.sources.buckets) : nil
-          subjects = total > 0 ? facet_by_key(response.aggregations.subjects.buckets) : nil
-          certificates = total > 0 ? facet_by_key(response.aggregations.certificates.buckets) : nil
+          affiliations = total.positive? ? facet_by_affiliation(response.aggregations.affiliations.buckets) : nil
+          sources = total.positive? ? facet_by_key(response.aggregations.sources.buckets) : nil
+          subjects = total.positive? ? facet_by_key(response.aggregations.subjects.buckets) : nil
+          certificates = total.positive? ? facet_by_key(response.aggregations.certificates.buckets) : nil
         }
         logger.warn method: "GET", path: "/dois", message: "AggregationsExtended /dois", duration: bm
 
         bm = Benchmark.ms {
-          link_checks_status = total > 0 ? facet_by_cumulative_year(response.aggregations.link_checks_status.buckets) : nil
-          links_with_schema_org = total > 0 ? facet_by_cumulative_year(response.aggregations.link_checks_has_schema_org.buckets) : nil
-          link_checks_schema_org_id = total > 0 ? response.aggregations.link_checks_schema_org_id.value : nil
-          link_checks_dc_identifier = total > 0 ? response.aggregations.link_checks_dc_identifier.value : nil
-          link_checks_citation_doi = total > 0 ? response.aggregations.link_checks_citation_doi.value : nil
-          links_checked = total > 0 ? response.aggregations.links_checked.value : nil
+          link_checks_status = total.positive? ? facet_by_cumulative_year(response.aggregations.link_checks_status.buckets) : nil
+          links_with_schema_org = total.positive? ? facet_by_cumulative_year(response.aggregations.link_checks_has_schema_org.buckets) : nil
+          link_checks_schema_org_id = total.positive? ? response.aggregations.link_checks_schema_org_id.value : nil
+          link_checks_dc_identifier = total.positive? ? response.aggregations.link_checks_dc_identifier.value : nil
+          link_checks_citation_doi = total.positive? ? response.aggregations.link_checks_citation_doi.value : nil
+          links_checked = total.positive? ? response.aggregations.links_checked.value : nil
         }
         logger.warn method: "GET", path: "/dois", message: "AggregationsLinkChecks /dois", duration: bm
 
         if params[:mix_in] == "metrics"
           dois_names =  results.map { |result| result.dig(:_source, :doi) }.join(',')
           metrics_array = get_metrics_array(dois_names)
           results = mix_in_metrics_array(results, metrics_array)
-        end
 
-        person_metrics = params[:mix_in] == "metrics" ? get_person_metrics(params[:user_id]) : {}
+          person_metrics = get_person_metrics(params[:user_id])
+          citations = person_metrics[:citations]
+          views = person_metrics[:views]
+          downloads = person_metrics[:downloads]
+        else
+          citations = total.positive? ? metric_facet_by_year(response.aggregations.citations.buckets) : nil
+          views = total.positive? ? metric_facet_by_year(response.aggregations.views.buckets) : nil
+          downloads = total.positive? ? metric_facet_by_year(response.aggregations.downloads.buckets) : nil
+        end
 
         respond_to do |format|
           format.json do
@@ -234,9 +241,9 @@ def index
               "linkChecksDcIdentifier" => link_checks_dc_identifier,
               "linkChecksCitationDoi" => link_checks_citation_doi,
               subjects: subjects,
-              citations: person_metrics[:citations],
-              views: person_metrics[:views],
-              downloads: person_metrics[:downloads],
+              citations: citations,
+              views: views,
+              downloads: downloads,
             }.compact
 
             options[:links] = {
@@ -261,7 +268,7 @@ def index
               mix_in: params[:mix_in],
               metrics: metrics_array,
               affiliation: params[:affiliation],
-              is_collection: options[:is_collection]
+              is_collection: options[:is_collection],
             }
 
             # sparse fieldsets

diff --git a/app/models/doi.rb b/app/models/doi.rb
@@ -534,9 +534,24 @@ def self.query_aggregations
       sources: { terms: { field: 'source', size: 15, min_doc_count: 1 } },
       subjects: { terms: { field: 'subjects.subject', size: 15, min_doc_count: 1 } },
       certificates: { terms: { field: 'client.certificate', size: 15, min_doc_count: 1 } },
-      views: { value_count: { field: "views_count" } },
-      downloads: { value_count: { field: "download_count" } },
-      citations: { value_count: { field: "citation_count" } },
+      views: {
+        date_histogram: { field: "publication_year", interval: "year", min_doc_count: 1 },
+        aggs: {
+          metric_count: { sum: { field: "view_count" } },
+        }
+      },
+      downloads: {
+        date_histogram: { field: "publication_year", interval: "year", min_doc_count: 1 }, 
+        aggs: {
+          metric_count: { sum: { field: "download_count" } }, 
+        }
+      },
+      citations: {
+        date_histogram: { field: "publication_year", interval: "year", min_doc_count: 1 }, 
+        aggs: {
+          metric_count: { sum: { field: "citation_count" } },
+        }
+      },
     }
   end
 

diff --git a/app/models/event.rb b/app/models/event.rb
@@ -196,7 +196,7 @@ def as_indexed_json(options = {})
       "occurred_at" => occurred_at,
       "citation_id" => citation_id,
       "citation_year" => citation_year,
-      "cache_key" => cache_key
+      "cache_key" => cache_key,
     }
   end
 

diff --git a/spec/fixtures/vcr_cassettes/dois/citations/has_downloads_meta.yml b/spec/fixtures/vcr_cassettes/dois/citations/has_downloads_meta.yml
diff --git a/spec/fixtures/vcr_cassettes/dois/downloads/has_downloads_meta.yml b/spec/fixtures/vcr_cassettes/dois/downloads/has_downloads_meta.yml
diff --git a/spec/fixtures/vcr_cassettes/dois/views/has_views_meta.yml b/spec/fixtures/vcr_cassettes/dois/views/has_views_meta.yml
diff --git a/spec/requests/dois_spec.rb b/spec/requests/dois_spec.rb
@@ -196,6 +196,21 @@
       expect(json.dig('data', 'attributes', 'viewCount')).to eq(75)
       expect(json.dig('data', 'attributes', 'viewsOverTime')).to eq([{"total"=>25, "year_month"=>"2015-06"}, {"total"=>25, "year_month"=>"2015-06"}, {"total"=>25, "year_month"=>"2015-06"}])
     end
+
+    it "has views meta" do
+      get "/dois", nil, headers
+
+      expect(last_response.status).to eq(200)
+      expect(json.dig('meta', 'views')).to eq([{"count"=>75, "id"=>"2011", "title"=>"2011"}])
+    end
+
+    it "repository shows summary count" do
+      get "/repositories/#{client.uid}", nil, headers
+
+      expect(last_response.status).to eq(200)
+      expect(json.dig('data', 'attributes', 'name')).to eq(client.name)
+      expect(json.dig('meta', 'views')).to eq([{"count"=>75, "id"=>"2011", "title"=>"2011"}])
+    end
   end
 
   describe "downloads", elasticsearch: true, vcr: true do
@@ -218,6 +233,21 @@
       expect(json.dig('data', 'attributes', 'downloadCount')).to eq(30)
       expect(json.dig('data', 'attributes', 'downloadsOverTime')).to eq([{"total"=>10, "year_month"=>"2015-06"}, {"total"=>10, "year_month"=>"2015-06"}, {"total"=>10, "year_month"=>"2015-06"}])
     end
+
+    it "has downloads meta" do
+      get "/dois", nil, headers
+
+      expect(last_response.status).to eq(200)
+      expect(json.dig('meta', 'downloads')).to eq([{"count"=>30, "id"=>"2011", "title"=>"2011"}])
+    end
+
+    it "repository shows summary count" do
+      get "/repositories/#{client.uid}", nil, headers
+
+      expect(last_response.status).to eq(200)
+      expect(json.dig('data', 'attributes', 'name')).to eq(client.name)
+      expect(json.dig('meta', 'downloads')).to eq([{"count"=>30, "id"=>"2011", "title"=>"2011"}])
+    end
   end
 
   describe "references", elasticsearch: true, vcr: true do
@@ -266,6 +296,22 @@
       expect(json.dig('included').length).to eq(2)
       expect(json.dig('included', 0, 'attributes', 'relationTypeId')).to eq("is-referenced-by")
     end
+
+    it "has downloads meta" do
+      get "/dois", nil, headers
+
+      expect(last_response.status).to eq(200)
+      expect(json.dig('meta', 'citations')).to eq([{"count"=>1, "id"=>"2011", "title"=>"2011"}])
+    end
+
+    it "repository shows summary count" do
+      get "/repositories/#{client.uid}", nil, headers
+
+      puts last_response.body
+      expect(last_response.status).to eq(200)
+      expect(json.dig('data', 'attributes', 'name')).to eq(client.name)
+      expect(json.dig('meta', 'citations')).to eq([{"count"=>1, "id"=>"2011", "title"=>"2011"}])
+    end
   end
 
   describe "parts", elasticsearch: true, vcr: true do