HSA Test User c16dcac362 User import | %!s(int64=7) %!d(string=hace) años | |
---|---|---|
.. | ||
.github | %!s(int64=7) %!d(string=hace) años | |
src | %!s(int64=7) %!d(string=hace) años | |
tests | %!s(int64=7) %!d(string=hace) años | |
.editorconfig | %!s(int64=7) %!d(string=hace) años | |
.gitignore | %!s(int64=7) %!d(string=hace) años | |
.travis.yml | %!s(int64=7) %!d(string=hace) años | |
CHANGELOG.md | %!s(int64=7) %!d(string=hace) años | |
CONTRIBUTING.md | %!s(int64=7) %!d(string=hace) años | |
LICENSE | %!s(int64=7) %!d(string=hace) años | |
README.md | %!s(int64=7) %!d(string=hace) años | |
appveyor.yml | %!s(int64=7) %!d(string=hace) años | |
composer.json | %!s(int64=7) %!d(string=hace) años | |
composer.lock | %!s(int64=7) %!d(string=hace) años | |
phpunit.xml.dist | %!s(int64=7) %!d(string=hace) años |
Initialize Symfony Console commands from annotated command class methods.
Currently in use in Robo (1.x+), Drush (9.x+) and Terminus (1.x+).
Symfony Console provides a set of classes that are widely used to implement command line tools. Increasingly, it is becoming popular to use annotations to describe the characteristics of the command (e.g. its arguments, options and so on) implemented by the annotated method.
Extant commandline tools that utilize this technique include:
This library provides routines to produce the Symfony\Component\Console\Command\Command from all public methods defined in the provided class.
Note If you are looking for a very fast way to write a Symfony Console-base command-line tool, you should consider using Robo, which is built on top of this library, and adds additional conveniences to get you going quickly. See Using Robo as a Framework. It is possible to use this project without Robo if desired, of course.
This is a library intended to be used in some other project. Require from your composer.json file:
"require": {
"consolidation/annotated-command": "~2"
},
The public methods of the command class define its commands, and the parameters of each method define its arguments and options. The command options, if any, are declared as the last parameter of the methods. The options will be passed in as an associative array; the default options of the last parameter should list the options recognized by the command.
The rest of the parameters are arguments. Parameters with a default value are optional; those without a default value are required.
class MyCommandClass
{
/**
* This is the my:cat command
*
* This command will concatenate two parameters. If the --flip flag
* is provided, then the result is the concatenation of two and one.
*
* @command my:cat
* @param integer $one The first parameter.
* @param integer $two The other parameter.
* @option arr An option that takes multiple values.
* @option flip Whether or not the second parameter should come first in the result.
* @aliases c
* @usage bet alpha --flip
* Concatenate "alpha" and "bet".
*/
public function myCat($one, $two, $options = ['flip' => false])
{
if ($options['flip']) {
return "{$two}{$one}";
}
return "{$one}{$two}";
}
}
The $options
array must be an associative array whose key is the name of the option, and whose value is one of:
false
, which indicates that the option takes no value.No other values should be used for the default value. For example, $options = ['a' => 1]
is incorrect; instead, use $options = ['a' => '1']
. Similarly, $options = ['a' => true]
is unsupported, or at least not useful, as this would indicate that the value of --a
was always true
, whether or not it appeared on the command line.
Commandfiles may provide hooks in addition to commands. A commandfile method that contains a @hook annotation is registered as a hook instead of a command. The format of the hook annotation is:
@hook type commandname|annotation
The commandname may be the command's primary name (e.g. my:command
), it's method name (e.g. myCommand) or any of its aliases.
If an annotation is given instead, then this hook function will run for all commands with the specified annotation.
There are ten types of hooks supported:
Most of these also have "pre" and "post" varieties, to give more flexibility vis-a-vis hook ordering (and for consistency). Note that many validate, process and alter hooks may run, but the first status or extract hook that successfully returns a result will halt processing of further hooks of the same type.
Each hook has an interface that defines its calling conventions; however, any callable may be used when registering a hook, which is convenient if versions of PHP prior to 7.0 (with no anonymous classes) need to be supported.
The command-event hook is called via the Symfony Console command event notification callback mechanism. This happens prior to event dispatching and command / option validation. Note that Symfony does not allow the $input object to be altered in this hook; any change made here will be reset, as Symfony re-parses the object. Changes to arguments and options should be done in the initialize hook (non-interactive alterations) or the interact hook (which is naturally for interactive alterations).
The option event hook (OptionHookInterface) is called for a specific command, whenever it is executed, or its help command is called. Any additional options for the command may be added here by instantiating and returnng an InputOption array.
The initialize hook (InitializeHookInterface) runs prior to the interact hook. It may supply command arguments and options from a configuration file or other sources. It should never do any user interaction.
The interact hook (InteractorInterface) runs prior to argument and option validation. Required arguments and options not supplied on the command line may be provided during this phase by prompting the user. Note that the interact hook is not called if the --no-interaction flag is supplied, whereas the command-event hook and the inject-configuration hook are.
The purpose of the validate hook (ValidatorInterface) is to ensure the state of the targets of the current command are usabe in the context required by that command. Symfony has already validated the arguments and options prior to this hook. It is possible to alter the values of the arguments and options if necessary, although this is better done in the configure hook. A validation hook may take one of several actions:
The validate hook may change the arguments and options of the command by modifying the Input object in the provided CommandData parameter. Any number of validation hooks may run, but if any fails, then execution of the command stops.
The command hook is provided for semantic purposes. The pre-command and command hooks are equivalent to the post-validate hook, and should confirm to the interface (ValidatorInterface). All of the post-validate hooks will be called before the first pre-command hook is called. Similarly, the post-command hook is equivalent to the pre-process hook, and should implement the interface (ProcessResultInterface).
The command callback itself (the method annotated @command) is called after the last command hook, and prior to the first post-command hook.
The process hook (ProcessResultInterface) is specifically designed to convert a series of processing instructions into a final result. An example of this is implemented in Robo; if a Robo command returns a TaskInterface, then a Robo process hook will execute the task and return the result. This allows a pre-process hook to alter the task, e.g. by adding more operations to a task collection.
The process hook should not be used for other purposes.
An alter hook (AlterResultInterface) changes the result object. Alter hooks should only operate on result objects of a type they explicitly recognize. They may return an object of the same type, or they may convert the object to some other type.
If something goes wrong, and the alter hooks wishes to force the command to fail, then it may either return a CommandError object, or throw an exception.
The status hook (StatusDeterminerInterface) is responsible for determing whether a command succeeded (status code 0) or failed (status code > 0). The result object returned by a command may be a compound object that contains multiple bits of information about the command result. If the result object implements ExitCodeInterface, then the getExitCode()
method of the result object is called to determine what the status result code for the command should be. If ExitCodeInterface is not implemented, then all of the status hooks attached to this command are executed; the first one that successfully returns a result will stop further execution of status hooks, and the result it returned will be used as the status result code for this operation.
If no status hook returns any result, then success is presumed.
The extract hook (ExtractOutputInterface) is responsible for determining what the actual rendered output for the command should be. The result object returned by a command may be a compound object that contains multiple bits of information about the command result. If the result object implements OutputDataInterface, then the getOutputData()
method of the result object is called to determine what information should be displayed to the user as a result of the command's execution. If OutputDataInterface is not implemented, then all of the extract hooks attached to this command are executed; the first one that successfully returns output data will stop further execution of extract hooks.
If no extract hook returns any data, then the result object itself is printed if it is a string; otherwise, no output is emitted (other than any produced by the command itself).
Commands can define their own custom events; to do so, they need only implement the CustomEventAwareInterface, and use the CustomEventAwareTrait. Event handlers for each custom event can then be defined using the on-event hook.
A handler using an on-event hook looks something like the following:
/**
* @hook on-event custom-event
*/
public function handlerForCustomEvent(/* arbitrary parameters, as defined by custom-event */)
{
// do the needful, return what custom-event expects
}
Then, to utilize this in a command:
class MyCommands implements CustomEventAwareInterface
{
use CustomEventAwareTrait;
/**
* @command my-command
*/
public myCommand($options = [])
{
$handlers = $this->getCustomEventHandlers('custom-event');
// iterate and call $handlers
}
}
It is up to the command that defines the custom event to declare what the expected parameters for the callback function should be, and what the return value is and how it should be used.
The replace-command (ReplaceCommandHookInterface) hook permits you to replace a command's method with another method of your own.
For instance, if you'd like to replace the foo:bar
command, you could utilize the following code:
<?php
class MyReplaceCommandHook {
/**
* @hook replace-command foo:bar
*
* Parameters must match original command method.
*/
public function myFooBarReplacement($value) {
print "Hello $value!";
}
}
If a command method returns an integer, it is used as the command exit status code. If the command method returns a string, it is printed.
If the Consolidation/OutputFormatters project is used, then users may specify a --format option to select the formatter to use to transform the output from whatever form the command provides to a string. To make this work, the application must provide a formatter to the AnnotatedCommandFactory. See API Usage below.
The Annotated-Command project is completely agnostic to logging. If a command wishes to log progress, then the CommandFile class should implement LoggerAwareInterface, and the Commandline tool should inject a logger for its use via the LoggerAwareTrait setLogger()
method. Using Robo is recommended.
If you want to use annotations, but still want access to the Symfony Command, e.g. to get a reference to the helpers in order to call some legacy code, you may create an ordinary Symfony Command that extends \Consolidation\AnnotatedCommand\AnnotatedCommand, which is a \Symfony\Component\Console\Command\Command. Omit the configure method, and place your annotations on the execute()
method.
It is also possible to add InputInterface or OutputInterface parameters to any annotated method of a command file.
If you would like to use Annotated Commands to build a commandline tool, it is recommended that you use Robo as a framework, as it will set up all of the various command classes for you. If you would like to integrate Annotated Commands into some other framework, see the sections below.
To use annotated commands in an application, pass an instance of your command class in to AnnotatedCommandFactory::createCommandsFromClass(). The result will be a list of Commands that may be added to your application.
$myCommandClassInstance = new MyCommandClass();
$commandFactory = new AnnotatedCommandFactory();
$commandFactory->setIncludeAllPublicMethods(true);
$commandFactory->commandProcessor()->setFormatterManager(new FormatterManager());
$commandList = $commandFactory->createCommandsFromClass($myCommandClassInstance);
foreach ($commandList as $command) {
$application->add($command);
}
You may have more than one command class, if you wish. If so, simply call AnnotatedCommandFactory::createCommandsFromClass() multiple times.
If you do not wish every public method in your classes to be added as commands, use AnnotatedCommandFactory::setIncludeAllPublicMethods(false)
, and only methods annotated with @command will become commands.
Note that the setFormatterManager()
operation is optional; omit this if not using Consolidation/OutputFormatters.
A CommandInfoAltererInterface can be added via AnnotatedCommandFactory::addCommandInfoAlterer(); it will be given the opportunity to adjust every CommandInfo object parsed from a command file prior to the creation of commands.
A discovery class, CommandFileDiscovery, is also provided to help find command files on the filesystem. Usage is as follows:
$discovery = new CommandFileDiscovery();
$myCommandFiles = $discovery->discover($path, '\Drupal');
foreach ($myCommandFiles as $myCommandClass) {
$myCommandClassInstance = new $myCommandClass();
// ... as above
}
For a discussion on command file naming conventions and search locations, see https://github.com/consolidation/annotated-command/issues/12.
If different namespaces are used at different command file paths, change the call to discover as follows:
$myCommandFiles = $discovery->discover(['\Ns1' => $path1, '\Ns2' => $path2]);
As a shortcut for the above, the method discoverNamespaced()
will take the last directory name of each path, and append it to the base namespace provided. This matches the conventions used by Drupal modules, for example.
The Output Formatters project supports automatic formatting of tabular output. In order for wordwrapping to work correctly, the terminal width must be passed in to the Output Formatters handlers via FormatterOptions::setWidth()
.
In the Annotated Commands project, this is done via dependency injection. If a PrepareFormatter
object is passed to CommandProcessor::addPrepareFormatter()
, then it will be given an opportunity to set properties on the FormatterOptions
when it is created.
A PrepareTerminalWidthOption
class is provided to use the Symfony Application class to fetch the terminal width, and provide it to the FormatterOptions. It is injected as follows:
$terminalWidthOption = new PrepareTerminalWidthOption();
$terminalWidthOption->setApplication($application);
$commandFactory->commandProcessor()->addPrepareFormatter($terminalWidthOption);
To provide greater control over the width used, create your own PrepareTerminalWidthOption
subclass, and adjust the width as needed.
In addition to the hooks provided by the hook manager, there are additional callbacks available to alter the way the annotated command library operates.
Factory listeners are notified every time a command file instance is used to create annotated commands.
public function AnnotatedCommandFactory::addListener(CommandCreationListenerInterface $listener);
Listeners can be used to construct command file instances as they are provided to the command factory.
An option provider is given an opportunity to add options to a command as it is being constructed.
public function AnnotatedCommandFactory::addAutomaticOptionProvider(AutomaticOptionsProviderInterface $listener);
The complete CommandInfo record with all of the annotation data is available, so you can, for example, add an option --foo
to every command whose method is annotated @fooable
.
CommandInfo alterers can adjust information about a command immediately before it is created. Typically, these will be used to supply default values for annotations custom to the command, or take other actions based on the interfaces implemented by the commandfile instance.
public function alterCommandInfo(CommandInfo $commandInfo, $commandFileInstance);
The existing solutions used their own hand-rolled regex-based parsers to process the contents of the DocBlock comments. consolidation/annotated-command uses the phpdocumentor/reflection-docblock project (which is itself a regex-based parser) to interpret DocBlock contents.