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          */
  46         function seed( $article_ids, $categories, $mode = 'AND' ) {
  47                 $this->articles = $article_ids;
  48                 $this->next = $article_ids;
  49                 $this->mode = $mode;
  50
  51                 # Set the list of target categories; convert them to DBKEY form first
  52                 $this->targets = array();
  53                 foreach ( $categories as $c ) {
  54                         $ct = Title::makeTitleSafe( NS_CATEGORY, $c );
  55                         if ( $ct ) {
  56                                 $c = $ct->getDBkey();
  57                                 $this->targets[$c] = $c;
  58                         }
  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
  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                 }
  98
  99                 $path[] = $id;
 100
 101                 # Shortcut (runtime paranoia): No contitions=all matched
 102                 if ( count( $conds ) == 0 ) {
 103                         return true;
 104                 }
 105
 106                 if ( !isset( $this->parents[$id] ) ) {
 107                         return false;
 108                 }
 109
 110                 # iterate through the parents
 111                 foreach ( $this->parents[$id] as $p ) {
 112                         $pname = $p->cl_to ;
 113
 114                         # Is this a condition?
 115                         if ( isset( $conds[$pname] ) ) {
 116                                 # This key is in the category list!
 117                                 if ( $this->mode == 'OR' ) {
 118                                         # One found, that's enough!
 119                                         $conds = array();
 120                                         return true;
 121                                 } else {
 122                                         # Assuming "AND" as default
 123                                         unset( $conds[$pname] );
 124                                         if ( count( $conds ) == 0 ) {
 125                                                 # All conditions met, done
 126                                                 return true;
 127                                         }
 128                                 }
 129                         }
 130
 131                         # Not done yet, try sub-parents
 132                         if ( !isset( $this->name2id[$pname] ) ) {
 133                                 # No sub-parent
 134                                 continue;
 135                         }
 136                         $done = $this->check( $this->name2id[$pname], $conds, $path );
 137                         if ( $done || count( $conds ) == 0 ) {
 138                                 # Subparents have done it!
 139                                 return true;
 140                         }
 141                 }
 142                 return false;
 143         }
 144
 145         /**
 146          * Scans a "parent layer" of the articles/categories in $this->next
 147          */
 148         function scan_next_layer() {
 149                 # Find all parents of the article currently in $this->next
 150                 $layer = array();
 151                 $res = $this->dbr->select(
 152                         /* FROM   */ 'categorylinks',
 153                         /* SELECT */ '*',
 154                         /* WHERE  */ array( 'cl_from' => $this->next ),
 155                         __METHOD__ . '-1'
 156                 );
 157                 while ( $o = $this->dbr->fetchObject( $res ) ) {
 158                         $k = $o->cl_to;
 159
 160                         # Update parent tree
 161                         if ( !isset( $this->parents[$o->cl_from] ) ) {
 162                                 $this->parents[$o->cl_from] = array();
 163                         }
 164                         $this->parents[$o->cl_from][$k] = $o;
 165
 166                         # Ignore those we already have
 167                         if ( in_array( $k, $this->deadend ) ) {
 168                                 continue;
 169                         }
 170
 171                         if ( isset( $this->name2id[$k] ) ) {
 172                                 continue;
 173                         }
 174
 175                         # Hey, new category!
 176                         $layer[$k] = $k;
 177                 }
 178
 179                 $this->next = array();
 180
 181                 # Find the IDs of all category pages in $layer, if they exist
 182                 if ( count( $layer ) > 0 ) {
 183                         $res = $this->dbr->select(
 184                                 /* FROM   */ 'page',
 185                                 /* SELECT */ array( 'page_id', 'page_title' ),
 186                                 /* WHERE  */ array( 'page_namespace' => NS_CATEGORY , 'page_title' => $layer ),
 187                                 __METHOD__ . '-2'
 188                         );
 189                         while ( $o = $this->dbr->fetchObject( $res ) ) {
 190                                 $id = $o->page_id;
 191                                 $name = $o->page_title;
 192                                 $this->name2id[$name] = $id;
 193                                 $this->next[] = $id;
 194                                 unset( $layer[$name] );
 195                         }
 196                 }
 197
 198                 # Mark dead ends
 199                 foreach ( $layer as $v ) {
 200                         $this->deadend[$v] = $v;
 201                 }
 202         }
 203
 204 }