wikimedia / mediawiki-extensions-Lingo

src/LingoParser.php

<?php

/**
 * File holding the Lingo\LingoParser class.
 *
 * This file is part of the MediaWiki extension Lingo.
 *
 * @copyright 2011 - 2018, Stephan Gambke
 * @license GPL-2.0-or-later
 *
 * The Lingo extension 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.
 *
 * The Lingo extension 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, see <http://www.gnu.org/licenses/>.
 *
 * @author Stephan Gambke
 *
 * @file
 * @ingroup Lingo
 */
namespace Lingo;

use DOMDocument;
use DOMXPath;
use ObjectCache;
use Parser;
use StubObject;
use Title;
use Wikimedia\AtEase\AtEase;

/**
 * This class parses the given text and enriches it with definitions for defined
 * terms.
 *
 * Contains a static function to initiate the parsing.
 *
 * @ingroup Lingo
 */
class LingoParser {

	const WORD_VALUE = 0;
	const WORD_OFFSET = 1;

	private $mLingoTree = null;

	/**
	 * @var Backend
	 */
	private $mLingoBackend = null;
	private static $parserSingleton = null;

	// Api params passed in from ApiMakeParserOptions Hook
	private $mApiParams = null;

	// The RegEx to split a chunk of text into words
	public $regex = null;

	/**
	 * Lingo\LingoParser constructor.
	 * @param MessageLog|null &$messages
	 */
	public function __construct( MessageLog &$messages = null ) {
		// The RegEx to split a chunk of text into words
		// Words are: placeholders for stripped items, sequences of letters and numbers, single characters that are neither letter nor number
		$this->regex = '/' . preg_quote( Parser::MARKER_PREFIX, '/' ) . '.*?' . preg_quote( Parser::MARKER_SUFFIX, '/' ) . '|[\p{L}\p{N}]+|[^\p{L}\p{N}]/u';
	}

	/**
	 * @param Parser $mwParser
	 *
	 * @return bool
	 */
	public function parse( $mwParser ) {
		if ( $this->shouldParse( $mwParser ) ) {
			$this->realParse( $mwParser );
		}

		return true;
	}

	/**
	 * @return LingoParser
	 * @since 2.0.1
	 */
	public static function getInstance() {
		if ( !self::$parserSingleton ) {
			self::$parserSingleton = new LingoParser();

		}

		return self::$parserSingleton;
	}

	/**
	 * @return string
	 */
	private function getCacheKey() {
		// FIXME: If Lingo ever stores the glossary tree per user, then the cache key also needs to include the user id (see T163608)
		return ObjectCache::getLocalClusterInstance()->makeKey( 'ext', 'lingo', 'lingotree', Tree::TREE_VERSION, get_class( $this->getBackend() ) );
	}

	/**
	 * @return Backend the backend used by the parser
	 * @throws \MWException
	 */
	public function getBackend() {
		if ( $this->mLingoBackend === null ) {
			throw new \MWException( 'No Lingo backend available!' );
		}

		return $this->mLingoBackend;
	}

	/**
	 * Returns the list of terms in the glossary
	 *
	 * @return array an array mapping terms (keys) to descriptions (values)
	 */
	public function getLingoArray() {
		return $this->getLingoTree()->getTermList();
	}

	/**
	 * Returns the list of terms in the glossary as a Lingo\Tree
	 *
	 * @return Tree a Lingo\Tree mapping terms (keys) to descriptions (values)
	 */
	public function getLingoTree() {
		// build glossary array only once per request
		if ( !$this->mLingoTree ) {

			// use cache if enabled
			if ( $this->getBackend()->useCache() ) {

				// Try cache first
				global $wgexLingoCacheType;
				$cache = ( $wgexLingoCacheType !== null ) ? wfGetCache( $wgexLingoCacheType ) : wfGetMainCache();
				$cachekey = $this->getCacheKey();
				$cachedLingoTree = $cache->get( $cachekey );

				// cache hit?
				if ( $cachedLingoTree !== false && $cachedLingoTree !== null ) {

					wfDebug( "Cache hit: Got lingo tree from cache.\n" );
					$this->mLingoTree = &$cachedLingoTree;

					wfDebug( "Re-cached lingo tree.\n" );
				} else {

					wfDebug( "Cache miss: Lingo tree not found in cache.\n" );
					$this->mLingoTree =& $this->buildLingo();
					wfDebug( "Cached lingo tree.\n" );
				}

				// Keep for one month
				// Limiting the cache validity will allow to purge stale cache
				// entries inserted by older versions after one month
				$cache->set( $cachekey, $this->mLingoTree, 60 * 60 * 24 * 30 );

			} else {
				wfDebug( "Caching of lingo tree disabled.\n" );
				$this->mLingoTree =& $this->buildLingo();
			}

		}

		return $this->mLingoTree;
	}

	/**
	 * @return Tree
	 */
	protected function &buildLingo() {
		$lingoTree = new Tree();
		$backend = &$this->mLingoBackend;

		// assemble the result array
		while ( $elementData = $backend->next() ) {
			$lingoTree->addTerm( $elementData[ Element::ELEMENT_TERM ], $elementData );
		}

		return $lingoTree;
	}

	/**
	 * Parses the given text and enriches applicable terms
	 *
	 * This method currently only recognizes terms consisting of max one word
	 *
	 * @param Parser &$parser
	 *
	 * @return bool
	 */
	protected function realParse( &$parser ) {
		// Parse text identical to options used in includes/api/ApiParse.php
		$params = $this->mApiParams;
		$text = is_null( $params ) ? $parser->getOutput()->getText() : $parser->getOutput()->getText( [
			'allowTOC' => !$params['disabletoc'],
			'enableSectionEditLinks' => !$params['disableeditsection'],
			'wrapperDivClass' => $params['wrapoutputclass'],
			'deduplicateStyles' => !$params['disablestylededuplication'],
		] );

		if ( $text === null || $text === '' ) {
			return true;
		}

		// Get array of terms
		$glossary = $this->getLingoTree();

		if ( $glossary == null ) {

Operator == prohibited; use === instead

			return true;
		}

		// Parse HTML from page

		// TODO: Remove call to \MediaWiki\suppressWarnings() for MW 1.34+.
		// \Wikimedia\AtEase\AtEase::suppressWarnings() is available from MW 1.34.
		if ( method_exists( AtEase::class, 'suppressWarnings' ) ) {
			\Wikimedia\AtEase\AtEase::suppressWarnings();
		} else {
			\MediaWiki\suppressWarnings();
		}

		$doc = new DOMDocument( '1.0', 'utf-8' );
		$doc->loadHTML( '<html><head><meta http-equiv="content-type" content="charset=utf-8"/></head><body>' . $text . '</body></html>' );

		// TODO: Remove call to \MediaWiki\restoreWarnings() for MW 1.34+.
		// \Wikimedia\AtEase\AtEase::restoreWarnings() is available from MW 1.34.
		if ( method_exists( AtEase::class, 'suppressWarnings' ) ) {
			\Wikimedia\AtEase\AtEase::restoreWarnings();
		} else {
			\MediaWiki\restoreWarnings();
		}

		// Find all text in HTML.
		$xpath = new DOMXPath( $doc );
		$textElements = $xpath->query(
			"//*[not(ancestor-or-self::*[@class='noglossary'] or ancestor-or-self::a)][text()!=' ']/text()"
		);

		// Iterate all HTML text matches
		$numberOfTextElements = $textElements->length;

		$definitions = [];

		for ( $textElementIndex = 0; $textElementIndex < $numberOfTextElements; $textElementIndex++ ) {
			$textElement = $textElements->item( $textElementIndex );

			if ( strlen( $textElement->nodeValue ) < $glossary->getMinTermLength() ) {
				continue;
			}

			$matches = [];
			preg_match_all(
				$this->regex,
				$textElement->nodeValue,
				$matches,
				PREG_OFFSET_CAPTURE | PREG_PATTERN_ORDER
			);

			if ( count( $matches ) === 0 || count( $matches[ 0 ] ) === 0 ) {
				continue;
			}

			$wordDescriptors = &$matches[ 0 ]; // See __construct() for definition of "word"
			$numberOfWordDescriptors = count( $wordDescriptors );

			$parentNode = &$textElement->parentNode;

			$wordDescriptorIndex = 0;
			$changedElem = false;

			while ( $wordDescriptorIndex < $numberOfWordDescriptors ) {

				/** @var \Lingo\Element $definition */
				list( $skippedWords, $usedWords, $definition ) =
					$glossary->findNextTerm( $wordDescriptors, $wordDescriptorIndex, $numberOfWordDescriptors );

				if ( $usedWords > 0 ) { // found a term

					if ( $skippedWords > 0 ) { // skipped some text, insert it as is

						$start = $wordDescriptors[ $wordDescriptorIndex ][ self::WORD_OFFSET ];
						$length = $wordDescriptors[ $wordDescriptorIndex + $skippedWords ][ self::WORD_OFFSET ] - $start;

						$parentNode->insertBefore(
							$doc->createTextNode(
								substr( $textElement->nodeValue, $start, $length )
							),
							$textElement
						);
					}

					$parentNode->insertBefore( $definition->getFormattedTerm( $doc ), $textElement );

					$definitions[ $definition->getId() ] = $definition->getFormattedDefinitions();

					$changedElem = true;

				} else { // did not find any term, just use the rest of the text

					// If we found no term now and no term before, there was no
					// term in the whole element. Might as well not change the
					// element at all.

					// Only change element if found term before
					if ( $changedElem === true ) {

						$start = $wordDescriptors[ $wordDescriptorIndex ][ self::WORD_OFFSET ];

						$parentNode->insertBefore(
							$doc->createTextNode(
								substr( $textElement->nodeValue, $start )
							),
							$textElement
						);

					}

					// In principle superfluous, the loop would run out anyway. Might save a bit of time.
					break;
				}

				$wordDescriptorIndex += $usedWords + $skippedWords;
			}

			if ( $changedElem ) {
				$parentNode->removeChild( $textElement );
			}
		}

		if ( count( $definitions ) > 0 ) {

			$this->loadModules( $parser );

			// U - Ungreedy, D - dollar matches only end of string, s - dot matches newlines
			$text = preg_replace( '%(^.*<body>)|(</body>.*$)%UDs', '', $doc->saveHTML() );
			$text .= $parser->recursiveTagParseFully( implode( $definitions ) );

			$parser->getOutput()->setText( $text );
		}

		return true;
	}

	/**
	 * @param Parser &$parser
	 */
	protected function loadModules( &$parser ) {
		global $wgOut;

		$parserOutput = $parser->getOutput();

		// load scripts
		$parserOutput->addModules( 'ext.Lingo' );

		if ( !$wgOut->isArticle() ) {
			$wgOut->addModules( 'ext.Lingo' );
		}
	}

	/**
	 * Purges the lingo tree from the cache.
	 *
	 * @deprecated 2.0.2
	 */
	public static function purgeCache() {
		self::getInstance()->purgeGlossaryFromCache();
	}

	/**
	 * Purges the lingo tree from the cache.
	 *
	 * @since 2.0.2
	 */
	public function purgeGlossaryFromCache() {
		global $wgexLingoCacheType;
		$cache = ( $wgexLingoCacheType !== null ) ? wfGetCache( $wgexLingoCacheType ) : wfGetMainCache();
		$cache->delete( $this->getCacheKey() );
	}

	/**
	 * @since 2.0.1
	 * @param Backend $backend
	 */
	public function setBackend( Backend $backend ) {
		$this->mLingoBackend = $backend;
		$backend->setLingoParser( $this );
	}

	/**
	 * Set parser options from API
	 *
	 * @param array $params
	 */
	public function setApiParams( array $params ) {
		$this->mApiParams = $params;
	}

	/**
	 * @param Parser &$parser
	 * @return bool
	 */
	protected function shouldParse( &$parser ) {
		global $wgexLingoUseNamespaces;

		if ( !( $parser instanceof Parser || $parser instanceof StubObject ) ) {
			return false;
		}

		if ( isset( $parser->mDoubleUnderscores[ 'noglossary' ] ) ) { // __NOGLOSSARY__ found in wikitext
			return false;
		}

		$title = $parser->getTitle();

		if ( !( $title instanceof Title ) ) {
			return false;
		}

		$namespace = $title->getNamespace();

		if ( isset( $wgexLingoUseNamespaces[ $namespace ] ) && $wgexLingoUseNamespaces[ $namespace ] === false ) {
			return false;
		};

		return true;
	}
}