includes/Categoryfinder.php

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