Consider contractors and students as external in metrics. (#79)

Signed-off-by: dblock <[email protected]>
opensearch-project · Mar 25, 2024 · ad44392 · ad44392
1 parent 25c77e2
commit ad44392
Show file tree

Hide file tree

Showing 9 changed files with 109 additions and 71 deletions.
diff --git a/bin/commands/maintainers.rb b/bin/commands/maintainers.rb
@@ -17,32 +17,34 @@ class Commands
                     GitHub::Organization.new(options.merge(org: options['org'] || 'opensearch-project')).repos
                   end
           maintainers = repos.maintainers(dt)
-          puts "As of #{dt || Date.today}, #{repos.count} repos have #{maintainers.unique_count} maintainers, where #{maintainers.external_unique_percent}% (#{maintainers.external_unique_count}/#{maintainers.unique_count}) are external."
-          puts "A total of #{repos.external_maintainers_percent}% (#{repos.external_maintained_size}/#{repos.count}) of repos have at least one of #{maintainers.external_unique_count} external maintainers."
-          puts "\n# Maintainers\n"
-          puts "unique: #{maintainers.unique_count}"
-          maintainers.each_pair do |bucket, logins|
-            puts "#{bucket}: #{logins.size} (#{logins.map(&:to_s).join(', ')})"
-          end
+          puts "As of #{dt || Date.today}, #{repos.count} repos have #{maintainers.unique_count} maintainers, where #{maintainers.all_external_unique_percent}% (#{maintainers.all_external_unique_count}/#{maintainers.unique_count}) are external."
+          puts "A total of #{repos.all_external_maintainers_percent}% (#{repos.all_external_maintained_size}/#{repos.count}) of repos have at least one of #{maintainers.all_external_unique_count} external maintainers."
+
           puts "\n# External Maintainers\n"
-          repos.maintained[:external]&.sort_by(&:name)&.each do |repo|
-            puts "#{repo.html_url}: #{repo.maintainers[:external]} (#{repo.maintainers.external_unique_percent}%, #{repo.maintainers.external_unique_count}/#{repo.maintainers.unique_count})"
+          repos.externally_maintained.each do |repo|
+            puts "#{repo.html_url}: #{repo.maintainers.all_external} (#{repo.maintainers.all_external_unique_percent}%, #{repo.maintainers.all_external_unique_count}/#{repo.maintainers.unique_count})"
           end
 
-          puts "\n# Student Maintainers\n"
-          repos.maintained[:students]&.sort_by(&:name)&.each do |repo|
-            puts "#{repo.html_url}: #{repo.maintainers[:students]}"
-          end
+          # GitHub::Maintainers::ALL_EXTERNAL.each do |bucket|
+          #   repos.maintained[bucket]&.sort_by(&:name)&.each do |repo|
+          #     puts "#{repo.html_url}: #{repo.maintainers.all_external} (#{repo.maintainers.all_external_unique_percent}%, #{repo.maintainers.all_external_unique_count}/#{repo.maintainers.unique_count})"
+          #   end
+          # end
 
-          puts "\n# Contractor Maintainers\n"
-          repos.maintained[:contractors]&.sort_by(&:name)&.each do |repo|
-            puts "#{repo.html_url}: #{repo.maintainers[:contractors]}"
-          end
+          # puts "\n# All Maintainers\n"
+          # puts "unique: #{maintainers.unique_count}"
+          # maintainers.each_pair do |bucket, logins|
+          #   puts "#{bucket}: #{logins.size} (#{logins.map(&:to_s).join(', ')})"
+          # end
 
-          puts "\n# Unknown Maintainers\n"
-          repos.maintained[:unknown]&.sort_by(&:name)&.each do |repo|
-            puts "#{repo.html_url}: #{repo.maintainers[:unknown]}"
-          end
+          # %i[external students contractors unknown].each do |bucket|
+          #   next unless repos.maintained[bucket]&.any?
+
+          #   puts "\n# #{bucket.capitalize} Maintainers\n"
+          #   repos.maintained[bucket]&.sort_by(&:name)&.each do |repo|
+          #     puts "#{repo.html_url}: #{repo.maintainers[bucket]}"
+          #   end
+          # end
         end
       end
 

diff --git a/bin/commands/members.rb b/bin/commands/members.rb
@@ -7,7 +7,7 @@ class Commands
       g.desc 'Check GitHub affiliation information for contributors.'
       g.command 'check' do |c|
         c.action do |_global_options, _options, _args|
-          GitHub::User.wrap(GitHub::Data.members).each do |contributor|
+          GitHub::User.wrap(GitHub::Data.members_data).each do |contributor|
             unless contributor.member?
               puts "#{contributor.login}: #{[contributor.company,
                                              contributor.bio].compact.join(' ')}"

diff --git a/bin/commands/org.rb b/bin/commands/org.rb
@@ -20,8 +20,8 @@ class Commands
           org = GitHub::Organization.new(options)
           puts "org: #{org.name}"
           puts "members: #{org.members.count}"
-          puts "missing in data/users/members.txt: #{(org.members.logins - GitHub::Data.members).join(' ')}"
-          puts "no longer members: #{(GitHub::Data.members - org.members.logins).join(' ')}"
+          puts "missing in data/users/members.txt: #{(org.members.logins - GitHub::Data.members_data).join(' ')}"
+          puts "no longer members: #{(GitHub::Data.members_data - org.members.logins).join(' ')}"
         end
       end
 

diff --git a/lib/github/buckets.rb b/lib/github/buckets.rb
@@ -35,11 +35,17 @@ def all_humans
     end
 
     def all_members
-      members.to_a + contractors.to_a
+      members.to_a
     end
 
     def all_external
-      external.to_a + students.to_a
+      all = []
+      Maintainers::ALL_EXTERNAL.each do |bucket|
+        next unless buckets[bucket]
+
+        all.concat(buckets[bucket].to_a)
+      end
+      all
     end
 
     def all_external_percent
@@ -56,15 +62,15 @@ def percent
     end
 
     def self.bucket(username)
-      if GitHub::Data.members.include?(username.to_s)
+      if GitHub::Data.members_data.include?(username.to_s)
         :members
-      elsif GitHub::Data.contractors.include?(username.to_s)
+      elsif GitHub::Data.contractors_data.include?(username.to_s)
         :contractors
-      elsif GitHub::Data.college_contributors.include?(username.to_s)
+      elsif GitHub::Data.students_data.include?(username.to_s)
         :students
-      elsif GitHub::Data.external_users.include?(username.to_s)
+      elsif GitHub::Data.external_data.include?(username.to_s)
         :external
-      elsif GitHub::Data.bots.include?(username.to_s)
+      elsif GitHub::Data.bots_data.include?(username.to_s)
         :bots
       else
         :unknown

diff --git a/lib/github/contributors.rb b/lib/github/contributors.rb
@@ -10,23 +10,23 @@ def initialize(arr)
     end
 
     def humans
-      reject { |item| item.type == 'Bot' || GitHub::Data.bots.include?(item.to_s) }
+      reject { |item| item.type == 'Bot' || GitHub::Data.bots_data.include?(item.to_s) }
     end
 
     def all
       humans
     end
 
     def self.bucket(username)
-      if GitHub::Data.members.include?(username.to_s)
+      if GitHub::Data.members_data.include?(username.to_s)
         :members
-      elsif GitHub::Data.contractors.include?(username.to_s)
+      elsif GitHub::Data.contractors_data.include?(username.to_s)
         :contractors
-      elsif GitHub::Data.students.include?(username.to_s)
-        :external # :students
-      elsif GitHub::Data.external_users.include?(username.to_s)
+      elsif GitHub::Data.students_data.include?(username.to_s)
+        :students
+      elsif GitHub::Data.external_data.include?(username.to_s)
         :external
-      elsif GitHub::Data.bots.include?(username.to_s)
+      elsif GitHub::Data.bots_data.include?(username.to_s)
         :bots
       else
         :unknown

diff --git a/lib/github/data.rb b/lib/github/data.rb
@@ -26,48 +26,52 @@ def data
         DATA
       end
 
-      def backports
-        @backports ||= load_list(BACKPORTS)
+      def backports_data
+        @backports_data ||= load_list(BACKPORTS)
       end
 
-      def members
-        @members ||= load_list(MEMBERS)
+      def members_data
+        @members_data ||= load_list(MEMBERS)
       end
 
-      def contractors
-        @contractors ||= load_list(CONTRACTORS)
+      def contractors_data
+        @contractors_data ||= load_list(CONTRACTORS)
       end
 
-      def external_users
-        @external_users ||= load_list(EXTERNAL)
+      def external_data
+        @external_data ||= load_list(EXTERNAL)
       end
 
-      def students
-        @students ||= load_list(STUDENTS)
+      def students_data
+        @students_data ||= load_list(STUDENTS)
       end
 
-      def check_dups!
-        %i[members contractors students external_users].combination(2).each do |l, r|
-          send(l).intersection(send(r)).each do |user|
-            warn "WARNING: #{user} is found in both #{l} and #{r}"
-          end
-        end
+      def bots_data
+        @bots_data ||= load_list(BOTS)
       end
 
-      def bots
-        @bots ||= load_list(BOTS)
+      def companies_data
+        @companies_data ||= load_list(COMPANIES)
       end
 
-      def companies
-        @companies ||= load_list(COMPANIES)
+      def all_members_data
+        members_data
       end
 
-      def all_members
-        members + contractors
+      def check_dups!
+        %i[members_data contractors_data students_data external_data].combination(2).each do |l, r|
+          send(l).intersection(send(r)).each do |user|
+            warn "WARNING: #{user} is found in both #{l} and #{r}"
+          end
+        end
       end
 
-      def all_external
-        external_users + students
+      def all_external_data
+        all = []
+        Maintainers::ALL_EXTERNAL.each do |bucket|
+          all.concat(send("#{bucket}_data"))
+        end
+        all
       end
     end
   end

diff --git a/lib/github/maintainers.rb b/lib/github/maintainers.rb
@@ -4,6 +4,8 @@ module GitHub
   class Maintainers < Array
     include GitHub::Buckets
 
+    ALL_EXTERNAL = %i[contractors external students].freeze
+
     def buckets
       @buckets ||= begin
         buckets = {}
@@ -16,18 +18,26 @@ def buckets
       end
     end
 
-    def external_unique_percent
+    def all_external
+      ALL_EXTERNAL.map do |bucket|
+        buckets[bucket]
+      end.flatten.compact.uniq
+    end
+
+    def all_external_unique_percent
       return 0 unless unique_count
 
-      ((external_unique_count.to_f / unique_count) * 100).to_i
+      ((all_external_unique_count.to_f / unique_count) * 100).to_i
     end
 
     def unique_count
       buckets.values.map(&:size).sum
     end
 
-    def external_unique_count
-      (buckets[:external]&.size || 0) + (buckets[:students]&.size || 0)
+    def all_external_unique_count
+      ALL_EXTERNAL.map do |bucket|
+        buckets[bucket]&.size || 0
+      end.sum
     end
 
     def each_pair(&_block)

diff --git a/lib/github/pull_requests.rb b/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) } || project_website_authors?(pr)
+        pr.user.type == 'Bot' || GitHub::Data.backports_data.any? { |b| pr.title&.downcase&.include?(b) } || project_website_authors?(pr)
       end
     end
 

diff --git a/lib/github/repos.rb b/lib/github/repos.rb
@@ -43,14 +43,30 @@ def maintained
       end
     end
 
-    def external_maintained_size
-      (maintained[:external]&.size || 0) + (maintained[:students]&.size || 0)
+    def externally_maintained
+      @externally_maintained ||= begin
+        all = Set.new
+        Maintainers::ALL_EXTERNAL.each do |bucket|
+          each do |repo|
+            next unless ((repo.maintainers & maintainers[bucket]) || []).any?
+
+            all.add(repo)
+          end
+        end
+        all
+      end
+    end
+
+    def all_external_maintained_size
+      Maintainers::ALL_EXTERNAL.map do |bucket|
+        maintained[bucket]&.size || 0
+      end.sum
     end
 
-    def external_maintainers_percent
+    def all_external_maintainers_percent
       return 0 unless any?
 
-      (((maintained[:external]&.size.to_f + maintained[:students]&.size.to_f) / size) * 100).to_i
+      (all_external_maintained_size.to_f * 100 / size).to_i
     end
   end
 end