licensee 9.7.0 → 9.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 688ece76934b3607042666300f8f38cf174db32da4c06e99823f32fc1f043375
-  data.tar.gz: 22d8028390732a2b75040153fdcc8fd630aa1af0f3ce9844fb023a39bd433955
+  metadata.gz: 6da17475ba74d0aea6d92013e6d141cc88f0d400af3aa9d0be99947c551f3ad6
+  data.tar.gz: 654876f0d889688a13752497759d2856be48a8714667a167e9781f80b9f1eef1
 SHA512:
-  metadata.gz: ba90beba409193357cf1a685b805e0c9acb68e1771b58bdc02114f4a1290a6a15b8d912e07d8c108b7c7adfcfc79454658154bb421a0fa9eeb4466cffb3b4d0e
-  data.tar.gz: f355dcb2386c642d745d791597015a2ce024775c0d07a24f95059baa58831e1835086a23f57a5290b99e750a9866a7e8255c65eb939849e010b9aeb4aa84b8b2
+  metadata.gz: 208fea169fb835f863d85ec6ac2bbcb87878c10e312938380fd262cfc3b3b7fe0e55b65d74cdd166a1c361d47d071ec5bfc785e4a7854031aa0855fc96105c54
+  data.tar.gz: 1c23cf247b226d85326ac7c9d388b22e228f07a13388d0e21cac091f6ca8fefd1c4b726448c13e44c429e83c828ef99b56cfaea5c39ae30ef22fc13655027d29

data/lib/licensee/content_helper.rb

@@ -16,7 +16,7 @@ module Licensee
     # A set of each word in the license, without duplicates
     def wordset
       @wordset ||= if content_normalized
-        content_normalized.scan(/[\w']+/).to_set
+        content_normalized.scan(/(?:\w(?:'s|(?<=s)')?)+/).to_set
       end
     end
 
@@ -79,6 +79,7 @@ module Licensee
         string = strip_all_rights_reserved(string)
         string, _partition, _instructions = string.partition(END_OF_TERMS_REGEX)
         string = strip_markup(string)
+        string = normalize_quotes(string)
         strip_whitespace(string)
       end
 
@@ -164,5 +165,12 @@ module Licensee
     def strip(string, regex)
       string.gsub(regex, ' ').squeeze(' ').strip
     end
+
+    # Replace all single quotes with double quotes
+    # Single versus double quotes don't alter the meaning, and it's easier to
+    # strip double quotes if we still want to allow possessives
+    def normalize_quotes(string)
+      string.gsub(/\s'([\w -]*?\w)'/, ' "\1"')
+    end
   end
 end

data/lib/licensee/matchers.rb

@@ -2,6 +2,7 @@ module Licensee
   module Matchers
     autoload :Matcher,   'licensee/matchers/matcher'
     autoload :Cabal,     'licensee/matchers/cabal'
+    autoload :Cargo,     'licensee/matchers/cargo'
     autoload :Copyright, 'licensee/matchers/copyright'
     autoload :Cran,      'licensee/matchers/cran'
     autoload :Dice,      'licensee/matchers/dice'

data/lib/licensee/matchers/cargo.rb

@@ -0,0 +1,16 @@
+module Licensee
+  module Matchers
+    class Cargo < Licensee::Matchers::Package
+      LICENSE_REGEX = %r{
+        ^\s*[\'\"]?license[\'\"]?\s*=\s*[\'\"]([a-z\-0-9\. +()\/]+)[\'\"]\s*
+      }ix
+
+      private
+
+      def license_property
+        match = @file.content.match LICENSE_REGEX
+        match[1].downcase if match && match[1]
+      end
+    end
+  end
+end

data/lib/licensee/project_files/package_manager_file.rb

@@ -12,15 +12,18 @@ module Licensee
       FILENAMES_EXTENSIONS = {
         'DESCRIPTION'  => [Matchers::Cran],
         'dist.ini'     => [Matchers::DistZilla],
-        'LICENSE.spdx' => [Matchers::Spdx]
+        'LICENSE.spdx' => [Matchers::Spdx],
+        'Cargo.toml'   => [Matchers::Cargo]
       }.freeze
 
       FILENAMES_SCORES = {
-        'package.json' => 1.0,
-        'LICENSE.spdx' => 1.0,
-        'DESCRIPTION'  => 0.9,
-        'dist.ini'     => 0.8,
-        'bower.json'   => 0.75
+        'package.json'     => 1.0,
+        'LICENSE.spdx'     => 1.0,
+        'Cargo.toml'       => 1.0,
+        'DESCRIPTION'      => 0.9,
+        'dist.ini'         => 0.8,
+        'bower.json'       => 0.75,
+        'elm-package.json' => 0.7
       }.freeze
 
       def possible_matchers

data/lib/licensee/project_files/readme_file.rb

@@ -1,21 +1,30 @@
 module Licensee
   module ProjectFiles
     class ReadmeFile < Licensee::ProjectFiles::LicenseFile
+      EXTENSIONS = %w[md markdown mdown txt rdoc].freeze
       SCORES = {
-        /\AREADME\z/i                          => 1.0,
-        /\AREADME\.(md|markdown|mdown|txt)\z/i => 0.9
+        /\AREADME\z/i                                       => 1.0,
+        /\AREADME\.(#{Regexp.union(EXTENSIONS).source})\z/i => 0.9
       }.freeze
 
+      TITLE_REGEX = /licen[sc]e:?/i
+      UNDERLINE_REGEX = /\n[-=]+/m
       CONTENT_REGEX = /^
-          (?:\#+\sLicen[sc]e     # Start of hashes-based license header
-             |
-             Licen[sc]e\n[-=]+)$ # Start of underlined license header
-          (.*?)                  # License content
-          (?=^(?:\#+             # Next hashes-based header
-                 |
-                 [^\n]+\n[-=]+)  # Next of underlined header
-             |
-             \z)                 # End of file
+          (?:                                # Header lookbehind
+            [\#=]+\s#{TITLE_REGEX}           # Start of hashes or rdoc header
+          |
+            #{TITLE_REGEX}#{UNDERLINE_REGEX} # Start of underlined header
+          )$
+          (.*?)                              # License content
+          (?=^                               # Header or end of file lookahead
+            (?:
+              [\#=]+                         # Next hash or rdoc header
+            |
+              [^\n]+#{UNDERLINE_REGEX}       # Next of underlined header
+            )
+          |
+            \z                               # End of file
+          )
         /mix
 
       def possible_matchers

data/lib/licensee/version.rb

@@ -1,3 +1,3 @@
 module Licensee
-  VERSION = '9.7.0'.freeze
+  VERSION = '9.8.0'.freeze
 end

data/spec/licensee/content_helper_spec.rb

@@ -21,6 +21,9 @@ RSpec.describe Licensee::ContentHelper do
   The made
   * * * *
   up  license.
+
+  This license provided 'as is'. Please respect the contributors' wishes when
+  implementing the license's "software".
   -----------
 LICENSE
   end
@@ -28,11 +31,17 @@ LICENSE
   let(:mit) { Licensee::License.find('mit') }
 
   it 'creates the wordset' do
-    expect(subject.wordset).to eql(Set.new(%w[the made up license]))
+    wordset = Set.new(
+      %w[
+        the made up license this provided as is please respect
+        contributors' wishes when implementing license's software
+      ]
+    )
+    expect(subject.wordset).to eql(wordset)
   end
 
   it 'knows the length' do
-    expect(subject.length).to eql(20)
+    expect(subject.length).to eql(135)
   end
 
   context 'a very long license' do
@@ -44,17 +53,17 @@ LICENSE
   end
 
   it 'knows the length delta' do
-    expect(subject.length_delta(mit)).to eql(999)
+    expect(subject.length_delta(mit)).to eql(884)
     expect(subject.length_delta(subject)).to eql(0)
   end
 
   it 'knows the similarity' do
-    expect(subject.similarity(mit)).to be_within(1).of(2)
+    expect(subject.similarity(mit)).to be_within(1).of(11)
     expect(subject.similarity(subject)).to eql(100.0)
   end
 
   it 'calculates the hash' do
-    content_hash = '3c59634b9fae4396a76a978f3f6aa718ed790a9a'
+    content_hash = '916b978940ecf8070c96bd3aca9321768e7f4901'
     expect(subject.content_hash).to eql(content_hash)
   end
 
@@ -120,6 +129,19 @@ LICENSE
       expect(normalized_content).to_not match(/[*=_-]+/)
     end
 
+    it 'normalizes quotes' do
+      expect(normalized_content).to_not match("'as is'")
+    end
+
+    it 'preserves possessives' do
+      expect(normalized_content).to match("contributors'")
+      expect(normalized_content).to match("license's")
+    end
+
+    it 'preserves double quotes' do
+      expect(normalized_content).to match('"software"')
+    end
+
     Licensee::License.all(hidden: true).each do |license|
       context license.name do
         let(:stripped_content) { subject.content_without_title_and_version }
@@ -158,7 +180,10 @@ LICENSE
     end
 
     it 'normalize the content' do
-      expect(normalized_content).to eql 'the made up license.'
+      expected = 'the made up license. this license provided "as is". '
+      expected << "please respect the contributors' wishes when implementing "
+      expected << "the license's \"software\"."
+      expect(normalized_content).to eql(expected)
     end
 
     context 'a title in parenthesis' do

data/spec/licensee/matchers/cargo_matcher_spec.rb

@@ -0,0 +1,48 @@
+RSpec.describe Licensee::Matchers::Cargo do
+  let(:mit) { Licensee::License.find('mit') }
+  let(:content) { 'license = "MIT"' }
+  let(:file) { Licensee::ProjectFiles::LicenseFile.new(content, 'Cargo.toml') }
+  subject { described_class.new(file) }
+
+  it 'stores the file' do
+    expect(subject.file).to eql(file)
+  end
+
+  it 'has confidence' do
+    expect(subject.confidence).to eql(90)
+  end
+
+  it 'matches' do
+    expect(subject.match).to eql(mit)
+  end
+
+  {
+    'spdx name'                => ['license = "MIT"', 'mit'],
+    'single quotes'            => ["license = 'mit'", 'mit'],
+    'quoted key'               => ["'license' = 'mit'", 'mit'],
+    'double quoted key'        => ['"license"="mit"', 'mit'],
+    'no whitespace'            => ["license='mit'", 'mit'],
+    'leading whitespace'       => [" license = 'mit'", 'mit'],
+    'other license'            => ['license = "Foo"', 'other'],
+    'multiple licenses /'      => ['license = "Apache-2.0/MIT"', 'other'],
+    'multiple licenses OR'     => ['license = "Apache-2.0 OR MIT"', 'other'],
+    'multiple licenses parens' => ['license = "(Apache-2.0 OR MIT)"', 'other']
+  }.each do |description, license_declaration_and_key|
+    context "with a #{description}" do
+      let(:content) { license_declaration_and_key[0] }
+      let(:license) { Licensee::License.find(license_declaration_and_key[1]) }
+
+      it 'matches' do
+        expect(subject.match).to eql(license)
+      end
+    end
+  end
+
+  context 'no license field' do
+    let(:content) { 'foo = "bar"' }
+
+    it 'returns nil' do
+      expect(subject.match).to be_nil
+    end
+  end
+end

data/spec/licensee/project_files/package_info_spec.rb

@@ -8,9 +8,11 @@ RSpec.describe Licensee::ProjectFiles::PackageManagerFile do
       'licensee.gemspec' => 1.0,
       'test.cabal'       => 1.0,
       'package.json'     => 1.0,
+      'Cargo.toml'       => 1.0,
       'DESCRIPTION'      => 0.9,
       'dist.ini'         => 0.8,
       'bower.json'       => 0.75,
+      'elm-package.json' => 0.70,
       'README.md'        => 0.0
     }.each do |filename, expected_score|
       context "a file named #{filename}" do

data/spec/licensee/project_files/{readme_spec.rb → readme_file_spec.rb}

@@ -11,6 +11,7 @@ RSpec.describe Licensee::ProjectFiles::ReadmeFile do
       'README.md'    => 0.9,
       'readme.txt'   => 0.9,
       'readme.mdown' => 0.9,
+      'readme.rdoc'  => 0.9,
       'LICENSE'      => 0.0
     }.each do |filename, expected_score|
       context "with a file named #{filename}" do
@@ -97,6 +98,22 @@ RSpec.describe Licensee::ProjectFiles::ReadmeFile do
         expect(license).to eql('hello world')
       end
     end
+
+    context 'With a trailing colon' do
+      let(:content) { "## License:\n\nhello world" }
+
+      it 'returns the license' do
+        expect(license).to eql('hello world')
+      end
+    end
+
+    context 'Rdoc' do
+      let(:content) { "== License:\n\nhello world" }
+
+      it 'returns the license' do
+        expect(license).to eql('hello world')
+      end
+    end
   end
 
   context 'a license reference' do

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: licensee
 version: !ruby/object:Gem::Version
-  version: 9.7.0
+  version: 9.8.0
 platform: ruby
 authors:
 - Ben Balter
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-01-02 00:00:00.000000000 Z
+date: 2018-01-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: octokit
@@ -161,6 +161,7 @@ files:
 - lib/licensee/license_rules.rb
 - lib/licensee/matchers.rb
 - lib/licensee/matchers/cabal.rb
+- lib/licensee/matchers/cargo.rb
 - lib/licensee/matchers/copyright.rb
 - lib/licensee/matchers/cran.rb
 - lib/licensee/matchers/dice.rb
@@ -227,6 +228,7 @@ files:
 - spec/licensee/license_rules_spec.rb
 - spec/licensee/license_spec.rb
 - spec/licensee/matchers/cabal_matcher_spec.rb
+- spec/licensee/matchers/cargo_matcher_spec.rb
 - spec/licensee/matchers/copyright_matcher_spec.rb
 - spec/licensee/matchers/cran_matcher_spec.rb
 - spec/licensee/matchers/dice_matcher_spec.rb
@@ -240,7 +242,7 @@ files:
 - spec/licensee/project_files/license_file_spec.rb
 - spec/licensee/project_files/package_info_spec.rb
 - spec/licensee/project_files/project_file_spec.rb
-- spec/licensee/project_files/readme_spec.rb
+- spec/licensee/project_files/readme_file_spec.rb
 - spec/licensee/project_spec.rb
 - spec/licensee/projects/github_project_spec.rb
 - spec/licensee/rule_spec.rb