MediaWiki  master
SearchUpdate.php
Go to the documentation of this file.
1 <?php
32 
38 class SearchUpdate implements DeferrableUpdate {
40  private $id = 0;
41 
43  private $page;
44 
46  private $content;
47 
49  private $latestPage = null;
50 
56  public function __construct( $id, $page, ?Content $c = null ) {
57  $this->page = $page;
58  $this->id = $id;
59  $this->content = $c;
60  }
61 
65  public function doUpdate() {
66  $services = MediaWikiServices::getInstance();
67  $config = $services->getSearchEngineConfig();
68 
69  if ( $config->getConfig()->get( MainConfigNames::DisableSearchUpdate ) || !$this->id ) {
70  LoggerFactory::getInstance( "search" )
71  ->debug( "Skipping update: search updates disabled by config" );
72  return;
73  }
74 
75  $seFactory = $services->getSearchEngineFactory();
76  foreach ( $config->getSearchTypes() as $type ) {
77  $search = $seFactory->create( $type );
78  if ( !$search->supports( 'search-update' ) ) {
79  continue;
80  }
81 
82  $normalTitle = $this->getNormalizedTitle( $search );
83 
84  if ( $this->getLatestPage() === null ) {
85  $search->delete( $this->id, $normalTitle );
86  continue;
87  } elseif ( $this->content === null ) {
88  $search->updateTitle( $this->id, $normalTitle );
89  continue;
90  }
91 
92  $text = $this->content !== null ? $this->content->getTextForSearchIndex() : '';
93  $text = $this->updateText( $text, $search );
94 
95  # Perform the actual update
96  $search->update( $this->id, $normalTitle, $search->normalizeText( $text ) );
97  }
98  }
99 
108  public function updateText( $text, SearchEngine $se = null ) {
109  $services = MediaWikiServices::getInstance();
110  $contLang = $services->getContentLanguage();
111  # Language-specific strip/conversion
112  $text = $contLang->normalizeForSearch( $text );
113  $se = $se ?: $services->newSearchEngine();
114  $lc = $se->legalSearchChars() . '&#;';
115 
116  # Strip HTML markup
117  $text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/",
118  ' ', $contLang->lc( " " . $text . " " ) );
119  $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/",
120  "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings
121 
122  # Strip external URLs
123  $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\x80-\\xFF";
124  $protos = "http|https|ftp|mailto|news|gopher";
125  $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
126  $text = preg_replace( $pat, "\\1 \\3", $text );
127 
128  $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
129  $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
130  $text = preg_replace( $p1, "\\1 ", $text );
131  $text = preg_replace( $p2, "\\1 \\3 ", $text );
132 
133  # Internal image links
134  $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
135  $text = preg_replace( $pat2, " \\1 \\3", $text );
136 
137  $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
138  "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
139 
140  # Strip all remaining non-search characters
141  $text = preg_replace( "/[^{$lc}]+/", " ", $text );
142 
159  $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
160  $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );
161 
162  # Strip wiki '' and '''
163  $text = preg_replace( "/''[']*/", " ", $text );
164 
165  return $text;
166  }
167 
177  private function getLatestPage() {
178  if ( !isset( $this->latestPage ) ) {
179  $this->latestPage = MediaWikiServices::getInstance()->getPageStore()
180  ->getPageById( $this->id, PageStore::READ_LATEST );
181  }
182 
183  return $this->latestPage;
184  }
185 
193  private function getNormalizedTitle( SearchEngine $search ) {
194  $contLang = MediaWikiServices::getInstance()->getContentLanguage();
195  $ns = $this->page->getNamespace();
196  $title = str_replace( '_', ' ', $this->page->getDBkey() );
197 
198  $lc = $search->legalSearchChars() . '&#;';
199  $t = $contLang->normalizeForSearch( $title );
200  $t = preg_replace( "/[^{$lc}]+/", ' ', $t );
201  $t = $contLang->lc( $t );
202 
203  # Handle 's, s'
204  $t = preg_replace( "/([{$lc}]+)'s( |$)/", "\\1 \\1's ", $t );
205  $t = preg_replace( "/([{$lc}]+)s'( |$)/", "\\1s ", $t );
206 
207  $t = preg_replace( "/\\s+/", ' ', $t );
208 
209  if ( $ns === NS_FILE ) {
210  $t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
211  }
212 
213  return $search->normalizeText( trim( $t ) );
214  }
215 }
const NS_FILE
Definition: Defines.php:70
PSR-3 logger instance factory.
A class containing constants representing the names of configuration variables.
Service locator for MediaWiki core services.
Contain a class for special pages.
delete( $id, $title)
Delete an indexed page Title should be pre-processed.
supports( $feature)
update( $id, $title, $text)
Create or update the search index record for the given page.
normalizeText( $string)
When overridden in derived class, performs database-specific conversions on text to be used for searc...
updateTitle( $id, $title)
Update a search index record's title only.
legalSearchChars( $type=self::CHARS_ALL)
Get chars legal for search.
Database independent search index updater.
updateText( $text, SearchEngine $se=null)
Clean text for indexing.
doUpdate()
Perform actual update for the entry.
__construct( $id, $page, ?Content $c=null)
Base interface for representing page content.
Definition: Content.php:39
Interface that deferrable updates should implement.
Data record representing a page that currently exists as an editable page on a wiki.
Interface for objects (potentially) representing an editable wiki page.