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
  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 __construct() {
  39         }
  40
  41         /**
  42          * Initializes the instance. Do this prior to calling run().
  43          * @param $article_ids Array of article IDs
  44          * @param $categories FIXME
  45          * @param $mode String: FIXME, default 'AND'.
  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                 foreach ( $this->articles as $article ) {
  77                         $conds = $this->targets;
  78                         if ( $this->check( $article, $conds ) ) {
  79                                 # Matches the conditions
  80                                 $ret[] = $article;
  81                         }
  82                 }
  83                 return $ret;
  84         }
  85
  86         /**
  87          * This functions recurses through the parent representation, trying to match the conditions
  88          * @param $id The article/category to check
  89          * @param $conds The array of categories to match
  90          * @param $path used to check for recursion loops
  91          * @return bool Does this match the conditions?
  92          */
  93         function check( $id , &$conds, $path = array() ) {
  94                 // Check for loops and stop!
  95                 if( in_array( $id, $path ) )
  96                         return false;
  97                 $path[] = $id;
  98
  99                 # Shortcut (runtime paranoia): No contitions=all matched
 100                 if ( count( $conds ) == 0 ) return true;
 101
 102                 if ( !isset( $this->parents[$id] ) ) return false;
 103
 104                 # iterate through the parents
 105                 foreach ( $this->parents[$id] as $p ) {
 106                         $pname = $p->cl_to ;
 107
 108                         # Is this a condition?
 109                         if ( isset( $conds[$pname] ) ) {
 110                                 # This key is in the category list!
 111                                 if ( $this->mode == "OR" ) {
 112                                         # One found, that's enough!
 113                                         $conds = array();
 114                                         return true;
 115                                 } else {
 116                                         # Assuming "AND" as default
 117                                         unset( $conds[$pname] ) ;
 118                                         if ( count( $conds ) == 0 ) {
 119                                                 # All conditions met, done
 120                                                 return true;
 121                                         }
 122                                 }
 123                         }
 124
 125                         # Not done yet, try sub-parents
 126                         if ( !isset( $this->name2id[$pname] ) ) {
 127                                 # No sub-parent
 128                                 continue ;
 129                         }
 130                         $done = $this->check( $this->name2id[$pname], $conds,$path );
 131                         if ( $done || count( $conds ) == 0 ) {
 132                                 # Subparents have done it!
 133                                 return true;
 134                         }
 135                 }
 136                 return false;
 137         }
 138
 139         /**
 140          * Scans a "parent layer" of the articles/categories in $this->next
 141          */
 142         function scan_next_layer() {
 143                 # Find all parents of the article currently in $this->next
 144                 $layer = array();
 145                 $res = $this->dbr->select(
 146                                 /* FROM   */ 'categorylinks',
 147                                 /* SELECT */ '*',
 148                                 /* WHERE  */ array( 'cl_from' => $this->next ),
 149                                 __METHOD__ . "-1"
 150                 );
 151                 while ( $o = $this->dbr->fetchObject( $res ) ) {
 152                         $k = $o->cl_to ;
 153
 154                         # Update parent tree
 155                         if ( !isset( $this->parents[$o->cl_from] ) ) {
 156                                 $this->parents[$o->cl_from] = array();
 157                         }
 158                         $this->parents[$o->cl_from][$k] = $o;
 159
 160                         # Ignore those we already have
 161                         if ( in_array ( $k , $this->deadend ) ) continue;
 162                         if ( isset ( $this->name2id[$k] ) ) continue;
 163
 164                         # Hey, new category!
 165                         $layer[$k] = $k;
 166                 }
 167
 168                 $this->next = array();
 169
 170                 # Find the IDs of all category pages in $layer, if they exist
 171                 if ( count ( $layer ) > 0 ) {
 172                         $res = $this->dbr->select(
 173                                         /* FROM   */ 'page',
 174                                         /* SELECT */ array( 'page_id', 'page_title' ),
 175                                         /* WHERE  */ array( 'page_namespace' => NS_CATEGORY , 'page_title' => $layer ),
 176                                         __METHOD__ . "-2"
 177                         );
 178                         while ( $o = $this->dbr->fetchObject( $res ) ) {
 179                                 $id = $o->page_id;
 180                                 $name = $o->page_title;
 181                                 $this->name2id[$name] = $id;
 182                                 $this->next[] = $id;
 183                                 unset( $layer[$name] );
 184                         }
 185                 }
 186
 187                 # Mark dead ends
 188                 foreach ( $layer as $v ) {
 189                         $this->deadend[$v] = $v;
 190                 }
 191         }
 192
 193 } # END OF CLASS "Categoryfinder"