class CookiesController
  def update
    cookies[:name] = 'demo'
  end
end
describe CookiesController
  it "should update cookies[:name] to 'demo'" do
    request.cookies[:name] = 'test'
    put :update
    request.cookies[:name].should == 'demo'
  end
end


如果你运行上面的测试，将会得到一个：expected ‘demo’ got ‘test’，这是怎么回事，cookie设置失败了？实际上这是正确的，之所以产生这样的结果是因为我们对Rails的Cookie机制不了解，首先我们需要区分request.cookies和Controller中的cookies，request.cookies是用于模拟客户端发送的 cookie的，它是一个Hash对象，因此我们可以像使用hash那样使用它：

>> request.cookies[:name] = ‘test’
>> request.cookies[:name]
test

但是Controller中的cookies则不同，它是一个ActionController::CookieJar对象，这个cookies对象实际上保存了2个cookie集合：incoming cookie和outgoing cookie。incoming cookie就是客户请求中所携带的cookie（也就是request.cookies），而outgoing cookie则是在请求处理完后将被设置回客户端的cookie。

因此尽管我们使用了[]=方法来设置cookie，但因为cookies对象重载了[]与[]=方法：

1. []方法返回的是incoming cookie，也就是request中携带的cookie
2. 而[]=方法设置的却是outgoing cookie，它只对下一个请求起作用，也就是说它会被包含在下一个请求的incoming cookie中。

因此设置controller cookies，我们只能得到：

>> cookies[:name] = ‘test’
>> cookies[:name]
nil

如果你需要测试cookies是否设置成功，那么一个简单的方法，就是将controller中的cookies替换成一个hash，比如：


describe CookiesController
  it "should update cookies[:name] to 'demo'" do
    request.cookies[:name] = 'test'
    @cookies = {}
    controller.stub!(:cookies).and_return(@cookies)
    put :update
    @cookies[:name].should == 'demo'
  end
end

无觅猜您也喜欢：
升级到Rails 3 Beta	Rails性能优化简明指南	Rails2.0新特性介绍（ 下）	Shanghai on Rails 7.26线下活动
无觅

1) Failure:
test_should_be_create(ForumCommentsControllerTest)
method assert_block in assertions.rb at line 48
method _wrap_assertion in assertions.rb at line 495
method assert_block in assertions.rb at line 46
method test_should_be_create in forum_comments_controller_test.rb at line 38
method __send__ in test_case_adapter.rb at line 19
method run in test_case_adapter.rb at line 19
method run in testsuite.rb at line 34
method each in testsuite.rb at line 33
method run in testsuite.rb at line 33
method run in testsuite.rb at line 34
method each in testsuite.rb at line 33
method run in testsuite.rb at line 33
method run_suite in testrunnermediator.rb at line 46
method start_mediator in testrunner.rb at line 67
method start in testrunner.rb at line 41
method run in testrunnerutilities.rb at line 29
method run in autorunner.rb at line 216
method run in autorunner.rb at line 12
at top level in unit.rb at line 278
at top level in forum_comments_controller_test.rb at line 63
response is not a redirection to all of the options supplied (redirection is <{"action"=>"show", "id"=>1, "controller"=>"forums"}>), difference: <{"id"=>"post-2"}>

thoughtbot提供的quietbacktrace gem或许可以帮你减轻一些痛苦：

$ sudo gem install quietbacktrace

要在Rails中使用quietbacktrace，你需要将它解包到vendor目录：

cd rails-app-folder/vendor/gems
gem unpack quietbacktrace

然后在config/environments/test.rb 中添加：

require 'quietbacktrace'

然后再在test/test_helper.rb中添加：

class Test::Unit::TestCase
  self.backtrace_silencers << :rails_vendor
  self.backtrace_filters << :rails_root
end

好了，现在再次运行测试，你会发现，上面冗长无比的输出已经变成了下面的样子：

1) Failure:
test_should_be_create(ForumCommentsControllerTest)
at top level in forum_comments_controller_test.rb at line 38
at top level in forum_comments_controller_test.rb at line 63
response is not a redirection to all of the options supplied (redirection is <{"action"=>"show", "id"=>1, "controller"=>"forums"}>), difference: <{"id"=>"post-2"}>

当然，如果你偶尔想要回首下往日的艰苦岁月，那也很简单，只需要在test/test_helper.rb中添加：

class Test::Unit::TestCase
  self.quiet_backtrace = false
end

再对着天空大喊一句：薄弱波罗蜜，一切就都恢复原样了，更多信息，请看这里。

无觅猜您也喜欢：
在test:unit中应用BDD	Rails Cookie测试	Merb: Rails的替代者？	NetBeans IDE: Windows平台最好的Ruby IDE
无觅

安装：

script/plugin install http://svn.techno-weenie.net/projects/plugins/simply_bdd/
script/plugin install http://svn.techno-weenie.net/projects/plugins/test_spec_on_rails/

使用：

1. simple_bdd:

context "New User" do
  def setup
  end
  specify "should be invalid without a username" do
  end
end

这相当于：

class NewUserTest < Test::Unit::TestCase
  def setup
  end
  def test_should_be_invalid_without_a_username
  end
end

test_spec_on_rails:

#model
@user.should.validate
@user.should.be.validated
#Redirection
response.should.be.redirected # assert_response :redirect
response.should.redirect foo_url(@foo) # assert_redirected_to foo_url(@foo)
should.redirect_to :action => 'show' # because "response" is optional

更多请参看README.

另外DUST也提供和simple_bdd类似的功能。

无觅猜您也喜欢：

quietbacktrace: 让Test::Unit不再唐僧

Rails应用性能优化3

使用YSlow提升应用性能

Rails应用性能优化4

无觅

def comment
  @post = Post.find(params[:id])
  @comment = Comment.new(params[:comment])
  respond_to do |format|
    if @comment.save && @post.comments.concat(@comment)
      flash[:notice] = 'Comment was successfully updated.'
      format.html { redirect_to post_path(@blogger, @post) }
      format.xml { head k }
    else
      flash[:notice] = 'Error saving comment.'
      format.html { redirect_to post_path(@blogger, @post) }
      format.xml { render ml => @comment.errors.to_xml }
    end
  end
end

我需要确保我能够对帖子发布评论：

it "should add comments to the post" do
  post = posts(:one)
  comment_count = post.comments.size
  post :comment, :id => 1, :comment => {:title => 'test', :body => 'very nice!'}
  post.reload
  post.comments.size.should == comment_count + 1
  flash[:notice].should match(/success/i)
  response.should redirect_to(posts_path(@blogger, @post))
end

看起来不错？但这个测试例实际测试了许多不该它来测试的东西，它应该只测试action，但是它却连同model也一并测试了，并且它还依赖于fixtures。

我们可以使用一个mock object来代替真是的post对象：

post = mock_model(Post)

现在我们需要关注comment的第一行：

@post = Post.find(params[:id])

我希望这一行在测试时得到执行，但是不用测试，我也知道这行代码肯定是工作正常的，因此，我现在要做的就是让这行代码返回我在上面创建的那个mock object：

Post.should_receive(:find).and_return(post)

或者这样：

Post.stub!(:find).and_return(post)

但 是请注意，should_receive和stub!虽然都可以让Post.find返回我的mock object，但是他们的区别却是巨大的，如果我调用了stub!，但是却没有调用Post.find，stub!不会抱怨什么，因为它是个老实人，但是 should_receive就不用，如果我告诉should_receive我会调用Post.find，但却没调用，那问题就来了，它会抱怨为何我不 守信用，并抛出一个异常来警告我。

但是跟在他们后面的and_return的作用却是完全相同的，它让你mock或者stub的方法返回你传递给它的任何参数。

接下来，我需要对comment做同样的事情：

comment = mock_model(Comment)
Comment.should_receive(:new).with({:title => 'comment title', :body => 'comment body'}).and_return(comment)

接下来我需要关注这条语句：

if @comment.save && @post.comments.concat(@comment)

同上面一样：

comment.should_receive(:save).and_return(true)
post.comments.should_receive(:concat).and_return(comment)

下面我需要测试flash及redirect：

flash[:notice].should match(/success/i)
response.should redirect_to(posts_path(blogger, post))

现在，还剩下最后一件事情：这个action必须使用blogger变量来进行重定向，这里我使用instance_variable_set来完成这个工作：

blogger = mock_model(User)
@controller.instance_variable_set(:@blogger, blogger)

基本就是这样，下面是将部分代码放入before中的完整代码：

describe PostsController, "comment", :behaviour_type => :controller, do
  before do
    @post = mock_model(Post)
    Post.should_receive(:find).and_return(@post)
    @comment = mock_model(Comment)
    Comment.should_receive(:new).with.({:title => 'comment title', :body => 'comment body'}).and_return(@comment)
    @blogger = mock_model(User)
    @controller.instance_variable_set(:@blogger, @blogger)
  end
  it "should add comments to the post" do
    @comment.should_receive(:save).and_return(true)
    @post.comments.should_receive(:concat).and_return(@comment)
    post :comment, :id => @post.id, :comment => {:title => 'comment title', :body => 'comment body'}
    flash[:notice].should match(/success/i)
    response.should redirect_to(posts_path(@blogger, @post))
  end
end

相关帖子：

• RSpec进阶指南(1): rspec_scaffold
• RSpec进阶指南(2): 基础知识

无觅猜您也喜欢：

RSpec进阶指南(2): 基础知识

RSpec进阶指南(1): rspec_scaffold

RSpec Story简介

RSpec简明指南

无觅

Strings:

'foo'.should == 'foo'
'foo'.should === 'foo'
'foo'.should_not equal('foo')
''.should be_empty
'foo with bar'.should include('with')
'http://fr.ivolo.us'.should match(/http:\/\/.+/i)
nil.should be_nil

Numbers:

100.should < 200
200.should >= 100
(200 - 100).should == 100

# (100 - 80) is less than 21
100.should be_close(80,21)

Arrays:

[1,2,3].should have(3).items
[].should be_empty
[1,2,3].should include(2)

Hashes:

{}.should be_empty
{:post => {:title => 'test'}}.should have_key(:post)
{:post => {:title => 'test'}}.should_not have_key(:title)
false.should be_false
true.should be_true

Records:

# assuming @post = Post.new(:title => 'test')
@post.should be_instance_of(Post)
@post.should respond_to(:title)

相关文章：

• RSpec进阶指南(1): rspec_scaffold
• RSpec简明指南

无觅猜您也喜欢：

RSpec进阶指南(1): rspec_scaffold

RSpec进阶指南（三）：Mocking

RSpec Story简介

RSpec简明指南

无觅

要使用RSpec来测试你的Rails应用，有一个比较简单的方法，使用rspec_scaffold生成器来代替scaffold_resource（Rails2.0中已更名为scaffold）。

$ ./script/generate rspec_scaffold post title:string body:text author:integer created_at:datetime updated_at:datetime
     exists app/models/
     exists app/controllers/
     exists app/helpers/
     create app/views/posts
     create spec/controllers/
     create spec/models/
     create spec/helpers/
     create spec/fixtures/
     create spec/views/posts
     create spec/controllers/posts_controller_spec.rb
     create app/controllers/posts_controller.rb
     create spec/helpers/posts_helper_spec.rb
     create app/helpers/posts_helper.rb
     create app/views/posts/index.rhtml
     create app/views/posts/show.rhtml
     create app/views/posts/new.rhtml
     create app/views/posts/edit.rhtml
     create app/models/post.rb
     create spec/fixtures/posts.yml
     create spec/models/post_spec.rb
     create spec/views/posts/edit.rhtml_spec.rb
     create spec/views/posts/index.rhtml_spec.rb
     create spec/views/posts/new.rhtml_spec.rb
     create spec/views/posts/show.rhtml_spec.rb
     create db/migrate
     create db/migrate/001_create_posts.rb
     route map.resources :posts

就像你看到的，rspec_scaffold会为你生成model，controller，helper，views以及其它一些辅助的spec文件。让我们先来看看spec/models/post_spec.rb吧：

# this is called a context (or a describe block)
require File.dirname(__FILE__) + '/../spec_helper'
describe Post do
  before(:each) do
    @post = Post.new
  end
  # this is called an example:
  it "should be valid" do
    @post.should be_valid
  end
end

首先，我们需要注意，所有的specs都拥有behaviour_type属性，spec使用它来确定需要加载哪些spec helpers，我们可以通过下面的语句来指定它：

describe Post, "a useful comment", :behaviour_type => :model do

这将会加载测试model所需的helpers，让我们继续回到post_spec.rb，这段代码其实没有任何意义，因为post model什么也没有做，所以让我们先来给post model加一些validations：

class Post < ActiveRecord::Base
  validates_presence_of :title
  validates_presence_of :body
  validates_uniqueness_of :title
end

一些人可能认为测试validations没什么用，因为几乎可以肯定它们肯定是工作正常的，但是作为应用的最底层代码，我还是愿意对它们的行为进行检查，以确保它们符合我的预期：

describe Post do
  before(:each) do
    @post = Post.new(valid_post_hash) # grabs the hash below
  end
  it "should be valid" do
    @post.should be_valid
  end
  it "should not be valid without a title" do
    @post.title = ''
    @post.should_not be_valid
  end
  it "should not be valid without a body" do
    @post.body = ''
    @post.should_not be_valid
  end
  def valid_post_hash
    {:title => 'test', :body => 'test body'}
  end
end

尽管RSpec的可读性并不是很好，但我认为至少上面这段代码写的相当清楚，应该不需要再进行解释，不过对于几个比较特殊的方法还是要说明一下：

  before(:all) do
  # 会在所有example运行前被调用一次
  end
  before do
  # 与before(:each)相同，会在每个example运行前被调用一次
  end
  after(:each) do
  # 会在每个example运行完后被调用一次
    @post.destroy unless @post.new_record?
  end
  after(:all) do
    # 会在所有examples运行完之后被调用一次
    Post.destroy_all
  end

无觅猜您也喜欢：

RSpec进阶指南(2): 基础知识

RSpec Story简介

RSpec进阶指南（三）：Mocking

RSpec简明指南

无觅

]]>