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, $c = null ) {
57  $this->page = $page;
58 
59  $this->id = $id;
60  // is_string() check is back-compat for ApprovedRevs
61  if ( is_string( $c ) ) {
62  wfDeprecated( __METHOD__ . " with a string for the content", '1.34' );
63  $c = new TextContent( $c );
64  } elseif ( is_bool( $c ) ) {
65  wfDeprecated( __METHOD__ . " with a boolean for the content", '1.34' );
66  $c = null;
67  }
68  $this->content = $c;
69  }
70 
74  public function doUpdate() {
75  $services = MediaWikiServices::getInstance();
76  $config = $services->getSearchEngineConfig();
77 
78  if ( $config->getConfig()->get( MainConfigNames::DisableSearchUpdate ) || !$this->id ) {
79  LoggerFactory::getInstance( "search" )
80  ->debug( "Skipping update: search updates disabled by config" );
81  return;
82  }
83 
84  $seFactory = $services->getSearchEngineFactory();
85  foreach ( $config->getSearchTypes() as $type ) {
86  $search = $seFactory->create( $type );
87  if ( !$search->supports( 'search-update' ) ) {
88  continue;
89  }
90 
91  $normalTitle = $this->getNormalizedTitle( $search );
92 
93  if ( $this->getLatestPage() === null ) {
94  $search->delete( $this->id, $normalTitle );
95  continue;
96  } elseif ( $this->content === null ) {
97  $search->updateTitle( $this->id, $normalTitle );
98  continue;
99  }
100 
101  $text = $this->content !== null ? $this->content->getTextForSearchIndex() : '';
102  $text = $this->updateText( $text, $search );
103 
104  # Perform the actual update
105  $search->update( $this->id, $normalTitle, $search->normalizeText( $text ) );
106  }
107  }
108 
117  public function updateText( $text, SearchEngine $se = null ) {
118  $services = MediaWikiServices::getInstance();
119  $contLang = $services->getContentLanguage();
120  # Language-specific strip/conversion
121  $text = $contLang->normalizeForSearch( $text );
122  $se = $se ?: $services->newSearchEngine();
123  $lc = $se->legalSearchChars() . '&#;';
124 
125  # Strip HTML markup
126  $text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/",
127  ' ', $contLang->lc( " " . $text . " " ) );
128  $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD",
129  "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings
130 
131  # Strip external URLs
132  $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\x80-\\xFF";
133  $protos = "http|https|ftp|mailto|news|gopher";
134  $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
135  $text = preg_replace( $pat, "\\1 \\3", $text );
136 
137  $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
138  $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
139  $text = preg_replace( $p1, "\\1 ", $text );
140  $text = preg_replace( $p2, "\\1 \\3 ", $text );
141 
142  # Internal image links
143  $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
144  $text = preg_replace( $pat2, " \\1 \\3", $text );
145 
146  $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
147  "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
148 
149  # Strip all remaining non-search characters
150  $text = preg_replace( "/[^{$lc}]+/", " ", $text );
151 
168  $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
169  $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );
170 
171  # Strip wiki '' and '''
172  $text = preg_replace( "/''[']*/", " ", $text );
173 
174  return $text;
175  }
176 
186  private function getLatestPage() {
187  if ( !isset( $this->latestPage ) ) {
188  $this->latestPage = MediaWikiServices::getInstance()->getPageStore()
189  ->getPageById( $this->id, PageStore::READ_LATEST );
190  }
191 
192  return $this->latestPage;
193  }
194 
202  private function getNormalizedTitle( SearchEngine $search ) {
203  $contLang = MediaWikiServices::getInstance()->getContentLanguage();
204  $ns = $this->page->getNamespace();
205  $title = str_replace( '_', ' ', $this->page->getDBkey() );
206 
207  $lc = $search->legalSearchChars() . '&#;';
208  $t = $contLang->normalizeForSearch( $title );
209  $t = preg_replace( "/[^{$lc}]+/", ' ', $t );
210  $t = $contLang->lc( $t );
211 
212  # Handle 's, s'
213  $t = preg_replace( "/([{$lc}]+)'s( |$)/", "\\1 \\1's ", $t );
214  $t = preg_replace( "/([{$lc}]+)s'( |$)/", "\\1s ", $t );
215 
216  $t = preg_replace( "/\\s+/", ' ', $t );
217 
218  if ( $ns === NS_FILE ) {
219  $t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
220  }
221 
222  return $search->normalizeText( trim( $t ) );
223  }
224 }
const NS_FILE
Definition: Defines.php:70
wfDeprecated( $function, $version=false, $component=false, $callerOffset=2)
Logs a warning that a deprecated feature was used.
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, $c=null)
Content object implementation for representing flat text.
Definition: TextContent.php:40
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.