includes/deferred/SearchUpdate.php

   1 <?php
   2 /**
   3  * Search index updater
   4  *
   5  * See deferred.txt
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @file
  23  * @ingroup Search
  24  */
  25
  26 use MediaWiki\MediaWikiServices;
  27
  28 /**
  29  * Database independant search index updater
  30  *
  31  * @ingroup Search
  32  */
  33 class SearchUpdate implements DeferrableUpdate {
  34         /** @var int Page id being updated */
  35         private $id = 0;
  36
  37         /** @var Title Title we're updating */
  38         private $title;
  39
  40         /** @var Content|bool Content of the page (not text) */
  41         private $content;
  42
  43         /** @var WikiPage **/
  44         private $page;
  45
  46         /**
  47          * Constructor
  48          *
  49          * @param int $id Page id to update
  50          * @param Title|string $title Title of page to update
  51          * @param Content|string|bool $c Content of the page to update. Default: false.
  52          *  If a Content object, text will be gotten from it. String is for back-compat.
  53          *  Passing false tells the backend to just update the title, not the content
  54          */
  55         public function __construct( $id, $title, $c = false ) {
  56                 if ( is_string( $title ) ) {
  57                         $nt = Title::newFromText( $title );
  58                 } else {
  59                         $nt = $title;
  60                 }
  61
  62                 if ( $nt ) {
  63                         $this->id = $id;
  64                         // is_string() check is back-compat for ApprovedRevs
  65                         if ( is_string( $c ) ) {
  66                                 $this->content = new TextContent( $c );
  67                         } else {
  68                                 $this->content = $c ?: false;
  69                         }
  70                         $this->title = $nt;
  71                 } else {
  72                         wfDebug( "SearchUpdate object created with invalid title '$title'\n" );
  73                 }
  74         }
  75
  76         /**
  77          * Perform actual update for the entry
  78          */
  79         public function doUpdate() {
  80                 $config = MediaWikiServices::getInstance()->getSearchEngineConfig();
  81
  82                 if ( $config->getConfig()->get( 'DisableSearchUpdate' ) || !$this->id ) {
  83                         return;
  84                 }
  85
  86                 $seFactory = MediaWikiServices::getInstance()->getSearchEngineFactory();
  87                 foreach ( $config->getSearchTypes() as $type ) {
  88                         $search = $seFactory->create( $type );
  89                         if ( !$search->supports( 'search-update' ) ) {
  90                                 continue;
  91                         }
  92
  93                         $normalTitle = $this->getNormalizedTitle( $search );
  94
  95                         if ( $this->getLatestPage() === null ) {
  96                                 $search->delete( $this->id, $normalTitle );
  97                                 continue;
  98                         } elseif ( $this->content === false ) {
  99                                 $search->updateTitle( $this->id, $normalTitle );
 100                                 continue;
 101                         }
 102
 103                         $text = $search->getTextFromContent( $this->title, $this->content );
 104                         if ( !$search->textAlreadyUpdatedForIndex() ) {
 105                                 $text = $this->updateText( $text, $search );
 106                         }
 107
 108                         # Perform the actual update
 109                         $search->update( $this->id, $normalTitle, $search->normalizeText( $text ) );
 110                 }
 111         }
 112
 113         /**
 114          * Clean text for indexing. Only really suitable for indexing in databases.
 115          * If you're using a real search engine, you'll probably want to override
 116          * this behavior and do something nicer with the original wikitext.
 117          * @param string $text
 118          * @param SearchEngine $se Search engine
 119          * @return string
 120          */
 121         public function updateText( $text, SearchEngine $se = null ) {
 122                 global $wgContLang;
 123
 124                 # Language-specific strip/conversion
 125                 $text = $wgContLang->normalizeForSearch( $text );
 126                 $se = $se ?: MediaWikiServices::getInstance()->newSearchEngine();
 127                 $lc = $se->legalSearchCharsForUpdate() . '&#;';
 128
 129                 $text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/",
 130                         ' ', $wgContLang->lc( " " . $text . " " ) ); # Strip HTML markup
 131                 $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD",
 132                         "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings
 133
 134                 # Strip external URLs
 135                 $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\x80-\\xFF";
 136                 $protos = "http|https|ftp|mailto|news|gopher";
 137                 $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
 138                 $text = preg_replace( $pat, "\\1 \\3", $text );
 139
 140                 $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
 141                 $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
 142                 $text = preg_replace( $p1, "\\1 ", $text );
 143                 $text = preg_replace( $p2, "\\1 \\3 ", $text );
 144
 145                 # Internal image links
 146                 $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
 147                 $text = preg_replace( $pat2, " \\1 \\3", $text );
 148
 149                 $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
 150                         "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
 151
 152                 # Strip all remaining non-search characters
 153                 $text = preg_replace( "/[^{$lc}]+/", " ", $text );
 154
 155                 /**
 156                  * Handle 's, s'
 157                  *
 158                  *   $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text );
 159                  *   $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text );
 160                  *
 161                  * These tail-anchored regexps are insanely slow. The worst case comes
 162                  * when Japanese or Chinese text (ie, no word spacing) is written on
 163                  * a wiki configured for Western UTF-8 mode. The Unicode characters are
 164                  * expanded to hex codes and the "words" are very long paragraph-length
 165                  * monstrosities. On a large page the above regexps may take over 20
 166                  * seconds *each* on a 1GHz-level processor.
 167                  *
 168                  * Following are reversed versions which are consistently fast
 169                  * (about 3 milliseconds on 1GHz-level processor).
 170                  */
 171                 $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
 172                 $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );
 173
 174                 # Strip wiki '' and '''
 175                 $text = preg_replace( "/''[']*/", " ", $text );
 176
 177                 return $text;
 178         }
 179
 180         /**
 181          * Get WikiPage for the SearchUpdate $id using WikiPage::READ_LATEST
 182          * and ensure using the same WikiPage object if there are multiple
 183          * SearchEngine types.
 184          *
 185          * Returns null if a page has been deleted or is not found.
 186          *
 187          * @return WikiPage|null
 188          */
 189         private function getLatestPage() {
 190                 if ( !isset( $this->page ) ) {
 191                         $this->page = WikiPage::newFromID( $this->id, WikiPage::READ_LATEST );
 192                 }
 193
 194                 return $this->page;
 195         }
 196
 197         /**
 198          * Get a normalized string representation of a title suitable for
 199          * including in a search index
 200          *
 201          * @param SearchEngine $search
 202          * @return string A stripped-down title string ready for the search index
 203          */
 204         private function getNormalizedTitle( SearchEngine $search ) {
 205                 global $wgContLang;
 206
 207                 $ns = $this->title->getNamespace();
 208                 $title = $this->title->getText();
 209
 210                 $lc = $search->legalSearchCharsForUpdate() . '&#;';
 211                 $t = $wgContLang->normalizeForSearch( $title );
 212                 $t = preg_replace( "/[^{$lc}]+/", ' ', $t );
 213                 $t = $wgContLang->lc( $t );
 214
 215                 # Handle 's, s'
 216                 $t = preg_replace( "/([{$lc}]+)'s( |$)/", "\\1 \\1's ", $t );
 217                 $t = preg_replace( "/([{$lc}]+)s'( |$)/", "\\1s ", $t );
 218
 219                 $t = preg_replace( "/\\s+/", ' ', $t );
 220
 221                 if ( $ns == NS_FILE ) {
 222                         $t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
 223                 }
 224
 225                 return $search->normalizeText( trim( $t ) );
 226         }
 227 }