MediaWiki  master
PrefixSearch.php
Go to the documentation of this file.
1 <?php
24 
33 abstract class PrefixSearch {
43  public function search( $search, $limit, $namespaces = [], $offset = 0 ) {
44  $search = trim( $search );
45  if ( $search == '' ) {
46  return []; // Return empty result
47  }
48 
49  $hasNamespace = SearchEngine::parseNamespacePrefixes( $search, false, true );
50  if ( $hasNamespace !== false ) {
51  list( $search, $namespaces ) = $hasNamespace;
52  }
53 
54  return $this->searchBackend( $namespaces, $search, $limit, $offset );
55  }
56 
66  public function searchWithVariants( $search, $limit, array $namespaces, $offset = 0 ) {
67  $searches = $this->search( $search, $limit, $namespaces, $offset );
68 
69  // if the content language has variants, try to retrieve fallback results
70  $fallbackLimit = $limit - count( $searches );
71  if ( $fallbackLimit > 0 ) {
72  $services = MediaWikiServices::getInstance();
73  $fallbackSearches = $services->getLanguageConverterFactory()
74  ->getLanguageConverter( $services->getContentLanguage() )
75  ->autoConvertToAllVariants( $search );
76  $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
77 
78  foreach ( $fallbackSearches as $fbs ) {
79  $fallbackSearchResult = $this->search( $fbs, $fallbackLimit, $namespaces );
80  $searches = array_merge( $searches, $fallbackSearchResult );
81  $fallbackLimit -= count( $fallbackSearchResult );
82 
83  if ( $fallbackLimit == 0 ) {
84  break;
85  }
86  }
87  }
88  return $searches;
89  }
90 
98  abstract protected function titles( array $titles );
99 
107  abstract protected function strings( array $strings );
108 
117  protected function searchBackend( $namespaces, $search, $limit, $offset ) {
118  if ( count( $namespaces ) == 1 ) {
119  $ns = $namespaces[0];
120  if ( $ns == NS_MEDIA ) {
121  $namespaces = [ NS_FILE ];
122  } elseif ( $ns == NS_SPECIAL ) {
123  return $this->titles( $this->specialSearch( $search, $limit, $offset ) );
124  }
125  }
126  $srchres = [];
127  if ( Hooks::runner()->onPrefixSearchBackend(
128  $namespaces, $search, $limit, $srchres, $offset )
129  ) {
130  return $this->titles( $this->defaultSearchBackend( $namespaces, $search, $limit, $offset ) );
131  }
132  return $this->strings(
133  $this->handleResultFromHook( $srchres, $namespaces, $search, $limit, $offset ) );
134  }
135 
136  private function handleResultFromHook( $srchres, $namespaces, $search, $limit, $offset ) {
137  if ( $offset === 0 ) {
138  // Only perform exact db match if offset === 0
139  // This is still far from perfect but at least we avoid returning the
140  // same title afain and again when the user is scrolling with a query
141  // that matches a title in the db.
142  $rescorer = new SearchExactMatchRescorer();
143  $srchres = $rescorer->rescore( $search, $namespaces, $srchres, $limit );
144  }
145  return $srchres;
146  }
147 
156  protected function specialSearch( $search, $limit, $offset ) {
157  $searchParts = explode( '/', $search, 2 );
158  $searchKey = $searchParts[0];
159  $subpageSearch = $searchParts[1] ?? null;
160 
161  // Handle subpage search separately.
162  $spFactory = MediaWikiServices::getInstance()->getSpecialPageFactory();
163  if ( $subpageSearch !== null ) {
164  // Try matching the full search string as a page name
165  $specialTitle = Title::makeTitleSafe( NS_SPECIAL, $searchKey );
166  if ( !$specialTitle ) {
167  return [];
168  }
169  $special = $spFactory->getPage( $specialTitle->getText() );
170  if ( $special ) {
171  $subpages = $special->prefixSearchSubpages( $subpageSearch, $limit, $offset );
172  return array_map( static function ( $sub ) use ( $specialTitle ) {
173  return $specialTitle->getSubpage( $sub );
174  }, $subpages );
175  } else {
176  return [];
177  }
178  }
179 
180  # normalize searchKey, so aliases with spaces can be found - T27675
181  $contLang = MediaWikiServices::getInstance()->getContentLanguage();
182  $searchKey = str_replace( ' ', '_', $searchKey );
183  $searchKey = $contLang->caseFold( $searchKey );
184 
185  // Unlike SpecialPage itself, we want the canonical forms of both
186  // canonical and alias title forms...
187  $keys = [];
188  foreach ( $spFactory->getNames() as $page ) {
189  $keys[$contLang->caseFold( $page )] = [ 'page' => $page, 'rank' => 0 ];
190  }
191 
192  foreach ( $contLang->getSpecialPageAliases() as $page => $aliases ) {
193  if ( !in_array( $page, $spFactory->getNames() ) ) {# T22885
194  continue;
195  }
196 
197  foreach ( $aliases as $key => $alias ) {
198  $keys[$contLang->caseFold( $alias )] = [ 'page' => $alias, 'rank' => $key ];
199  }
200  }
201  ksort( $keys );
202 
203  $matches = [];
204  foreach ( $keys as $pageKey => $page ) {
205  if ( $searchKey === '' || strpos( $pageKey, $searchKey ) === 0 ) {
206  // T29671: Don't use SpecialPage::getTitleFor() here because it
207  // localizes its input leading to searches for e.g. Special:All
208  // returning Spezial:MediaWiki-Systemnachrichten and returning
209  // Spezial:Alle_Seiten twice when $wgLanguageCode == 'de'
210  $matches[$page['rank']][] = Title::makeTitleSafe( NS_SPECIAL, $page['page'] );
211 
212  if ( isset( $matches[0] ) && count( $matches[0] ) >= $limit + $offset ) {
213  // We have enough items in primary rank, no use to continue
214  break;
215  }
216  }
217 
218  }
219 
220  // Ensure keys are in order
221  ksort( $matches );
222  // Flatten the array
223  $matches = array_reduce( $matches, 'array_merge', [] );
224 
225  return array_slice( $matches, $offset, $limit );
226  }
227 
240  public function defaultSearchBackend( $namespaces, $search, $limit, $offset ) {
241  if ( !$namespaces ) {
242  $namespaces = [ NS_MAIN ];
243  }
244 
245  if ( in_array( NS_SPECIAL, $namespaces ) ) {
246  // For now, if special is included, ignore the other namespaces
247  return $this->specialSearch( $search, $limit, $offset );
248  }
249 
250  // Construct suitable prefix for each namespace. They differ in cases where
251  // some namespaces always capitalize and some don't.
252  $prefixes = [];
253  // Allow to do a prefix search for e.g. "Talk:"
254  if ( $search === '' ) {
255  $prefixes[$search] = $namespaces;
256  } else {
257  // Don't just ignore input like "[[Foo]]", but try to search for "Foo"
258  $search = preg_replace( MediaWikiTitleCodec::getTitleInvalidRegex(), '', $search );
259  foreach ( $namespaces as $namespace ) {
260  $title = Title::makeTitleSafe( $namespace, $search );
261  if ( $title ) {
262  $prefixes[ $title->getDBkey() ][] = $namespace;
263  }
264  }
265  }
266  if ( !$prefixes ) {
267  return [];
268  }
269 
270  $dbr = wfGetDB( DB_REPLICA );
271  // Often there is only one prefix that applies to all requested namespaces,
272  // but sometimes there are two if some namespaces do not always capitalize.
273  $conds = [];
274  foreach ( $prefixes as $prefix => $namespaces ) {
275  $condition = [ 'page_namespace' => $namespaces ];
276  if ( $prefix !== '' ) {
277  $condition[] = 'page_title' . $dbr->buildLike( $prefix, $dbr->anyString() );
278  }
279  $conds[] = $dbr->makeList( $condition, LIST_AND );
280  }
281 
282  $table = 'page';
283  $fields = [ 'page_id', 'page_namespace', 'page_title' ];
284  $conds = $dbr->makeList( $conds, LIST_OR );
285  $options = [
286  'LIMIT' => $limit,
287  'ORDER BY' => [ 'page_title', 'page_namespace' ],
288  'OFFSET' => $offset
289  ];
290 
291  $res = $dbr->select( $table, $fields, $conds, __METHOD__, $options );
292 
293  return iterator_to_array( TitleArray::newFromResult( $res ) );
294  }
295 
302  protected function validateNamespaces( $namespaces ) {
303  // We will look at each given namespace against content language namespaces
304  $validNamespaces = MediaWikiServices::getInstance()->getContentLanguage()->getNamespaces();
305  if ( is_array( $namespaces ) && count( $namespaces ) > 0 ) {
306  $valid = [];
307  foreach ( $namespaces as $ns ) {
308  if ( is_numeric( $ns ) && array_key_exists( $ns, $validNamespaces ) ) {
309  $valid[] = $ns;
310  }
311  }
312  if ( count( $valid ) > 0 ) {
313  return $valid;
314  }
315  }
316 
317  return [ NS_MAIN ];
318  }
319 }
LIST_OR
const LIST_OR
Definition: Defines.php:46
PrefixSearch\specialSearch
specialSearch( $search, $limit, $offset)
Prefix search special-case for Special: namespace.
Definition: PrefixSearch.php:156
TitleArray\newFromResult
static newFromResult( $res)
Definition: TitleArray.php:44
MediaWiki\MediaWikiServices
MediaWikiServices is the service locator for the application scope of MediaWiki.
Definition: MediaWikiServices.php:193
LIST_AND
const LIST_AND
Definition: Defines.php:43
MediaWikiTitleCodec\getTitleInvalidRegex
static getTitleInvalidRegex()
Returns a simple regex that will match on characters and sequences invalid in titles.
Definition: MediaWikiTitleCodec.php:590
PrefixSearch\handleResultFromHook
handleResultFromHook( $srchres, $namespaces, $search, $limit, $offset)
Definition: PrefixSearch.php:136
SearchEngine\parseNamespacePrefixes
static parseNamespacePrefixes( $query, $withAllKeyword=true, $withPrefixSearchExtractNamespaceHook=false)
Parse some common prefixes: all (search everything) or namespace names.
Definition: SearchEngine.php:399
$res
$res
Definition: testCompression.php:57
NS_MAIN
const NS_MAIN
Definition: Defines.php:64
$dbr
$dbr
Definition: testCompression.php:54
NS_SPECIAL
const NS_SPECIAL
Definition: Defines.php:53
SearchExactMatchRescorer
An utility class to rescore search results by looking for an exact match in the db and add the page f...
Definition: SearchExactMatchRescorer.php:33
PrefixSearch\searchWithVariants
searchWithVariants( $search, $limit, array $namespaces, $offset=0)
Do a prefix search for all possible variants of the prefix.
Definition: PrefixSearch.php:66
wfGetDB
wfGetDB( $db, $groups=[], $wiki=false)
Get a Database object.
Definition: GlobalFunctions.php:2203
$matches
$matches
Definition: NoLocalSettings.php:24
PrefixSearch\search
search( $search, $limit, $namespaces=[], $offset=0)
Do a prefix search of titles and return a list of matching page names.
Definition: PrefixSearch.php:43
PrefixSearch\strings
strings(array $strings)
When implemented in a descendant class, receives an array of titles as strings and returns either an ...
$title
$title
Definition: testCompression.php:38
DB_REPLICA
const DB_REPLICA
Definition: defines.php:25
Title\makeTitleSafe
static makeTitleSafe( $ns, $title, $fragment='', $interwiki='')
Create a new Title from a namespace index and a DB key.
Definition: Title.php:676
NS_MEDIA
const NS_MEDIA
Definition: Defines.php:52
Hooks\runner
static runner()
Get a HookRunner instance for calling hooks using the new interfaces.
Definition: Hooks.php:173
PrefixSearch\validateNamespaces
validateNamespaces( $namespaces)
Validate an array of numerical namespace indexes.
Definition: PrefixSearch.php:302
PrefixSearch\searchBackend
searchBackend( $namespaces, $search, $limit, $offset)
Do a prefix search of titles and return a list of matching page names.
Definition: PrefixSearch.php:117
PrefixSearch\defaultSearchBackend
defaultSearchBackend( $namespaces, $search, $limit, $offset)
Unless overridden by PrefixSearchBackend hook...
Definition: PrefixSearch.php:240
PrefixSearch
Handles searching prefixes of titles and finding any page names that match.
Definition: PrefixSearch.php:33
$keys
$keys
Definition: testCompression.php:72
NS_FILE
const NS_FILE
Definition: Defines.php:70
PrefixSearch\titles
titles(array $titles)
When implemented in a descendant class, receives an array of Title objects and returns either an unmo...