license_file.rb in licensee-9.15.0

- old
+ new

@@ -8,12 +8,19 @@
       # List of extensions to give preference to
       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.freeze
+      LICENSE_EXT_REGEX = %r{\.(?!spdx|header)[^./]+\z}i.freeze
 
+      # Regex to match any extension except a few unlikely as license
+      # texts with complex filenames
+      OTHER_EXT_REGEX = %r{\.(?!xml|go|gemspec)[^./]+\z}i.freeze
+
+      # Regex to match any extension
+      ANY_EXT_REGEX = %r{\.[^./]+\z}i.freeze
+
       # Regex to match, LICENSE, LICENCE, unlicense, etc.
       LICENSE_REGEX = /(un)?licen[sc]e/i.freeze
 
       # Regex to match COPYING, COPYRIGHT, etc.
       COPYING_REGEX = /copy(ing|right)/i.freeze
@@ -24,26 +31,26 @@
       # BSD + PATENTS patent file
       PATENTS_REGEX = /patents/i.freeze
 
       # Hash of Regex => score with which to score potential license files
       FILENAME_REGEXES = {
-        /\A#{LICENSE_REGEX}\z/                       => 1.00,  # LICENSE
-        /\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}#{OTHER_EXT_REGEX}\z/     => 0.80,  # LICENSE.textile
-        /\A#{COPYING_REGEX}#{OTHER_EXT_REGEX}\z/     => 0.75,  # COPYING.textile
-        /\A#{LICENSE_REGEX}[-_]/                     => 0.70,  # LICENSE-MIT
-        /\A#{COPYING_REGEX}[-_]/                     => 0.65,  # COPYING-MIT
-        /\A\w+[-_]#{LICENSE_REGEX}/                  => 0.60,  # MIT-LICENSE-MIT
-        /\A\w+[-_]#{COPYING_REGEX}/                  => 0.55,  # MIT-COPYING
-        /\A#{OFL_REGEX}#{PREFERRED_EXT_REGEX}/       => 0.50,  # OFL.md
-        /\A#{OFL_REGEX}#{OTHER_EXT_REGEX}/           => 0.45,  # OFL.textile
-        /\A#{OFL_REGEX}\z/                           => 0.40,  # OFL
-        /\A#{PATENTS_REGEX}\z/                       => 0.35,  # PATENTS
-        /\A#{PATENTS_REGEX}#{OTHER_EXT_REGEX}\z/     => 0.30,  # PATENTS.txt
-        //                                           => 0.00   # Catch all
+        /\A#{LICENSE_REGEX}\z/                                => 1.00,  # LICENSE
+        /\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}#{LICENSE_EXT_REGEX}\z/            => 0.80,  # LICENSE.textile
+        /\A#{COPYING_REGEX}#{ANY_EXT_REGEX}\z/                => 0.75,  # COPYING.textile
+        /\A#{LICENSE_REGEX}[-_][^.]*#{OTHER_EXT_REGEX}?\z/    => 0.70,  # LICENSE-MIT
+        /\A#{COPYING_REGEX}[-_][^.]*#{OTHER_EXT_REGEX}?\z/    => 0.65,  # COPYING-MIT
+        /\A\w+[-_]#{LICENSE_REGEX}[^.]*#{OTHER_EXT_REGEX}?\z/ => 0.60,  # MIT-LICENSE-MIT
+        /\A\w+[-_]#{COPYING_REGEX}[^.]*#{OTHER_EXT_REGEX}?\z/ => 0.55,  # MIT-COPYING
+        /\A#{OFL_REGEX}#{PREFERRED_EXT_REGEX}/                => 0.50,  # OFL.md
+        /\A#{OFL_REGEX}#{OTHER_EXT_REGEX}/                    => 0.45,  # OFL.textile
+        /\A#{OFL_REGEX}\z/                                    => 0.40,  # OFL
+        /\A#{PATENTS_REGEX}\z/                                => 0.35,  # PATENTS
+        /\A#{PATENTS_REGEX}#{OTHER_EXT_REGEX}\z/              => 0.30,  # PATENTS.txt
+        //                                                    => 0.00   # Catch all
       }.freeze
 
       # CC-NC and CC-ND are not open source licenses and should not be
       # detected as CC-BY or CC-BY-SA which are 98%+ similar
       CC_FALSE_POSITIVE_REGEX = /
@@ -54,14 +61,14 @@
         [Matchers::Copyright, Matchers::Exact, Matchers::Dice]
       end
 
       def attribution
         @attribution ||= begin
-          return unless copyright? || license.content&.include?('[fullname]')
-
-          matches = Matchers::Copyright::REGEX
-                    .match(content_without_title_and_version)
-          matches[0] if matches
+          if copyright? || license.content&.include?('[fullname]')
+            matches = Matchers::Copyright::REGEX
+                      .match(content_without_title_and_version)
+            matches[0] if matches
+          end
         end
       end
 
       # Is this file likely to result in a creative commons false positive?
       def potential_false_positive?