Logging with PSR-3 to Improve Reusability
可以用composer的autoload来,导入自己写的类库。


composer dump-autoload -o ----------------> 改成 composer update 也可以。
怎么使用co'mposer呢?
1.composer.json

2. index.php

3. 执行 php index.php


---------------------------------------------------------------------------------
Logging is one of the most ubiquitous tasks encountered in PHP. We use logs to track error messages, record important events, and debug problems with our code. In any PHP project, the code is likely to be full of calls to a logging library which handles these actions for us.
Unfortunately, having calls to a logging library scattered throughout our code makes the code dependent on the availability of that library, a clear violation of the Dependency Inversion Principle. Even if we use dependency injection to give our objects access to the logging library, the differences between logging libraries means that it can be difficult and time consuming to switch between them, requiring a major refactoring of our entire codebase.
To promote compatibility between logging libraries, the PHP-FIG group recently released PSR-3, a common interface for logger objects. In this article, I’ll discuss how the logger interface defined by PSR-3 allows us to write reusable code that isn’t dependent on any particular logging implementation.
First, a Quick Primer
Before we look at how PSR-3 can make our code more reusable, it is necessary to understand what PSR-3 is. If you are already familiar with PSR-3, you can skip this section.
The heart of the specification is an interface for logger objects. This interface exposes eight methods for handling messages of different severity, and a generic log() method which can accept an arbitrary severity level.
The eight severity levels supported by PSR-3 are based on RFC 5424, and are described below:
Emergency– the system is unusableAlert– immediate action is requiredCritical– critical conditionsError– errors that do not require immediate attention but should be monitoredWarning– unusual or undesirable occurrences that are not errorsNotice– normal but significant eventsInfo– interesting eventsDebug– detailed information for debugging purposes
Each of the logging methods accepts a message, which must be a string or an object with a __toString() method. An additional argument accepts an array, which may be given to provide contextual information for the log message.
A full explanation of these methods and parameters can be found in the PSR-3 specification.
Getting PSR-3 Files
Getting the files you need to work with PSR-3 is easy – you can find them in the Psr/Log GitHub repository. You can also use Composer to get the files from Packagist. Below is a sample composer.json file to retrieve the Psr/Log files:
{
"require": {
"psr/log": "dev-master"
}
}
How Logging Can Limit Code Reuse
There are many different logging libraries for PHP, each with its own approach to collecting and recording data. While there are some common ideas among them, each library has its own unique set of logging methods. This means that switching between loggers can be challenging, often requiring code changes wherever logging is used. This works against the idea of code reuse and the SOLID principles of object-oriented design. We’re left with a situation that requires us to either declare a dependency on a specific logging library, or avoid logging entirely.
To illustrate this problem more clearly, a concrete example is needed. Let’s say we’re creating a simple Mailer object to handle sending emails. We want the Mailer to log a message whenever an email is sent, and we decide to use the excellent Monologlibrary to handle our logging needs.
<?php
namespace Email;
class Mailer
{
private $logger;
public function __construct($logger)
{
$this->logger = $logger;
}
public function sendEmail($emailAddress)
{
// code to send an email...
// log a message
$this->logger->addInfo("Email sent to $emailAddress");
}
}
We can use this class with the following code:
<?php
// create a Monolog object
$logger = new MonologLogger("Mail");
$logger->pushHandler(new MonologHandlerStreamHandler("mail.log"));
// create the mailer and send an email
$mailer = new EmailMailer($logger);
$mailer->sendEmail("email@example.com");
Running this code will result in a new entry in the mail.log file, recording that the email was sent.
At this point, we might be tempted to say that we’ve written a reusable Mailer object. We’re making the logger available to the Mailer with dependency injection so we can swap out different logger configurations without having to touch our Mailer code. It appears that we’ve successfully followed the SOLID principles and avoided creating any hard dependencies.
But suppose we want to reuse our Mailer class in a different project which uses Analog to handle logging interactions. Now we run into a problem, because Analog doesn’t have an addInfo() method. To log an info level message with Analog, we call Analog::log($message, Analog::INFO).
We could modify our Mailer class to use Analog’s methods, as seen below.
<?php
namespace Email;
class Mailer
{
public function sendEmail($emailAddress)
{
// code to send an email...
// log a message
Analog::log("Email sent to $emailAddress", Analog::INFO);
}
}
We can consume the updated Mailer class using the following code:
<?php
// set up Analog
Analog::handler(AnalogHandlerFile::init("mail.log"));
// create the mailer and send an email
$mailer = new EmailMailer();
$mailer->sendEmail("email@example.com");
While this will work, it is far from ideal. We’ve run into Mailer’s dependency on a specific logging implementation, which requires the class to change whenever a new logger is introduced. This makes the class less reusable and forces us to choose between being dependent on the availability of a specific logger or abandoning logging from within our class entirely.
Using PSR-3 to Avoid the Logger Dependency
As Alejandro Gervasio explained in his excellent article on the subject, the Dependency Inversion Principle tells us that we should depend upon abstractions rather than concretions. In the case of logging, our problem so far has been a lack of a decent abstraction on which to depend.
This is where PSR-3 comes in. PSR-3 is designed to overcome the problem of incompatible logging implementations by providing a universal interface for loggers, aptly named LoggerInterface. By providing an interface which is not bound to any particular implementation, PSR-3 frees us from having to rely on a specific logger – we can instead type-hint against LoggerInterface to acquire a PSR-3 compliant logger. I have updated the Mailer class below to demonstrate this:
<?php
namespace Email;
class Mailer
{
private $logger;
public function __construct(PsrLogLoggerInterface $logger)
{
$this->logger = $logger;
}
public function sendEmail($emailAddress)
{
// code to send an email...
// log a message
$this->logger->info("Email sent to $emailAddress");
}
}
The constructor has been modified to accept an implementor of LoggerInterface, and the sendEmail() method now calls the info() method specified in PSR-3.
Monolog is already PSR-3 compliant, and Analog supplies a wrapper object that implements LoggerInterface, so we can now use both loggers without modifying our Mailer class.
Here’s how you would call the class with Monolog:
<?php
// create a Monolog object
$logger = new MonologLogger("Mail");
$logger->pushHandler(new MonologHandlerStreamHandler("mail.log"));
// create the mailer and send an email
$mailer = new EmailMailer($logger);
$mailer->sendEmail("email@example.com");
And with Analog:
<?php
// create a PSR-3 compatible Analog wrapper
$logger = new AnalogLogger();
$logger->handler(AnalogHandlerFile::init("mail.log"));
// create the mailer and send an email
$mailer = new EmailMailer($logger);
$mailer->sendEmail("email@example.com");
Now we’re able to use our Mailer object with either library without having to edit the Mailer class or change the way we consume it.
Using the Adapter Pattern for Loggers that Don’t Support PSR-3
So far we’ve successfully decoupled the Mailer object from any particular logging implementation by asking for an implementor of LoggerInterface. But what about loggers that have yet to add support for PSR-3? For example, the popular KLoggerlibrary has not been updated for some time and is currently not compatible with PSR-3.
Luckily, it’s simple for us to map the methods exposed by KLogger to those defined by LoggerInterface by harnessing the power of the Adapter Pattern. The supporting files in the Psr/Log repository make it easy to create adapter classes by providing us with an AbstractLogger class that we can extend. The abstract class simply forwards the eight level-specific logging methods defined in LoggerInterface to a generic log() method. By extending the AbstractLoggerclass and defining our own log() method, we can easily create PSR-3 compliant adapters for loggers that do not natively support PSR-3. I’ll demonstrate this below by creating a simple adapter for KLogger:
<?php
namespace Logger;
class KloggerAdapter extends PsrLogAbstractLogger implements PsrLogLoggerInterface
{
private $logger;
public function __construct($logger)
{
$this->logger = $logger;
}
public function log($level, $message, array $context = array())
{
// PSR-3 states that $message should be a string
$message = (string)$message;
// map $level to the relevant KLogger method
switch ($level) {
case PsrLogLogLevel::EMERGENCY:
$this->logger->logEmerg($message, $context);
break;
case PsrLogLogLevel::ALERT:
$this->logger->logAlert($message, $context);
break;
case PsrLogLogLevel::CRITICAL:
$this->logger->logCrit($message, $context);
break;
case PsrLogLogLevel::ERROR:
$this->logger->logError($message, $context);
break;
case PsrLogLogLevel::WARNING:
$this->logger->logWarn($message, $context);
break;
case PsrLogLogLevel::NOTICE:
$this->logger->logNotice($message, $context);
break;
case PsrLogLogLevel::INFO:
$this->logger->logInfo($message, $context);
break;
case PsrLogLogLevel::DEBUG:
// KLogger logDebug() method does not accept a second
// argument
$this->logger->logDebug($message);
break;
default:
// PSR-3 states that we must throw a
// PsrLogInvalidArgumentException if we don't
// recognize the level
throw new PsrLogInvalidArgumentException(
"Unknown severity level"
);
}
}
}
The log() method simply maps the LoggerInterface methods to the respective KLogger methods, and KLogger handles the actual logging activity. By wrapping the KLogger class in this way, we’re able to use it without breaking the LoggerInterface contract.
We can now use the KLogger adapter to work with the Mailer class:
<?php
// create a new KLogger object which will log to the "logs" directory
$klogger = KLogger::instance("logs");
// inject KLoggger to a PSR-3 compatible KloggerAdapter
$logger = new LoggerKloggerAdapter($klogger);
// send an email
$mailer = new EmailMailer($logger);
$mailer->sendEmail("email@example.com");
Using the adapter class, we’re able to use KLogger without modifying the Mailer class and still adhering to LoggerInterface.
KLogger doesn’t accept a second argument for debug level messages, so it’s not technically PSR-3 compliant even with an adapter. Extending KLogger to make it fully compatible with PSR-3 would be a trivial task, but one which falls outside the scope of this article. However, it is safe to say that using our adapter class gets us very close to full PSR-3 compliance and allows us to use LoggerInterface with the KLogger class.
Conclusion
In this article we’ve seen how using PSR-3 can help us to write logger-agnostic code which does not depend on a particular logging implementation. Many major PHP projects have already added support for PSR-3, including Monolog, Symfony, and Mustache.php, and other big names such as Drupal are discussing how to best integrate it. As PSR-3 makes logging less of a barrier to code reuse, we should see more libraries and frameworks making proper use of logging to provide useful information to developers.
Will PSR-3 affect how you use logging in your applications? Let us know in the comments section below.
Logging with PSR-3 to Improve Reusability的更多相关文章
- [php]laravel框架容器管理的一些要点
本文面向php语言的laravel框架的用户,介绍一些laravel框架里面容器管理方面的使用要点.文章很长,但是内容应该很有用,希望有需要的朋友能看到.php经验有限,不到位的地方,欢迎帮忙指正. ...
- laravel5.1启动详解
laravel的启动过程 如果没有使用过类似Yii之类的框架,直接去看laravel,会有点一脸迷糊的感觉,起码我是这样的.laravel的启动过程,也是laravel的核心,对这个过程有一个了解,有 ...
- Java Interview Reference Guide--reference
Part 1 http://techmytalk.com/2014/01/24/java-interview-reference-guide-part-1/ Posted on January 24, ...
- laravel 服务容器
服务容器,也叫IOC容器,其实包含了依赖注入(DI)和控制反转(IOC)两部分,是laravel的真正核心.其他的各种功能模块比如 Route(路由).Eloquent ORM(数据库 ORM 组件) ...
- laravel启动过程简单解析
:first-child{margin-top:0!important}img.plugin{box-shadow:0 1px 3px rgba(0,0,0,.1);border-radius:3px ...
- laravel框架容器管理
来自http://www.cnblogs.com/chy1000/p/7072936.html 本文面向php语言的laravel框架的用户,介绍一些laravel框架里面容器管理方面的使用要点.文章 ...
- laravel的启动过程---摘自网络博客个人学习之用
如果没有使用过类似Yii之类的框架,直接去看laravel,会有点一脸迷糊的感觉,起码我是这样的.laravel的启动过程,也是laravel的核心,对这个过程有一个了解,有助于得心应手的使用框架,希 ...
- laravel的启动过程解析
laravel的启动过程,也是laravel的核心,对这个过程有一个了解,有助于得心应手的使用框架,希望能对大家有点帮助. 统一入口 laravel框架使用了统一入口,入口文件:/public/ind ...
- laravel框架容器管理的一些要点(转)
本文面向php语言的laravel框架的用户,介绍一些laravel框架里面容器管理方面的使用要点.文章很长,但是内容应该很有用,希望有需要的朋友能看到.php经验有限,不到位的地方,欢迎帮忙指正. ...
随机推荐
- Django项目启动之前执行流程剖析
下面,我们只看看主要的步骤: 1.项目启动,遍历settings下面的INSTALLED_APPS,导入默认配置. INSTALLED_APPS = [ 'django.contrib.admin', ...
- VMWare虚拟机下CentOS 配置网络实现远程连接,提供Web访问
最近使用VMWARE虚拟机当中redhat操作系统,感觉直接使用很不方便,于是就决定配置下redhat网络,通过本机远程工具SecureCRT来连接redhat使用. 环境说明:本机操 ...
- 使用jQuery的插件qrcode生成二维码(静态+动态生成)及常见问题解决方法
一.简介 1.说明 qrcode其实是通过使用jQuery实现图形渲染,画图,支持canvas(HTML5)和table两种方式,您可以到https://github.com/jeromeetienn ...
- Codeforces Round #254 (Div. 1) A. DZY Loves Physics 智力题
A. DZY Loves Physics 题目连接: http://codeforces.com/contest/444/problem/A Description DZY loves Physics ...
- 体感设备:因特尔 Intel RealSense R200,乐视LeTV Pro Xtion和Orb奥比中光bec Astra比较
最近调试三个个厂家的体感设备,第一个是Intel的RealSense R200(参数规格:分辨率:1080p,深度有效距离:0.51-4,USB3.0),第二个是乐视LeTV Pro Xtion(参数 ...
- Go语言Web框架gwk介绍 (五)
Session Go的net/http本身不带session的机制,需要开发人员自行实现,gwk实现了内存中的session存储机制,如果需要将session存在其他地方比如redis或者memcac ...
- 小程序获取openid unionid session_key
.wxml <button bindtap="paytap">授权</button> .js Page({ paytap: function () { va ...
- 丢失或损坏NDF文件如何附加数据库
在论坛看到有人遇到 NDF文件丢失并且没有备份,所以无法成功附加数据库.在网上也看到过很多回答是如果没有NDF就无法附加成功. 其实我自己测试下来即使没有NDF也是可以成功附加的.但是有条件,丢失的N ...
- java基础学习总结——网络编程
一.网络基础概念 首先理清一个概念:网络编程 != 网站编程,网络编程现在一般称为TCP/IP编程.
- 我对NHibernate的感受(1):对延迟加载方式的误解
NHibernate是.NET平台上最著名的ORM框架,虽说出身于Java平台上的Hibernate,但是从外部看来这几乎就是一个.NET平台上的原生产品:有自己的社区,有自己的用户,有自己的商业支持 ...