Implement uniqueness (hash equality)

lib/twingly/url.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require "digest"
+
 require "addressable/idna/pure"
 require "addressable/uri"
 require "public_suffix"
@@ -23,6 +25,11 @@ class URL
       PublicSuffix::DomainInvalid,
     ].freeze
 
+    # Instantiate digest classes in a thread-safe manner
+    # This is important since we don't know how people will
+    # use this gem (if they require it in a thread safe way)
+    SHA256_DIGEST = Digest(:SHA256)
+
     private_constant :ACCEPTED_SCHEMES
     private_constant :CUSTOM_PSL
     private_constant :STARTS_WITH_WWW
@@ -193,6 +200,14 @@ def <=>(other)
       self.to_s <=> other.to_s
     end
 
+    def eql?(other)
+      self.hash == other.hash
+    end
+
+    def hash
+      SHA256_DIGEST.digest(self.to_s).unpack1("q")
+    end
+
     def to_s
       addressable_uri.to_s
     end

lib/twingly/url/null_url.rb

@@ -24,6 +24,14 @@ def <=>(other)
         self.to_s <=> other.to_s
       end
 
+      def eql?(other)
+        self.hash == other.hash
+      end
+
+      def hash
+        "".hash
+      end
+
       def to_s
         ""
       end

spec/lib/twingly/url/null_url_spec.rb

@@ -97,4 +97,12 @@
       expect { url.method_does_not_exist }.to raise_error(NoMethodError)
     end
   end
+
+  describe "uniqueness" do
+    context "a list with multiple NullURLs should only have one unique item" do
+      subject(:list) { 10.times.map { described_class.new }.uniq }
+
+      it { is_expected.to eq([described_class.new]) }
+    end
+  end
 end

spec/lib/twingly/url_spec.rb

@@ -847,6 +847,36 @@ def valid_urls
     end
   end
 
+  describe "uniqueness" do
+    context do "with the same URL twice"
+      let(:a) { described_class.parse("https://www.twingly.com/") }
+      let(:b) { described_class.parse("https://www.google.com/") }
+      let(:c) { described_class.parse("https://www.twingly.com/") }
+
+      it "should give only unique URLs" do
+        expect([a,b,c].uniq).to eq([a,b])
+      end
+    end
+
+    context do "two similar URLs, but not exactly the same"
+      let(:a) { described_class.parse("https://www.twingly.com") }
+      let(:b) { described_class.parse("https://www.twingly.com/") }
+
+      it "should be two unique URLs" do
+        expect([a,b].uniq).to eq([a,b])
+      end
+    end
+
+    context do "the same URL but with some whitespace should be the same"
+      let(:a) { described_class.parse(" https://www.twingly.com/") }
+      let(:b) { described_class.parse("https://www.twingly.com/ ") }
+
+      it "should be one unique URL" do
+        expect([a,b].uniq).to eq([a])
+      end
+    end
+  end
+
   describe "#inspect" do
     let(:url_object) { described_class.parse(url) }
     subject { url_object.inspect }