licensee 9.11.0 → 9.12.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 613429f0350489c8dbdbdc80973d2557ad655efd2b89786096f290cf52632344
-  data.tar.gz: 87fca6ceddb6972dd946720355e6efb2661793d671ff92ce8934e281affa6ad5
+  metadata.gz: d7dc009b0467cfb305e8dac051ed4e78d2f35d0454f2e14cef0952338540f8ae
+  data.tar.gz: 3c27bb3dd3cea6d62fab826b81fab93d9152893851b541c91d69406cdf9fcbd8
 SHA512:
-  metadata.gz: 7bcb1c30bdebf4cad0a84ef7fd9dd4afe922ba8acac1a9124bdd96dc64d4ef736787f1b70c4ab0d01c3c05ba3547300caa3c882b2d99ea15117f72335f56b0a5
-  data.tar.gz: bb2f30c4f70a4610f55a530c36b65852270558b8b36342b1fe8ad1113afa4dca711214545b3abbeb19cdba8d7e2d05643b0e138c8964a5b3f51106ce8c1bf3db
+  metadata.gz: 07f19b33f70b0b73611d34e474f2aa4e4d7f62c7451cdf70f76774beceac2c75ab3d1cc5048061a848b979a54032aad6dd1ba278c79cd798029efd6873d54425
+  data.tar.gz: 96c5e66f65307e7feb2c00b3f06661b093c60995d049f7fd19cc27b76881965a1d33768a16b0a3a3b085e9392ef828dce9cf692ee04255dd9ea2c6d22da38da6

data/bin/licensee

@@ -1,4 +1,5 @@
 #!/usr/bin/env ruby
+# frozen_string_literal: true
 
 require 'dotenv/load'
 require 'thor'

data/lib/licensee.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'licensee/version'
 require 'forwardable'
 require 'pathname'
@@ -19,7 +21,7 @@ module Licensee
   CONFIDENCE_THRESHOLD = 98
 
   # Base domain from which to build license URLs
-  DOMAIN = 'http://choosealicense.com'.freeze
+  DOMAIN = 'http://choosealicense.com'
 
   class << self
     attr_writer :confidence_threshold
@@ -49,7 +51,7 @@ module Licensee
     end
 
     # Inverse of the confidence threshold, represented as a float
-    # By default this will be 0.05
+    # By default this will be 0.02
     def inverse_confidence_threshold
       @inverse_confidence_threshold ||=
         (1 - Licensee.confidence_threshold / 100.0).round(2)

data/lib/licensee/commands/detect.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class LicenseeCLI < Thor
   # Methods to call when displaying information about ProjectFiles
   MATCHED_FILE_METHODS = %i[

data/lib/licensee/commands/diff.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'tmpdir'
 
 class LicenseeCLI < Thor
@@ -47,7 +49,9 @@ class LicenseeCLI < Thor
   end
 
   def expected_license
-    @expected_license ||= Licensee::License.find options[:license] if options[:license]
+    if options[:license]
+      @expected_license ||= Licensee::License.find options[:license]
+    end
     return @expected_license if @expected_license
 
     if options[:license]

data/lib/licensee/commands/license_path.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class LicenseeCLI < Thor
   desc 'license-path [PATH]', "Returns the path to the given project's license file"
   def license_path(_path)

data/lib/licensee/commands/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class LicenseeCLI < Thor
   desc 'version', 'Return the Licensee version'
   def version

data/lib/licensee/content_helper.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'set'
 require 'digest'
 
@@ -22,7 +24,8 @@ module Licensee
       bullet:              /\n\n\s*(?:[*-]|\(?[\da-z]{1,2}[)\.])\s+/i,
       developed_by:        /#{START_REGEX}developed by:.*?\n\n/im,
       quote_begin:         /[`'"‘“]/,
-      quote_end:           /[`'"’”]/
+      quote_end:           /[`'"’”]/,
+      mit_optional:        /\(including the next paragraph\)/i
     }.freeze
     NORMALIZATIONS = {
       lists:      { from: /^\s*(?:\d\.|\*)\s+([^\n])/, to: '- \1' },
@@ -85,13 +88,12 @@ module Licensee
       hrs markdown_headings borders title version url copyright
       block_markup span_markup link_markup
       all_rights_reserved developed_by end_of_terms whitespace
+      mit_optional
     ].freeze
 
     # A set of each word in the license, without duplicates
     def wordset
-      @wordset ||= if content_normalized
-        content_normalized.scan(/(?:\w(?:'s|(?<=s)')?)+/).to_set
-      end
+      @wordset ||= content_normalized&.scan(/(?:\w(?:'s|(?<=s)')?)+/)&.to_set
     end
 
     # Number of characteres in the normalized content
@@ -115,8 +117,10 @@ module Licensee
     # Given another license or project file, calculates the similarity
     # as a percentage of words in common
     def similarity(other)
-      overlap = (wordset & other.wordset).size
-      total = wordset.size + other.wordset.size
+      wordset_fieldless = wordset - LicenseField.keys
+      fields_removed = wordset.size - wordset_fieldless.size
+      overlap = (wordset_fieldless & other.wordset).size
+      total = wordset_fieldless.size + other.wordset.size - fields_removed
       100.0 * (overlap * 2.0 / total)
     end
 
@@ -132,7 +136,7 @@ module Licensee
     def content_without_title_and_version
       @content_without_title_and_version ||= begin
         @_content = nil
-        ops = %i[hrs comments markdown_headings title version]
+        ops = %i[html hrs comments markdown_headings title version]
         ops.each { |op| strip(op) }
         _content
       end
@@ -183,7 +187,7 @@ module Licensee
     end
 
     def self.format_percent(float)
-      "#{format('%.2f', float)}%"
+      "#{format('%<float>.2f', float: float)}%"
     end
 
     def self.title_regex
@@ -206,11 +210,9 @@ module Licensee
 
     private
 
-    # rubocop:disable Naming/MemoizedInstanceVariableName
     def _content
       @_content ||= content.to_s.dup.strip
     end
-    # rubocop:enable Naming/MemoizedInstanceVariableName
 
     def strip(regex_or_sym)
       return unless _content
@@ -265,6 +267,14 @@ module Licensee
       normalize(REGEXES[:link_markup], '\1')
     end
 
+    def strip_html
+      return unless respond_to?(:filename) && filename
+      return unless File.extname(filename) =~ /\.html?/i
+
+      require 'reverse_markdown'
+      @_content = ReverseMarkdown.convert(_content, unknown_tags: :bypass)
+    end
+
     def normalize(from_or_key, to = nil)
       operation = { from: from_or_key, to: to } if to
       operation ||= NORMALIZATIONS[from_or_key]

data/lib/licensee/hash_helper.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module HashHelper
     def to_h

data/lib/licensee/license.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'uri'
 
 module Licensee

data/lib/licensee/license_field.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   class LicenseField < Struct.new(:name, :description)
     class << self

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.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

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

data/lib/licensee/matchers/copyright.rb

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

data/lib/licensee/matchers/cran.rb

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

data/lib/licensee/matchers/dice.rb

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

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

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

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

data/lib/licensee/matchers/matcher.rb

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

data/lib/licensee/matchers/npm_bower.rb

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

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Licensee
   module Matchers
     # Matches README files that include a license by reference

data/lib/licensee/matchers/spdx.rb

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

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

@@ -1,10 +1,12 @@
+# 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 = %w[md markdown txt html].freeze
       PREFERRED_EXT_REGEX = /\.#{Regexp.union(PREFERRED_EXT)}\z/.freeze
 
       # Regex to match any extension except .spdx or .header
@@ -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
@@ -32,7 +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?
@@ -99,6 +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

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