licensee 9.10.0 → 9.13.1

data/lib/licensee/project_files.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module ProjectFiles
     autoload :ProjectFile, 'licensee/project_files/project_file'

data/lib/licensee/project_files/license_file.rb

@@ -1,26 +1,28 @@
+# frozen_string_literal: true
+
 module Licensee
   module ProjectFiles
     class LicenseFile < Licensee::ProjectFiles::ProjectFile
       include Licensee::ContentHelper
 
       # List of extensions to give preference to
-      PREFERRED_EXT = %w[md markdown txt].freeze
-      PREFERRED_EXT_REGEX = /\.#{Regexp.union(PREFERRED_EXT)}\z/
+      PREFERRED_EXT = %w[md markdown txt html].freeze
+      PREFERRED_EXT_REGEX = /\.#{Regexp.union(PREFERRED_EXT)}\z/.freeze
 
       # Regex to match any extension except .spdx or .header
-      OTHER_EXT_REGEX = %r{\.(?!spdx|header|gemspec)[^./]+\z}i
+      OTHER_EXT_REGEX = %r{\.(?!spdx|header|gemspec)[^./]+\z}i.freeze
 
       # Regex to match, LICENSE, LICENCE, unlicense, etc.
-      LICENSE_REGEX = /(un)?licen[sc]e/i
+      LICENSE_REGEX = /(un)?licen[sc]e/i.freeze
 
       # Regex to match COPYING, COPYRIGHT, etc.
-      COPYING_REGEX = /copy(ing|right)/i
+      COPYING_REGEX = /copy(ing|right)/i.freeze
 
       # Regex to match OFL.
-      OFL_REGEX = /ofl/i
+      OFL_REGEX = /ofl/i.freeze
 
       # BSD + PATENTS patent file
-      PATENTS_REGEX = /patents/i
+      PATENTS_REGEX = /patents/i.freeze
 
       # Hash of Regex => score with which to score potential license files
       FILENAME_REGEXES = {
@@ -46,7 +48,7 @@ module Licensee
       # detected as CC-BY or CC-BY-SA which are 98%+ similar
       CC_FALSE_POSITIVE_REGEX = /
         ^(creative\ commons\ )?Attribution-(NonCommercial|NoDerivatives)
-      /xi
+      /xi.freeze
 
       def possible_matchers
         [Matchers::Copyright, Matchers::Exact, Matchers::Dice]
@@ -55,6 +57,7 @@ module Licensee
       def attribution
         @attribution ||= begin
           return unless copyright? || license.content =~ /\[fullname\]/
+
           matches = Matchers::Copyright::REGEX
                     .match(content_without_title_and_version)
           matches[0] if matches
@@ -71,11 +74,11 @@ module Licensee
       end
 
       def gpl?
-        license && license.gpl?
+        license&.gpl?
       end
 
       def license
-        if matcher && matcher.match
+        if matcher&.match
           matcher.match
         else
           License.find('other')

data/lib/licensee/project_files/package_manager_file.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module ProjectFiles
     class PackageManagerFile < Licensee::ProjectFiles::ProjectFile
@@ -32,6 +34,7 @@ module Licensee
 
       def self.name_score(filename)
         return 1.0 if ['.gemspec', '.cabal'].include?(File.extname(filename))
+
         FILENAMES_SCORES[filename] || 0.0
       end
 

data/lib/licensee/project_files/project_file.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # A project file is a file within a project that contains license information
 # Currently extended by LicenseFile, PackageManagerFile, and ReadmeFile
 #
@@ -13,6 +15,7 @@ module Licensee
       include Licensee::HashHelper
       HASH_METHODS = %i[
         filename content content_hash content_normalized matcher matched_license
+        attribution
       ].freeze
 
       ENCODING = Encoding::UTF_8
@@ -32,7 +35,7 @@ module Licensee
       #
       # Returns a new Licensee::ProjectFile
       def initialize(content, metadata = {})
-        @content = content
+        @content = content.dup
         @content.force_encoding(ENCODING)
         unless @content.valid_encoding?
           @content.encode!(ENCODING, ENCODING_OPTIONS)
@@ -70,17 +73,17 @@ module Licensee
 
       # Returns the percent confident with the match
       def confidence
-        matcher && matcher.confidence
+        matcher&.confidence
       end
 
       def license
-        matcher && matcher.match
+        matcher&.match
       end
 
       alias match license
 
       def matched_license
-        license.spdx_id if license
+        license&.spdx_id
       end
 
       # Is this file a COPYRIGHT file with only a copyright statement?
@@ -88,6 +91,7 @@ module Licensee
       def copyright?
         return false unless is_a?(LicenseFile)
         return false unless matcher.is_a?(Matchers::Copyright)
+
         filename =~ /\Acopyright(?:#{LicenseFile::OTHER_EXT_REGEX})?\z/i
       end
 
@@ -98,6 +102,10 @@ module Licensee
       def content_normalized
         nil
       end
+
+      def attribution
+        nil
+      end
     end
   end
 end

data/lib/licensee/project_files/readme_file.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module ProjectFiles
     class ReadmeFile < Licensee::ProjectFiles::LicenseFile
@@ -7,8 +9,8 @@ module Licensee
         /\AREADME\.(#{Regexp.union(EXTENSIONS).source})\z/i => 0.9
       }.freeze
 
-      TITLE_REGEX = /licen[sc]e:?/i
-      UNDERLINE_REGEX = /\n[-=]+/m
+      TITLE_REGEX = /licen[sc]e:?/i.freeze
+      UNDERLINE_REGEX = /\n[-=]+/m.freeze
       CONTENT_REGEX = /^
           (?:                                # Header lookbehind
             [\#=]+\s#{TITLE_REGEX}\s*[\#=]*  # Start of hashes or rdoc header
@@ -25,7 +27,7 @@ module Licensee
           |
             \z                               # End of file
           )
-        /mix
+        /mix.freeze
 
       def possible_matchers
         super.push(Matchers::Reference)

data/lib/licensee/projects.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Projects
     autoload :Project, 'licensee/projects/project'

data/lib/licensee/projects/fs_project.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Filesystem-based project
 #
 # Analyze a folder on the filesystem for license information
@@ -36,6 +38,7 @@ module Licensee
           relative_dir = Pathname.new(dir).relative_path_from(dir_path).to_s
           Dir.glob(::File.join(dir, @pattern).tr('\\', '/')).map do |file|
             next unless ::File.file?(file)
+
             { name: ::File.basename(file), dir: relative_dir }
           end.compact
         end

data/lib/licensee/projects/git_project.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Git-based project
 #
 # Analyze a given (bare) Git repository for license information
@@ -11,34 +13,40 @@ autoload :Rugged, 'rugged'
 module Licensee
   module Projects
     class GitProject < Licensee::Projects::Project
-      attr_reader :repository, :revision
+      attr_reader :revision
 
       class InvalidRepository < ArgumentError; end
 
       def initialize(repo, revision: nil, **args)
-        @repository = if repo.is_a? Rugged::Repository
-          repo
-        else
-          Rugged::Repository.new(repo)
-        end
-
+        @raw_repo = repo
         @revision = revision
+
+        raise InvalidRepository if repository.head_unborn?
+
         super(**args)
+      end
+
+      def repository
+        @repository ||= begin
+          return @raw_repo if @raw_repo.is_a? Rugged::Repository
+
+          Rugged::Repository.new(@raw_repo)
+        end
       rescue Rugged::OSError, Rugged::RepositoryError
         raise InvalidRepository
       end
 
       def close
-        @repository.close
+        repository.close
       end
 
       private
 
       def commit
         @commit ||= if revision
-          repository.lookup(revision)
-        else
-          repository.last_commit
+                      repository.lookup(revision)
+                    else
+                      repository.last_commit
         end
       end
 

data/lib/licensee/projects/github_project.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # GitHub project
 #
 # Analyses a remote GitHub repository for license information
@@ -13,13 +15,14 @@ module Licensee
       # If there's any trailing data (e.g. `.git`) this pattern will ignore it:
       # we're going to use the API rather than clone the repo.
       GITHUB_REPO_PATTERN =
-        %r{https://github.com/([^\/]+\/([^\/]+(?=\.git)|[^\/]+)).*}
+        %r{https://github.com/([^\/]+\/([^\/]+(?=\.git)|[^\/]+)).*}.freeze
 
       class RepoNotFound < StandardError; end
 
       def initialize(github_url, **args)
         @repo = github_url[GITHUB_REPO_PATTERN, 1]
         raise ArgumentError, "Not a github URL: #{github_url}" unless @repo
+
         super(**args)
       end
 
@@ -29,8 +32,10 @@ module Licensee
 
       def files
         return @files if defined? @files_from_tree
+
         @files = dir_files
         return @files unless @files.empty?
+
         msg = "Could not load GitHub repo #{repo}, it may be private or deleted"
         raise RepoNotFound, msg
       end

data/lib/licensee/projects/project.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Licensee::Project represents an open source project on disk
 # It is not used directly, but rather is extended by FSProject and GitProject
 # depending on the type of file system access available
@@ -21,10 +23,11 @@ module Licensee
       # Returns the matching License instance if a license can be detected
       def license
         return @license if defined? @license
+
         @license = if licenses_without_copyright.count == 1 || lgpl?
-          licenses_without_copyright.first
-        elsif licenses_without_copyright.count > 1
-          Licensee::License.find('other')
+                     licenses_without_copyright.first
+                   elsif licenses_without_copyright.count > 1
+                     Licensee::License.find('other')
         end
       end
 
@@ -53,6 +56,7 @@ module Licensee
       def license_files
         @license_files ||= begin
           return [] if files.empty? || files.nil?
+
           files = find_files do |n|
             Licensee::ProjectFiles::LicenseFile.name_score(n)
           end
@@ -67,6 +71,7 @@ module Licensee
       def readme_file
         return unless detect_readme?
         return @readme if defined? @readme
+
         @readme = begin
           content, file = find_file do |n|
             Licensee::ProjectFiles::ReadmeFile.name_score(n)
@@ -74,6 +79,7 @@ module Licensee
           content = Licensee::ProjectFiles::ReadmeFile.license_content(content)
 
           return unless content && file
+
           Licensee::ProjectFiles::ReadmeFile.new(content, file)
         end
       end
@@ -82,12 +88,14 @@ module Licensee
       def package_file
         return unless detect_packages?
         return @package_file if defined? @package_file
+
         @package_file = begin
           content, file = find_file do |n|
             Licensee::ProjectFiles::PackageManagerFile.name_score(n)
           end
 
           return unless content && file
+
           Licensee::ProjectFiles::PackageManagerFile.new(content, file)
         end
       end
@@ -96,6 +104,7 @@ module Licensee
 
       def lgpl?
         return false unless licenses.count == 2 && license_files.count == 2
+
         license_files[0].lgpl? && license_files[1].gpl?
       end
 
@@ -104,8 +113,9 @@ module Licensee
       # sorted by file score descending
       def find_files
         return [] if files.empty? || files.nil?
+
         found = files.map { |file| file.merge(score: yield(file[:name])) }
-        found.select! { |file| file[:score] > 0 }
+        found.select! { |file| file[:score].positive? }
         found.sort { |a, b| b[:score] <=> a[:score] }
       end
 
@@ -114,6 +124,7 @@ module Licensee
       # or nil, if no file scored > 0
       def find_file(&block)
         return if files.empty? || files.nil?
+
         file = find_files(&block).first
         [load_file(file), file] if file
       end
@@ -128,7 +139,7 @@ module Licensee
       # Returns an array of LicenseFiles with LPGL first
       def prioritize_lgpl(files)
         return files if files.empty?
-        return files unless files.first.license && files.first.license.gpl?
+        return files unless files.first.license&.gpl?
 
         lesser = files.find_index(&:lgpl?)
         files.unshift(files.delete_at(lesser)) if lesser

data/lib/licensee/rule.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   class Rule
     attr_reader :tag, :label, :description, :group

data/lib/licensee/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
-  VERSION = '9.10.0'.freeze
+  VERSION = '9.13.1'
 end

data/licensee.gemspec

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+require File.expand_path('lib/licensee/version', __dir__)
+
+Gem::Specification.new do |gem|
+  gem.name    = 'licensee'
+  gem.version = Licensee::VERSION
+
+  gem.summary = 'A Ruby Gem to detect open source project licenses'
+  gem.description = <<-DESC
+    Licensee automates the process of reading LICENSE files and
+    compares their contents to known licenses using a fancy maths.
+  DESC
+
+  gem.authors  = ['Ben Balter']
+  gem.email    = 'ben.balter@github.com'
+  gem.homepage = 'https://github.com/benbalter/licensee'
+  gem.license  = 'MIT'
+
+  gem.bindir = 'bin'
+  gem.executables << 'licensee'
+
+  gem.add_dependency('dotenv', '~> 2.0')
+  gem.add_dependency('octokit', '~> 4.17')
+  gem.add_dependency('reverse_markdown', '~> 1.0')
+  gem.add_dependency('rugged', '~> 0.24')
+  gem.add_dependency('thor', '~> 0.19')
+
+  gem.add_development_dependency('mustache', '>= 0.9', '< 2.0')
+  gem.add_development_dependency('pry', '~> 0.9')
+  gem.add_development_dependency('rake', '~> 10.3')
+  gem.add_development_dependency('rspec', '~> 3.5')
+  gem.add_development_dependency('rubocop', '~> 0.80')
+  gem.add_development_dependency('rubocop-performance', '~> 1.5')
+  gem.add_development_dependency('rubocop-rspec', '~> 1.36')
+  gem.add_development_dependency('simplecov', '~> 0.16')
+  gem.add_development_dependency('webmock', '~> 3.1')
+
+  gem.required_ruby_version = '> 2.3'
+
+  # ensure the gem is built out of versioned files
+  gem.files = Dir[
+    'Rakefile',
+    '{bin,lib,man,test,vendor,spec}/**/*',
+    'README*', 'LICENSE*'
+  ] & `git ls-files -z`.split("\0")
+end

data/spec/bin_spec.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 RSpec.describe 'command line invocation' do
   let(:command) { ['bundle', 'exec', 'bin/licensee', 'help'] }
   let(:arguments) { [] }
@@ -12,7 +14,7 @@ RSpec.describe 'command line invocation' do
   let(:status) { output[2] }
 
   it 'Returns a zero exit code' do
-    expect(status.exitstatus).to eql(0)
+    expect(status.exitstatus).to be(0)
   end
 
   it 'returns the help text' do