wiki.techinc.nl/includes/OutputPage.php

4082 lines
120 KiB
PHP
Raw Normal View History

<?php
/**
* Preparation for the final page rendering.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
* @file
*/
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use MediaWiki\Session\SessionManager;
use WrappedString\WrappedString;
use WrappedString\WrappedStringList;
/**
* This class should be covered by a general architecture document which does
* not exist as of January 2011. This is one of the Core classes and should
* be read at least once by any new developers.
*
* This class is used to prepare the final rendering. A skin is then
* applied to the output parameters (links, javascript, html, categories ...).
2011-04-23 19:28:35 +00:00
*
* @todo FIXME: Another class handles sending the whole page to the client.
2011-04-23 19:28:35 +00:00
*
* Some comments comes from a pairing session between Zak Greant and Antoine Musso
* in November 2010.
*
* @todo document
*/
class OutputPage extends ContextSource {
/** @var array Should be private. Used with addMeta() which adds "<meta>" */
protected $mMetatags = [];
/** @var array */
protected $mLinktags = [];
/** @var bool */
protected $mCanonicalUrl = false;
/**
* @var array Additional stylesheets. Looks like this is for extensions.
* Might be replaced by ResourceLoader.
*/
protected $mExtStyles = [];
/**
* @var string Should be private - has getter and setter. Contains
* the HTML title */
public $mPagetitle = '';
/**
* @var string Contains all of the "<body>" content. Should be private we
* got set/get accessors and the append() method.
*/
public $mBodytext = '';
2010-07-08 08:12:19 +00:00
/** @var string Stores contents of "<title>" tag */
private $mHTMLtitle = '';
/**
* @var bool Is the displayed content related to the source of the
* corresponding wiki article.
*/
private $mIsarticle = false;
/** @var bool Stores "article flag" toggle. */
private $mIsArticleRelated = true;
2011-01-14 07:54:28 +00:00
/**
* @var bool We have to set isPrintable(). Some pages should
2011-01-14 07:54:28 +00:00
* never be printed (ex: redirections).
*/
private $mPrintable = false;
2011-01-14 07:54:28 +00:00
/**
* @var array Contains the page subtitle. Special pages usually have some
* links here. Don't confuse with site subtitle added by skins.
*/
private $mSubtitle = [];
/** @var string */
public $mRedirect = '';
/** @var int */
protected $mStatusCode;
2011-01-14 07:54:28 +00:00
/**
* @var string Used for sending cache control.
* The whole caching system should probably be moved into its own class.
2011-01-14 07:54:28 +00:00
*/
protected $mLastModified = '';
/** @var array */
protected $mCategoryLinks = [];
/** @var array */
protected $mCategories = [
'hidden' => [],
'normal' => [],
];
Implement page status indicators Page status indicators are icons (or short text snippets) usually displayed in the top-right corner of the page, outside of the main content. Basically, <indicator name="foo">[[File:Foo.svg|20px]]</indicator> may be used on a page to place the icon in the indicator area. They are also known as top icons, page icons, heading icons or title icons. I found the discussion on bug 23796 highly illuminating. I suggest that everyone read it before suggesting different design choices. I spent some time with a thesaurus pondering the name. "Emblems" and "badges" were also considered, but the former has a much more limited meaning and the latter is already taken by Wikidata, with a similar but subtly different feature set. I am not aware of any naming conflicts ;) besides new talk page message "indicator" (used by core and Echo in some documents) and OOjs UI indicators (tiny icons like the arrow on a dropdown form element), which shouldn't be confusing. Potential use cases include: * "Lock" indicators for page protection levels * Featured/good article indicators * Redirect shortcuts display ("WP:VPT") * Links to help/manual for special pages * Coordinates?… or globe icon for inline pop-up maps Design features: * Skin-customizable. Skins can fully control where and how indicators are shown, or may just do <?php echo $this->getIndicators(); ?> to output the default structure. By default they are not shown at all. * Extension-customizable. Extensions can call ParserOutput::addIndicator() to insert an indicator from one of the numerous parser hooks. * Wiki-customizable. In addition to just using the parser functions, on-wiki styles and scripts can use the provided classes and ids (.mw-indicator, #mw-indicator-<name>) to customize their display. Design limitations: * Every indicator must have a unique identifier (name). It's not possible to create arrays, or to have several indicators with the same name. In case of duplicates, the latest occurrence of the parser function wins. * Indicators are displayed ordered by their names (and not occurrence order). This ensures consistency across pages and provides a simple means of ordering or grouping them. * Indicators are not stored, tracked or accessible outside of ParserOutput (in particular they're not in the page_props table). They are intended to merely reflect the content or metadata that is already present on the page, and not be data themselves. If you ever think you need to list pages with a given status indicator, instead figure out what it means and use the appropriate tracking category, special page report, already existing page_prop, or other means. Corresponding patch in Vector: I90a8ae15ac8275d084ea5f47b6b2684d5e6c7412. I'll implement support in the other three skins included in the tarball and document it on mediawiki.org after this is merged. Bug: 23796 Change-Id: I2389ff9a5332a2b1d033eb75f0946e5241cfaaf4
2014-09-24 10:44:16 +00:00
/** @var array */
protected $mIndicators = [];
Implement page status indicators Page status indicators are icons (or short text snippets) usually displayed in the top-right corner of the page, outside of the main content. Basically, <indicator name="foo">[[File:Foo.svg|20px]]</indicator> may be used on a page to place the icon in the indicator area. They are also known as top icons, page icons, heading icons or title icons. I found the discussion on bug 23796 highly illuminating. I suggest that everyone read it before suggesting different design choices. I spent some time with a thesaurus pondering the name. "Emblems" and "badges" were also considered, but the former has a much more limited meaning and the latter is already taken by Wikidata, with a similar but subtly different feature set. I am not aware of any naming conflicts ;) besides new talk page message "indicator" (used by core and Echo in some documents) and OOjs UI indicators (tiny icons like the arrow on a dropdown form element), which shouldn't be confusing. Potential use cases include: * "Lock" indicators for page protection levels * Featured/good article indicators * Redirect shortcuts display ("WP:VPT") * Links to help/manual for special pages * Coordinates?… or globe icon for inline pop-up maps Design features: * Skin-customizable. Skins can fully control where and how indicators are shown, or may just do <?php echo $this->getIndicators(); ?> to output the default structure. By default they are not shown at all. * Extension-customizable. Extensions can call ParserOutput::addIndicator() to insert an indicator from one of the numerous parser hooks. * Wiki-customizable. In addition to just using the parser functions, on-wiki styles and scripts can use the provided classes and ids (.mw-indicator, #mw-indicator-<name>) to customize their display. Design limitations: * Every indicator must have a unique identifier (name). It's not possible to create arrays, or to have several indicators with the same name. In case of duplicates, the latest occurrence of the parser function wins. * Indicators are displayed ordered by their names (and not occurrence order). This ensures consistency across pages and provides a simple means of ordering or grouping them. * Indicators are not stored, tracked or accessible outside of ParserOutput (in particular they're not in the page_props table). They are intended to merely reflect the content or metadata that is already present on the page, and not be data themselves. If you ever think you need to list pages with a given status indicator, instead figure out what it means and use the appropriate tracking category, special page report, already existing page_prop, or other means. Corresponding patch in Vector: I90a8ae15ac8275d084ea5f47b6b2684d5e6c7412. I'll implement support in the other three skins included in the tarball and document it on mediawiki.org after this is merged. Bug: 23796 Change-Id: I2389ff9a5332a2b1d033eb75f0946e5241cfaaf4
2014-09-24 10:44:16 +00:00
/** @var array Array of Interwiki Prefixed (non DB key) Titles (e.g. 'fr:Test page') */
private $mLanguageLinks = [];
2011-01-14 07:54:28 +00:00
/**
* Used for JavaScript (predates ResourceLoader)
* @todo We should split JS / CSS.
* mScripts content is inserted as is in "<head>" by Skin. This might
* contain either a link to a stylesheet or inline CSS.
*/
private $mScripts = '';
2011-01-14 07:54:28 +00:00
/** @var string Inline CSS styles. Use addInlineStyle() sparingly */
protected $mInlineStyles = '';
/**
* @var string Used by skin template.
* Example: $tpl->set( 'displaytitle', $out->mPageLinkTitle );
*/
public $mPageLinkTitle = '';
/** @var array Array of elements in "<head>". Parser might add its own headers! */
protected $mHeadItems = [];
/** @var array Additional <body> classes; there are also <body> classes from other sources */
protected $mAdditionalBodyClasses = [];
/** @var array */
protected $mModules = [];
/** @var array */
protected $mModuleScripts = [];
/** @var array */
protected $mModuleStyles = [];
2010-07-08 08:12:19 +00:00
/** @var ResourceLoader */
protected $mResourceLoader;
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
/** @var ResourceLoaderClientHtml */
private $rlClient;
/** @var ResourceLoaderContext */
private $rlClientContext;
/** @var string */
private $rlUserModuleState;
/** @var array */
private $rlExemptStyleModules;
/** @var array */
protected $mJsConfigVars = [];
/** @var array */
protected $mTemplateIds = [];
/** @var array */
protected $mImageTimeKeys = [];
2011-07-06 17:55:44 +00:00
/** @var string */
public $mRedirectCode = '';
2011-07-06 17:55:44 +00:00
protected $mFeedLinksAppendQuery = null;
/** @var array
* What level of 'untrustworthiness' is allowed in CSS/JS modules loaded on this page?
* @see ResourceLoaderModule::$origin
* ResourceLoaderModule::ORIGIN_ALL is assumed unless overridden;
*/
protected $mAllowedModules = [
ResourceLoaderModule::TYPE_COMBINED => ResourceLoaderModule::ORIGIN_ALL,
];
/** @var bool Whether output is disabled. If this is true, the 'output' method will do nothing. */
protected $mDoNothing = false;
// Parser related.
/** @var int */
protected $mContainsNewMagic = 0;
/**
* lazy initialised, use parserOptions()
* @var ParserOptions
*/
protected $mParserOptions = null;
2011-01-14 07:54:28 +00:00
/**
* Handles the Atom / RSS links.
* We probably only support Atom in 2011.
2011-01-14 07:54:28 +00:00
* @see $wgAdvertisedFeedTypes
*/
private $mFeedLinks = [];
2009-11-07 09:26:02 +00:00
// Gwicke work on squid caching? Roughly from 2003.
protected $mEnableClientCache = true;
/** @var bool Flag if output should only contain the body of the article. */
private $mArticleBodyOnly = false;
/** @var bool */
protected $mNewSectionLink = false;
/** @var bool */
protected $mHideNewSectionLink = false;
2011-01-14 07:54:28 +00:00
/**
* @var bool Comes from the parser. This was probably made to load CSS/JS
* only if we had "<gallery>". Used directly in CategoryPage.php.
* Looks like ResourceLoader can replace this.
*/
public $mNoGallery = false;
/** @var string */
private $mPageTitleActionText = '';
/** @var int Cache stuff. Looks like mEnableClientCache */
protected $mCdnMaxage = 0;
/** @var int Upper limit on mCdnMaxage */
protected $mCdnMaxageLimit = INF;
/**
* @var bool Controls if anti-clickjacking / frame-breaking headers will
* be sent. This should be done for pages where edit actions are possible.
* Setters: $this->preventClickjacking() and $this->allowClickjacking().
*/
protected $mPreventClickjacking = true;
/** @var int To include the variable {{REVISIONID}} */
private $mRevisionId = null;
/** @var string */
private $mRevisionTimestamp = null;
/** @var array */
protected $mFileVersion = null;
/**
* @var array An array of stylesheet filenames (relative from skins path),
* with options for CSS media, IE conditions, and RTL/LTR direction.
* For internal use; add settings in the skin via $this->addStyle()
*
* Style again! This seems like a code duplication since we already have
* mStyles. This is what makes Open Source amazing.
*/
protected $styles = [];
private $mIndexPolicy = 'index';
private $mFollowPolicy = 'follow';
private $mVaryHeader = [
'Accept-Encoding' => [ 'match=gzip' ],
];
/**
* If the current page was reached through a redirect, $mRedirectedFrom contains the Title
* of the redirect.
*
* @var Title
*/
private $mRedirectedFrom = null;
/**
* Additional key => value data
*/
private $mProperties = [];
/**
* @var string|null ResourceLoader target for load.php links. If null, will be omitted
*/
private $mTarget = null;
/**
* @var bool Whether parser output contains a table of contents
*/
private $mEnableTOC = false;
/**
* @var bool Whether parser output should contain section edit links
*/
private $mEnableSectionEditLinks = true;
/**
* @var string|null The URL to send in a <link> element with rel=license
*/
private $copyrightUrl;
/** @var array Profiling data */
private $limitReportJSData = [];
/**
* Link: header contents
*/
private $mLinkHeader = [];
/**
* Constructor for OutputPage. This should not be called directly.
* Instead a new RequestContext should be created and it will implicitly create
* a OutputPage tied to that context.
* @param IContextSource|null $context
*/
function __construct( IContextSource $context = null ) {
if ( $context === null ) {
# Extensions should use `new RequestContext` instead of `new OutputPage` now.
wfDeprecated( __METHOD__, '1.18' );
} else {
$this->setContext( $context );
}
}
/**
* Redirect to $url rather than displaying the normal page
*
* @param string $url URL
* @param string $responsecode HTTP status code
*/
public function redirect( $url, $responsecode = '302' ) {
# Strip newlines as a paranoia check for header injection in PHP<5.1.2
$this->mRedirect = str_replace( "\n", '', $url );
$this->mRedirectCode = $responsecode;
}
/**
* Get the URL to redirect to, or an empty string if not redirect URL set
*
* @return string
*/
public function getRedirect() {
return $this->mRedirect;
}
2003-04-14 23:10:40 +00:00
/**
* Set the copyright URL to send with the output.
* Empty string to omit, null to reset.
*
* @since 1.26
*
* @param string|null $url
*/
public function setCopyrightUrl( $url ) {
$this->copyrightUrl = $url;
}
/**
* Set the HTTP status code to send with the output.
*
* @param int $statusCode
*/
public function setStatusCode( $statusCode ) {
$this->mStatusCode = $statusCode;
}
/**
* Add a new "<meta>" tag
* To add an http-equiv meta tag, precede the name with "http:"
*
* @param string $name Tag name
* @param string $val Tag value
*/
function addMeta( $name, $val ) {
array_push( $this->mMetatags, [ $name, $val ] );
}
/**
* Returns the current <meta> tags
*
* @since 1.25
* @return array
*/
public function getMetaTags() {
return $this->mMetatags;
}
/**
* Add a new \<link\> tag to the page header.
*
* Note: use setCanonicalUrl() for rel=canonical.
*
* @param array $linkarr Associative array of attributes.
*/
function addLink( array $linkarr ) {
array_push( $this->mLinktags, $linkarr );
}
/**
* Returns the current <link> tags
*
* @since 1.25
* @return array
*/
public function getLinkTags() {
return $this->mLinktags;
}
/**
* Add a new \<link\> with "rel" attribute set to "meta"
*
* @param array $linkarr Associative array mapping attribute names to their
* values, both keys and values will be escaped, and the
* "rel" attribute will be automatically added
*/
function addMetadataLink( array $linkarr ) {
$linkarr['rel'] = $this->getMetadataAttribute();
$this->addLink( $linkarr );
}
/**
* Set the URL to be used for the <link rel=canonical>. This should be used
* in preference to addLink(), to avoid duplicate link tags.
* @param string $url
*/
function setCanonicalUrl( $url ) {
$this->mCanonicalUrl = $url;
}
/**
* Returns the URL to be used for the <link rel=canonical> if
* one is set.
*
* @since 1.25
* @return bool|string
*/
public function getCanonicalUrl() {
return $this->mCanonicalUrl;
}
/**
* Get the value of the "rel" attribute for metadata links
*
* @return string
*/
public function getMetadataAttribute() {
# note: buggy CC software only reads first "meta" link
static $haveMeta = false;
if ( $haveMeta ) {
return 'alternate meta';
} else {
$haveMeta = true;
return 'meta';
}
}
/**
* Add raw HTML to the list of scripts (including \<script\> tag, etc.)
* Internal use only. Use OutputPage::addModules() or OutputPage::addJsConfigVars()
* if possible.
*
* @param string $script Raw HTML
*/
2009-08-07 00:16:54 +00:00
function addScript( $script ) {
$this->mScripts .= $script;
2009-08-07 00:16:54 +00:00
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
/**
2009-08-07 00:16:54 +00:00
* Register and add a stylesheet from an extension directory.
*
* @deprecated since 1.27 use addModuleStyles() or addStyle() instead
* @param string $url Path to sheet. Provide either a full url (beginning
* with 'http', etc) or a relative path from the document root
* (beginning with '/'). Otherwise it behaves identically to
* addStyle() and draws from the /skins folder.
*/
public function addExtensionStyle( $url ) {
wfDeprecated( __METHOD__, '1.27' );
array_push( $this->mExtStyles, $url );
2008-08-10 07:14:08 +00:00
}
/**
2011-01-14 07:54:28 +00:00
* Get all styles added by extensions
*
* @deprecated since 1.27
* @return array
*/
function getExtStyle() {
wfDeprecated( __METHOD__, '1.27' );
return $this->mExtStyles;
}
/**
* Add a JavaScript file out of skins/common, or a given relative path.
* Internal use only. Use OutputPage::addModules() if possible.
*
* @param string $file Filename in skins/common or complete on-server path
* (/foo/bar.js)
* @param string $version Style version of the file. Defaults to $wgStyleVersion
*/
public function addScriptFile( $file, $version = null ) {
2010-08-20 01:47:52 +00:00
// See if $file parameter is an absolute URL or begins with a slash
if ( substr( $file, 0, 1 ) == '/' || preg_match( '#^[a-z]*://#i', $file ) ) {
$path = $file;
} else {
$path = $this->getConfig()->get( 'StylePath' ) . "/common/{$file}";
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
}
if ( is_null( $version ) ) {
$version = $this->getConfig()->get( 'StyleVersion' );
}
$this->addScript( Html::linkedScript( wfAppendQuery( $path, $version ) ) );
}
/**
* Add a self-contained script tag with the given contents
* Internal use only. Use OutputPage::addModules() if possible.
*
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* @param string $script JavaScript text, no script tags
*/
public function addInlineScript( $script ) {
$this->mScripts .= Html::inlineScript( $script );
}
/**
* Filter an array of modules to remove insufficiently trustworthy members, and modules
* which are no longer registered (eg a page is cached before an extension is disabled)
* @param array $modules
* @param string|null $position If not null, only return modules with this position
* @param string $type
* @return array
*/
protected function filterModules( array $modules, $position = null,
$type = ResourceLoaderModule::TYPE_COMBINED
) {
$resourceLoader = $this->getResourceLoader();
$filteredModules = [];
foreach ( $modules as $val ) {
$module = $resourceLoader->getModule( $val );
if ( $module instanceof ResourceLoaderModule
&& $module->getOrigin() <= $this->getAllowedModules( $type )
&& ( is_null( $position ) || $module->getPosition() == $position )
) {
if ( $this->mTarget && !in_array( $this->mTarget, $module->getTargets() ) ) {
$this->warnModuleTargetFilter( $module->getName() );
continue;
}
$filteredModules[] = $val;
}
}
return $filteredModules;
}
private function warnModuleTargetFilter( $moduleName ) {
static $warnings = [];
if ( isset( $warnings[$this->mTarget][$moduleName] ) ) {
return;
}
$warnings[$this->mTarget][$moduleName] = true;
$this->getResourceLoader()->getLogger()->debug(
'Module "{module}" not loadable on target "{target}".',
[
'module' => $moduleName,
'target' => $this->mTarget,
]
);
}
/**
* Get the list of modules to include on this page
2010-09-07 21:05:09 +00:00
*
* @param bool $filter Whether to filter out insufficiently trustworthy modules
* @param string|null $position If not null, only return modules with this position
* @param string $param
* @param string $type
* @return array Array of module names
*/
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
public function getModules( $filter = false, $position = null, $param = 'mModules',
$type = ResourceLoaderModule::TYPE_COMBINED
) {
$modules = array_values( array_unique( $this->$param ) );
return $filter
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
? $this->filterModules( $modules, $position, $type )
: $modules;
}
/**
* Add one or more modules recognized by ResourceLoader. Modules added
* through this function will be loaded by ResourceLoader when the
* page loads.
2010-09-07 21:05:09 +00:00
*
* @param string|array $modules Module name (string) or array of module names
*/
public function addModules( $modules ) {
$this->mModules = array_merge( $this->mModules, (array)$modules );
}
/**
* Get the list of module JS to include on this page
*
* @param bool $filter
* @param string|null $position
* @return array Array of module names
*/
public function getModuleScripts( $filter = false, $position = null ) {
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return $this->getModules( $filter, $position, 'mModuleScripts',
ResourceLoaderModule::TYPE_SCRIPTS
);
}
/**
* Add only JS of one or more modules recognized by ResourceLoader. Module
* scripts added through this function will be loaded by ResourceLoader when
* the page loads.
2010-09-07 21:05:09 +00:00
*
* @param string|array $modules Module name (string) or array of module names
*/
public function addModuleScripts( $modules ) {
$this->mModuleScripts = array_merge( $this->mModuleScripts, (array)$modules );
}
/**
* Get the list of module CSS to include on this page
2010-09-07 21:05:09 +00:00
*
* @param bool $filter
* @param string|null $position
* @return array Array of module names
*/
public function getModuleStyles( $filter = false, $position = null ) {
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return $this->getModules( $filter, $position, 'mModuleStyles',
ResourceLoaderModule::TYPE_STYLES
);
}
/**
* Add only CSS of one or more modules recognized by ResourceLoader.
*
* Module styles added through this function will be added using standard link CSS
* tags, rather than as a combined Javascript and CSS package. Thus, they will
* load when JavaScript is disabled (unless CSS also happens to be disabled).
2010-09-07 21:05:09 +00:00
*
* @param string|array $modules Module name (string) or array of module names
*/
public function addModuleStyles( $modules ) {
$this->mModuleStyles = array_merge( $this->mModuleStyles, (array)$modules );
}
/**
* @return null|string ResourceLoader target
*/
public function getTarget() {
return $this->mTarget;
}
/**
* Sets ResourceLoader target for load.php links. If null, will be omitted
*
* @param string|null $target
*/
public function setTarget( $target ) {
$this->mTarget = $target;
}
/**
* Get an array of head items
*
* @return array
*/
function getHeadItemsArray() {
return $this->mHeadItems;
}
/**
* Add or replace a head item to the output
*
* Whenever possible, use more specific options like ResourceLoader modules,
* OutputPage::addLink(), OutputPage::addMetaLink() and OutputPage::addFeedLink()
* Fallback options for those are: OutputPage::addStyle, OutputPage::addScript(),
* OutputPage::addInlineScript() and OutputPage::addInlineStyle()
* This would be your very LAST fallback.
*
* @param string $name Item name
* @param string $value Raw HTML
*/
public function addHeadItem( $name, $value ) {
2007-04-29 10:15:05 +00:00
$this->mHeadItems[$name] = $value;
}
/**
* Add one or more head items to the output
*
* @since 1.28
* @param string|string[] $values Raw HTML
*/
public function addHeadItems( $values ) {
$this->mHeadItems = array_merge( $this->mHeadItems, (array)$values );
}
/**
* Check if the header item $name is already set
*
* @param string $name Item name
* @return bool
*/
public function hasHeadItem( $name ) {
return isset( $this->mHeadItems[$name] );
}
/**
* Add a class to the <body> element
*
* @since 1.30
* @param string|string[] $classes One or more classes to add
*/
public function addBodyClasses( $classes ) {
$this->mAdditionalBodyClasses = array_merge( $this->mAdditionalBodyClasses, (array)$classes );
}
/**
* @deprecated since 1.28 Obsolete - wgUseETag experiment was removed.
* @param string $tag
*/
public function setETag( $tag ) {
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
/**
* Set whether the output should only contain the body of the article,
* without any skin, sidebar, etc.
* Used e.g. when calling with "action=render".
*
* @param bool $only Whether to output only the body of the article
*/
public function setArticleBodyOnly( $only ) {
$this->mArticleBodyOnly = $only;
2008-08-10 07:14:08 +00:00
}
2003-04-14 23:10:40 +00:00
/**
* Return whether the output will contain only the body of the article
*
* @return bool
*/
public function getArticleBodyOnly() {
return $this->mArticleBodyOnly;
}
/**
* Set an additional output property
* @since 1.21
*
* @param string $name
* @param mixed $value
*/
public function setProperty( $name, $value ) {
$this->mProperties[$name] = $value;
}
/**
* Get an additional output property
* @since 1.21
*
* @param string $name
* @return mixed Property value or null if not found
*/
public function getProperty( $name ) {
if ( isset( $this->mProperties[$name] ) ) {
return $this->mProperties[$name];
} else {
return null;
}
}
/**
* checkLastModified tells the client to use the client-cached page if
* possible. If successful, the OutputPage is disabled so that
* any future call to OutputPage->output() have no effect.
*
* Side effect: sets mLastModified for Last-Modified header
*
* @param string $timestamp
*
* @return bool True if cache-ok headers was sent.
*/
public function checkLastModified( $timestamp ) {
if ( !$timestamp || $timestamp == '19700101000000' ) {
wfDebug( __METHOD__ . ": CACHE DISABLED, NO TIMESTAMP\n" );
return false;
}
$config = $this->getConfig();
if ( !$config->get( 'CachePages' ) ) {
wfDebug( __METHOD__ . ": CACHE DISABLED\n" );
return false;
2003-07-10 04:55:41 +00:00
}
$timestamp = wfTimestamp( TS_MW, $timestamp );
$modifiedTimes = [
'page' => $timestamp,
'user' => $this->getUser()->getTouched(),
'epoch' => $config->get( 'CacheEpoch' )
];
if ( $config->get( 'UseSquid' ) ) {
// T46570: the core page itself may not change, but resources might
$modifiedTimes['sepoch'] = wfTimestamp( TS_MW, time() - $config->get( 'SquidMaxage' ) );
}
Hooks::run( 'OutputPageCheckLastModified', [ &$modifiedTimes, $this ] );
$maxModified = max( $modifiedTimes );
$this->mLastModified = wfTimestamp( TS_RFC2822, $maxModified );
$clientHeader = $this->getRequest()->getHeader( 'If-Modified-Since' );
if ( $clientHeader === false ) {
wfDebug( __METHOD__ . ": client did not send If-Modified-Since header", 'private' );
return false;
2003-04-14 23:10:40 +00:00
}
# IE sends sizes after the date like this:
# Wed, 20 Aug 2003 06:51:19 GMT; length=5202
# this breaks strtotime().
$clientHeader = preg_replace( '/;.*$/', '', $clientHeader );
MediaWiki\suppressWarnings(); // E_STRICT system time bitching
$clientHeaderTime = strtotime( $clientHeader );
MediaWiki\restoreWarnings();
if ( !$clientHeaderTime ) {
wfDebug( __METHOD__
. ": unable to parse the client's If-Modified-Since header: $clientHeader\n" );
return false;
}
$clientHeaderTime = wfTimestamp( TS_MW, $clientHeaderTime );
# Make debug info
$info = '';
foreach ( $modifiedTimes as $name => $value ) {
if ( $info !== '' ) {
$info .= ', ';
}
$info .= "$name=" . wfTimestamp( TS_ISO_8601, $value );
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
wfDebug( __METHOD__ . ": client sent If-Modified-Since: " .
wfTimestamp( TS_ISO_8601, $clientHeaderTime ), 'private' );
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
wfDebug( __METHOD__ . ": effective Last-Modified: " .
wfTimestamp( TS_ISO_8601, $maxModified ), 'private' );
if ( $clientHeaderTime < $maxModified ) {
wfDebug( __METHOD__ . ": STALE, $info", 'private' );
return false;
}
# Not modified
# Give a 304 Not Modified response code and disable body output
wfDebug( __METHOD__ . ": NOT MODIFIED, $info", 'private' );
ini_set( 'zlib.output_compression', 0 );
$this->getRequest()->response()->statusHeader( 304 );
$this->sendCacheControl();
$this->disable();
// Don't output a compressed blob when using ob_gzhandler;
// it's technically against HTTP spec and seems to confuse
// Firefox when the response gets split over two packets.
wfClearOutputBuffers();
return true;
2003-04-14 23:10:40 +00:00
}
/**
* Override the last modified timestamp
*
* @param string $timestamp New timestamp, in a format readable by
* wfTimestamp()
*/
public function setLastModified( $timestamp ) {
$this->mLastModified = wfTimestamp( TS_RFC2822, $timestamp );
}
/**
* Set the robot policy for the page: <http://www.robotstxt.org/meta.html>
*
* @param string $policy The literal string to output as the contents of
* the meta tag. Will be parsed according to the spec and output in
* standardized form.
* @return null
*/
public function setRobotPolicy( $policy ) {
$policy = Article::formatRobotPolicy( $policy );
if ( isset( $policy['index'] ) ) {
$this->setIndexPolicy( $policy['index'] );
}
if ( isset( $policy['follow'] ) ) {
$this->setFollowPolicy( $policy['follow'] );
}
}
/**
* Set the index policy for the page, but leave the follow policy un-
* touched.
*
* @param string $policy Either 'index' or 'noindex'.
* @return null
*/
public function setIndexPolicy( $policy ) {
$policy = trim( $policy );
if ( in_array( $policy, [ 'index', 'noindex' ] ) ) {
$this->mIndexPolicy = $policy;
}
}
/**
* Set the follow policy for the page, but leave the index policy un-
* touched.
*
* @param string $policy Either 'follow' or 'nofollow'.
* @return null
*/
public function setFollowPolicy( $policy ) {
$policy = trim( $policy );
if ( in_array( $policy, [ 'follow', 'nofollow' ] ) ) {
$this->mFollowPolicy = $policy;
}
}
/**
* Set the new value of the "action text", this will be added to the
* "HTML title", separated from it with " - ".
*
* @param string $text New value of the "action text"
*/
public function setPageTitleActionText( $text ) {
$this->mPageTitleActionText = $text;
}
/**
* Get the value of the "action text"
*
* @return string
*/
public function getPageTitleActionText() {
return $this->mPageTitleActionText;
}
/**
* "HTML title" means the contents of "<title>".
* It is stored as plain, unescaped text and will be run through htmlspecialchars in the skin file.
*
* @param string|Message $name
*/
public function setHTMLTitle( $name ) {
if ( $name instanceof Message ) {
$this->mHTMLtitle = $name->setContext( $this->getContext() )->text();
} else {
$this->mHTMLtitle = $name;
}
}
/**
* Return the "HTML title", i.e. the content of the "<title>" tag.
*
* @return string
*/
public function getHTMLTitle() {
return $this->mHTMLtitle;
}
/**
* Set $mRedirectedFrom, the Title of the page which redirected us to the current page.
*
* @param Title $t
*/
public function setRedirectedFrom( $t ) {
$this->mRedirectedFrom = $t;
}
/**
* "Page title" means the contents of \<h1\>. It is stored as a valid HTML
* fragment. This function allows good tags like \<sup\> in the \<h1\> tag,
* but not bad tags like \<script\>. This function automatically sets
* \<title\> to the same content as \<h1\> but with all tags removed. Bad
* tags that were escaped in \<h1\> will still be escaped in \<title\>, and
* good tags like \<i\> will be dropped entirely.
*
* @param string|Message $name
*/
public function setPageTitle( $name ) {
if ( $name instanceof Message ) {
$name = $name->setContext( $this->getContext() )->text();
}
# change "<script>foo&bar</script>" to "&lt;script&gt;foo&amp;bar&lt;/script&gt;"
# but leave "<i>foobar</i>" alone
$nameWithTags = Sanitizer::normalizeCharReferences( Sanitizer::removeHTMLtags( $name ) );
$this->mPagetitle = $nameWithTags;
# change "<i>foo&amp;bar</i>" to "foo&bar"
$this->setHTMLTitle(
$this->msg( 'pagetitle' )->rawParams( Sanitizer::stripAllTags( $nameWithTags ) )
->inContentLanguage()
);
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
/**
* Return the "page title", i.e. the content of the \<h1\> tag.
*
* @return string
*/
public function getPageTitle() {
return $this->mPagetitle;
}
/**
* Set the Title object to use
*
* @param Title $t
*/
public function setTitle( Title $t ) {
$this->getContext()->setTitle( $t );
}
/**
* Replace the subtitle with $str
*
* @param string|Message $str New value of the subtitle. String should be safe HTML.
*/
public function setSubtitle( $str ) {
$this->clearSubtitle();
$this->addSubtitle( $str );
}
/**
* Add $str to the subtitle
*
* @param string|Message $str String or Message to add to the subtitle. String should be safe HTML.
*/
public function addSubtitle( $str ) {
if ( $str instanceof Message ) {
$this->mSubtitle[] = $str->setContext( $this->getContext() )->parse();
} else {
$this->mSubtitle[] = $str;
}
}
/**
* Build message object for a subtitle containing a backlink to a page
*
* @param Title $title Title to link to
* @param array $query Array of additional parameters to include in the link
* @return Message
* @since 1.25
*/
public static function buildBacklinkSubtitle( Title $title, $query = [] ) {
if ( $title->isRedirect() ) {
$query['redirect'] = 'no';
}
$linkRenderer = MediaWikiServices::getInstance()->getLinkRenderer();
return wfMessage( 'backlinksubtitle' )
->rawParams( $linkRenderer->makeLink( $title, null, [], $query ) );
}
/**
* Add a subtitle containing a backlink to a page
*
* @param Title $title Title to link to
* @param array $query Array of additional parameters to include in the link
*/
public function addBacklinkSubtitle( Title $title, $query = [] ) {
$this->addSubtitle( self::buildBacklinkSubtitle( $title, $query ) );
}
/**
* Clear the subtitles
*/
public function clearSubtitle() {
$this->mSubtitle = [];
}
/**
* Get the subtitle
*
* @return string
*/
public function getSubtitle() {
return implode( "<br />\n\t\t\t\t", $this->mSubtitle );
}
/**
* Set the page as printable, i.e. it'll be displayed with all
* print styles included
*/
public function setPrintable() {
$this->mPrintable = true;
}
/**
* Return whether the page is "printable"
*
* @return bool
*/
public function isPrintable() {
return $this->mPrintable;
}
/**
* Disable output completely, i.e. calling output() will have no effect
*/
public function disable() {
$this->mDoNothing = true;
}
/**
* Return whether the output will be completely disabled
*
* @return bool
*/
public function isDisabled() {
return $this->mDoNothing;
}
/**
* Show an "add new section" link?
*
* @return bool
*/
public function showNewSectionLink() {
return $this->mNewSectionLink;
}
/**
* Forcibly hide the new section link?
*
* @return bool
*/
public function forceHideNewSectionLink() {
return $this->mHideNewSectionLink;
}
/**
* Add or remove feed links in the page header
* This is mainly kept for backward compatibility, see OutputPage::addFeedLink()
* for the new version
* @see addFeedLink()
*
* @param bool $show True: add default feeds, false: remove all feeds
*/
public function setSyndicated( $show = true ) {
if ( $show ) {
$this->setFeedAppendQuery( false );
} else {
$this->mFeedLinks = [];
}
}
/**
* Add default feeds to the page header
* This is mainly kept for backward compatibility, see OutputPage::addFeedLink()
* for the new version
* @see addFeedLink()
*
* @param string $val Query to append to feed links or false to output
* default links
*/
public function setFeedAppendQuery( $val ) {
$this->mFeedLinks = [];
foreach ( $this->getConfig()->get( 'AdvertisedFeedTypes' ) as $type ) {
$query = "feed=$type";
if ( is_string( $val ) ) {
$query .= '&' . $val;
}
$this->mFeedLinks[$type] = $this->getTitle()->getLocalURL( $query );
}
}
/**
* Add a feed link to the page header
*
* @param string $format Feed type, should be a key of $wgFeedClasses
* @param string $href URL
*/
public function addFeedLink( $format, $href ) {
if ( in_array( $format, $this->getConfig()->get( 'AdvertisedFeedTypes' ) ) ) {
$this->mFeedLinks[$format] = $href;
}
}
/**
2010-01-27 00:11:13 +00:00
* Should we output feed links for this page?
* @return bool
*/
public function isSyndicated() {
return count( $this->mFeedLinks ) > 0;
}
/**
* Return URLs for each supported syndication format for this page.
* @return array Associating format keys with URLs
*/
public function getSyndicationLinks() {
return $this->mFeedLinks;
}
/**
* Will currently always return null
*
* @return null
*/
public function getFeedAppendQuery() {
return $this->mFeedLinksAppendQuery;
}
/**
* Set whether the displayed content is related to the source of the
* corresponding article on the wiki
* Setting true will cause the change "article related" toggle to true
*
* @param bool $v
*/
public function setArticleFlag( $v ) {
$this->mIsarticle = $v;
if ( $v ) {
$this->mIsArticleRelated = $v;
}
}
/**
* Return whether the content displayed page is related to the source of
* the corresponding article on the wiki
*
* @return bool
*/
public function isArticle() {
return $this->mIsarticle;
}
/**
* Set whether this page is related an article on the wiki
* Setting false will cause the change of "article flag" toggle to false
*
* @param bool $v
*/
public function setArticleRelated( $v ) {
$this->mIsArticleRelated = $v;
if ( !$v ) {
$this->mIsarticle = false;
}
}
/**
* Return whether this page is related an article on the wiki
*
* @return bool
*/
public function isArticleRelated() {
return $this->mIsArticleRelated;
}
/**
* Add new language links
*
* @param string[] $newLinkArray Array of interwiki-prefixed (non DB key) titles
* (e.g. 'fr:Test page')
*/
public function addLanguageLinks( array $newLinkArray ) {
2004-06-01 18:29:31 +00:00
$this->mLanguageLinks += $newLinkArray;
}
/**
* Reset the language links and add new language links
*
* @param string[] $newLinkArray Array of interwiki-prefixed (non DB key) titles
* (e.g. 'fr:Test page')
*/
public function setLanguageLinks( array $newLinkArray ) {
2004-06-01 18:29:31 +00:00
$this->mLanguageLinks = $newLinkArray;
}
/**
* Get the list of language links
*
* @return string[] Array of interwiki-prefixed (non DB key) titles (e.g. 'fr:Test page')
*/
public function getLanguageLinks() {
return $this->mLanguageLinks;
}
/**
* Add an array of categories, with names in the keys
*
* @param array $categories Mapping category name => sort key
*/
public function addCategoryLinks( array $categories ) {
global $wgContLang;
if ( !is_array( $categories ) || count( $categories ) == 0 ) {
return;
}
$res = $this->addCategoryLinksToLBAndGetResult( $categories );
# Set all the values to 'normal'.
$categories = array_fill_keys( array_keys( $categories ), 'normal' );
# Mark hidden categories
foreach ( $res as $row ) {
if ( isset( $row->pp_value ) ) {
$categories[$row->page_title] = 'hidden';
}
}
// Avoid PHP 7.1 warning of passing $this by reference
$outputPage = $this;
# Add the remaining categories to the skin
if ( Hooks::run(
'OutputPageMakeCategoryLinks',
[ &$outputPage, $categories, &$this->mCategoryLinks ] )
) {
$linkRenderer = MediaWikiServices::getInstance()->getLinkRenderer();
foreach ( $categories as $category => $type ) {
// array keys will cast numeric category names to ints, so cast back to string
$category = (string)$category;
$origcategory = $category;
$title = Title::makeTitleSafe( NS_CATEGORY, $category );
if ( !$title ) {
continue;
}
$wgContLang->findVariantLink( $category, $title, true );
if ( $category != $origcategory && array_key_exists( $category, $categories ) ) {
continue;
}
$text = $wgContLang->convertHtml( $title->getText() );
$this->mCategories[$type][] = $title->getText();
$this->mCategoryLinks[$type][] = $linkRenderer->makeLink( $title, new HtmlArmor( $text ) );
}
}
2004-06-19 06:46:54 +00:00
}
/**
* @param array $categories
* @return bool|ResultWrapper
*/
protected function addCategoryLinksToLBAndGetResult( array $categories ) {
# Add the links to a LinkBatch
$arr = [ NS_CATEGORY => $categories ];
$lb = new LinkBatch;
$lb->setArray( $arr );
# Fetch existence plus the hiddencat property
$dbr = wfGetDB( DB_REPLICA );
$fields = array_merge(
LinkCache::getSelectFields(),
[ 'page_namespace', 'page_title', 'pp_value' ]
);
$res = $dbr->select( [ 'page', 'page_props' ],
$fields,
$lb->constructSet( 'page', $dbr ),
__METHOD__,
[],
[ 'page_props' => [ 'LEFT JOIN', [
'pp_propname' => 'hiddencat',
'pp_page = page_id'
] ] ]
);
# Add the results to the link cache
$lb->addResultToCache( LinkCache::singleton(), $res );
return $res;
}
/**
* Reset the category links (but not the category list) and add $categories
*
* @param array $categories Mapping category name => sort key
*/
public function setCategoryLinks( array $categories ) {
$this->mCategoryLinks = [];
$this->addCategoryLinks( $categories );
}
/**
* Get the list of category links, in a 2-D array with the following format:
* $arr[$type][] = $link, where $type is either "normal" or "hidden" (for
* hidden categories) and $link a HTML fragment with a link to the category
* page
*
* @return array
*/
public function getCategoryLinks() {
return $this->mCategoryLinks;
}
/**
* Get the list of category names this page belongs to.
*
* @param string $type The type of categories which should be returned. Possible values:
* * all: all categories of all types
* * hidden: only the hidden categories
* * normal: all categories, except hidden categories
* @return array Array of strings
*/
public function getCategories( $type = 'all' ) {
if ( $type === 'all' ) {
$allCategories = [];
foreach ( $this->mCategories as $categories ) {
$allCategories = array_merge( $allCategories, $categories );
}
return $allCategories;
}
if ( !isset( $this->mCategories[$type] ) ) {
throw new InvalidArgumentException( 'Invalid category type given: ' . $type );
}
return $this->mCategories[$type];
2004-06-19 06:46:54 +00:00
}
Implement page status indicators Page status indicators are icons (or short text snippets) usually displayed in the top-right corner of the page, outside of the main content. Basically, <indicator name="foo">[[File:Foo.svg|20px]]</indicator> may be used on a page to place the icon in the indicator area. They are also known as top icons, page icons, heading icons or title icons. I found the discussion on bug 23796 highly illuminating. I suggest that everyone read it before suggesting different design choices. I spent some time with a thesaurus pondering the name. "Emblems" and "badges" were also considered, but the former has a much more limited meaning and the latter is already taken by Wikidata, with a similar but subtly different feature set. I am not aware of any naming conflicts ;) besides new talk page message "indicator" (used by core and Echo in some documents) and OOjs UI indicators (tiny icons like the arrow on a dropdown form element), which shouldn't be confusing. Potential use cases include: * "Lock" indicators for page protection levels * Featured/good article indicators * Redirect shortcuts display ("WP:VPT") * Links to help/manual for special pages * Coordinates?… or globe icon for inline pop-up maps Design features: * Skin-customizable. Skins can fully control where and how indicators are shown, or may just do <?php echo $this->getIndicators(); ?> to output the default structure. By default they are not shown at all. * Extension-customizable. Extensions can call ParserOutput::addIndicator() to insert an indicator from one of the numerous parser hooks. * Wiki-customizable. In addition to just using the parser functions, on-wiki styles and scripts can use the provided classes and ids (.mw-indicator, #mw-indicator-<name>) to customize their display. Design limitations: * Every indicator must have a unique identifier (name). It's not possible to create arrays, or to have several indicators with the same name. In case of duplicates, the latest occurrence of the parser function wins. * Indicators are displayed ordered by their names (and not occurrence order). This ensures consistency across pages and provides a simple means of ordering or grouping them. * Indicators are not stored, tracked or accessible outside of ParserOutput (in particular they're not in the page_props table). They are intended to merely reflect the content or metadata that is already present on the page, and not be data themselves. If you ever think you need to list pages with a given status indicator, instead figure out what it means and use the appropriate tracking category, special page report, already existing page_prop, or other means. Corresponding patch in Vector: I90a8ae15ac8275d084ea5f47b6b2684d5e6c7412. I'll implement support in the other three skins included in the tarball and document it on mediawiki.org after this is merged. Bug: 23796 Change-Id: I2389ff9a5332a2b1d033eb75f0946e5241cfaaf4
2014-09-24 10:44:16 +00:00
/**
* Add an array of indicators, with their identifiers as array
* keys and HTML contents as values.
Implement page status indicators Page status indicators are icons (or short text snippets) usually displayed in the top-right corner of the page, outside of the main content. Basically, <indicator name="foo">[[File:Foo.svg|20px]]</indicator> may be used on a page to place the icon in the indicator area. They are also known as top icons, page icons, heading icons or title icons. I found the discussion on bug 23796 highly illuminating. I suggest that everyone read it before suggesting different design choices. I spent some time with a thesaurus pondering the name. "Emblems" and "badges" were also considered, but the former has a much more limited meaning and the latter is already taken by Wikidata, with a similar but subtly different feature set. I am not aware of any naming conflicts ;) besides new talk page message "indicator" (used by core and Echo in some documents) and OOjs UI indicators (tiny icons like the arrow on a dropdown form element), which shouldn't be confusing. Potential use cases include: * "Lock" indicators for page protection levels * Featured/good article indicators * Redirect shortcuts display ("WP:VPT") * Links to help/manual for special pages * Coordinates?… or globe icon for inline pop-up maps Design features: * Skin-customizable. Skins can fully control where and how indicators are shown, or may just do <?php echo $this->getIndicators(); ?> to output the default structure. By default they are not shown at all. * Extension-customizable. Extensions can call ParserOutput::addIndicator() to insert an indicator from one of the numerous parser hooks. * Wiki-customizable. In addition to just using the parser functions, on-wiki styles and scripts can use the provided classes and ids (.mw-indicator, #mw-indicator-<name>) to customize their display. Design limitations: * Every indicator must have a unique identifier (name). It's not possible to create arrays, or to have several indicators with the same name. In case of duplicates, the latest occurrence of the parser function wins. * Indicators are displayed ordered by their names (and not occurrence order). This ensures consistency across pages and provides a simple means of ordering or grouping them. * Indicators are not stored, tracked or accessible outside of ParserOutput (in particular they're not in the page_props table). They are intended to merely reflect the content or metadata that is already present on the page, and not be data themselves. If you ever think you need to list pages with a given status indicator, instead figure out what it means and use the appropriate tracking category, special page report, already existing page_prop, or other means. Corresponding patch in Vector: I90a8ae15ac8275d084ea5f47b6b2684d5e6c7412. I'll implement support in the other three skins included in the tarball and document it on mediawiki.org after this is merged. Bug: 23796 Change-Id: I2389ff9a5332a2b1d033eb75f0946e5241cfaaf4
2014-09-24 10:44:16 +00:00
*
* In case of duplicate keys, existing values are overwritten.
*
* @param array $indicators
* @since 1.25
*/
public function setIndicators( array $indicators ) {
$this->mIndicators = $indicators + $this->mIndicators;
// Keep ordered by key
ksort( $this->mIndicators );
}
/**
* Get the indicators associated with this page.
*
* The array will be internally ordered by item keys.
*
* @return array Keys: identifiers, values: HTML contents
* @since 1.25
*/
public function getIndicators() {
return $this->mIndicators;
}
/**
* Adds help link with an icon via page indicators.
* Link target can be overridden by a local message containing a wikilink:
* the message key is: lowercase action or special page name + '-helppage'.
* @param string $to Target MediaWiki.org page title or encoded URL.
* @param bool $overrideBaseUrl Whether $url is a full URL, to avoid MW.o.
* @since 1.25
*/
public function addHelpLink( $to, $overrideBaseUrl = false ) {
$this->addModuleStyles( 'mediawiki.helplink' );
$text = $this->msg( 'helppage-top-gethelp' )->escaped();
if ( $overrideBaseUrl ) {
$helpUrl = $to;
} else {
$toUrlencoded = wfUrlencode( str_replace( ' ', '_', $to ) );
$helpUrl = "//www.mediawiki.org/wiki/Special:MyLanguage/$toUrlencoded";
}
$link = Html::rawElement(
'a',
[
'href' => $helpUrl,
'target' => '_blank',
'class' => 'mw-helplink',
],
$text
);
$this->setIndicators( [ 'mw-helplink' => $link ] );
}
/**
* Do not allow scripts which can be modified by wiki users to load on this page;
* only allow scripts bundled with, or generated by, the software.
* Site-wide styles are controlled by a config setting, since they can be
* used to create a custom skin/theme, but not user-specific ones.
*
* @todo this should be given a more accurate name
*/
public function disallowUserJs() {
$this->reduceAllowedModules(
ResourceLoaderModule::TYPE_SCRIPTS,
ResourceLoaderModule::ORIGIN_CORE_INDIVIDUAL
);
// Site-wide styles are controlled by a config setting, see T73621
// for background on why. User styles are never allowed.
if ( $this->getConfig()->get( 'AllowSiteCSSOnRestrictedPages' ) ) {
$styleOrigin = ResourceLoaderModule::ORIGIN_USER_SITEWIDE;
} else {
$styleOrigin = ResourceLoaderModule::ORIGIN_CORE_INDIVIDUAL;
}
$this->reduceAllowedModules(
ResourceLoaderModule::TYPE_STYLES,
$styleOrigin
);
}
/**
* Show what level of JavaScript / CSS untrustworthiness is allowed on this page
* @see ResourceLoaderModule::$origin
* @param string $type ResourceLoaderModule TYPE_ constant
* @return int ResourceLoaderModule ORIGIN_ class constant
*/
public function getAllowedModules( $type ) {
if ( $type == ResourceLoaderModule::TYPE_COMBINED ) {
return min( array_values( $this->mAllowedModules ) );
} else {
return isset( $this->mAllowedModules[$type] )
? $this->mAllowedModules[$type]
: ResourceLoaderModule::ORIGIN_ALL;
}
}
/**
* Limit the highest level of CSS/JS untrustworthiness allowed.
*
* If passed the same or a higher level than the current level of untrustworthiness set, the
* level will remain unchanged.
*
* @param string $type
* @param int $level ResourceLoaderModule class constant
*/
public function reduceAllowedModules( $type, $level ) {
$this->mAllowedModules[$type] = min( $this->getAllowedModules( $type ), $level );
}
/**
* Prepend $text to the body HTML
*
* @param string $text HTML
*/
public function prependHTML( $text ) {
$this->mBodytext = $text . $this->mBodytext;
}
/**
* Append $text to the body HTML
*
* @param string $text HTML
*/
public function addHTML( $text ) {
$this->mBodytext .= $text;
}
/**
* Shortcut for adding an Html::element via addHTML.
*
* @since 1.19
*
* @param string $element
* @param array $attribs
* @param string $contents
*/
public function addElement( $element, array $attribs = [], $contents = '' ) {
$this->addHTML( Html::element( $element, $attribs, $contents ) );
}
/**
* Clear the body HTML
*/
public function clearHTML() {
$this->mBodytext = '';
}
/**
* Get the body HTML
*
* @return string HTML
*/
public function getHTML() {
return $this->mBodytext;
}
/**
* Get/set the ParserOptions object to use for wikitext parsing
*
* @param ParserOptions|null $options Either the ParserOption to use or null to only get the
* current ParserOption object
* @return ParserOptions
*/
public function parserOptions( $options = null ) {
if ( $options !== null && !empty( $options->isBogus ) ) {
// Someone is trying to set a bogus pre-$wgUser PO. Check if it has
// been changed somehow, and keep it if so.
$anonPO = ParserOptions::newFromAnon();
$anonPO->setEditSection( false );
$anonPO->setAllowUnsafeRawHtml( false );
if ( !$options->matches( $anonPO ) ) {
wfLogWarning( __METHOD__ . ': Setting a changed bogus ParserOptions: ' . wfGetAllCallers( 5 ) );
$options->isBogus = false;
}
}
Merged localisation-work branch: * Made lines from initialiseMessages() appear as list items during installation * Moved the bulk of the localisation data from the Language*.php files to the Messages*.php files. Deleted most of the Languages*.php files. * Introduced "stub global" framework to provide deferred initialisation of core modules. * Removed placeholder values for $wgTitle and $wgArticle, these variables will now be null during the initialisation process, until they are set by index.php or another entry point. * Added DBA cache type, for BDB-style caches. * Removed custom date format functions, replacing them with a format string in the style of PHP's date(). Used string identifiers instead of integer identifiers, in both the language files and user preferences. Migration should be transparent in most cases. * Simplified the initialisation API for LoadBalancer objects. * Removed the broken altencoding feature. * Moved default user options and toggles from Language to User. Language objects are still able to define default preference overrides and extra user toggles, via a slightly different interface. * Don't include the date option in the parser cache rendering hash unless $wgUseDynamicDates is enabled. * Merged LanguageUtf8 with Language. Removed LanguageUtf8.php. * Removed inclusion of language files from the bottom of Language.php. This is now consistently done from Language::factory(). * Add the name of the executing maintenance script to the debug log. Start the profiler during maintenance scripts. * Added "serialized" directory, for storing precompiled data in serialized form.
2006-07-26 07:15:39 +00:00
if ( !$this->mParserOptions ) {
if ( !$this->getContext()->getUser()->isSafeToLoad() ) {
// $wgUser isn't loaded yet, so don't try to get a
// ParserOptions for it. And don't cache this ParserOptions
// either.
$po = ParserOptions::newFromAnon();
$po->setEditSection( false );
$po->setAllowUnsafeRawHtml( false );
$po->isBogus = true;
if ( $options !== null ) {
$this->mParserOptions = empty( $options->isBogus ) ? $options : null;
}
return $po;
}
$this->mParserOptions = ParserOptions::newFromContext( $this->getContext() );
$this->mParserOptions->setEditSection( false );
$this->mParserOptions->setAllowUnsafeRawHtml( false );
Merged localisation-work branch: * Made lines from initialiseMessages() appear as list items during installation * Moved the bulk of the localisation data from the Language*.php files to the Messages*.php files. Deleted most of the Languages*.php files. * Introduced "stub global" framework to provide deferred initialisation of core modules. * Removed placeholder values for $wgTitle and $wgArticle, these variables will now be null during the initialisation process, until they are set by index.php or another entry point. * Added DBA cache type, for BDB-style caches. * Removed custom date format functions, replacing them with a format string in the style of PHP's date(). Used string identifiers instead of integer identifiers, in both the language files and user preferences. Migration should be transparent in most cases. * Simplified the initialisation API for LoadBalancer objects. * Removed the broken altencoding feature. * Moved default user options and toggles from Language to User. Language objects are still able to define default preference overrides and extra user toggles, via a slightly different interface. * Don't include the date option in the parser cache rendering hash unless $wgUseDynamicDates is enabled. * Merged LanguageUtf8 with Language. Removed LanguageUtf8.php. * Removed inclusion of language files from the bottom of Language.php. This is now consistently done from Language::factory(). * Add the name of the executing maintenance script to the debug log. Start the profiler during maintenance scripts. * Added "serialized" directory, for storing precompiled data in serialized form.
2006-07-26 07:15:39 +00:00
}
if ( $options !== null && !empty( $options->isBogus ) ) {
// They're trying to restore the bogus pre-$wgUser PO. Do the right
// thing.
return wfSetVar( $this->mParserOptions, null, true );
} else {
return wfSetVar( $this->mParserOptions, $options );
}
}
2006-01-07 13:31:29 +00:00
/**
* Set the revision ID which will be seen by the wiki text parser
* for things such as embedded {{REVISIONID}} variable use.
*
* @param int|null $revid An positive integer, or null
* @return mixed Previous value
*/
public function setRevisionId( $revid ) {
$val = is_null( $revid ) ? null : intval( $revid );
return wfSetVar( $this->mRevisionId, $val );
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
/**
* Get the displayed revision ID
*
* @return int
*/
2008-08-11 13:23:45 +00:00
public function getRevisionId() {
return $this->mRevisionId;
}
/**
* Set the timestamp of the revision which will be displayed. This is used
* to avoid a extra DB call in Skin::lastModified().
*
* @param string|null $timestamp
* @return mixed Previous value
*/
public function setRevisionTimestamp( $timestamp ) {
return wfSetVar( $this->mRevisionTimestamp, $timestamp );
}
/**
* Get the timestamp of displayed revision.
* This will be null if not filled by setRevisionTimestamp().
*
* @return string|null
*/
public function getRevisionTimestamp() {
return $this->mRevisionTimestamp;
}
/**
* Set the displayed file version
*
* @param File|bool $file
* @return mixed Previous value
*/
public function setFileVersion( $file ) {
$val = null;
if ( $file instanceof File && $file->exists() ) {
$val = [ 'time' => $file->getTimestamp(), 'sha1' => $file->getSha1() ];
}
return wfSetVar( $this->mFileVersion, $val, true );
}
/**
* Get the displayed file version
*
* @return array|null ('time' => MW timestamp, 'sha1' => sha1)
*/
public function getFileVersion() {
return $this->mFileVersion;
}
/**
* Get the templates used on this page
*
* @return array (namespace => dbKey => revId)
* @since 1.18
*/
public function getTemplateIds() {
return $this->mTemplateIds;
}
/**
* Get the files used on this page
*
* @return array (dbKey => array('time' => MW timestamp or null, 'sha1' => sha1 or ''))
* @since 1.18
*/
public function getFileSearchOptions() {
return $this->mImageTimeKeys;
}
/**
* Convert wikitext to HTML and add it to the buffer
* Default assumes that the current page title will be used.
*
* @param string $text
* @param bool $linestart Is this the start of a line?
* @param bool $interface Is this text in the user interface language?
* @throws MWException
*/
public function addWikiText( $text, $linestart = true, $interface = true ) {
$title = $this->getTitle(); // Work around E_STRICT
if ( !$title ) {
throw new MWException( 'Title is null' );
}
$this->addWikiTextTitle( $text, $title, $linestart, /*tidy*/false, $interface );
}
/**
* Add wikitext with a custom Title object
*
* @param string $text Wikitext
* @param Title &$title
* @param bool $linestart Is this the start of a line?
*/
public function addWikiTextWithTitle( $text, &$title, $linestart = true ) {
$this->addWikiTextTitle( $text, $title, $linestart );
}
/**
* Add wikitext with a custom Title object and tidy enabled.
*
* @param string $text Wikitext
* @param Title &$title
* @param bool $linestart Is this the start of a line?
*/
function addWikiTextTitleTidy( $text, &$title, $linestart = true ) {
$this->addWikiTextTitle( $text, $title, $linestart, true );
}
/**
* Add wikitext with tidy enabled
*
* @param string $text Wikitext
* @param bool $linestart Is this the start of a line?
*/
public function addWikiTextTidy( $text, $linestart = true ) {
$title = $this->getTitle();
$this->addWikiTextTitleTidy( $text, $title, $linestart );
}
/**
* Add wikitext with a custom Title object
*
* @param string $text Wikitext
* @param Title $title
* @param bool $linestart Is this the start of a line?
* @param bool $tidy Whether to use tidy
* @param bool $interface Whether it is an interface message
* (for example disables conversion)
*/
public function addWikiTextTitle( $text, Title $title, $linestart,
$tidy = false, $interface = false
) {
2005-12-04 18:27:59 +00:00
global $wgParser;
$popts = $this->parserOptions();
$oldTidy = $popts->setTidy( $tidy );
$popts->setInterfaceMessage( (bool)$interface );
$parserOutput = $wgParser->getFreshParser()->parse(
$text, $title, $popts,
$linestart, true, $this->mRevisionId
);
$popts->setTidy( $oldTidy );
2006-01-01 20:08:08 +00:00
$this->addParserOutput( $parserOutput );
}
/**
* Add all metadata associated with a ParserOutput object, but without the actual HTML. This
* includes categories, language links, ResourceLoader modules, effects of certain magic words,
* and so on.
*
* @since 1.24
* @param ParserOutput $parserOutput
*/
public function addParserOutputMetadata( $parserOutput ) {
$this->mLanguageLinks += $parserOutput->getLanguageLinks();
$this->addCategoryLinks( $parserOutput->getCategories() );
Implement page status indicators Page status indicators are icons (or short text snippets) usually displayed in the top-right corner of the page, outside of the main content. Basically, <indicator name="foo">[[File:Foo.svg|20px]]</indicator> may be used on a page to place the icon in the indicator area. They are also known as top icons, page icons, heading icons or title icons. I found the discussion on bug 23796 highly illuminating. I suggest that everyone read it before suggesting different design choices. I spent some time with a thesaurus pondering the name. "Emblems" and "badges" were also considered, but the former has a much more limited meaning and the latter is already taken by Wikidata, with a similar but subtly different feature set. I am not aware of any naming conflicts ;) besides new talk page message "indicator" (used by core and Echo in some documents) and OOjs UI indicators (tiny icons like the arrow on a dropdown form element), which shouldn't be confusing. Potential use cases include: * "Lock" indicators for page protection levels * Featured/good article indicators * Redirect shortcuts display ("WP:VPT") * Links to help/manual for special pages * Coordinates?… or globe icon for inline pop-up maps Design features: * Skin-customizable. Skins can fully control where and how indicators are shown, or may just do <?php echo $this->getIndicators(); ?> to output the default structure. By default they are not shown at all. * Extension-customizable. Extensions can call ParserOutput::addIndicator() to insert an indicator from one of the numerous parser hooks. * Wiki-customizable. In addition to just using the parser functions, on-wiki styles and scripts can use the provided classes and ids (.mw-indicator, #mw-indicator-<name>) to customize their display. Design limitations: * Every indicator must have a unique identifier (name). It's not possible to create arrays, or to have several indicators with the same name. In case of duplicates, the latest occurrence of the parser function wins. * Indicators are displayed ordered by their names (and not occurrence order). This ensures consistency across pages and provides a simple means of ordering or grouping them. * Indicators are not stored, tracked or accessible outside of ParserOutput (in particular they're not in the page_props table). They are intended to merely reflect the content or metadata that is already present on the page, and not be data themselves. If you ever think you need to list pages with a given status indicator, instead figure out what it means and use the appropriate tracking category, special page report, already existing page_prop, or other means. Corresponding patch in Vector: I90a8ae15ac8275d084ea5f47b6b2684d5e6c7412. I'll implement support in the other three skins included in the tarball and document it on mediawiki.org after this is merged. Bug: 23796 Change-Id: I2389ff9a5332a2b1d033eb75f0946e5241cfaaf4
2014-09-24 10:44:16 +00:00
$this->setIndicators( $parserOutput->getIndicators() );
$this->mNewSectionLink = $parserOutput->getNewSection();
$this->mHideNewSectionLink = $parserOutput->getHideNewSection();
if ( !$parserOutput->isCacheable() ) {
$this->enableClientCache( false );
}
$this->mNoGallery = $parserOutput->getNoGallery();
$this->mHeadItems = array_merge( $this->mHeadItems, $parserOutput->getHeadItems() );
$this->addModules( $parserOutput->getModules() );
$this->addModuleScripts( $parserOutput->getModuleScripts() );
$this->addModuleStyles( $parserOutput->getModuleStyles() );
$this->addJsConfigVars( $parserOutput->getJsConfigVars() );
$this->mPreventClickjacking = $this->mPreventClickjacking
|| $parserOutput->preventClickjacking();
// Template versioning...
foreach ( (array)$parserOutput->getTemplateIds() as $ns => $dbks ) {
if ( isset( $this->mTemplateIds[$ns] ) ) {
$this->mTemplateIds[$ns] = $dbks + $this->mTemplateIds[$ns];
} else {
$this->mTemplateIds[$ns] = $dbks;
}
}
// File versioning...
foreach ( (array)$parserOutput->getFileSearchOptions() as $dbk => $data ) {
$this->mImageTimeKeys[$dbk] = $data;
}
// Hooks registered in the object
$parserOutputHooks = $this->getConfig()->get( 'ParserOutputHooks' );
foreach ( $parserOutput->getOutputHooks() as $hookInfo ) {
list( $hookName, $data ) = $hookInfo;
if ( isset( $parserOutputHooks[$hookName] ) ) {
call_user_func( $parserOutputHooks[$hookName], $this, $parserOutput, $data );
}
}
// Enable OOUI if requested via ParserOutput
if ( $parserOutput->getEnableOOUI() ) {
$this->enableOOUI();
}
// Include parser limit report
if ( !$this->limitReportJSData ) {
$this->limitReportJSData = $parserOutput->getLimitReportJSData();
}
// Link flags are ignored for now, but may in the future be
// used to mark individual language links.
$linkFlags = [];
// Avoid PHP 7.1 warning of passing $this by reference
$outputPage = $this;
Hooks::run( 'LanguageLinks', [ $this->getTitle(), &$this->mLanguageLinks, &$linkFlags ] );
Hooks::runWithoutAbort( 'OutputPageParserOutput', [ &$outputPage, $parserOutput ] );
// This check must be after 'OutputPageParserOutput' runs in addParserOutputMetadata
// so that extensions may modify ParserOutput to toggle TOC.
// This cannot be moved to addParserOutputText because that is not
// called by EditPage for Preview.
if ( $parserOutput->getTOCEnabled() && $parserOutput->getTOCHTML() ) {
$this->mEnableTOC = true;
}
2006-01-01 20:08:08 +00:00
}
2006-01-07 13:31:29 +00:00
/**
* Add the HTML and enhancements for it (like ResourceLoader modules) associated with a
* ParserOutput object, without any other metadata.
*
* @since 1.24
* @param ParserOutput $parserOutput
*/
public function addParserOutputContent( $parserOutput ) {
$this->addParserOutputText( $parserOutput );
$this->addModules( $parserOutput->getModules() );
$this->addModuleScripts( $parserOutput->getModuleScripts() );
$this->addModuleStyles( $parserOutput->getModuleStyles() );
$this->addJsConfigVars( $parserOutput->getJsConfigVars() );
}
/**
* Add the HTML associated with a ParserOutput object, without any metadata.
*
* @since 1.24
* @param ParserOutput $parserOutput
*/
public function addParserOutputText( $parserOutput ) {
$text = $parserOutput->getText();
// Avoid PHP 7.1 warning of passing $this by reference
$outputPage = $this;
Hooks::runWithoutAbort( 'OutputPageBeforeHTML', [ &$outputPage, &$text ] );
$this->addHTML( $text );
}
/**
* Add everything from a ParserOutput object.
*
* @param ParserOutput $parserOutput
*/
function addParserOutput( $parserOutput ) {
$this->addParserOutputMetadata( $parserOutput );
// Touch section edit links only if not previously disabled
if ( $parserOutput->getEditSectionTokens() ) {
$parserOutput->setEditSectionTokens( $this->mEnableSectionEditLinks );
}
$this->addParserOutputText( $parserOutput );
}
/**
* Add the output of a QuickTemplate to the output buffer
*
* @param QuickTemplate &$template
*/
public function addTemplate( &$template ) {
$this->addHTML( $template->getHTML() );
}
/**
* Parse wikitext and return the HTML.
*
* @param string $text
* @param bool $linestart Is this the start of a line?
* @param bool $interface Use interface language ($wgLang instead of
* $wgContLang) while parsing language sensitive magic words like GRAMMAR and PLURAL.
* This also disables LanguageConverter.
* @param Language $language Target language object, will override $interface
* @throws MWException
* @return string HTML
*/
public function parse( $text, $linestart = true, $interface = false, $language = null ) {
global $wgParser;
if ( is_null( $this->getTitle() ) ) {
throw new MWException( 'Empty $mTitle in ' . __METHOD__ );
}
Merged localisation-work branch: * Made lines from initialiseMessages() appear as list items during installation * Moved the bulk of the localisation data from the Language*.php files to the Messages*.php files. Deleted most of the Languages*.php files. * Introduced "stub global" framework to provide deferred initialisation of core modules. * Removed placeholder values for $wgTitle and $wgArticle, these variables will now be null during the initialisation process, until they are set by index.php or another entry point. * Added DBA cache type, for BDB-style caches. * Removed custom date format functions, replacing them with a format string in the style of PHP's date(). Used string identifiers instead of integer identifiers, in both the language files and user preferences. Migration should be transparent in most cases. * Simplified the initialisation API for LoadBalancer objects. * Removed the broken altencoding feature. * Moved default user options and toggles from Language to User. Language objects are still able to define default preference overrides and extra user toggles, via a slightly different interface. * Don't include the date option in the parser cache rendering hash unless $wgUseDynamicDates is enabled. * Merged LanguageUtf8 with Language. Removed LanguageUtf8.php. * Removed inclusion of language files from the bottom of Language.php. This is now consistently done from Language::factory(). * Add the name of the executing maintenance script to the debug log. Start the profiler during maintenance scripts. * Added "serialized" directory, for storing precompiled data in serialized form.
2006-07-26 07:15:39 +00:00
$popts = $this->parserOptions();
if ( $interface ) {
$popts->setInterfaceMessage( true );
}
if ( $language !== null ) {
$oldLang = $popts->setTargetLanguage( $language );
}
$parserOutput = $wgParser->getFreshParser()->parse(
$text, $this->getTitle(), $popts,
$linestart, true, $this->mRevisionId
);
if ( $interface ) {
$popts->setInterfaceMessage( false );
}
if ( $language !== null ) {
$popts->setTargetLanguage( $oldLang );
}
return $parserOutput->getText();
}
/**
* Parse wikitext, strip paragraphs, and return the HTML.
*
* @param string $text
* @param bool $linestart Is this the start of a line?
* @param bool $interface Use interface language ($wgLang instead of
* $wgContLang) while parsing language sensitive magic
* words like GRAMMAR and PLURAL
* @return string HTML
*/
public function parseInline( $text, $linestart = true, $interface = false ) {
$parsed = $this->parse( $text, $linestart, $interface );
return Parser::stripOuterParagraph( $parsed );
}
/**
* @param int $maxage
* @deprecated since 1.27 Use setCdnMaxage() instead
*/
public function setSquidMaxage( $maxage ) {
$this->setCdnMaxage( $maxage );
}
/**
* Set the value of the "s-maxage" part of the "Cache-control" HTTP header
*
* @param int $maxage Maximum cache time on the CDN, in seconds.
*/
public function setCdnMaxage( $maxage ) {
$this->mCdnMaxage = min( $maxage, $this->mCdnMaxageLimit );
}
/**
* Lower the value of the "s-maxage" part of the "Cache-control" HTTP header
*
* @param int $maxage Maximum cache time on the CDN, in seconds
* @since 1.27
*/
public function lowerCdnMaxage( $maxage ) {
$this->mCdnMaxageLimit = min( $maxage, $this->mCdnMaxageLimit );
$this->setCdnMaxage( $this->mCdnMaxage );
}
/**
* Get TTL in [$minTTL,$maxTTL] in pass it to lowerCdnMaxage()
*
* This sets and returns $minTTL if $mtime is false or null. Otherwise,
* the TTL is higher the older the $mtime timestamp is. Essentially, the
* TTL is 90% of the age of the object, subject to the min and max.
*
* @param string|int|float|bool|null $mtime Last-Modified timestamp
* @param int $minTTL Mimimum TTL in seconds [default: 1 minute]
* @param int $maxTTL Maximum TTL in seconds [default: $wgSquidMaxage]
* @return int TTL in seconds
* @since 1.28
*/
public function adaptCdnTTL( $mtime, $minTTL = 0, $maxTTL = 0 ) {
$minTTL = $minTTL ?: IExpiringStore::TTL_MINUTE;
$maxTTL = $maxTTL ?: $this->getConfig()->get( 'SquidMaxage' );
if ( $mtime === null || $mtime === false ) {
return $minTTL; // entity does not exist
}
$age = time() - wfTimestamp( TS_UNIX, $mtime );
$adaptiveTTL = max( 0.9 * $age, $minTTL );
$adaptiveTTL = min( $adaptiveTTL, $maxTTL );
$this->lowerCdnMaxage( (int)$adaptiveTTL );
return $adaptiveTTL;
}
/**
* Use enableClientCache(false) to force it to send nocache headers
*
* @param bool $state
*
* @return bool
*/
public function enableClientCache( $state ) {
2004-03-23 10:19:31 +00:00
return wfSetVar( $this->mEnableClientCache, $state );
}
/**
* Get the list of cookies that will influence on the cache
*
* @return array
*/
function getCacheVaryCookies() {
static $cookies;
if ( $cookies === null ) {
$config = $this->getConfig();
$cookies = array_merge(
SessionManager::singleton()->getVaryCookies(),
[
'forceHTTPS',
],
$config->get( 'CacheVaryCookies' )
);
Hooks::run( 'GetCacheVaryCookies', [ $this, &$cookies ] );
}
return $cookies;
}
/**
* Check if the request has a cache-varying cookie header
* If it does, it's very important that we don't allow public caching
*
* @return bool
*/
function haveCacheVaryCookies() {
$request = $this->getRequest();
foreach ( $this->getCacheVaryCookies() as $cookieName ) {
if ( $request->getCookie( $cookieName, '', '' ) !== '' ) {
wfDebug( __METHOD__ . ": found $cookieName\n" );
return true;
}
}
wfDebug( __METHOD__ . ": no cache-varying cookies found\n" );
return false;
}
/**
* Add an HTTP header that will influence on the cache
*
* @param string $header Header name
* @param string[]|null $option Options for the Key header. See
* https://datatracker.ietf.org/doc/draft-fielding-http-key/
* for the list of valid options.
*/
public function addVaryHeader( $header, array $option = null ) {
if ( !array_key_exists( $header, $this->mVaryHeader ) ) {
$this->mVaryHeader[$header] = [];
}
if ( !is_array( $option ) ) {
$option = [];
}
$this->mVaryHeader[$header] = array_unique( array_merge( $this->mVaryHeader[$header], $option ) );
}
/**
* Return a Vary: header on which to vary caches. Based on the keys of $mVaryHeader,
* such as Accept-Encoding or Cookie
*
* @return string
*/
public function getVaryHeader() {
// If we vary on cookies, let's make sure it's always included here too.
if ( $this->getCacheVaryCookies() ) {
$this->addVaryHeader( 'Cookie' );
}
foreach ( SessionManager::singleton()->getVaryHeaders() as $header => $options ) {
$this->addVaryHeader( $header, $options );
}
return 'Vary: ' . implode( ', ', array_keys( $this->mVaryHeader ) );
}
/**
* Add an HTTP Link: header
*
* @param string $header Header value
*/
public function addLinkHeader( $header ) {
$this->mLinkHeader[] = $header;
}
/**
* Return a Link: header. Based on the values of $mLinkHeader.
*
* @return string
*/
public function getLinkHeader() {
if ( !$this->mLinkHeader ) {
return false;
}
return 'Link: ' . implode( ',', $this->mLinkHeader );
}
/**
* Get a complete Key header
*
* @return string
*/
public function getKeyHeader() {
$cvCookies = $this->getCacheVaryCookies();
2010-05-22 11:47:56 +00:00
$cookiesOption = [];
foreach ( $cvCookies as $cookieName ) {
$cookiesOption[] = 'param=' . $cookieName;
}
$this->addVaryHeader( 'Cookie', $cookiesOption );
2010-05-22 11:47:56 +00:00
foreach ( SessionManager::singleton()->getVaryHeaders() as $header => $options ) {
$this->addVaryHeader( $header, $options );
}
$headers = [];
foreach ( $this->mVaryHeader as $header => $option ) {
$newheader = $header;
if ( is_array( $option ) && count( $option ) > 0 ) {
$newheader .= ';' . implode( ';', $option );
}
$headers[] = $newheader;
}
$key = 'Key: ' . implode( ',', $headers );
2010-05-22 11:47:56 +00:00
return $key;
}
/**
* T23672: Add Accept-Language to Vary and Key headers
* if there's no 'variant' parameter existed in GET.
*
* For example:
* /w/index.php?title=Main_page should always be served; but
* /w/index.php?title=Main_page&variant=zh-cn should never be served.
*/
function addAcceptLanguage() {
$title = $this->getTitle();
if ( !$title instanceof Title ) {
return;
}
$lang = $title->getPageLanguage();
if ( !$this->getRequest()->getCheck( 'variant' ) && $lang->hasVariants() ) {
$variants = $lang->getVariants();
$aloption = [];
foreach ( $variants as $variant ) {
if ( $variant === $lang->getCode() ) {
continue;
} else {
$aloption[] = 'substr=' . $variant;
// IE and some other browsers use BCP 47 standards in
// their Accept-Language header, like "zh-CN" or "zh-Hant".
// We should handle these too.
$variantBCP47 = wfBCP47( $variant );
if ( $variantBCP47 !== $variant ) {
$aloption[] = 'substr=' . $variantBCP47;
}
}
}
$this->addVaryHeader( 'Accept-Language', $aloption );
}
}
/**
2011-04-23 19:28:35 +00:00
* Set a flag which will cause an X-Frame-Options header appropriate for
* edit pages to be sent. The header value is controlled by
* $wgEditPageFrameOptions.
*
2011-04-23 19:28:35 +00:00
* This is the default for special pages. If you display a CSRF-protected
* form on an ordinary view page, then you need to call this function.
*
* @param bool $enable
*/
public function preventClickjacking( $enable = true ) {
$this->mPreventClickjacking = $enable;
}
/**
* Turn off frame-breaking. Alias for $this->preventClickjacking(false).
* This can be called from pages which do not contain any CSRF-protected
* HTML form.
*/
public function allowClickjacking() {
$this->mPreventClickjacking = false;
}
/**
* Get the prevent-clickjacking flag
*
* @since 1.24
* @return bool
*/
public function getPreventClickjacking() {
return $this->mPreventClickjacking;
}
/**
2011-04-23 19:28:35 +00:00
* Get the X-Frame-Options header value (without the name part), or false
* if there isn't one. This is used by Skin to determine whether to enable
* JavaScript frame-breaking, for clients that don't support X-Frame-Options.
*
* @return string|false
*/
public function getFrameOptions() {
$config = $this->getConfig();
if ( $config->get( 'BreakFrames' ) ) {
return 'DENY';
} elseif ( $this->mPreventClickjacking && $config->get( 'EditPageFrameOptions' ) ) {
return $config->get( 'EditPageFrameOptions' );
}
return false;
}
/**
* Send cache control HTTP headers
*/
public function sendCacheControl() {
$response = $this->getRequest()->response();
$config = $this->getConfig();
$this->addVaryHeader( 'Cookie' );
$this->addAcceptLanguage();
# don't serve compressed data to clients who can't handle it
2005-03-08 02:58:43 +00:00
# maintain different caches for logged-in users and non-logged in ones
$response->header( $this->getVaryHeader() );
if ( $config->get( 'UseKeyHeader' ) ) {
$response->header( $this->getKeyHeader() );
}
if ( $this->mEnableClientCache ) {
if (
$config->get( 'UseSquid' ) &&
!$response->hasCookies() &&
!SessionManager::getGlobalSession()->isPersistent() &&
!$this->isPrintable() &&
$this->mCdnMaxage != 0 &&
!$this->haveCacheVaryCookies()
) {
if ( $config->get( 'UseESI' ) ) {
# We'll purge the proxy cache explicitly, but require end user agents
# to revalidate against the proxy on each visit.
# Surrogate-Control controls our CDN, Cache-Control downstream caches
wfDebug( __METHOD__ .
": proxy caching with ESI; {$this->mLastModified} **", 'private' );
# start with a shorter timeout for initial testing
# header( 'Surrogate-Control: max-age=2678400+2678400, content="ESI/1.0"');
$response->header(
"Surrogate-Control: max-age={$config->get( 'SquidMaxage' )}" .
"+{$this->mCdnMaxage}, content=\"ESI/1.0\""
);
$response->header( 'Cache-Control: s-maxage=0, must-revalidate, max-age=0' );
} else {
# We'll purge the proxy cache for anons explicitly, but require end user agents
# to revalidate against the proxy on each visit.
# IMPORTANT! The CDN needs to replace the Cache-Control header with
# Cache-Control: s-maxage=0, must-revalidate, max-age=0
wfDebug( __METHOD__ .
": local proxy caching; {$this->mLastModified} **", 'private' );
# start with a shorter timeout for initial testing
# header( "Cache-Control: s-maxage=2678400, must-revalidate, max-age=0" );
$response->header( "Cache-Control: " .
"s-maxage={$this->mCdnMaxage}, must-revalidate, max-age=0" );
}
} else {
# We do want clients to cache if they can, but they *must* check for updates
# on revisiting the page.
wfDebug( __METHOD__ . ": private caching; {$this->mLastModified} **", 'private' );
$response->header( 'Expires: ' . gmdate( 'D, d M Y H:i:s', 0 ) . ' GMT' );
$response->header( "Cache-Control: private, must-revalidate, max-age=0" );
}
if ( $this->mLastModified ) {
$response->header( "Last-Modified: {$this->mLastModified}" );
}
} else {
wfDebug( __METHOD__ . ": no caching **", 'private' );
2004-03-23 10:19:31 +00:00
# In general, the absence of a last modified header should be enough to prevent
# the client from using its cache. We send a few other things just to make sure.
$response->header( 'Expires: ' . gmdate( 'D, d M Y H:i:s', 0 ) . ' GMT' );
$response->header( 'Cache-Control: no-cache, no-store, max-age=0, must-revalidate' );
$response->header( 'Pragma: no-cache' );
}
}
/**
* Finally, all the text has been munged and accumulated into
* the object, let's actually output it:
*
* @param bool $return Set to true to get the result as a string rather than sending it
* @return string|null
* @throws Exception
* @throws FatalError
* @throws MWException
*/
public function output( $return = false ) {
global $wgContLang;
if ( $this->mDoNothing ) {
return $return ? '' : null;
}
$response = $this->getRequest()->response();
$config = $this->getConfig();
if ( $this->mRedirect != '' ) {
# Standards require redirect URLs to be absolute
$this->mRedirect = wfExpandUrl( $this->mRedirect, PROTO_CURRENT );
$redirect = $this->mRedirect;
$code = $this->mRedirectCode;
if ( Hooks::run( "BeforePageRedirect", [ $this, &$redirect, &$code ] ) ) {
if ( $code == '301' || $code == '303' ) {
if ( !$config->get( 'DebugRedirects' ) ) {
$response->statusHeader( $code );
}
$this->mLastModified = wfTimestamp( TS_RFC2822 );
}
if ( $config->get( 'VaryOnXFP' ) ) {
$this->addVaryHeader( 'X-Forwarded-Proto' );
}
$this->sendCacheControl();
$response->header( "Content-Type: text/html; charset=utf-8" );
if ( $config->get( 'DebugRedirects' ) ) {
$url = htmlspecialchars( $redirect );
print "<!DOCTYPE html>\n<html>\n<head>\n<title>Redirect</title>\n</head>\n<body>\n";
print "<p>Location: <a href=\"$url\">$url</a></p>\n";
print "</body>\n</html>\n";
} else {
$response->header( 'Location: ' . $redirect );
}
}
return $return ? '' : null;
} elseif ( $this->mStatusCode ) {
$response->statusHeader( $this->mStatusCode );
}
2010-05-22 11:47:56 +00:00
# Buffer output; final headers may depend on later processing
ob_start();
$response->header( 'Content-type: ' . $config->get( 'MimeType' ) . '; charset=UTF-8' );
$response->header( 'Content-language: ' . $wgContLang->getHtmlCode() );
2003-04-14 23:10:40 +00:00
// Avoid Internet Explorer "compatibility view" in IE 8-10, so that
// jQuery etc. can work correctly.
$response->header( 'X-UA-Compatible: IE=Edge' );
if ( !$this->mArticleBodyOnly ) {
$sk = $this->getSkin();
if ( $sk->shouldPreloadLogo() ) {
$this->addLogoPreloadLinkHeaders();
}
}
$linkHeader = $this->getLinkHeader();
if ( $linkHeader ) {
$response->header( $linkHeader );
}
// Prevent framing, if requested
$frameOptions = $this->getFrameOptions();
if ( $frameOptions ) {
$response->header( "X-Frame-Options: $frameOptions" );
}
if ( $this->mArticleBodyOnly ) {
echo $this->mBodytext;
2005-07-03 04:00:33 +00:00
} else {
// Enable safe mode if requested
if ( $this->getRequest()->getBool( 'safemode' ) ) {
$this->disallowUserJs();
}
$sk = $this->getSkin();
foreach ( $sk->getDefaultModules() as $group ) {
$this->addModules( $group );
}
MWDebug::addModules( $this );
// Avoid PHP 7.1 warning of passing $this by reference
$outputPage = $this;
// Hook that allows last minute changes to the output page, e.g.
// adding of CSS or Javascript by extensions.
Hooks::runWithoutAbort( 'BeforePageDisplay', [ &$outputPage, &$sk ] );
try {
$sk->outputPage();
} catch ( Exception $e ) {
ob_end_clean(); // bug T129657
throw $e;
}
2005-07-03 04:00:33 +00:00
}
try {
// This hook allows last minute changes to final overall output by modifying output buffer
Hooks::runWithoutAbort( 'AfterFinalPageOutput', [ $this ] );
} catch ( Exception $e ) {
ob_end_clean(); // bug T129657
throw $e;
}
$this->sendCacheControl();
if ( $return ) {
return ob_get_clean();
} else {
ob_end_flush();
return null;
}
2003-04-14 23:10:40 +00:00
}
/**
* Prepare this object to display an error page; disable caching and
* indexing, clear the current text and redirect, set the page's title
* and optionally an custom HTML title (content of the "<title>" tag).
*
* @param string|Message $pageTitle Will be passed directly to setPageTitle()
* @param string|Message $htmlTitle Will be passed directly to setHTMLTitle();
* optional, if not passed the "<title>" attribute will be
* based on $pageTitle
*/
public function prepareErrorPage( $pageTitle, $htmlTitle = false ) {
$this->setPageTitle( $pageTitle );
if ( $htmlTitle !== false ) {
$this->setHTMLTitle( $htmlTitle );
}
$this->setRobotPolicy( 'noindex,nofollow' );
$this->setArticleRelated( false );
$this->enableClientCache( false );
$this->mRedirect = '';
$this->clearSubtitle();
$this->clearHTML();
}
/**
* Output a standard error page
*
* showErrorPage( 'titlemsg', 'pagetextmsg' );
* showErrorPage( 'titlemsg', 'pagetextmsg', [ 'param1', 'param2' ] );
* showErrorPage( 'titlemsg', $messageObject );
* showErrorPage( $titleMessageObject, $messageObject );
*
* @param string|Message $title Message key (string) for page title, or a Message object
* @param string|Message $msg Message key (string) for page text, or a Message object
* @param array $params Message parameters; ignored if $msg is a Message object
*/
public function showErrorPage( $title, $msg, $params = [] ) {
if ( !$title instanceof Message ) {
$title = $this->msg( $title );
}
$this->prepareErrorPage( $title );
if ( $msg instanceof Message ) {
if ( $params !== [] ) {
trigger_error( 'Argument ignored: $params. The message parameters argument '
. 'is discarded when the $msg argument is a Message object instead of '
. 'a string.', E_USER_NOTICE );
}
$this->addHTML( $msg->parseAsBlock() );
} else {
2011-08-25 10:56:46 +00:00
$this->addWikiMsgArray( $msg, $params );
}
$this->returnToMain();
2003-04-14 23:10:40 +00:00
}
/**
* Output a standard permission error page
*
* @param array $errors Error message keys or [key, param...] arrays
* @param string $action Action that was denied or null if unknown
*/
public function showPermissionsErrorPage( array $errors, $action = null ) {
foreach ( $errors as $key => $error ) {
$errors[$key] = (array)$error;
}
// For some action (read, edit, create and upload), display a "login to do this action"
// error if all of the following conditions are met:
// 1. the user is not logged in
// 2. the only error is insufficient permissions (i.e. no block or something else)
// 3. the error can be avoided simply by logging in
if ( in_array( $action, [ 'read', 'edit', 'createpage', 'createtalk', 'upload' ] )
&& $this->getUser()->isAnon() && count( $errors ) == 1 && isset( $errors[0][0] )
&& ( $errors[0][0] == 'badaccess-groups' || $errors[0][0] == 'badaccess-group0' )
&& ( User::groupHasPermission( 'user', $action )
|| User::groupHasPermission( 'autoconfirmed', $action ) )
) {
$displayReturnto = null;
# Due to T34276, if a user does not have read permissions,
2011-11-29 21:04:20 +00:00
# $this->getTitle() will just give Special:Badtitle, which is
# not especially useful as a returnto parameter. Use the title
# from the request instead, if there was one.
$request = $this->getRequest();
$returnto = Title::newFromText( $request->getVal( 'title', '' ) );
if ( $action == 'edit' ) {
$msg = 'whitelistedittext';
$displayReturnto = $returnto;
} elseif ( $action == 'createpage' || $action == 'createtalk' ) {
$msg = 'nocreatetext';
} elseif ( $action == 'upload' ) {
$msg = 'uploadnologintext';
} else { # Read
$msg = 'loginreqpagetext';
$displayReturnto = Title::newMainPage();
}
$query = [];
if ( $returnto ) {
$query['returnto'] = $returnto->getPrefixedText();
if ( !$request->wasPosted() ) {
$returntoquery = $request->getValues();
unset( $returntoquery['title'] );
unset( $returntoquery['returnto'] );
unset( $returntoquery['returntoquery'] );
$query['returntoquery'] = wfArrayToCgi( $returntoquery );
}
}
$linkRenderer = MediaWikiServices::getInstance()->getLinkRenderer();
$loginLink = $linkRenderer->makeKnownLink(
SpecialPage::getTitleFor( 'Userlogin' ),
$this->msg( 'loginreqlink' )->text(),
[],
$query
);
$this->prepareErrorPage( $this->msg( 'loginreqtitle' ) );
$this->addHTML( $this->msg( $msg )->rawParams( $loginLink )->parse() );
# Don't return to a page the user can't read otherwise
# we'll end up in a pointless loop
if ( $displayReturnto && $displayReturnto->userCan( 'read', $this->getUser() ) ) {
$this->returnToMain( null, $displayReturnto );
}
} else {
$this->prepareErrorPage( $this->msg( 'permissionserrors' ) );
$this->addWikiText( $this->formatPermissionsErrorMessage( $errors, $action ) );
}
}
/**
* Display an error page indicating that a given version of MediaWiki is
* required to use it
*
* @param mixed $version The version of MediaWiki needed to use the page
*/
public function versionRequired( $version ) {
$this->prepareErrorPage( $this->msg( 'versionrequired', $version ) );
$this->addWikiMsg( 'versionrequiredtext', $version );
$this->returnToMain();
}
/**
* Format a list of error messages
*
* @param array $errors Array of arrays returned by Title::getUserPermissionsErrors
* @param string $action Action that was denied or null if unknown
* @return string The wikitext error-messages, formatted into a list.
*/
public function formatPermissionsErrorMessage( array $errors, $action = null ) {
if ( $action == null ) {
$text = $this->msg( 'permissionserrorstext', count( $errors ) )->plain() . "\n\n";
} else {
$action_desc = $this->msg( "action-$action" )->plain();
$text = $this->msg(
'permissionserrorstext-withaction',
count( $errors ),
$action_desc
)->plain() . "\n\n";
}
if ( count( $errors ) > 1 ) {
$text .= '<ul class="permissions-errors">' . "\n";
foreach ( $errors as $error ) {
$text .= '<li>';
$text .= call_user_func_array( [ $this, 'msg' ], $error )->plain();
$text .= "</li>\n";
}
$text .= '</ul>';
} else {
$text .= "<div class=\"permissions-errors\">\n" .
call_user_func_array( [ $this, 'msg' ], reset( $errors ) )->plain() .
"\n</div>";
}
return $text;
}
/**
* Display a page stating that the Wiki is in read-only mode.
* Should only be called after wfReadOnly() has returned true.
*
* Historically, this function was used to show the source of the page that the user
* was trying to edit and _also_ permissions error messages. The relevant code was
* moved into EditPage in 1.19 (r102024 / d83c2a431c2a) and removed here in 1.25.
*
* @deprecated since 1.25; throw the exception directly
* @throws ReadOnlyError
*/
public function readOnlyPage() {
if ( func_num_args() > 0 ) {
throw new MWException( __METHOD__ . ' no longer accepts arguments since 1.25.' );
}
throw new ReadOnlyError;
2003-04-14 23:10:40 +00:00
}
/**
* Turn off regular page output and return an error response
* for when rate limiting has triggered.
*
* @deprecated since 1.25; throw the exception directly
*/
public function rateLimited() {
wfDeprecated( __METHOD__, '1.25' );
throw new ThrottledError;
}
/**
* Show a warning about replica DB lag
*
* If the lag is higher than $wgSlaveLagCritical seconds,
* then the warning is a bit more obvious. If the lag is
* lower than $wgSlaveLagWarning, then no warning is shown.
*
* @param int $lag Slave lag
*/
public function showLagWarning( $lag ) {
$config = $this->getConfig();
if ( $lag >= $config->get( 'SlaveLagWarning' ) ) {
$lag = floor( $lag ); // floor to avoid nano seconds to display
$message = $lag < $config->get( 'SlaveLagCritical' )
? 'lag-warn-normal'
: 'lag-warn-high';
$wrap = Html::rawElement( 'div', [ 'class' => "mw-{$message}" ], "\n$1\n" );
$this->wrapWikiMsg( "$wrap\n", [ $message, $this->getLanguage()->formatNum( $lag ) ] );
}
}
public function showFatalError( $message ) {
$this->prepareErrorPage( $this->msg( 'internalerror' ) );
$this->addHTML( $message );
2003-04-14 23:10:40 +00:00
}
public function showUnexpectedValueError( $name, $val ) {
$this->showFatalError( $this->msg( 'unexpected', $name, $val )->text() );
2003-04-14 23:10:40 +00:00
}
public function showFileCopyError( $old, $new ) {
$this->showFatalError( $this->msg( 'filecopyerror', $old, $new )->text() );
2003-04-14 23:10:40 +00:00
}
public function showFileRenameError( $old, $new ) {
$this->showFatalError( $this->msg( 'filerenameerror', $old, $new )->text() );
2003-04-14 23:10:40 +00:00
}
public function showFileDeleteError( $name ) {
$this->showFatalError( $this->msg( 'filedeleteerror', $name )->text() );
2003-04-14 23:10:40 +00:00
}
public function showFileNotFoundError( $name ) {
$this->showFatalError( $this->msg( 'filenotfound', $name )->text() );
2003-04-14 23:10:40 +00:00
}
/**
* Add a "return to" link pointing to a specified title
*
* @param Title $title Title to link
* @param array $query Query string parameters
* @param string $text Text of the link (input is not escaped)
* @param array $options Options array to pass to Linker
*/
public function addReturnTo( $title, array $query = [], $text = null, $options = [] ) {
$linkRenderer = MediaWikiServices::getInstance()
->getLinkRendererFactory()->createFromLegacyOptions( $options );
$link = $this->msg( 'returnto' )->rawParams(
$linkRenderer->makeLink( $title, $text, [], $query ) )->escaped();
$this->addHTML( "<p id=\"mw-returnto\">{$link}</p>\n" );
}
/**
* Add a "return to" link pointing to a specified title,
* or the title indicated in the request, or else the main page
*
* @param mixed $unused
* @param Title|string $returnto Title or String to return to
* @param string $returntoquery Query string for the return to link
*/
public function returnToMain( $unused = null, $returnto = null, $returntoquery = null ) {
if ( $returnto == null ) {
$returnto = $this->getRequest()->getText( 'returnto' );
}
2009-08-07 00:16:54 +00:00
if ( $returntoquery == null ) {
$returntoquery = $this->getRequest()->getText( 'returntoquery' );
}
if ( $returnto === '' ) {
$returnto = Title::newMainPage();
2003-04-14 23:10:40 +00:00
}
if ( is_object( $returnto ) ) {
$titleObj = $returnto;
} else {
$titleObj = Title::newFromText( $returnto );
}
// We don't want people to return to external interwiki. That
// might potentially be used as part of a phishing scheme
if ( !is_object( $titleObj ) || $titleObj->isExternal() ) {
2006-06-23 06:53:05 +00:00
$titleObj = Title::newMainPage();
}
$this->addReturnTo( $titleObj, wfCgiToArray( $returntoquery ) );
2003-04-14 23:10:40 +00:00
}
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
private function getRlClientContext() {
if ( !$this->rlClientContext ) {
$query = ResourceLoader::makeLoaderQuery(
[], // modules; not relevant
$this->getLanguage()->getCode(),
$this->getSkin()->getSkinName(),
$this->getUser()->isLoggedIn() ? $this->getUser()->getName() : null,
null, // version; not relevant
ResourceLoader::inDebugMode(),
null, // only; not relevant
$this->isPrintable(),
$this->getRequest()->getBool( 'handheld' )
);
$this->rlClientContext = new ResourceLoaderContext(
$this->getResourceLoader(),
new FauxRequest( $query )
);
}
return $this->rlClientContext;
}
/**
* Call this to freeze the module queue and JS config and create a formatter.
*
* Depending on the Skin, this may get lazy-initialised in either headElement() or
* getBottomScripts(). See SkinTemplate::prepareQuickTemplate(). Calling this too early may
* cause unexpected side-effects since disallowUserJs() may be called at any time to change
* the module filters retroactively. Skins and extension hooks may also add modules until very
* late in the request lifecycle.
*
* @return ResourceLoaderClientHtml
*/
public function getRlClient() {
if ( !$this->rlClient ) {
$context = $this->getRlClientContext();
$rl = $this->getResourceLoader();
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$this->addModules( [
'user.options',
'user.tokens',
] );
$this->addModuleStyles( [
'site.styles',
'noscript',
'user.styles',
] );
$this->getSkin()->setupSkinUserCss( $this );
// Prepare exempt modules for buildExemptModules()
$exemptGroups = [ 'site' => [], 'noscript' => [], 'private' => [], 'user' => [] ];
$exemptStates = [];
$moduleStyles = $this->getModuleStyles( /*filter*/ true );
// Preload getTitleInfo for isKnownEmpty calls below and in ResourceLoaderClientHtml
// Separate user-specific batch for improved cache-hit ratio.
$userBatch = [ 'user.styles', 'user' ];
$siteBatch = array_diff( $moduleStyles, $userBatch );
$dbr = wfGetDB( DB_REPLICA );
ResourceLoaderWikiModule::preloadTitleInfo( $context, $dbr, $siteBatch );
ResourceLoaderWikiModule::preloadTitleInfo( $context, $dbr, $userBatch );
// Filter out modules handled by buildExemptModules()
$moduleStyles = array_filter( $moduleStyles,
function ( $name ) use ( $rl, $context, &$exemptGroups, &$exemptStates ) {
$module = $rl->getModule( $name );
if ( $module ) {
if ( $name === 'user.styles' && $this->isUserCssPreview() ) {
$exemptStates[$name] = 'ready';
// Special case in buildExemptModules()
return false;
}
$group = $module->getGroup();
if ( isset( $exemptGroups[$group] ) ) {
$exemptStates[$name] = 'ready';
if ( !$module->isKnownEmpty( $context ) ) {
// E.g. Don't output empty <styles>
$exemptGroups[$group][] = $name;
}
return false;
}
}
return true;
}
);
$this->rlExemptStyleModules = $exemptGroups;
$isUserModuleFiltered = !$this->filterModules( [ 'user' ] );
// If this page filters out 'user', makeResourceLoaderLink will drop it.
// Avoid indefinite "loading" state or untrue "ready" state (T145368).
if ( !$isUserModuleFiltered ) {
// Manually handled by getBottomScripts()
$userModule = $rl->getModule( 'user' );
$userState = $userModule->isKnownEmpty( $context ) && !$this->isUserJsPreview()
? 'ready'
: 'loading';
$this->rlUserModuleState = $exemptStates['user'] = $userState;
}
$rlClient = new ResourceLoaderClientHtml( $context, $this->getTarget() );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$rlClient->setConfig( $this->getJSVars() );
$rlClient->setModules( $this->getModules( /*filter*/ true ) );
$rlClient->setModuleStyles( $moduleStyles );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$rlClient->setModuleScripts( $this->getModuleScripts( /*filter*/ true ) );
$rlClient->setExemptStates( $exemptStates );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$this->rlClient = $rlClient;
}
return $this->rlClient;
}
/**
* @param Skin $sk The given Skin
* @param bool $includeStyle Unused
* @return string The doctype, opening "<html>", and head element.
*/
public function headElement( Skin $sk, $includeStyle = true ) {
global $wgContLang;
$userdir = $this->getLanguage()->getDir();
2011-07-19 20:36:09 +00:00
$sitedir = $wgContLang->getDir();
2003-04-14 23:10:40 +00:00
$pieces = [];
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$pieces[] = Html::htmlHeader( Sanitizer::mergeAttributes(
$this->getRlClient()->getDocumentAttributes(),
$sk->getHtmlElementAttributes()
) );
$pieces[] = Html::openElement( 'head' );
if ( $this->getHTMLTitle() == '' ) {
$this->setHTMLTitle( $this->msg( 'pagetitle', $this->getPageTitle() )->inContentLanguage() );
2003-04-14 23:10:40 +00:00
}
if ( !Html::isXmlMimeType( $this->getConfig()->get( 'MimeType' ) ) ) {
// Add <meta charset="UTF-8">
// This should be before <title> since it defines the charset used by
// text including the text inside <title>.
// The spec recommends defining XHTML5's charset using the XML declaration
// instead of meta.
// Our XML declaration is output by Html::htmlHeader.
// https://html.spec.whatwg.org/multipage/semantics.html#attr-meta-http-equiv-content-type
// https://html.spec.whatwg.org/multipage/semantics.html#charset
$pieces[] = Html::element( 'meta', [ 'charset' => 'UTF-8' ] );
}
$pieces[] = Html::element( 'title', null, $this->getHTMLTitle() );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$pieces[] = $this->getRlClient()->getHeadHtml();
$pieces[] = $this->buildExemptModules();
$pieces = array_merge( $pieces, array_values( $this->getHeadLinksArray() ) );
$pieces = array_merge( $pieces, array_values( $this->mHeadItems ) );
// Use an IE conditional comment to serve the script only to old IE
$pieces[] = '<!--[if lt IE 9]>' .
ResourceLoaderClientHtml::makeLoad(
ResourceLoaderContext::newDummyContext(),
[ 'html5shiv' ],
ResourceLoaderModule::TYPE_SCRIPTS,
[ 'sync' => true ]
) .
'<![endif]-->';
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$pieces[] = Html::closeElement( 'head' );
$bodyClasses = $this->mAdditionalBodyClasses;
$bodyClasses[] = 'mediawiki';
# Classes for LTR/RTL directionality support
$bodyClasses[] = $userdir;
$bodyClasses[] = "sitedir-$sitedir";
$underline = $this->getUser()->getOption( 'underline' );
if ( $underline < 2 ) {
// The following classes can be used here:
// * mw-underline-always
// * mw-underline-never
$bodyClasses[] = 'mw-underline-' . ( $underline ? 'always' : 'never' );
}
if ( $this->getLanguage()->capitalizeAllNouns() ) {
# A <body> class is probably not the best way to do this . . .
$bodyClasses[] = 'capitalize-all-nouns';
}
// Parser feature migration class
// The idea is that this will eventually be removed, after the wikitext
// which requires it is cleaned up.
$bodyClasses[] = 'mw-hide-empty-elt';
$bodyClasses[] = $sk->getPageClasses( $this->getTitle() );
$bodyClasses[] = 'skin-' . Sanitizer::escapeClass( $sk->getSkinName() );
$bodyClasses[] =
'action-' . Sanitizer::escapeClass( Action::getActionName( $this->getContext() ) );
$bodyAttrs = [];
// While the implode() is not strictly needed, it's used for backwards compatibility
// (this used to be built as a string and hooks likely still expect that).
$bodyAttrs['class'] = implode( ' ', $bodyClasses );
// Allow skins and extensions to add body attributes they need
$sk->addToBodyAttributes( $this, $bodyAttrs );
Hooks::run( 'OutputPageBodyAttributes', [ $this, $sk, &$bodyAttrs ] );
$pieces[] = Html::openElement( 'body', $bodyAttrs );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return self::combineWrappedStrings( $pieces );
}
2010-10-02 13:45:35 +00:00
* Made Resources.php return a pure-data array instead of an ugly mix of data and code. This allows the class code to be lazy-loaded with the autoloader, for a performance advantage especially on non-APC installs. And using the convention where if the class is omitted, ResourceLoaderFileModule is assumed, the registration code becomes shorter and simpler. * Modified ResourceLoader to lazy-initialise module objects, for a further performance advantage. * Deleted ResourceLoader::getModules(), provided getModuleNames() instead. Although the startup module needs this functionality, it's slow to generate, so to avoid misuse, it's better to provide a foolproof fast interface and let the startup module do the slow thing itself. * Modified ResourceLoader::register() to optionally accept an info array instead of an object. * Added $wgResourceModules, allowing extensions to efficiently define their own resource loader modules. The trouble with hooks is that they contain code, and code is slow. We've been through all this before with i18n. Hooks are useful as a performance tool only if you call them very rarely. * Moved ResourceLoader settings to their own section in DefaultSettings.php * Added options to ResourceLoaderFileModule equivalent to the $localBasePath and $remoteBasePath parameters, to allow it to be instantiated via the new array style. Also added remoteExtPath, which allows modules to be registered before $wgExtensionAssetsPath is known. * Added OutputPage::getResourceLoader(), mostly for debugging. * The time saving at the moment is about 5ms per request with no extensions, which is significant already with 6 load.php requests for a cold cache page view. This is a much more scalable interface; the relative saving will grow as more extensions are added which use this interface, especially for non-APC installs. Although the interface is backwards compatible, extension updates will follow in a subsequent commit.
2010-11-19 10:41:06 +00:00
/**
* Get a ResourceLoader object associated with this OutputPage
2011-02-21 17:12:33 +00:00
*
* @return ResourceLoader
* Made Resources.php return a pure-data array instead of an ugly mix of data and code. This allows the class code to be lazy-loaded with the autoloader, for a performance advantage especially on non-APC installs. And using the convention where if the class is omitted, ResourceLoaderFileModule is assumed, the registration code becomes shorter and simpler. * Modified ResourceLoader to lazy-initialise module objects, for a further performance advantage. * Deleted ResourceLoader::getModules(), provided getModuleNames() instead. Although the startup module needs this functionality, it's slow to generate, so to avoid misuse, it's better to provide a foolproof fast interface and let the startup module do the slow thing itself. * Modified ResourceLoader::register() to optionally accept an info array instead of an object. * Added $wgResourceModules, allowing extensions to efficiently define their own resource loader modules. The trouble with hooks is that they contain code, and code is slow. We've been through all this before with i18n. Hooks are useful as a performance tool only if you call them very rarely. * Moved ResourceLoader settings to their own section in DefaultSettings.php * Added options to ResourceLoaderFileModule equivalent to the $localBasePath and $remoteBasePath parameters, to allow it to be instantiated via the new array style. Also added remoteExtPath, which allows modules to be registered before $wgExtensionAssetsPath is known. * Added OutputPage::getResourceLoader(), mostly for debugging. * The time saving at the moment is about 5ms per request with no extensions, which is significant already with 6 load.php requests for a cold cache page view. This is a much more scalable interface; the relative saving will grow as more extensions are added which use this interface, especially for non-APC installs. Although the interface is backwards compatible, extension updates will follow in a subsequent commit.
2010-11-19 10:41:06 +00:00
*/
public function getResourceLoader() {
if ( is_null( $this->mResourceLoader ) ) {
$this->mResourceLoader = new ResourceLoader(
$this->getConfig(),
LoggerFactory::getInstance( 'resourceloader' )
);
* Made Resources.php return a pure-data array instead of an ugly mix of data and code. This allows the class code to be lazy-loaded with the autoloader, for a performance advantage especially on non-APC installs. And using the convention where if the class is omitted, ResourceLoaderFileModule is assumed, the registration code becomes shorter and simpler. * Modified ResourceLoader to lazy-initialise module objects, for a further performance advantage. * Deleted ResourceLoader::getModules(), provided getModuleNames() instead. Although the startup module needs this functionality, it's slow to generate, so to avoid misuse, it's better to provide a foolproof fast interface and let the startup module do the slow thing itself. * Modified ResourceLoader::register() to optionally accept an info array instead of an object. * Added $wgResourceModules, allowing extensions to efficiently define their own resource loader modules. The trouble with hooks is that they contain code, and code is slow. We've been through all this before with i18n. Hooks are useful as a performance tool only if you call them very rarely. * Moved ResourceLoader settings to their own section in DefaultSettings.php * Added options to ResourceLoaderFileModule equivalent to the $localBasePath and $remoteBasePath parameters, to allow it to be instantiated via the new array style. Also added remoteExtPath, which allows modules to be registered before $wgExtensionAssetsPath is known. * Added OutputPage::getResourceLoader(), mostly for debugging. * The time saving at the moment is about 5ms per request with no extensions, which is significant already with 6 load.php requests for a cold cache page view. This is a much more scalable interface; the relative saving will grow as more extensions are added which use this interface, especially for non-APC installs. Although the interface is backwards compatible, extension updates will follow in a subsequent commit.
2010-11-19 10:41:06 +00:00
}
return $this->mResourceLoader;
2011-04-23 19:28:35 +00:00
}
* Made Resources.php return a pure-data array instead of an ugly mix of data and code. This allows the class code to be lazy-loaded with the autoloader, for a performance advantage especially on non-APC installs. And using the convention where if the class is omitted, ResourceLoaderFileModule is assumed, the registration code becomes shorter and simpler. * Modified ResourceLoader to lazy-initialise module objects, for a further performance advantage. * Deleted ResourceLoader::getModules(), provided getModuleNames() instead. Although the startup module needs this functionality, it's slow to generate, so to avoid misuse, it's better to provide a foolproof fast interface and let the startup module do the slow thing itself. * Modified ResourceLoader::register() to optionally accept an info array instead of an object. * Added $wgResourceModules, allowing extensions to efficiently define their own resource loader modules. The trouble with hooks is that they contain code, and code is slow. We've been through all this before with i18n. Hooks are useful as a performance tool only if you call them very rarely. * Moved ResourceLoader settings to their own section in DefaultSettings.php * Added options to ResourceLoaderFileModule equivalent to the $localBasePath and $remoteBasePath parameters, to allow it to be instantiated via the new array style. Also added remoteExtPath, which allows modules to be registered before $wgExtensionAssetsPath is known. * Added OutputPage::getResourceLoader(), mostly for debugging. * The time saving at the moment is about 5ms per request with no extensions, which is significant already with 6 load.php requests for a cold cache page view. This is a much more scalable interface; the relative saving will grow as more extensions are added which use this interface, especially for non-APC installs. Although the interface is backwards compatible, extension updates will follow in a subsequent commit.
2010-11-19 10:41:06 +00:00
/**
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* Explicily load or embed modules on a page.
resourceloader: Async all the way Page startup: * Due to the startup module and top queue being asynchronous now, move client-nojs/client-js class handling to OutputPage to ensure there is no flashes of wrongly styled or unstyled content. To preserve compatibility for unsupported browsers, undo the class swap at runtime after the isCompatible() check. ResourceLoader startup module: * Load the startup module with <script async>. * Use DOM methods instead of 'document.write' to create base module request (jquery|mediawiki). mw.loader: * Drop 'async' parameter from mw.loader.load(). * Remove the now-unused code paths for synchronous requests. OutputPage: * Drop '$loadCall' parameter from makeResourceLoaderLink(). Asynchronous is now the default and only way to load JavaScript. This means the 'user' module "conditional document-write scripts" are now a simple "mw.loader.load( url )" call. * Fix incorrect @return of makeResourceLoaderLink(). This returns an array not a string. * Improve documentation of makeResourceLoaderLink(). * Drop '$inHead' parameter from getScriptsForBottomQueue(). No longer used. Compatibility with the $wgResourceLoaderExperimentalAsyncLoading feature is maintained. It just no longer needs to change the way the queue works since it's always asynchronous. The feature flag now only controls whether the bottom queue starts at the bottom or starts at the top. * Remove jQuery.ready() optimisation. This was mostly there to avoid the setTimeout() loop jQuery does to detect dom-ready in IE6/IE7 (which we no longer serve JavaScript at all). And for a bug in Firefox with document.write (which is no longer used as of this commit). Bug: T107399 Change-Id: Icba6d7a87b239bf127a221bc6bc432cfa71a4a72
2015-07-28 02:46:00 +00:00
*
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
* @param array|string $modules One or more module names
* @param string $only ResourceLoaderModule TYPE_ class constant
resourceloader: Async all the way Page startup: * Due to the startup module and top queue being asynchronous now, move client-nojs/client-js class handling to OutputPage to ensure there is no flashes of wrongly styled or unstyled content. To preserve compatibility for unsupported browsers, undo the class swap at runtime after the isCompatible() check. ResourceLoader startup module: * Load the startup module with <script async>. * Use DOM methods instead of 'document.write' to create base module request (jquery|mediawiki). mw.loader: * Drop 'async' parameter from mw.loader.load(). * Remove the now-unused code paths for synchronous requests. OutputPage: * Drop '$loadCall' parameter from makeResourceLoaderLink(). Asynchronous is now the default and only way to load JavaScript. This means the 'user' module "conditional document-write scripts" are now a simple "mw.loader.load( url )" call. * Fix incorrect @return of makeResourceLoaderLink(). This returns an array not a string. * Improve documentation of makeResourceLoaderLink(). * Drop '$inHead' parameter from getScriptsForBottomQueue(). No longer used. Compatibility with the $wgResourceLoaderExperimentalAsyncLoading feature is maintained. It just no longer needs to change the way the queue works since it's always asynchronous. The feature flag now only controls whether the bottom queue starts at the bottom or starts at the top. * Remove jQuery.ready() optimisation. This was mostly there to avoid the setTimeout() loop jQuery does to detect dom-ready in IE6/IE7 (which we no longer serve JavaScript at all). And for a bug in Firefox with document.write (which is no longer used as of this commit). Bug: T107399 Change-Id: Icba6d7a87b239bf127a221bc6bc432cfa71a4a72
2015-07-28 02:46:00 +00:00
* @param array $extraQuery [optional] Array with extra query parameters for the request
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* @return string|WrappedStringList HTML
resourceloader: Async all the way Page startup: * Due to the startup module and top queue being asynchronous now, move client-nojs/client-js class handling to OutputPage to ensure there is no flashes of wrongly styled or unstyled content. To preserve compatibility for unsupported browsers, undo the class swap at runtime after the isCompatible() check. ResourceLoader startup module: * Load the startup module with <script async>. * Use DOM methods instead of 'document.write' to create base module request (jquery|mediawiki). mw.loader: * Drop 'async' parameter from mw.loader.load(). * Remove the now-unused code paths for synchronous requests. OutputPage: * Drop '$loadCall' parameter from makeResourceLoaderLink(). Asynchronous is now the default and only way to load JavaScript. This means the 'user' module "conditional document-write scripts" are now a simple "mw.loader.load( url )" call. * Fix incorrect @return of makeResourceLoaderLink(). This returns an array not a string. * Improve documentation of makeResourceLoaderLink(). * Drop '$inHead' parameter from getScriptsForBottomQueue(). No longer used. Compatibility with the $wgResourceLoaderExperimentalAsyncLoading feature is maintained. It just no longer needs to change the way the queue works since it's always asynchronous. The feature flag now only controls whether the bottom queue starts at the bottom or starts at the top. * Remove jQuery.ready() optimisation. This was mostly there to avoid the setTimeout() loop jQuery does to detect dom-ready in IE6/IE7 (which we no longer serve JavaScript at all). And for a bug in Firefox with document.write (which is no longer used as of this commit). Bug: T107399 Change-Id: Icba6d7a87b239bf127a221bc6bc432cfa71a4a72
2015-07-28 02:46:00 +00:00
*/
public function makeResourceLoaderLink( $modules, $only, array $extraQuery = [] ) {
// Apply 'target' and 'origin' filters
$modules = $this->filterModules( (array)$modules, null, $only );
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return ResourceLoaderClientHtml::makeLoad(
$this->getRlClientContext(),
$modules,
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$only,
$extraQuery
);
}
2010-10-02 13:45:35 +00:00
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
/**
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* Combine WrappedString chunks and filter out empty ones
*
* @param array $chunks
* @return string|WrappedStringList HTML
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
*/
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
protected static function combineWrappedStrings( array $chunks ) {
// Filter out empty values
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$chunks = array_filter( $chunks, 'strlen' );
return WrappedString::join( "\n", $chunks );
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
}
private function isUserJsPreview() {
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return $this->getConfig()->get( 'AllowUserJs' )
&& $this->getTitle()
&& $this->getTitle()->isJsSubpage()
&& $this->userCanPreview();
}
protected function isUserCssPreview() {
return $this->getConfig()->get( 'AllowUserCss' )
&& $this->getTitle()
&& $this->getTitle()->isCssSubpage()
&& $this->userCanPreview();
}
/**
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* JS stuff to put at the bottom of the `<body>`. These are modules with position 'bottom',
* legacy scripts ($this->mScripts), and user JS.
*
* @return string|WrappedStringList HTML
*/
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
public function getBottomScripts() {
$chunks = [];
$chunks[] = $this->getRlClient()->getBodyHtml();
// Legacy non-ResourceLoader scripts
$chunks[] = $this->mScripts;
// Exempt 'user' module
// - May need excludepages for live preview. (T28283)
// - Must use TYPE_COMBINED so its response is handled by mw.loader.implement() which
// ensures execution is scheduled after the "site" module.
// - Don't load if module state is already resolved as "ready".
if ( $this->rlUserModuleState === 'loading' ) {
if ( $this->isUserJsPreview() ) {
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$chunks[] = $this->makeResourceLoaderLink( 'user', ResourceLoaderModule::TYPE_COMBINED,
[ 'excludepage' => $this->getTitle()->getPrefixedDBkey() ]
);
$chunks[] = ResourceLoader::makeInlineScript(
Xml::encodeJsCall( 'mw.loader.using', [
[ 'user', 'site' ],
new XmlJsCode(
'function () {'
. Xml::encodeJsCall( '$.globalEval', [
$this->getRequest()->getText( 'wpTextbox1' )
] )
. '}'
)
] )
);
// FIXME: If the user is previewing, say, ./vector.js, his ./common.js will be loaded
// asynchronously and may arrive *after* the inline script here. So the previewed code
// may execute before ./common.js runs. Normally, ./common.js runs before ./vector.js.
// Similarly, when previewing ./common.js and the user module does arrive first,
// it will arrive without common.js and the inline script runs after.
// Thus running common after the excluded subpage.
} else {
// Load normally
$chunks[] = $this->makeResourceLoaderLink( 'user', ResourceLoaderModule::TYPE_COMBINED );
}
}
if ( $this->limitReportJSData ) {
$chunks[] = ResourceLoader::makeInlineScript(
ResourceLoader::makeConfigSetScript(
[ 'wgPageParseReport' => $this->limitReportJSData ]
)
);
}
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return self::combineWrappedStrings( $chunks );
2012-01-05 23:32:41 +00:00
}
/**
* Get the javascript config vars to include on this page
*
* @return array Array of javascript config vars
* @since 1.23
*/
public function getJsConfigVars() {
return $this->mJsConfigVars;
}
/**
* Add one or more variables to be set in mw.config in JavaScript
2011-04-23 19:28:35 +00:00
*
* @param string|array $keys Key or array of key/value pairs
* @param mixed $value [optional] Value of the configuration variable
*/
public function addJsConfigVars( $keys, $value = null ) {
if ( is_array( $keys ) ) {
foreach ( $keys as $key => $value ) {
$this->mJsConfigVars[$key] = $value;
}
return;
}
$this->mJsConfigVars[$keys] = $value;
}
/**
* Get an array containing the variables to be set in mw.config in JavaScript.
*
* Do not add things here which can be evaluated in ResourceLoaderStartUpModule
2011-10-03 14:59:56 +00:00
* - in other words, page-independent/site-wide variables (without state).
* You will only be adding bloat to the html page and causing page caches to
* have to be purged on configuration changes.
2011-11-29 21:04:20 +00:00
* @return array
*/
SpecialJavaScriptTest: Add export feature Add an 'export' subpage to SpecialJavaScriptTest which allows one to request a self-sufficient JavaScript payload that will bootstrap a ResourceLoader client and load the test suites. This is needed for using Karma (which only loads JavaScript, no full html pages). As such elements from the Skin and OutputPage will not exist. While all QUnit tests in MediaWiki core and most extensions I've seen already use #qunit-fixture, this is now required. This to prevent leakage of elements from one test to another, but it also prevents tests from depending on elements provided by the server. While the Karma setup is still in the pipeline (might land before this commit loses WIP status), for now this can be tested via the 'Special:JavaScriptTest/qunit/plain' subpage. Refactor: * Use HTTP status code 404 in the response for "noframework". * Simplify HTML footprint by using <div id="qunit"> instead of hardcoding the full structure. This feature was added to QUnit since v1.3.0 (Feb 2012), we're using v1.14.0 (Jan 2014). QUnit's header is automatically derived from document.title. * Remove redundant addModules() for 'test.mediawiki.qunit.testrunner'. This is already added by default. * Move allowClickjacking() call so that it applies to other modes as well. The exported javascript needs to have wgBreakFrame set to false so that test runners can frame it. * Change mediawiki.special.javaScriptTest to not depend on QUnit. It caused QUnit to load on error pages. And in theory the page is suited for other frameworks and shouldn't load QUnit this way. Bug: T74063 Change-Id: I3d4d0df43bb426d9579eb0349b8b5477281a7cfc
2014-12-02 21:48:21 +00:00
public function getJSVars() {
global $wgContLang;
$curRevisionId = 0;
$articleId = 0;
$canonicalSpecialPageName = false; # T23115
$title = $this->getTitle();
$ns = $title->getNamespace();
$canonicalNamespace = MWNamespace::exists( $ns )
? MWNamespace::getCanonicalName( $ns )
: $title->getNsText();
$sk = $this->getSkin();
2012-03-13 17:57:54 +00:00
// Get the relevant title so that AJAX features can use the correct page name
// when making API requests from certain special pages (T36972).
$relevantTitle = $sk->getRelevantTitle();
$relevantUser = $sk->getRelevantUser();
2012-03-13 17:57:54 +00:00
if ( $ns == NS_SPECIAL ) {
list( $canonicalSpecialPageName, /*...*/ ) =
SpecialPageFactory::resolveAlias( $title->getDBkey() );
} elseif ( $this->canUseWikiPage() ) {
$wikiPage = $this->getWikiPage();
$curRevisionId = $wikiPage->getLatest();
$articleId = $wikiPage->getId();
}
$lang = $title->getPageViewLanguage();
// Pre-process information
$separatorTransTable = $lang->separatorTransformTable();
$separatorTransTable = $separatorTransTable ? $separatorTransTable : [];
$compactSeparatorTransTable = [
implode( "\t", array_keys( $separatorTransTable ) ),
implode( "\t", $separatorTransTable ),
];
$digitTransTable = $lang->digitTransformTable();
$digitTransTable = $digitTransTable ? $digitTransTable : [];
$compactDigitTransTable = [
implode( "\t", array_keys( $digitTransTable ) ),
implode( "\t", $digitTransTable ),
];
$user = $this->getUser();
$vars = [
'wgCanonicalNamespace' => $canonicalNamespace,
'wgCanonicalSpecialPageName' => $canonicalSpecialPageName,
'wgNamespaceNumber' => $title->getNamespace(),
'wgPageName' => $title->getPrefixedDBkey(),
'wgTitle' => $title->getText(),
'wgCurRevisionId' => $curRevisionId,
'wgRevisionId' => (int)$this->getRevisionId(),
'wgArticleId' => $articleId,
'wgIsArticle' => $this->isArticle(),
'wgIsRedirect' => $title->isRedirect(),
'wgAction' => Action::getActionName( $this->getContext() ),
'wgUserName' => $user->isAnon() ? null : $user->getName(),
'wgUserGroups' => $user->getEffectiveGroups(),
'wgCategories' => $this->getCategories(),
'wgBreakFrames' => $this->getFrameOptions() == 'DENY',
'wgPageContentLanguage' => $lang->getCode(),
'wgPageContentModel' => $title->getContentModel(),
'wgSeparatorTransformTable' => $compactSeparatorTransTable,
'wgDigitTransformTable' => $compactDigitTransTable,
'wgDefaultDateFormat' => $lang->getDefaultDateFormat(),
'wgMonthNames' => $lang->getMonthNamesArray(),
'wgMonthNamesShort' => $lang->getMonthAbbreviationsArray(),
'wgRelevantPageName' => $relevantTitle->getPrefixedDBkey(),
'wgRelevantArticleId' => $relevantTitle->getArticleID(),
Provide a unique request identifier When MediaWiki encounters an unhandled exception, the error message it produces includes a randomly-generated token, which allows the exception details to be looked up in the error logs. This is useful but narrow: would it not be useful to have the ability to retrieve all log records associated with a particular request, rather than just exception details? (Hint: yes.) So: introduce the notion of a request-global unique ID, retrievable via WebRequest::getRequestId(). When MediaWiki is behind Apache + mod_unique_id (which provides the same facility) or some other software which sets a UNIQUE_ID envvar, the value of that envvar is used as the request ID. Otherwise, it is a randomly-generated 24-character string. The request ID supplants exception-specific IDs; MWExceptionHandler::getLogId() is deprecated, accordingly. The request ID is also added as an annotation to all Monolog-processed log records, and is exposed client-side as 'wgRequestId'. This allows developers to associate a page view with log records even when the page view does not result in an unhandled exception. (For the WMF, I also intend to add it as an annotation to profiling data). The request ID is not a tracking token; it does not persist, and it is associated with a backend request, not with a particular user or a particular session. Like the data in the NewPP report, the request ID is designed to be cacheable, so that if, for example, a developer notices something weird in the HTML, s/he can associate the output with a backend request regardless of whether the response was served from the cache or directly from the backend. Some prior art: * https://httpd.apache.org/docs/2.4/mod/mod_unique_id.html * http://api.rubyonrails.org/classes/ActionDispatch/RequestId.html * https://github.com/dabapps/django-log-request-id * https://packagist.org/packages/php-middleware/request-id * https://github.com/rhyselsmore/flask-request-id Change-Id: Iaf90c20c330e0470b9b98627a0228cadefd301d1
2016-03-25 01:43:23 +00:00
'wgRequestId' => WebRequest::getRequestId(),
];
if ( $user->isLoggedIn() ) {
$vars['wgUserId'] = $user->getId();
$vars['wgUserEditCount'] = $user->getEditCount();
$userReg = $user->getRegistration();
$vars['wgUserRegistration'] = $userReg ? wfTimestamp( TS_UNIX, $userReg ) * 1000 : null;
// Get the revision ID of the oldest new message on the user's talk
// page. This can be used for constructing new message alerts on
// the client side.
$vars['wgUserNewMsgRevisionId'] = $user->getNewMessageRevisionId();
}
if ( $wgContLang->hasVariants() ) {
$vars['wgUserVariant'] = $wgContLang->getPreferredVariant();
}
// Same test as SkinTemplate
$vars['wgIsProbablyEditable'] = $title->quickUserCan( 'edit', $user )
&& ( $title->exists() || $title->quickUserCan( 'create', $user ) );
$vars['wgRelevantPageIsProbablyEditable'] = $relevantTitle
&& $relevantTitle->quickUserCan( 'edit', $user )
&& ( $relevantTitle->exists() || $relevantTitle->quickUserCan( 'create', $user ) );
foreach ( $title->getRestrictionTypes() as $type ) {
$vars['wgRestriction' . ucfirst( $type )] = $title->getRestrictions( $type );
}
if ( $title->isMainPage() ) {
$vars['wgIsMainPage'] = true;
}
if ( $this->mRedirectedFrom ) {
$vars['wgRedirectedFrom'] = $this->mRedirectedFrom->getPrefixedDBkey();
}
if ( $relevantUser ) {
$vars['wgRelevantUserName'] = $relevantUser->getName();
}
2011-04-23 19:28:35 +00:00
// Allow extensions to add their custom variables to the mw.config map.
// Use the 'ResourceLoaderGetConfigVars' hook if the variable is not
// page-dependant but site-wide (without state).
// Alternatively, you may want to use OutputPage->addJsConfigVars() instead.
Hooks::run( 'MakeGlobalVariablesScript', [ &$vars, $this ] );
2011-04-23 19:28:35 +00:00
// Merge in variables from addJsConfigVars last
return array_merge( $vars, $this->getJsConfigVars() );
}
/**
* To make it harder for someone to slip a user a fake
* user-JavaScript or user-CSS preview, a random token
* is associated with the login session. If it's not
* passed back with the preview request, we won't render
* the code.
*
* @return bool
*/
public function userCanPreview() {
$request = $this->getRequest();
if (
$request->getVal( 'action' ) !== 'submit' ||
!$request->getCheck( 'wpPreview' ) ||
!$request->wasPosted()
) {
return false;
}
$user = $this->getUser();
if ( !$user->isLoggedIn() ) {
// Anons have predictable edit tokens
return false;
}
if ( !$user->matchEditToken( $request->getVal( 'wpEditToken' ) ) ) {
return false;
}
$title = $this->getTitle();
if ( !$title->isJsSubpage() && !$title->isCssSubpage() ) {
return false;
}
if ( !$title->isSubpageOf( $user->getUserPage() ) ) {
// Don't execute another user's CSS or JS on preview (T85855)
return false;
}
$errors = $title->getUserPermissionsErrors( 'edit', $user );
if ( count( $errors ) !== 0 ) {
return false;
}
return true;
}
/**
* @return array Array in format "link name or number => 'link html'".
*/
public function getHeadLinksArray() {
global $wgVersion;
$tags = [];
$config = $this->getConfig();
$canonicalUrl = $this->mCanonicalUrl;
$tags['meta-generator'] = Html::element( 'meta', [
'name' => 'generator',
'content' => "MediaWiki $wgVersion",
] );
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
if ( $config->get( 'ReferrerPolicy' ) !== false ) {
$tags['meta-referrer'] = Html::element( 'meta', [
'name' => 'referrer',
'content' => $config->get( 'ReferrerPolicy' )
] );
}
$p = "{$this->mIndexPolicy},{$this->mFollowPolicy}";
if ( $p !== 'index,follow' ) {
// http://www.robotstxt.org/wc/meta-user.html
// Only show if it's different from the default robots policy
$tags['meta-robots'] = Html::element( 'meta', [
'name' => 'robots',
'content' => $p,
] );
}
2003-04-14 23:10:40 +00:00
foreach ( $this->mMetatags as $tag ) {
if ( strncasecmp( $tag[0], 'http:', 5 ) === 0 ) {
$a = 'http-equiv';
$tag[0] = substr( $tag[0], 5 );
} elseif ( strncasecmp( $tag[0], 'og:', 3 ) === 0 ) {
$a = 'property';
} else {
$a = 'name';
2003-04-14 23:10:40 +00:00
}
$tagName = "meta-{$tag[0]}";
if ( isset( $tags[$tagName] ) ) {
$tagName .= $tag[1];
}
$tags[$tagName] = Html::element( 'meta',
[
$a => $tag[0],
'content' => $tag[1]
]
);
2003-04-14 23:10:40 +00:00
}
2003-04-14 23:10:40 +00:00
foreach ( $this->mLinktags as $tag ) {
2009-09-18 20:10:25 +00:00
$tags[] = Html::element( 'link', $tag );
2003-04-14 23:10:40 +00:00
}
# Universal edit button
if ( $config->get( 'UniversalEditButton' ) && $this->isArticleRelated() ) {
$user = $this->getUser();
if ( $this->getTitle()->quickUserCan( 'edit', $user )
&& ( $this->getTitle()->exists() ||
$this->getTitle()->quickUserCan( 'create', $user ) )
) {
// Original UniversalEditButton
$msg = $this->msg( 'edit' )->text();
$tags['universal-edit-button'] = Html::element( 'link', [
'rel' => 'alternate',
'type' => 'application/x-wiki',
'title' => $msg,
'href' => $this->getTitle()->getEditURL(),
] );
// Alternate edit link
$tags['alternative-edit'] = Html::element( 'link', [
'rel' => 'edit',
'title' => $msg,
'href' => $this->getTitle()->getEditURL(),
] );
}
}
# Generally the order of the favicon and apple-touch-icon links
# should not matter, but Konqueror (3.5.9 at least) incorrectly
# uses whichever one appears later in the HTML source. Make sure
# apple-touch-icon is specified first to avoid this.
if ( $config->get( 'AppleTouchIcon' ) !== false ) {
$tags['apple-touch-icon'] = Html::element( 'link', [
'rel' => 'apple-touch-icon',
'href' => $config->get( 'AppleTouchIcon' )
] );
}
if ( $config->get( 'Favicon' ) !== false ) {
$tags['favicon'] = Html::element( 'link', [
'rel' => 'shortcut icon',
'href' => $config->get( 'Favicon' )
] );
}
# OpenSearch description link
$tags['opensearch'] = Html::element( 'link', [
'rel' => 'search',
'type' => 'application/opensearchdescription+xml',
'href' => wfScript( 'opensearch_desc' ),
'title' => $this->msg( 'opensearch-desc' )->inContentLanguage()->text(),
] );
if ( $config->get( 'EnableAPI' ) ) {
# Real Simple Discovery link, provides auto-discovery information
# for the MediaWiki API (and potentially additional custom API
# support such as WordPress or Twitter-compatible APIs for a
# blogging extension, etc)
$tags['rsd'] = Html::element( 'link', [
'rel' => 'EditURI',
'type' => 'application/rsd+xml',
// Output a protocol-relative URL here if $wgServer is protocol-relative.
// Whether RSD accepts relative or protocol-relative URLs is completely
// undocumented, though.
'href' => wfExpandUrl( wfAppendQuery(
wfScript( 'api' ),
[ 'action' => 'rsd' ] ),
PROTO_RELATIVE
),
] );
}
# Language variants
if ( !$config->get( 'DisableLangConversion' ) ) {
$lang = $this->getTitle()->getPageLanguage();
if ( $lang->hasVariants() ) {
$variants = $lang->getVariants();
foreach ( $variants as $variant ) {
$tags["variant-$variant"] = Html::element( 'link', [
'rel' => 'alternate',
'hreflang' => wfBCP47( $variant ),
'href' => $this->getTitle()->getLocalURL(
[ 'variant' => $variant ] )
]
);
}
# x-default link per https://support.google.com/webmasters/answer/189077?hl=en
$tags["variant-x-default"] = Html::element( 'link', [
'rel' => 'alternate',
'hreflang' => 'x-default',
'href' => $this->getTitle()->getLocalURL() ] );
}
}
# Copyright
if ( $this->copyrightUrl !== null ) {
$copyright = $this->copyrightUrl;
} else {
$copyright = '';
if ( $config->get( 'RightsPage' ) ) {
$copy = Title::newFromText( $config->get( 'RightsPage' ) );
if ( $copy ) {
$copyright = $copy->getLocalURL();
}
}
if ( !$copyright && $config->get( 'RightsUrl' ) ) {
$copyright = $config->get( 'RightsUrl' );
}
}
if ( $copyright ) {
$tags['copyright'] = Html::element( 'link', [
'rel' => 'license',
'href' => $copyright ]
);
}
# Feeds
if ( $config->get( 'Feed' ) ) {
$feedLinks = [];
foreach ( $this->getSyndicationLinks() as $format => $link ) {
# Use the page name for the title. In principle, this could
# lead to issues with having the same name for different feeds
# corresponding to the same page, but we can't avoid that at
# this low a level.
$feedLinks[] = $this->feedLink(
$format,
$link,
# Used messages: 'page-rss-feed' and 'page-atom-feed' (for an easier grep)
$this->msg(
"page-{$format}-feed", $this->getTitle()->getPrefixedText()
)->text()
);
}
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
# Recent changes feed should appear on every page (except recentchanges,
# that would be redundant). Put it after the per-page feed to avoid
# changing existing behavior. It's still available, probably via a
# menu in your browser. Some sites might have a different feed they'd
# like to promote instead of the RC feed (maybe like a "Recent New Articles"
# or "Breaking news" one). For this, we see if $wgOverrideSiteFeed is defined.
# If so, use it instead.
$sitename = $config->get( 'Sitename' );
if ( $config->get( 'OverrideSiteFeed' ) ) {
foreach ( $config->get( 'OverrideSiteFeed' ) as $type => $feedUrl ) {
// Note, this->feedLink escapes the url.
$feedLinks[] = $this->feedLink(
$type,
$feedUrl,
$this->msg( "site-{$type}-feed", $sitename )->text()
);
}
} elseif ( !$this->getTitle()->isSpecial( 'Recentchanges' ) ) {
$rctitle = SpecialPage::getTitleFor( 'Recentchanges' );
foreach ( $config->get( 'AdvertisedFeedTypes' ) as $format ) {
$feedLinks[] = $this->feedLink(
$format,
$rctitle->getLocalURL( [ 'feed' => $format ] ),
# For grep: 'site-rss-feed', 'site-atom-feed'
$this->msg( "site-{$format}-feed", $sitename )->text()
);
}
}
# Allow extensions to change the list pf feeds. This hook is primarily for changing,
# manipulating or removing existing feed tags. If you want to add new feeds, you should
# use OutputPage::addFeedLink() instead.
Hooks::run( 'AfterBuildFeedLinks', [ &$feedLinks ] );
$tags += $feedLinks;
}
# Canonical URL
if ( $config->get( 'EnableCanonicalServerLink' ) ) {
if ( $canonicalUrl !== false ) {
$canonicalUrl = wfExpandUrl( $canonicalUrl, PROTO_CANONICAL );
} else {
Construct clean canonical URLs for wiki pages, ignoring request URL Canonical URLs allow webmasters to indicate the preferred URL form for accessing some content that can be reached via a multitude of URL patterns. It is usually (but not always) distinct from the request URL, which may feature things like aliases and session-specific query parameters. We currently derive canonical URLs from request URLs, which is backwards: it is the web application, not the client, that ought to know the canonical way to refer to some content. This patch ensures MediaWiki derives a clean canonical URL for all wiki pages from the request context's title object and action. For some assurance that this is the correct approach, see: http://googlewebmastercentral.blogspot.com/2009/02/specify-your-canonical.html This Google blog post identifies Wikia as exemplary in its usage of canonical URLs. Wikia disregards things like the requested revision ID (oldid=NNN) when constructing the canonical URL. See, for example: http://fallout.wikia.com/wiki/Aqua_Pura_delivery_program?oldid=2171222 Wikia goes as far as canonicalizing the action=history to the page view URL. I think that this is incorrect, because the history and info actions are not views of the page content, but rather its associated metadata. This affects all requests where "setArticleRelated" is true. This is typically all urls that show content (title query, curid, oldid, diff), and all actions thereof (edit, delete, purge, info, history etc.). It does not apply to File pages and Special pages. Bug: T67402 Change-Id: I1549ca056637981a0d751020c634b9fab387f7bc
2015-06-22 04:26:22 +00:00
if ( $this->isArticleRelated() ) {
// This affects all requests where "setArticleRelated" is true. This is
// typically all requests that show content (query title, curid, oldid, diff),
// and all wikipage actions (edit, delete, purge, info, history etc.).
// It does not apply to File pages and Special pages.
// 'history' and 'info' actions address page metadata rather than the page
// content itself, so they may not be canonicalized to the view page url.
// TODO: this ought to be better encapsulated in the Action class.
$action = Action::getActionName( $this->getContext() );
if ( in_array( $action, [ 'history', 'info' ] ) ) {
Construct clean canonical URLs for wiki pages, ignoring request URL Canonical URLs allow webmasters to indicate the preferred URL form for accessing some content that can be reached via a multitude of URL patterns. It is usually (but not always) distinct from the request URL, which may feature things like aliases and session-specific query parameters. We currently derive canonical URLs from request URLs, which is backwards: it is the web application, not the client, that ought to know the canonical way to refer to some content. This patch ensures MediaWiki derives a clean canonical URL for all wiki pages from the request context's title object and action. For some assurance that this is the correct approach, see: http://googlewebmastercentral.blogspot.com/2009/02/specify-your-canonical.html This Google blog post identifies Wikia as exemplary in its usage of canonical URLs. Wikia disregards things like the requested revision ID (oldid=NNN) when constructing the canonical URL. See, for example: http://fallout.wikia.com/wiki/Aqua_Pura_delivery_program?oldid=2171222 Wikia goes as far as canonicalizing the action=history to the page view URL. I think that this is incorrect, because the history and info actions are not views of the page content, but rather its associated metadata. This affects all requests where "setArticleRelated" is true. This is typically all urls that show content (title query, curid, oldid, diff), and all actions thereof (edit, delete, purge, info, history etc.). It does not apply to File pages and Special pages. Bug: T67402 Change-Id: I1549ca056637981a0d751020c634b9fab387f7bc
2015-06-22 04:26:22 +00:00
$query = "action={$action}";
} else {
$query = '';
}
$canonicalUrl = $this->getTitle()->getCanonicalURL( $query );
} else {
$reqUrl = $this->getRequest()->getRequestURL();
$canonicalUrl = wfExpandUrl( $reqUrl, PROTO_CANONICAL );
}
}
}
if ( $canonicalUrl !== false ) {
$tags[] = Html::element( 'link', [
'rel' => 'canonical',
'href' => $canonicalUrl
] );
}
return $tags;
}
/**
* Generate a "<link rel/>" for a feed.
*
* @param string $type Feed type
* @param string $url URL to the feed
* @param string $text Value of the "title" attribute
* @return string HTML fragment
*/
private function feedLink( $type, $url, $text ) {
return Html::element( 'link', [
'rel' => 'alternate',
'type' => "application/$type+xml",
'title' => $text,
'href' => $url ]
);
}
/**
* Add a local or specified stylesheet, with the given media options.
* Internal use only. Use OutputPage::addModuleStyles() if possible.
*
* @param string $style URL to the file
* @param string $media To specify a media type, 'screen', 'printable', 'handheld' or any.
* @param string $condition For IE conditional comments, specifying an IE version
* @param string $dir Set to 'rtl' or 'ltr' for direction-specific sheets
*/
public function addStyle( $style, $media = '', $condition = '', $dir = '' ) {
$options = [];
if ( $media ) {
$options['media'] = $media;
}
if ( $condition ) {
$options['condition'] = $condition;
}
if ( $dir ) {
$options['dir'] = $dir;
}
$this->styles[$style] = $options;
}
/**
* Adds inline CSS styles
* Internal use only. Use OutputPage::addModuleStyles() if possible.
*
* @param mixed $style_css Inline CSS
* @param string $flip Set to 'flip' to flip the CSS if needed
*/
public function addInlineStyle( $style_css, $flip = 'noflip' ) {
if ( $flip === 'flip' && $this->getLanguage()->isRTL() ) {
# If wanted, and the interface is right-to-left, flip the CSS
$style_css = CSSJanus::transform( $style_css, true, false );
}
$this->mInlineStyles .= Html::inlineStyle( $style_css );
here it is ... the upload-api, script-server, js2 (javascript phase2) branch merge 1st attempt. Here is a short overview of changes and associated default configuration variables (most everything is off by default) also see ~soon to be updated~: http://www.mediawiki.org/wiki/Media_Projects_Overview = Upload Improvements = ==Upload API == * Based on the early work of Bryan Tong and others it adds the upload option to the api. * We rewrite Special:Upload page to include use the new refactoring * Added in token checks in both the SpecialUpload.php page so avoids DOS / xss copy-by-url JavaScript based cross site POST file submissions == Copy by URL== $wgAllowCopyUploads = false; * http class rewrite includes a new http background download see: includes/HttpFunctions.php * spins off a php process that calls: maintenance/http_session_download.php * pushes updates to the session and gives the user a progress bar on http copy uploads from other server progress (using js2 upload interface) (if not using the js2 upload interface it does the request in-place but the download is limited to the php ini timeout time) == Firefogg == * Firefogg enables resumable upload by chunks * progress indicators and conditional invokation (js2 system) * and of-course client side transcoding. = Script Server = $wgEnableScriptLoader = false; * off by default if $wgEnableScriptLoader is turned on script files are grouped, gziped, cached etc. for more info see: http://www.mediawiki.org/wiki/Extension:ScriptLoader * Includes some early skin js include fixes (skin/script system still lots of love) * Includes a "javascript class autoloader" this is packaged into mwEmbed so that the mwEmbed library can work in stand alone mode (while retaining localization and script serving) (one such application is the make page for firefogg.org : http://www.firefogg.org/make/index.html ) * The file that contains the autojavascript loading classes is: js2/php/jsAutoloadLocalClasses.php * One can use this auto class loading dependency system with extensions and add-ons but I need to better document that. = js2 system / mwEmbed= $wgEnableJS2system = false * includes initial rewrite towards more jquery based javascript code * especially for the Special:Upload page. * Also the edit page include support for the "add-media-wizard" * includes dependency loader for javascript that optionally takes advantage of the script-loader * remote embedding of javascript interfaces (like embedding video, or commons media searching) * $wgDebugJavaScript = false; .. .this variable lets you always get "always fresh javascript". When used with the script-loader it does not minify the script-loader output. = mwEmbed = * Will commit a separate patch to oggHandler that conditionally outputs <video tag> to use the new javascript video player. ** mv_embed player includes: play-head, volume control, remote embedding, oggz-chop support across plugins. * add-media-wizard adds easy inserts of media to pages (with import) == jQuery== * we include a base install of jQuery, jQuery ui and some plugins. * all the javascript classes are in the scriptloader so its easy to load any set of jquery ui components that you may need using the script-server. You get a callback so you can then execute js with dependencies loaded. == other stuff == there is a bit more code in js2 that pertains to sequence editing, timed text display and basic image editing. We include a base import of pixastic-lib & pixastic-editor... will work with the pixastic developer to try and ensure upstream compatibility on our usage of the library for in-browser photo and sequence manipulation.
2009-07-14 23:52:14 +00:00
}
/**
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
* Build exempt modules and legacy non-ResourceLoader styles.
*
* @return string|WrappedStringList HTML
*/
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
protected function buildExemptModules() {
global $wgContLang;
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$chunks = [];
// Things that go after the ResourceLoaderDynamicStyles marker
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$append = [];
// Exempt 'user' styles module (may need 'excludepages' for live preview)
if ( $this->isUserCssPreview() ) {
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$append[] = $this->makeResourceLoaderLink(
'user.styles',
ResourceLoaderModule::TYPE_STYLES,
[ 'excludepage' => $this->getTitle()->getPrefixedDBkey() ]
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
);
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
// Load the previewed CSS. Janus it if needed.
// User-supplied CSS is assumed to in the wiki's content language.
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
$previewedCSS = $this->getRequest()->getText( 'wpTextbox1' );
if ( $this->getLanguage()->getDir() !== $wgContLang->getDir() ) {
$previewedCSS = CSSJanus::transform( $previewedCSS, true, false );
}
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
$append[] = Html::inlineStyle( $previewedCSS );
}
// We want site, private and user styles to override dynamically added styles from
// general modules, but we want dynamically added styles to override statically added
// style modules. So the order has to be:
// - page style modules (formatted by ResourceLoaderClientHtml::getHeadHtml())
// - dynamically loaded styles (added by mw.loader before ResourceLoaderDynamicStyles)
// - ResourceLoaderDynamicStyles marker
// - site/private/user styles
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
// Add legacy styles added through addStyle()/addInlineStyle() here
$chunks[] = implode( '', $this->buildCssLinksArray() ) . $this->mInlineStyles;
$chunks[] = Html::element(
'meta',
[ 'name' => 'ResourceLoaderDynamicStyles', 'content' => '' ]
);
2011-04-23 19:28:35 +00:00
$separateReq = [ 'site.styles', 'user.styles' ];
foreach ( $this->rlExemptStyleModules as $group => $moduleNames ) {
// Combinable modules
$chunks[] = $this->makeResourceLoaderLink(
array_diff( $moduleNames, $separateReq ),
resourceloader: Refactor module links output Changes: * Removed hardcoded logic in OutputPage regarding modules being "enabled". Previously we would always output state=loading and use $wgAllowUserJs (and others) to decide whether to output state=ready or makeResourceLoaderLink. Now, we no longer unconditionally output state=loading and simply always call makeResourceLoaderLink. That method takes care of checking whether modules are enabled and non-empty and returns state=ready when that is the case. This cleans up cases where the duplicated and incomplete logic in OutputPage thought the module was non-empty but turned out to be empty and thus would output both state=loading and later state=ready for the same module. * Clean up documentation for makeResourceLoaderLink (inconsistent ordering of type hint and $var, and @return was missing the fact that the returned html can also contain <link>). * makeResourceLoaderLink now returns an array of html and module states. This allows the consumer of this method to combine the states in 1 larger script tag on top of multiple makeResourceLoaderLink calls (e.g. one state script followed by multiple <script src=load.php>). This isn't to reduce html output, but to make sure we inform mw.loader about modules before the <script src=load.php>. If we were to mix/alternate the state script and load.php requests (which are blocking in html), it is possible for those scripts to request other modules. We need to prevent duplicate loading of modules we already know are going to be requested by the HTML output futher down. * Removed spurious new line. Example of change in HTML output: * The output has been reduced from: - loader.state( site: loading, user: loading, user.groups: loading ) - loader.load( .. ) - <script src="load.php?modules=site .."> - loader.state( user: ready, user.groups: ready ) to: - loader.state( site: loading, user: ready, user.groups: ready ) - loader.load( .. ) - <script src="load.php?modules=site .."> Change-Id: I91754ce5fae3d05b4bfa7372372eba81ee2fc579
2013-11-14 16:54:19 +00:00
ResourceLoaderModule::TYPE_STYLES
);
foreach ( array_intersect( $moduleNames, $separateReq ) as $name ) {
// These require their own dedicated request in order to support "@import"
// syntax, which is incompatible with concatenation. (T147667, T37562)
$chunks[] = $this->makeResourceLoaderLink( $name,
ResourceLoaderModule::TYPE_STYLES
);
}
}
resourceloader: Move queue formatting out of OutputPage HTML formatting of the queue was distributed over several OutputPage methods. Each method demanding a snippet of HTML by calling makeResourceLoaderLink() with a limited amount of information. As such, makeResourceLoaderLink() was unable to provide the client with the proper state information. Centralising it also allows it to better reduce duplication in HTML output and maintain a more accurate state. Problems fixed by centralising: 1. The 'user' module is special (due to per-user 'version' and 'user' params). It is manually requested via script-src. To avoid a separate (and wrong) request from something that requires it, we set state=loading directly. However, because the module is in the bottom, the old HTML formatter could only put state=loading in the bottom also. This sometimes caused a wrong request to be fired for modules=user if something in the top queue triggered a requirement for it. 2. Since a464d1d4 (T87871) we track states of page-style modules, with purpose of allowing dependencies on style modules without risking duplicate loading on pages where the styles are loaded already. This didn't work, because the state information about page-style modules is output near the stylesheet, which is after the script tag with mw.loader.load(). That runs first, and mw.loader would still make a duplicate request before it learns the state. Changes: * Document reasons for style/script tag order in getHeadHtml (per 09537e83). * Pass $type from getModuleStyles() to getAllowedModules(). This wasn't needed before since a duplicate check in makeResourceLoaderLink() verified the origin a second time. * Declare explicit position 'top' on 'user.options' and 'user.tokens' module. Previously, OutputPage hardcoded them in the top. The new formatter doesn't. * Remove getHeadScripts(). * Remove getInlineHeadScripts(). * Remove getExternalHeadScripts(). * Remove buildCssLinks(). * Remove getScriptsForBottomQueue(). * Change where Skin::setupSkinUserCss() is called. This methods lets the skin add modules to the queue. Previously it was called from buildCssLinks(), via headElement(), via prepareQuickTemplate(), via OutputPage::output(). It's now in OutputPage::output() directly (slightly earlier). This is needed because prepareQuickTemplate() calls bottomScripts() before headElement(). And bottomScript() would lazy-initialise the queue and lock it before setupSkinUserCss() is called from headElement(). This makes execution order more predictable instead of being dependent on the arbitrary order of data extraction in prepareQuickTemplate (which varies from one skin to another). * Compute isUserModulePreview() and isKnownEmpty() for the 'user' module early on so. This avoids wrongful loading and fixes problem 1. Effective changes in output: * mw.loader.state() is now before mw.loader.load(). This fixes problem 2. * mw.loader.state() now sets 'user.options' and 'user.tokens' to "loading". * mw.loader.state() now sets 'user' (as "loading" or "ready"). Fixes problem 1. * The <script async src> tag for 'startup' changed position (slightly). Previously it was after all inline scripts and stylesheets. It's still after all inline scripts and after most stylesheets, but before any user styles. Since the queue is now formatted outside OutputPage, it can't inject the meta-ResourceLoaderDynamicStyles tag and user-stylesheet hack in the middle of existing output. This shouldn't have any noticable impact. Bug: T87871 Change-Id: I605b8cd1e1fc009b4662a0edbc54d09dd65ee1df
2016-07-15 14:13:09 +00:00
return self::combineWrappedStrings( array_merge( $chunks, $append ) );
}
/**
* @return array
*/
public function buildCssLinksArray() {
$links = [];
// Add any extension CSS
foreach ( $this->mExtStyles as $url ) {
$this->addStyle( $url );
}
$this->mExtStyles = [];
foreach ( $this->styles as $file => $options ) {
$link = $this->styleLink( $file, $options );
if ( $link ) {
$links[$file] = $link;
}
}
return $links;
}
/**
* Generate \<link\> tags for stylesheets
*
* @param string $style URL to the file
* @param array $options Option, can contain 'condition', 'dir', 'media' keys
* @return string HTML fragment
*/
protected function styleLink( $style, array $options ) {
if ( isset( $options['dir'] ) ) {
if ( $this->getLanguage()->getDir() != $options['dir'] ) {
return '';
}
}
if ( isset( $options['media'] ) ) {
$media = self::transformCssMedia( $options['media'] );
if ( is_null( $media ) ) {
return '';
}
} else {
$media = 'all';
}
if ( substr( $style, 0, 1 ) == '/' ||
substr( $style, 0, 5 ) == 'http:' ||
substr( $style, 0, 6 ) == 'https:' ) {
$url = $style;
} else {
$config = $this->getConfig();
$url = $config->get( 'StylePath' ) . '/' . $style . '?' .
$config->get( 'StyleVersion' );
}
$link = Html::linkedStyle( $url, $media );
if ( isset( $options['condition'] ) ) {
$condition = htmlspecialchars( $options['condition'] );
$link = "<!--[if $condition]>$link<![endif]-->";
}
return $link;
}
/**
* Transform path to web-accessible static resource.
*
* This is used to add a validation hash as query string.
* This aids various behaviors:
*
* - Put long Cache-Control max-age headers on responses for improved
* cache performance.
* - Get the correct version of a file as expected by the current page.
* - Instantly get the updated version of a file after deployment.
*
* Avoid using this for urls included in HTML as otherwise clients may get different
* versions of a resource when navigating the site depending on when the page was cached.
* If changes to the url propagate, this is not a problem (e.g. if the url is in
* an external stylesheet).
*
* @since 1.27
* @param Config $config
* @param string $path Path-absolute URL to file (from document root, must start with "/")
* @return string URL
*/
public static function transformResourcePath( Config $config, $path ) {
global $IP;
$localDir = $IP;
$remotePathPrefix = $config->get( 'ResourceBasePath' );
if ( $remotePathPrefix === '' ) {
// The configured base path is required to be empty string for
// wikis in the domain root
$remotePath = '/';
} else {
$remotePath = $remotePathPrefix;
}
if ( strpos( $path, $remotePath ) !== 0 || substr( $path, 0, 2 ) === '//' ) {
// - Path is outside wgResourceBasePath, ignore.
// - Path is protocol-relative. Fixes T155310. Not supported by RelPath lib.
return $path;
}
// For files in resources, extensions/ or skins/, ResourceBasePath is preferred here.
// For other misc files in $IP, we'll fallback to that as well. There is, however, a fourth
// supported dir/path pair in the configuration (wgUploadDirectory, wgUploadPath)
// which is not expected to be in wgResourceBasePath on CDNs. (T155146)
$uploadPath = $config->get( 'UploadPath' );
if ( strpos( $path, $uploadPath ) === 0 ) {
$localDir = $config->get( 'UploadDirectory' );
$remotePathPrefix = $remotePath = $uploadPath;
}
$path = RelPath\getRelativePath( $path, $remotePath );
return self::transformFilePath( $remotePathPrefix, $localDir, $path );
}
/**
* Utility method for transformResourceFilePath().
*
* Caller is responsible for ensuring the file exists. Emits a PHP warning otherwise.
*
* @since 1.27
* @param string $remotePathPrefix URL path prefix that points to $localPath
* @param string $localPath File directory exposed at $remotePath
* @param string $file Path to target file relative to $localPath
* @return string URL
*/
public static function transformFilePath( $remotePathPrefix, $localPath, $file ) {
$hash = md5_file( "$localPath/$file" );
if ( $hash === false ) {
wfLogWarning( __METHOD__ . ": Failed to hash $localPath/$file" );
$hash = '';
}
return "$remotePathPrefix/$file?" . substr( $hash, 0, 5 );
}
/**
* Transform "media" attribute based on request parameters
*
* @param string $media Current value of the "media" attribute
* @return string Modified value of the "media" attribute, or null to skip
* this stylesheet
*/
public static function transformCssMedia( $media ) {
global $wgRequest;
// https://www.w3.org/TR/css3-mediaqueries/#syntax
$screenMediaQueryRegex = '/^(?:only\s+)?screen\b/i';
// Switch in on-screen display for media testing
$switches = [
'printable' => 'print',
'handheld' => 'handheld',
];
foreach ( $switches as $switch => $targetMedia ) {
if ( $wgRequest->getBool( $switch ) ) {
if ( $media == $targetMedia ) {
$media = '';
} elseif ( preg_match( $screenMediaQueryRegex, $media ) === 1 ) {
/* This regex will not attempt to understand a comma-separated media_query_list
*
* Example supported values for $media:
* 'screen', 'only screen', 'screen and (min-width: 982px)' ),
* Example NOT supported value for $media:
* '3d-glasses, screen, print and resolution > 90dpi'
*
* If it's a print request, we never want any kind of screen stylesheets
* If it's a handheld request (currently the only other choice with a switch),
* we don't want simple 'screen' but we might want screen queries that
* have a max-width or something, so we'll pass all others on and let the
* client do the query.
*/
if ( $targetMedia == 'print' || $media == 'screen' ) {
return null;
}
}
}
}
return $media;
}
/**
* Add a wikitext-formatted message to the output.
* This is equivalent to:
*
* $wgOut->addWikiText( wfMessage( ... )->plain() )
*/
public function addWikiMsg( /*...*/ ) {
$args = func_get_args();
$name = array_shift( $args );
$this->addWikiMsgArray( $name, $args );
}
/**
* Add a wikitext-formatted message to the output.
* Like addWikiMsg() except the parameters are taken as an array
* instead of a variable argument list.
*
* @param string $name
* @param array $args
*/
public function addWikiMsgArray( $name, $args ) {
$this->addHTML( $this->msg( $name, $args )->parseAsBlock() );
}
/**
* This function takes a number of message/argument specifications, wraps them in
* some overall structure, and then parses the result and adds it to the output.
*
* In the $wrap, $1 is replaced with the first message, $2 with the second,
* and so on. The subsequent arguments may be either
* 1) strings, in which case they are message names, or
* 2) arrays, in which case, within each array, the first element is the message
* name, and subsequent elements are the parameters to that message.
*
* Don't use this for messages that are not in the user's interface language.
*
* For example:
*
* $wgOut->wrapWikiMsg( "<div class='error'>\n$1\n</div>", 'some-error' );
*
* Is equivalent to:
*
* $wgOut->addWikiText( "<div class='error'>\n"
* . wfMessage( 'some-error' )->plain() . "\n</div>" );
*
* The newline after the opening div is needed in some wikitext. See T21226.
*
* @param string $wrap
*/
public function wrapWikiMsg( $wrap /*, ...*/ ) {
$msgSpecs = func_get_args();
array_shift( $msgSpecs );
$msgSpecs = array_values( $msgSpecs );
$s = $wrap;
foreach ( $msgSpecs as $n => $spec ) {
if ( is_array( $spec ) ) {
$args = $spec;
$name = array_shift( $args );
if ( isset( $args['options'] ) ) {
unset( $args['options'] );
wfDeprecated(
'Adding "options" to ' . __METHOD__ . ' is no longer supported',
'1.20'
);
}
} else {
$args = [];
$name = $spec;
}
$s = str_replace( '$' . ( $n + 1 ), $this->msg( $name, $args )->plain(), $s );
}
$this->addWikiText( $s );
}
/**
* Whether the output has a table of contents
* @return bool
* @since 1.22
*/
public function isTOCEnabled() {
return $this->mEnableTOC;
}
/**
* Enables/disables section edit links, doesn't override __NOEDITSECTION__
* @param bool $flag
* @since 1.23
*/
public function enableSectionEditLinks( $flag = true ) {
$this->mEnableSectionEditLinks = $flag;
}
/**
* @return bool
* @since 1.23
*/
public function sectionEditLinksEnabled() {
return $this->mEnableSectionEditLinks;
}
/**
* Helper function to setup the PHP implementation of OOUI to use in this request.
*
* @since 1.26
* @param String $skinName The Skin name to determine the correct OOUI theme
* @param String $dir Language direction
*/
public static function setupOOUI( $skinName = 'default', $dir = 'ltr' ) {
$themes = ResourceLoaderOOUIModule::getSkinThemeMap();
$theme = isset( $themes[$skinName] ) ? $themes[$skinName] : $themes['default'];
// For example, 'OOUI\WikimediaUITheme'.
$themeClass = "OOUI\\{$theme}Theme";
OOUI\Theme::setSingleton( new $themeClass() );
OOUI\Element::setDefaultDir( $dir );
}
/**
* Add ResourceLoader module styles for OOUI and set up the PHP implementation of it for use with
* MediaWiki and this OutputPage instance.
*
* @since 1.25
*/
public function enableOOUI() {
self::setupOOUI(
strtolower( $this->getSkin()->getSkinName() ),
$this->getLanguage()->getDir()
);
$this->addModuleStyles( [
'oojs-ui-core.styles',
'oojs-ui.styles.indicators',
'oojs-ui.styles.textures',
'mediawiki.widgets.styles',
'oojs-ui.styles.icons-content',
'oojs-ui.styles.icons-alerts',
'oojs-ui.styles.icons-interactions',
] );
}
/**
* Add Link headers for preloading the wiki's logo.
*
* @since 1.26
*/
protected function addLogoPreloadLinkHeaders() {
$logo = ResourceLoaderSkinModule::getLogo( $this->getConfig() );
$tags = [];
$logosPerDppx = [];
$logos = [];
if ( !is_array( $logo ) ) {
// No media queries required if we only have one variant
$this->addLinkHeader( '<' . $logo . '>;rel=preload;as=image' );
return;
}
foreach ( $logo as $dppx => $src ) {
// Keys are in this format: "1.5x"
$dppx = substr( $dppx, 0, -1 );
$logosPerDppx[$dppx] = $src;
}
// Because PHP can't have floats as array keys
uksort( $logosPerDppx, function ( $a , $b ) {
$a = floatval( $a );
$b = floatval( $b );
if ( $a == $b ) {
return 0;
}
// Sort from smallest to largest (e.g. 1x, 1.5x, 2x)
return ( $a < $b ) ? -1 : 1;
} );
foreach ( $logosPerDppx as $dppx => $src ) {
$logos[] = [ 'dppx' => $dppx, 'src' => $src ];
}
$logosCount = count( $logos );
// Logic must match ResourceLoaderSkinModule:
// - 1x applies to resolution < 1.5dppx
// - 1.5x applies to resolution >= 1.5dppx && < 2dppx
// - 2x applies to resolution >= 2dppx
// Note that min-resolution and max-resolution are both inclusive.
for ( $i = 0; $i < $logosCount; $i++ ) {
if ( $i === 0 ) {
// Smallest dppx
// min-resolution is ">=" (larger than or equal to)
// "not min-resolution" is essentially "<"
$media_query = 'not all and (min-resolution: ' . $logos[ 1 ]['dppx'] . 'dppx)';
} elseif ( $i !== $logosCount - 1 ) {
// In between
// Media query expressions can only apply "not" to the entire expression
// (e.g. can't express ">= 1.5 and not >= 2).
// Workaround: Use <= 1.9999 in place of < 2.
$upper_bound = floatval( $logos[ $i + 1 ]['dppx'] ) - 0.000001;
$media_query = '(min-resolution: ' . $logos[ $i ]['dppx'] .
'dppx) and (max-resolution: ' . $upper_bound . 'dppx)';
} else {
// Largest dppx
$media_query = '(min-resolution: ' . $logos[ $i ]['dppx'] . 'dppx)';
}
$this->addLinkHeader(
'<' . $logos[$i]['src'] . '>;rel=preload;as=image;media=' . $media_query
);
}
}
}