TYPO3 /
TYPO3.CMS
| Total Complexity | 128 |
| Total Lines | 1299 |
| Duplicated Lines | 7.7 % |
| Changes | 0 | ||
Duplicate code is one of the most pungent code smells. A rule that is often used is to re-structure code once it is duplicated in three or more places.
Common duplication problems, and corresponding solutions are:
Complex classes like IndexSearchRepository often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use IndexSearchRepository, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 31 | class IndexSearchRepository |
||
| 32 | { |
||
| 33 | /** |
||
| 34 | * Indexer object |
||
| 35 | * |
||
| 36 | * @var Indexer |
||
| 37 | */ |
||
| 38 | protected $indexerObj; |
||
| 39 | |||
| 40 | /** |
||
| 41 | * External Parsers |
||
| 42 | * |
||
| 43 | * @var array |
||
| 44 | */ |
||
| 45 | protected $externalParsers = []; |
||
| 46 | |||
| 47 | /** |
||
| 48 | * Frontend User Group List |
||
| 49 | * |
||
| 50 | * @var string |
||
| 51 | */ |
||
| 52 | protected $frontendUserGroupList = ''; |
||
| 53 | |||
| 54 | /** |
||
| 55 | * Sections |
||
| 56 | * formally known as $this->piVars['sections'] |
||
| 57 | * |
||
| 58 | * @var string |
||
| 59 | */ |
||
| 60 | protected $sections = null; |
||
| 61 | |||
| 62 | /** |
||
| 63 | * Search type |
||
| 64 | * formally known as $this->piVars['type'] |
||
| 65 | * |
||
| 66 | * @var string |
||
| 67 | */ |
||
| 68 | protected $searchType = null; |
||
| 69 | |||
| 70 | /** |
||
| 71 | * Language uid |
||
| 72 | * formally known as $this->piVars['lang'] |
||
| 73 | * |
||
| 74 | * @var int |
||
| 75 | */ |
||
| 76 | protected $languageUid = null; |
||
| 77 | |||
| 78 | /** |
||
| 79 | * Media type |
||
| 80 | * formally known as $this->piVars['media'] |
||
| 81 | * |
||
| 82 | * @var int |
||
| 83 | */ |
||
| 84 | protected $mediaType = null; |
||
| 85 | |||
| 86 | /** |
||
| 87 | * Sort order |
||
| 88 | * formally known as $this->piVars['sort_order'] |
||
| 89 | * |
||
| 90 | * @var string |
||
| 91 | */ |
||
| 92 | protected $sortOrder = null; |
||
| 93 | |||
| 94 | /** |
||
| 95 | * Descending sort order flag |
||
| 96 | * formally known as $this->piVars['desc'] |
||
| 97 | * |
||
| 98 | * @var bool |
||
| 99 | */ |
||
| 100 | protected $descendingSortOrderFlag = null; |
||
| 101 | |||
| 102 | /** |
||
| 103 | * Result page pointer |
||
| 104 | * formally known as $this->piVars['pointer'] |
||
| 105 | * |
||
| 106 | * @var int |
||
| 107 | */ |
||
| 108 | protected $resultpagePointer = 0; |
||
| 109 | |||
| 110 | /** |
||
| 111 | * Number of results |
||
| 112 | * formally known as $this->piVars['result'] |
||
| 113 | * |
||
| 114 | * @var int |
||
| 115 | */ |
||
| 116 | protected $numberOfResults = 10; |
||
| 117 | |||
| 118 | /** |
||
| 119 | * list of all root pages that will be used |
||
| 120 | * If this value is set to less than zero (eg. -1) searching will happen |
||
| 121 | * in ALL of the page tree with no regard to branches at all. |
||
| 122 | * |
||
| 123 | * @var string |
||
| 124 | */ |
||
| 125 | protected $searchRootPageIdList; |
||
| 126 | |||
| 127 | /** |
||
| 128 | * formally known as $conf['search.']['searchSkipExtendToSubpagesChecking'] |
||
| 129 | * enabled through settings.searchSkipExtendToSubpagesChecking |
||
| 130 | * |
||
| 131 | * @var bool |
||
| 132 | */ |
||
| 133 | protected $joinPagesForQuery = false; |
||
| 134 | |||
| 135 | /** |
||
| 136 | * Select clauses for individual words, will be filled during the search |
||
| 137 | * |
||
| 138 | * @var array |
||
| 139 | */ |
||
| 140 | protected $wSelClauses = []; |
||
| 141 | |||
| 142 | /** |
||
| 143 | * Flag for exact search count |
||
| 144 | * formally known as $conf['search.']['exactCount'] |
||
| 145 | * |
||
| 146 | * Continue counting and checking of results even if we are sure |
||
| 147 | * they are not displayed in this request. This will slow down your |
||
| 148 | * page rendering, but it allows precise search result counters. |
||
| 149 | * enabled through settings.exactCount |
||
| 150 | * |
||
| 151 | * @var bool |
||
| 152 | */ |
||
| 153 | protected $useExactCount = false; |
||
| 154 | |||
| 155 | /** |
||
| 156 | * Display forbidden records |
||
| 157 | * formally known as $this->conf['show.']['forbiddenRecords'] |
||
| 158 | * |
||
| 159 | * enabled through settings.displayForbiddenRecords |
||
| 160 | * |
||
| 161 | * @var bool |
||
| 162 | */ |
||
| 163 | protected $displayForbiddenRecords = false; |
||
| 164 | |||
| 165 | /** |
||
| 166 | * initialize all options that are necessary for the search |
||
| 167 | * |
||
| 168 | * @param array $settings the extbase plugin settings |
||
| 169 | * @param array $searchData the search data |
||
| 170 | * @param array $externalParsers |
||
| 171 | * @param string $searchRootPageIdList |
||
| 172 | */ |
||
| 173 | public function initialize($settings, $searchData, $externalParsers, $searchRootPageIdList) |
||
| 174 | { |
||
| 175 | // Initialize the indexer-class - just to use a few function (for making hashes) |
||
| 176 | $this->indexerObj = GeneralUtility::makeInstance(Indexer::class); |
||
| 177 | $this->externalParsers = $externalParsers; |
||
| 178 | $this->searchRootPageIdList = $searchRootPageIdList; |
||
| 179 | $this->frontendUserGroupList = $this->getTypoScriptFrontendController()->gr_list; |
||
| 180 | // Should we use joinPagesForQuery instead of long lists of uids? |
||
| 181 | if ($settings['searchSkipExtendToSubpagesChecking']) { |
||
| 182 | $this->joinPagesForQuery = 1; |
||
|
|
|||
| 183 | } |
||
| 184 | if ($settings['exactCount']) { |
||
| 185 | $this->useExactCount = true; |
||
| 186 | } |
||
| 187 | if ($settings['displayForbiddenRecords']) { |
||
| 188 | $this->displayForbiddenRecords = true; |
||
| 189 | } |
||
| 190 | $this->sections = $searchData['sections']; |
||
| 191 | $this->searchType = $searchData['searchType']; |
||
| 192 | $this->languageUid = $searchData['languageUid']; |
||
| 193 | $this->mediaType = isset($searchData['mediaType']) ? $searchData['mediaType'] : false; |
||
| 194 | $this->sortOrder = $searchData['sortOrder']; |
||
| 195 | $this->descendingSortOrderFlag = $searchData['desc']; |
||
| 196 | $this->resultpagePointer = $searchData['pointer']; |
||
| 197 | if (isset($searchData['numberOfResults']) && is_numeric($searchData['numberOfResults'])) { |
||
| 198 | $this->numberOfResults = (int)$searchData['numberOfResults']; |
||
| 199 | } |
||
| 200 | } |
||
| 201 | |||
| 202 | /** |
||
| 203 | * Get search result rows / data from database. Returned as data in array. |
||
| 204 | * |
||
| 205 | * @param array $searchWords Search word array |
||
| 206 | * @param int $freeIndexUid Pointer to which indexing configuration you want to search in. -1 means no filtering. 0 means only regular indexed content. |
||
| 207 | * @return bool|array FALSE if no result, otherwise an array with keys for first row, result rows and total number of results found. |
||
| 208 | */ |
||
| 209 | public function doSearch($searchWords, $freeIndexUid = -1) |
||
| 210 | { |
||
| 211 | // unserializing the configuration so we can use it here: |
||
| 212 | $extConf = []; |
||
| 213 | View Code Duplication | if (isset($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search'])) { |
|
| 214 | $extConf = unserialize( |
||
| 215 | $GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search'], |
||
| 216 | ['allowed_classes' => false] |
||
| 217 | ); |
||
| 218 | } |
||
| 219 | |||
| 220 | // Getting SQL result pointer: |
||
| 221 | $this->getTimeTracker()->push('Searching result'); |
||
| 222 | if ($hookObj = &$this->hookRequest('getResultRows_SQLpointer')) { |
||
| 223 | $result = $hookObj->getResultRows_SQLpointer($searchWords, $freeIndexUid); |
||
| 224 | } elseif (isset($extConf['useMysqlFulltext']) && $extConf['useMysqlFulltext'] === '1') { |
||
| 225 | $result = $this->getResultRows_SQLpointerMysqlFulltext($searchWords, $freeIndexUid); |
||
| 226 | } else { |
||
| 227 | $result = $this->getResultRows_SQLpointer($searchWords, $freeIndexUid); |
||
| 228 | } |
||
| 229 | $this->getTimeTracker()->pull(); |
||
| 230 | // Organize and process result: |
||
| 231 | if ($result) { |
||
| 232 | // Total search-result count |
||
| 233 | $count = $result->rowCount(); |
||
| 234 | // The pointer is set to the result page that is currently being viewed |
||
| 235 | $pointer = MathUtility::forceIntegerInRange($this->resultpagePointer, 0, floor($count / $this->numberOfResults)); |
||
| 236 | // Initialize result accumulation variables: |
||
| 237 | $c = 0; |
||
| 238 | // Result pointer: Counts up the position in the current search-result |
||
| 239 | $grouping_phashes = []; |
||
| 240 | // Used to filter out duplicates. |
||
| 241 | $grouping_chashes = []; |
||
| 242 | // Used to filter out duplicates BASED ON cHash. |
||
| 243 | $firstRow = []; |
||
| 244 | // Will hold the first row in result - used to calculate relative hit-ratings. |
||
| 245 | $resultRows = []; |
||
| 246 | // Will hold the results rows for display. |
||
| 247 | // Now, traverse result and put the rows to be displayed into an array |
||
| 248 | // Each row should contain the fields from 'ISEC.*, IP.*' combined |
||
| 249 | // + artificial fields "show_resume" (bool) and "result_number" (counter) |
||
| 250 | while ($row = $result->fetch()) { |
||
| 251 | // Set first row |
||
| 252 | if (!$c) { |
||
| 253 | $firstRow = $row; |
||
| 254 | } |
||
| 255 | // Tells whether we can link directly to a document |
||
| 256 | // or not (depends on possible right problems) |
||
| 257 | $row['show_resume'] = $this->checkResume($row); |
||
| 258 | $phashGr = !in_array($row['phash_grouping'], $grouping_phashes); |
||
| 259 | $chashGr = !in_array(($row['contentHash'] . '.' . $row['data_page_id']), $grouping_chashes); |
||
| 260 | if ($phashGr && $chashGr) { |
||
| 261 | // Only if the resume may be shown are we going to filter out duplicates... |
||
| 262 | if ($row['show_resume'] || $this->displayForbiddenRecords) { |
||
| 263 | // Only on documents which are not multiple pages documents |
||
| 264 | if (!$this->multiplePagesType($row['item_type'])) { |
||
| 265 | $grouping_phashes[] = $row['phash_grouping']; |
||
| 266 | } |
||
| 267 | $grouping_chashes[] = $row['contentHash'] . '.' . $row['data_page_id']; |
||
| 268 | // Increase the result pointer |
||
| 269 | $c++; |
||
| 270 | // All rows for display is put into resultRows[] |
||
| 271 | if ($c > $pointer * $this->numberOfResults && $c <= $pointer * $this->numberOfResults + $this->numberOfResults) { |
||
| 272 | $row['result_number'] = $c; |
||
| 273 | $resultRows[] = $row; |
||
| 274 | // This may lead to a problem: If the result check is not stopped here, the search will take longer. |
||
| 275 | // However the result counter will not filter out grouped cHashes/pHashes that were not processed yet. |
||
| 276 | // You can change this behavior using the "search.exactCount" property (see above). |
||
| 277 | if (!$this->useExactCount && $c + 1 > ($pointer + 1) * $this->numberOfResults) { |
||
| 278 | break; |
||
| 279 | } |
||
| 280 | } |
||
| 281 | } else { |
||
| 282 | // Skip this row if the user cannot |
||
| 283 | // view it (missing permission) |
||
| 284 | $count--; |
||
| 285 | } |
||
| 286 | } else { |
||
| 287 | // For each time a phash_grouping document is found |
||
| 288 | // (which is thus not displayed) the search-result count is reduced, |
||
| 289 | // so that it matches the number of rows displayed. |
||
| 290 | $count--; |
||
| 291 | } |
||
| 292 | } |
||
| 293 | |||
| 294 | $result->closeCursor(); |
||
| 295 | |||
| 296 | return [ |
||
| 297 | 'resultRows' => $resultRows, |
||
| 298 | 'firstRow' => $firstRow, |
||
| 299 | 'count' => $count |
||
| 300 | ]; |
||
| 301 | } |
||
| 302 | // No results found |
||
| 303 | return false; |
||
| 304 | } |
||
| 305 | |||
| 306 | /** |
||
| 307 | * Gets a SQL result pointer to traverse for the search records. |
||
| 308 | * |
||
| 309 | * @param array $searchWords Search words |
||
| 310 | * @param int $freeIndexUid Pointer to which indexing configuration you want to search in. -1 means no filtering. 0 means only regular indexed content. |
||
| 311 | * @return Statement |
||
| 312 | */ |
||
| 313 | protected function getResultRows_SQLpointer($searchWords, $freeIndexUid = -1) |
||
| 314 | { |
||
| 315 | // This SEARCHES for the searchwords in $searchWords AND returns a |
||
| 316 | // COMPLETE list of phash-integers of the matches. |
||
| 317 | $list = $this->getPhashList($searchWords); |
||
| 318 | // Perform SQL Search / collection of result rows array: |
||
| 319 | if ($list) { |
||
| 320 | // Do the search: |
||
| 321 | $this->getTimeTracker()->push('execFinalQuery'); |
||
| 322 | $res = $this->execFinalQuery($list, $freeIndexUid); |
||
| 323 | $this->getTimeTracker()->pull(); |
||
| 324 | return $res; |
||
| 325 | } |
||
| 326 | return false; |
||
| 327 | } |
||
| 328 | |||
| 329 | /** |
||
| 330 | * Gets a SQL result pointer to traverse for the search records. |
||
| 331 | * |
||
| 332 | * mysql fulltext specific version triggered by ext_conf_template setting 'useMysqlFulltext' |
||
| 333 | * |
||
| 334 | * @param array $searchWordsArray Search words |
||
| 335 | * @param int $freeIndexUid Pointer to which indexing configuration you want to search in. -1 means no filtering. 0 means only regular indexed content. |
||
| 336 | * @return bool|\mysqli_result|object MySQLi result object / DBAL object |
||
| 337 | */ |
||
| 338 | protected function getResultRows_SQLpointerMysqlFulltext($searchWordsArray, $freeIndexUid = -1) |
||
| 339 | { |
||
| 340 | $connection = GeneralUtility::makeInstance(ConnectionPool::class)->getConnectionForTable('index_fulltext'); |
||
| 341 | if (strpos($connection->getServerVersion(), 'MySQL') !== 0) { |
||
| 342 | throw new \RuntimeException( |
||
| 343 | 'Extension indexed_search is configured to use mysql fulltext, but table \'index_fulltext\'' |
||
| 344 | . ' is running on a different DBMS.', |
||
| 345 | 1472585525 |
||
| 346 | ); |
||
| 347 | } |
||
| 348 | // Build the search string, detect which fulltext index to use, and decide whether boolean search is needed or not |
||
| 349 | $searchData = $this->getSearchString($searchWordsArray); |
||
| 350 | // Perform SQL Search / collection of result rows array: |
||
| 351 | $resource = false; |
||
| 352 | if ($searchData) { |
||
| 353 | /** @var TimeTracker $timeTracker */ |
||
| 354 | $timeTracker = GeneralUtility::makeInstance(TimeTracker::class); |
||
| 355 | // Do the search: |
||
| 356 | $timeTracker->push('execFinalQuery'); |
||
| 357 | $resource = $this->execFinalQuery_fulltext($searchData, $freeIndexUid); |
||
| 358 | $timeTracker->pull(); |
||
| 359 | } |
||
| 360 | return $resource; |
||
| 361 | } |
||
| 362 | |||
| 363 | /** |
||
| 364 | * Returns a search string for use with MySQL FULLTEXT query |
||
| 365 | * |
||
| 366 | * mysql fulltext specific helper method |
||
| 367 | * |
||
| 368 | * @param array $searchWordArray Search word array |
||
| 369 | * @return string Search string |
||
| 370 | */ |
||
| 371 | protected function getSearchString($searchWordArray) |
||
| 372 | { |
||
| 373 | // Initialize variables: |
||
| 374 | $count = 0; |
||
| 375 | // Change this to TRUE to force BOOLEAN SEARCH MODE (useful if fulltext index is still empty) |
||
| 376 | $searchBoolean = false; |
||
| 377 | $fulltextIndex = 'index_fulltext.fulltextdata'; |
||
| 378 | // This holds the result if the search is natural (doesn't contain any boolean operators) |
||
| 379 | $naturalSearchString = ''; |
||
| 380 | // This holds the result if the search is boolen (contains +/-/| operators) |
||
| 381 | $booleanSearchString = ''; |
||
| 382 | |||
| 383 | $searchType = (string)$this->getSearchType(); |
||
| 384 | |||
| 385 | // Traverse searchwords and prefix them with corresponding operator |
||
| 386 | foreach ($searchWordArray as $searchWordData) { |
||
| 387 | // Making the query for a single search word based on the search-type |
||
| 388 | $searchWord = $searchWordData['sword']; |
||
| 389 | $wildcard = ''; |
||
| 390 | if (strstr($searchWord, ' ')) { |
||
| 391 | $searchType = '20'; |
||
| 392 | } |
||
| 393 | switch ($searchType) { |
||
| 394 | case '1': |
||
| 395 | case '2': |
||
| 396 | case '3': |
||
| 397 | // First part of word |
||
| 398 | $wildcard = '*'; |
||
| 399 | // Part-of-word search requires boolean mode! |
||
| 400 | $searchBoolean = true; |
||
| 401 | break; |
||
| 402 | case '10': |
||
| 403 | $indexerObj = GeneralUtility::makeInstance(Indexer::class); |
||
| 404 | // Initialize the indexer-class |
||
| 405 | /** @var Indexer $indexerObj */ |
||
| 406 | $searchWord = $indexerObj->metaphone($searchWord, $indexerObj->storeMetaphoneInfoAsWords); |
||
| 407 | unset($indexerObj); |
||
| 408 | $fulltextIndex = 'index_fulltext.metaphonedata'; |
||
| 409 | break; |
||
| 410 | case '20': |
||
| 411 | $searchBoolean = true; |
||
| 412 | // Remove existing quotes and fix misplaced quotes. |
||
| 413 | $searchWord = trim(str_replace('"', ' ', $searchWord)); |
||
| 414 | break; |
||
| 415 | } |
||
| 416 | // Perform search for word: |
||
| 417 | switch ($searchWordData['oper']) { |
||
| 418 | case 'AND NOT': |
||
| 419 | $booleanSearchString .= ' -' . $searchWord . $wildcard; |
||
| 420 | $searchBoolean = true; |
||
| 421 | break; |
||
| 422 | case 'OR': |
||
| 423 | $booleanSearchString .= ' ' . $searchWord . $wildcard; |
||
| 424 | $searchBoolean = true; |
||
| 425 | break; |
||
| 426 | default: |
||
| 427 | $booleanSearchString .= ' +' . $searchWord . $wildcard; |
||
| 428 | $naturalSearchString .= ' ' . $searchWord; |
||
| 429 | } |
||
| 430 | $count++; |
||
| 431 | } |
||
| 432 | if ($searchType == '20') { |
||
| 433 | $searchString = '"' . trim($naturalSearchString) . '"'; |
||
| 434 | } elseif ($searchBoolean) { |
||
| 435 | $searchString = trim($booleanSearchString); |
||
| 436 | } else { |
||
| 437 | $searchString = trim($naturalSearchString); |
||
| 438 | } |
||
| 439 | return [ |
||
| 440 | 'searchBoolean' => $searchBoolean, |
||
| 441 | 'searchString' => $searchString, |
||
| 442 | 'fulltextIndex' => $fulltextIndex |
||
| 443 | ]; |
||
| 444 | } |
||
| 445 | |||
| 446 | /** |
||
| 447 | * Execute final query, based on phash integer list. The main point is sorting the result in the right order. |
||
| 448 | * |
||
| 449 | * mysql fulltext specific helper method |
||
| 450 | * |
||
| 451 | * @param array $searchData Array with search string, boolean indicator, and fulltext index reference |
||
| 452 | * @param int $freeIndexUid Pointer to which indexing configuration you want to search in. -1 means no filtering. 0 means only regular indexed content. |
||
| 453 | * @return Statement |
||
| 454 | */ |
||
| 455 | protected function execFinalQuery_fulltext($searchData, $freeIndexUid = -1) |
||
| 456 | { |
||
| 457 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('index_fulltext'); |
||
| 458 | $queryBuilder->getRestrictions()->removeAll(); |
||
| 459 | $queryBuilder->select('index_fulltext.*', 'ISEC.*', 'IP.*') |
||
| 460 | ->from('index_fulltext') |
||
| 461 | ->join( |
||
| 462 | 'index_fulltext', |
||
| 463 | 'index_phash', |
||
| 464 | 'IP', |
||
| 465 | $queryBuilder->expr()->eq('index_fulltext.phash', $queryBuilder->quoteIdentifier('IP.phash')) |
||
| 466 | ) |
||
| 467 | ->join( |
||
| 468 | 'IP', |
||
| 469 | 'index_section', |
||
| 470 | 'ISEC', |
||
| 471 | $queryBuilder->expr()->eq('IP.phash', $queryBuilder->quoteIdentifier('ISEC.phash')) |
||
| 472 | ); |
||
| 473 | |||
| 474 | // Calling hook for alternative creation of page ID list |
||
| 475 | $searchRootPageIdList = $this->getSearchRootPageIdList(); |
||
| 476 | if ($hookObj = &$this->hookRequest('execFinalQuery_idList')) { |
||
| 477 | $pageWhere = $hookObj->execFinalQuery_idList(''); |
||
| 478 | $queryBuilder->andWhere(QueryHelper::stripLogicalOperatorPrefix($pageWhere)); |
||
| 479 | } elseif ($this->joinPagesForQuery) { |
||
| 480 | // Alternative to getting all page ids by ->getTreeList() where "excludeSubpages" is NOT respected. |
||
| 481 | $queryBuilder |
||
| 482 | ->join( |
||
| 483 | 'ISEC', |
||
| 484 | 'pages', |
||
| 485 | 'pages', |
||
| 486 | $queryBuilder->expr()->eq('ISEC.page_id', $queryBuilder->quoteIdentifier('pages.uid')) |
||
| 487 | ) |
||
| 488 | ->andWhere( |
||
| 489 | $queryBuilder->expr()->eq( |
||
| 490 | 'pages.no_search', |
||
| 491 | $queryBuilder->createNamedParameter(0, \PDO::PARAM_INT) |
||
| 492 | ) |
||
| 493 | ) |
||
| 494 | ->andWhere( |
||
| 495 | $queryBuilder->expr()->lt( |
||
| 496 | 'pages.doktype', |
||
| 497 | $queryBuilder->createNamedParameter(200, \PDO::PARAM_INT) |
||
| 498 | ) |
||
| 499 | ); |
||
| 500 | $queryBuilder->setRestrictions(GeneralUtility::makeInstance(FrontendRestrictionContainer::class)); |
||
| 501 | } elseif ($searchRootPageIdList[0] >= 0) { |
||
| 502 | // Collecting all pages IDs in which to search; |
||
| 503 | // filtering out ALL pages that are not accessible due to restriction containers. Does NOT look for "no_search" field! |
||
| 504 | $idList = []; |
||
| 505 | foreach ($searchRootPageIdList as $rootId) { |
||
| 506 | /** @var \TYPO3\CMS\Frontend\ContentObject\ContentObjectRenderer $cObj */ |
||
| 507 | $cObj = GeneralUtility::makeInstance(\TYPO3\CMS\Frontend\ContentObject\ContentObjectRenderer::class); |
||
| 508 | $idList[] = $cObj->getTreeList(-1 * $rootId, 9999); |
||
| 509 | } |
||
| 510 | $idList = GeneralUtility::intExplode(',', implode(',', $idList)); |
||
| 511 | $queryBuilder->andWhere( |
||
| 512 | $queryBuilder->expr()->in( |
||
| 513 | 'ISEC.page_id', |
||
| 514 | $queryBuilder->createNamedParameter($idList, Connection::PARAM_INT_ARRAY) |
||
| 515 | ) |
||
| 516 | ); |
||
| 517 | } |
||
| 518 | |||
| 519 | $searchBoolean = ''; |
||
| 520 | if ($searchData['searchBoolean']) { |
||
| 521 | $searchBoolean = ' IN BOOLEAN MODE'; |
||
| 522 | } |
||
| 523 | $queryBuilder->andWhere( |
||
| 524 | 'MATCH (' . $queryBuilder->quoteIdentifier($searchData['fulltextIndex']) . ')' |
||
| 525 | . ' AGAINST (' . $queryBuilder->createNamedParameter($searchData['searchString']) |
||
| 526 | . $searchBoolean |
||
| 527 | . ')' |
||
| 528 | ); |
||
| 529 | |||
| 530 | $queryBuilder->andWhere( |
||
| 531 | QueryHelper::stripLogicalOperatorPrefix($this->mediaTypeWhere()), |
||
| 532 | QueryHelper::stripLogicalOperatorPrefix($this->languageWhere()), |
||
| 533 | QueryHelper::stripLogicalOperatorPrefix($this->freeIndexUidWhere($freeIndexUid)), |
||
| 534 | QueryHelper::stripLogicalOperatorPrefix($this->sectionTableWhere()) |
||
| 535 | ); |
||
| 536 | |||
| 537 | $queryBuilder->groupBy( |
||
| 538 | 'IP.phash', |
||
| 539 | 'ISEC.phash', |
||
| 540 | 'ISEC.phash_t3', |
||
| 541 | 'ISEC.rl0', |
||
| 542 | 'ISEC.rl1', |
||
| 543 | 'ISEC.rl2', |
||
| 544 | 'ISEC.page_id', |
||
| 545 | 'ISEC.uniqid', |
||
| 546 | 'IP.phash_grouping', |
||
| 547 | 'IP.data_filename', |
||
| 548 | 'IP.data_page_id', |
||
| 549 | 'IP.data_page_reg1', |
||
| 550 | 'IP.data_page_type', |
||
| 551 | 'IP.data_page_mp', |
||
| 552 | 'IP.gr_list', |
||
| 553 | 'IP.item_type', |
||
| 554 | 'IP.item_title', |
||
| 555 | 'IP.item_description', |
||
| 556 | 'IP.item_mtime', |
||
| 557 | 'IP.tstamp', |
||
| 558 | 'IP.item_size', |
||
| 559 | 'IP.contentHash', |
||
| 560 | 'IP.crdate', |
||
| 561 | 'IP.parsetime', |
||
| 562 | 'IP.sys_language_uid', |
||
| 563 | 'IP.item_crdate', |
||
| 564 | 'IP.cHashParams', |
||
| 565 | 'IP.externalUrl', |
||
| 566 | 'IP.recordUid', |
||
| 567 | 'IP.freeIndexUid', |
||
| 568 | 'IP.freeIndexSetId' |
||
| 569 | ); |
||
| 570 | |||
| 571 | return $queryBuilder->execute(); |
||
| 572 | } |
||
| 573 | |||
| 574 | /*********************************** |
||
| 575 | * |
||
| 576 | * Helper functions on searching (SQL) |
||
| 577 | * |
||
| 578 | ***********************************/ |
||
| 579 | /** |
||
| 580 | * Returns a COMPLETE list of phash-integers matching the search-result composed of the search-words in the $searchWords array. |
||
| 581 | * The list of phash integers are unsorted and should be used for subsequent selection of index_phash records for display of the result. |
||
| 582 | * |
||
| 583 | * @param array $searchWords Search word array |
||
| 584 | * @return string List of integers |
||
| 585 | */ |
||
| 586 | protected function getPhashList($searchWords) |
||
| 587 | { |
||
| 588 | // Initialize variables: |
||
| 589 | $c = 0; |
||
| 590 | // This array accumulates the phash-values |
||
| 591 | $totalHashList = []; |
||
| 592 | $this->wSelClauses = []; |
||
| 593 | // Traverse searchwords; for each, select all phash integers and merge/diff/intersect them with previous word (based on operator) |
||
| 594 | foreach ($searchWords as $k => $v) { |
||
| 595 | // Making the query for a single search word based on the search-type |
||
| 596 | $sWord = $v['sword']; |
||
| 597 | $theType = (string)$this->searchType; |
||
| 598 | // If there are spaces in the search-word, make a full text search instead. |
||
| 599 | if (strstr($sWord, ' ')) { |
||
| 600 | $theType = 20; |
||
| 601 | } |
||
| 602 | $this->getTimeTracker()->push('SearchWord "' . $sWord . '" - $theType=' . $theType); |
||
| 603 | // Perform search for word: |
||
| 604 | switch ($theType) { |
||
| 605 | case '1': |
||
| 606 | // Part of word |
||
| 607 | $res = $this->searchWord($sWord, Utility\LikeWildcard::BOTH); |
||
| 608 | break; |
||
| 609 | case '2': |
||
| 610 | // First part of word |
||
| 611 | $res = $this->searchWord($sWord, Utility\LikeWildcard::RIGHT); |
||
| 612 | break; |
||
| 613 | case '3': |
||
| 614 | // Last part of word |
||
| 615 | $res = $this->searchWord($sWord, Utility\LikeWildcard::LEFT); |
||
| 616 | break; |
||
| 617 | case '10': |
||
| 618 | // Sounds like |
||
| 619 | /** |
||
| 620 | * Indexer object |
||
| 621 | * |
||
| 622 | * @var Indexer |
||
| 623 | */ |
||
| 624 | $indexerObj = GeneralUtility::makeInstance(Indexer::class); |
||
| 625 | // Perform metaphone search |
||
| 626 | $storeMetaphoneInfoAsWords = !$this->isTableUsed('index_words'); |
||
| 627 | $res = $this->searchMetaphone($indexerObj->metaphone($sWord, $storeMetaphoneInfoAsWords)); |
||
| 628 | unset($indexerObj); |
||
| 629 | break; |
||
| 630 | case '20': |
||
| 631 | // Sentence |
||
| 632 | $res = $this->searchSentence($sWord); |
||
| 633 | // If there is a fulltext search for a sentence there is |
||
| 634 | // a likeliness that sorting cannot be done by the rankings |
||
| 635 | // from the rel-table (because no relations will exist for the |
||
| 636 | // sentence in the word-table). So therefore mtime is used instead. |
||
| 637 | // It is not required, but otherwise some hits may be left out. |
||
| 638 | $this->sortOrder = 'mtime'; |
||
| 639 | break; |
||
| 640 | default: |
||
| 641 | // Distinct word |
||
| 642 | $res = $this->searchDistinct($sWord); |
||
| 643 | } |
||
| 644 | // If there was a query to do, then select all phash-integers which resulted from this. |
||
| 645 | if ($res) { |
||
| 646 | // Get phash list by searching for it: |
||
| 647 | $phashList = []; |
||
| 648 | while ($row = $res->fetch()) { |
||
| 649 | $phashList[] = $row['phash']; |
||
| 650 | } |
||
| 651 | // Here the phash list are merged with the existing result based on whether we are dealing with OR, NOT or AND operations. |
||
| 652 | if ($c) { |
||
| 653 | switch ($v['oper']) { |
||
| 654 | case 'OR': |
||
| 655 | $totalHashList = array_unique(array_merge($phashList, $totalHashList)); |
||
| 656 | break; |
||
| 657 | case 'AND NOT': |
||
| 658 | $totalHashList = array_diff($totalHashList, $phashList); |
||
| 659 | break; |
||
| 660 | default: |
||
| 661 | // AND... |
||
| 662 | $totalHashList = array_intersect($totalHashList, $phashList); |
||
| 663 | } |
||
| 664 | } else { |
||
| 665 | // First search |
||
| 666 | $totalHashList = $phashList; |
||
| 667 | } |
||
| 668 | } |
||
| 669 | $this->getTimeTracker()->pull(); |
||
| 670 | $c++; |
||
| 671 | } |
||
| 672 | return implode(',', $totalHashList); |
||
| 673 | } |
||
| 674 | |||
| 675 | /** |
||
| 676 | * Returns a query which selects the search-word from the word/rel tables. |
||
| 677 | * |
||
| 678 | * @param string $wordSel WHERE clause selecting the word from phash |
||
| 679 | * @param string $additionalWhereClause Additional AND clause in the end of the query. |
||
| 680 | * @return Statement |
||
| 681 | */ |
||
| 682 | protected function execPHashListQuery($wordSel, $additionalWhereClause = '') |
||
| 683 | { |
||
| 684 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('index_words'); |
||
| 685 | $queryBuilder->select('IR.phash') |
||
| 686 | ->from('index_words', 'IW') |
||
| 687 | ->from('index_rel', 'IR') |
||
| 688 | ->from('index_section', 'ISEC') |
||
| 689 | ->where( |
||
| 690 | QueryHelper::stripLogicalOperatorPrefix($wordSel), |
||
| 691 | $queryBuilder->expr()->eq('IW.wid', $queryBuilder->quoteIdentifier('IR.wid')), |
||
| 692 | $queryBuilder->expr()->eq('ISEC.phash', $queryBuilder->quoteIdentifier('IR.phash')), |
||
| 693 | QueryHelper::stripLogicalOperatorPrefix($this->sectionTableWhere()), |
||
| 694 | QueryHelper::stripLogicalOperatorPrefix($additionalWhereClause) |
||
| 695 | ) |
||
| 696 | ->groupBy('IR.phash'); |
||
| 697 | |||
| 698 | return $queryBuilder->execute(); |
||
| 699 | } |
||
| 700 | |||
| 701 | /** |
||
| 702 | * Search for a word |
||
| 703 | * |
||
| 704 | * @param string $sWord the search word |
||
| 705 | * @param int $wildcard Bit-field of Utility\LikeWildcard |
||
| 706 | * @return Statement |
||
| 707 | */ |
||
| 708 | protected function searchWord($sWord, $wildcard) |
||
| 709 | { |
||
| 710 | $likeWildcard = Utility\LikeWildcard::cast($wildcard); |
||
| 711 | $wSel = $likeWildcard->getLikeQueryPart( |
||
| 712 | 'index_words', |
||
| 713 | 'IW.baseword', |
||
| 714 | $sWord |
||
| 715 | ); |
||
| 716 | $this->wSelClauses[] = $wSel; |
||
| 717 | return $this->execPHashListQuery($wSel, ' AND is_stopword=0'); |
||
| 718 | } |
||
| 719 | |||
| 720 | /** |
||
| 721 | * Search for one distinct word |
||
| 722 | * |
||
| 723 | * @param string $sWord the search word |
||
| 724 | * @return Statement |
||
| 725 | */ |
||
| 726 | View Code Duplication | protected function searchDistinct($sWord) |
|
| 727 | { |
||
| 728 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 729 | ->getQueryBuilderForTable('index_words') |
||
| 730 | ->expr(); |
||
| 731 | $wSel = $expressionBuilder->eq('IW.wid', $this->md5inthash($sWord)); |
||
| 732 | $this->wSelClauses[] = $wSel; |
||
| 733 | return $this->execPHashListQuery($wSel, $expressionBuilder->eq('is_stopword', 0)); |
||
| 734 | } |
||
| 735 | |||
| 736 | /** |
||
| 737 | * Search for a sentence |
||
| 738 | * |
||
| 739 | * @param string $sWord the search word |
||
| 740 | * @return Statement |
||
| 741 | */ |
||
| 742 | protected function searchSentence($sWord) |
||
| 743 | { |
||
| 744 | $this->wSelClauses[] = '1=1'; |
||
| 745 | $likeWildcard = Utility\LikeWildcard::cast(Utility\LikeWildcard::BOTH); |
||
| 746 | $likePart = $likeWildcard->getLikeQueryPart( |
||
| 747 | 'index_fulltext', |
||
| 748 | 'IFT.fulltextdata', |
||
| 749 | $sWord |
||
| 750 | ); |
||
| 751 | |||
| 752 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('index_section'); |
||
| 753 | return $queryBuilder->select('ISEC.phash') |
||
| 754 | ->from('index_section', 'ISEC') |
||
| 755 | ->from('index_fulltext', 'IFT') |
||
| 756 | ->where( |
||
| 757 | QueryHelper::stripLogicalOperatorPrefix($likePart), |
||
| 758 | $queryBuilder->expr()->eq('ISEC.phash', $queryBuilder->quoteIdentifier(('IFT.phash'))), |
||
| 759 | QueryHelper::stripLogicalOperatorPrefix($this->sectionTableWhere()) |
||
| 760 | ) |
||
| 761 | ->groupBy('ISEC.phash') |
||
| 762 | ->execute(); |
||
| 763 | } |
||
| 764 | |||
| 765 | /** |
||
| 766 | * Search for a metaphone word |
||
| 767 | * |
||
| 768 | * @param string $sWord the search word |
||
| 769 | * @return Statement |
||
| 770 | */ |
||
| 771 | View Code Duplication | protected function searchMetaphone($sWord) |
|
| 772 | { |
||
| 773 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 774 | ->getQueryBuilderForTable('index_words') |
||
| 775 | ->expr(); |
||
| 776 | $wSel = $expressionBuilder->eq('IW.metaphone', $expressionBuilder->literal($sWord)); |
||
| 777 | $this->wSelClauses[] = $wSel; |
||
| 778 | return $this->execPHashListQuery($wSel, $expressionBuilder->eq('is_stopword', 0)); |
||
| 779 | } |
||
| 780 | |||
| 781 | /** |
||
| 782 | * Returns AND statement for selection of section in database. (rootlevel 0-2 + page_id) |
||
| 783 | * |
||
| 784 | * @return string AND clause for selection of section in database. |
||
| 785 | */ |
||
| 786 | public function sectionTableWhere() |
||
| 787 | { |
||
| 788 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 789 | ->getQueryBuilderForTable('index_section') |
||
| 790 | ->expr(); |
||
| 791 | |||
| 792 | $whereClause = $expressionBuilder->andX(); |
||
| 793 | $match = false; |
||
| 794 | if (!($this->searchRootPageIdList < 0)) { |
||
| 795 | $whereClause->add( |
||
| 796 | $expressionBuilder->in('ISEC.rl0', GeneralUtility::intExplode(',', $this->searchRootPageIdList, true)) |
||
| 797 | ); |
||
| 798 | } |
||
| 799 | if (substr($this->sections, 0, 4) === 'rl1_') { |
||
| 800 | $whereClause->add( |
||
| 801 | $expressionBuilder->in('ISEC.rl1', GeneralUtility::intExplode(',', substr($this->sections, 4))) |
||
| 802 | ); |
||
| 803 | $match = true; |
||
| 804 | } elseif (substr($this->sections, 0, 4) === 'rl2_') { |
||
| 805 | $whereClause->add( |
||
| 806 | $expressionBuilder->in('ISEC.rl2', GeneralUtility::intExplode(',', substr($this->sections, 4))) |
||
| 807 | ); |
||
| 808 | $match = true; |
||
| 809 | } else { |
||
| 810 | // Traversing user configured fields to see if any of those are used to limit search to a section: |
||
| 811 | foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['addRootLineFields'] ?? [] as $fieldName => $rootLineLevel) { |
||
| 812 | if (substr($this->sections, 0, strlen($fieldName) + 1) == $fieldName . '_') { |
||
| 813 | $whereClause->add( |
||
| 814 | $expressionBuilder->in( |
||
| 815 | 'ISEC.' . $fieldName, |
||
| 816 | GeneralUtility::intExplode(',', substr($this->sections, strlen($fieldName) + 1)) |
||
| 817 | ) |
||
| 818 | ); |
||
| 819 | $match = true; |
||
| 820 | break; |
||
| 821 | } |
||
| 822 | } |
||
| 823 | } |
||
| 824 | // If no match above, test the static types: |
||
| 825 | if (!$match) { |
||
| 826 | switch ((string)$this->sections) { |
||
| 827 | case '-1': |
||
| 828 | $whereClause->add( |
||
| 829 | $expressionBuilder->eq('ISEC.page_id', (int)$this->getTypoScriptFrontendController()->id) |
||
| 830 | ); |
||
| 831 | break; |
||
| 832 | case '-2': |
||
| 833 | $whereClause->add($expressionBuilder->eq('ISEC.rl2', 0)); |
||
| 834 | break; |
||
| 835 | case '-3': |
||
| 836 | $whereClause->add($expressionBuilder->gt('ISEC.rl2', 0)); |
||
| 837 | break; |
||
| 838 | } |
||
| 839 | } |
||
| 840 | |||
| 841 | return $whereClause->count() ? ' AND ' . $whereClause : ''; |
||
| 842 | } |
||
| 843 | |||
| 844 | /** |
||
| 845 | * Returns AND statement for selection of media type |
||
| 846 | * |
||
| 847 | * @return string AND statement for selection of media type |
||
| 848 | */ |
||
| 849 | public function mediaTypeWhere() |
||
| 850 | { |
||
| 851 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 852 | ->getQueryBuilderForTable('index_phash') |
||
| 853 | ->expr(); |
||
| 854 | switch ($this->mediaType) { |
||
| 855 | case '0': |
||
| 856 | // '0' => 'only TYPO3 pages', |
||
| 857 | $whereClause = $expressionBuilder->eq('IP.item_type', $expressionBuilder->literal('0')); |
||
| 858 | break; |
||
| 859 | case '-2': |
||
| 860 | // All external documents |
||
| 861 | $whereClause = $expressionBuilder->neq('IP.item_type', $expressionBuilder->literal('0')); |
||
| 862 | break; |
||
| 863 | case false: |
||
| 864 | // Intentional fall-through |
||
| 865 | case '-1': |
||
| 866 | // All content |
||
| 867 | $whereClause = ''; |
||
| 868 | break; |
||
| 869 | default: |
||
| 870 | $whereClause = $expressionBuilder->eq('IP.item_type', $expressionBuilder->literal($this->mediaType)); |
||
| 871 | } |
||
| 872 | return $whereClause ? ' AND ' . $whereClause : ''; |
||
| 873 | } |
||
| 874 | |||
| 875 | /** |
||
| 876 | * Returns AND statement for selection of language |
||
| 877 | * |
||
| 878 | * @return string AND statement for selection of language |
||
| 879 | */ |
||
| 880 | public function languageWhere() |
||
| 881 | { |
||
| 882 | // -1 is the same as ALL language. |
||
| 883 | if ($this->languageUid < 0) { |
||
| 884 | return ''; |
||
| 885 | } |
||
| 886 | |||
| 887 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 888 | ->getQueryBuilderForTable('index_phash') |
||
| 889 | ->expr(); |
||
| 890 | |||
| 891 | return ' AND ' . $expressionBuilder->eq('IP.sys_language_uid', (int)$this->languageUid); |
||
| 892 | } |
||
| 893 | |||
| 894 | /** |
||
| 895 | * Where-clause for free index-uid value. |
||
| 896 | * |
||
| 897 | * @param int $freeIndexUid Free Index UID value to limit search to. |
||
| 898 | * @return string WHERE SQL clause part. |
||
| 899 | */ |
||
| 900 | public function freeIndexUidWhere($freeIndexUid) |
||
| 901 | { |
||
| 902 | $freeIndexUid = (int)$freeIndexUid; |
||
| 903 | if ($freeIndexUid < 0) { |
||
| 904 | return ''; |
||
| 905 | } |
||
| 906 | // First, look if the freeIndexUid is a meta configuration: |
||
| 907 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 908 | ->getQueryBuilderForTable('index_config'); |
||
| 909 | $indexCfgRec = $queryBuilder->select('indexcfgs') |
||
| 910 | ->from('index_config') |
||
| 911 | ->where( |
||
| 912 | $queryBuilder->expr()->eq('type', $queryBuilder->createNamedParameter(5, \PDO::PARAM_INT)), |
||
| 913 | $queryBuilder->expr()->eq( |
||
| 914 | 'uid', |
||
| 915 | $queryBuilder->createNamedParameter($freeIndexUid, \PDO::PARAM_INT) |
||
| 916 | ) |
||
| 917 | ) |
||
| 918 | ->execute() |
||
| 919 | ->fetch(); |
||
| 920 | |||
| 921 | if (is_array($indexCfgRec)) { |
||
| 922 | $refs = GeneralUtility::trimExplode(',', $indexCfgRec['indexcfgs']); |
||
| 923 | // Default value to protect against empty array. |
||
| 924 | $list = [-99]; |
||
| 925 | foreach ($refs as $ref) { |
||
| 926 | list($table, $uid) = GeneralUtility::revExplode('_', $ref, 2); |
||
| 927 | $uid = (int)$uid; |
||
| 928 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 929 | ->getQueryBuilderForTable('index_config'); |
||
| 930 | $queryBuilder->select('uid') |
||
| 931 | ->from('index_config'); |
||
| 932 | switch ($table) { |
||
| 933 | View Code Duplication | case 'index_config': |
|
| 934 | $idxRec = $queryBuilder |
||
| 935 | ->where( |
||
| 936 | $queryBuilder->expr()->eq( |
||
| 937 | 'uid', |
||
| 938 | $queryBuilder->createNamedParameter($uid, \PDO::PARAM_INT) |
||
| 939 | ) |
||
| 940 | ) |
||
| 941 | ->execute() |
||
| 942 | ->fetch(); |
||
| 943 | if ($idxRec) { |
||
| 944 | $list[] = $uid; |
||
| 945 | } |
||
| 946 | break; |
||
| 947 | View Code Duplication | case 'pages': |
|
| 948 | $indexCfgRecordsFromPid = $queryBuilder |
||
| 949 | ->where( |
||
| 950 | $queryBuilder->expr()->eq( |
||
| 951 | 'pid', |
||
| 952 | $queryBuilder->createNamedParameter($uid, \PDO::PARAM_INT) |
||
| 953 | ) |
||
| 954 | ) |
||
| 955 | ->execute(); |
||
| 956 | while ($idxRec = $indexCfgRecordsFromPid->fetch()) { |
||
| 957 | $list[] = $idxRec['uid']; |
||
| 958 | } |
||
| 959 | break; |
||
| 960 | } |
||
| 961 | } |
||
| 962 | $list = array_unique($list); |
||
| 963 | } else { |
||
| 964 | $list = [$freeIndexUid]; |
||
| 965 | } |
||
| 966 | |||
| 967 | $expressionBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 968 | ->getQueryBuilderForTable('index_phash') |
||
| 969 | ->expr(); |
||
| 970 | return ' AND ' . $expressionBuilder->in('IP.freeIndexUid', array_map('intval', $list)); |
||
| 971 | } |
||
| 972 | |||
| 973 | /** |
||
| 974 | * Execute final query, based on phash integer list. The main point is sorting the result in the right order. |
||
| 975 | * |
||
| 976 | * @param string $list List of phash integers which match the search. |
||
| 977 | * @param int $freeIndexUid Pointer to which indexing configuration you want to search in. -1 means no filtering. 0 means only regular indexed content. |
||
| 978 | * @return Statement |
||
| 979 | */ |
||
| 980 | protected function execFinalQuery($list, $freeIndexUid = -1) |
||
| 981 | { |
||
| 982 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('index_words'); |
||
| 983 | $queryBuilder->select('ISEC.*', 'IP.*') |
||
| 984 | ->from('index_phash', 'IP') |
||
| 985 | ->from('index_section', 'ISEC') |
||
| 986 | ->where( |
||
| 987 | $queryBuilder->expr()->in( |
||
| 988 | 'IP.phash', |
||
| 989 | $queryBuilder->createNamedParameter( |
||
| 990 | GeneralUtility::intExplode(',', $list, true), |
||
| 991 | Connection::PARAM_INT_ARRAY |
||
| 992 | ) |
||
| 993 | ), |
||
| 994 | QueryHelper::stripLogicalOperatorPrefix($this->mediaTypeWhere()), |
||
| 995 | QueryHelper::stripLogicalOperatorPrefix($this->languageWhere()), |
||
| 996 | QueryHelper::stripLogicalOperatorPrefix($this->freeIndexUidWhere($freeIndexUid)), |
||
| 997 | $queryBuilder->expr()->eq('ISEC.phash', $queryBuilder->quoteIdentifier('IP.phash')) |
||
| 998 | ) |
||
| 999 | ->groupBy( |
||
| 1000 | 'IP.phash', |
||
| 1001 | 'ISEC.phash', |
||
| 1002 | 'ISEC.phash_t3', |
||
| 1003 | 'ISEC.rl0', |
||
| 1004 | 'ISEC.rl1', |
||
| 1005 | 'ISEC.rl2', |
||
| 1006 | 'ISEC.page_id', |
||
| 1007 | 'ISEC.uniqid', |
||
| 1008 | 'IP.phash_grouping', |
||
| 1009 | 'IP.data_filename', |
||
| 1010 | 'IP.data_page_id', |
||
| 1011 | 'IP.data_page_reg1', |
||
| 1012 | 'IP.data_page_type', |
||
| 1013 | 'IP.data_page_mp', |
||
| 1014 | 'IP.gr_list', |
||
| 1015 | 'IP.item_type', |
||
| 1016 | 'IP.item_title', |
||
| 1017 | 'IP.item_description', |
||
| 1018 | 'IP.item_mtime', |
||
| 1019 | 'IP.tstamp', |
||
| 1020 | 'IP.item_size', |
||
| 1021 | 'IP.contentHash', |
||
| 1022 | 'IP.crdate', |
||
| 1023 | 'IP.parsetime', |
||
| 1024 | 'IP.sys_language_uid', |
||
| 1025 | 'IP.item_crdate', |
||
| 1026 | 'IP.cHashParams', |
||
| 1027 | 'IP.externalUrl', |
||
| 1028 | 'IP.recordUid', |
||
| 1029 | 'IP.freeIndexUid', |
||
| 1030 | 'IP.freeIndexSetId' |
||
| 1031 | ); |
||
| 1032 | |||
| 1033 | // Setting up methods of filtering results |
||
| 1034 | // based on page types, access, etc. |
||
| 1035 | if ($hookObj = $this->hookRequest('execFinalQuery_idList')) { |
||
| 1036 | // Calling hook for alternative creation of page ID list |
||
| 1037 | $hookWhere = QueryHelper::stripLogicalOperatorPrefix($hookObj->execFinalQuery_idList($list)); |
||
| 1038 | if (!empty($hookWhere)) { |
||
| 1039 | $queryBuilder->andWhere($hookWhere); |
||
| 1040 | } |
||
| 1041 | } elseif ($this->joinPagesForQuery) { |
||
| 1042 | // Alternative to getting all page ids by ->getTreeList() where |
||
| 1043 | // "excludeSubpages" is NOT respected. |
||
| 1044 | $queryBuilder->setRestrictions(GeneralUtility::makeInstance(FrontendRestrictionContainer::class)); |
||
| 1045 | $queryBuilder->from('pages'); |
||
| 1046 | $queryBuilder->andWhere( |
||
| 1047 | $queryBuilder->expr()->eq('pages.uid', $queryBuilder->quoteIdentifier('ISEC.page_id')), |
||
| 1048 | $queryBuilder->expr()->eq( |
||
| 1049 | 'pages.no_search', |
||
| 1050 | $queryBuilder->createNamedParameter(0, \PDO::PARAM_INT) |
||
| 1051 | ), |
||
| 1052 | $queryBuilder->expr()->lt( |
||
| 1053 | 'pages.doktype', |
||
| 1054 | $queryBuilder->createNamedParameter(200, \PDO::PARAM_INT) |
||
| 1055 | ) |
||
| 1056 | ); |
||
| 1057 | } elseif ($this->searchRootPageIdList >= 0) { |
||
| 1058 | // Collecting all pages IDs in which to search; |
||
| 1059 | // filtering out ALL pages that are not accessible due to restriction containers. |
||
| 1060 | // Does NOT look for "no_search" field! |
||
| 1061 | $siteIdNumbers = GeneralUtility::intExplode(',', $this->searchRootPageIdList); |
||
| 1062 | $pageIdList = []; |
||
| 1063 | foreach ($siteIdNumbers as $rootId) { |
||
| 1064 | $pageIdList[] = $this->getTypoScriptFrontendController()->cObj->getTreeList(-1 * $rootId, 9999); |
||
| 1065 | } |
||
| 1066 | $queryBuilder->andWhere( |
||
| 1067 | $queryBuilder->expr()->in( |
||
| 1068 | 'ISEC.page_id', |
||
| 1069 | $queryBuilder->createNamedParameter( |
||
| 1070 | array_unique(GeneralUtility::intExplode(',', implode(',', $pageIdList), true)), |
||
| 1071 | Connection::PARAM_INT_ARRAY |
||
| 1072 | ) |
||
| 1073 | ) |
||
| 1074 | ); |
||
| 1075 | } |
||
| 1076 | // otherwise select all / disable everything |
||
| 1077 | // If any of the ranking sortings are selected, we must make a |
||
| 1078 | // join with the word/rel-table again, because we need to |
||
| 1079 | // calculate ranking based on all search-words found. |
||
| 1080 | if (substr($this->sortOrder, 0, 5) === 'rank_') { |
||
| 1081 | $queryBuilder |
||
| 1082 | ->from('index_words', 'IW') |
||
| 1083 | ->from('index_rel', 'IR') |
||
| 1084 | ->andWhere( |
||
| 1085 | $queryBuilder->expr()->eq('IW.wid', $queryBuilder->quoteIdentifier('IR.wid')), |
||
| 1086 | $queryBuilder->expr()->eq('ISEC.phash', $queryBuilder->quoteIdentifier('IR.phash')) |
||
| 1087 | ); |
||
| 1088 | switch ($this->sortOrder) { |
||
| 1089 | case 'rank_flag': |
||
| 1090 | // This gives priority to word-position (max-value) so that words in title, keywords, description counts more than in content. |
||
| 1091 | // The ordering is refined with the frequency sum as well. |
||
| 1092 | $queryBuilder |
||
| 1093 | ->addSelectLiteral( |
||
| 1094 | $queryBuilder->expr()->max('IR.flags', 'order_val1'), |
||
| 1095 | $queryBuilder->expr()->sum('IR.freq', 'order_val2') |
||
| 1096 | ) |
||
| 1097 | ->orderBy('order_val1', $this->getDescendingSortOrderFlag()) |
||
| 1098 | ->addOrderBy('order_val2', $this->getDescendingSortOrderFlag()); |
||
| 1099 | break; |
||
| 1100 | case 'rank_first': |
||
| 1101 | // Results in average position of search words on page. |
||
| 1102 | // Must be inversely sorted (low numbers are closer to top) |
||
| 1103 | $queryBuilder |
||
| 1104 | ->addSelectLiteral($queryBuilder->expr()->avg('IR.first', 'order_val')) |
||
| 1105 | ->orderBy('order_val', $this->getDescendingSortOrderFlag(true)); |
||
| 1106 | break; |
||
| 1107 | case 'rank_count': |
||
| 1108 | // Number of words found |
||
| 1109 | $queryBuilder |
||
| 1110 | ->addSelectLiteral($queryBuilder->expr()->sum('IR.count', 'order_val')) |
||
| 1111 | ->orderBy('order_val', $this->getDescendingSortOrderFlag()); |
||
| 1112 | break; |
||
| 1113 | default: |
||
| 1114 | // Frequency sum. I'm not sure if this is the best way to do |
||
| 1115 | // it (make a sum...). Or should it be the average? |
||
| 1116 | $queryBuilder |
||
| 1117 | ->addSelectLiteral($queryBuilder->expr()->sum('IR.freq', 'order_val')) |
||
| 1118 | ->orderBy('order_val', $this->getDescendingSortOrderFlag()); |
||
| 1119 | } |
||
| 1120 | |||
| 1121 | if (!empty($this->wSelClauses)) { |
||
| 1122 | // So, words are combined in an OR statement |
||
| 1123 | // (no "sentence search" should be done here - may deselect results) |
||
| 1124 | $wordSel = $queryBuilder->expr()->orX(); |
||
| 1125 | foreach ($this->wSelClauses as $wSelClause) { |
||
| 1126 | $wordSel->add(QueryHelper::stripLogicalOperatorPrefix($wSelClause)); |
||
| 1127 | } |
||
| 1128 | $queryBuilder->andWhere($wordSel); |
||
| 1129 | } |
||
| 1130 | } else { |
||
| 1131 | // Otherwise, if sorting are done with the pages table or other fields, |
||
| 1132 | // there is no need for joining with the rel/word tables: |
||
| 1133 | switch ((string)$this->sortOrder) { |
||
| 1134 | case 'title': |
||
| 1135 | $queryBuilder->orderBy('IP.item_title', $this->getDescendingSortOrderFlag()); |
||
| 1136 | break; |
||
| 1137 | case 'crdate': |
||
| 1138 | $queryBuilder->orderBy('IP.item_crdate', $this->getDescendingSortOrderFlag()); |
||
| 1139 | break; |
||
| 1140 | case 'mtime': |
||
| 1141 | $queryBuilder->orderBy('IP.item_mtime', $this->getDescendingSortOrderFlag()); |
||
| 1142 | break; |
||
| 1143 | } |
||
| 1144 | } |
||
| 1145 | |||
| 1146 | return $queryBuilder->execute(); |
||
| 1147 | } |
||
| 1148 | |||
| 1149 | /** |
||
| 1150 | * Checking if the resume can be shown for the search result |
||
| 1151 | * (depending on whether the rights are OK) |
||
| 1152 | * ? Should it also check for gr_list "0,-1"? |
||
| 1153 | * |
||
| 1154 | * @param array $row Result row array. |
||
| 1155 | * @return bool Returns TRUE if resume can safely be shown |
||
| 1156 | */ |
||
| 1157 | protected function checkResume($row) |
||
| 1158 | { |
||
| 1159 | // If the record is indexed by an indexing configuration, just show it. |
||
| 1160 | // At least this is needed for external URLs and files. |
||
| 1161 | // For records we might need to extend this - for instance block display if record is access restricted. |
||
| 1162 | if ($row['freeIndexUid']) { |
||
| 1163 | return true; |
||
| 1164 | } |
||
| 1165 | // Evaluate regularly indexed pages based on item_type: |
||
| 1166 | // External media: |
||
| 1167 | $connection = GeneralUtility::makeInstance(ConnectionPool::class)->getConnectionForTable('index_grlist'); |
||
| 1168 | View Code Duplication | if ($row['item_type']) { |
|
| 1169 | // For external media we will check the access of the parent page on which the media was linked from. |
||
| 1170 | // "phash_t3" is the phash of the parent TYPO3 page row which initiated the indexing of the documents |
||
| 1171 | // in this section. So, selecting for the grlist records belonging to the parent phash-row where the |
||
| 1172 | // current users gr_list exists will help us to know. If this is NOT found, there is still a theoretical |
||
| 1173 | // possibility that another user accessible page would display a link, so maybe the resume of such a |
||
| 1174 | // document here may be unjustified hidden. But better safe than sorry. |
||
| 1175 | if (!$this->isTableUsed('index_grlist')) { |
||
| 1176 | return false; |
||
| 1177 | } |
||
| 1178 | |||
| 1179 | return (bool)$connection->count( |
||
| 1180 | 'phash', |
||
| 1181 | 'index_grlist', |
||
| 1182 | [ |
||
| 1183 | 'phash' => (int)$row['phash_t3'], |
||
| 1184 | 'gr_list' => $this->frontendUserGroupList |
||
| 1185 | ] |
||
| 1186 | ); |
||
| 1187 | } |
||
| 1188 | // Ordinary TYPO3 pages: |
||
| 1189 | View Code Duplication | if ((string)$row['gr_list'] !== (string)$this->frontendUserGroupList) { |
|
| 1190 | // Selecting for the grlist records belonging to the phash-row where the current users gr_list exists. |
||
| 1191 | // If it is found it is proof that this user has direct access to the phash-rows content although |
||
| 1192 | // he did not himself initiate the indexing... |
||
| 1193 | if (!$this->isTableUsed('index_grlist')) { |
||
| 1194 | return false; |
||
| 1195 | } |
||
| 1196 | |||
| 1197 | return (bool)$connection->count( |
||
| 1198 | 'phash', |
||
| 1199 | 'index_grlist', |
||
| 1200 | [ |
||
| 1201 | 'phash' => (int)$row['phash'], |
||
| 1202 | 'gr_list' => $this->frontendUserGroupList |
||
| 1203 | ] |
||
| 1204 | ); |
||
| 1205 | } |
||
| 1206 | return true; |
||
| 1207 | } |
||
| 1208 | |||
| 1209 | /** |
||
| 1210 | * Returns "DESC" or "" depending on the settings of the incoming |
||
| 1211 | * highest/lowest result order (piVars['desc']) |
||
| 1212 | * |
||
| 1213 | * @param bool $inverse If TRUE, inverse the order which is defined by piVars['desc'] |
||
| 1214 | * @return string " DESC" or formerly known as tx_indexedsearch_pi->isDescending |
||
| 1215 | */ |
||
| 1216 | protected function getDescendingSortOrderFlag($inverse = false) |
||
| 1223 | } |
||
| 1224 | |||
| 1225 | /** |
||
| 1226 | * Returns if an item type is a multipage item type |
||
| 1227 | * |
||
| 1228 | * @param string $itemType Item type |
||
| 1229 | * @return bool TRUE if multipage capable |
||
| 1230 | */ |
||
| 1231 | protected function multiplePagesType($itemType) |
||
| 1232 | { |
||
| 1233 | /** @var \TYPO3\CMS\IndexedSearch\FileContentParser $fileContentParser */ |
||
| 1234 | $fileContentParser = $this->externalParsers[$itemType]; |
||
| 1235 | return is_object($fileContentParser) && $fileContentParser->isMultiplePageExtension($itemType); |
||
| 1236 | } |
||
| 1237 | |||
| 1238 | /** |
||
| 1239 | * md5 integer hash |
||
| 1240 | * Using 7 instead of 8 just because that makes the integers lower than |
||
| 1241 | * 32 bit (28 bit) and so they do not interfere with UNSIGNED integers |
||
| 1242 | * or PHP-versions which has varying output from the hexdec function. |
||
| 1243 | * |
||
| 1244 | * @param string $str String to hash |
||
| 1245 | * @return int Integer interpretation of the md5 hash of input string. |
||
| 1246 | */ |
||
| 1247 | protected function md5inthash($str) |
||
| 1248 | { |
||
| 1249 | return Utility\IndexedSearchUtility::md5inthash($str); |
||
| 1250 | } |
||
| 1251 | |||
| 1252 | /** |
||
| 1253 | * Check if the tables provided are configured for usage. |
||
| 1254 | * This becomes necessary for extensions that provide additional database |
||
| 1255 | * functionality like indexed_search_mysql. |
||
| 1256 | * |
||
| 1257 | * @param string $table_list Comma-separated list of tables |
||
| 1258 | * @return bool TRUE if given tables are enabled |
||
| 1259 | */ |
||
| 1260 | protected function isTableUsed($table_list) |
||
| 1261 | { |
||
| 1262 | return Utility\IndexedSearchUtility::isTableUsed($table_list); |
||
| 1263 | } |
||
| 1264 | |||
| 1265 | /** |
||
| 1266 | * Returns an object reference to the hook object if any |
||
| 1267 | * |
||
| 1268 | * @param string $functionName Name of the function you want to call / hook key |
||
| 1269 | * @return object|null Hook object, if any. Otherwise NULL. |
||
| 1270 | */ |
||
| 1271 | View Code Duplication | public function hookRequest($functionName) |
|
| 1272 | { |
||
| 1273 | // Hook: menuConfig_preProcessModMenu |
||
| 1274 | if ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['pi1_hooks'][$functionName]) { |
||
| 1275 | $hookObj = GeneralUtility::makeInstance($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['pi1_hooks'][$functionName]); |
||
| 1276 | if (method_exists($hookObj, $functionName)) { |
||
| 1277 | $hookObj->pObj = $this; |
||
| 1278 | return $hookObj; |
||
| 1279 | } |
||
| 1280 | } |
||
| 1281 | return null; |
||
| 1282 | } |
||
| 1283 | |||
| 1284 | /** |
||
| 1285 | * Search type |
||
| 1286 | * e.g. sentence (20), any part of the word (1) |
||
| 1287 | * |
||
| 1288 | * @return int |
||
| 1289 | */ |
||
| 1290 | public function getSearchType() |
||
| 1291 | { |
||
| 1292 | return (int)$this->searchType; |
||
| 1293 | } |
||
| 1294 | |||
| 1295 | /** |
||
| 1296 | * A list of integer which should be root-pages to search from |
||
| 1297 | * |
||
| 1298 | * @return int[] |
||
| 1299 | */ |
||
| 1300 | public function getSearchRootPageIdList() |
||
| 1301 | { |
||
| 1302 | return GeneralUtility::intExplode(',', $this->searchRootPageIdList); |
||
| 1303 | } |
||
| 1304 | |||
| 1305 | /** |
||
| 1306 | * Getter for joinPagesForQuery flag |
||
| 1307 | * enabled through TypoScript 'settings.skipExtendToSubpagesChecking' |
||
| 1308 | * |
||
| 1309 | * @return bool |
||
| 1310 | */ |
||
| 1311 | public function getJoinPagesForQuery() |
||
| 1314 | } |
||
| 1315 | |||
| 1316 | /** |
||
| 1317 | * @return \TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController |
||
| 1318 | */ |
||
| 1319 | protected function getTypoScriptFrontendController() |
||
| 1320 | { |
||
| 1321 | return $GLOBALS['TSFE']; |
||
| 1322 | } |
||
| 1323 | |||
| 1324 | /** |
||
| 1325 | * @return TimeTracker |
||
| 1326 | */ |
||
| 1327 | protected function getTimeTracker() |
||
| 1330 | } |
||
| 1331 | } |
||
| 1332 |
Loading history...
This check looks for assignments to scalar types that may be of the wrong type.
To ensure the code behaves as expected, it may be a good idea to add an explicit type cast.