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.
11 * # Determines whether the article with the page_id 12345 is in both
12 * # "Category 1" and "Category 2" or their subcategories, respectively
14 * $cf = new Categoryfinder ;
17 * array ( "Category 1","Category 2" ) ,
21 * print implode ( "," , $a ) ;
25 class Categoryfinder {
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
37 * Constructor (currently empty).
39 function __construct() {
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'.
48 function seed ( $article_ids , $categories , $mode = "AND" ) {
49 $this->articles = $article_ids ;
50 $this->next = $article_ids ;
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 ;
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)
68 $this->dbr = wfGetDB( DB_SLAVE );
69 while ( count ( $this->next ) > 0 ) {
70 $this->scan_next_layer () ;
73 # Now check if this applies to the individual articles
75 foreach ( $this->articles AS $article ) {
76 $conds = $this->targets ;
77 if ( $this->check ( $article , $conds ) ) {
78 # Matches the conditions
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 * @return bool Does this match the conditions?
91 function check ( $id , &$conds ) {
92 # Shortcut (runtime paranoia): No contitions=all matched
93 if ( count ( $conds ) == 0 ) return true ;
95 if ( !isset ( $this->parents[$id] ) ) return false ;
97 # iterate through the parents
98 foreach ( $this->parents[$id] AS $p ) {
101 # Is this a condition?
102 if ( isset ( $conds[$pname] ) ) {
103 # This key is in the category list!
104 if ( $this->mode == "OR" ) {
105 # One found, that's enough!
109 # Assuming "AND" as default
110 unset ( $conds[$pname] ) ;
111 if ( count ( $conds ) == 0 ) {
112 # All conditions met, done
118 # Not done yet, try sub-parents
119 if ( !isset ( $this->name2id[$pname] ) ) {
123 $done = $this->check ( $this->name2id[$pname] , $conds ) ;
124 if ( $done OR count ( $conds ) == 0 ) {
125 # Subparents have done it!
133 * Scans a "parent layer" of the articles/categories in $this->next
135 function scan_next_layer () {
136 $fname = "Categoryfinder::scan_next_layer" ;
138 # Find all parents of the article currently in $this->next
140 $res = $this->dbr->select(
141 /* FROM */ 'categorylinks',
143 /* WHERE */ array( 'cl_from' => $this->next ),
146 while ( $o = $this->dbr->fetchObject( $res ) ) {
150 if ( !isset ( $this->parents[$o->cl_from] ) ) {
151 $this->parents[$o->cl_from] = array () ;
153 $this->parents[$o->cl_from][$k] = $o ;
155 # Ignore those we already have
156 if ( in_array ( $k , $this->deadend ) ) continue ;
157 if ( isset ( $this->name2id[$k] ) ) continue ;
162 $this->dbr->freeResult( $res ) ;
164 $this->next = array() ;
166 # Find the IDs of all category pages in $layer, if they exist
167 if ( count ( $layer ) > 0 ) {
168 $res = $this->dbr->select(
170 /* SELECT */ 'page_id,page_title',
171 /* WHERE */ array( 'page_namespace' => NS_CATEGORY , 'page_title' => $layer ),
174 while ( $o = $this->dbr->fetchObject( $res ) ) {
176 $name = $o->page_title ;
177 $this->name2id[$name] = $id ;
178 $this->next[] = $id ;
179 unset ( $layer[$name] ) ;
181 $this->dbr->freeResult( $res ) ;
185 foreach ( $layer AS $v ) {
186 $this->deadend[$v] = $v ;
190 } # END OF CLASS "Categoryfinder"