| Index: trunk/extensions/SphinxSearch/SphinxSearch.php |
| — | — | @@ -65,7 +65,7 @@ |
| 66 | 66 | $wgSphinxSearch_index_weights = null; |
| 67 | 67 | |
| 68 | 68 | # Default Sphinx search mode |
| 69 | | -$wgSphinxSearch_mode = SPH_MATCH_EXTENDED; |
| | 69 | +$wgSphinxSearch_mode = SPH_MATCH_EXTENDED2; |
| 70 | 70 | |
| 71 | 71 | # Default sort mode |
| 72 | 72 | $wgSphinxSearch_sortmode = SPH_SORT_RELEVANCE; |
| Index: trunk/extensions/SphinxSearch/SphinxMWSearch.php |
| — | — | @@ -18,6 +18,12 @@ |
| 19 | 19 | var $exc_categories = array(); |
| 20 | 20 | var $db; |
| 21 | 21 | var $sphinx_client = null; |
| | 22 | + var $prefix_handlers = array( |
| | 23 | + 'all' => 'searchAllNamespaces', |
| | 24 | + 'intitle' => 'filterByTitle', |
| | 25 | + 'incategory' => 'filterByCategory', |
| | 26 | + 'prefix' => 'filterByPrefix', |
| | 27 | + ); |
| 22 | 28 | |
| 23 | 29 | /** |
| 24 | 30 | * Do not go to a near match if query prefixed with ~ |
| — | — | @@ -117,29 +123,26 @@ |
| 118 | 124 | |
| 119 | 125 | $cl = new SphinxClient(); |
| 120 | 126 | |
| 121 | | - // setup the options for searching |
| 122 | | - if ( isset( $wgSphinxSearch_host ) && isset( $wgSphinxSearch_port ) ) { |
| 123 | | - $cl->SetServer( $wgSphinxSearch_host, $wgSphinxSearch_port ); |
| 124 | | - } |
| 125 | | - if ( count( $wgSphinxSearch_weights ) ) { |
| | 127 | + $cl->SetServer( $wgSphinxSearch_host, $wgSphinxSearch_port ); |
| | 128 | + if ( $wgSphinxSearch_weights && count( $wgSphinxSearch_weights ) ) { |
| 126 | 129 | $cl->SetFieldWeights( $wgSphinxSearch_weights ); |
| 127 | 130 | } |
| 128 | 131 | if ( is_array( $wgSphinxSearch_index_weights ) ) { |
| 129 | 132 | $cl->SetIndexWeights( $wgSphinxSearch_index_weights ); |
| 130 | 133 | } |
| 131 | | - if ( isset( $wgSphinxSearch_mode ) ) { |
| | 134 | + if ( $wgSphinxSearch_mode ) { |
| 132 | 135 | $cl->SetMatchMode( $wgSphinxSearch_mode ); |
| 133 | 136 | } |
| 134 | | - if ( count( $this->namespaces ) ) { |
| | 137 | + if ( $this->namespaces && count( $this->namespaces ) ) { |
| 135 | 138 | $cl->SetFilter( 'page_namespace', $this->namespaces ); |
| 136 | 139 | } |
| 137 | 140 | if( !$this->showRedirects ) { |
| 138 | 141 | $cl->SetFilter( 'page_is_redirect', array( 0 ) ); |
| 139 | 142 | } |
| 140 | | - if ( count( $this->categories ) ) { |
| | 143 | + if ( $this->categories && count( $this->categories ) ) { |
| 141 | 144 | $cl->SetFilter( 'category', $this->categories ); |
| 142 | 145 | } |
| 143 | | - if ( count( $this->exc_categories ) ) { |
| | 146 | + if ( $this->exc_categories && count( $this->exc_categories ) ) { |
| 144 | 147 | $cl->SetFilter( 'category', $this->exc_categories, true ); |
| 145 | 148 | } |
| 146 | 149 | $cl->SetSortMode( $wgSphinxSearch_sortmode, $wgSphinxSearch_sortby ); |
| — | — | @@ -167,6 +170,136 @@ |
| 168 | 171 | return array( $contextlines, $contextchars ); |
| 169 | 172 | } |
| 170 | 173 | |
| | 174 | + /** |
| | 175 | + * Prepare query for sphinx search daemon |
| | 176 | + * |
| | 177 | + * @param string $query |
| | 178 | + * @return string rewritten query |
| | 179 | + */ |
| | 180 | + function replacePrefixes( $query ) { |
| | 181 | + // ~ prefix is used to avoid near-term search, remove it now |
| | 182 | + if ( $query[ 0 ] === '~' ) { |
| | 183 | + $query = substr( $query, 1 ); |
| | 184 | + } |
| | 185 | + |
| | 186 | + $parts = preg_split( '/(")/', $query, -1, PREG_SPLIT_DELIM_CAPTURE ); |
| | 187 | + $inquotes = false; |
| | 188 | + $rewritten = ''; |
| | 189 | + foreach ( $parts as $part ) { |
| | 190 | + if ( $part == '"' ) { // stuff in quotes doesn't get rewritten |
| | 191 | + $rewritten .= $part; |
| | 192 | + $inquotes = !$inquotes; |
| | 193 | + } elseif ( $inquotes ) { |
| | 194 | + $rewritten .= $part; |
| | 195 | + } else { |
| | 196 | + if ( strpos( $query, ':' ) !== false ) { |
| | 197 | + $regexp = $this->preparePrefixRegexp(); |
| | 198 | + $part = preg_replace_callback( |
| | 199 | + '/(^|[| :])(' . $regexp . '):([^ ]+)/i', |
| | 200 | + array( $this, 'replaceQueryPrefix' ), |
| | 201 | + $part |
| | 202 | + ); |
| | 203 | + } |
| | 204 | + $rewritten .= str_replace( |
| | 205 | + array( ' OR ', ' AND ' ), |
| | 206 | + array( ' | ', ' & ' ), |
| | 207 | + $part |
| | 208 | + ); |
| | 209 | + } |
| | 210 | + } |
| | 211 | + return $rewritten; |
| | 212 | + } |
| | 213 | + |
| | 214 | + /** |
| | 215 | + * @return string Regexp to match namespaces and other prefixes |
| | 216 | + */ |
| | 217 | + function preparePrefixRegexp() { |
| | 218 | + global $wgContLang, $wgCanonicalNamespaceNames, $wgNamespaceAliases; |
| | 219 | + |
| | 220 | + $nsNamesRaw = array_merge( |
| | 221 | + $wgContLang->getNamespaces(), |
| | 222 | + $wgCanonicalNamespaceNames, |
| | 223 | + array_keys( array_merge( $wgNamespaceAliases, $wgContLang->getNamespaceAliases() ) ) |
| | 224 | + ); |
| | 225 | + |
| | 226 | + // add all namespace names w/o spaces |
| | 227 | + $nsNames = array(); |
| | 228 | + foreach ( $nsNamesRaw as $ns ) { |
| | 229 | + if ( $ns != '' ) { |
| | 230 | + $nsNames[] = str_replace( ' ', '_', $ns ); |
| | 231 | + } |
| | 232 | + } |
| | 233 | + |
| | 234 | + // "search everything" keyword |
| | 235 | + $allkeyword = wfMsgForContent( 'searchall' ); |
| | 236 | + $this->prefix_handlers[ $allkeyword ] = 'searchAllNamespaces'; |
| | 237 | + |
| | 238 | + // add other kinds of prefixes we support |
| | 239 | + $nsNames = array_merge( $nsNames, array_keys( $this->prefix_handlers ) ); |
| | 240 | + |
| | 241 | + return implode( '|', array_unique( $nsNames ) ); |
| | 242 | + } |
| | 243 | + |
| | 244 | + /** |
| | 245 | + * preg callback to process foo: prefixes in the query |
| | 246 | + * |
| | 247 | + * @param array $matches |
| | 248 | + * @return string |
| | 249 | + */ |
| | 250 | + function replaceQueryPrefix( $matches ) { |
| | 251 | + if ( isset( $this->prefix_handlers[ $matches[ 2 ] ] ) ) { |
| | 252 | + $callback = $this->prefix_handlers[ $matches[ 2 ] ]; |
| | 253 | + return $this->$callback( $matches ); |
| | 254 | + } else { |
| | 255 | + return $this->filterByNamespace( $matches ); |
| | 256 | + } |
| | 257 | + } |
| | 258 | + |
| | 259 | + function filterByNamespace( $matches ) { |
| | 260 | + global $wgContLang; |
| | 261 | + $inx = $wgContLang->getNsIndex( str_replace( ' ', '_', $matches[ 2 ] ) ); |
| | 262 | + if ( $inx === false ) { |
| | 263 | + return $matches[ 0 ]; |
| | 264 | + } else { |
| | 265 | + $this->namespaces[] = $inx; |
| | 266 | + return $matches[ 3 ]; |
| | 267 | + } |
| | 268 | + } |
| | 269 | + |
| | 270 | + function searchAllNamespaces( $matches ) { |
| | 271 | + $this->namespaces = null; |
| | 272 | + return $matches[ 3 ]; |
| | 273 | + } |
| | 274 | + |
| | 275 | + function filterByTitle( $matches ) { |
| | 276 | + return '@page_title ' . $matches[ 3 ]; |
| | 277 | + } |
| | 278 | + |
| | 279 | + function filterByPrefix( $matches ) { |
| | 280 | + $prefix = $matches[ 3 ]; |
| | 281 | + if ( strpos( $matches[ 3 ], ':' ) !== false ) { |
| | 282 | + global $wgContLang; |
| | 283 | + list( $ns, $prefix ) = explode( ':', $matches[ 3 ] ); |
| | 284 | + $inx = $wgContLang->getNsIndex( str_replace( ' ', '_', $ns ) ); |
| | 285 | + if ( $inx !== false ) { |
| | 286 | + $this->namespaces = array( $inx ); |
| | 287 | + } |
| | 288 | + } |
| | 289 | + return '@page_title ^' . $prefix . '*'; |
| | 290 | + } |
| | 291 | + |
| | 292 | + function filterByCategory( $matches ) { |
| | 293 | + $page_id = $this->db->selectField( 'page', 'page_id', |
| | 294 | + array( |
| | 295 | + 'page_title' => $matches[ 3 ], |
| | 296 | + 'page_namespace' => NS_CATEGORY |
| | 297 | + ), |
| | 298 | + __METHOD__ |
| | 299 | + ); |
| | 300 | + $this->categories[] = intval( $page_id ); |
| | 301 | + return ''; |
| | 302 | + } |
| | 303 | + |
| 171 | 304 | } |
| 172 | 305 | |
| 173 | 306 | class SphinxMWSearchResultSet extends SearchResultSet { |