/monolog-cascade

Configure multiple loggers and handlers in the blink of an eye

Primary LanguagePHPMIT LicenseMIT

Monolog Cascade Build Status Coverage Status

What is Monolog Cascade?

Monolog Cascade is a Monolog extension that allows you to set up and configure multiple loggers and handlers from a single config file.

It's been inspired by the logging.config Python module.


Installation

Add monolog-cascade as a requirement in your composer.json file or run

$ composer require theorchard/monolog-cascade

Note: Monolog Cascade requires PHP 5.3.9 or higher.

Usage

<?php
use Cascade\Cascade;

// configure your loggers
Cascade::fileConfig('path/to/some/config.yaml');

// or use php array
$config = require 'config.php';
Cascade::fileConfig($config);

Then just use your logger as shown below

Cascade::getLogger('myLogger')->info('Well, that works!');
Cascade::getLogger('myLogger')->error('Maybe not...');

Configuring your loggers

Monolog Cascade supports the following config formats:

  • Yaml
  • JSON
  • PHP File returning Array
  • PHP Array

Configuration structure

Here is a sample Yaml config file:

formatters:
    dashed:
        class: Monolog\Formatter\LineFormatter
        format: "%datetime%-%channel%.%level_name% - %message%\n"
handlers:
    console:
        class: Monolog\Handler\StreamHandler
        level: DEBUG
        formatter: dashed
        processors: [memory_processor]
        stream: php://stdout
    info_file_handler:
        class: Monolog\Handler\StreamHandler
        level: INFO
        formatter: dashed
        stream: ./example_info.log
processors:
    web_processor:
        class: Monolog\Processor\WebProcessor
    memory_processor:
        class: Monolog\Processor\MemoryUsageProcessor
loggers:
    myLogger:
        handlers: [console, info_file_handler]
        processors: [web_processor]

Here is a sample PHP config file:

<?php

return array(
    'version' => 1,

    'formatters' => array(
        'spaced' => array(
            'format' => "%datetime% %channel%.%level_name%  %message%\n",
            'include_stacktraces' => true
        ),
        'dashed' => array(
            'format' => "%datetime%-%channel%.%level_name% - %message%\n"
        ),
    ),
    'handlers' => array(
        'console' => array(
            'class' => 'Monolog\Handler\StreamHandler',
            'level' => 'DEBUG',
            'formatter' => 'spaced',
            'stream' => 'php://stdout'
        ),

        'info_file_handler' => array(
            'class' => 'Monolog\Handler\StreamHandler',
            'level' => 'INFO',
            'formatter' => 'dashed',
            'stream' => './demo_info.log'
        ),

        'error_file_handler' => array(
            'class' => 'Monolog\Handler\StreamHandler',
            'level' => 'ERROR',
            'stream' => './demo_error.log',
            'formatter' => 'spaced'
        )
    ),
    'processors' => array(
        'tag_processor' => array(
            'class' => 'Monolog\Processor\TagProcessor'
        )
    ),
    'loggers' => array(
        'my_logger' => array(
            'handlers' => array('console', 'info_file_handler')
        )
    )
);

More information on how the Cascade config parser loads and reads the parameters:

Only the loggers key is required. If formatters and/or handlers are ommitted, Monolog's default will be used. processors is optional and if ommitted, no processors will be used. (See the "Optional Keys" section further below).

Other keys are optional and would be interpreted as described below:

  • formatters - the derived associative array (from the Yaml or JSON) in which each key is the formatter identifier holds keys/values to configure your formatters. The only reserved key is class and it should contain the classname of the formatter you would like to use. Other parameters will be interpreted as constructor parameters for that class and passed in when the formatter object is instanced by the Cascade config loader.
    If some parameters are not present in the constructor, they will be treated as extra parameters and Cascade will try to interpret them should they match any custom handler functions that are able to use them. (see Extra Parameters section below)

    If class is not provided Cascade will default to Monolog\Formatter\LineFormatter

  • handlers - the derived associative array (from the Yaml or JSON) in which each key is the handler identifier holds keys/values to configure your handlers.
    The following keys are reserved:

    • class (optional): classname of the handler you would like to use
    • formatter (optional): formatter identifier that you have defined
    • processors (optional): array of processor identifiers that you have defined
    • handlers (optional): array of handler identifiers that you have defined
    • handler (optional): single handler identifier that you have defined

    Other parameters will be interpreted as constructor parameters for that Handler class and passed in when the handler object is instantiated by the Cascade config loader.
    If some parameters are not present in the constructor, they will be interpreted as extra parameters and Cascade will try to interpret them should they match any custom handler functions that are able to use them. (see Extra Parameters section below)

    If class is not provided Cascade will default to Monolog\Handler\StreamHandler

  • processors - the derived associative array (from the Yaml or JSON) in which each key is the processor identifier holds keys/values to configure your processors.
    The following key is reserved:

    • class (required): classname of the processor you would like to use
  • loggers - the derived array (from the Yaml or JSON) in which each key is the logger identifier may contain only a handlers key and/or a processors key. You can decide what handler(s) and/or processor(s) you would like your logger to use.

Note: If you would like to use objects as parameters for your handlers, you can pass a class name (using the class option) with the corresponding arguments just like you would configure your handler. Cascade recursively instantiates and loads those objects as it parses the config file. See this sample config file.

Parameter case

You can use either underscored or camelCased style in your config files, it does not matter. However, it is important that they match the names of the arguments from the constructor method.

public function __construct($level = Logger::ERROR, $bubble = true, $appName = null)

Using a Yaml file:

    level: ERROR,
    bubble: true,
    app_name: "some app that I wrote"

Cascade will camelCase all the names of your parameters internally prior to be passed to the constructors.

Optional keys

formatters, handlers and processors keys are optional. If ommitted Cascade will default to Monolog's default formatter and handler: Monolog\Formatter\LineFormatter and Monolog\Handler\StreamHandler to stderr. If processors is ommitted, your logger(s) won't use any.

Default parameters

If a constructor method provides default value(s) in their declaration, Cascade will look it up and identify those parameters as optional with their default values. It can therefore be ommitted in your config file.

Order of sections and params

Order of the sections within the config file has no impact as long as they are formatted properly.
Order of parameters does not matter either.

Extra parameters (other than constructor's)

You may want to have your Formatters and/or Handlers consume values other than via the constructor. Some methods may be called to do additional set up when configuring your loggers. Cascade interprets those extra params 3 different ways and will try do so in that order:

  1. Instance method
    Your Formatter or Handler has a defined method that takes a param as input. In that case you can write it as follow in your config file:

    formatters:
      spaced:
          class: Monolog\Formatter\LineFormatter
          format: "%datetime% %channel%.%level_name%  %message%\n"
          include_stacktraces: true

    In this example, the LineFormatter class has an includeStacktraces method that takes a boolean. This method will be called upon instantiation.

  2. Public member
    Your Formatter or Handler has a public member that can be set.

    formatters:
        spaced:
            class: Monolog\Formatter\SomeFormatter
            some_public_member: "some value"

    In this example, the public member will be set to the passed in value upon instantiation.

  3. Custom handler function
    See FormatterLoader::initExtraOptionsHandlers and HandlerLoader::initExtraOptionsHandlers. Those methods hold closures that can call instance methods if needed. The closure takes the instance and the parameter value as input.

    self::$extraOptionHandlers = array(
        'Monolog\Formatter\LineFormatter' => array(
            'includeStacktraces' => function ($instance, $include) {
                $instance->includeStacktraces($include);
            }
        )
    );

    You can add handlers at runtime if needed. (i.e. if you write your logger handler for instance)

Running Tests

Just run Phpunit:

$ phpunit tests/

Contributing

This extension is open source. Feel free to contribute and send a pull request!

Make sure your code follows the PSR-2 standards, is documented and has unit tests.

What's next?

  • add support for .ini config files
  • add support for namespaced Loggers with message propagation (through handler inheritance) so children loggers log messages using parent's handlers
  • add more custom function handlers to cover all the possible options of the current Monolog Formatters and Handlers
  • add support for Processors (DONE)
  • add support for DB/Store and other handlers requiring injection into the constructor (issue #30) (DONE)
  • other suggestions?

Symfony Users

You may want to use MonologBundle as it integrates directly with your favorite framework.

Under The Hood

Here is a Medium post if you want to know more about the implementation.