Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 98 |
|
0.00% |
0 / 6 |
CRAP | |
0.00% |
0 / 1 |
| SearchPostgres | |
0.00% |
0 / 97 |
|
0.00% |
0 / 6 |
380 | |
0.00% |
0 / 1 |
| doSearchTitleInDB | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchTextInDB | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| parseQuery | |
0.00% |
0 / 27 |
|
0.00% |
0 / 1 |
72 | |||
| searchQuery | |
0.00% |
0 / 42 |
|
0.00% |
0 / 1 |
56 | |||
| update | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
2 | |||
| updateTitle | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * PostgreSQL search engine |
| 4 | * |
| 5 | * Copyright © 2006-2007 Greg Sabino Mullane <greg@turnstep.com> |
| 6 | * https://www.mediawiki.org/ |
| 7 | * |
| 8 | * @license GPL-2.0-or-later |
| 9 | * @file |
| 10 | * @ingroup Search |
| 11 | */ |
| 12 | |
| 13 | namespace MediaWiki\Search; |
| 14 | |
| 15 | use MediaWiki\MediaWikiServices; |
| 16 | use MediaWiki\Revision\SlotRecord; |
| 17 | use Wikimedia\Rdbms\IDatabase; |
| 18 | |
| 19 | /** |
| 20 | * Search engine hook base class for Postgres |
| 21 | * @ingroup Search |
| 22 | */ |
| 23 | class SearchPostgres extends SearchDatabase { |
| 24 | /** |
| 25 | * Perform a full text search query via tsearch2 and return a result set. |
| 26 | * Currently searches a page's current title (page.page_title) and |
| 27 | * latest revision article text (text.old_text) |
| 28 | * |
| 29 | * @param string $term Raw search term |
| 30 | * @return SqlSearchResultSet |
| 31 | */ |
| 32 | protected function doSearchTitleInDB( $term ) { |
| 33 | $q = $this->searchQuery( $term, 'titlevector' ); |
| 34 | $olderror = error_reporting( E_ERROR ); |
| 35 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 36 | // The real type is still IDatabase, but IReplicaDatabase is used for safety. |
| 37 | '@phan-var IDatabase $dbr'; |
| 38 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 39 | $resultSet = $dbr->query( $q, 'SearchPostgres', IDatabase::QUERY_SILENCE_ERRORS ); |
| 40 | error_reporting( $olderror ); |
| 41 | return new SqlSearchResultSet( $resultSet, $this->searchTerms ); |
| 42 | } |
| 43 | |
| 44 | /** @inheritDoc */ |
| 45 | protected function doSearchTextInDB( $term ) { |
| 46 | $q = $this->searchQuery( $term, 'textvector' ); |
| 47 | $olderror = error_reporting( E_ERROR ); |
| 48 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 49 | // The real type is still IDatabase, but IReplicaDatabase is used for safety. |
| 50 | '@phan-var IDatabase $dbr'; |
| 51 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 52 | $resultSet = $dbr->query( $q, 'SearchPostgres', IDatabase::QUERY_SILENCE_ERRORS ); |
| 53 | error_reporting( $olderror ); |
| 54 | return new SqlSearchResultSet( $resultSet, $this->searchTerms ); |
| 55 | } |
| 56 | |
| 57 | /** |
| 58 | * Transform the user's search string into a better form for tsearch2 |
| 59 | * Returns an SQL fragment consisting of quoted text to search for. |
| 60 | * |
| 61 | * @param string $term |
| 62 | * |
| 63 | * @return string |
| 64 | */ |
| 65 | private function parseQuery( $term ) { |
| 66 | wfDebug( "parseQuery received: $term" ); |
| 67 | |
| 68 | // No backslashes allowed |
| 69 | $term = preg_replace( '/\\\\/', '', $term ); |
| 70 | |
| 71 | // Collapse parens into nearby words: |
| 72 | $term = preg_replace( '/\s*\(\s*/', ' (', $term ); |
| 73 | $term = preg_replace( '/\s*\)\s*/', ') ', $term ); |
| 74 | |
| 75 | // Treat colons as word separators: |
| 76 | $term = preg_replace( '/:/', ' ', $term ); |
| 77 | |
| 78 | $searchstring = ''; |
| 79 | $m = []; |
| 80 | if ( preg_match_all( '/([-!]?)(\S+)\s*/', $term, $m, PREG_SET_ORDER ) ) { |
| 81 | foreach ( $m as $terms ) { |
| 82 | if ( $terms[1] !== '' ) { |
| 83 | $searchstring .= ' & !'; |
| 84 | } |
| 85 | if ( strtolower( $terms[2] ) === 'and' ) { |
| 86 | $searchstring .= ' & '; |
| 87 | } elseif ( strtolower( $terms[2] ) === 'or' || $terms[2] === '|' ) { |
| 88 | $searchstring .= ' | '; |
| 89 | } elseif ( strtolower( $terms[2] ) === 'not' ) { |
| 90 | $searchstring .= ' & !'; |
| 91 | } else { |
| 92 | $searchstring .= " & $terms[2]"; |
| 93 | } |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | // Strip out leading junk |
| 98 | $searchstring = preg_replace( '/^[\s\&\|]+/', '', $searchstring ); |
| 99 | |
| 100 | // Remove any doubled-up operators |
| 101 | $searchstring = preg_replace( '/([\!\&\|]) +(?:[\&\|] +)+/', "$1 ", $searchstring ); |
| 102 | |
| 103 | // Remove any non-spaced operators (e.g. "Zounds!") |
| 104 | $searchstring = preg_replace( '/([^ ])[\!\&\|]/', "$1", $searchstring ); |
| 105 | |
| 106 | // Remove any trailing whitespace or operators |
| 107 | $searchstring = preg_replace( '/[\s\!\&\|]+$/', '', $searchstring ); |
| 108 | |
| 109 | // Remove unnecessary quotes around everything |
| 110 | $searchstring = preg_replace( '/^[\'"](.*)[\'"]$/', "$1", $searchstring ); |
| 111 | |
| 112 | // Quote the whole thing |
| 113 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 114 | $searchstring = $dbr->addQuotes( $searchstring ); |
| 115 | |
| 116 | wfDebug( "parseQuery returned: $searchstring" ); |
| 117 | |
| 118 | return $searchstring; |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * Construct the full SQL query to do the search. |
| 123 | * @param string $term |
| 124 | * @param string $fulltext |
| 125 | * @return string |
| 126 | */ |
| 127 | private function searchQuery( $term, $fulltext ) { |
| 128 | # Get the SQL fragment for the given term |
| 129 | $searchstring = $this->parseQuery( $term ); |
| 130 | |
| 131 | // We need a separate query here so gin does not complain about empty searches |
| 132 | $sql = "SELECT to_tsquery($searchstring)"; |
| 133 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 134 | // The real type is still IDatabase, but IReplicaDatabase is used for safety. |
| 135 | '@phan-var IDatabase $dbr'; |
| 136 | // phpcs:ignore MediaWiki.Usage.DbrQueryUsage.DbrQueryFound |
| 137 | $res = $dbr->query( $sql, __METHOD__ ); |
| 138 | if ( !$res ) { |
| 139 | // TODO: Better output (example to catch: one 'two) |
| 140 | die( "Sorry, that was not a valid search string. Please go back and try again" ); |
| 141 | } |
| 142 | $top = $res->fetchRow()[0]; |
| 143 | |
| 144 | $this->searchTerms = []; |
| 145 | $slotRoleStore = MediaWikiServices::getInstance()->getSlotRoleStore(); |
| 146 | if ( $top === "" ) { // e.g. if only stopwords are used XXX return something better |
| 147 | $query = "SELECT page_id, page_namespace, page_title, 0 AS score " . |
| 148 | "FROM page p, revision r, slots s, content c, \"text\" pc " . |
| 149 | "WHERE p.page_latest = r.rev_id " . |
| 150 | "AND s.slot_revision_id = r.rev_id " . |
| 151 | "AND s.slot_role_id = " . |
| 152 | $dbr->addQuotes( $slotRoleStore->acquireId( SlotRecord::MAIN ) ) . " " . |
| 153 | "AND c.content_id = s.slot_content_id " . |
| 154 | "AND pc.old_id = substring( c.content_address from '^tt:([0-9]+)$' )::int " . |
| 155 | "AND 1=0"; |
| 156 | } else { |
| 157 | $m = []; |
| 158 | if ( preg_match_all( "/'([^']+)'/", $top, $m, PREG_SET_ORDER ) ) { |
| 159 | foreach ( $m as $terms ) { |
| 160 | $this->searchTerms[$terms[1]] = $terms[1]; |
| 161 | } |
| 162 | } |
| 163 | |
| 164 | $query = "SELECT page_id, page_namespace, page_title, " . |
| 165 | "ts_rank($fulltext, to_tsquery($searchstring), 5) AS score " . |
| 166 | "FROM page p, revision r, slots s, content c, \"text\" pc " . |
| 167 | "WHERE p.page_latest = r.rev_id " . |
| 168 | "AND s.slot_revision_id = r.rev_id " . |
| 169 | "AND s.slot_role_id = " . $dbr->addQuotes( |
| 170 | $slotRoleStore->acquireId( SlotRecord::MAIN ) ) . " " . |
| 171 | "AND c.content_id = s.slot_content_id " . |
| 172 | "AND pc.old_id = substring( c.content_address from '^tt:([0-9]+)$' )::int " . |
| 173 | "AND $fulltext @@ to_tsquery($searchstring)"; |
| 174 | } |
| 175 | // Namespaces - defaults to main |
| 176 | if ( $this->namespaces !== null ) { // null -> search all |
| 177 | if ( count( $this->namespaces ) < 1 ) { |
| 178 | $query .= ' AND page_namespace = ' . NS_MAIN; |
| 179 | } else { |
| 180 | $namespaces = $dbr->makeList( $this->namespaces ); |
| 181 | $query .= " AND page_namespace IN ($namespaces)"; |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | $query .= " ORDER BY score DESC, page_id DESC"; |
| 186 | |
| 187 | $query .= $dbr->limitResult( '', $this->limit, $this->offset ); |
| 188 | |
| 189 | wfDebug( "searchQuery returned: $query" ); |
| 190 | |
| 191 | return $query; |
| 192 | } |
| 193 | |
| 194 | // Most of the work of these two functions are done automatically via triggers |
| 195 | |
| 196 | /** @inheritDoc */ |
| 197 | public function update( $pageid, $title, $text ) { |
| 198 | // We don't want to index older revisions |
| 199 | $slotRoleStore = MediaWikiServices::getInstance()->getSlotRoleStore(); |
| 200 | $dbw = $this->dbProvider->getPrimaryDatabase(); |
| 201 | $sql = "UPDATE \"text\" SET textvector = NULL " . |
| 202 | "WHERE textvector IS NOT NULL " . |
| 203 | "AND old_id IN " . |
| 204 | "(SELECT DISTINCT substring( c.content_address from '^tt:([0-9]+)$' )::int AS old_rev_text_id " . |
| 205 | " FROM content c, slots s, revision r " . |
| 206 | " WHERE r.rev_page = $pageid " . |
| 207 | " AND s.slot_revision_id = r.rev_id " . |
| 208 | " AND s.slot_role_id = " . |
| 209 | $dbw->addQuotes( $slotRoleStore->acquireId( SlotRecord::MAIN ) ) . " " . |
| 210 | " AND c.content_id = s.slot_content_id " . |
| 211 | " ORDER BY old_rev_text_id DESC OFFSET 1)"; |
| 212 | |
| 213 | $dbw->query( $sql, __METHOD__ ); |
| 214 | |
| 215 | return true; |
| 216 | } |
| 217 | |
| 218 | /** @inheritDoc */ |
| 219 | public function updateTitle( $id, $title ) { |
| 220 | return true; |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | /** @deprecated class alias since 1.46 */ |
| 225 | class_alias( SearchPostgres::class, 'SearchPostgres' ); |