New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Allow to define the default column serializer #47463
Merged
byroot
merged 1 commit into
rails:main
from
Shopify:configurable-default-column-serializer
Feb 22, 2023
Merged
Changes from all commits
Commits
File filter
Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -15,6 +15,10 @@ def initialize(name, type) | |
end | ||
end | ||
|
||
included do | ||
class_attribute :default_column_serializer, instance_accessor: false, default: Coders::YAMLColumn | ||
end | ||
|
||
module ClassMethods | ||
# If you have an attribute that needs to be saved to the database as a | ||
# serialized object, and retrieved by deserializing into the same object, | ||
|
@@ -36,21 +40,16 @@ module ClassMethods | |
# ==== Parameters | ||
# | ||
# * +attr_name+ - The name of the attribute to serialize. | ||
# * +class_name_or_coder+ - Optional. May be one of the following: | ||
# * <em>default</em> - The attribute value will be serialized as YAML. | ||
# The attribute value must respond to +to_yaml+. | ||
# * +Array+ - The attribute value will be serialized as YAML, but an | ||
# empty +Array+ will be serialized as +NULL+. The attribute value | ||
# must be an +Array+. | ||
# * +Hash+ - The attribute value will be serialized as YAML, but an | ||
# empty +Hash+ will be serialized as +NULL+. The attribute value | ||
# must be a +Hash+. | ||
# * +JSON+ - The attribute value will be serialized as JSON. The | ||
# attribute value must respond to +to_json+. | ||
# * <em>custom coder</em> - The attribute value will be serialized | ||
# * +coder+ The serializer implementation to use, e.g. +JSON+. | ||
# * The attribute value will be serialized | ||
# using the coder's <tt>dump(value)</tt> method, and will be | ||
# deserialized using the coder's <tt>load(string)</tt> method. The | ||
# +dump+ method may return +nil+ to serialize the value as +NULL+. | ||
# * +type+ - Optional. What the type of the serialized object should be. | ||
# * Attempting to serialize another type will raise an | ||
# <tt>ActiveRecord::SerializationTypeMismatch</tt> error. | ||
# * If the column is +NULL+ or starting from a new record, the default value | ||
# will set to +type.new+ | ||
# * +yaml+ - Optional. Yaml specific options. The allowed config is: | ||
# * +:permitted_classes+ - +Array+ with the permitted classes. | ||
# * +:unsafe_load+ - Unsafely load YAML blobs, allow YAML to load any class. | ||
|
@@ -61,30 +60,101 @@ module ClassMethods | |
# this option is not passed, the previous default value (if any) will | ||
# be used. Otherwise, the default will be +nil+. | ||
# | ||
# ==== Choosing a serializer | ||
# | ||
# While any serialization format can be used, it is recommended to carefully | ||
# evaludate the properties of a serializer before using it, as migrating to | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. 👀 typo: evaludate -> evaluate |
||
# another format later on can be difficult. | ||
# | ||
# ===== Avoid accepting arbitrary types | ||
# | ||
# When serializing data in a column, it is heavily recommended to make sure | ||
# only expected types will be serialized. For instance some serializer like | ||
# +Marshal+ or +YAML+ are capable of serializing almost any Ruby object. | ||
# | ||
# This can lead to unexpected types being serialized, and it it happens | ||
# that type serialization must remain backward and forward compatible as long | ||
# as some database record still contain these serialized types. | ||
# | ||
# class Address | ||
# def initialize(line, city, country) | ||
# @line, @city, @country = line, city, country | ||
# end | ||
# end | ||
# | ||
# In the above example, if any of the +Address+ attributes is renamed, | ||
# instances that were persisted before the change will be loaded with the | ||
# old attributes. This problem is even worse when the serialized type comes | ||
# from a dependency which doesn't expect to be serialized this way and may | ||
# change its internal representation without notice. | ||
# | ||
# As such, it is heavily recommended to instead convert these objects into | ||
# primitives of the serialization format, for example: | ||
# | ||
# class Address | ||
# attr_reader :line, :city, :country | ||
# | ||
# def self.load(payload) | ||
# data = YAML.safe_load(payload) | ||
# new(data["line"], data["city"], data["country"]) | ||
# end | ||
# | ||
# def self.dump(address) | ||
# YAML.safe_dump( | ||
# "line" => address.line, | ||
# "city" => address.city, | ||
# "country" => address.country, | ||
# ) | ||
# end | ||
# | ||
# def initialize(line, city, country) | ||
# @line, @city, @country = line, city, country | ||
# end | ||
# end | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :address, coder: Address | ||
# end | ||
# | ||
# This patterns allow to be more deliberate about what is serialized, and | ||
# to evolve the format in a backward compatible way. | ||
# | ||
# ===== Ensure serialization stability | ||
# | ||
# Some serialization methods may accept some types they don't support by | ||
# silently casting them to another types. This can cause bugs when the | ||
# data is deserialized. | ||
# | ||
# For instance the +JSON+ serializer provided in the standard library will | ||
# silently cast unsupported types to +String+: | ||
# | ||
# >> JSON.parse(JSON.dump(Struct.new(:foo))) | ||
# => "#<Class:0x000000013090b4c0>" | ||
# | ||
# ==== Examples | ||
# | ||
# ===== Serialize the +preferences+ attribute using YAML | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :preferences | ||
# serialize :preferences, coder: YAML | ||
# end | ||
# | ||
# ===== Serialize the +preferences+ attribute using JSON | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :preferences, JSON | ||
# serialize :preferences, coder: JSON | ||
# end | ||
# | ||
# ===== Serialize the +preferences+ +Hash+ using YAML | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :preferences, Hash | ||
# serialize :preferences, type: Hash, coder: YAML | ||
# end | ||
# | ||
# ===== Serializes +preferences+ to YAML, permitting select classes | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :preferences, yaml: { permitted_classes: [Symbol, Time] } | ||
# serialize :preferences, coder: YAML, yaml: { permitted_classes: [Symbol, Time] } | ||
# end | ||
# | ||
# ===== Serialize the +preferences+ attribute using a custom coder | ||
|
@@ -106,35 +176,74 @@ module ClassMethods | |
# end | ||
# | ||
# class User < ActiveRecord::Base | ||
# serialize :preferences, Rot13JSON | ||
# serialize :preferences, coder: Rot13JSON | ||
# end | ||
# | ||
def serialize(attr_name, class_name_or_coder = Object, yaml: {}, **options) | ||
# When ::JSON is used, force it to go through the Active Support JSON encoder | ||
# to ensure special objects (e.g. Active Record models) are dumped correctly | ||
# using the #as_json hook. | ||
coder = if class_name_or_coder == ::JSON | ||
Coders::JSON | ||
elsif [:load, :dump].all? { |x| class_name_or_coder.respond_to?(x) } | ||
class_name_or_coder | ||
else | ||
Coders::YAMLColumn.new(attr_name, class_name_or_coder, **yaml) | ||
def serialize(attr_name, class_name_or_coder = nil, coder: nil, type: Object, yaml: {}, **options) | ||
unless class_name_or_coder.nil? | ||
if class_name_or_coder.respond_to?(:new) | ||
ActiveRecord.deprecator.warn(<<-MSG) | ||
Passing the class as positional argument is deprecated and will be remove in Rails 7.2. | ||
|
||
Please pass the class as a keyword argument: | ||
|
||
serialize #{attr_name.inspect}, type: #{class_name_or_coder.name} | ||
MSG | ||
type = class_name_or_coder | ||
else | ||
ActiveRecord.deprecator.warn(<<-MSG) | ||
Passing the coder as positional argument is deprecated and will be remove in Rails 7.2. | ||
|
||
Please pass the coder as a keyword argument: | ||
|
||
serialize #{attr_name.inspect}, coder: #{class_name_or_coder} | ||
MSG | ||
coder = class_name_or_coder | ||
end | ||
end | ||
|
||
coder ||= default_column_serializer | ||
unless coder | ||
raise ArgumentError, <<~MSG.squish | ||
missing keyword: :coder | ||
|
||
If no default coder is configured, a coder must be provided to `serialize`. | ||
MSG | ||
end | ||
|
||
column_serializer = build_column_serializer(attr_name, coder, type, yaml) | ||
|
||
attribute(attr_name, **options) do |cast_type| | ||
if type_incompatible_with_serialize?(cast_type, class_name_or_coder) | ||
if type_incompatible_with_serialize?(cast_type, coder, type) | ||
raise ColumnNotSerializableError.new(attr_name, cast_type) | ||
end | ||
|
||
cast_type = cast_type.subtype if Type::Serialized === cast_type | ||
Type::Serialized.new(cast_type, coder, default: columns_hash[attr_name.to_s]&.default) | ||
Type::Serialized.new(cast_type, column_serializer, default: columns_hash[attr_name.to_s]&.default) | ||
end | ||
end | ||
|
||
private | ||
def type_incompatible_with_serialize?(type, class_name) | ||
type.is_a?(ActiveRecord::Type::Json) && class_name == ::JSON || | ||
type.respond_to?(:type_cast_array, true) && class_name == ::Array | ||
def build_column_serializer(attr_name, coder, type, yaml = nil) | ||
# When ::JSON is used, force it to go through the Active Support JSON encoder | ||
# to ensure special objects (e.g. Active Record models) are dumped correctly | ||
# using the #as_json hook. | ||
coder = Coders::JSON if coder == ::JSON | ||
|
||
if coder == ::YAML | ||
Coders::YAMLColumn.new(attr_name, type, **(yaml || {})) | ||
elsif coder.respond_to?(:new) && !coder.respond_to?(:load) | ||
coder.new(attr_name, type) | ||
elsif type && type != Object | ||
Coders::ColumnSerializer.new(attr_name, coder, type) | ||
else | ||
coder | ||
end | ||
end | ||
|
||
def type_incompatible_with_serialize?(cast_type, coder, type) | ||
cast_type.is_a?(ActiveRecord::Type::Json) && coder == ::JSON || | ||
cast_type.respond_to?(:type_cast_array, true) && type == ::Array | ||
end | ||
end | ||
end | ||
|
61 changes: 61 additions & 0 deletions
61
activerecord/lib/active_record/coders/column_serializer.rb
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,61 @@ | ||
# frozen_string_literal: true | ||
|
||
module ActiveRecord | ||
module Coders # :nodoc: | ||
class ColumnSerializer # :nodoc: | ||
attr_reader :object_class | ||
attr_reader :coder | ||
|
||
def initialize(attr_name, coder, object_class = Object) | ||
@attr_name = attr_name | ||
@object_class = object_class | ||
@coder = coder | ||
check_arity_of_constructor | ||
end | ||
|
||
def init_with(coder) # :nodoc: | ||
@attr_name = coder["attr_name"] | ||
@object_class = coder["object_class"] | ||
@coder = coder["coder"] | ||
end | ||
|
||
def dump(object) | ||
return if object.nil? | ||
|
||
assert_valid_value(object, action: "dump") | ||
coder.dump(object) | ||
end | ||
|
||
def load(payload) | ||
if payload.nil? | ||
if @object_class != ::Object | ||
return @object_class.new | ||
end | ||
return nil | ||
end | ||
|
||
object = coder.load(payload) | ||
|
||
assert_valid_value(object, action: "load") | ||
object ||= object_class.new if object_class != Object | ||
|
||
object | ||
end | ||
|
||
# Public because it's called by Type::Serialized | ||
def assert_valid_value(object, action:) | ||
unless object.nil? || object_class === object | ||
raise SerializationTypeMismatch, | ||
"can't #{action} `#{@attr_name}`: was supposed to be a #{object_class}, but was a #{object.class}. -- #{object.inspect}" | ||
end | ||
end | ||
|
||
private | ||
def check_arity_of_constructor | ||
load(nil) | ||
rescue ArgumentError | ||
raise ArgumentError, "Cannot serialize #{object_class}. Classes passed to `serialize` must have a 0 argument constructor." | ||
end | ||
end | ||
end | ||
end |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This does not create a global default but instead is configurable for each subclass of
ActiveRecord::Base
. The ability to configure eachserialize
is enough, plus a global default for compatibility should be enough.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can make it a global, but I thought having it inheritable made sense for say engines so that they can have their base abstract model, and define the serializer they use across the engine centrally.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hum, they could also redefine
ApplicationController.serialize
to provide defaults. 🤔There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
And break if we change the signature. I prefer a class attribute.