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