Avoid double serialization of message data

Prior to this commit, messages with metadata were always serialized in
the following way:

  ```ruby
  Base64.strict_encode64(
    ActiveSupport::JSON.encode({
      "_rails" => {
        "message" => Base64.strict_encode64(
          serializer.dump(data)
        ),
        "pur" => "the purpose",
        "exp" => "the expiration"
      },
    })
  )
  ```

in which the message data is serialized and URL-encoded twice.

This commit changes message serialization such that, when possible, the
data is serialized and URL-encoded only once:

  ```ruby
  Base64.strict_encode64(
    serializer.dump({
      "_rails" => {
        "data" => data,
        "pur" => "the purpose",
        "exp" => "the expiration"
      },
    })
  )
  ```

This improves performance in proportion to the size of the data:

**Benchmark**

  ```ruby
  # frozen_string_literal: true
  require "benchmark/ips"
  require "active_support/all"

  verifier = ActiveSupport::MessageVerifier.new("secret", serializer: JSON)

  payloads = [
    { "content" => "x" * 100 },
    { "content" => "x" * 2000 },
    { "content" => "x" * 1_000_000 },
  ]

  if ActiveSupport::Messages::Metadata.respond_to?(:use_message_serializer_for_metadata)
    ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata = true
  end

  Benchmark.ips do |x|
    payloads.each do |payload|
      x.report("generate ~#{payload["content"].size}B") do
        $generated_message = verifier.generate(payload, purpose: "x")
      end

      x.report("verify ~#{payload["content"].size}B") do
        verifier.verify($generated_message, purpose: "x")
      end
    end
  end

  puts

  puts "Message size:"
  payloads.each do |payload|
    puts "  ~#{payload["content"].size} bytes of data => " \
      "#{verifier.generate(payload, purpose: "x").size} byte message"
  end
  ```

**Before**

  ```
  Warming up --------------------------------------
        generate ~100B     1.578k i/100ms
          verify ~100B     2.506k i/100ms
       generate ~2000B   447.000  i/100ms
         verify ~2000B     1.409k i/100ms
    generate ~1000000B     1.000  i/100ms
      verify ~1000000B     6.000  i/100ms
  Calculating -------------------------------------
        generate ~100B     15.807k (± 1.8%) i/s -     80.478k in   5.093161s
          verify ~100B     25.240k (± 2.1%) i/s -    127.806k in   5.066096s
       generate ~2000B      4.530k (± 2.4%) i/s -     22.797k in   5.035398s
         verify ~2000B     14.136k (± 2.3%) i/s -     71.859k in   5.086267s
    generate ~1000000B     11.673  (± 0.0%) i/s -     59.000  in   5.060598s
      verify ~1000000B     64.372  (± 6.2%) i/s -    324.000  in   5.053304s

  Message size:
    ~100 bytes of data => 306 byte message
    ~2000 bytes of data => 3690 byte message
    ~1000000 bytes of data => 1777906 byte message
  ```

**After**

  ```
  Warming up --------------------------------------
        generate ~100B     4.689k i/100ms
          verify ~100B     3.183k i/100ms
       generate ~2000B     2.722k i/100ms
         verify ~2000B     2.066k i/100ms
    generate ~1000000B    12.000  i/100ms
      verify ~1000000B    11.000  i/100ms
  Calculating -------------------------------------
        generate ~100B     46.984k (± 1.2%) i/s -    239.139k in   5.090540s
          verify ~100B     32.043k (± 1.2%) i/s -    162.333k in   5.066903s
       generate ~2000B     27.163k (± 1.2%) i/s -    136.100k in   5.011254s
         verify ~2000B     20.726k (± 1.7%) i/s -    105.366k in   5.085442s
    generate ~1000000B    125.600  (± 1.6%) i/s -    636.000  in   5.064607s
      verify ~1000000B    122.039  (± 4.1%) i/s -    616.000  in   5.058386s

  Message size:
    ~100 bytes of data => 234 byte message
    ~2000 bytes of data => 2770 byte message
    ~1000000 bytes of data => 1333434 byte message
  ```

This optimization is only applied for recognized serializers that are
capable of serializing a `Hash`.

Additionally, because the optimization changes the message format, a
`config.active_support.use_message_serializer_for_metadata` option has
been added to disable it.  The optimization is disabled by default, but
enabled with `config.load_defaults 7.1`.

Regardless of whether the optimization is enabled, messages using either
format can still be read.

In the case of a rolling deploy of a Rails upgrade, wherein servers that
have not yet been upgraded must be able to read messages from upgraded
servers, the optimization can be disabled on first deploy, then safely
enabled on a subsequent deploy.

Author: jonathanhefner

activesupport/lib/active_support/message_encryptor.rb

@@ -85,6 +85,7 @@ module ActiveSupport
   #
   #   crypt.rotate old_secret, cipher: "aes-256-cbc"
   class MessageEncryptor
+    include Messages::Metadata
     prepend Messages::Rotator::Encryptor
 
     cattr_accessor :use_authenticated_message_encryption, instance_accessor: false, default: false
@@ -221,13 +222,7 @@ def self.key_len(cipher = default_cipher)
     end
 
     private
-      def serialize(value)
-        @serializer.dump(value)
-      end
-
-      def deserialize(value)
-        @serializer.load(value)
-      end
+      attr_reader :serializer
 
       def encode(data)
         @url_safe ? ::Base64.urlsafe_encode64(data, padding: false) : ::Base64.strict_encode64(data)
@@ -246,7 +241,7 @@ def _encrypt(value, **metadata_options)
         iv = cipher.random_iv
         cipher.auth_data = "" if aead_mode?
 
-        encrypted_data = cipher.update(Messages::Metadata.wrap(serialize(value), **metadata_options))
+        encrypted_data = cipher.update(serialize_with_metadata(value, **metadata_options))
         encrypted_data << cipher.final
 
         parts = [encrypted_data, iv]
@@ -275,8 +270,7 @@ def _decrypt(encrypted_message, purpose)
         decrypted_data = cipher.update(encrypted_data)
         decrypted_data << cipher.final
 
-        message = Messages::Metadata.verify(decrypted_data, purpose)
-        deserialize(message) if message
+        deserialize_with_metadata(decrypted_data, purpose: purpose)
       rescue OpenSSLCipherError, TypeError, ArgumentError, ::JSON::ParserError
         raise InvalidMessage
       end

activesupport/lib/active_support/message_verifier.rb

@@ -119,6 +119,7 @@ module ActiveSupport
   #   @verifier = ActiveSupport::MessageVerifier.new("secret", url_safe: true)
   #   @verifier.generate("signed message") #=> URL-safe string
   class MessageVerifier
+    include Messages::Metadata
     prepend Messages::Rotator::Verifier
 
     class InvalidSignature < StandardError; end
@@ -198,8 +199,7 @@ def verified(signed_message, purpose: nil, **)
       data, digest = get_data_and_digest_from(signed_message)
       if digest_matches_data?(digest, data)
         begin
-          message = Messages::Metadata.verify(decode(data), purpose)
-          @serializer.load(message) if message
+          deserialize_with_metadata(decode(data), purpose: purpose)
         rescue ArgumentError => argument_error
           return if argument_error.message.include?("invalid base64")
           raise
@@ -274,11 +274,14 @@ def verify(*args, **options)
     #   specified when verifying the message; otherwise, verification will fail.
     #   (See #verified and #verify.)
     def generate(value, expires_at: nil, expires_in: nil, purpose: nil)
-      data = encode(Messages::Metadata.wrap(@serializer.dump(value), expires_at: expires_at, expires_in: expires_in, purpose: purpose))
-      "#{data}#{SEPARATOR}#{generate_digest(data)}"
+      data = encode(serialize_with_metadata(value, expires_at: expires_at, expires_in: expires_in, purpose: purpose))
+      digest = generate_digest(data)
+      data << SEPARATOR << digest
     end
 
     private
+      attr_reader :serializer
+
       def encode(data)
         @url_safe ? Base64.urlsafe_encode64(data, padding: false) : Base64.strict_encode64(data)
       end

activesupport/lib/active_support/messages/metadata.rb

@@ -1,83 +1,101 @@
 # frozen_string_literal: true
 
 require "time"
+require "active_support/json"
 
 module ActiveSupport
   module Messages # :nodoc:
-    class Metadata # :nodoc:
-      def initialize(message, expires_at = nil, purpose = nil)
-        @message, @purpose = message, purpose
-        @expires_at = expires_at.is_a?(String) ? parse_expires_at(expires_at) : expires_at
-      end
+    module Metadata # :nodoc:
+      singleton_class.attr_accessor :use_message_serializer_for_metadata
 
-      def as_json(options = {})
-        { _rails: { message: @message, exp: @expires_at, pur: @purpose } }
-      end
+      ENVELOPE_SERIALIZERS = [
+        ::JSON,
+        ActiveSupport::JSON,
+        ActiveSupport::JsonWithMarshalFallback,
+        Marshal,
+      ]
 
-      class << self
-        def wrap(message, expires_at: nil, expires_in: nil, purpose: nil)
-          if expires_at || expires_in || purpose
-            JSON.encode new(encode(message), pick_expiry(expires_at, expires_in), purpose)
+      private
+        def serialize_with_metadata(data, **metadata)
+          has_metadata = metadata.any? { |k, v| v }
+
+          if has_metadata && !use_message_serializer_for_metadata?
+            data_string = serialize_to_json_safe_string(data)
+            envelope = wrap_in_metadata_envelope({ "message" => data_string }, **metadata)
+            ActiveSupport::JSON.encode(envelope)
           else
-            message
+            data = wrap_in_metadata_envelope({ "data" => data }, **metadata) if has_metadata
+            serializer.dump(data)
           end
         end
 
-        def verify(message, purpose)
-          extract_metadata(message).verify(purpose)
-        end
-
-        private
-          def pick_expiry(expires_at, expires_in)
-            if expires_at
-              expires_at.utc.iso8601(3)
-            elsif expires_in
-              Time.now.utc.advance(seconds: expires_in).iso8601(3)
-            end
-          end
-
-          def extract_metadata(message)
-            begin
-              data = JSON.decode(message) if message.start_with?('{"_rails":')
-            rescue ::JSON::JSONError
-            end
-
-            if data
-              new(decode(data["_rails"]["message"]), data["_rails"]["exp"], data["_rails"]["pur"])
+        def deserialize_with_metadata(message, **expected_metadata)
+          if dual_serialized_metadata_envelope_json?(message)
+            envelope = ActiveSupport::JSON.decode(message)
+            extracted = extract_from_metadata_envelope(envelope, **expected_metadata)
+            deserialize_from_json_safe_string(extracted["message"]) if extracted
+          else
+            deserialized = serializer.load(message)
+            if metadata_envelope?(deserialized)
+              extracted = extract_from_metadata_envelope(deserialized, **expected_metadata)
+              extracted["data"] if extracted
             else
-              new(message)
+              deserialized if expected_metadata.none? { |k, v| v }
             end
           end
+        end
 
-          def encode(message)
-            ::Base64.strict_encode64(message)
-          end
+        def use_message_serializer_for_metadata?
+          Metadata.use_message_serializer_for_metadata && Metadata::ENVELOPE_SERIALIZERS.include?(serializer)
+        end
 
-          def decode(message)
-            ::Base64.strict_decode64(message)
-          end
-      end
+        def wrap_in_metadata_envelope(hash, expires_at: nil, expires_in: nil, purpose: nil)
+          expiry = pick_expiry(expires_at, expires_in)
+          hash["exp"] = expiry if expiry
+          hash["pur"] = purpose.to_s if purpose
+          { "_rails" => hash }
+        end
 
-      def verify(purpose)
-        @message if match?(purpose) && fresh?
-      end
+        def extract_from_metadata_envelope(envelope, purpose: nil)
+          hash = envelope["_rails"]
+          return if hash["exp"] && Time.now.utc >= parse_expiry(hash["exp"])
+          return if hash["pur"] != purpose&.to_s
+          hash
+        end
 
-      private
-        def match?(purpose)
-          @purpose.to_s == purpose.to_s
+        def metadata_envelope?(object)
+          object.is_a?(Hash) && object.key?("_rails")
+        end
+
+        def dual_serialized_metadata_envelope_json?(string)
+          string.start_with?('{"_rails":{"message":')
         end
 
-        def fresh?
-          @expires_at.nil? || Time.now.utc < @expires_at
+        def pick_expiry(expires_at, expires_in)
+          if expires_at
+            expires_at.utc.iso8601(3)
+          elsif expires_in
+            Time.now.utc.advance(seconds: expires_in).iso8601(3)
+          end
         end
 
-        def parse_expires_at(expires_at)
-          if ActiveSupport.use_standard_json_time_format
+        def parse_expiry(expires_at)
+          if !expires_at.is_a?(String)
+            expires_at
+          elsif ActiveSupport.use_standard_json_time_format
             Time.iso8601(expires_at)
           else
             Time.parse(expires_at)
           end
         end
+
+        def serialize_to_json_safe_string(data)
+          ::Base64.strict_encode64(serializer.dump(data))
+        end
+
+        def deserialize_from_json_safe_string(string)
+          serializer.load(::Base64.strict_decode64(string))
+        end
     end
   end
 end

activesupport/lib/active_support/railtie.rb

@@ -192,5 +192,12 @@ class Railtie < Rails::Railtie # :nodoc:
         end
       end
     end
+
+    initializer "active_support.set_use_message_serializer_for_metadata" do |app|
+      config.after_initialize do
+        ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata =
+          app.config.active_support.use_message_serializer_for_metadata
+      end
+    end
   end
 end

activesupport/test/messages/message_metadata_tests.rb

@@ -2,6 +2,7 @@
 
 require "active_support/json"
 require "active_support/time"
+require "active_support/messages/metadata"
 
 module MessageMetadataTests
   extend ActiveSupport::Concern
@@ -89,6 +90,17 @@ module MessageMetadataTests
       codec = make_codec(serializer: ActiveSupport::MessageEncryptor::NullSerializer)
       assert_roundtrip "a string", codec, { purpose: "x", expires_in: 1.year }, { purpose: "x" }
     end
+
+    test "messages are readable regardless of use_message_serializer_for_metadata" do
+      each_scenario do |data, codec|
+        message = encode(data, codec, purpose: "x")
+        message_setting = ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata
+
+        using_message_serializer_for_metadata(!message_setting) do
+          assert_equal data, decode(message, codec, purpose: "x")
+        end
+      end
+    end
   end
 
   private
@@ -116,11 +128,23 @@ def self.load(value)
       ["a string", 123, Time.local(2004), { "key" => "value" }],
     ]
 
+    def using_message_serializer_for_metadata(value = true)
+      original = ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata
+      ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata = value
+      yield
+    ensure
+      ActiveSupport::Messages::Metadata.use_message_serializer_for_metadata = original
+    end
+
     def each_scenario
-      SERIALIZERS.each do |serializer|
-        codec = make_codec(serializer: serializer)
-        DATA.each do |data|
-          yield data, codec
+      [false, true].each do |use_message_serializer_for_metadata|
+        using_message_serializer_for_metadata(use_message_serializer_for_metadata) do
+          SERIALIZERS.each do |serializer|
+            codec = make_codec(serializer: serializer)
+            DATA.each do |data|
+              yield data, codec
+            end
+          end
         end
       end
     end

activesupport/test/messages/message_verifier_metadata_test.rb

@@ -56,6 +56,21 @@ class MessageVerifierMetadataTest < ActiveSupport::TestCase
     end
   end
 
+  test "messages are readable by legacy versions when use_message_serializer_for_metadata = false" do
+    # Message generated by Rails 7.0 using:
+    #
+    #   verifier = ActiveSupport::MessageVerifier.new("secret", serializer: JSON)
+    #   legacy_message = verifier.generate("legacy", purpose: "test", expires_at: Time.utc(3000))
+    #
+    legacy_message = "eyJfcmFpbHMiOnsibWVzc2FnZSI6IklteGxaMkZqZVNJPSIsImV4cCI6IjMwMDAtMDEtMDFUMDA6MDA6MDAuMDAwWiIsInB1ciI6InRlc3QifX0=--81b11c317dba91cedd86ab79b7d7e68de8d290b3"
+
+    verifier = ActiveSupport::MessageVerifier.new("secret", serializer: JSON)
+
+    using_message_serializer_for_metadata(false) do
+      assert_equal legacy_message, verifier.generate("legacy", purpose: "test", expires_at: Time.utc(3000))
+    end
+  end
+
   private
     def make_codec(**options)
       ActiveSupport::MessageVerifier.new("secret", **options)

guides/source/configuring.md

@@ -2195,6 +2195,21 @@ The default value depends on the `config.load_defaults` target version:
 | (original)            | `false`              |
 | 5.2                   | `true`               |
 
+#### `config.active_support.use_message_serializer_for_metadata`
+
+When `true`, enables a performance optimization that serializes message data and
+metadata together. This changes the message format, so messages serialized this
+way cannot be read by older (< 7.1) versions of Rails. However, messages that
+use the old format can still be read, regardless of whether this optimization is
+enabled.
+
+The default value depends on the `config.load_defaults` target version:
+
+| Starting with version | The default value is |
+| --------------------- | -------------------- |
+| (original)            | `false`              |
+| 7.1                   | `true`               |
+
 #### `config.active_support.cache_format_version`
 
 Specifies which version of the cache serializer to use. Possible values are `6.1` and `7.0`.

railties/lib/rails/application/configuration.rb

@@ -306,6 +306,7 @@ def load_defaults(target_version)
           if respond_to?(:active_support)
             active_support.default_message_encryptor_serializer = :json
             active_support.default_message_verifier_serializer = :json
+            active_support.use_message_serializer_for_metadata = true
             active_support.raise_on_invalid_cache_expiration_time = true
           end
 

railties/lib/rails/generators/rails/app/templates/config/initializers/new_framework_defaults_7_1.rb.tt

@@ -99,6 +99,17 @@
 #
 # For detailed migration steps, check out https://guides.rubyonrails.org/v7.1/upgrading_ruby_on_rails.html#new-activesupport-messageverifier-default-serializer
 
+# Enable a performance optimization that serializes message data and metadata
+# together. This changes the message format, so messages serialized this way
+# cannot be read by older versions of Rails. However, messages that use the old
+# format can still be read, regardless of whether this optimization is enabled.
+#
+# To perform a rolling deploy of a Rails 7.1 upgrade, wherein servers that have
+# not yet been upgraded must be able to read messages from upgraded servers,
+# leave this optimization off on the first deploy, then enable it on a
+# subsequent deploy.
+# Rails.application.config.active_support.use_message_serializer_for_metadata = true
+
 # Set the maximum size for Rails log files.
 #
 # `config.load_defaults 7.1` does not set this value for environments other than