includes/specials/SpecialLinkSearch.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup SpecialPage
   5  *
   6  * @author Brion Vibber
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License 2.0 or later
   8  */
   9
  10 /**
  11  * Special:LinkSearch to search the external-links table.
  12  * @ingroup SpecialPage
  13  */
  14
  15 function wfSpecialLinkSearch( $par ) {
  16
  17         list( $limit, $offset ) = wfCheckLimits();
  18         global $wgOut, $wgRequest, $wgUrlProtocols, $wgMiserMode;
  19         $target = $GLOBALS['wgRequest']->getVal( 'target', $par );
  20         $namespace = $GLOBALS['wgRequest']->getIntorNull( 'namespace', null );
  21
  22         $protocols_list[] = '';
  23         foreach( $wgUrlProtocols as $prot ) {
  24                 $protocols_list[] = $prot;
  25         }
  26
  27         $target2 = $target;
  28         $protocol = '';
  29         $pr_sl = strpos($target2, '//' );
  30         $pr_cl = strpos($target2, ':' );
  31         if ( $pr_sl ) {
  32                 // For protocols with '//'
  33                 $protocol = substr( $target2, 0 , $pr_sl+2 );
  34                 $target2 = substr( $target2, $pr_sl+2 );
  35         } elseif ( !$pr_sl && $pr_cl ) {
  36                 // For protocols without '//' like 'mailto:'
  37                 $protocol = substr( $target2, 0 , $pr_cl+1 );
  38                 $target2 = substr( $target2, $pr_cl+1 );
  39         } elseif ( $protocol == '' && $target2 != '' ) {
  40                 // default
  41                 $protocol = 'http://';
  42         }
  43         if ( !in_array( $protocol, $protocols_list ) ) {
  44                 // unsupported protocol, show original search request
  45                 $target2 = $target;
  46                 $protocol = '';
  47         }
  48
  49         $self = Title::makeTitle( NS_SPECIAL, 'Linksearch' );
  50
  51         $wgOut->addWikiText( wfMsg( 'linksearch-text', '<nowiki>' . implode( ', ',  $wgUrlProtocols) . '</nowiki>' ) );
  52         $s =    Xml::openElement( 'form', array( 'id' => 'mw-linksearch-form', 'method' => 'get', 'action' => $GLOBALS['wgScript'] ) ) .
  53                 Xml::hidden( 'title', $self->getPrefixedDbKey() ) .
  54                 '<fieldset>' .
  55                 Xml::element( 'legend', array(), wfMsg( 'linksearch' ) ) .
  56                 Xml::inputLabel( wfMsg( 'linksearch-pat' ), 'target', 'target', 50, $target ) . ' ';
  57         if ( !$wgMiserMode ) {
  58                 $s .= Xml::label( wfMsg( 'linksearch-ns' ), 'namespace' ) . ' ' .
  59                         XML::namespaceSelector( $namespace, '' );
  60         }
  61         $s .=   Xml::submitButton( wfMsg( 'linksearch-ok' ) ) .
  62                 '</fieldset>' .
  63                 Xml::closeElement( 'form' );
  64         $wgOut->addHtml( $s );
  65
  66         if( $target != '' ) {
  67                 $searcher = new LinkSearchPage( $target2, $namespace, $protocol );
  68                 $searcher->doQuery( $offset, $limit );
  69         }
  70 }
  71
  72 class LinkSearchPage extends QueryPage {
  73
  74         function __construct( $query, $ns, $prot ) {
  75                 $this->mQuery = $query;
  76                 $this->mNs = $ns;
  77                 $this->mProt = $prot;
  78         }
  79
  80         function getName() {
  81                 return 'LinkSearch';
  82         }
  83
  84         /**
  85          * Disable RSS/Atom feeds
  86          */
  87         function isSyndicated() {
  88                 return false;
  89         }
  90
  91         /**
  92          * Return an appropriately formatted LIKE query and the clause
  93          */
  94         static function mungeQuery( $query , $prot ) {
  95                 $field = 'el_index';
  96                 $rv = LinkFilter::makeLike( $query , $prot );
  97                 if ($rv === false) {
  98                         //makeLike doesn't handle wildcard in IP, so we'll have to munge here.
  99                         if (preg_match('/^(:?[0-9]{1,3}\.)+\*\s*$|^(:?[0-9]{1,3}\.){3}[0-9]{1,3}:[0-9]*\*\s*$/', $query)) {
 100                                 $rv = $prot . rtrim($query, " \t*") . '%';
 101                                 $field = 'el_to';
 102                         }
 103                 }
 104                 return array( $rv, $field );
 105         }
 106
 107         function linkParameters() {
 108                 global $wgMiserMode;
 109                 $params = array();
 110                 $params['target'] = $this->mProt . $this->mQuery;
 111                 if( isset( $this->mNs ) && !$wgMiserMode ) {
 112                         $params['namespace'] = $this->mNs;
 113                 }
 114                 return $params;
 115         }
 116
 117         function getSQL() {
 118                 global $wgMiserMode;
 119                 $dbr = wfGetDB( DB_SLAVE );
 120                 $page = $dbr->tableName( 'page' );
 121                 $externallinks = $dbr->tableName( 'externallinks' );
 122
 123                 /* strip everything past first wildcard, so that index-based-only lookup would be done */
 124                 list( $munged, $clause ) = self::mungeQuery( $this->mQuery, $this->mProt );
 125                 $stripped = substr($munged,0,strpos($munged,'%')+1);
 126                 $encSearch = $dbr->addQuotes( $stripped );
 127
 128                 $encSQL = '';
 129                 if ( isset ($this->mNs) && !$wgMiserMode )
 130                         $encSQL = 'AND page_namespace=' . $dbr->addQuotes( $this->mNs );
 131
 132                 $use_index = $dbr->useIndexClause( $clause );
 133                 return
 134                         "SELECT
 135                                 page_namespace AS namespace,
 136                                 page_title AS title,
 137                                 el_index AS value,
 138                                 el_to AS url
 139                         FROM
 140                                 $page,
 141                                 $externallinks $use_index
 142                         WHERE
 143                                 page_id=el_from
 144                                 AND $clause LIKE $encSearch
 145                                 $encSQL";
 146         }
 147
 148         function formatResult( $skin, $result ) {
 149                 $title = Title::makeTitle( $result->namespace, $result->title );
 150                 $url = $result->url;
 151                 $pageLink = $skin->makeKnownLinkObj( $title );
 152                 $urlLink = $skin->makeExternalLink( $url, $url );
 153
 154                 return wfMsgHtml( 'linksearch-line', $urlLink, $pageLink );
 155         }
 156
 157         /**
 158          * Override to check query validity.
 159          */
 160         function doQuery( $offset, $limit, $shownavigation=true ) {
 161                 global $wgOut;
 162                 list( $this->mMungedQuery, $clause ) = LinkSearchPage::mungeQuery( $this->mQuery, $this->mProt );
 163                 if( $this->mMungedQuery === false ) {
 164                         $wgOut->addWikiText( wfMsg( 'linksearch-error' ) );
 165                 } else {
 166                         // For debugging
 167                         // Generates invalid xhtml with patterns that contain --
 168                         //$wgOut->addHtml( "\n<!-- " . htmlspecialchars( $this->mMungedQuery ) . " -->\n" );
 169                         parent::doQuery( $offset, $limit, $shownavigation );
 170                 }
 171         }
 172
 173         /**
 174          * Override to squash the ORDER BY.
 175          * We do a truncated index search, so the optimizer won't trust
 176          * it as good enough for optimizing sort. The implicit ordering
 177          * from the scan will usually do well enough for our needs.
 178          */
 179         function getOrder() {
 180                 return '';
 181         }
 182 }