LinkHolderArray.php

Go to the documentation of this file.

<?php
class LinkHolderArray {
    public $internals = [];
    public $interwikis = [];
    public $size = 0;

    public $parent;
    protected $tempIdOffset;

    public function __construct( $parent ) {
        $this->parent = $parent;
    }

    public function __destruct() {
        foreach ( $this as $name => $value ) {
            unset( $this->$name );
        }
    }

    public function __sleep() {
        foreach ( $this->internals as &$nsLinks ) {
            foreach ( $nsLinks as &$entry ) {
                unset( $entry['title'] );
            }
        }
        unset( $nsLinks );
        unset( $entry );

        foreach ( $this->interwikis as &$entry ) {
            unset( $entry['title'] );
        }
        unset( $entry );

        return [ 'internals', 'interwikis', 'size' ];
    }

    public function __wakeup() {
        foreach ( $this->internals as &$nsLinks ) {
            foreach ( $nsLinks as &$entry ) {
                $entry['title'] = Title::newFromText( $entry['pdbk'] );
            }
        }
        unset( $nsLinks );
        unset( $entry );

        foreach ( $this->interwikis as &$entry ) {
            $entry['title'] = Title::newFromText( $entry['pdbk'] );
        }
        unset( $entry );
    }

    public function merge( $other ) {
        foreach ( $other->internals as $ns => $entries ) {
            $this->size += count( $entries );
            if ( !isset( $this->internals[$ns] ) ) {
                $this->internals[$ns] = $entries;
            } else {
                $this->internals[$ns] += $entries;
            }
        }
        $this->interwikis += $other->interwikis;
    }

    public function mergeForeign( $other, $texts ) {
        $this->tempIdOffset = $idOffset = $this->parent->nextLinkID();
        $maxId = 0;

        # Renumber internal links
        foreach ( $other->internals as $ns => $nsLinks ) {
            foreach ( $nsLinks as $key => $entry ) {
                $newKey = $idOffset + $key;
                $this->internals[$ns][$newKey] = $entry;
                $maxId = $newKey > $maxId ? $newKey : $maxId;
            }
        }
        $texts = preg_replace_callback( '/(<!--LINK \d+:)(\d+)(-->)/',
            [ $this, 'mergeForeignCallback' ], $texts );

        # Renumber interwiki links
        foreach ( $other->interwikis as $key => $entry ) {
            $newKey = $idOffset + $key;
            $this->interwikis[$newKey] = $entry;
            $maxId = $newKey > $maxId ? $newKey : $maxId;
        }
        $texts = preg_replace_callback( '/(<!--IWLINK )(\d+)(-->)/',
            [ $this, 'mergeForeignCallback' ], $texts );

        # Set the parent link ID to be beyond the highest used ID
        $this->parent->setLinkID( $maxId + 1 );
        $this->tempIdOffset = null;
        return $texts;
    }

    protected function mergeForeignCallback( $m ) {
        return $m[1] . ( $m[2] + $this->tempIdOffset ) . $m[3];
    }

    public function getSubArray( $text ) {
        $sub = new LinkHolderArray( $this->parent );

        # Internal links
        $pos = 0;
        while ( $pos < strlen( $text ) ) {
            if ( !preg_match( '/<!--LINK (\d+):(\d+)-->/',
                $text, $m, PREG_OFFSET_CAPTURE, $pos )
            ) {
                break;
            }
            $ns = $m[1][0];
            $key = $m[2][0];
            $sub->internals[$ns][$key] = $this->internals[$ns][$key];
            $pos = $m[0][1] + strlen( $m[0][0] );
        }

        # Interwiki links
        $pos = 0;
        while ( $pos < strlen( $text ) ) {
            if ( !preg_match( '/<!--IWLINK (\d+)-->/', $text, $m, PREG_OFFSET_CAPTURE, $pos ) ) {
                break;
            }
            $key = $m[1][0];
            $sub->interwikis[$key] = $this->interwikis[$key];
            $pos = $m[0][1] + strlen( $m[0][0] );
        }
        return $sub;
    }

    public function isBig() {
        global $wgLinkHolderBatchSize;
        return $this->size > $wgLinkHolderBatchSize;
    }

    public function clear() {
        $this->internals = [];
        $this->interwikis = [];
        $this->size = 0;
    }

    public function makeHolder( $nt, $text = '', $query = [], $trail = '', $prefix = '' ) {
        if ( !is_object( $nt ) ) {
            # Fail gracefully
            $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
        } else {
            # Separate the link trail from the rest of the link
            list( $inside, $trail ) = Linker::splitTrail( $trail );

            $entry = [
                'title' => $nt,
                'text' => $prefix . $text . $inside,
                'pdbk' => $nt->getPrefixedDBkey(),
            ];
            if ( $query !== [] ) {
                $entry['query'] = $query;
            }

            if ( $nt->isExternal() ) {
                // Use a globally unique ID to keep the objects mergable
                $key = $this->parent->nextLinkID();
                $this->interwikis[$key] = $entry;
                $retVal = "<!--IWLINK $key-->{$trail}";
            } else {
                $key = $this->parent->nextLinkID();
                $ns = $nt->getNamespace();
                $this->internals[$ns][$key] = $entry;
                $retVal = "<!--LINK $ns:$key-->{$trail}";
            }
            $this->size++;
        }
        return $retVal;
    }

    public function replace( &$text ) {

        $this->replaceInternal( $text );
        $this->replaceInterwiki( $text );

    }

    protected function replaceInternal( &$text ) {
        if ( !$this->internals ) {
            return;
        }

        global $wgContLang, $wgContentHandlerUseDB, $wgPageLanguageUseDB;

        $colours = [];
        $linkCache = LinkCache::singleton();
        $output = $this->parent->getOutput();

        $dbr = wfGetDB( DB_SLAVE );
        $threshold = $this->parent->getOptions()->getStubThreshold();

        # Sort by namespace
        ksort( $this->internals );

        $linkcolour_ids = [];

        # Generate query
        $queries = [];
        foreach ( $this->internals as $ns => $entries ) {
            foreach ( $entries as $entry ) {
                $title = $entry['title'];
                $pdbk = $entry['pdbk'];

                # Skip invalid entries.
                # Result will be ugly, but prevents crash.
                if ( is_null( $title ) ) {
                    continue;
                }

                # Check if it's a static known link, e.g. interwiki
                if ( $title->isAlwaysKnown() ) {
                    $colours[$pdbk] = '';
                } elseif ( $ns == NS_SPECIAL ) {
                    $colours[$pdbk] = 'new';
                } else {
                    $id = $linkCache->getGoodLinkID( $pdbk );
                    if ( $id != 0 ) {
                        $colours[$pdbk] = Linker::getLinkColour( $title, $threshold );
                        $output->addLink( $title, $id );
                        $linkcolour_ids[$id] = $pdbk;
                    } elseif ( $linkCache->isBadLink( $pdbk ) ) {
                        $colours[$pdbk] = 'new';
                    } else {
                        # Not in the link cache, add it to the query
                        $queries[$ns][] = $title->getDBkey();
                    }
                }
            }
        }
        if ( $queries ) {
            $where = [];
            foreach ( $queries as $ns => $pages ) {
                $where[] = $dbr->makeList(
                    [
                        'page_namespace' => $ns,
                        'page_title' => array_unique( $pages ),
                    ],
                    LIST_AND
                );
            }

            $fields = [ 'page_id', 'page_namespace', 'page_title',
                'page_is_redirect', 'page_len', 'page_latest' ];

            if ( $wgContentHandlerUseDB ) {
                $fields[] = 'page_content_model';
            }
            if ( $wgPageLanguageUseDB ) {
                $fields[] = 'page_lang';
            }

            $res = $dbr->select(
                'page',
                $fields,
                $dbr->makeList( $where, LIST_OR ),
                __METHOD__
            );

            # Fetch data and form into an associative array
            # non-existent = broken
            foreach ( $res as $s ) {
                $title = Title::makeTitle( $s->page_namespace, $s->page_title );
                $pdbk = $title->getPrefixedDBkey();
                $linkCache->addGoodLinkObjFromRow( $title, $s );
                $output->addLink( $title, $s->page_id );
                # @todo FIXME: Convoluted data flow
                # The redirect status and length is passed to getLinkColour via the LinkCache
                # Use formal parameters instead
                $colours[$pdbk] = Linker::getLinkColour( $title, $threshold );
                // add id to the extension todolist
                $linkcolour_ids[$s->page_id] = $pdbk;
            }
            unset( $res );
        }
        if ( count( $linkcolour_ids ) ) {
            // pass an array of page_ids to an extension
            Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
        }

        # Do a second query for different language variants of links and categories
        if ( $wgContLang->hasVariants() ) {
            $this->doVariants( $colours );
        }

        # Construct search and replace arrays
        $replacePairs = [];
        foreach ( $this->internals as $ns => $entries ) {
            foreach ( $entries as $index => $entry ) {
                $pdbk = $entry['pdbk'];
                $title = $entry['title'];
                $query = isset( $entry['query'] ) ? $entry['query'] : [];
                $key = "$ns:$index";
                $searchkey = "<!--LINK $key-->";
                $displayText = $entry['text'];
                if ( isset( $entry['selflink'] ) ) {
                    $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
                    continue;
                }
                if ( $displayText === '' ) {
                    $displayText = null;
                }
                if ( !isset( $colours[$pdbk] ) ) {
                    $colours[$pdbk] = 'new';
                }
                $attribs = [];
                if ( $colours[$pdbk] == 'new' ) {
                    $linkCache->addBadLinkObj( $title );
                    $output->addLink( $title, 0 );
                    $type = [ 'broken' ];
                } else {
                    if ( $colours[$pdbk] != '' ) {
                        $attribs['class'] = $colours[$pdbk];
                    }
                    $type = [ 'known', 'noclasses' ];
                }
                $replacePairs[$searchkey] = Linker::link( $title, $displayText,
                        $attribs, $query, $type );
            }
        }
        $replacer = new HashtableReplacer( $replacePairs, 1 );

        # Do the thing
        $text = preg_replace_callback(
            '/(<!--LINK .*?-->)/',
            $replacer->cb(),
            $text
        );

    }

    protected function replaceInterwiki( &$text ) {
        if ( empty( $this->interwikis ) ) {
            return;
        }

        # Make interwiki link HTML
        $output = $this->parent->getOutput();
        $replacePairs = [];
        $options = [
            'stubThreshold' => $this->parent->getOptions()->getStubThreshold(),
        ];
        foreach ( $this->interwikis as $key => $link ) {
            $replacePairs[$key] = Linker::link( $link['title'], $link['text'], [], [], $options );
            $output->addInterwikiLink( $link['title'] );
        }
        $replacer = new HashtableReplacer( $replacePairs, 1 );

        $text = preg_replace_callback(
            '/<!--IWLINK (.*?)-->/',
            $replacer->cb(),
            $text );
    }

    protected function doVariants( &$colours ) {
        global $wgContLang, $wgContentHandlerUseDB, $wgPageLanguageUseDB;
        $linkBatch = new LinkBatch();
        $variantMap = []; // maps $pdbkey_Variant => $keys (of link holders)
        $output = $this->parent->getOutput();
        $linkCache = LinkCache::singleton();
        $threshold = $this->parent->getOptions()->getStubThreshold();
        $titlesToBeConverted = '';
        $titlesAttrs = [];

        // Concatenate titles to a single string, thus we only need auto convert the
        // single string to all variants. This would improve parser's performance
        // significantly.
        foreach ( $this->internals as $ns => $entries ) {
            if ( $ns == NS_SPECIAL ) {
                continue;
            }
            foreach ( $entries as $index => $entry ) {
                $pdbk = $entry['pdbk'];
                // we only deal with new links (in its first query)
                if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
                    $titlesAttrs[] = [ $index, $entry['title'] ];
                    // separate titles with \0 because it would never appears
                    // in a valid title
                    $titlesToBeConverted .= $entry['title']->getText() . "\0";
                }
            }
        }

        // Now do the conversion and explode string to text of titles
        $titlesAllVariants = $wgContLang->autoConvertToAllVariants( rtrim( $titlesToBeConverted, "\0" ) );
        $allVariantsName = array_keys( $titlesAllVariants );
        foreach ( $titlesAllVariants as &$titlesVariant ) {
            $titlesVariant = explode( "\0", $titlesVariant );
        }

        // Then add variants of links to link batch
        $parentTitle = $this->parent->getTitle();
        foreach ( $titlesAttrs as $i => $attrs ) {
            list( $index, $title ) = $attrs;
            $ns = $title->getNamespace();
            $text = $title->getText();

            foreach ( $allVariantsName as $variantName ) {
                $textVariant = $titlesAllVariants[$variantName][$i];
                if ( $textVariant === $text ) {
                    continue;
                }

                $variantTitle = Title::makeTitle( $ns, $textVariant );
                if ( is_null( $variantTitle ) ) {
                    continue;
                }

                // Self-link checking for mixed/different variant titles. At this point, we
                // already know the exact title does not exist, so the link cannot be to a
                // variant of the current title that exists as a separate page.
                if ( $variantTitle->equals( $parentTitle ) && !$title->hasFragment() ) {
                    $this->internals[$ns][$index]['selflink'] = true;
                    continue 2;
                }

                $linkBatch->addObj( $variantTitle );
                $variantMap[$variantTitle->getPrefixedDBkey()][] = "$ns:$index";
            }
        }

        // process categories, check if a category exists in some variant
        $categoryMap = []; // maps $category_variant => $category (dbkeys)
        $varCategories = []; // category replacements oldDBkey => newDBkey
        foreach ( $output->getCategoryLinks() as $category ) {
            $categoryTitle = Title::makeTitleSafe( NS_CATEGORY, $category );
            $linkBatch->addObj( $categoryTitle );
            $variants = $wgContLang->autoConvertToAllVariants( $category );
            foreach ( $variants as $variant ) {
                if ( $variant !== $category ) {
                    $variantTitle = Title::makeTitleSafe( NS_CATEGORY, $variant );
                    if ( is_null( $variantTitle ) ) {
                        continue;
                    }
                    $linkBatch->addObj( $variantTitle );
                    $categoryMap[$variant] = [ $category, $categoryTitle ];
                }
            }
        }

        if ( !$linkBatch->isEmpty() ) {
            // construct query
            $dbr = wfGetDB( DB_SLAVE );
            $fields = [ 'page_id', 'page_namespace', 'page_title',
                'page_is_redirect', 'page_len', 'page_latest' ];

            if ( $wgContentHandlerUseDB ) {
                $fields[] = 'page_content_model';
            }
            if ( $wgPageLanguageUseDB ) {
                $fields[] = 'page_lang';
            }

            $varRes = $dbr->select( 'page',
                $fields,
                $linkBatch->constructSet( 'page', $dbr ),
                __METHOD__
            );

            $linkcolour_ids = [];

            // for each found variants, figure out link holders and replace
            foreach ( $varRes as $s ) {
                $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
                $varPdbk = $variantTitle->getPrefixedDBkey();
                $vardbk = $variantTitle->getDBkey();

                $holderKeys = [];
                if ( isset( $variantMap[$varPdbk] ) ) {
                    $holderKeys = $variantMap[$varPdbk];
                    $linkCache->addGoodLinkObjFromRow( $variantTitle, $s );
                    $output->addLink( $variantTitle, $s->page_id );
                }

                // loop over link holders
                foreach ( $holderKeys as $key ) {
                    list( $ns, $index ) = explode( ':', $key, 2 );
                    $entry =& $this->internals[$ns][$index];
                    $pdbk = $entry['pdbk'];

                    if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
                        // found link in some of the variants, replace the link holder data
                        $entry['title'] = $variantTitle;
                        $entry['pdbk'] = $varPdbk;

                        // set pdbk and colour
                        # @todo FIXME: Convoluted data flow
                        # The redirect status and length is passed to getLinkColour via the LinkCache
                        # Use formal parameters instead
                        $colours[$varPdbk] = Linker::getLinkColour( $variantTitle, $threshold );
                        $linkcolour_ids[$s->page_id] = $pdbk;
                    }
                }

                // check if the object is a variant of a category
                if ( isset( $categoryMap[$vardbk] ) ) {
                    list( $oldkey, $oldtitle ) = $categoryMap[$vardbk];
                    if ( !isset( $varCategories[$oldkey] ) && !$oldtitle->exists() ) {
                        $varCategories[$oldkey] = $vardbk;
                    }
                }
            }
            Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );

            // rebuild the categories in original order (if there are replacements)
            if ( count( $varCategories ) > 0 ) {
                $newCats = [];
                $originalCats = $output->getCategories();
                foreach ( $originalCats as $cat => $sortkey ) {
                    // make the replacement
                    if ( array_key_exists( $cat, $varCategories ) ) {
                        $newCats[$varCategories[$cat]] = $sortkey;
                    } else {
                        $newCats[$cat] = $sortkey;
                    }
                }
                $output->setCategoryLinks( $newCats );
            }
        }
    }

    public function replaceText( $text ) {

        $text = preg_replace_callback(
            '/<!--(LINK|IWLINK) (.*?)-->/',
            [ &$this, 'replaceTextCallback' ],
            $text );

        return $text;
    }

    public function replaceTextCallback( $matches ) {
        $type = $matches[1];
        $key = $matches[2];
        if ( $type == 'LINK' ) {
            list( $ns, $index ) = explode( ':', $key, 2 );
            if ( isset( $this->internals[$ns][$index]['text'] ) ) {
                return $this->internals[$ns][$index]['text'];
            }
        } elseif ( $type == 'IWLINK' ) {
            if ( isset( $this->interwikis[$key]['text'] ) ) {
                return $this->interwikis[$key]['text'];
            }
        }
        return $matches[0];
    }
}