better no license matching

lib/licensee/license_file.rb

@@ -4,7 +4,13 @@ class Licensee
 
     def initialize(blob)
       @blob = blob
+    end
+
+    def similarity(other)
       blob.hashsig(Rugged::Blob::HashSignature::WHITESPACE_SMART)
+      other.hashsig ? blob.similarity(other.hashsig) : 0
+    rescue Rugged::InvalidError
+      0
     end
 
     # Raw file contents

lib/licensee/matchers/copyright_matcher.rb

@@ -1,11 +1,7 @@
 class Licensee
   class CopyrightMatcher < Matcher
 
-    REGEX = /^(©|\(c\)|Copyright) \d{4}(.*)\n?$/i
-
-    def no_license
-      @no_license ||= Licensee::Licenses.find("no-license")
-    end
+    REGEX = /\A(Copyright )?(©|\(c\)) \d{4}(.*)\n?\z/i
 
     def match
       no_license if file.content_normalized =~ REGEX
@@ -14,5 +10,11 @@ class Licensee
     def confidence
       100
     end
+
+    private
+
+    def no_license
+      @no_license ||= Licensee::Licenses.find("no-license")
+    end
   end
 end

lib/licensee/matchers/git_matcher.rb

@@ -12,11 +12,11 @@ class Licensee
     private
 
     def matches
-      @matches ||= Licensee.licenses.map { |l| [l, similarity(l)] }.select { |l,sim| sim > 0 }
+      @matches ||= Licensee.licenses.map { |l| [l, file.similarity(l)] }.select { |l,sim| sim > 0 }
     end
 
     def similarity(other)
-      other.hashsig ? file.blob.similarity(other.hashsig) : 0
+      file.similarity(other)
     end
 
     # Pulled out for easier testing