tripleloop 0.0.5 → 0.0.6

data/lib/tripleloop/document_processor.rb

@@ -1,8 +1,9 @@
 module Tripleloop
   class DocumentProcessor
-    attr_reader :document
+    attr_reader :document, :options
 
-    def initialize(document)
+    def initialize(document, options={})
+      @options = options
       @document = Util.with_nested_fetch(document)
     end
 
@@ -21,9 +22,9 @@ module Tripleloop
       }]
     end
 
-    def self.batch_process(documents)
+    def self.batch_process(documents, options={})
       documents.map { |doc|
-        self.new(doc).extracted_statements
+        self.new(doc, options).extracted_statements
       }.reduce(Hash.new([])) { |accu, statements|
         accu.merge(statements) { |k, olds, news|
           olds.concat(news)
@@ -34,14 +35,25 @@ module Tripleloop
   private
     def extractor_instances
       extractors = self.class.instance_variable_get(:@extractors)
+      filtered_extractors = apply_filters(extractors)
 
-      @extractor_instances ||= extractors.map { |ext, opts|
+      @extractor_instances ||= filtered_extractors.map { |ext, opts|
         klass   = extractor_class(ext)
         context = get_context(opts[:context])
         klass.new(context)
       }
     end
 
+    def apply_filters(extractors)
+      if options[:only]
+        extractors.select { |k,_| Array(options[:only]).include?(k) }
+      elsif options[:except]
+        extractors.reject { |k, _| Array(options[:except]).include?(k) }
+      else
+        extractors
+      end
+    end
+
     def extractor_class(extractor)
       class_name = Tripleloop::Util::String.classify("#{extractor}_extractor")
       scope.const_get(class_name)

data/spec/tripleloop/document_processor_spec.rb

@@ -36,8 +36,10 @@ describe Tripleloop::DocumentProcessor do
     }
   }}
 
+  let(:options) {{}}
+
   describe "#extracted_statements" do
-    subject { Example::SampleProcessor.new(document) }
+    subject { Example::SampleProcessor.new(document, options) }
 
     context "when some of the registered extractors cannot be found" do
       it "raises an ExtractorNotFound error" do
@@ -70,6 +72,28 @@ describe Tripleloop::DocumentProcessor do
           :extractor_2 => :extracted
         })
       end
+
+      context "and the :only option is present" do
+        let(:options) {{ :only => [:foo, :baz] }}
+
+        it "executes only the extractors specified" do
+          subject.extracted_statements.should eq({
+            :foo => [[:subject, "foo-value", :object]],
+            :baz => [[:subject, "baz a", :object],
+                     [:subject, "baz b", :object]]
+          })
+        end
+      end
+
+      context "and the :except option is present" do
+        let(:options) {{ :except=> [:bar, :baz] }}
+
+        it "executes only the extractors specified" do
+          subject.extracted_statements.should eq({
+            :foo => [[:subject, "foo-value", :object]]
+          })
+        end
+      end
     end
   end
 
@@ -86,10 +110,10 @@ describe Tripleloop::DocumentProcessor do
       }
     }
 
-    subject { Example::SampleProcessor.batch_process(documents) }
+    let(:options) {{}} 
 
     it "returns a hash of combined statements, grouped by extractor name" do
-      subject.should eq({
+      Example::SampleProcessor.batch_process(documents).should eq({
         :foo => [
           [:subject, "foo-value 0", :object],
           [:subject, "foo-value 1", :object],
@@ -110,5 +134,15 @@ describe Tripleloop::DocumentProcessor do
         ]
       })
     end
+
+    it "Accepts an options array and forwards it to the constructor" do
+      Example::SampleProcessor.batch_process(documents, {:only => :foo}).should eq({
+        :foo => [
+          [:subject, "foo-value 0", :object],
+          [:subject, "foo-value 1", :object],
+          [:subject, "foo-value 2", :object]
+        ],
+      })
+    end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tripleloop
 version: !ruby/object:Gem::Version
-  version: 0.0.5
+  version: 0.0.6
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-03-05 00:00:00.000000000 Z
+date: 2013-03-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rdf