laravel最佳实践

发表于 2018-06-02

着意闻时不肯香，香在无心处。

Single responsibility principle

A class and a method should have only one responsibility.

Bad:

public function getFullNameAttribute()
{
    if (auth()->user() && auth()->user()->hasRole('client') && auth()->user()->isVerified()) {
        return 'Mr. ' . $this->first_name . ' ' . $this->middle_name . ' ' $this->last_name;
    } else {
        return $this->first_name[0] . '. ' . $this->last_name;
    }
}

Good:

public function getFullNameAttribute()
{
    return $this->isVerifiedClient() ? $this->getFullNameLong() : $this->getFullNameShort();
}
public function isVerfiedClient()
{
    return auth()->user() && auth()->user()->hasRole('client') && auth()->user()->isVerified();
}
public function getFullNameLong()
{
    return 'Mr. ' . $this->first_name . ' ' . $this->middle_name . ' ' . $this->last_name;
}
public function getFullNameShort()
{
    return $this->first_name[0] . '. ' . $this->last_name;
}

Fat models, skinny controllers

Put all DB related logic into Eloquent models or into Repository classes if you’re using Query Builder or raw SQL queries.

Bad:

public function index()
{
    $clients = Client::verified()
        ->with(['orders' => function ($q) {
            $q->where('created_at', '>', Carbon::today()->subWeek());
        }])
        ->get();
    return view('index', ['clients' => $clients]);
}

Good:

public function index()
{
    return view('index', ['clients' => $this->client->getWithNewOrders()]);
}
Class Client extends Model
{
    public function getWithNewOrders()
    {
        return $this->verified()
            ->with(['orders' => function ($q) {
                $q->where('created_at', '>', Carbon::today()->subWeek());
            }])
            ->get();
    }
}

Validation

Move validation from controllers to Request classes.

Bad:

public function store(Request $request)
{
    $request->validate([
        'title' => 'required|unique:posts|max:255',
        'body' => 'required',
        'publish_at' => 'nullable|date',
    ]);
    ....
}

Good:

public function store(PostRequest $request)
{    
    ....
}
class PostRequest extends Request
{
    public function rules()
    {
        return [
            'title' => 'required|unique:posts|max:255',
            'body' => 'required',
            'publish_at' => 'nullable|date',
        ];
    }
}

Business logic should be in service class

A controller must have only one responsibility, so move business logic from controllers to service classes.

Bad:

public function store(Request $request)
{
    if ($request->hasFile('image')) {
        $request->file('image')->move(public_path('images') . 'temp');
    }
    
    ....
}

Good:

public function store(Request $request)
{
    $this->articleService->handleUploadedImage($request->file('image'));
    ....
}
class ArticleService
{
    public function handleUploadedImage($image)
    {
        if (!is_null($image)) {
            $image->move(public_path('images') . 'temp');
        }
    }
}

Don’t repeat yourself (DRY)

Reuse code when you can. SRP is helping you to avoid duplication. Also, reuse Blade templates, use Eloquent scopes etc.

Bad:

public function getActive()
{
    return $this->where('verified', 1)->whereNotNull('deleted_at')->get();
}
public function getArticles()
{
    return $this->whereHas('user', function ($q) {
            $q->where('verified', 1)->whereNotNull('deleted_at');
        })->get();
}

Good:

public function scopeActive($q)
{
    return $q->where('verified', 1)->whereNotNull('deleted_at');
}
public function getActive()
{
    return $this->active()->get();
}
public function getArticles()
{
    return $this->whereHas('user', function ($q) {
            $q->active();
        })->get();
}

Prefer to use Eloquent over using Query Builder and raw SQL queries. Prefer collections over arrays

Eloquent allows you to write readable and maintainable code. Also, Eloquent has great built-in tools like soft deletes, events, scopes etc.

Bad:

SELECT *
FROM `articles`
WHERE EXISTS (SELECT *
              FROM `users`
              WHERE `articles`.`user_id` = `users`.`id`
              AND EXISTS (SELECT *
                          FROM `profiles`
                          WHERE `profiles`.`user_id` = `users`.`id`) 
              AND `users`.`deleted_at` IS NULL)
AND `verified` = '1'
AND `active` = '1'
ORDER BY `created_at` DESC

Good:

1	Article::has('user.profile')->verified()->latest()->get();

Mass assignment

Bad:

$article = new Article;
$article->title = $request->title;
$article->content = $request->content;
$article->verified = $request->verified;
// Add category to article
$article->category_id = $category->id;
$article->save();

Good:

1	$category->article()->create($request->all());

Do not execute queries in Blade templates and use eager loading (N + 1 problem)

Bad (for 100 users, 101 DB queries will be executed):

1
2
3

@foreach (User::all() as $user)
    {{ $user->profile->name }}
@endforeach

Good (for 100 users, 2 DB queries will be executed):

$users = User::with('profile')->get();
...
@foreach ($users as $user)
    {{ $user->profile->name }}
@endforeach

Comment your code, but prefer descriptive method and variable names over comments

Bad:

1	if (count((array) $builder->getQuery()->joins) > 0)

Better:

1 2	// Determine if there are any joins. if (count((array) $builder->getQuery()->joins) > 0)

Good:

1	if ($this->hasJoins())

Do not put JS and CSS in Blade templates and do not put any HTML in PHP classes

Bad:

1	let article = `{{ json_encode($article) }}`;

Better:

1	<input id="article" type="hidden" value="{{ json_encode($article) }}">

1	<button class="js-fav-article" data-article="{{ json_encode($article) }}">{{ $article->name }}<button>

In a Javascript file:

1 2	let article = $('#article').val(); The best way is to use specialized PHP to JS package to transfer the data.

Use config and language files, constants instead of text in the code

Bad:

public function isNormal()
{
    return $article->type === 'normal';
}
return back()->with('message', 'Your article has been added!');

Good:

public function isNormal()
{
    return $article->type === Article::TYPE_NORMAL;
}
return back()->with('message', __('app.article_added'));

Use shorter and more readable syntax where possible

Bad:

1 2	$request->session()->get('cart'); $request->input('name');

Good:

1 2	session('cart'); $request->name;

Use IoC container or facades instead of new Class

new Class syntax creates tight coupling between classes and complicates testing. Use IoC container or facades instead.

Bad:

1 2	$user = new User; $user->create($request->all());

Good:

public function __construct(User $user)
{
    $this->user = $user;
}
$this->user->create($request->all());

Do not get data from the .env file directly

Pass the data to config files instead and then use the config() helper function to use the data in an application.

Bad:

1	$apiKey = env('API_KEY');

Good:

// config/api.php
'key' => env('API_KEY'),
// Use the data
$apiKey = config('api.key');

Store dates in the standard format. Use accessors and mutators to modify date format

Bad:

1 2	{{ Carbon::createFromFormat('Y-d-m H-i', $object->ordered_at)->toDateString() }} {{ Carbon::createFromFormat('Y-d-m H-i', $object->ordered_at)->format('m-d') }}

Good:

// Model
protected $dates = ['ordered_at', 'created_at', 'updated_at']
public function getMonthDayAttribute($date)
{
    return $date->format('m-d');
}
// View
{{ $object->ordered_at->toDateString() }}
{{ $object->ordered_at->monthDay }}

Other good practices

Never put any logic in routes files.

Minimize usage of vanilla PHP in Blade templates.

https://github.com/alexeymezenin/laravel-best-practices
https://laravel-china.org/articles/12762/eighteen-best-practices-of-laravel

Laravel中使用多个数据库操作数据

发表于 2018-06-02

人人尽说江南好，游人只合江南老。春水碧于天，画船听雨眠。
垆边人似月，皓腕凝霜雪。未老莫还乡，还乡须断肠。

定义连接

在你的数据库配置文件app/config/ database.php你可以定义任何类型的多个数据库连接。事实上，你可以定义尽可能多的连接，只要你愿意。例如，如果你的应用程序有2个 MySQL数据库中提取数据，则可以分别定义他们两个。

<?php
return array(
    'default' => 'mysql',
    'connections' => array(
        # 主数据库连接
        'mysql' => array(
            'driver'    => 'mysql',
            'host'      => 'host1',
            'database'  => 'database1',
            'username'  => 'user1',
            'password'  => 'pass1'
            'charset'   => 'utf8',
            'collation' => 'utf8_unicode_ci',
            'prefix'    => '',
        ),
        # 次数据连接
        'mysql2' => array(
            'driver'    => 'mysql',
            'host'      => 'host2',
            'database'  => 'database2',
            'username'  => 'user2',
            'password'  => 'pass2'
            'charset'   => 'utf8',
            'collation' => 'utf8_unicode_ci',
            'prefix'    => '',
        ),
    ),
);

我们默认的连接(default)仍设置为 mysql,这意味着，除非我们特别指定，应用程序将继续使用 mysql 连接.

Schema

Schema::connection('mysql2')->create('lp_table', function($table)
{
    $table->increments('id'):
});

DB类

1	DB::connection('mysql2')->table('article')->where...

Eloquent

<?php
class SomeModel extends Eloquent {
    protected $connection = 'mysql2';
    .......
}
?>

或者

<?php
class SomeController extends BaseController {
    public function someMethod()
    {
        $someModel = new SomeModel;
        $someModel->setConnection('mysql2');
        $something = $someModel->find(1);
        return $something;
    }
    .......
}
?>

https://www.itlipeng.cn/2015/12/24/%E5%9C%A8-laravel-%E4%B8%AD%E4%BD%BF%E7%94%A8%E5%A4%9A%E4%B8%AA%E6%95%B0%E6%8D%AE%E5%BA%93%E6%93%8D%E4%BD%9C%E6%95%B0%E6%8D%AE/

browser-sync文件监听失败的解决方案

发表于 2018-06-02

两岸青山相对出，孤帆一片日边来

问题

为了方便实时预览前端开发过程中修改源码后的页面，推荐一个非常实用的工具，browser-sync。

安装使用方式请自行到官网https://browsersync.io/参考文档，仓库地址在这里https://github.com/BrowserSync/browser-sync

GetStart中官网给出的CLI示例命令为：

1	browser-sync start --server --files "css/*.css"

我将其写到到npm命令中，package.json 相关内容如下：

{
  ...
  "scripts": {
    "dev": "browser-sync start --server --files 'css/*.css'"
  },
  "devDependencies": {
    "browser-sync": "^2.18.13"
  },
  ...
}

接着执行 npm run dev，控制台输出一切正常。

然而，当我修改 css/style.css 这个文件的时候，发现浏览器并没有刷新，这说明 browser-sync 并未成功监听 css/*.css 文件的修改。

分析

为此，我翻了一遍 browser-sync 的issue，发现有人遇到相同的问题，也给出了解决方案。

问题出在命令行参数上，仔细对比，我们也会发现：

我写的CLI命令为

1	browser-sync start --server --files 'css/*.css'

而官方CLI命令为

1	browser-sync start --server --files "css/*.css"

问题就出在分号的不同上（browser-sync没能解析单引号的内容）

解决
因此，我将 npm命令中的 ‘ 替换为 \” 即可解决问题。更改后的 package.json 内容如下：

{
  ...
  "scripts": {
    "dev": "browser-sync start --server --files \"css/*.css\""
  },
  "devDependencies": {
    "browser-sync": "^2.18.13"
  },
  ...
}

https://blog.csdn.net/pwc1996/article/details/76849876

Lavavel拒绝软删除的用户邮箱再次验证

发表于 2018-06-01

松下问童子，言师采药去。只在此山中，云深不知处。

When creating user authorization system with soft-deletable data we might encounter a problem: deleted user might try to register with same email address and gets an error that it is in use. What to do in order to prevent it? Here is a quite simple example of how it could be solved.

First of all – by default Laravel migrations for users table have a unique index on email field. This needs to be modified – we need to have unique values on email and deleted_at fields at the same time. So let’s write our migration like this:

public function up()
{
    Schema::create('users', function (Blueprint $table) {
        $table->increments('id');
        $table->string('name');
        $table->string('email');
        $table->string('password');
        $table->rememberToken();
        $table->timestamps();
        $table->softDeletes();
        $table->unique(['email', 'deleted_at']);
    });
}

As you can see, we have a unique index for email and deleted_at at the same time. It is called a composite index. From now on – it is impossible to have two entries that would have identical information in both fields as long as none of them are NULL (except for a situation where your deleted_at field is set to NULL. This is not a bug due the fact that unique allows multiple NULL values in a column: http://dev.mysql.com/doc/refman/5.7/en/create-index.html – see comment below)

A UNIQUE index creates a constraint such that all values in the index must be distinct. An error occurs if you try to add a new row with a key value that matches an existing row. For all engines, a UNIQUE index permits multiple NULL values for columns that can contain NULL. If you specify a prefix value for a column in a UNIQUE index, the column values must be unique within the prefix.

Now, to match a case where our user might not be deleted yet and we don’t want him to register with same email again – we need to change email validation rule:

Open our app/Http/Controllers/Auth/AuthController.php file (Request or other Controller where you have the validation rule) and change your email validation to this:

1	'email' => 'required\|email\|max:255\|unique:users,email,NULL,id,deleted_at,NULL',

You might need to modify table name, column name and etc. for your needs.

And that’s it. Your user is able to register again with the same email as he did before and Laravel will make sure that the email is not within active users. Just don’t forget that when restoring it you need to check if there are no active users with identical email. This might not be the best solution for you, so we made a tiny list of other possible solutions. Feel free to choose any other if this doesn’t work for you or suggest us a new one!

Make a second table where you would store deleted users email and set a random string in the original database. On restore just copy the email back and delete the dummy row.
On user delete (using an observer or manually) prepend users email with a prefix: _deleted or something like that.
… your suggestions?

http://laraveldaily.com/make-soft-deleted-user-email-available/

Laravel之开发测试和生产环境区分

发表于 2018-05-31

不做动机揣测，少做价值判断

Laravel环境检测的方法是很容易的，因为大家都知道Laravel中使用了Dotenv来做文件配置，这就牵扯到了我们的都很熟悉的.env。

比如一个典型的.env是:

APP_ENV=local
APP_DEBUG=true
APP_KEY=base64:3csTuw5O1me1PF4j9xErbUR+seyH1xdf6uRfvjuQ22Q=
APP_URL=http://localhost
DB_CONNECTION=mysql
DB_HOST=127.0.0.1
DB_PORT=3306
DB_DATABASE=db_name
DB_USERNAME=db_user
DB_PASSWORD=db_pass
//...省略

今天我的这篇分享主要介绍两种区分不同环境的方法。

不同环境维护自己的.env文件

这种办法也是Laravel默认的办法，也就是开发环境、测试环境和生产环境各自维护自己的.env文件，也就是说.env不要加到版本控制系统中，通过不同的配置可以做到环境的区分，比如:

开发环境

1 2	APP_ENV=dev //...省略

测试环境

1 2	APP_ENV=staging //...省略

生产环境

1 2	APP_ENV=production //...省略

这样，在我们的代码中即可以通过判断APP_ENV来判断是哪种环境。但是这种办法我个人是不太喜欢的，原因有以下几点:

环境的判断过多的写到代码里面，维护起来很麻烦
如果某次上线需要修改配置文件，则每次上线代码还需要到线上机器修改.env文件，部署起来很麻烦

不同环境加载自己的.env.文件

这里面的.env.文件延伸开来就是.env.dev、.env.test和.env.prod，比如开发环境会自动加载.env.dev，依此类推，那么如果是这样的话，不同环境的机器又怎么知道加载哪个文件呢？其实这里面还是耍了点小聪明，别忘了我们有php.ini

1.在php.ini中追加一行配置
当然开发、测试和生产环境中env所对应的值也需要不一样
1
2
3
4
5
6
//开发环境
env=dev
//测试环境
env=staging
//生产环境
env=production

2.加载不同的配置文件
在bootstrap/app.php文件中添加如下判断，在这里将通过获取php.ini中的env的值，然后从而加载不同的配置文件。

//...省略
$env = get_cfg_var('env');
$env = !empty($env) ? $env : 'production';
if(!defined('APP_MODE')){
    define('APP_MODE', $env);
}
$app->loadEnvironmentFrom('.env.'.$env);
return $app;

3.新建.env.dev、.env.staging和.env.production

好，就到此为止了，当然虽然我非常喜欢第二种办法，但第二种毕竟需要对线上机器做一次大的改动，需要运维同学支持下。但是也需要具体情况具体来选择吧。

http://xuwenzhi.com/2016/07/31/laravel%E4%B9%8B%E5%BC%80%E5%8F%91%E3%80%81%E6%B5%8B%E8%AF%95%E5%92%8C%E7%94%9F%E4%BA%A7%E7%8E%AF%E5%A2%83%E5%8C%BA%E5%88%86/

lnmput@gmail.com

一个记录PHP, Laravel, MySQL, Linux, Redis, ElasticSearch等web开发实践经验的博客

GitHub