disable aggregation unique citations

datacite · Jul 30, 2019 · 2ab6f96 · 2ab6f96
1 parent 6e35753
commit 2ab6f96
Show file tree

Hide file tree

Showing 3 changed files with 67 additions and 37 deletions.
diff --git a/app/controllers/concerns/facetable.rb b/app/controllers/concerns/facetable.rb
@@ -134,12 +134,26 @@ def facet_by_source(arr)
       end
     end
 
+    # def facet_citations_by_year(hash)
+    #   hash.map do |hsh|
+    #     { "id" => hsh["key"].to_i,
+    #       "title" => hsh["key"],
+    #       "count" => hsh["doc_count"] }
+    #   end
+    # end
+
     def facet_citations_by_year(hash)
-      hash.map do |hsh|
-        { "id" => hsh["key"].to_i,
-          "title" => hsh["key"],
-          "count" => hsh["doc_count"] }
+      arr = hash.dig('years', 'buckets').map do |h|
+        year = h['key_as_string'][0..3].to_i
+        title = h['key_as_string'][0..3]
+
+        {
+          'id' => year,
+          'title' => title,
+          'sum' => h.dig('total_by_year', 'value') }
       end
+      { "count" => hash.dig("sum_distribution", "value"),
+        "years" => arr }
     end
 
     def facet_by_relation_type(arr)

diff --git a/app/controllers/events_controller.rb b/app/controllers/events_controller.rb
@@ -121,8 +121,9 @@ def index
     pairings = total.positive? && params[:extra] ? facet_by_pairings(response.response.aggregations.pairings.buckets) : nil
     dois = total.positive? && params[:extra] ? facet_by_dois(response.response.aggregations.dois.buckets) : nil
     dois_usage = total.positive? && params[:extra] ? facet_by_dois(response.response.aggregations.dois_usage.dois.buckets) : nil
-    citations_histogram = total.positive? && params[:extra] ? facet_citations_by_year(response.response.aggregations.citations_histogram.years.buckets) : nil
-    citations = total.positive? && params[:extra] ? facet_citations_by_dois(response.response.aggregations.citations.dois.buckets) : nil
+    citations_histogram = total.positive? && params[:extra] ? facet_citations_by_year(response.response.aggregations.dois_citations) : nil
+    # citations_histogram = total.positive? && params[:extra] ? facet_citations_by_year(response.response.aggregations.citations_histogram.years.buckets) : nil
+    # citations = total.positive? && params[:extra] ? facet_citations_by_dois(response.response.aggregations.citations.dois.buckets) : nil
 
     results = response.results
 
@@ -139,8 +140,8 @@ def index
       registrants: registrants,
       "doisRelationTypes": dois,
       "doisUsageTypes": dois_usage,
-      "doisCitations": citations_histogram,
-      "uniqueCitations": citations
+      "doisCitations": citations_histogram
+      # "uniqueCitations": citations
     }.compact
 
     options[:links] = {
@@ -165,14 +166,16 @@ def index
         "page[size]" => page[:size] }.compact.to_query
       }.compact
 
-    options[:include] = []
     options[:is_collection] = true
 
     bmr = Benchmark.ms {
-      doi_names = (results.map { |event| event.doi}).uniq().join(",")
       events_serialized = EventSerializer.new(results, options).serializable_hash
-      events_serialized[:included] = DoiSerializer.new((Doi.find_by_id(doi_names).results), {is_collection: true}).serializable_hash.dig(:data) 
-      render json: events_serialized, status: :ok
+      if @include.include?(:dois)
+        options[:include] =[]
+        doi_names = (results.map { |event| event.doi}).uniq().join(",")
+        events_serialized[:included] = DoiSerializer.new((Doi.find_by_id(doi_names).results), {is_collection: true}).serializable_hash.dig(:data) 
+      end
+        render json: events_serialized, status: :ok
     }
 
     if bmr > 3000

diff --git a/app/models/event.rb b/app/models/event.rb
@@ -180,6 +180,12 @@ def self.query_aggregations
       }
     }
 
+    sum_year_distribution = {
+      sum_bucket: {
+        buckets_path: "years>total_by_year"
+      }
+    }
+
     {
       sources: { terms: { field: 'source_id', size: 50, min_doc_count: 1 } },
       prefixes: { terms: { field: 'prefix', size: 50, min_doc_count: 1 } },
@@ -193,37 +199,44 @@ def self.query_aggregations
         aggs: {
           dois: { terms: { field: 'obj_id', size: 50, min_doc_count: 1 }, aggs: { relation_types: { terms: { field: 'relation_type_id',size: 50, min_doc_count: 1 }, aggs: { "total_by_type" => { sum: { field: 'total' }}}}} } }
       },
-      citations_histogram: {
+      dois_citations: {
         filter: {
           script: {
             script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"
           }
         },
-        aggs: { years: { terms: { script:  { source: "
-          String subjDatePublished = params['_source']['subj']['date_published']?.substring(0, 4);
-          String objDatePublished  = params['_source']['obj']['date_published']?.substring(0, 4);
+        aggs: { years: { date_histogram: { field: 'occurred_at', interval: 'year', min_doc_count: 1 }, aggs: { "total_by_year" => { sum: { field: 'total' }}}},"sum_distribution"=>sum_year_distribution}
+      # citations_histogram: {
+      #   filter: {
+      #     script: {
+      #       script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"
+      #     }
+      #   },
+      #   aggs: { years: { terms: { script:  { source: "
+      #     String subjDatePublished = params['_source']['subj']['date_published']?.substring(0, 4);
+      #     String objDatePublished  = params['_source']['obj']['date_published']?.substring(0, 4);
 
-          if( params['_source']['subj']['date_published']?.substring(0, 4) !== null && params['_source']['obj']['date_published']?.substring(0, 4) !== null){
-
-            if(Integer.parseInt(objDatePublished) > Integer.parseInt(subjDatePublished) )
-            {
-              objDatePublished
-            }
-            else{
-              subjDatePublished
-            }
-          } 
-        " }}}}
-      },
-      citations: {
-        filter: {
-          script: {
-            script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"
-          }
-        },
-        aggs: { dois: {
-           terms: { field: 'obj_id', size: 50, min_doc_count: 1 }, aggs: { unique_citations: { cardinality: { field: 'citation_id' }}}
-        }}
+      #     if( params['_source']['subj']['date_published']?.substring(0, 4) !== null && params['_source']['obj']['date_published']?.substring(0, 4) !== null){
+
+      #       if(Integer.parseInt(objDatePublished) > Integer.parseInt(subjDatePublished) )
+      #       {
+      #         objDatePublished
+      #       }
+      #       else{
+      #         subjDatePublished
+      #       }
+      #     } 
+      #   " }}}}
+      # },
+      # citations: {
+      #   filter: {
+      #     script: {
+      #       script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"
+      #     }
+      #   },
+      #   aggs: { dois: {
+      #      terms: { field: 'obj_id', size: 50, min_doc_count: 1 }, aggs: { unique_citations: { cardinality: { field: 'citation_id' }}}
+      #   }}
       }
     }
   end