Merge branch 'master' into sim-scaled

lib/licensee/matchers.rb

@@ -11,5 +11,6 @@ module Matchers
     autoload :NpmBower,  'licensee/matchers/npm_bower'
     autoload :Package,   'licensee/matchers/package'
     autoload :Reference, 'licensee/matchers/reference'
+    autoload :Spdx,      'licensee/matchers/spdx'
   end
 end

lib/licensee/matchers/npm_bower.rb

@@ -4,7 +4,7 @@ class NpmBower < Licensee::Matchers::Package
       # While we could parse the package.json or bower.json file, prefer
       # a lenient regex for speed and security. Moar parsing moar problems.
       LICENSE_REGEX = /
-        \s*[\"\']license[\"\']\s*\:\s*[\'\"]([a-z\-0-9\.]+)[\'\"],?\s*
+        \s*[\"\']license[\"\']\s*\:\s*[\'\"]([a-z\-0-9\.+ ()]+)[\'\"],?\s*
       /ix
 
       private

lib/licensee/matchers/spdx.rb

@@ -0,0 +1,16 @@
+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
+
+      private
+
+      def license_property
+        match = @file.content.match LICENSE_REGEX
+        match[1].downcase if match && match[1]
+      end
+    end
+  end
+end

lib/licensee/project_files/license_file.rb

@@ -7,8 +7,8 @@ class LicenseFile < Licensee::ProjectFiles::ProjectFile
       PREFERRED_EXT = %w[md markdown txt].freeze
       PREFERRED_EXT_REGEX = /\.#{Regexp.union(PREFERRED_EXT)}\z/
 
-      # Regex to match any extension
-      ANY_EXT_REGEX = %r{\.[^./]+\z}
+      # Regex to match any extension except .spdx
+      NONSPDX_EXT_REGEX = %r{\.(?!spdx)[^./]+\z}
 
       # Regex to match, LICENSE, LICENCE, unlicense, etc.
       LICENSE_REGEX = /(un)?licen[sc]e/i
@@ -28,17 +28,17 @@ class LicenseFile < Licensee::ProjectFiles::ProjectFile
         /\A#{LICENSE_REGEX}#{PREFERRED_EXT_REGEX}\z/ => 0.95,  # LICENSE.md
         /\A#{COPYING_REGEX}\z/                       => 0.90,  # COPYING
         /\A#{COPYING_REGEX}#{PREFERRED_EXT_REGEX}\z/ => 0.85,  # COPYING.md
-        /\A#{LICENSE_REGEX}#{ANY_EXT_REGEX}\z/       => 0.80,  # LICENSE.textile
-        /\A#{COPYING_REGEX}#{ANY_EXT_REGEX}\z/       => 0.75,  # COPYING.textile
+        /\A#{LICENSE_REGEX}#{NONSPDX_EXT_REGEX}\z/   => 0.80,  # LICENSE.textile
+        /\A#{COPYING_REGEX}#{NONSPDX_EXT_REGEX}\z/   => 0.75,  # COPYING.textile
         /\A#{LICENSE_REGEX}[-_]/                     => 0.70,  # LICENSE-MIT
         /\A#{COPYING_REGEX}[-_]/                     => 0.65,  # COPYING-MIT
         /[-_]#{LICENSE_REGEX}/                       => 0.60,  # MIT-LICENSE-MIT
         /[-_]#{COPYING_REGEX}/                       => 0.55,  # MIT-COPYING
         /\A#{OFL_REGEX}#{PREFERRED_EXT_REGEX}/       => 0.50,  # OFL.md
-        /\A#{OFL_REGEX}#{ANY_EXT_REGEX}/             => 0.45,  # OFL.textile
+        /\A#{OFL_REGEX}#{NONSPDX_EXT_REGEX}/         => 0.45,  # OFL.textile
         /\A#{OFL_REGEX}\z/                           => 0.40,  # OFL
         /\A#{PATENTS_REGEX}\z/                       => 0.35,  # PATENTS
-        /\A#{PATENTS_REGEX}#{ANY_EXT_REGEX}\z/       => 0.30,  # PATENTS.txt
+        /\A#{PATENTS_REGEX}#{NONSPDX_EXT_REGEX}\z/   => 0.30,  # PATENTS.txt
         //                                           => 0.00   # Catch all
       }.freeze
 

lib/licensee/project_files/package_manager_file.rb

@@ -10,21 +10,26 @@ class PackageManagerFile < Licensee::ProjectFiles::ProjectFile
 
       # Hash of Filename => [possible matchers]
       FILENAMES_EXTENSIONS = {
-        'DESCRIPTION' => [Matchers::Cran],
-        'dist.ini'    => [Matchers::DistZilla]
+        'DESCRIPTION'  => [Matchers::Cran],
+        'dist.ini'     => [Matchers::DistZilla],
+        'LICENSE.spdx' => [Matchers::Spdx]
+      }.freeze
+
+      FILENAMES_SCORES = {
+        'package.json' => 1.0,
+        'LICENSE.spdx' => 1.0,
+        'DESCRIPTION'  => 0.9,
+        'dist.ini'     => 0.8,
+        'bower.json'   => 0.75
       }.freeze
 
       def possible_matchers
         MATCHERS_EXTENSIONS[extension] || FILENAMES_EXTENSIONS[filename] || []
       end
 
       def self.name_score(filename)
-        return 1.0  if ['.gemspec', '.cabal'].include?(File.extname(filename))
-        return 1.0  if filename == 'package.json'
-        return 0.8  if filename == 'dist.ini'
-        return 0.9  if filename == 'DESCRIPTION'
-        return 0.75 if filename == 'bower.json'
-        0.0
+        return 1.0 if ['.gemspec', '.cabal'].include?(File.extname(filename))
+        FILENAMES_SCORES[filename] || 0.0
       end
 
       private

lib/licensee/project_files/project_file.rb

@@ -66,7 +66,7 @@ def license
       def copyright?
         return false unless is_a?(LicenseFile)
         return false unless matcher.is_a?(Matchers::Copyright)
-        filename =~ /\Acopyright(?:#{LicenseFile::ANY_EXT_REGEX})?\z/i
+        filename =~ /\Acopyright(?:#{LicenseFile::NONSPDX_EXT_REGEX})?\z/i
       end
     end
   end

script/diff

@@ -3,6 +3,7 @@
 # Example usage: cat LICENSE.md | be script/diff mit
 
 require 'licensee'
+require 'tmpdir'
 
 if STDIN.tty? || ARGV[0].to_s.empty?
   puts 'USAGE: [LICENSE_CONTENT] | script/diff [LICENSE_ID]'

spec/licensee/matchers/npm_bower_matcher_spec.rb

@@ -2,7 +2,8 @@
   let(:content) { '"license": "mit"' }
   let(:file) { Licensee::ProjectFiles::LicenseFile.new(content, 'LICENSE.txt') }
   let(:mit) { Licensee::License.find('mit') }
-  let(:no_license) { Licensee::License.find('no-license') }
+  let(:other) { Licensee::License.find('other') }
+
   subject { described_class.new(file) }
 
   it 'matches' do
@@ -42,7 +43,15 @@
     let(:content) { "'license': 'foo'" }
 
     it 'returns other' do
-      expect(subject.match).to eql(Licensee::License.find('other'))
+      expect(subject.match).to eql(other)
+    end
+  end
+
+  context 'a license expression' do
+    let(:content) { "'license': '(MIT OR Apache-2.0 OR AGPL-3.0+)'" }
+
+    it 'returns other' do
+      expect(subject.match).to eql(other)
     end
   end
 end

spec/licensee/matchers/spdx_matcher_spec.rb

@@ -0,0 +1,41 @@
+RSpec.describe Licensee::Matchers::Spdx do
+  let(:content) { 'PackageLicenseDeclared: MIT' }
+  let(:file) do
+    Licensee::ProjectFiles::LicenseFile.new(content, 'LICENSE.spdx')
+  end
+  let(:mit) { Licensee::License.find('mit') }
+  let(:other) { Licensee::License.find('other') }
+  subject { described_class.new(file) }
+
+  it 'matches' do
+    expect(subject.match).to eql(mit)
+  end
+
+  it 'has a confidence' do
+    expect(subject.confidence).to eql(90)
+  end
+
+  context 'no license field' do
+    let(:content) { 'foo: bar' }
+
+    it 'returns nil' do
+      expect(subject.match).to be_nil
+    end
+  end
+
+  context 'an unknown license' do
+    let(:content) { 'PackageLicenseDeclared: xyz' }
+
+    it 'returns other' do
+      expect(subject.match).to eql(other)
+    end
+  end
+
+  context 'a license expression' do
+    let(:content) { 'PackageLicenseDeclared: (MIT OR Apache-2.0)' }
+
+    it 'returns other' do
+      expect(subject.match).to eql(other)
+    end
+  end
+end

spec/licensee/project_files/license_file_spec.rb

@@ -109,11 +109,11 @@
 
     context 'any extension regex' do
       it 'matches .foo' do
-        expect(described_class::ANY_EXT_REGEX).to match('.foo')
+        expect(described_class::NONSPDX_EXT_REGEX).to match('.foo')
       end
 
       it 'does not match .md/foo' do
-        expect(described_class::ANY_EXT_REGEX).to_not match('.md/foo')
+        expect(described_class::NONSPDX_EXT_REGEX).to_not match('.md/foo')
       end
     end