wayfarer 0.4.7 → 0.4.9

data/spec/wayfarer/uri/normalization_spec.rb

@@ -0,0 +1,98 @@
+# frozen_string_literal: true
+
+require "spec_helpers"
+
+RSpec.describe Wayfarer::URI::Normalization do
+  using RSpec::Parameterized::TableSyntax
+
+  let(:default_tracking_params) { Wayfarer.config.dig(:normalization, :tracking_params) }
+  let(:default_schemes) { Wayfarer.config.dig(:normalization, :schemes) }
+
+  # rubocop:disable Layout/LineLength, Layout/ExtraSpacing
+  context "when the URI is valid and absolute" do
+    where(
+      :input,                                                                               :remove_www, :remove_trailing_slash, :remove_fragment, :remove_tracking_parameters, :order_query_parameters, :output
+    ) do
+      "https://www.example.com"                                                            | true       | true                    | true            | true                       | true                     | "https://example.com"
+      "https://example.com/path/"                                                          | false      | true                    | false           | false                      | false                    | "https://example.com/path"
+      "https://example.com/page#section"                                                   | false      | false                   | true            | false                      | false                    | "https://example.com/page"
+      "https://example.com?fbclid=123"                                                     | false      | false                   | false           | true                       | false                    | "https://example.com"
+      "https://www.example.com/?b=2&a=1"                                                   | true       | true                    | false           | false                      | true                     | "https://example.com?a=1&b=2"
+      "https://example.com/p/#f"                                                           | false      | true                    | true            | false                      | false                    | "https://example.com/p"
+      "https://www.example.com/path/?z=9&fbclid=1#frag"                                    | false      | false                   | false           | false                      | false                    | "https://www.example.com/path/?z=9&fbclid=1#frag"
+      "https://www.example.com/p/?z=9&a=1&utm_source=c#f"                                  | true       | true                    | true            | true                       | true                     | "https://example.com/p?a=1&z=9"
+      "HTTP://WWW.Example.COM/A/Path?b=2&a=1"                                              | true       | true                    | true            | true                       | true                     | "http://example.com/A/Path?a=1&b=2"
+      "https://blog.example.com/"                                                          | true       | true                    | true            | true                       | true                     | "https://blog.example.com"
+      "https://www.youtube.com/watch?v=dQw4w9WgXcQ&utm_campaign=test&feature=share"        | true       | true                    | true            | true                       | true                     | "https://youtube.com/watch?feature=share&v=dQw4w9WgXcQ"
+      "https://example.com/search?q=ruby&is_awesome=&sort=asc"                             | true       | true                    | true            | true                       | true                     | "https://example.com/search?q=ruby&sort=asc"
+      "https://example.com/a%20path/?q=a%26b"                                              | true       | true                    | true            | true                       | true                     | "https://example.com/a%20path?q=a%26b"
+      "https://example.com/path?a=1"                                                       | true       | true                    | true            | true                       | true                     | "https://example.com/path?a=1"
+      "https://example.com/"                                                               | true       | true                    | true            | true                       | true                     | "https://example.com"
+      "http://example.com:8080"                                                            | true       | true                    | true            | true                       | true                     | "http://example.com:8080"
+      "https://user:password@example.com"                                                  | true       | true                    | true            | true                       | true                     | "https://user:password@example.com"
+      "https://example.com/Path/To/Resource"                                               | true       | true                    | true            | true                       | true                     | "https://example.com/Path/To/Resource"
+      "https://example.com"                                                                | true       | true                    | true            | true                       | true                     | "https://example.com"
+      "https://example.com/../../etc/passwd"                                               | true       | true                    | true            | true                       | true                     | "https://example.com/etc/passwd"
+      "https://www.microsоft.com"                                                          | true       | true                    | true            | true                       | true                     | "https://xn--microsft-sbh.com"
+      "https://example.com/a%252fb"                                                        | true       | true                    | true            | true                       | true                     | "https://example.com/a%252fb"
+      "https://example.com/path%00/something"                                              | true       | true                    | true            | true                       | true                     | "https://example.com/path%00/something"
+      "https://www.example.com/path/?z=9&a=1#frag"                                         | false      | false                   | false           | false                      | false                    | "https://www.example.com/path/?z=9&a=1#frag"
+      "https://www.example.com/path/?z=9&a=1#frag"                                         | true       | false                   | false           | false                      | false                    | "https://example.com/path/?z=9&a=1#frag"
+      "https://www.example.com/path/?z=9&a=1#frag"                                         | false      | true                    | false           | false                      | false                    | "https://www.example.com/path?z=9&a=1#frag"
+      "https://www.example.com/path/?z=9&a=1#frag"                                         | false      | false                   | true            | false                      | false                    | "https://www.example.com/path/?z=9&a=1"
+      "https://www.example.com/path/?z=9&fbclid=123#frag"                                  | false      | false                   | false           | true                       | false                    | "https://www.example.com/path/?z=9#frag"
+      "https://www.example.com/path/?z=9&a=1#frag"                                         | false      | false                   | false           | false                      | true                     | "https://www.example.com/path/?a=1&z=9#frag"
+      "https://www.example.com/path/?z=9&fbclid=1&a=1#frag"                                | true       | true                    | true            | true                       | true                     | "https://example.com/path?a=1&z=9"
+      "https://www.example.com/"                                                           | false      | false                   | false           | false                      | false                    | "https://www.example.com/"
+      "https://example.com?b=2&a=1"                                                        | true       | true                    | true            | false                      | true                     | "https://example.com?a=1&b=2"
+      "https://example.com?utm_source=test&a=1"                                            | false      | true                    | true            | false                      | true                     | "https://example.com?a=1&utm_source=test"
+      "https://example.com?utm_source=test&a=1"                                            | false      | true                    | true            | true                       | false                    | "https://example.com?a=1"
+      "https://www.example.com/trailing/"                                                  | false      | true                    | true            | true                       | true                     | "https://www.example.com/trailing"
+      "https://www.example.com/trailing/"                                                  | false      | false                   | true            | true                       | true                     | "https://www.example.com/trailing/"
+    end
+    # rubocop:enable Layout/LineLength, Layout/ExtraSpacing
+
+    with_them do
+      subject { described_class.canonical!(Addressable::URI.parse(input)).to_s }
+
+      before do
+        Wayfarer.config.fetch(:normalization).merge!(
+          remove_www: remove_www,
+          remove_trailing_slash: remove_trailing_slash,
+          remove_fragment: remove_fragment,
+          remove_tracking_parameters: remove_tracking_parameters,
+          order_query_parameters: order_query_parameters,
+          tracking_params: default_tracking_params,
+          schemes: default_schemes
+        )
+      end
+
+      it { is_expected.to eq(output) }
+    end
+  end
+
+  context "with invalid URLs" do
+    before do
+      # Consistent configuration setting
+      Wayfarer.config.fetch(:normalization).merge!(
+        remove_www: true,
+        remove_trailing_slash: true,
+        remove_fragment: true,
+        remove_tracking_parameters: true,
+        order_query_parameters: true,
+        tracking_params: default_tracking_params,
+        schemes: default_schemes
+      )
+    end
+
+    specify do
+      expect { described_class.canonical!(Addressable::URI.parse("")) }
+        .to raise_error(Wayfarer::URI::Normalization::RelativeURIError)
+    end
+
+    specify do
+      expect { described_class.canonical!(Addressable::URI.parse("/path/to/resource")) }
+        .to raise_error(Wayfarer::URI::Normalization::RelativeURIError)
+    end
+  end
+end

data/spec/wayfarer_spec.rb

@@ -4,10 +4,10 @@ require "spec_helpers"
 
 describe Wayfarer do
   describe "::VERSION" do
-    it("is present") { expect(defined? Wayfarer::VERSION).not_to be nil }
+    it("is present") { expect(defined? Wayfarer::VERSION).not_to be_nil }
   end
 
   describe "::config" do
-    specify { expect(Wayfarer.config).to be_a(Hash) }
+    specify { expect(described_class.config).to be_a(Hash) }
   end
 end

data/wayfarer.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name = "wayfarer"
-  s.version = "0.4.7"
+  s.version = "0.4.9"
   s.license = "MIT"
 
   s.homepage = "http://github.com/bauerd/wayfarer"
@@ -14,39 +14,28 @@ Gem::Specification.new do |s|
   s.email = "mail@bauerd.com"
 
   s.files = `git ls-files`.split("\n")
-  s.test_files = `git ls-files -- {spec}/*`.split("\n")
+  s.test_files = `git ls-files -- spec`.split("\n")
 
   s.require_paths = ["lib"]
 
   s.executables << "wayfarer"
 
-  s.required_ruby_version = "~> 2.7"
+  s.required_ruby_version = "~> 3.4"
 
   s.add_runtime_dependency "activejob", ">= 7.1"
   s.add_runtime_dependency "addressable", "~> 2.8"
-  s.add_runtime_dependency "capybara", "~> 3.0"
-  s.add_runtime_dependency "connection_pool", "~> 2.2"
-  s.add_runtime_dependency "docile", "~> 1.1"
-  s.add_runtime_dependency "ferrum", "~> 0.9"
-  s.add_runtime_dependency "metainspector", "~> 5.0"
-  s.add_runtime_dependency "mime-types", "~> 3.0"
-  s.add_runtime_dependency "mock_redis", "~> 0.29"
-  s.add_runtime_dependency "mustermann", "~> 1.1"
-  s.add_runtime_dependency "net-http-persistent", "~> 3.0"
-  s.add_runtime_dependency "nokogiri", "~> 1.11"
-  s.add_runtime_dependency "normalize_url", "~> 0.0.6"
-  s.add_runtime_dependency "redis", "~> 4.4", "< 4.5" # redis 4.5.x broke SMISMEMBER
-  s.add_runtime_dependency "selenium-webdriver", "~> 3.4"
-  s.add_runtime_dependency "thor", "~> 1.0"
-  s.add_runtime_dependency "zeitwerk", "~> 2.4"
-
-  s.add_development_dependency "cuprite", "~> 0.13"
-  s.add_development_dependency "factory_bot", "~> 6.0"
-  s.add_development_dependency "faker", "~> 1.7"
-  s.add_development_dependency "pry", "~> 0.10"
-  s.add_development_dependency "rake", "~> 10.4"
-  s.add_development_dependency "rspec", "~> 3.4"
-  s.add_development_dependency "rubocop", "~> 0.49"
-  s.add_development_dependency "sinatra", "~> 1.4"
-  s.add_development_dependency "yard", "~> 0.9"
+  s.add_runtime_dependency "capybara", "~> 3.4"
+  s.add_runtime_dependency "connection_pool", "~> 2.5"
+  s.add_runtime_dependency "ferrum", "~> 0.17"
+  s.add_runtime_dependency "metainspector", "~> 5.16"
+  s.add_runtime_dependency "mime-types", "~> 3.7"
+  s.add_runtime_dependency "mock_redis", "~> 0.52"
+  s.add_runtime_dependency "mustermann", "~> 3.0"
+  s.add_runtime_dependency "net-http-persistent", "~> 4.0" # updated
+  s.add_runtime_dependency "nokogiri", "~> 1.18"
+  s.add_runtime_dependency "redis", "~> 5.4"
+  s.add_runtime_dependency "selenium-webdriver", "~> 4.35" # updated
+  s.add_runtime_dependency "thor", "~> 1.0" # problematic
+  s.add_runtime_dependency "zeitwerk", "~> 2.7" # updated
+  s.add_runtime_dependency "sorted_set", "~> 1.0"
 end