CommentParser.php

Go to the documentation of this file.

<?php
 
namespace MediaWiki\CommentFormatter;
 
use MediaWiki\FileRepo\File\File;
use MediaWiki\FileRepo\RepoGroup;
use MediaWiki\HookContainer\HookContainer;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\Html\Html;
use MediaWiki\Language\Language;
use MediaWiki\Linker\Linker;
use MediaWiki\Linker\LinkRenderer;
use MediaWiki\Linker\LinkTarget;
use MediaWiki\Page\LinkAlwaysKnownLookup;
use MediaWiki\Page\LinkBatch;
use MediaWiki\Page\LinkBatchFactory;
use MediaWiki\Page\LinkCache;
use MediaWiki\Parser\Parser;
use MediaWiki\Parser\Sanitizer;
use MediaWiki\SpecialPage\SpecialPage;
use MediaWiki\Title\MalformedTitleException;
use MediaWiki\Title\NamespaceInfo;
use MediaWiki\Title\Title;
use MediaWiki\Title\TitleParser;
use MediaWiki\Title\TitleValue;
use MediaWiki\WikiMap\WikiMap;
use Wikimedia\HtmlArmor\HtmlArmor;
use Wikimedia\StringUtils\StringUtils;
 
class CommentParser {
    private $linkRenderer;
    private $linkBatchFactory;
    private $repoGroup;
    private $userLang;
    private $contLang;
    private $titleParser;
    private $namespaceInfo;
    private $hookRunner;
    private $linkCache;
    private LinkAlwaysKnownLookup $linkAlwaysKnownLookup;
 
    private $links = [];
    private $linkBatch;
 
    private $fileBatch;
    private $files = [];
 
    private const MAX_ID_SIZE = 7;
    private const MARKER_PREFIX = "\x1B\"'";
 
    public function __construct(
        LinkRenderer $linkRenderer,
        LinkBatchFactory $linkBatchFactory,
        LinkCache $linkCache,
        RepoGroup $repoGroup,
        Language $userLang,
        Language $contLang,
        TitleParser $titleParser,
        NamespaceInfo $namespaceInfo,
        HookContainer $hookContainer,
        LinkAlwaysKnownLookup $linkAlwaysKnownLookup,
    ) {
        $this->linkRenderer = $linkRenderer;
        $this->linkBatchFactory = $linkBatchFactory;
        $this->linkCache = $linkCache;
        $this->repoGroup = $repoGroup;
        $this->userLang = $userLang;
        $this->contLang = $contLang;
        $this->titleParser = $titleParser;
        $this->namespaceInfo = $namespaceInfo;
        $this->hookRunner = new HookRunner( $hookContainer );
        $this->linkAlwaysKnownLookup = $linkAlwaysKnownLookup;
    }
 
    public function preprocess( string $comment, ?LinkTarget $selfLinkTarget = null,
        $samePage = false, $wikiId = false, $enableSectionLinks = true
    ) {
        return $this->preprocessInternal( $comment, false, $selfLinkTarget,
            $samePage, $wikiId, $enableSectionLinks );
    }
 
    public function preprocessUnsafe( $comment, ?LinkTarget $selfLinkTarget = null,
        $samePage = false, $wikiId = false, $enableSectionLinks = true
    ) {
        return $this->preprocessInternal( $comment, true, $selfLinkTarget,
            $samePage, $wikiId, $enableSectionLinks );
    }
 
    public function finalize( $comments ) {
        $this->flushLinkBatches();
        return preg_replace_callback(
            '/' . self::MARKER_PREFIX . '([0-9]{' . self::MAX_ID_SIZE . '})/',
            function ( $m ) {
                $callback = $this->links[(int)$m[1]] ?? null;
                if ( $callback ) {
                    return $callback();
                } else {
                    return '<!-- MISSING -->';
                }
            },
            $comments
        );
    }
 
    private function preprocessInternal( $comment, $unsafe, $selfLinkTarget, $samePage, $wikiId,
        $enableSectionLinks
    ) {
        // Sanitize text a bit
        // \x1b needs to be stripped because it is used for link markers
        $comment = strtr( $comment, "\n\x1b", "  " );
        // Allow HTML entities (for T15815)
        if ( !$unsafe ) {
            $comment = Sanitizer::escapeHtmlAllowEntities( $comment );
        }
        if ( $enableSectionLinks ) {
            $comment = $this->doSectionLinks( $comment, $selfLinkTarget, $samePage, $wikiId );
        }
        return $this->doWikiLinks( $comment, $selfLinkTarget, $samePage, $wikiId );
    }
 
    private function doSectionLinks(
        $comment,
        $selfLinkTarget = null,
        $samePage = false,
        $wikiId = false
    ) {
        $comment = preg_replace_callback(
            // To detect the presence of content before or after the
            // auto-comment, we use capturing groups inside optional zero-width
            // assertions. But older versions of PCRE can not directly make
            // zero-width assertions optional, so wrap them in a non-capturing
            // group.
            '!(?:(?<=(.)))?/\*\s*(.*?)\s*\*/(?:(?=(.)))?!',
            function ( $match ) use ( $selfLinkTarget, $samePage, $wikiId ) {
                $pre = ( $match[1] ?? '' ) !== '';
                $section = ( $match[2] ?? '' );
                $post = ( $match[3] ?? '' ) !== '';
                $comment = null;
 
                $this->hookRunner->onFormatAutocomments(
                    $comment, $pre, $section, $post,
                    Title::castFromLinkTarget( $selfLinkTarget ),
                    $samePage,
                    $wikiId );
                if ( $comment !== null ) {
                    return $comment;
                }
 
                // HTML has already been escaped in preprocessInternal(), so treat this as HTML from this point
                $parsedSection = $section;
 
                if ( $selfLinkTarget ) {
                    $decodedSection = substr( Parser::guessSectionNameFromStrippedText(
                        // Remove links that a user may have manually put in the autosummary
                        // This could be improved by copying as much of Parser::stripSectionName as desired.
                        str_replace( [ '[[:', '[[', ']]' ], '', $section )
                    ), 1 );
                    if ( $decodedSection !== '' || $section === '' ) {
                        if ( $samePage ) {
                            $targetWithSection = new TitleValue( NS_MAIN, '', $decodedSection );
                        } else {
                            $targetWithSection = $selfLinkTarget->createFragmentTarget( $decodedSection );
                        }
                        if ( $section === '' ) {
                            // Special case for edits to the zeroth section (T412472).
                            $linkHtml = wfMessage( 'autocomment-top' )->inLanguage( $this->userLang )->escaped();
                        } else {
                            $linkHtml = $parsedSection;
                        }
                        $parsedSection = $this->makeSectionLink(
                            $targetWithSection,
                            $this->userLang->getArrow() .
                                Html::rawElement( 'bdi', [ 'dir' => $this->userLang->getDir() ], $linkHtml ),
                            $wikiId,
                            $selfLinkTarget
                        );
                    }
                }
                if ( $post ) {
                    # autocomment $postsep written summary (/* section */ summary)
                    $parsedSection .= wfMessage( 'colon-separator' )->inContentLanguage()->escaped();
                }
                if ( $parsedSection ) {
                    $parsedSection = Html::rawElement( 'span', [ 'class' => 'autocomment' ], $parsedSection );
                }
                if ( $pre ) {
                    # written summary $presep autocomment (summary /* section */)
                    $parsedSection = wfMessage( 'autocomment-prefix' )->inContentLanguage()->escaped()
                        . $parsedSection;
                }
 
                // Make sure any brackets (which the user could have input in the edit summary)
                // in the generated autocomment HTML don't trigger additional link processing (T406664).
                return str_replace( [ '[', ']' ], [ '&#91;', '&#93;' ], $parsedSection );
            },
            $comment
        );
        return $comment;
    }
 
    private function makeSectionLink(
        LinkTarget $target, $html, $wikiId, LinkTarget $contextTitle
    ) {
        if ( $wikiId !== null && $wikiId !== false && !$target->isExternal() ) {
            return $this->linkRenderer->makeExternalLink(
                WikiMap::getForeignURL(
                    $wikiId,
                    $target->getNamespace() === 0
                        ? $target->getDBkey()
                        : $this->namespaceInfo->getCanonicalName( $target->getNamespace() ) .
                        ':' . $target->getDBkey(),
                    $target->getFragment()
                ),
                new HtmlArmor( $html ), // Already escaped
                $contextTitle
            );
        } elseif ( !$target->hasFragment() && $target->getText() === '' && $target->getNamespace() === NS_MAIN ) {
            // Special case for edits to the zeroth section (T412472) when linking within the same page.
            // All of the LinkTarget classes treat this as a link to the main page.
            // We want to link to the top of the current page.
            return Html::rawElement( 'a', [
                'href' => '#',
            ], $html );
        }
        return $this->linkRenderer->makePreloadedLink( $target, new HtmlArmor( $html ), '' );
    }
 
    private function doWikiLinks( $comment, $selfLinkTarget = null, $samePage = false, $wikiId = false ) {
        return preg_replace_callback(
            '/
                \[\[
                \s*+ # ignore leading whitespace, the *+ quantifier disallows backtracking
                :? # ignore optional leading colon
                ([^[\]|]+) # 1. link target; page names cannot include [, ] or |
                (?:\|
                    # 2. link text
                    # Stop matching at ]] without relying on backtracking.
                    ((?:]?[^\]])*+)
                )?
                \]\]
                ([^[]*) # 3. link trail (the text up until the next link)
            /x',
            function ( $match ) use ( $selfLinkTarget, $samePage, $wikiId ) {
                $medians = '(?:';
                $medians .= preg_quote(
                    $this->namespaceInfo->getCanonicalName( NS_MEDIA ), '/' );
                $medians .= '|';
                $medians .= preg_quote(
                        $this->contLang->getNsText( NS_MEDIA ),
                        '/'
                    ) . '):';
 
                $comment = $match[0];
 
                // Fix up urlencoded title texts (copied from Parser::replaceInternalLinks)
                if ( str_contains( $match[1], '%' ) ) {
                    $match[1] = strtr(
                        rawurldecode( $match[1] ),
                        [ '<' => '&lt;', '>' => '&gt;' ]
                    );
                }
 
                // Handle link renaming [[foo|text]] will show link as "text"
                if ( $match[2] != "" ) {
                    $text = $match[2];
                } else {
                    $text = $match[1];
                }
                $submatch = [];
                $linkMarker = null;
                if ( preg_match( '/^' . $medians . '(.*)$/i', $match[1], $submatch ) ) {
                    // Media link; trail not supported.
                    $linkRegexp = '/\[\[(.*?)\]\]/';
                    $linkTarget = $this->titleParser->makeTitleValueSafe( NS_FILE, $submatch[1] );
                    if ( $linkTarget ) {
                        $linkMarker = $this->addFileLink( $linkTarget, $text );
                    }
                } else {
                    // Other kind of link
                    // Make sure its target is non-empty
                    if ( isset( $match[1][0] ) && $match[1][0] == ':' ) {
                        $match[1] = substr( $match[1], 1 );
                    }
                    if ( $match[1] !== false && $match[1] !== null && $match[1] !== '' ) {
                        if ( preg_match(
                            $this->contLang->linkTrail(),
                            $match[3],
                            $submatch
                        ) ) {
                            $trail = $submatch[1];
                        } else {
                            $trail = "";
                        }
                        $linkRegexp = '/\[\[(.*?)\]\]' . preg_quote( $trail, '/' ) . '/';
                        [ $inside, $trail ] = Linker::splitTrail( $trail );
 
                        $linkText = $text;
                        $linkTarget = Linker::normalizeSubpageLink( $selfLinkTarget, $match[1], $linkText );
 
                        try {
                            $target = $this->titleParser->parseTitle( $linkTarget );
 
                            if ( $target->getText() == '' && !$target->isExternal()
                                && !$samePage && $selfLinkTarget
                            ) {
                                $target = $selfLinkTarget->createFragmentTarget( $target->getFragment() );
                            }
 
                            // We should deprecate `null` as a valid value for
                            // $selfLinkTarget to ensure that we can use it as
                            // the title context for the external link.
                            // phpcs:ignore MediaWiki.Usage.DeprecatedGlobalVariables.Deprecated$wgTitle
                            global $wgTitle;
                            $linkMarker = $this->addPageLink(
                                $target,
                                $linkText . $inside,
                                $wikiId,
                                $selfLinkTarget ?? $wgTitle ?? SpecialPage::getTitleFor( 'Badtitle' )
                            );
                            $linkMarker .= $trail;
                        } catch ( MalformedTitleException ) {
                            // Fall through
                        }
                    }
                }
                if ( $linkMarker ) {
                    // If the link is still valid, go ahead and replace it in!
                    $comment = preg_replace(
                        // @phan-suppress-next-line PhanPossiblyUndeclaredVariable linkRegexp set when used
                        $linkRegexp,
                        StringUtils::escapeRegexReplacement( $linkMarker ),
                        $comment,
                        1
                    );
                }
 
                return $comment;
            },
            $comment
        );
    }
 
    private function addLinkMarker( $callback ) {
        $nextId = count( $this->links );
        if ( strlen( (string)$nextId ) > self::MAX_ID_SIZE ) {
            throw new \RuntimeException( 'Too many links in comment batch' );
        }
        $this->links[] = $callback;
        return sprintf( self::MARKER_PREFIX . "%0" . self::MAX_ID_SIZE . 'd', $nextId );
    }
 
    private function addPageLink( LinkTarget $target, $text, $wikiId, LinkTarget $contextTitle ) {
        if ( $wikiId !== null && $wikiId !== false && !$target->isExternal() ) {
            // Handle links from a foreign wiki ID
            return $this->linkRenderer->makeExternalLink(
                WikiMap::getForeignURL(
                    $wikiId,
                    $target->getNamespace() === 0
                        ? $target->getDBkey()
                        : $this->namespaceInfo->getCanonicalName( $target->getNamespace() ) .
                        ':' . $target->getDBkey(),
                    $target->getFragment()
                ),
                new HtmlArmor( $text ), // Already escaped
                $contextTitle
            );
        } elseif ( $this->linkCache->getGoodLinkID( $target ) ||
            $this->linkAlwaysKnownLookup->isAlwaysKnown( $target )
        ) {
            // Already known
            return $this->linkRenderer->makeKnownLink( $target, new HtmlArmor( $text ) );
        } elseif ( $this->linkCache->isBadLink( $target ) ) {
            // Already cached as unknown
            return $this->linkRenderer->makeBrokenLink( $target, new HtmlArmor( $text ) );
        }
 
        // Defer page link
        if ( !$this->linkBatch ) {
            $this->linkBatch = $this->linkBatchFactory->newLinkBatch();
            $this->linkBatch->setCaller( __METHOD__ );
        }
        $this->linkBatch->addObj( $target );
        return $this->addLinkMarker( function () use ( $target, $text ) {
            return $this->linkRenderer->makeLink( $target, new HtmlArmor( $text ) );
        } );
    }
 
    private function addFileLink( LinkTarget $target, $html ) {
        $this->fileBatch[] = [
            'title' => $target
        ];
        return $this->addLinkMarker( function () use ( $target, $html ) {
            return Linker::makeMediaLinkFile(
                $target,
                $this->files[$target->getDBkey()] ?? false,
                $html
            );
        } );
    }
 
    private function flushLinkBatches() {
        if ( $this->linkBatch ) {
            $this->linkBatch->execute();
            $this->linkBatch = null;
        }
        if ( $this->fileBatch ) {
            $this->files += $this->repoGroup->findFiles( $this->fileBatch );
            $this->fileBatch = [];
        }
    }
 
}