关于 minitest
minitest 是面向 Rails 项目的 Minitest 测试编写技能,适用于新增、迁移和优化测试。它强调使用 fixtures、减少数据库和浏览器开销,并通过验证行为而非实现来降低脆弱性。
主要功能
- 01按 Rails 测试类型组织测试
- 02优先使用 fixtures 与最少数据
- 03编写关注行为的稳定断言
- 04提供 RSpec 到 Minitest 的映射
- 05涵盖多租户与权限测试
- 06给出测试提速和调试建议
适用场景
- 01为模型、请求或策略编写 Minitest
- 02将 RSpec 测试迁移到 Minitest
- 03排查缓慢或不稳定的测试套件
- 04为 Rails 项目制定测试驱动流程
安装
复制下面的命令,在终端中运行。
$
npx skills add https:/ / github.com/ thinkoodle/ rails-skills --skill minitest安装后在 Claude Code 中输入/minitest确认已加载。
附带文件reference.md
详细介绍
Rails Minitest 专家
使用 Minitest 和 fixture 为 Rails 应用编写高性能、易维护且不脆弱的测试。
理念
核心原则:
- 测试行为,而非实现 - 关注代码做什么,而不是如何实现
- 快速反馈循环 - 优先选择单元测试而非集成测试,优先使用 fixture 而非工厂
- 测试即文档 - 测试名称应描述预期行为
- 尽量减少测试数据 - 只创建必要的数据;2 条记录就足够代表很多条记录
- 使用不脆弱的断言 - 测试结果,而非可能变化的确切值
测试金字塔:
/\ System Tests (Few - critical paths only)
/ \
/____\ Request/Integration Tests (Some)
/ \
/________\ Unit Tests (Many - models, policies, services)
何时使用此技能
- 为 Rails 模型、策略、控制器或请求编写新的 Minitest 测试
- 将 RSpec 测试转换为 Minitest
- 调试运行缓慢或不稳定的测试
- 提升测试套件性能
- 遵循 Rails 测试惯例
- 编写基于 fixture 的测试数据
- 实施 TDD 工作流
指南
步骤 1:确定测试类型
编写测试前,先确定合适的测试类型:
| 测试类型 | 位置 | 适用场景 |
|---|---|---|
| 模型 | test/models/ | 验证、关联、业务逻辑方法 |
| 策略 | test/policies/ | Pundit 授权策略 |
| 请求 | test/requests/ | 完整的 HTTP 请求/响应周期 |
| 控制器 | test/controllers/ | 控制器操作(优先使用请求测试) |
| 系统 | test/system/ | 使用真实浏览器测试关键用户流程 |
| 服务 | test/services/ | 服务对象和复杂操作 |
| 作业 | test/jobs/ | 后台作业行为 |
| 邮件程序 | test/mailers/ | 电子邮件内容和发送 |
步骤 2:检查现有模式
务必先搜索现有测试:
# Find similar test files
rg "class.*Test < " test/
# Find existing fixtures
ls test/fixtures/
# Check for test helpers
cat test/test_helper.rb
cat test/support/*.rb
遵循项目现有惯例 - 保持一致比采用“最佳”模式更重要。
步骤 3:使用 fixture(而非工厂)
fixture 的速度比 Factory Bot 快 10-100 倍。
# AVOID - Factory Bot (slow, implicit)
let(:user) { create(:user) }
let(:project) { create(:project, workspace: workspace) }
# PREFER - Fixtures (fast, explicit)
setup do
@workspace = workspaces(:main_workspace)
@user = users(:admin_user)
@project = projects(:active_project)
end
Fixture 最佳实践:
- 创建用途明确且名称清晰的 fixture
- 使用
<%= %>生成动态值和 UUID - 通过 fixture 名称而非 ID 引用关联
- 保持 fixture 精简 - 只包含必需属性
# test/fixtures/users.yml
admin_user:
id: <%= Digest::UUID.uuid_v5(Digest::UUID::OID_NAMESPACE, "admin_user") %>
email: "[email protected]"
name: "Admin User"
created_at: <%= 1.week.ago %>
member_user:
id: <%= Digest::UUID.uuid_v5(Digest::UUID::OID_NAMESPACE, "member_user") %>
email: "[email protected]"
name: "Member User"
workspace: main_workspace # Reference by fixture name
步骤 4:编写测试结构
标准测试结构:
require "test_helper"
class ModelTest < ActiveSupport::TestCase
setup do
# Load fixtures - MINIMAL setup only
@record = models(:fixture_name)
end
# Group related tests with comments or test naming
# Validation tests
test "requires name" do
@record.name = nil
refute @record.valid?
assert_includes @record.errors[:name], "can't be blank"
end
# Method tests
test "#full_name returns formatted name" do
@record.first_name = "John"
@record.last_name = "Doe"
assert_equal "John Doe", @record.full_name
end
end
步骤 5:遵循性能指南
尽可能避免数据库操作:
# SLOW - Creates database records
test "validates email format" do
user = User.create!(email: "invalid", name: "Test")
refute user.valid?
end
# FAST - Uses in-memory object
test "validates email format" do
user = User.new(email: "invalid", name: "Test")
refute user.valid?
end
尽量减少创建的记录:
# SLOW - Creates 25 records
test "paginates results" do
create_list(:post, 25)
# ...
end
# FAST - Configure pagination threshold for tests
# config/environments/test.rb: Pagy::DEFAULT[:limit] = 2
test "paginates results" do
# Only need 3 records to test pagination with limit of 2
assert_operator posts.count, :>=, 3
# ...
end
尽可能避免浏览器测试:
# SLOW - Full browser simulation
class PostsSystemTest < ApplicationSystemTestCase
test "creates a post" do
visit new_post_path
fill_in "Title", with: "Test"
click_on "Create"
assert_text "Post created"
end
end
# FAST - Request test (no browser)
class PostsRequestTest < ActionDispatch::IntegrationTest
test "creates a post" do
post posts_path, params: { post: { title: "Test" } }
assert_response :redirect
follow_redirect!
assert_response :success
end
end
步骤 6:编写不脆弱的断言
测试行为,而非确切值:
# BRITTLE - Exact timestamp match
assert_equal "2025-01-15T10:00:00Z", response["created_at"]
# ROBUST - Just verify presence
assert response["created_at"].present?
# BRITTLE - Exact error message
assert_equal "Name can't be blank", record.errors.full_messages.first
# ROBUST - Check for key content
assert_includes record.errors[:name], "can't be blank"
使用包含式断言:
# BRITTLE - Exact match
assert_equal({ id: 1, name: "Test", email: "[email protected]" }, response)
# ROBUST - Check key attributes only
assert_equal 1, response[:id]
assert_equal "Test", response[:name]
# OR
assert response.slice(:id, :name) == { id: 1, name: "Test" }
步骤 7:处理多租户
对于使用 acts_as_tenant 的项目,务必在租户上下文中执行:
# WRONG - Missing tenant context
test "admin can view project" do
assert policy(@admin, @project).show?
end
# CORRECT - Proper tenant scoping
test "admin can view project" do
with_workspace(@workspace) do
assert policy(@admin, @project).show?
end
end
步骤 8:正确测试权限流程
务必先测试拒绝,再授予权限,之后测试允许:
test "member requires permission to create" do
with_workspace(@workspace) do
# 1. Test denial WITHOUT permission
refute policy(@member, Project).create?
# 2. Grant permission
set_workspace_permissions(@member, @workspace, :allowed_to_create_projects)
# 3. Test allow WITH permission
assert policy(@member, Project).create?
end
end
快速参考
断言对照(RSpec 到 Minitest)
| RSpec | Minitest |
|---|---|
expect(x).to eq(y) | assert_equal y, x |
expect(x).to be_truthy | assert x |
expect(x).to be_falsey | refute x |
expect(x).to be_nil | assert_nil x |
expect(arr).to include(x) | assert_includes arr, x |
expect(arr).not_to include(x) | refute_includes arr, x |
expect { }.to change { X.count }.by(1) | assert_difference "X.count", 1 do ... end |
expect { }.to raise_error(E) | assert_raises(E) { ... } |
expect(x).to be_valid | assert x.valid? |
expect(x).not_to be_valid | refute x.valid? |
expect(x).to match(/pattern/) | assert_match /pattern/, x |
Rails 专用断言
# Record changes
assert_difference "Post.count", 1 do
Post.create!(title: "Test")
end
assert_no_difference "Post.count" do
Post.new.save # Invalid, doesn't save
end
# Value changes
assert_changes -> { post.reload.title }, from: "Old", to: "New" do
post.update!(title: "New")
end
# Response assertions
assert_response :success
assert_response :redirect
assert_redirected_to post_path(post)
# DOM assertions
assert_select "h1", "Expected Title"
assert_select ".post", count: 3
# Query assertions
assert_queries_count(2) { User.find(1); User.find(2) }
assert_no_queries { cached_value }
测试文件模板
模型测试:
require "test_helper"
class UserTest < ActiveSupport::TestCase
setup do
@user = users(:active_user)
end
test "valid fixture" do
assert @user.valid?
end
test "requires email" do
@user.email = nil
refute @user.valid?
assert_includes @user.errors[:email], "can't be blank"
end
test "#display_name returns formatted name" do
@user.name = "John Doe"
assert_equal "John Doe", @user.display_name
end
end
请求测试:
require "test_helper"
class PostsRequestTest < ActionDispatch::IntegrationTest
setup do
@user = users(:active_user)
@post = posts(:published_post)
sign_in @user
end
test "GET /posts returns success" do
get posts_path
assert_response :success
end
test "POST /posts creates record" do
assert_difference "Post.count", 1 do
post posts_path, params: { post: { title: "New Post", body: "Content" } }
end
assert_redirected_to post_path(Post.last)
end
test "POST /posts with invalid data returns error" do
assert_no_difference "Post.count" do
post posts_path, params: { post: { title: "" } }
end
assert_response :unprocessable_entity
end
end
策略测试:
require "test_helper"
class PostPolicyTest < ActiveSupport::TestCase
include PolicyTestHelpers
setup do
@workspace = workspaces(:main_workspace)
@admin = users(:admin_user)
@member = users(:member_user)
@post = posts(:workspace_post)
Current.user = nil
end
test "admin can always edit" do
with_workspace(@workspace) do
assert policy(@admin, @post).edit?
end
end
test "member requires permission to edit" do
with_workspace(@workspace) do
refute policy(@member, @post).edit?
set_workspace_permissions(@member, @workspace, :allowed_to_edit_posts)
assert policy(@member, @post).edit?
end
end
test "scope excludes other workspace posts" do
with_workspace(@other_workspace) do
scope = PostPolicy::Scope.new(@admin, Post.all).resolve
refute_includes scope, @post
end
end
end
性能优化检查清单
提交测试前,请确认:
- 使用 Fixtures,而不是工厂
- 无需数据库时,使用
User.new而不是User.create - 在内存对象上测试验证错误
- Fixture 数据最少化(仅包含所需内容)
- 尽可能使用请求测试,而不是系统测试
- 将测试用的分页阈值设得较低
- Fixture 中没有不必要的关联
- 在测试环境中将 BCrypt 成本设为最低
- 在测试环境中禁用日志记录
- 适用时使用
build_stubbed模式
应避免的反模式
- 测试实现细节 - 测试结果,而不是内部方法调用
- 设置过于复杂 - 如果设置超过 10 行,请重构为 Fixtures
- 测试间共享状态 - 每个测试都应相互独立
- 测试私有方法 - 仅测试公共接口
- 脆弱的断言 - 不要断言时间戳、确切错误或顺序
- 系统测试过多 - 仅为关键用户路径保留
- 缺少负向测试 - 始终测试不应成功或应被拒绝的情况
- 工厂级联 - 避免创建多个关联记录的工厂
运行测试
# Run all tests
bin/rails test
# Run specific file
bin/rails test test/models/user_test.rb
# Run specific test by line
bin/rails test test/models/user_test.rb:42
# Run specific test by name
bin/rails test -n "test_requires_email"
# Run directory
bin/rails test test/policies/
# Run with verbose output
bin/rails test -v
# Run in parallel
bin/rails test --parallel
# Run with coverage
COVERAGE=true bin/rails test
# Run and fail fast
bin/rails test --fail-fast
调试提示
# Print response body in request tests
puts response.body
# Print validation errors
pp @record.errors.full_messages
# Use breakpoint (Rails 7+)
debugger
# Check SQL queries
ActiveRecord::Base.logger = Logger.new(STDOUT)
# Inspect fixture data
pp users(:admin_user).attributes
评论
支持 Markdown