yiffspace 0.0.22 → 0.0.23

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 55b429503eee52d6ce298af0e2ce3e65327f630c6f7fc0114e58d49a324c2fd3
-  data.tar.gz: 017a1eaaf6a5ee571fa8e29cc3f7a2da617d1b6de78f9d5d834bb248a9212326
+  metadata.gz: d7568ca2851a68d6ec6e5a66588f514bf6b95597eb8ed9fba59b6e3c204668e3
+  data.tar.gz: 29f61b714e981320ea89c3ea179334a0bbc2ae3fb14db8fc25129e135cb54935
 SHA512:
-  metadata.gz: 62391e56655903180b26837749b124f5c4a7f6ac4180815be23bcf83295f221af0f6433d824664caf8f1c10f4d81f0074630e66b3bf956c82472eecf9e4a2ea8
-  data.tar.gz: 86671d5ef0e7c213ae10721e9024f0dafc590f4b464b2311c549e666f87d2368838efa56e54ead37d16e17d1918ad465b611b3df25f3130c9f2675ed4d49dda5
+  metadata.gz: 4f214f02c6c540cfe1f9d64eb13c457d3ff2d37d6606ecd0e7656e630880103d8d5744fb04327d822e29d8c34b43fac42372f4c62ed89dbe54c34f32efa9e811
+  data.tar.gz: 07d6de0d59c2700a5543b8b1d5cf3c9691af11a5fe9696b03fdbb293a169959c07f0b3e2a6f29116c231dc313736b57da77a89a435a4be6a43645b1f2147ab33

data/lib/tasks/yiff_space_tasks.rake

@@ -1,6 +1,41 @@
 # frozen_string_literal: true
 
-# desc "Explaining what the task does"
-# task :yiffspace do
-#   # Task goes here
-# end
+namespace(:yiff_space) do
+  desc("Report and interactively remove duplicate/orphaned Logto users left behind by " \
+       "get_or_create_user races. Prompts for confirmation before deleting each one. " \
+       "Pass AUTH_CLIENT=name to use a registered auth client other than :default " \
+       "(e.g. AUTH_CLIENT=yiffyapi_manage).")
+  task(dedupe_users: :environment) do
+    client_name = (ENV["AUTH_CLIENT"] || YiffSpace::Auth::DEFAULT_CLIENT_NAME).to_sym
+    management = YiffSpace::Auth[client_name].logto_management
+
+    puts("Scanning users via auth client #{client_name.inspect}...")
+    result = YiffSpace::Utils::UserDeduper.scan(management)
+
+    puts("\n#{result[:conflicts].size} discord id(s) with more than one linked user (ambiguous - not handled by this task, review manually):")
+    result[:conflicts].each do |discord_id, group|
+      puts("  discordId=#{discord_id}: #{group.map { |u| "#{u.id} (#{u.name})" }.join(', ')}")
+    end
+
+    puts("\n#{result[:unresolved_orphans].size} orphaned user(s) with no matching linked account (not handled by this task, review manually):")
+    result[:unresolved_orphans].each do |user|
+      puts("  #{user.id} (name=#{user.name}, avatar=#{user.data.avatar})")
+    end
+
+    puts("\n#{result[:deletable_orphans].size} confirmed duplicate(s) - orphan left by a losing get_or_create_user race:")
+    if result[:deletable_orphans].empty?
+      puts("  none found")
+    else
+      result[:deletable_orphans].each do |orphan, keeper|
+        print("  delete #{orphan.id} (name=#{orphan.name}, avatar=#{orphan.data.avatar}), keeping #{keeper.id} (name=#{keeper.name}) for discordId=#{YiffSpace::Utils::UserDeduper.discord_id_for(keeper)}? [y/N] ")
+        answer = $stdin.gets&.strip&.downcase
+        if answer == "y"
+          management.delete_user(orphan.id)
+          puts("    deleted #{orphan.id}")
+        else
+          puts("    skipped")
+        end
+      end
+    end
+  end
+end

data/lib/yiffspace/logto_management_client.rb

@@ -75,6 +75,35 @@ module YiffSpace
       get_user(id) || create_user(id)
     end
 
+    # Pages through every user in the tenant. Used by maintenance tasks (e.g. the
+    # dedupe_users rake task) - there's no discordId to filter by when scanning the
+    # whole user base for duplicates/orphans.
+    def list_users(page_size: 100)
+      users = []
+      page = 1
+      loop do
+        response = HTTParty.get("#{auth.server_url}/api/users", {
+          headers: { "Authorization" => "Bearer #{get_token}" },
+          query:   { page: page, page_size: page_size },
+        })
+        raise("failed to list users: #{response.code} #{response.message}\n#{response.parsed_response.inspect}") if response.code != 200 || !response.parsed_response.is_a?(Array)
+
+        batch = response.parsed_response
+        users.concat(batch.map { |u| Auth::ApiUser.new(u) })
+        break if batch.length < page_size
+
+        page += 1
+      end
+      users
+    end
+
+    def delete_user(logto_id)
+      response = HTTParty.delete("#{auth.server_url}/api/users/#{logto_id}", { headers: { "Authorization" => "Bearer #{get_token}" } })
+      return true if [204, 404].include?(response.code)
+
+      raise("failed to delete user #{logto_id}: #{response.code} #{response.message}\n#{response.parsed_response.inspect}")
+    end
+
     def get_user_by_id(logto_id)
       response = HTTParty.get("#{auth.server_url}/api/users/#{logto_id}", { headers: { "Authorization" => "Bearer #{get_token}" } })
       return nil if response.code == 404

data/lib/yiffspace/utils/user_deduper.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module YiffSpace
+  module Utils
+    # get_or_create_user is check-then-act (query Logto, then create) with no locking
+    # of its own - two concurrent calls for a discord id that doesn't exist yet will
+    # both create a user, and the loser's discord-identity-attach step fails once the
+    # winner's identity claims it first. That leaves an identity-less "orphan" user
+    # behind in Logto. This scans the whole user base to find those orphans (and any
+    # other duplicate/ambiguous state) so they can be cleaned up.
+    module UserDeduper
+      # create_user sets a new user's avatar to the discord CDN URL for the discord id
+      # it was created for, before the identity-attach step runs - so even an orphan
+      # that never got its identity attached still reveals which discord id it was
+      # meant for.
+      AVATAR_ID_PATTERN = %r{cdn\.discordapp\.com/avatars/(\d+)/}
+
+      module_function
+
+      def discord_id_for(user)
+        user.data.identities&.discord&.userId || avatar_discord_id_for(user)
+      end
+
+      def avatar_discord_id_for(user)
+        match = user.data.avatar.to_s.match(AVATAR_ID_PATTERN)
+        match && match[1]
+      end
+
+      def linked?(user)
+        user.data.identities&.discord&.userId.present?
+      end
+
+      # Returns a hash with:
+      # - :conflicts - discord id => users, for ids where more than one user has the
+      #   identity actually attached. Ambiguous; always needs manual review, never
+      #   auto-deleted.
+      # - :deletable_orphans - [orphan, keeper] pairs, where orphan has no identity
+      #   attached but its avatar points at a discord id claimed by exactly one other
+      #   (linked) user. This is the confirmed race-loser case - safe to delete.
+      # - :unresolved_orphans - users with no identity attached and no corresponding
+      #   linked user (still mid-race, or the winner can't be identified). Reported
+      #   only, never auto-deleted.
+      def scan(management)
+        users = management.list_users
+        linked, unlinked = users.partition { |u| linked?(u) }
+
+        by_discord_id = linked.group_by { |u| u.data.identities.discord.userId }
+        conflicts = by_discord_id.select { |_id, group| group.length > 1 }
+
+        deletable_orphans = []
+        unresolved_orphans = []
+        unlinked.each do |user|
+          discord_id = avatar_discord_id_for(user)
+          keeper = discord_id && by_discord_id[discord_id]
+          if keeper&.length == 1
+            deletable_orphans << [user, keeper.first]
+          else
+            unresolved_orphans << user
+          end
+        end
+
+        { conflicts: conflicts, deletable_orphans: deletable_orphans, unresolved_orphans: unresolved_orphans }
+      end
+    end
+  end
+end

data/lib/yiffspace/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module YiffSpace
-  VERSION = "0.0.22"
+  VERSION = "0.0.23"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: yiffspace
 version: !ruby/object:Gem::Version
-  version: 0.0.22
+  version: 0.0.23
 platform: ruby
 authors:
 - Donovan_DMC
@@ -231,6 +231,7 @@ files:
 - lib/yiffspace/utils/table_builder.rb
 - lib/yiffspace/utils/trace_logger.rb
 - lib/yiffspace/utils/user_attribute.rb
+- lib/yiffspace/utils/user_deduper.rb
 - lib/yiffspace/utils/user_like.rb
 - lib/yiffspace/utils/user_resolvable.rb
 - lib/yiffspace/utils/user_to_id.rb