includes/Categoryfinder.php

   1 <?php
   2 /**
   3  * The "Categoryfinder" class takes a list of articles, creates an internal
   4  * representation of all their parent categories (as well as parents of
   5  * parents etc.). From this representation, it determines which of these
   6  * articles are in one or all of a given subset of categories.
   7  *
   8  * Example use :
   9  * <code>
  10  *      # Determines whether the article with the page_id 12345 is in both
  11  *      # "Category 1" and "Category 2" or their subcategories, respectively
  12  *
  13  *      $cf = new Categoryfinder;
  14  *      $cf->seed(
  15  *              array( 12345 ),
  16  *              array( 'Category 1', 'Category 2' ),
  17  *              'AND'
  18  *      );
  19  *      $a = $cf->run();
  20  *      print implode( ',' , $a );
  21  * </code>
  22  *
  23  */
  24 class Categoryfinder {
  25         var $articles = array(); # The original article IDs passed to the seed function
  26         var $deadend = array(); # Array of DBKEY category names for categories that don't have a page
  27         var $parents = array(); # Array of [ID => array()]
  28         var $next = array(); # Array of article/category IDs
  29         var $targets = array(); # Array of DBKEY category names
  30         var $name2id = array();
  31         var $mode; # "AND" or "OR"
  32         var $dbr; # Read-DB slave
  33
  34         /**
  35          * Constructor (currently empty).
  36          */
  37         function __construct() {
  38         }
  39
  40         /**
  41          * Initializes the instance. Do this prior to calling run().
  42          * @param $article_ids Array of article IDs
  43          * @param $categories FIXME
  44          * @param $mode String: FIXME, default 'AND'.
  45          * @todo FIXME: $categories/$mode
  46          */
  47         function seed( $article_ids, $categories, $mode = 'AND' ) {
  48                 $this->articles = $article_ids;
  49                 $this->next = $article_ids;
  50                 $this->mode = $mode;
  51
  52                 # Set the list of target categories; convert them to DBKEY form first
  53                 $this->targets = array();
  54                 foreach ( $categories as $c ) {
  55                         $ct = Title::makeTitleSafe( NS_CATEGORY, $c );
  56                         if ( $ct ) {
  57                                 $c = $ct->getDBkey();
  58                                 $this->targets[$c] = $c;
  59                         }
  60                 }
  61         }
  62
  63         /**
  64          * Iterates through the parent tree starting with the seed values,
  65          * then checks the articles if they match the conditions
  66          * @return array of page_ids (those given to seed() that match the conditions)
  67          */
  68         function run() {
  69                 $this->dbr = wfGetDB( DB_SLAVE );
  70                 while ( count( $this->next ) > 0 ) {
  71                         $this->scan_next_layer();
  72                 }
  73
  74                 # Now check if this applies to the individual articles
  75                 $ret = array();
  76
  77                 foreach ( $this->articles as $article ) {
  78                         $conds = $this->targets;
  79                         if ( $this->check( $article, $conds ) ) {
  80                                 # Matches the conditions
  81                                 $ret[] = $article;
  82                         }
  83                 }
  84                 return $ret;
  85         }
  86
  87         /**
  88          * This functions recurses through the parent representation, trying to match the conditions
  89          * @param $id int The article/category to check
  90          * @param $conds array The array of categories to match
  91          * @param $path array used to check for recursion loops
  92          * @return bool Does this match the conditions?
  93          */
  94         function check( $id, &$conds, $path = array() ) {
  95                 // Check for loops and stop!
  96                 if ( in_array( $id, $path ) ) {
  97                         return false;
  98                 }
  99
 100                 $path[] = $id;
 101
 102                 # Shortcut (runtime paranoia): No contitions=all matched
 103                 if ( count( $conds ) == 0 ) {
 104                         return true;
 105                 }
 106
 107                 if ( !isset( $this->parents[$id] ) ) {
 108                         return false;
 109                 }
 110
 111                 # iterate through the parents
 112                 foreach ( $this->parents[$id] as $p ) {
 113                         $pname = $p->cl_to ;
 114
 115                         # Is this a condition?
 116                         if ( isset( $conds[$pname] ) ) {
 117                                 # This key is in the category list!
 118                                 if ( $this->mode == 'OR' ) {
 119                                         # One found, that's enough!
 120                                         $conds = array();
 121                                         return true;
 122                                 } else {
 123                                         # Assuming "AND" as default
 124                                         unset( $conds[$pname] );
 125                                         if ( count( $conds ) == 0 ) {
 126                                                 # All conditions met, done
 127                                                 return true;
 128                                         }
 129                                 }
 130                         }
 131
 132                         # Not done yet, try sub-parents
 133                         if ( !isset( $this->name2id[$pname] ) ) {
 134                                 # No sub-parent
 135                                 continue;
 136                         }
 137                         $done = $this->check( $this->name2id[$pname], $conds, $path );
 138                         if ( $done || count( $conds ) == 0 ) {
 139                                 # Subparents have done it!
 140                                 return true;
 141                         }
 142                 }
 143                 return false;
 144         }
 145
 146         /**
 147          * Scans a "parent layer" of the articles/categories in $this->next
 148          */
 149         function scan_next_layer() {
 150                 # Find all parents of the article currently in $this->next
 151                 $layer = array();
 152                 $res = $this->dbr->select(
 153                         /* FROM   */ 'categorylinks',
 154                         /* SELECT */ '*',
 155                         /* WHERE  */ array( 'cl_from' => $this->next ),
 156                         __METHOD__ . '-1'
 157                 );
 158                 foreach ( $res as $o ) {
 159                         $k = $o->cl_to;
 160
 161                         # Update parent tree
 162                         if ( !isset( $this->parents[$o->cl_from] ) ) {
 163                                 $this->parents[$o->cl_from] = array();
 164                         }
 165                         $this->parents[$o->cl_from][$k] = $o;
 166
 167                         # Ignore those we already have
 168                         if ( in_array( $k, $this->deadend ) ) {
 169                                 continue;
 170                         }
 171
 172                         if ( isset( $this->name2id[$k] ) ) {
 173                                 continue;
 174                         }
 175
 176                         # Hey, new category!
 177                         $layer[$k] = $k;
 178                 }
 179
 180                 $this->next = array();
 181
 182                 # Find the IDs of all category pages in $layer, if they exist
 183                 if ( count( $layer ) > 0 ) {
 184                         $res = $this->dbr->select(
 185                                 /* FROM   */ 'page',
 186                                 /* SELECT */ array( 'page_id', 'page_title' ),
 187                                 /* WHERE  */ array( 'page_namespace' => NS_CATEGORY , 'page_title' => $layer ),
 188                                 __METHOD__ . '-2'
 189                         );
 190                         foreach ( $res as $o ) {
 191                                 $id = $o->page_id;
 192                                 $name = $o->page_title;
 193                                 $this->name2id[$name] = $id;
 194                                 $this->next[] = $id;
 195                                 unset( $layer[$name] );
 196                         }
 197                 }
 198
 199                 # Mark dead ends
 200                 foreach ( $layer as $v ) {
 201                         $this->deadend[$v] = $v;
 202                 }
 203         }
 204
 205 }