Merge pull request #11893 from rioug/10809-patch-dfc-connector-parser

[DFC] Patch dfc connector parser to improve SKOS concept parsing

engines/dfc_provider/lib/data_food_consortium/connector/connector.rb

@@ -6,6 +6,9 @@
 # Then our tools for monky-patching:
 require_relative "importer"
 require_relative "context"
+require_relative "skos_parser_element"
+require_relative "skos_concept"
+require_relative "skos_parser"
 
 module DataFoodConsortium
   module Connector

engines/dfc_provider/lib/data_food_consortium/connector/skos_concept.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+# Patch: Improve parsing of SKOS Concept. Will be fixed upstream
+require_relative 'skos_helper'
+
+module DataFoodConsortium
+  module Connector
+    class SKOSConcept
+      include DataFoodConsortium::Connector::SKOSHelper
+    end
+  end
+end

engines/dfc_provider/lib/data_food_consortium/connector/skos_helper.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+# Patch: Improve parsing of SKOS Concept. Will be fixed upstream
+module DataFoodConsortium
+  module Connector
+    module SKOSHelper
+      def addAttribute(name, value) # rubocop:disable Naming/MethodName
+        instance_variable_set("@#{name}", value)
+        define_singleton_method(name) do
+          instance_variable_get("@#{name}")
+        end
+      end
+
+      def hasAttribute(name) # rubocop:disable Naming/MethodName
+        methods.include?(:"#{name}")
+      end
+    end
+  end
+end

engines/dfc_provider/lib/data_food_consortium/connector/skos_parser.rb

@@ -1,23 +1,144 @@
 # frozen_string_literal: true
 
-# Overriding the current implementation to store all parsed concepts for
-# lookup later. Otherwise the importer can't associate these.
-# This is just a workaround and needs to be solved upstream.
+# patches :
+# - Maikel: Overriding the current implementation to store all parsed concepts for
+#           lookup later. Otherwise the importer can't associate these.
+#           This is just a workaround and needs to be solved upstream.
+
+# - Gaetan: Improve parsing of SKOS Concept. Will be fixed upstream
+
+require_relative 'skos_helper'
+
+module DataFoodConsortium
+  module Connector
+    class SKOSInstance
+      include DataFoodConsortium::Connector::SKOSHelper
+
+      # Return a list of singelton methods, ie the list of Concept available
+      def topConcepts # rubocop:disable Naming/MethodName
+        methods(false).sort
+      end
+    end
+  end
+end
+
+# rubocop:disable Naming/VariableName
 module DataFoodConsortium
   module Connector
     class SKOSParser
+      CONCEPT_SCHEMES = ["Facet", "productTypes"].freeze
+
+      def initialize
+        @results = DataFoodConsortium::Connector::SKOSInstance.new
+        @skosConcepts = {}
+        @rootElements = []
+        @broaders = {}
+        # Flag used to tell the parser to use SkosConcept object when parsing data from
+        # Concept Scheme.
+        # defined in CONCEPT_SCHEMES
+        @useSkosConcept = false
+      end
+
+      def parse(data) # rubocop:disable Metrics/CyclomaticComplexity
+        init
+
+        data.each do |element|
+          current = DataFoodConsortium::Connector::SKOSParserElement.new(element)
+
+          setSkosConceptFlag(current)
+
+          next unless current.isConcept? || current.isCollection?
+
+          if !@skosConcepts.key?(current.id)
+            concept = createSKOSConcept(current)
+            @skosConcepts[current.id] = concept
+          end
+
+          if current.hasBroader
+            current.broader.each do |broader_id|
+              if !@broaders.key?(broader_id)
+                @broaders[broader_id] = []
+              end
+
+              @broaders[broader_id].push(current.id)
+            end
+          # No broader, save the concept to the root
+          else
+            @rootElements.push(current.id)
+          end
+        end
+
+        @rootElements.each do |root_element_id|
+          setResults(@results, root_element_id)
+        end
+
+        @results
+      end
+
+      # Maikel's patch
       def self.concepts
         @concepts ||= {}
       end
 
       protected
 
       def createSKOSConcept(element) # rubocop:disable Naming/MethodName
-        concept = DataFoodConsortium::Connector::SKOSConcept.new(element.id)
-        concept.semanticType = element.type
-        self.class.concepts[element.id] = concept
-        concept
+        skosConcept = DataFoodConsortium::Connector::SKOSConcept.new(
+          element.id,
+          broaders: element.broader,
+          narrowers: element.narrower,
+          prefLabels: element.label
+        )
+        skosConcept.semanticType = element.type
+        # Maikel's patch
+        self.class.concepts[element.id] = skosConcept
+        skosConcept
+      end
+
+      private
+
+      def init
+        @results = DataFoodConsortium::Connector::SKOSInstance.new
+        @skosConcepts = {}
+        @rootElements = []
+        @broaders = {}
+        @useSkosConcept = false
+      end
+
+      def setResults(parent, id) # rubocop:disable Naming/MethodName
+        name = getValueWithoutPrefix(id)
+
+        if !parent.hasAttribute(name)
+          if @useSkosConcept && @skosConcepts[id]
+            parent.addAttribute(name, @skosConcepts[id])
+          else
+            parent.addAttribute(name, DataFoodConsortium::Connector::SKOSInstance.new)
+          end
+        end
+
+        # Leaf concepts, stop the process
+        if !@broaders.key?(id)
+          parent.instance_variable_set("@#{name}", @skosConcepts[id])
+          return
+        end
+
+        @broaders[id].each do |narrower|
+          parentSkosInstance = parent.instance_variable_get("@#{name}")
+
+          setResults(parentSkosInstance, narrower) # recursive call
+        end
+      end
+
+      def setSkosConceptFlag(current) # rubocop:disable Naming/MethodName
+        @useSkosConcept = true if current.isConceptScheme? && matchingConceptSchemes(current)
+      end
+
+      def matchingConceptSchemes(current) # rubocop:disable Naming/MethodName
+        regex = /#{CONCEPT_SCHEMES.join('|')}/
+
+        current.id =~ regex
       end
     end
   end
 end
+# rubocop:enable Naming/VariableName

engines/dfc_provider/lib/data_food_consortium/connector/skos_parser_element.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+# Patch: Improve parsing of SKOS Concept. Will be fixed upstream
+module DataFoodConsortium
+  module Connector
+    class SKOSParserElement
+      attr_reader :narrower, :label
+
+      def initialize(element) # rubocop:disable Metrics/CyclomaticComplexity
+        @broader = []
+        @narrower = []
+        @label = {}
+
+        if element
+          @id = element["@id"]
+
+          @type = if element["http://www.w3.org/1999/02/22-rdf-syntax-ns#type"]
+                    extractId(element["http://www.w3.org/1999/02/22-rdf-syntax-ns#type"])
+                  elsif element["@type"]
+                    extractId(element["@type"])
+                  else
+                    "undefined"
+                  end
+
+          element["http://www.w3.org/2004/02/skos/core#broader"]&.each do |broader|
+            @broader.push(broader["@id"])
+          end
+
+          element["http://www.w3.org/2004/02/skos/core#narrower"]&.each do |narrower|
+            @narrower.push(narrower["@id"])
+          end
+
+          element["http://www.w3.org/2004/02/skos/core#prefLabel"]&.each do |label|
+            @label[label["@language"].to_sym] = label["@value"]
+          end
+        else
+          @id = ""
+          @type = ""
+        end
+      end
+
+      def isConceptScheme? # rubocop:disable Naming/MethodName
+        @type == "http://www.w3.org/2004/02/skos/core#ConceptScheme"
+      end
+    end
+  end
+end