licensee 9.10.1 → 9.13.2

data/lib/licensee/hash_helper.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module HashHelper
     def to_h
@@ -6,11 +8,11 @@ module Licensee
         key = method.to_s.delete('?').to_sym
         value = public_send(method)
         hash[key] = if value.is_a?(Array)
-          value.map { |v| v.respond_to?(:to_h) ? v.to_h : v }
-        elsif value.respond_to?(:to_h) && !value.nil?
-          value.to_h
-        else
-          value
+                      value.map { |v| v.respond_to?(:to_h) ? v.to_h : v }
+                    elsif value.respond_to?(:to_h) && !value.nil?
+                      value.to_h
+                    else
+                      value
         end
       end
 

data/lib/licensee/license.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'uri'
 
 module Licensee
@@ -23,6 +25,7 @@ module Licensee
           output = licenses.dup
           output.reject!(&:hidden?) unless options[:hidden]
           output.reject!(&:pseudo_license?) unless options[:pseudo]
+          output.sort_by!(&:key)
           return output if options[:featured].nil?
 
           output.select { |l| l.featured? == options[:featured] }
@@ -89,14 +92,8 @@ module Licensee
       pseudo:   true
     }.freeze
 
-    ALT_TITLE_REGEX = {
-      'bsd-2-clause'       => /bsd 2-clause(?: \"simplified\")?/i,
-      'bsd-3-clause'       => /bsd 3-clause(?: \"new\" or \"revised\")?/i,
-      'bsd-3-clause-clear' => /(?:clear bsd|bsd 3-clause(?: clear)?)/i
-    }.freeze
-
-    SOURCE_PREFIX = %r{https?://(?:www\.)?}i
-    SOURCE_SUFFIX = %r{(?:\.html?|\.txt|\/)(?:\?[^\s]*)?}i
+    SOURCE_PREFIX = %r{https?://(?:www\.)?}i.freeze
+    SOURCE_SUFFIX = %r{(?:\.html?|\.txt|\/)(?:\?[^\s]*)?}i.freeze
 
     HASH_METHODS = %i[
       key spdx_id meta url rules fields other? gpl? lgpl? cc?
@@ -141,20 +138,22 @@ module Licensee
     def title_regex
       return @title_regex if defined? @title_regex
 
-      title_regex = ALT_TITLE_REGEX[key]
-      title_regex ||= begin
-        string = name.downcase.sub('*', 'u')
-        string.sub!(/\Athe /i, '')
-        string.sub!(/ license\z/i, '')
-        string.sub!(/,? version /, ' ')
-        string.sub!(/v(\d+\.\d+)/, '\1')
-        string = Regexp.escape(string)
-        string = string.sub(/ (\d+\\.\d+)/, ',?\ (?:version\ |v)?\1')
-        string = string.sub(/\bgnu\\ /, '(?:GNU )?')
-        Regexp.new string, 'i'
-      end
-
-      parts = [title_regex, key]
+      string = name.downcase.sub('*', 'u')
+      string.sub!(/\Athe /i, '')
+      string.sub!(/,? version /, ' ')
+      string.sub!(/v(\d+\.\d+)/, '\1')
+      string = Regexp.escape(string)
+      string = string.sub(/\\ licen[sc]e/i, '(?:\ licen[sc]e)?')
+      string = string.sub(/\\ (\d+\\.\d+)/, ',?\s+(?:version\ |v(?:\. )?)?\1')
+      string = string.sub(/\bgnu\\ /, '(?:GNU )?')
+      title_regex = Regexp.new string, 'i'
+
+      string = key.sub('-', '[- ]')
+      string.sub!('.', '\.')
+      string << '(?:\ licen[sc]e)?'
+      key_regex = Regexp.new string, 'i'
+
+      parts = [title_regex, key_regex]
       if meta.nickname
         parts.push Regexp.new meta.nickname.sub(/\bGNU /i, '(?:GNU )?')
       end

data/lib/licensee/license_field.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   class LicenseField < Struct.new(:name, :description)
     class << self
@@ -45,7 +47,7 @@ module Licensee
     end
 
     alias key name
-    FIELD_REGEX = /\[(#{Regexp.union(LicenseField.keys)})\]/
+    FIELD_REGEX = /\[(#{Regexp.union(LicenseField.keys)})\]/.freeze
 
     # The human-readable field name
     def label

data/lib/licensee/license_meta.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   class LicenseMeta < Struct.new(
     :title, :spdx_id, :source, :description, :how, :conditions, :permissions,

data/lib/licensee/license_rules.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   # Exposes #conditions, #permissions, and #limitation arrays of LicenseRules
   class LicenseRules < Struct.new(:conditions, :permissions, :limitations)

data/lib/licensee/matchers.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     autoload :Matcher,   'licensee/matchers/matcher'

data/lib/licensee/matchers/cabal.rb

@@ -1,15 +1,29 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Cabal < Licensee::Matchers::Package
       # While we could parse the cabal file, prefer
       # a lenient regex for speed and security. Moar parsing moar problems.
-      LICENSE_REGEX = /^\s*license\s*\:\s*([a-z\-0-9\.]+)\s*$/ix
+      LICENSE_REGEX = /^\s*license\s*\:\s*([a-z\-0-9\.]+)\s*$/ix.freeze
+      LICENSE_CONVERSIONS = {
+        'GPL-2'  => 'GPL-2.0',
+        'GPL-3'  => 'GPL-3.0',
+        'LGPL-3' => 'LGPL-3.0',
+        'AGPL-3' => 'AGPL-3.0',
+        'BSD2'   => 'BSD-2-Clause',
+        'BSD3'   => 'BSD-3-Clause'
+      }.freeze
 
       private
 
       def license_property
         match = @file.content.match LICENSE_REGEX
-        match[1].downcase if match && match[1]
+        spdx_name(match[1]).downcase if match && match[1]
+      end
+
+      def spdx_name(cabal_name)
+        LICENSE_CONVERSIONS[cabal_name] || cabal_name
       end
     end
   end

data/lib/licensee/matchers/cargo.rb

@@ -1,9 +1,11 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Cargo < Licensee::Matchers::Package
       LICENSE_REGEX = %r{
         ^\s*[\'\"]?license[\'\"]?\s*=\s*[\'\"]([a-z\-0-9\. +()\/]+)[\'\"]\s*
-      }ix
+      }ix.freeze
 
       private
 

data/lib/licensee/matchers/copyright.rb

@@ -1,16 +1,18 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Copyright < Licensee::Matchers::Matcher
       attr_reader :file
 
-      # rubocop:disable Metrics/LineLength
+      # rubocop:disable Layout/LineLength
       COPYRIGHT_SYMBOLS = Regexp.union([/copyright/i, /\(c\)/i, "\u00A9", "\xC2\xA9"])
-      REGEX = /\A(\s*#{COPYRIGHT_SYMBOLS}.*$)+/i
-      # rubocop:enable Metrics/LineLength
+      REGEX = /#{ContentHelper::START_REGEX}(?:portions )?(\s*#{COPYRIGHT_SYMBOLS}.*$)+$/i.freeze
+      # rubocop:enable Layout/LineLength
 
       def match
         # Note: must use content, and not content_normalized here
-        if @file.content.strip =~ /\A#{REGEX}\z/i
+        if file.content.strip =~ /#{REGEX}+\z/i
           Licensee::License.find('no-license')
         end
       rescue Encoding::CompatibilityError

data/lib/licensee/matchers/cran.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Cran < Licensee::Matchers::Package
@@ -5,9 +7,9 @@ module Licensee
 
       # While we could parse the DESCRIPTION file, prefer
       # a lenient regex for speed and security. Moar parsing moar problems.
-      LICENSE_FIELD_REGEX = /^license:\s*(.+)/i
-      PLUS_FILE_LICENSE_REGEX = /\s*\+\s*file\s+LICENSE$/i
-      GPL_VERSION_REGEX = /^GPL(?:-([23])|\s*\(\s*>=\s*([23])\s*\))$/i
+      LICENSE_FIELD_REGEX = /^license:\s*(.+)/i.freeze
+      PLUS_FILE_LICENSE_REGEX = /\s*\+\s*file\s+LICENSE$/i.freeze
+      GPL_VERSION_REGEX = /^GPL(?:-([23])|\s*\(\s*>=\s*([23])\s*\))$/i.freeze
 
       private
 

data/lib/licensee/matchers/dice.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Dice < Licensee::Matchers::Matcher
@@ -5,9 +7,9 @@ module Licensee
       # than the confidence threshold
       def match
         @match ||= if matches.empty?
-          nil
-        else
-          matches.first[0]
+                     nil
+                   else
+                     matches.first[0]
         end
       end
 
@@ -49,7 +51,7 @@ module Licensee
 
       # Confidence that the matched license is a match
       def confidence
-        @confidence ||= match ? file.similarity(match) : 0
+        @confidence ||= match ? match.similarity(file) : 0
       end
 
       private

data/lib/licensee/matchers/dist_zilla.rb

@@ -1,9 +1,11 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class DistZilla < Licensee::Matchers::Package
       attr_reader :file
 
-      LICENSE_REGEX = /^license\s*=\s*([a-z\-0-9\._]+)/i
+      LICENSE_REGEX = /^license\s*=\s*([a-z\-0-9\._]+)/i.freeze
 
       private
 

data/lib/licensee/matchers/exact.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Exact < Licensee::Matchers::Matcher

data/lib/licensee/matchers/gemspec.rb

@@ -1,26 +1,28 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Gemspec < Licensee::Matchers::Package
       # a value is a string surrounded by any amount of whitespace
       # optionally ended with (non-captured) ".freeze"
-      VALUE_REGEX = /\s*[\'\"]([a-z\-0-9\.]+)[\'\"](?:\.freeze)?\s*/i
+      VALUE_REGEX = /\s*[\'\"]([a-z\-0-9\.]+)[\'\"](?:\.freeze)?\s*/i.freeze
 
       # an array contains one or more values. all values, or array itself,
       # can be surrounded by any amount of whitespace.  do not capture
       # non-value groups
-      ARRAY_REGEX = /\s*\[#{VALUE_REGEX}(?:,#{VALUE_REGEX})*\]\s*/i
+      ARRAY_REGEX = /\s*\[#{VALUE_REGEX}(?:,#{VALUE_REGEX})*\]\s*/i.freeze
 
       DECLARATION_REGEX = /
         ^\s*[a-z0-9_]+\.([a-z0-9_]+)\s*\=#{VALUE_REGEX}$
-      /ix
+      /ix.freeze
 
       LICENSE_REGEX = /
         ^\s*[a-z0-9_]+\.license\s*\=#{VALUE_REGEX}$
-      /ix
+      /ix.freeze
 
       LICENSE_ARRAY_REGEX = /
         ^\s*[a-z0-9_]+\.licenses\s*\=#{ARRAY_REGEX}$
-      /ix
+      /ix.freeze
 
       private
 

data/lib/licensee/matchers/matcher.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Matcher
@@ -25,7 +27,7 @@ module Licensee
       private
 
       def potential_matches
-        @potential_matches ||= Licensee.licenses(hidden: true)
+        @potential_matches ||= Licensee.licenses(hidden: true, psuedo: false)
       end
     end
   end

data/lib/licensee/matchers/npm_bower.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class NpmBower < Licensee::Matchers::Package
@@ -5,7 +7,7 @@ module Licensee
       # a lenient regex for speed and security. Moar parsing moar problems.
       LICENSE_REGEX = /
         \s*[\"\']license[\"\']\s*\:\s*[\'\"]([a-z\-0-9\.+ ()]+)[\'\"],?\s*
-      /ix
+      /ix.freeze
 
       private
 

data/lib/licensee/matchers/package.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Package < Licensee::Matchers::Matcher

data/lib/licensee/matchers/reference.rb

@@ -1,9 +1,11 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     # Matches README files that include a license by reference
     class Reference < Licensee::Matchers::Matcher
       def match
-        License.all(hidden: true, psuedo: false).find do |license|
+        potential_matches.find do |license|
           title_or_source = [license.title_regex, license.source_regex].compact
           /\b#{Regexp.union(title_or_source)}\b/ =~ file.content
         end

data/lib/licensee/matchers/spdx.rb

@@ -1,9 +1,11 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     class Spdx < Licensee::Matchers::Package
       # While we could parse the LICENSE.spdx file, prefer
       # a lenient regex for speed and security. Moar parsing moar problems.
-      LICENSE_REGEX = /PackageLicenseDeclared:\s*([a-z\-0-9\. +()]+)\s*/i
+      LICENSE_REGEX = /PackageLicenseDeclared:\s*([a-z\-0-9\. +()]+)\s*/i.freeze
 
       private
 

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]
@@ -72,11 +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

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
@@ -32,10 +35,10 @@ 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)
+          @content.encode!(ENCODING, **ENCODING_OPTIONS)
         end
 
         metadata = { name: metadata } if metadata.is_a? String
@@ -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?
@@ -99,6 +102,10 @@ module Licensee
       def content_normalized
         nil
       end
+
+      def attribution
+        nil
+      end
     end
   end
 end