Merge pull request #249 from samvera/fix/247_predicates_nil

Raise an exception if data can’t be normalized
samvera · Jun 4, 2019 · 13f1051 · 13f1051
2 parents 7c24d4b + 55dbc05
commit 13f1051
Show file tree

Hide file tree

Showing 7 changed files with 222 additions and 79 deletions.
diff --git a/app/controllers/qa/linked_data_terms_controller.rb b/app/controllers/qa/linked_data_terms_controller.rb
@@ -80,6 +80,10 @@ def show # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
                 "was not identified as a valid RDF format.  You may need to include the linkeddata gem."
     logger.warn msg
     render json: { errors: msg }, status: :internal_server_error
+  rescue Qa::DataNormalizationError => e
+    msg = "Data Normalization Error - #{e.message}"
+    logger.warn msg
+    render json: { errors: msg }, status: :internal_server_error
   end
 
   # Return all the information for a given term given a URI

diff --git a/lib/qa.rb b/lib/qa.rb
@@ -64,4 +64,7 @@ class TermNotFound < ArgumentError; end
   module IriTemplate
     class MissingParameter < StandardError; end
   end
+
+  # Raised when data is returned but cannot be normalized
+  class DataNormalizationError < StandardError; end
 end
diff --git a/lib/qa/authorities/linked_data/find_term.rb b/lib/qa/authorities/linked_data/find_term.rb
@@ -133,16 +133,26 @@ def expects_uri?
 
         def extract_uri
           return @uri = RDF::URI.new(id) if expects_uri?
-          @uri = graph_service.subjects_for_object_value(graph: @filtered_graph, predicate: RDF::URI.new(term_config.term_results_id_predicate), object_value: URI.unescape(id)).first
-          return @uri unless loc? && @uri.blank?
-          # for backward compatibility, if an loc id as passed in fails to extract the URI, try to adding a blank to the id
-          @uri = graph_service.subjects_for_object_value(graph: @filtered_graph, predicate: RDF::URI.new(term_config.term_results_id_predicate), object_value: loc_id).first
-          if @uri.present?
-            Qa.deprecation_warning(
-              in_msg: 'Qa::Authorities::LinkedData::FindTerm',
-              msg: 'Special processing of LOC ids is deprecated; id should be an exact match of the id in the graph'
-            )
-          end
+          extract_uri_by_id
+        end
+
+        def extract_uri_by_id
+          @uri = get_uri_from_graph_by_id(id)
+        rescue Qa::DataNormalizationError
+          raise unless loc?
+          @uri = get_uri_from_graph_by_id(loc_id)
+          Qa.deprecation_warning(
+            in_msg: 'Qa::Authorities::LinkedData::FindTerm',
+            msg: 'Special processing of LOC ids is deprecated; id should be an exact match of the id in the graph'
+          )
+          @uri
+        end
+
+        def get_uri_from_graph_by_id(gid)
+          @uri = graph_service.subjects_for_object_value(graph: @filtered_graph,
+                                                         predicate: RDF::URI.new(term_config.term_results_id_predicate),
+                                                         object_value: URI.unescape(gid)).first
+          raise Qa::DataNormalizationError, "Unable to extract URI based on ID: #{id}" if @uri.blank?
           @uri
         end
 
@@ -240,7 +250,8 @@ def predicates_with_subject_uri(expected_uri) # rubocop:disable Metrics/MethodLe
         end
 
         def append_performance_data(results)
-          performance = { predicate_count: results['predicates'].size,
+          pred_count = results['predicates'].present? ? results['predicates'].size : 0
+          performance = { predicate_count: pred_count,
                           fetch_time_s: access_time_s,
                           normalization_time_s: normalize_time_s,
                           total_time_s: (access_time_s + normalize_time_s) }

diff --git a/spec/controllers/linked_data_terms_controller_spec.rb b/spec/controllers/linked_data_terms_controller_spec.rb
@@ -336,6 +336,18 @@
         end
       end
 
+      context 'when data normalization error' do
+        before do
+          stub_request(:get, 'http://id.worldcat.org/fast/530369')
+            .to_return(status: 200, body: webmock_fixture('lod_oclc_term_bad_id.nt'), headers: { 'Content-Type' => 'application/ntriples' })
+        end
+        it 'returns 500' do
+          expect(Rails.logger).to receive(:warn).with("Data Normalization Error - Unable to extract URI based on ID: 530369")
+          get :show, params: { id: '530369', vocab: 'OCLC_FAST' }
+          expect(response.code).to eq('500')
+        end
+      end
+
       context 'when rdf format error' do
         before do
           stub_request(:get, 'http://id.worldcat.org/fast/530369').to_return(status: 200)

diff --git a/spec/fixtures/lod_loc_term_bad_id.rdf.xml b/spec/fixtures/lod_loc_term_bad_id.rdf.xml
@@ -0,0 +1,68 @@
+<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
+  <madsrdf:Topic rdf:about="http://id.loc.gov/authorities/subjects/sh1234" xmlns:madsrdf="http://www.loc.gov/mads/rdf/v1#">
+    <rdf:type rdf:resource="http://www.loc.gov/mads/rdf/v1#Authority"/>
+    <madsrdf:authoritativeLabel xml:lang="en">More Science</madsrdf:authoritativeLabel>
+    <madsrdf:elementList rdf:parseType="Collection">
+      <madsrdf:TopicElement>
+	<madsrdf:elementValue xml:lang="en">More Science</madsrdf:elementValue>
+      </madsrdf:TopicElement>
+    </madsrdf:elementList>
+    <madsrdf:hasVariant>
+      <madsrdf:Topic>
+	<rdf:type rdf:resource="http://www.loc.gov/mads/rdf/v1#Variant"/>
+	<madsrdf:variantLabel xml:lang="en">More Natural science</madsrdf:variantLabel>
+	<madsrdf:elementList rdf:parseType="Collection">
+	  <madsrdf:TopicElement>
+	    <madsrdf:elementValue xml:lang="en">More Natural science</madsrdf:elementValue>
+	  </madsrdf:TopicElement>
+	</madsrdf:elementList>
+      </madsrdf:Topic>
+    </madsrdf:hasVariant>
+    <madsrdf:hasVariant>
+      <madsrdf:Topic>
+	<rdf:type rdf:resource="http://www.loc.gov/mads/rdf/v1#Variant"/>
+	<madsrdf:variantLabel xml:lang="en">More Science of science</madsrdf:variantLabel>
+	<madsrdf:elementList rdf:parseType="Collection">
+	  <madsrdf:TopicElement>
+	    <madsrdf:elementValue xml:lang="en">More Science of science</madsrdf:elementValue>
+	  </madsrdf:TopicElement>
+	</madsrdf:elementList>
+      </madsrdf:Topic>
+    </madsrdf:hasVariant>
+    <madsrdf:hasVariant>
+      <madsrdf:Topic>
+	<rdf:type rdf:resource="http://www.loc.gov/mads/rdf/v1#Variant"/>
+	<madsrdf:variantLabel xml:lang="en">More Sciences</madsrdf:variantLabel>
+	<madsrdf:elementList rdf:parseType="Collection">
+	  <madsrdf:TopicElement>
+	    <madsrdf:elementValue xml:lang="en">More Sciences</madsrdf:elementValue>
+	  </madsrdf:TopicElement>
+	</madsrdf:elementList>
+      </madsrdf:Topic>
+    </madsrdf:hasVariant>
+    <identifiers:lccn xmlns:identifiers="http://id.loc.gov/vocabulary/identifiers/">BAD ID sh 1234</identifiers:lccn>
+    <rdf:type rdf:resource="http://www.w3.org/2004/02/skos/core#Concept"/>
+    <skos:prefLabel xml:lang="en" xmlns:skos="http://www.w3.org/2004/02/skos/core#">More Science</skos:prefLabel>
+    <skosxl:altLabel xmlns:skosxl="http://www.w3.org/2008/05/skos-xl#">
+      <rdf:Description>
+	<rdf:type rdf:resource="http://www.w3.org/2008/05/skos-xl#Label"/>
+	<skosxl:literalForm xml:lang="en">More Natural science</skosxl:literalForm>
+      </rdf:Description>
+    </skosxl:altLabel>
+    <skosxl:altLabel xmlns:skosxl="http://www.w3.org/2008/05/skos-xl#">
+      <rdf:Description>
+	<rdf:type rdf:resource="http://www.w3.org/2008/05/skos-xl#Label"/>
+	<skosxl:literalForm xml:lang="en">More Science of science</skosxl:literalForm>
+      </rdf:Description>
+    </skosxl:altLabel>
+    <skosxl:altLabel xmlns:skosxl="http://www.w3.org/2008/05/skos-xl#">
+      <rdf:Description>
+	<rdf:type rdf:resource="http://www.w3.org/2008/05/skos-xl#Label"/>
+	<skosxl:literalForm xml:lang="en">More Sciences</skosxl:literalForm>
+      </rdf:Description>
+    </skosxl:altLabel>
+    <skos:altLabel xml:lang="en" xmlns:skos="http://www.w3.org/2004/02/skos/core#">More Natural science</skos:altLabel>
+    <skos:altLabel xml:lang="en" xmlns:skos="http://www.w3.org/2004/02/skos/core#">More Science of science</skos:altLabel>
+    <skos:altLabel xml:lang="en" xmlns:skos="http://www.w3.org/2004/02/skos/core#">More Sciences</skos:altLabel>
+  </madsrdf:Topic>
+</rdf:RDF>
diff --git a/spec/fixtures/lod_oclc_term_bad_id.nt b/spec/fixtures/lod_oclc_term_bad_id.nt
@@ -0,0 +1,4 @@
+<http://id.worldcat.org/fast/530369> <http://purl.org/dc/terms/identifier> "BAD_ID 530369" .
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#prefLabel> "Cornell University" .
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#altLabel> "Ithaca (N.Y.). Cornell University" .
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#sameAs> <http://id.loc.gov/authorities/names/n79021621> .