3 * SQLite search backend, based upon SearchMysql
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
25 * Search engine hook for SQLite
28 class SearchSqlite
extends SearchDatabase
{
30 * Whether fulltext search is supported by current schema
33 function fulltextSearchSupported() {
34 return $this->db
->checkForEnabledSearch();
38 * Parse the user's query and transform it into an SQL fragment which will
39 * become part of a WHERE clause
41 * @param string $filteredText
42 * @param bool $fulltext
45 function parseQuery( $filteredText, $fulltext ) {
47 $lc = $this->legalSearchChars( self
::CHARS_NO_SYNTAX
); // Minus syntax chars (" and *)
49 $this->searchTerms
= [];
52 if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
53 $filteredText, $m, PREG_SET_ORDER
) ) {
54 foreach ( $m as $bits ) {
55 MediaWiki\
suppressWarnings();
56 list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits;
57 MediaWiki\restoreWarnings
();
59 if ( $nonQuoted != '' ) {
63 $term = str_replace( '"', '', $term );
67 if ( $searchon !== '' ) {
71 // Some languages such as Serbian store the input form in the search index,
72 // so we may need to search for matches in multiple writing system variants.
73 $convertedVariants = $wgContLang->autoConvertToAllVariants( $term );
74 if ( is_array( $convertedVariants ) ) {
75 $variants = array_unique( array_values( $convertedVariants ) );
77 $variants = [ $term ];
80 // The low-level search index does some processing on input to work
81 // around problems with minimum lengths and encoding in MySQL's
83 // For Chinese this also inserts spaces between adjacent Han characters.
84 $strippedVariants = array_map(
85 [ $wgContLang, 'normalizeForSearch' ],
88 // Some languages such as Chinese force all variants to a canonical
89 // form when stripping to the low-level search index, so to be sure
90 // let's check our variants list for unique items after stripping.
91 $strippedVariants = array_unique( $strippedVariants );
93 $searchon .= $modifier;
94 if ( count( $strippedVariants ) > 1 ) {
97 foreach ( $strippedVariants as $stripped ) {
98 if ( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
99 // Hack for Chinese: we need to toss in quotes for
100 // multiple-character phrases since normalizeForSearch()
101 // added spaces between them to make word breaks.
102 $stripped = '"' . trim( $stripped ) . '"';
104 $searchon .= "$quote$stripped$quote$wildcard ";
106 if ( count( $strippedVariants ) > 1 ) {
110 // Match individual terms or quoted phrase in result highlighting...
111 // Note that variants will be introduced in a later stage for highlighting!
112 $regexp = $this->regexTerm( $term, $wildcard );
113 $this->searchTerms
[] = $regexp;
117 wfDebug( __METHOD__
. ": Can't understand search query '{$filteredText}'\n" );
120 $searchon = $this->db
->addQuotes( $searchon );
121 $field = $this->getIndexField( $fulltext );
122 return " $field MATCH $searchon ";
125 function regexTerm( $string, $wildcard ) {
128 $regex = preg_quote( $string, '/' );
129 if ( $wgContLang->hasWordBreaks() ) {
131 // Don't cut off the final bit!
134 $regex = "\b$regex\b";
137 // For Chinese, words may legitimately abut other words in the text literal.
138 // Don't add \b boundary checks... note this could cause false positives
144 public static function legalSearchChars( $type = self
::CHARS_ALL
) {
145 $searchChars = parent
::legalSearchChars( $type );
146 if ( $type === self
::CHARS_ALL
) {
147 // " for phrase, * for wildcard
148 $searchChars = "\"*" . $searchChars;
154 * Perform a full text search query and return a result set.
156 * @param string $term Raw search term
157 * @return SqlSearchResultSet
159 function searchText( $term ) {
160 return $this->searchInternal( $term, true );
164 * Perform a title-only search query and return a result set.
166 * @param string $term Raw search term
167 * @return SqlSearchResultSet
169 function searchTitle( $term ) {
170 return $this->searchInternal( $term, false );
173 protected function searchInternal( $term, $fulltext ) {
176 if ( !$this->fulltextSearchSupported() ) {
180 $filteredTerm = $this->filter( $wgContLang->lc( $term ) );
181 $resultSet = $this->db
->query( $this->getQuery( $filteredTerm, $fulltext ) );
184 $totalResult = $this->db
->query( $this->getCountQuery( $filteredTerm, $fulltext ) );
185 $row = $totalResult->fetchObject();
187 $total = intval( $row->c
);
189 $totalResult->free();
191 return new SqlSearchResultSet( $resultSet, $this->searchTerms
, $total );
195 * Return a partial WHERE clause to limit the search to the given namespaces
198 function queryNamespaces() {
199 if ( is_null( $this->namespaces
) ) {
200 return ''; # search all
202 if ( !count( $this->namespaces
) ) {
205 $namespaces = $this->db
->makeList( $this->namespaces
);
207 return 'AND page_namespace IN (' . $namespaces . ')';
211 * Returns a query with limit for number of results set.
215 function limitResult( $sql ) {
216 return $this->db
->limitResult( $sql, $this->limit
, $this->offset
);
220 * Construct the full SQL query to do the search.
221 * The guts shoulds be constructed in queryMain()
222 * @param string $filteredTerm
223 * @param bool $fulltext
226 function getQuery( $filteredTerm, $fulltext ) {
227 return $this->limitResult(
228 $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
229 $this->queryNamespaces()
234 * Picks which field to index on, depending on what type of query.
235 * @param bool $fulltext
238 function getIndexField( $fulltext ) {
239 return $fulltext ?
'si_text' : 'si_title';
243 * Get the base part of the search query.
245 * @param string $filteredTerm
246 * @param bool $fulltext
249 function queryMain( $filteredTerm, $fulltext ) {
250 $match = $this->parseQuery( $filteredTerm, $fulltext );
251 $page = $this->db
->tableName( 'page' );
252 $searchindex = $this->db
->tableName( 'searchindex' );
253 return "SELECT $searchindex.rowid, page_namespace, page_title " .
254 "FROM $page,$searchindex " .
255 "WHERE page_id=$searchindex.rowid AND $match";
258 function getCountQuery( $filteredTerm, $fulltext ) {
259 $match = $this->parseQuery( $filteredTerm, $fulltext );
260 $page = $this->db
->tableName( 'page' );
261 $searchindex = $this->db
->tableName( 'searchindex' );
262 return "SELECT COUNT(*) AS c " .
263 "FROM $page,$searchindex " .
264 "WHERE page_id=$searchindex.rowid AND $match " .
265 $this->queryNamespaces();
269 * Create or update the search index record for the given page.
270 * Title and text should be pre-processed.
273 * @param string $title
274 * @param string $text
276 function update( $id, $title, $text ) {
277 if ( !$this->fulltextSearchSupported() ) {
280 // @todo find a method to do it in a single request,
281 // couldn't do it so far due to typelessness of FTS3 tables.
282 $dbw = wfGetDB( DB_MASTER
);
284 $dbw->delete( 'searchindex', [ 'rowid' => $id ], __METHOD__
);
286 $dbw->insert( 'searchindex',
289 'si_title' => $title,
295 * Update a search index record's title only.
296 * Title should be pre-processed.
299 * @param string $title
301 function updateTitle( $id, $title ) {
302 if ( !$this->fulltextSearchSupported() ) {
305 $dbw = wfGetDB( DB_MASTER
);
307 $dbw->update( 'searchindex',
308 [ 'si_title' => $title ],