Merge pull request #349 from datacite/feature_refenreces_citaitons

Feature references, citations, relations links

app/controllers/concerns/facetable.rb

@@ -312,7 +312,7 @@ def facet_citations_by_dois(arr)
       arr.map do |hsh|
         { "id" => hsh["key"],
           "title" => hsh["key"],
-          "citations" => hsh.dig("unique_citations", "value")}
+          "count" => hsh.dig("total", "value")}
       end
     end
 

app/controllers/events_controller.rb

@@ -126,8 +126,10 @@ def index
     dois = total.positive? && aggregations.blank? || aggregations.include?("query_aggregations") ? facet_by_dois(response.response.aggregations.dois.buckets) : nil
     dois_usage = total.positive? && aggregations.blank? || aggregations.include?("query_aggregations") ? facet_by_dois(response.response.aggregations.dois_usage.dois.buckets) : nil
     dois_citations = total.positive? && aggregations.blank? || aggregations.include?("query_aggregations") ? facet_citations_by_year_v1(response.response.aggregations.dois_citations) : nil
-    citations_histogram = total.positive? && aggregations.include?("metrics_aggregations") ? facet_citations_by_year(response.response.aggregations.citations_histogram) : nil
-    citations = total.positive? && aggregations.include?("metrics_aggregations") ? facet_citations_by_dois(response.response.aggregations.citations.dois.buckets) : nil
+    citations_histogram = total.positive? && params[:doi].present? && aggregations.include?("citations_aggregations") ? facet_citations_by_year(response.response.aggregations.citations_histogram) : nil
+    citations = total.positive? && params[:doi].present? &&  aggregations.include?("citations_aggregations") ? facet_citations_by_dois(response.response.aggregations.citations.dois.buckets) : nil
+    references = total.positive? && params[:doi].present? &&  aggregations.include?("citations_aggregations") ? facet_citations_by_dois(response.response.aggregations.references.dois.buckets) : nil
+    relations = total.positive? && params[:doi].present? &&  aggregations.include?("citations_aggregations") ? facet_citations_by_dois(response.response.aggregations.relations.dois.buckets) : nil
     views_histogram = total.positive? && aggregations.include?("metrics_aggregations") ? facet_counts_by_year_month(response.response.aggregations.views_histogram) : nil
     downloads_histogram = total.positive? && aggregations.include?("metrics_aggregations") ? facet_counts_by_year_month(response.response.aggregations.downloads_histogram) : nil
     views = total.positive? && aggregations.include?("metrics_aggregations") ? facet_by_source(response.response.aggregations.views.dois.buckets) : nil
@@ -153,6 +155,8 @@ def index
       "doisCitations": dois_citations,
       "citationsHistogram": citations_histogram,
       "uniqueCitations": citations,
+      "references": references,
+      "relations": relations,
       "uniqueNodes": {
         "objCount": unique_obj_count,
         "subjCount": unique_subj_count

app/graphql/types/metric_interface.rb

@@ -6,7 +6,10 @@ module MetricInterface
   field :view_count, Integer, null: true, description: "The count of DOI views according to the COUNTER code of Practice"
   field :download_count, Integer, null: true, description: "The count of  DOI dowloands according to the COUNTER code of Practice"
   field :citation_count, Integer, null: true, description: "The count of DOI events that represents citations"
-
+  field :citations_list, [String], null: true, description: "List of DOIS citing a given DOI"
+  field :referenceslist, [String], null: true, description: "List of DOIS that a given DOI references to"
+  field :relations_list, [String], null: true, description: "List of DOIS relations a given DOI has"
+
   def aggregation_results(**args)
     Event.query(nil, doi: doi_from_url(args[:id]), "page[size]": 0, aggregations: "metrics_aggregations", source_id: args[:source_id] || nil).response.aggregations
   end
@@ -22,7 +25,45 @@ def download_count
   end
 
   def citation_count
-    meta = aggregation_results(id: object.identifier).citations.dois.buckets
+    meta = citations_aggs
     meta.first.fetch("unique_citations", {}).fetch("value", nil) if meta.any?
   end
+
+  def references_count
+    meta = references_aggs
+    meta.first.fetch("references", {}).fetch("value", nil) if meta.any?
+  end
+
+  def relations_count
+    meta = relations_aggs
+    meta.first.fetch("relations", {}).fetch("value", nil) if meta.any?
+  end
+
+  def references_list
+    references_aggs.map { |item| item[:key]}
+  end
+
+  def relations_list
+    relations_aggs.map { |item| item[:key]}
+  end
+
+  def citations_list
+    citations_aggs.map { |item| item[:key]}
+    # citations_aggs.map do |item| 
+    #   puts item[:key]
+    #   puts Doi.find_by_id(item[:key]).results.first
+    # end
+  end
+
+  def citations_aggs
+    aggregation_results(id: object.identifier).citations.dois.buckets 
+  end
+
+  def references_aggs
+    aggregation_results(id: object.identifier).references.dois.buckets
+  end
+
+  def relations_aggs
+    aggregation_results(id: object.identifier).relations.dois.buckets
+  end
 end

app/models/concerns/indexable.rb

@@ -95,18 +95,20 @@ def find_by_id_list(ids, options={})
       })
     end
 
-    def get_aggregations_hash(aggregations = "")
+    def get_aggregations_hash(options={})
+      aggregations = options[:aggregations] || ""
       return send(:query_aggregations) if aggregations.blank?
       aggs = {}
       aggregations.split(",").each do |agg|
         agg = :query_aggregations if agg.blank? || !respond_to?(agg)
-        aggs.merge! send(agg)
+        doi = options[:doi].present? ? options[:doi].downcase.split(",").first : nil
+        aggs.merge! send(agg,doi)
       end
       aggs
     end
 
     def query(query, options={})
-      aggregations = options[:totals_agg] == true ? totals_aggregations : get_aggregations_hash(options[:aggregations])
+      aggregations = options[:totals_agg] == true ? totals_aggregations : get_aggregations_hash(options)
       options[:page] ||= {}
       options[:page][:number] ||= 1
       options[:page][:size] ||= 25

app/models/event.rb

@@ -62,14 +62,37 @@ class Event < ActiveRecord::Base
 
   INCLUDED_RELATION_TYPES = [
     "cites", "is-cited-by",
+    "is-supplement-to", "is-supplemented-by",
+    "references", "is-referenced-by"
+  ]
+
+  ACTIVE_RELATION_TYPES = [
+    "cites", 
+    "is-supplement-to",
+    "references"
+  ]
+
+  PASSIVE_RELATION_TYPES = [
+     "is-cited-by",
+     "is-supplemented-by",
+     "is-referenced-by"
+  ]
+
+
+  RELATIONS_RELATION_TYPES = [
     "compiles", "is-compiled-by",
     "documents", "is-documented-by",
     "has-metadata", "is-metadata-for",
-    "is-supplement-to", "is-supplemented-by",
     "is-derived-from", "is-source-of",
-    "references", "is-referenced-by",
     "reviews", "is-reviewed-by",
     "requires", "is-required-by",
+    "continues", "is-coutinued-by",
+    "has-version", "is-version-of",
+    "has-part", "is-part-of",
+    "is-variant-from-of", "is-original-form-of",
+    "is-identical-to", "obsoletes",
+    "is-obsolete-by",
+    "is-new-version-of", "is-previous-version-of",
     "describes", "is-described-by"
   ]
 
@@ -183,7 +206,7 @@ def self.query_fields
     ['subj_id^10', 'obj_id^10', 'subj.name^5', 'subj.author^5', 'subj.periodical^5', 'subj.publisher^5', 'obj.name^5', 'obj.author^5', 'obj.periodical^5', 'obj.publisher^5', '_all']
   end
 
-  def self.query_aggregations
+  def self.query_aggregations(doi=nil) 
     sum_distribution = {
       sum_bucket: {
         buckets_path: "year_months>total_by_year_month"
@@ -220,35 +243,18 @@ def self.query_aggregations
     }
   end
 
-  def self.metrics_aggregations
+  def self.metrics_aggregations(doi=nil)  
     sum_distribution = {
       sum_bucket: {
         buckets_path: "year_months>total_by_year_month"
       }
     }
-    sum_year_distribution = {
-      sum_bucket: {
-        buckets_path: "years>total_by_year"
-      }
-    }
-
+
     views_filter = {script: {script: "#{VIEWS_RELATION_TYPES}.contains(doc['relation_type_id'].value) && doc['source_id'].value == 'datacite-usage' && doc['occurred_at'].value.getMillis() >= doc['obj.datePublished'].value.getMillis() && doc['occurred_at'].value.getMillis() < new Date().getTime()"}}
 
     downloads_filter = {script: {script: "#{DOWNLOADS_RELATION_TYPES}.contains(doc['relation_type_id'].value) && doc['source_id'].value == 'datacite-usage' && doc['occurred_at'].value.getMillis() >= doc['obj.datePublished'].value.getMillis() && doc['occurred_at'].value.getMillis() < new Date().getTime()"} }
 
-    {
-      citations_histogram: {
-        filter: {script: {script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"}
-        },
-        aggs: { years: { histogram: { field: 'citation_year', interval: 1 , min_doc_count: 1 }, aggs: { "total_by_year" => { sum: { field: 'total' }}}},"sum_distribution"=>sum_year_distribution}
-      },
-      citations: {
-        filter: {script: {script: "#{INCLUDED_RELATION_TYPES}.contains(doc['relation_type_id'].value)"}
-        },
-        aggs: { dois: {
-           terms: { field: 'doi', size: 100, min_doc_count: 1 }, aggs: { unique_citations: { cardinality: { field: 'citation_id' }}}
-        }}
-      },
+   {
       views: {
         filter: views_filter,
         aggs: { dois: {
@@ -273,7 +279,48 @@ def self.metrics_aggregations
           year_months: { date_histogram: { field: 'occurred_at', interval: 'month', min_doc_count: 1 }, aggs: { "total_by_year_month" => { sum: { field: 'total' } } } }, "sum_distribution" => sum_distribution
           }
       }   
-  }
+    }
+
+  end
+
+  def self.citations_aggregations(doi)  
+    doi = Event.new.normalize_doi(doi) if doi.present?
+
+    sum_year_distribution = {
+      sum_bucket: {
+        buckets_path: "years>total_by_year"
+      }
+    }
+
+    citations_filter = {script: {script: "(#{PASSIVE_RELATION_TYPES}.contains(doc['relation_type_id'].value) && '#{doi}' == doc['subj_id'].value) || (#{ACTIVE_RELATION_TYPES}.contains(doc['relation_type_id'].value) && '#{doi}' == doc['obj_id'].value)"}}
+
+    references_filter = {script: {script: "(#{PASSIVE_RELATION_TYPES}.contains(doc['relation_type_id'].value) && '#{doi}' == doc['obj_id'].value) || (#{ACTIVE_RELATION_TYPES}.contains(doc['relation_type_id'].value) && '#{doi}' == doc['subj_id'].value)"}}
+
+    {
+      citations_histogram: {
+        filter: citations_filter,
+        aggs: { years: { histogram: { field: 'citation_year', interval: 1 , min_doc_count: 1 }, aggs: { "total_by_year" => { sum: { field: 'total' }}}},"sum_distribution"=>sum_year_distribution}
+      },
+      citations: {
+        filter: citations_filter,
+        aggs: { dois: {
+          terms: { field: 'doi', size: 100, min_doc_count: 1 }, aggs: { total: { cardinality: { field: 'citation_id' }}}
+        }}
+      },
+      references: {
+        filter: references_filter,
+        aggs: { dois: {
+          terms: { field: 'doi', size: 100, min_doc_count: 1 }, aggs: { total: { cardinality: { field: 'citation_id' }}}
+        }}
+      },
+      relations: {
+        filter: {script: {script: "#{RELATIONS_RELATION_TYPES}.contains(doc['relation_type_id'].value)"}
+        },
+        aggs: { dois: {
+          terms: { field: 'doi', size: 100, min_doc_count: 1 }, aggs: { total: { cardinality: { field: 'citation_id' }}}
+        }}
+      }
+    } 
   end
 
   def self.advanced_aggregations
@@ -608,7 +655,7 @@ def obj_cache_key
   end
 
   def citation_year
-    "" unless INCLUDED_RELATION_TYPES.include?(relation_type_id)
+    "" unless (INCLUDED_RELATION_TYPES+RELATIONS_RELATION_TYPES).include?(relation_type_id)
     subj_publication = subj['date_published'] || (date_published(subj_id) || year_month)
     obj_publication =  obj['date_published']  || (date_published(obj_id) || year_month)
     [subj_publication[0..3].to_i, obj_publication[0..3].to_i].max

spec/concerns/indexable_spec.rb

@@ -126,15 +126,15 @@
 
     context "aggregations" do
       it 'returns query_aggregation when filters aggregation with empty' do
-        aggregations = Doi.get_aggregations_hash("")
+        aggregations = Doi.get_aggregations_hash({aggregations:""})
         expect(aggregations[:resource_types]).not_to be_nil
         expect(aggregations[:states]).not_to be_nil
         expect(aggregations[:created]).not_to be_nil
         expect(aggregations[:schema_versions]).not_to be_nil
       end
 
       it 'returns multiple aggregations when filters aggregations with multiple' do
-        aggregations = Doi.get_aggregations_hash("query_aggregations,metrics_aggregations")
+        aggregations = Doi.get_aggregations_hash({aggregations:""})
         expect(aggregations[:resource_types]).not_to be_nil
         expect(aggregations[:states]).not_to be_nil
         expect(aggregations[:created]).not_to be_nil
@@ -154,7 +154,7 @@
 
     context "aggregations" do
       it 'returns query_aggregation when filters aggregation with empty' do
-        aggregations = Event.get_aggregations_hash("")
+        aggregations = Event.get_aggregations_hash({aggregations:""})
         expect(aggregations[:sources]).not_to be_nil
         expect(aggregations[:prefixes]).not_to be_nil
         expect(aggregations[:citation_types]).not_to be_nil
@@ -167,7 +167,7 @@
       end
 
       it 'returns multiple aggregations when filters aggregations with multiple' do
-        aggregations = Event.get_aggregations_hash("query_aggregations,metrics_aggregations")
+        aggregations = Event.get_aggregations_hash({aggregations:"query_aggregations,metrics_aggregations"})
         expect(aggregations[:sources]).not_to be_nil
         expect(aggregations[:prefixes]).not_to be_nil
         expect(aggregations[:citation_types]).not_to be_nil
@@ -176,8 +176,8 @@
         expect(aggregations[:pairings]).not_to be_nil
         expect(aggregations[:dois]).not_to be_nil
         expect(aggregations[:dois_usage]).not_to be_nil
-        expect(aggregations[:citations_histogram]).not_to be_nil
-        expect(aggregations[:citations]).not_to be_nil
+        expect(aggregations[:citations_histogram]).to be_nil
+        expect(aggregations[:citations]).to be_nil
       end
     end
   end