Merge "Make interwiki transclusion use the WAN cache"
[lhc/web/wiklou.git] / includes / parser / Parser.php
index bd7d4ac..78265e8 100644 (file)
@@ -22,6 +22,7 @@
  */
 use MediaWiki\Linker\LinkRenderer;
 use MediaWiki\MediaWikiServices;
+use MediaWiki\Special\SpecialPageFactory;
 use Wikimedia\ScopedCallback;
 
 /**
@@ -51,9 +52,6 @@ use Wikimedia\ScopedCallback;
  * - Parser::getPreloadText()
  *     removes <noinclude> sections and <includeonly> tags
  *
- * Globals used:
- *    object: $wgContLang
- *
  * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
  *
  * @par Settings:
@@ -266,13 +264,29 @@ class Parser {
        /** @var MagicWordFactory */
        private $magicWordFactory;
 
+       /** @var Language */
+       private $contLang;
+
+       /** @var ParserFactory */
+       private $factory;
+
+       /** @var SpecialPageFactory */
+       private $specialPageFactory;
+
        /**
-        * @param array $conf
+        * @param array $conf See $wgParserConf documentation
         * @param MagicWordFactory|null $magicWordFactory
-        */
-       public function __construct( $conf = [], MagicWordFactory $magicWordFactory = null ) {
+        * @param Language|null $contLang Content language
+        * @param ParserFactory|null $factory
+        * @param string|null $urlProtocols As returned from wfUrlProtocols()
+        * @param SpecialPageFactory|null $spFactory
+        */
+       public function __construct(
+               array $conf = [], MagicWordFactory $magicWordFactory = null, Language $contLang = null,
+               ParserFactory $factory = null, $urlProtocols = null, SpecialPageFactory $spFactory = null
+       ) {
                $this->mConf = $conf;
-               $this->mUrlProtocols = wfUrlProtocols();
+               $this->mUrlProtocols = $urlProtocols ?? wfUrlProtocols();
                $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
                        self::EXT_LINK_ADDR .
                        self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
@@ -292,10 +306,14 @@ class Parser {
                }
                wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 
-               $this->magicWordFactory = $magicWordFactory;
-               if ( !$magicWordFactory ) {
-                       $this->magicWordFactory = MediaWikiServices::getInstance()->getMagicWordFactory();
-               }
+               $services = MediaWikiServices::getInstance();
+               $this->magicWordFactory = $magicWordFactory ??
+                       $services->getMagicWordFactory();
+
+               $this->contLang = $contLang ?? $services->getContentLanguage();
+
+               $this->factory = $factory ?? $services->getParserFactory();
+               $this->specialPageFactory = $spFactory ?? $services->getSpecialPageFactory();
        }
 
        /**
@@ -968,6 +986,16 @@ class Parser {
                return $this->magicWordFactory;
        }
 
+       /**
+        * Get the content language that this Parser is using
+        *
+        * @since 1.32
+        * @return Language
+        */
+       public function getContentLanguage() {
+               return $this->contLang;
+       }
+
        /**
         * Replaces all occurrences of HTML-style comments and the given tags
         * in the text with a random marker and returns the next text. The output
@@ -2147,8 +2175,7 @@ class Parser {
                if ( $useLinkPrefixExtension ) {
                        # Match the end of a line for a word that's not followed by whitespace,
                        # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
-                       global $wgContLang;
-                       $charset = $wgContLang->linkPrefixCharset();
+                       $charset = $this->contLang->linkPrefixCharset();
                        $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
                }
 
@@ -2509,7 +2536,7 @@ class Parser {
         * @return string
         */
        public function getVariableValue( $index, $frame = false ) {
-               global $wgContLang, $wgSitename, $wgServer, $wgServerName;
+               global $wgSitename, $wgServer, $wgServerName;
                global $wgArticlePath, $wgScriptPath, $wgStylePath;
 
                if ( is_null( $this->mTitle ) ) {
@@ -2703,10 +2730,11 @@ class Parser {
                                $value = $this->getRevisionSize();
                                break;
                        case 'namespace':
-                               $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
+                               $value = str_replace( '_', ' ',
+                                       $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
                                break;
                        case 'namespacee':
-                               $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
+                               $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
                                break;
                        case 'namespacenumber':
                                $value = $this->mTitle->getNamespace();
@@ -2849,15 +2877,13 @@ class Parser {
         * @return string
         */
        private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
-               global $wgContLang;
-
                # Get the timezone-adjusted timestamp to be used for this revision
                $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
                # Possibly set vary-revision if there is not yet an associated revision
                if ( !$this->getRevisionObject() ) {
                        # Get the timezone-adjusted timestamp $mtts seconds in the future
                        $resThen = substr(
-                               $wgContLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
+                               $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
                                $start,
                                $len
                        );
@@ -3225,7 +3251,7 @@ class Parser {
                                        && $this->mOptions->getAllowSpecialInclusion()
                                        && $this->ot['html']
                                ) {
-                                       $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
+                                       $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
                                        // Pass the template arguments as URL parameters.
                                        // "uselang" will have no effect since the Language object
                                        // is forced to the one defined in ParserOptions.
@@ -3251,8 +3277,7 @@ class Parser {
                                                $context->setUser( User::newFromName( '127.0.0.1', false ) );
                                        }
                                        $context->setLanguage( $this->mOptions->getUserLangObj() );
-                                       $ret = SpecialPageFactory::capturePath(
-                                               $title, $context, $this->getLinkRenderer() );
+                                       $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
                                        if ( $ret ) {
                                                $text = $context->getOutput()->getHTML();
                                                $this->mOutput->addOutputPageMetadata( $context->getOutput() );
@@ -3404,14 +3429,12 @@ class Parser {
         * @return array
         */
        public function callParserFunction( $frame, $function, array $args = [] ) {
-               global $wgContLang;
-
                # Case sensitive functions
                if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
                        $function = $this->mFunctionSynonyms[1][$function];
                } else {
                        # Case insensitive functions
-                       $function = $wgContLang->lc( $function );
+                       $function = $this->contLang->lc( $function );
                        if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
                                $function = $this->mFunctionSynonyms[0][$function];
                        } else {
@@ -3676,8 +3699,8 @@ class Parser {
                                        break;
                                }
                        } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
-                               global $wgContLang;
-                               $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
+                               $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
+                                       lcfirst( $title->getText() ) )->inContentLanguage();
                                if ( !$message->exists() ) {
                                        $text = false;
                                        break;
@@ -3760,57 +3783,68 @@ class Parser {
         * Transclude an interwiki link.
         *
         * @param Title $title
-        * @param string $action
+        * @param string $action Usually one of (raw, render)
         *
         * @return string
         */
        public function interwikiTransclude( $title, $action ) {
-               global $wgEnableScaryTranscluding;
+               global $wgEnableScaryTranscluding, $wgTranscludeCacheExpiry;
 
                if ( !$wgEnableScaryTranscluding ) {
                        return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
                }
 
                $url = $title->getFullURL( [ 'action' => $action ] );
-
-               if ( strlen( $url ) > 255 ) {
+               if ( strlen( $url ) > 1024 ) {
                        return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
                }
-               return $this->fetchScaryTemplateMaybeFromCache( $url );
-       }
 
-       /**
-        * @param string $url
-        * @return mixed|string
-        */
-       public function fetchScaryTemplateMaybeFromCache( $url ) {
-               global $wgTranscludeCacheExpiry;
-               $dbr = wfGetDB( DB_REPLICA );
-               $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
-               $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
-                               [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
-               if ( $obj ) {
-                       return $obj->tc_contents;
-               }
-
-               $req = MWHttpRequest::factory( $url, [], __METHOD__ );
-               $status = $req->execute(); // Status object
-               if ( $status->isOK() ) {
-                       $text = $req->getContent();
-               } elseif ( $req->getStatus() != 200 ) {
+               $wikiId = $title->getTransWikiID(); // remote wiki ID or false
+
+               $fname = __METHOD__;
+               $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
+
+               $data = $cache->getWithSetCallback(
+                       $cache->makeGlobalKey(
+                               'interwiki-transclude',
+                               ( $wikiId !== false ) ? $wikiId : 'external',
+                               sha1( $url )
+                       ),
+                       $wgTranscludeCacheExpiry,
+                       function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
+                               $req = MWHttpRequest::factory( $url, [], $fname );
+
+                               $status = $req->execute(); // Status object
+                               if ( !$status->isOK() ) {
+                                       $ttl = $cache::TTL_UNCACHEABLE;
+                               } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
+                                       $ttl = min( $cache::TTL_LAGGED, $ttl );
+                               }
+
+                               return [
+                                       'text' => $status->isOK() ? $req->getContent() : null,
+                                       'code' => $req->getStatus()
+                               ];
+                       },
+                       [
+                               'checkKeys' => ( $wikiId !== false )
+                                       ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
+                                       : [],
+                               'pcGroup' => 'interwiki-transclude:5',
+                               'pcTTL' => $cache::TTL_PROC_LONG
+                       ]
+               );
+
+               if ( is_string( $data['text'] ) ) {
+                       $text = $data['text'];
+               } elseif ( $data['code'] != 200 ) {
                        // Though we failed to fetch the content, this status is useless.
-                       return wfMessage( 'scarytranscludefailed-httpstatus' )
-                               ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
+                       $text = wfMessage( 'scarytranscludefailed-httpstatus' )
+                               ->params( $url, $data['code'] )->inContentLanguage()->text();
                } else {
-                       return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
+                       $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
                }
 
-               $dbw = wfGetDB( DB_MASTER );
-               $dbw->replace( 'transcache', [ 'tc_url' ], [
-                       'tc_url' => $url,
-                       'tc_time' => $dbw->timestamp( time() ),
-                       'tc_contents' => $text
-               ] );
                return $text;
        }
 
@@ -4514,19 +4548,15 @@ class Parser {
         * @return string
         */
        private function pstPass2( $text, $user ) {
-               global $wgContLang;
-
-               # Note: This is the timestamp saved as hardcoded wikitext to
-               # the database, we use $wgContLang here in order to give
-               # everyone the same signature and use the default one rather
-               # than the one selected in each user's preferences.
-               # (see also T14815)
+               # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
+               # $this->contLang here in order to give everyone the same signature and use the default one
+               # rather than the one selected in each user's preferences.  (see also T14815)
                $ts = $this->mOptions->getTimestamp();
                $timestamp = MWTimestamp::getLocalInstance( $ts );
                $ts = $timestamp->format( 'YmdHis' );
                $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
 
-               $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
+               $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
 
                # Variable replacement
                # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
@@ -4877,8 +4907,6 @@ class Parser {
         * @return string|callable The old callback function for this name, if any
         */
        public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
-               global $wgContLang;
-
                $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
                $this->mFunctionHooks[$id] = [ $callback, $flags ];
 
@@ -4894,7 +4922,7 @@ class Parser {
                foreach ( $synonyms as $syn ) {
                        # Case
                        if ( !$sensitive ) {
-                               $syn = $wgContLang->lc( $syn );
+                               $syn = $this->contLang->lc( $syn );
                        }
                        # Add leading hash
                        if ( !( $flags & self::SFH_NO_HASH ) ) {
@@ -4915,6 +4943,7 @@ class Parser {
         * @return array
         */
        public function getFunctionHooks() {
+               $this->firstCallInit();
                return array_keys( $this->mFunctionHooks );
        }
 
@@ -5471,6 +5500,7 @@ class Parser {
         * @return array
         */
        public function getTags() {
+               $this->firstCallInit();
                return array_merge(
                        array_keys( $this->mTransparentTagHooks ),
                        array_keys( $this->mTagHooks ),
@@ -5478,6 +5508,23 @@ class Parser {
                );
        }
 
+       /**
+        * @since 1.32
+        * @return array
+        */
+       public function getFunctionSynonyms() {
+               $this->firstCallInit();
+               return $this->mFunctionSynonyms;
+       }
+
+       /**
+        * @since 1.32
+        * @return string
+        */
+       public function getUrlProtocols() {
+               return $this->mUrlProtocols;
+       }
+
        /**
         * Replace transparent tags in $text with the values given by the callbacks.
         *
@@ -5730,8 +5777,6 @@ class Parser {
         */
        public function getRevisionTimestamp() {
                if ( is_null( $this->mRevisionTimestamp ) ) {
-                       global $wgContLang;
-
                        $revObject = $this->getRevisionObject();
                        $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
 
@@ -5740,8 +5785,7 @@ class Parser {
                        # Since this value will be saved into the parser cache, served
                        # to other users, and potentially even used inside links and such,
                        # it needs to be consistent for all visitors.
-                       $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
-
+                       $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
                }
                return $this->mRevisionTimestamp;
        }
@@ -6196,9 +6240,8 @@ class Parser {
         * @return Parser A parser object that is not parsing anything
         */
        public function getFreshParser() {
-               global $wgParserConf;
                if ( $this->mInParse ) {
-                       return new $wgParserConf['class']( $wgParserConf );
+                       return $this->factory->create();
                } else {
                        return $this;
                }