licensee 8.4.0 → 8.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0ea7d783ce6fc8e59946877a34fc32d84b01f449
-  data.tar.gz: 65142dc4101a66360d080b7eede89977d5ca10c6
+  metadata.gz: '09da33c25905df581cb047d1eed3c138ede5f440'
+  data.tar.gz: c0d34f622f3f6dde90f32c859b221b6b4af02bce
 SHA512:
-  metadata.gz: 9b93a5da892c334a4e3bce03be2ef7342ffda08ae64b6db412216f691af53ee9d8a7d54596c9aadaff6cd6d02038c04da21f4d4e98af222283788fdb77a340cc
-  data.tar.gz: d066ee5ed370b2c840cd5282f19fccf525f26cde2469a6205039ab886b5b1f2ca5e63e9ac88fc2888fa1c4657cda2b50d2f57f4da79632ac22ad4f96779bc33a
+  metadata.gz: 0c253e00b53aa258a70e80fad5a94454ff9ae080d9911f20e999e09eba9db14b306052dbf04b0e66de8ae2256ee87d81b4d458e4fa5317bd84e129bfe5ef2084
+  data.tar.gz: 7d0e97bbc08a7d687b6011690e8d346d76f97a6c1d9ac5ee2cede3400771c480c05950e904f5f632d01ff0bed0e699233cb3961f8e79f75ed4a91b7265fbac2c

data/bin/licensee

@@ -4,23 +4,33 @@ require_relative '../lib/licensee'
 
 path = ARGV[0] || Dir.pwd
 
-def print_file(license_file)
-  if license_file
-    puts "License file: #{license_file.filename}"
-    puts "Attribution: #{license_file.attribution}" if license_file.attribution
-  end
+def format_percent(float)
+  "#{format('%.2f', float)}%"
 end
 
-def print_evaluation(file)
-  if file
-    puts "License: #{file.license ? file.license.meta['title'] : 'no license'}"
-    puts "Confidence: #{format('%.2f', file.confidence)}%" if file.confidence
-    puts "Method: #{file.matcher.class}" if file.matcher
+project = Licensee.project(path, detect_packages: true, detect_readme: true)
+license_file = project.license_file
+matched_file = project.matched_file
+
+if license_file
+  puts "License file: #{license_file.filename}"
+  puts "Attribution: #{license_file.attribution}" if license_file.attribution
+end
+
+if matched_file
+  if matched_file.license
+    puts "License: #{matched_file.license.meta['title']}"
+    puts "Confidence: #{format_percent(matched_file.confidence)}" if matched_file.confidence
+    puts "Method: #{matched_file.matcher.class}" if matched_file.matcher
   else
-    puts 'Unknown'
+    puts 'License: Not detected'
+    puts
+    puts "Here's the closest licenses:"
+    matcher = Licensee::Matchers::Dice.new(matched_file)
+    matcher.licenses_by_similiarity[0...3].each do |license, similarity|
+      puts "* #{license.meta['spdx-id']} similarity: #{format_percent(similarity)}"
+    end
   end
+else
+  puts 'Unknown'
 end
-
-project = Licensee.project(path, detect_packages: true, detect_readme: true)
-print_file(project.license_file)
-print_evaluation(project.matched_file)

data/lib/licensee/content_helper.rb

@@ -48,6 +48,7 @@ module Licensee
       @content_normalized ||= begin
         content_normalized = content.downcase.strip
         content_normalized.gsub!(/^#{Matchers::Copyright::REGEX}$/i, '')
+        content_normalized.gsub!(/[=-]{4,}/, '') # Strip HRs from MPL
         content_normalized.tr("\n", ' ').squeeze(' ').strip
       end
     end

data/lib/licensee/license.rb

@@ -140,11 +140,10 @@ module Licensee
     # Raw content of license file, including YAML front matter
     def raw_content
       return if pseudo_license?
-      @raw_content ||= if File.exist?(path)
-        File.read(path, encoding: 'utf-8')
-      else
+      unless File.exist?(path)
         raise Licensee::InvalidLicense, "'#{key}' is not a valid license key"
       end
+      @raw_content ||= File.read(path, encoding: 'utf-8')
     end
 
     def parts

data/lib/licensee/matchers/dice_matcher.rb

@@ -10,16 +10,10 @@ module Licensee
       # Return the first potential license that is more similar
       # than the confidence threshold
       def match
-        return @match if defined? @match
-        matches = potential_licenses.map do |license|
-          similarity = license.similarity(file)
-          [license, similarity] if similarity >= Licensee.confidence_threshold
-        end
-        matches.compact!
-        @match = if matches.empty?
+        @match ||= if matches.empty?
           nil
         else
-          matches.max_by { |_l, sim| sim }.first
+          matches.first[0]
         end
       end
 
@@ -34,6 +28,19 @@ module Licensee
         end
       end
 
+      def licenses_by_similiarity
+        @licenses_by_similiarity ||= begin
+          licenses = potential_licenses.map { |l| [l, l.similarity(file)] }
+          licenses.sort_by { |_, similarity| similarity }.reverse
+        end
+      end
+
+      def matches
+        @matches ||= licenses_by_similiarity.select do |_, similarity|
+          similarity >= Licensee.confidence_threshold
+        end
+      end
+
       # Confidence that the matched license is a match
       def confidence
         @confidence ||= match ? file.similarity(match) : 0

data/lib/licensee/version.rb

@@ -1,3 +1,3 @@
 module Licensee
-  VERSION = '8.4.0'.freeze
+  VERSION = '8.5.0'.freeze
 end

data/test/fixtures/mpl-without-hrs.git/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/master

data/test/fixtures/mpl-without-hrs.git/config

@@ -0,0 +1,5 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = true
+	ignorecase = true

data/test/fixtures/mpl-without-hrs.git/refs/heads/master

@@ -0,0 +1 @@
+baf5649b61fac4ed64ec2949fe5cd616b85a8298

data/test/licensee/matchers/test_dice_matcher.rb

@@ -4,6 +4,9 @@ class TestLicenseeDiceMatchers < Minitest::Test
   def setup
     text = license_from_path(Licensee::License.find('mit').path)
     @mit = Licensee::Project::LicenseFile.new(text)
+
+    text = license_from_path(Licensee::License.find('gpl-2.0').path)
+    @gpl = Licensee::Project::LicenseFile.new(text)
   end
 
   def concat_licenses(*args)
@@ -25,6 +28,33 @@ class TestLicenseeDiceMatchers < Minitest::Test
     text = concat_licenses('mit', 'gpl-2.0')
     license = Licensee::Project::LicenseFile.new(text)
     matcher = Licensee::Matchers::Dice.new(license)
-    refute matcher.match
+    if matcher.match
+      msg = "Expected no-license, got #{matcher.match.key}"
+      msg << " (#{matcher.match.similarity(license).round(2)}% similar)"
+    end
+    refute matcher.match, msg
+  end
+
+  should 'build the list of licenses by similarity' do
+    matcher = Licensee::Matchers::Dice.new(@gpl)
+
+    match = matcher.licenses_by_similiarity.first
+    assert_equal Licensee::License, match[0].class
+    assert_equal 'gpl-2.0', match[0].key
+    assert_equal 100.0, match[1]
+
+    match = matcher.licenses_by_similiarity[1]
+    assert_equal 'lppl-1.3c', match[0].key
+    assert_equal 49.52, match[1].round(2)
+  end
+
+  should 'build the list of matches' do
+    matcher = Licensee::Matchers::Dice.new(@gpl)
+
+    assert_equal 1, matcher.matches.count
+    match = matcher.matches.first
+    assert_equal Licensee::License, match[0].class
+    assert_equal 'gpl-2.0', match[0].key
+    assert_equal 100.0, match[1]
   end
 end

data/test/licensee/test_content_helper.rb

@@ -8,7 +8,8 @@ class ContentHelperTestHelper
 Copyright 2016 Ben Balter
 
 The made
-up license.
+up  license.
+-----------
   EOS
 
   def initialize(content = nil)
@@ -21,8 +22,32 @@ class TestLicenseeContentHelper < Minitest::Test
     @helper = ContentHelperTestHelper.new
   end
 
-  should 'normalize the content' do
-    assert_equal 'the made up license.', @helper.content_normalized
+  context 'normalizing' do
+    should 'strips copyright' do
+      refute_match('Copyright', @helper.content_normalized)
+      refute_match('Ben Balter', @helper.content_normalized)
+    end
+
+    should 'downcases' do
+      refute_match('The', @helper.content_normalized)
+      assert_match('the', @helper.content_normalized)
+    end
+
+    should 'strips HRs' do
+      refute_match('---', @helper.content_normalized)
+    end
+
+    should 'squeeze whitespace' do
+      refute_match('  ', @helper.content_normalized)
+    end
+
+    should 'strips whitespace' do
+      refute_match(/\n/i, @helper.content_normalized)
+    end
+
+    should 'normalize the content' do
+      assert_equal 'the made up license.', @helper.content_normalized
+    end
   end
 
   should 'generate the hash' do

data/test/licensee/test_project.rb

@@ -77,6 +77,11 @@ class TestLicenseeProject < Minitest::Test
         project = make_project 'lgpl.git'
         assert_equal 'lgpl-3.0', project.license.key
       end
+
+      should 'detect the MPL even with HRs removed' do
+        project = make_project 'mpl-without-hrs.git'
+        assert_equal 'mpl-2.0', project.license.key
+      end
     end
   end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: licensee
 version: !ruby/object:Gem::Version
-  version: 8.4.0
+  version: 8.5.0
 platform: ruby
 authors:
 - Ben Balter
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-09-15 00:00:00.000000000 Z
+date: 2016-09-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rugged
@@ -173,6 +173,12 @@ files:
 - test/fixtures/mit-with-redundant-title/mit.txt
 - test/fixtures/mit-without-title-rewrapped/mit.txt
 - test/fixtures/mit-without-title/mit.txt
+- test/fixtures/mpl-without-hrs.git/HEAD
+- test/fixtures/mpl-without-hrs.git/config
+- test/fixtures/mpl-without-hrs.git/objects/45/26aa2d1d78ea20fb0faf59f2360a68fa897774
+- test/fixtures/mpl-without-hrs.git/objects/ba/f5649b61fac4ed64ec2949fe5cd616b85a8298
+- test/fixtures/mpl-without-hrs.git/objects/be/2cc4dfb609fb6c38f6365ec345bded3350dd63
+- test/fixtures/mpl-without-hrs.git/refs/heads/master
 - test/fixtures/named-license-file-prefix.git/HEAD
 - test/fixtures/named-license-file-prefix.git/config
 - test/fixtures/named-license-file-prefix.git/objects/64/3983d3f82ecc2a7d8e4227946220ebffd477d2