A Rubymotion wrapper for syncing JSON resources to Core Data.
-- ###Setup Gemfile
gem 'afmotion'
gem 'motion-courier', '~>0.5.2', git: 'https://github.com/pachun/Courier'bundle exec rake pod:install
bundle
-- ###Models Quickly
class League < Courier::Base
has_many :teams, as: :teams, on_delete: :cascade, inverse_name: :league
has_many :players, through: [:teams, :players]
attr_accessor :unpersisted_variables_here, :commissioner
end
class Team < Courier::Base
belongs_to :league, as: :league, on_delete: :nullify, inverse_name: :teams
has_many :players, as: :players, on_delete: :nullify, inverse_name: :team
property :id, Integer32, required: true, key: true
property :name, String
property :location, String
self.json_to_local = {:ID => :id, :TeamName => :name, :TeamTown => :location}
self.collection_path = "teams/"
self.individual_path = "teams/:id"
end
class Player < Courier::Base
belongs_to :team, as: :team, on_delete: :nullify, inverse_name: :players
property :id, Integer32, required: true
property :name, String, required: true, default_value: "Unknown"
property :age, Integer16
property :num_championships, Integer16
scope :seasoned, :or => [ :and => ["age >= 40", "num_championships >= 2"],
:and => ["age >= 30", "num_championships >= 4"],
]
end-- ###Schema Defining the models doesn't implicitly define a schema (managed object model) for you. You have to tell Courier the names of the models. You use the singleton instance's parcels=() to do that.
class AppDelegate
def application(application, didFinishLaunchingWithOptions:launchOptions)
Courier::Courier.instance.parcels = [League, Team, Player]
true
end
end-- ###Tests
Every rspec test that wants to initialize a new schema should begin with the line
Courier::nuke.everything.right.nowThis erases the old store completely.
describe "The Courier Base Class" do
before do
Courier::nuke.everything.right.now
Object.send(:remove_const, :Keyboard) if Object.constants.include?(:Keyboard)
class Keyboard < Courier::Base; end
Courier::Courier.instance.parcels = [Keyboard]
end
endIf you're declaring Courier::Base models in any block that'll be run twice, during your test suite, courier::base tries to re-register the class with core data, which will error because one by that name was already registered. Therefor, you should delete the old class constant prior to your new class definition.
--
class AppDelegate
def application(application, didFinishLaunchingWithOptions:launchOptions)
Courier::nuke.everything.right.now
Courier::Courier.instance.parcels = [League, Team, Player]
Fixtures::seed_players
true
end
end
module Fixtures
def self.seed_players
["Nick", "Chris", "Matt", "Josh"].each_with_index do |name, pos|
p = Player.create
p.id = pos
p.name = name
end
Courier::Courier.instance.save
end
end-- ###Create, Save, and Delete .create() will create a model instance, but it will not be persisted to subsequent app runs. Calling .save() on the Courier singleton will persist everything created up to that point
p1 = Person.create
p1.name = "John"
p2 = Person.create
Courier::Courier.instance.save.delete() is similar to create in that it will delete the model in the context of the current app run, but the deletion won't be persisted in successive app runs, unless you also call .save() is called on the courier instance after .delete() is called on the object. (.delete!() exists and is different from .delete(); don't ever call .delete!() on an object created with .create())
-- ###Loading from Core Data .all on a subclass of Courier::Base will return an array of all the created models. eg
Player.all # => [<Player1>, <Player2>, etc]To set relationships, you can do
team = Team.create
some_player = Player.create
team.add_to_players(Player.create) # either
some_player.team = team # or
team.players # => [those 2 players] # both work exactly the sameIf a league has many teams, and a team has many players, define a league has many players through teams as shown in the first README.md model, to get this shortcut
league.players-- ###Named Scopes You saw one earlier in the player model:
class Player < Courier::Base
belongs_to :team, as: :team, on_delete: :nullify, inverse_name: :players
property :id, Integer32, required: true
property :name, String, required: true, default_value: "Unknown"
property :age, Integer16
property :num_championships, Integer16
scope :seasoned, :or => [ :and => ["age >= 40", "num_championships >= 2"],
:and => ["age >= 30", "num_championships >= 4"],
]
endYou can nest those as deeply as you want, and there's a lot of them. Check out the courier_scope_spec.rb file to see them all. Everything in here is provisioned for by the Courier::Scope module. Some of the string comparisons like LIKE to find a string that contains another string can be tricky to use with those "x >= y" string format of comparison above. If you need to be really specific you can use
scope :males_and_doctors_whos_names_start_with_mc, Courier::Scope.where(:name, contains: "r. Mc")(for example) to search for names that have either Dr. Mc or Mr. Mc. A list of those and examples on how to use each of them is also in the courier_scope_spec.rb file.
-- ###Dynamic Scopes
seasoned_scope = Courier::Scope.where(:or => [ :and => ["age >= 40", "num_championships >= 2"],
:and => ["age >= 30", "num_championships >= 4"],
])
team.players.where(seasoned_scope) # => [player1, player2, etc]-- ###JSON Resources
class Team < Courier::Base
has_many :players, as: players, on_delete: :nullify, inverse_name: :team
property :id, Integer32, key: true
property :name, String
property :location, String
self.json_to_local = {:ID => :id, :TeamName => :name, :TeamTown => :location}
self.collection_path = "teams/"
self.individual_path = "teams/:id"
end
class Player < Courier::Base
belongs_to :team, as: :team, on_delete: :nullify, inverse_name: :players
property :id, Integer32, key: true
property :name, String
end
Courier::Courier.instance.url = "http://hello.world.me"
Courier::Courier.instance.parcels = [Team, Player]Fetching single resources
# pass in anything needed to resolve the individual_path set before
Team.find(id: 5) do |response|
if response.success?
response.resource # fetched resource
response.resource.merge_if { conditional } # do things with it
else
response.error_message # useful information
end
endFetching a collection of resources
Team.find_all do |response|
# response[:response].success? returns true/false ([:response] is the AFMotion response)
# response[:error_message] is an error message, if there was a problem
# response[:conflicts] will be an array, in this format:
#
# [
# {
# local: <TeamInstance0x123>,
# foreign: <TeamInstance0x456>,
# },
# {
# local: <TeamInstance0x789>,
# foreign: <TeamInstance0x012>,
# },
# {
# local: <TeamInstance0x345>,
# foreign: <TeamInstance0x678>,
# }
# ]
#
# resolve the conflicts one at a time the same way you would with single resources
response[:conflicts].each do |conflict|
conflict[:foreign].merge_if do
# true or false
end
end
# or you could add "conflict_policy :overwrite_local" to the team model to automatically do
# what was formerly accomplished with the following:
response[:conflicts].each do |conflict|
conflict[:foreign].merge!
end
# you can still do this, but I think the conflict_policy route is cleaner
endFetching has_many related resources:
# getting all the players on a team
team = Team.create
team.id = 1
team.find_players do |response| # endpoint inferred to be /teams/1/players (based on collection/individual url settings)
# response[:conflicts] will be an array of players this time:
#
# [
# {
# local: <PlayerInstance0x123>,
# foreign: <PlayerInstance0x456>,
# },
# {
# local: <PlayerInstance0x789>,
# foreign: <PlayerInstance0x012>,
# },
# {
# local: <PlayerInstance0x345>,
# foreign: <PlayerInstance0x678>,
# }
# ]
endPushing single resources back up to the server
team.push do |result|
if result.success?
# worked
else
# uh oh
puts "#{result.operation.response.statusCode.to_s}"
end
end-- ###To Do
-
Remove need for .save call.
-
lightweight command line migrations.