CategoryFinder.php

Go to the documentation of this file.

<?php
use Wikimedia\Rdbms\IDatabase;
 
class CategoryFinder {
    protected $articles = [];
 
    protected $deadend = [];
 
    protected $parents = [];
 
    protected $next = [];
 
    protected $maxdepth = -1;
 
    protected $targets = [];
 
    protected $name2id = [];
 
    protected $mode;
 
    protected $dbr;
 
    public function seed( $articleIds, $categories, $mode = 'AND', $maxdepth = -1 ) {
        $this->articles = $articleIds;
        $this->next = $articleIds;
        $this->mode = $mode;
        $this->maxdepth = $maxdepth;
 
        # Set the list of target categories; convert them to DBKEY form first
        $this->targets = [];
        foreach ( $categories as $c ) {
            $ct = Title::makeTitleSafe( NS_CATEGORY, $c );
            if ( $ct ) {
                $c = $ct->getDBkey();
                $this->targets[$c] = $c;
            }
        }
    }
 
    public function run() {
        $this->dbr = wfGetDB( DB_REPLICA );
 
        $i = 0;
        $dig = true;
        while ( count( $this->next ) && $dig ) {
            $this->scanNextLayer();
 
            // Is there any depth limit?
            if ( $this->maxdepth !== -1 ) {
                $dig = $i < $this->maxdepth;
                $i++;
            }
        }
 
        # Now check if this applies to the individual articles
        $ret = [];
 
        foreach ( $this->articles as $article ) {
            $conds = $this->targets;
            if ( $this->check( $article, $conds ) ) {
                # Matches the conditions
                $ret[] = $article;
            }
        }
        return $ret;
    }
 
    public function getParents() {
        return $this->parents;
    }
 
    private function check( $id, &$conds, $path = [] ) {
        // Check for loops and stop!
        if ( in_array( $id, $path ) ) {
            return false;
        }
 
        $path[] = $id;
 
        # Shortcut (runtime paranoia): No conditions=all matched
        if ( count( $conds ) == 0 ) {
            return true;
        }
 
        if ( !isset( $this->parents[$id] ) ) {
            return false;
        }
 
        # iterate through the parents
        foreach ( $this->parents[$id] as $p ) {
            $pname = $p->cl_to;
 
            # Is this a condition?
            if ( isset( $conds[$pname] ) ) {
                # This key is in the category list!
                if ( $this->mode == 'OR' ) {
                    # One found, that's enough!
                    $conds = [];
                    return true;
                } else {
                    # Assuming "AND" as default
                    unset( $conds[$pname] );
                    if ( count( $conds ) == 0 ) {
                        # All conditions met, done
                        return true;
                    }
                }
            }
 
            # Not done yet, try sub-parents
            if ( !isset( $this->name2id[$pname] ) ) {
                # No sub-parent
                continue;
            }
            $done = $this->check( $this->name2id[$pname], $conds, $path );
            if ( $done || count( $conds ) == 0 ) {
                # Subparents have done it!
                return true;
            }
        }
        return false;
    }
 
    private function scanNextLayer() {
        # Find all parents of the article currently in $this->next
        $layer = [];
        $res = $this->dbr->select(
            /* FROM   */ 'categorylinks',
            /* SELECT */ [ 'cl_to', 'cl_from' ],
            /* WHERE  */ [ 'cl_from' => $this->next ],
            __METHOD__ . '-1'
        );
        foreach ( $res as $row ) {
            $k = $row->cl_to;
 
            # Update parent tree
            if ( !isset( $this->parents[$row->cl_from] ) ) {
                $this->parents[$row->cl_from] = [];
            }
            $this->parents[$row->cl_from][$k] = $row;
 
            # Ignore those we already have
            if ( in_array( $k, $this->deadend ) ) {
                continue;
            }
 
            if ( isset( $this->name2id[$k] ) ) {
                continue;
            }
 
            # Hey, new category!
            $layer[$k] = $k;
        }
 
        $this->next = [];
 
        # Find the IDs of all category pages in $layer, if they exist
        if ( count( $layer ) > 0 ) {
            $res = $this->dbr->select(
                /* FROM   */ 'page',
                /* SELECT */ [ 'page_id', 'page_title' ],
                /* WHERE  */ [ 'page_namespace' => NS_CATEGORY, 'page_title' => $layer ],
                __METHOD__ . '-2'
            );
            foreach ( $res as $row ) {
                $id = $row->page_id;
                $name = $row->page_title;
                $this->name2id[$name] = $id;
                $this->next[] = $id;
                unset( $layer[$name] );
            }
        }
 
        # Mark dead ends
        foreach ( $layer as $v ) {
            $this->deadend[$v] = $v;
        }
    }
}