From 4551c356c5975da9fc86c204f4aac4dee969feb9 Mon Sep 17 00:00:00 2001
From: mo khan <mo.khan@gmail.com>
Date: Tue, 22 Oct 2019 17:10:03 -0600
Subject: Extract spdx id from known source urls

---
 lib/license/management/repository.rb | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

(limited to 'lib')

diff --git a/lib/license/management/repository.rb b/lib/license/management/repository.rb
index fdff30d..ed2519f 100644
--- a/lib/license/management/repository.rb
+++ b/lib/license/management/repository.rb
@@ -5,6 +5,11 @@ module License
     class Repository
       include Loggable
       include Verifiable
+      KNOWN_SOURCES = [
+        'licenses.nuget.org',
+        'opensource.org',
+        'www.opensource.org',
+      ].freeze
 
       def initialize(
         compatibility_path: License::Management.root.join('normalized-licenses.yml'),
@@ -34,15 +39,15 @@ module License
             'name' => data['name'],
             'url' => data['seeAlso'][-1]
           }
-        else
-          log_info("Could not find license `#{id}` in SPDX")
-          nil
         end
       end
 
       def id_for(license)
         ids = compatibility_data['ids']
-        ids[license.send(:short_name)] || ids[license.url] || known_sources(license.url)
+        ids[license.send(:short_name)] ||
+          ids[license.url] ||
+          known_sources(license.send(:short_name)) ||
+          known_sources(license.url)
       end
 
       # When `license_finder` is unable to determine the license it will use the full
@@ -55,7 +60,7 @@ module License
       end
 
       def generate_item_for(license)
-        log_info("detected unknown license named `#{license.send(:short_name)}`")
+        log_info("detected unknown license named `#{license.send(:short_name)}`:`#{license.url}`")
         name = take_first_line_from(license.name)
         {
           'id' => name.downcase,
@@ -76,13 +81,14 @@ module License
       end
 
       def known_sources(url)
+        return if blank?(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)
-
+        return unless KNOWN_SOURCES.include?(uri.host.downcase)
         uri.path.split('/')[-1]
-      rescue
+      rescue => error
+        log_info(error)
         nil
       end
     end
-- 
cgit v1.2.3