standard_ledger 0.5.1 → 0.7.0

data/lib/tasks/standard_ledger.rake

@@ -1,60 +0,0 @@
-namespace :standard_ledger do
-  # PostgreSQL-only: queries `pg_trigger` directly. The only mode that
-  # registers a trigger today is `:trigger`, and the only adopter is
-  # nutripod-web (Postgres). SQLite has no comparable per-statement
-  # trigger introspection that makes sense for this gem's contract — if
-  # a non-Postgres host adopts `:trigger` mode in the future, they'll
-  # need to extend this task with a connection-adapter dispatch. The
-  # task is loud about its Postgres assumption: a `pg_trigger` query
-  # against a non-Postgres connection will raise, which is preferable
-  # to silently passing.
-  desc "Verify that every :trigger projection's trigger exists in the database (Postgres-only)"
-  task doctor: :environment do
-    require "standard_ledger"
-
-    # Discover entry classes by walking ActiveRecord descendants for
-    # ones that include `StandardLedger::Projector` and have at least
-    # one `:trigger` projection registered. The host's eager loading
-    # (Rails default in production / when explicitly invoked in dev)
-    # ensures all entry classes are loaded before this iterates.
-    entry_classes = ActiveRecord::Base.descendants.select { |klass|
-      klass.respond_to?(:standard_ledger_projections) &&
-        klass.standard_ledger_projections.any? { |d| d.mode == :trigger }
-    }
-
-    missing = []
-    entry_classes.each do |klass|
-      klass.standard_ledger_projections_for(:trigger).each do |definition|
-        # `pg_trigger.tgname` is the trigger's name as known to Postgres,
-        # but trigger names are scoped per-table — two tables can each
-        # have a trigger called e.g. `update_counts`. Join `pg_trigger`
-        # to `pg_class` and filter by the entry class's table name so the
-        # doctor reports presence on the *correct* table, not "anywhere
-        # in the schema". Use `klass.connection` (rather than
-        # `ActiveRecord::Base.connection`) so multi-DB setups query the
-        # connection that owns the entry class's table.
-        result = klass.connection.exec_query(
-          "SELECT 1 FROM pg_trigger t " \
-          "JOIN pg_class c ON c.oid = t.tgrelid " \
-          "WHERE t.tgname = $1 AND c.relname = $2 " \
-          "LIMIT 1",
-          "standard_ledger:doctor",
-          [ definition.trigger_name, klass.table_name ]
-        )
-        if result.rows.empty?
-          missing << "  #{klass.name}##{definition.target_association}: trigger #{definition.trigger_name.inspect} not found"
-        end
-      end
-    end
-
-    if missing.empty?
-      puts "All :trigger projections have their triggers present."
-    else
-      warn "Missing triggers detected:"
-      missing.each { |line| warn line }
-      warn ""
-      warn "Run the migration that creates the trigger, or check that the trigger name in `projects_onto` matches the actual trigger name in the schema."
-      exit 1
-    end
-  end
-end