Merge pull request #238 from benbalter/match-source

Allow ReferenceMatcher to match references to licenses by their source

lib/licensee/license.rb

@@ -91,6 +91,9 @@ module Licensee
       'bsd-3-clause-clear' => /(?:clear bsd|bsd 3-clause(?: clear)?)/i
     }.freeze
 
+    SOURCE_PREFIX = %r{https?://(?:www\.)?}i
+    SOURCE_SUFFIX = %r{(?:\.html?|\.txt|\/)}i
+
     include Licensee::ContentHelper
     extend Forwardable
     def_delegators :meta, *LicenseMeta.helper_methods
@@ -119,7 +122,7 @@ module Licensee
     end
 
     def title_regex
-      return @regex if defined? @regex
+      return @title_regex if defined? @title_regex
 
       title_regex = ALT_TITLE_REGEX[key]
       title_regex ||= begin
@@ -139,7 +142,26 @@ module Licensee
         parts.push Regexp.new meta.nickname.sub(/\bGNU /i, '(?:GNU )?')
       end
 
-      @regex = Regexp.union parts
+      @title_regex = Regexp.union parts
+    end
+
+    # Returns a regex that will match the license source
+    #
+    # The following variations are supported (as presumed identical):
+    # 1. HTTP or HTTPS
+    # 2. www or non-www
+    # 3. .txt, .html, .htm, or / suffix
+    #
+    # Returns the regex, or nil if no source exists
+    def source_regex
+      return @source_regex if defined? @source_regex
+      return unless meta.source
+
+      source = meta.source.dup.sub(/\A#{SOURCE_PREFIX}/, '')
+      source = source.sub(/#{SOURCE_SUFFIX}\z/, '')
+
+      escaped_source = Regexp.escape(source)
+      @source_regex = /#{SOURCE_PREFIX}#{escaped_source}(?:#{SOURCE_SUFFIX})?/i
     end
 
     def other?

lib/licensee/matchers/reference.rb

@@ -4,7 +4,8 @@ module Licensee
     class Reference < Licensee::Matchers::Matcher
       def match
         License.all(hidden: true, psuedo: false).find do |license|
-          /\b#{license.title_regex}\b/ =~ file.content
+          title_or_source = [license.title_regex, license.source_regex].compact
+          /\b#{Regexp.union(title_or_source)}\b/ =~ file.content
         end
       end
 

spec/licensee/license_spec.rb

@@ -421,4 +421,41 @@ RSpec.describe Licensee::License do
       end
     end
   end
+
+  context 'source regex' do
+    Licensee::License.all(hidden: true, psuedo: false).each do |license|
+      context "the #{license.title} license" do
+        let(:source) { URI.parse(license.source) }
+
+        %w[http https].each do |scheme|
+          context "with a #{scheme}:// scheme" do
+            before { source.scheme = scheme }
+
+            ['www.', ''].each do |prefix|
+              context "with '#{prefix}' before the host" do
+                before do
+                  source.host = "#{prefix}#{source.host.sub(/\Awww\./, '')}"
+                end
+
+                ['.html', '.htm', '.txt', ''].each do |suffix|
+                  context "with '#{suffix}' after the path" do
+                    before do
+                      next if license.key == 'wtfpl'
+                      regex = /#{Licensee::License::SOURCE_SUFFIX}\z/
+                      source.path = source.path.sub(regex, '')
+                      source.path = "#{source.path}#{suffix}"
+                    end
+
+                    it 'matches' do
+                      expect(source.to_s).to match(license.source_regex)
+                    end
+                  end
+                end
+              end
+            end
+          end
+        end
+      end
+    end
+  end
 end

spec/licensee/matchers/reference_matcher_spec.rb

@@ -39,4 +39,13 @@ RSpec.describe Licensee::Matchers::Reference do
       expect(subject.match).to eql(license)
     end
   end
+
+  context 'with a license source' do
+    let(:license) { Licensee::License.find('mpl-2.0') }
+    let(:content) { "The [license](#{license.source})" }
+
+    it 'matches' do
+      expect(subject.match).to eql(license)
+    end
+  end
 end