Add ability to ignore tables in the schema cache

In cases where an application uses pt-osc or lhm they may have
temporary tables being used for migrations. Those tables shouldn't be
included by the schema cache because it makes the cache bigger and those
tables shouldn't ever be queried by the application. This feature allows
applications to configure a list of or regex of tables to ignore. The
behavior for ignored tables for each method in the schema cache
(`columns`, `columns_hash`, `primary_keys`, and `indexes`) matches the
behavior of a non-existent table. `columns` and `columns_hash` will
raise a database not found error, `primary_keys` will return `nil`. and
`indexes` will return an empty array. See rails#43105 which make behavior
across adapters consistent.

To use in an application configure `config.active_record.schema_cache_ignored_tables`
to an array of tables or regex's. Those tables will not be included in
the schema cache dump.

Author: eileencodes

activerecord/CHANGELOG.md

@@ -1,3 +1,21 @@
+*   Add config option for ignoring tables when dumping the schema cache.
+
+    Applications can now be configured to ignore certain tables when dumping the schema cache.
+
+    The configuration option can table an array of tables:
+
+    ```ruby
+    config.active_record.schema_cache_ignored_tables = ["ignored_table", "another_ignored_table"]
+    ```
+
+    Or a regex:
+
+    ```ruby
+    config.active_record.schema_cache_ignored_tables = [/^_/]
+    ```
+
+    *Eileen M. Uchitelle*
+
 *   Make schema cache methods return consistent results.
 
     Previously the schema cache methods `primary_keys`, `columns, `columns_hash`, and `indexes`

activerecord/lib/active_record.rb

@@ -170,6 +170,12 @@ module Tasks
   autoload :TestDatabases, "active_record/test_databases"
   autoload :TestFixtures, "active_record/fixtures"
 
+  # A list of tables or regex's to match tables to ignore when
+  # dumping the schema cache. For example if this is set to +[/^_/]+
+  # the schema cache will not dump tables named with an underscore.
+  singleton_class.attr_accessor :schema_cache_ignored_tables
+  self.schema_cache_ignored_tables = []
+
   singleton_class.attr_accessor :legacy_connection_handling
   self.legacy_connection_handling = true
 

activerecord/lib/active_record/connection_adapters/schema_cache.rb

@@ -86,6 +86,7 @@ def primary_keys(table_name)
 
       # A cached lookup for table existence.
       def data_source_exists?(name)
+        return if ignored_table?(name)
         prepare_data_sources if @data_sources.empty?
         return @data_sources[name] if @data_sources.key? name
 
@@ -108,6 +109,10 @@ def data_sources(name)
 
       # Get the columns for a table
       def columns(table_name)
+        if ignored_table?(table_name)
+          raise ActiveRecord::StatementInvalid, "Table '#{table_name}' doesn't exist"
+        end
+
         @columns.fetch(table_name) do
           @columns[deep_deduplicate(table_name)] = deep_deduplicate(connection.columns(table_name))
         end
@@ -166,7 +171,7 @@ def clear_data_source_cache!(name)
 
       def dump_to(filename)
         clear!
-        connection.data_sources.each { |table| add(table) }
+        tables_to_cache.each { |table| add(table) }
         open(filename) { |f|
           if filename.include?(".dump")
             f.write(Marshal.dump(self))
@@ -190,6 +195,18 @@ def marshal_load(array)
       end
 
       private
+        def tables_to_cache
+          connection.data_sources.reject do |table|
+            ignored_table?(table)
+          end
+        end
+
+        def ignored_table?(table_name)
+          ActiveRecord.schema_cache_ignored_tables.any? do |ignored|
+            ignored === table_name
+          end
+        end
+
         def reset_version!
           @version = connection.migration_context.current_version
         end
@@ -216,7 +233,9 @@ def deep_deduplicate(value)
         end
 
         def prepare_data_sources
-          connection.data_sources.each { |source| @data_sources[source] = true }
+          tables_to_cache.each do |source|
+            @data_sources[source] = true
+          end
         end
 
         def open(filename)

activerecord/test/cases/connection_adapters/schema_cache_test.rb

@@ -220,6 +220,43 @@ def test_marshal_dump_and_load_via_disk
         tempfile.unlink
       end
 
+      def test_marshal_dump_and_load_with_ignored_tables
+        old_ignore = ActiveRecord.schema_cache_ignored_tables
+        ActiveRecord.schema_cache_ignored_tables = ["p_schema_migrations"]
+        # Create an empty cache.
+        cache = SchemaCache.new @connection
+
+        tempfile = Tempfile.new(["schema_cache-", ".dump"])
+        # Dump it. It should get populated before dumping.
+        cache.dump_to(tempfile.path)
+
+        # Load a new cache.
+        cache = SchemaCache.load_from(tempfile.path)
+        cache.connection = @connection
+
+        # Assert a table in the cache
+        assert cache.data_sources("posts"), "expected posts to be in the cached data_sources"
+        assert_equal 12, cache.columns("posts").size
+        assert_equal 12, cache.columns_hash("posts").size
+        assert cache.data_sources("posts")
+        assert_equal "id", cache.primary_keys("posts")
+        assert_equal 1, cache.indexes("posts").size
+
+        # Assert ignored table. Behavior should match non-existent table.
+        assert_nil cache.data_sources("p_schema_migrations"), "expected comments to not be in the cached data_sources"
+        assert_raises ActiveRecord::StatementInvalid do
+          cache.columns("p_schema_migrations")
+        end
+        assert_raises ActiveRecord::StatementInvalid do
+          cache.columns_hash("p_schema_migrations").size
+        end
+        assert_nil cache.primary_keys("p_schema_migrations")
+        assert_equal [], cache.indexes("p_schema_migrations")
+      ensure
+        tempfile.unlink
+        ActiveRecord.schema_cache_ignored_tables = old_ignore
+      end
+
       def test_marshal_dump_and_load_with_gzip
         # Create an empty cache.
         cache = SchemaCache.new @connection