Exclude website authors that skews numbers.

Signed-off-by: dblock <[email protected]>

bin/commands/pr.rb

@@ -29,7 +29,7 @@ class Commands
 
           if !options['ignore-unknown'] && prs.contributors[:unknown]&.any?
             puts 'Add the following users to either data/users/members.txt, external.txt, students.txt or contractors.txt and re-run.'
-            prs.contributors[:unknown].keys.take(10).each do |user|
+            prs.contributors[:unknown].keys.each do |user|
               puts user
               system "open https://github.com/#{user}"
             end

lib/github/pull_requests.rb

@@ -29,7 +29,7 @@ def page(options)
       raise 'There are 1000+ PRs returned from a single query, reduce --page.' if data.size >= 1000
 
       data.reject do |pr|
-        pr.user.type == 'Bot' || GitHub::Data.backports.any? { |b| pr.title&.downcase&.include?(b) }
+        pr.user.type == 'Bot' || GitHub::Data.backports.any? { |b| pr.title&.downcase&.include?(b) } || project_website_authors?(pr)
       end
     end
 
@@ -44,5 +44,15 @@ def query(options = {})
         ].compact
       ).compact.join(' ')
     end
+
+    # exclude a high number of misleading contributions from a student program
+    # modifying https://github.com/opensearch-project/project-website/commits/main/_authors
+    def project_website_authors?(pr)
+      return false unless pr.repository_url == 'https://api.github.com/repos/opensearch-project/project-website'
+
+      repo = pr.repository_url.split('/')[4, 2].join('/')
+      files = $github.pull_request_files(repo, pr.number)
+      files.all? { |file| file.filename.start_with?('_authors/') }
+    end
   end
 end