Parse spdx id from known sources

1 files changed, 17 insertions, 3 deletions

diff --git a/lib/license/management/repository.rb b/lib/license/management/repository.rb
index 0c428dd..fdff30d 100644
--- a/lib/license/management/repository.rb
+++ b/lib/license/management/repository.rb
@@ -25,6 +25,8 @@ module License
       attr_reader :spdx_data, :compatibility_data
 
       def spdx_data_for(id)
+        return if blank?(id)
+
         data = spdx_data[id]
         if data
           {
@@ -40,7 +42,7 @@ module License
 
       def id_for(license)
         ids = compatibility_data['ids']
-        ids[license.send(:short_name)] || ids[license.url]
+        ids[license.send(:short_name)] || ids[license.url] || known_sources(license.url)
       end
 
       # When `license_finder` is unable to determine the license it will use the full
@@ -54,9 +56,10 @@ module License
 
       def generate_item_for(license)
         log_info("detected unknown license named `#{license.send(:short_name)}`")
+        name = take_first_line_from(license.name)
         {
-          'id' => 'unknown',
-          'name' => take_first_line_from(license.name),
+          'id' => name.downcase,
+          'name' => name,
           'url' => present?(license.url) ? license.url : ''
         }
       end
@@ -71,6 +74,17 @@ module License
           memo
         end
       end
+
+      def known_sources(url)
+        return unless url =~ /\A#{::URI::DEFAULT_PARSER.make_regexp(['http', 'https'])}\z/
+
+        uri = URI.parse(url)
+        return unless ['opensource.org', 'licenses.nuget.org'].include?(uri.host)
+
+        uri.path.split('/')[-1]
+      rescue
+        nil
+      end
     end
   end
 end