Manual:Magic words

MediaWiki extensions

Magic words are a technique for mapping a variety of wiki text strings to a single ID that is associated with a function. Both variables and parser functions use this technique. All text mapped to that ID will be replaced with the return value of the function. The mapping between the text strings and the ID is stored in the variable $magicWords in a file that can be loaded using $wgExtensionMessagesFiles[] .

The default magic words are implemented in CoreParserFunctions.php .

How magic words work edit

Whenever MediaWiki finds text between double braces ({{XXX ...}}) it must decide whether XXX is a variable, parser function, or template. To do so, it asks a series of questions:

  1. Does it have an associated magic word ID? As a first step in resolving markup of the form {{XXX...}}, MediaWiki attempts to translate XXX to a magic word ID. The translation table is defined by $magicWords.
    • If no magic word ID is associated with XXX, XXX is presumed to be a template.

  2. Is it a variable? If a magic word ID is found, MediaWiki next checks to see if it has any parameters.
    • If no parameters are found, MediaWiki checks to see if the magic word ID has been declared as a variable ID. To check this, it retrieves the list of magic words serving by calling MagicWord::getVariableIDs(). This method gets its list of variable IDs from a hard coded list of variable IDs (see Help:Variables ) and from a list of custom variable IDs provided by all functions attached to the hook MagicWordwgVariableIDs .
      • If the magic word ID has been classified as a variable, MediaWiki calls the ParserGetVariableValueSwitch function to get the value associated with the variable name.

  3. Is it a parser function? If there are any parameters or if the magic word ID is missing from the list of variable magic word IDs, then MediaWiki assumes that the magic word is a parser function or template. If the magic word ID is found in the list of parser functions declared via a call to $wgParser->setFunctionHook($magicWordId, $renderingFunctionName), it is treated as a parser function and rendered using the function named $renderingFunctionName. Otherwise, it is presumed to be a template.
By convention:
  • The magic words called variables are capitalised, case-sensitive and do not have space characters.
  • Parserfunctions are prefixed with a hash sign (#), are case insensitive and do not include space characters.

This is however a convention and one not consistently applied (for historic reasons).

  • Variables do not have space characters, but some translations of variables in other languages DO have spaces
  • Variables generally are capitalised and case-sensitive, but some parser functions also use this convention.
  • Some parser functions start with a hash sign, but some do not.

Where possible you should follow the conventions when defining or translating magic words. Magic words are higher in priority than templates, so any magic word defined, will block the usage of that defined name as a template.

Following the conventions avoids adding more and more potential collisions.

Defining magic words edit

For magic words to do their magic we must define two things:

  • a mapping between wiki text and a magic word ID
  • a mapping between a magic word ID and some PHP function that interprets the magic word.

Mapping wiki text to magic word IDs edit

The variable $magicWords is used to associate each magic word ID with a language-dependent array that describes all the text strings that mapped to the magic word ID. Important: This only sets up the back end i18n mapping, you still have to write other code to make MediaWiki use the magic word for anything. Also, make sure that you initialize $magicWords as an empty array before adding language-specific values or you will get errors when trying to load the magic word and will need to rebuild your localization cache before it will work.

The first element of this array is an integer flag indicating whether or not the magic word is case sensitive. The remaining elements are a list of text that should be associated with the magic word ID. If the case sensitive flag is 0, any case variant of the names in the array will match. If the case sensitive flag is 1, only exact case matches will be associated with the magic word ID. Thus the format is $magicWords['en'] = [ 'InternalName' => [ 0, 'NameUserTypes', 'AdditionalAliasUserCanType' ] ];

This association is created by $magicWords in a file registered using $wgExtensionMessagesFiles[] .

In the example below, a Spanish MediaWiki installation will associate the magic word ID 'MAG_CUSTOM' with "personalizado", "custom", "PERSONALIZADO", "CUSTOM" and all other case variants. In an English MediaWiki only "custom" in various case combinations will be mapped to 'MAG_CUSTOM':

File Example.i18n.magic.php:

<?php

$magicWords = [];

$magicWords['en'] = [
	'MAG_CUSTOM' => [ 0, 'custom' ],
];

$magicWords['es'] = [
	'MAG_CUSTOM' => [ 0, 'personalizado' ],
];

In part of the extension.json file:

"ExtensionMessagesFiles": {
	"ExampleMagic": "Example.i18n.magic.php"
}

Note that "ExampleMagic" is a different to the key you would use for a plain internationalization file (normally just the title of the extension, i.e. "Example"). "Magic" has been appended deliberately so one does not overwrite the other.

In inline PHP edit

You can associate magic words inline in PHP rather than through a i18n file. This is useful when defining hooks in LocalSettings.php but should not be done in extensions.

MediaWiki\MediaWikiServices::getInstance()->getContentLanguage()->mMagicExtensions['wikicodeToHtml'] = ['MAG_CUSTOM', 'custom'];

Associating a magic word ID with a PHP function edit

The mechanism for associating magic word IDs with rendering functions depends on whether the magic word will be used as a parser function or a variable. For more information, please see:

Localisation edit

See Help:Magic words#Localisation for help.

You can read more on definition and usage of magic words for localisation at Manual:Messages API, Manual:Language#Namespaces; Avoid {{SITENAME}} in messages.

Behavior switches (double underscore magic words) edit

Behavior switches are a special type of magic word. They can be recognized by their use of double underscores (rather than double braces). Example: __NOTOC__

These magic words typically do not output any content, but instead change the behavior of a page and/or set a page property. These magic words are listed in MagicWordFactory::mDoubleUnderscoreIDs and also at Help:Magic words#Behavior switches. The effect of most standard behavior switches is defined in Parser::handleDoubleUnderscore(). If no specific effect is defined, the magic word will simply set a page property in the page_props table. This can also be checked later by testing if $parser->getOutput()->getPageProperty( 'MAGIC_WORD' ) is null or the empty string

Custom behavior switch edit

Here is an example extension implementing a custom __CUSTOM__ behaviour switch

custom/extension.json - This is minimal, a real extension would fill out more fields.

{
	"name": "Custom",
	"type": "parserhook",
	"AutoloadClasses": {
		"MyHooks": "MyHooks.php"
	},
	"Hooks": {
		"GetDoubleUnderscoreIDs": [
			"MyHooks::onGetDoubleUnderscoreIDs"
		],
		"ParserAfterParse": [
			"MyHooks::onParserAfterParse"
		]
	},
	"ExtensionMessagesFiles": {
		"CustomMagic": "custom.i18n.php"
	},
	"manifest_version": 1
}

custom/custom.i18n.php

<?php
$magicWords = [];
$magicWords['en'] = [
	'MAG_CUSTOM' => [ 0, '__CUSTOM__' ],
];

custom/MyHooks.php

<?php
class MyHooks {
	public static function onGetDoubleUnderscoreIDs( &$ids ) {
		$ids[] = 'MAG_CUSTOM';
	}

	public static function onParserAfterParse( Parser $parser, &$text, StripState $stripState ) {
		if ( $parser->getOutput()->getPageProperty( 'MAG_CUSTOM' ) !== null ) {
			// Do behavior switching here ...
			// e.g. If you wanted to add some JS, you would do $parser->getOutput()->addModules( 'moduleName' );
		}
	}
}

See also edit