includes/Categoryfinder.php

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