LogQueryBuilder
5 years ago
Actions.php
5 years ago
ArchiveSelector.php
5 years ago
ArchiveTableCreator.php
5 years ago
ArchiveTableDao.php
5 years ago
ArchiveWriter.php
5 years ago
ArchivingDbAdapter.php
5 years ago
LogAggregator.php
5 years ago
LogQueryBuilder.php
5 years ago
LogTableTemporary.php
5 years ago
Model.php
5 years ago
RawLogDao.php
5 years ago
TableMetadata.php
5 years ago
Model.php
968 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Matomo - free/libre analytics platform |
| 4 | * |
| 5 | * @link https://matomo.org |
| 6 | * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 7 | * |
| 8 | */ |
| 9 | namespace Piwik\DataAccess; |
| 10 | |
| 11 | use Exception; |
| 12 | use Piwik\Archive\ArchiveInvalidator; |
| 13 | use Piwik\ArchiveProcessor\ArchivingStatus; |
| 14 | use Piwik\ArchiveProcessor\Parameters; |
| 15 | use Piwik\ArchiveProcessor\Rules; |
| 16 | use Piwik\Common; |
| 17 | use Piwik\Container\StaticContainer; |
| 18 | use Piwik\Date; |
| 19 | use Piwik\Db; |
| 20 | use Piwik\DbHelper; |
| 21 | use Piwik\Period; |
| 22 | use Piwik\Segment; |
| 23 | use Piwik\Sequence; |
| 24 | use Piwik\SettingsServer; |
| 25 | use Piwik\Site; |
| 26 | use Psr\Log\LoggerInterface; |
| 27 | |
| 28 | /** |
| 29 | * Cleans up outdated archives |
| 30 | */ |
| 31 | class Model |
| 32 | { |
| 33 | /** |
| 34 | * @var LoggerInterface |
| 35 | */ |
| 36 | private $logger; |
| 37 | |
| 38 | /** |
| 39 | * @var ArchivingStatus |
| 40 | */ |
| 41 | private $archivingStatus; |
| 42 | |
| 43 | public function __construct(LoggerInterface $logger = null) |
| 44 | { |
| 45 | $this->logger = $logger ?: StaticContainer::get('Psr\Log\LoggerInterface'); |
| 46 | $this->archivingStatus = StaticContainer::get(ArchivingStatus::class); |
| 47 | } |
| 48 | |
| 49 | /** |
| 50 | * Returns the archives IDs that have already been invalidated and have been since re-processed. |
| 51 | * |
| 52 | * These archives { archive name (includes segment hash) , idsite, date, period } will be deleted. |
| 53 | * |
| 54 | * @param string $archiveTable |
| 55 | * @param array $idSites |
| 56 | * @param bool $setGroupContentMaxLen for tests only |
| 57 | * @return array |
| 58 | * @throws Exception |
| 59 | */ |
| 60 | public function getInvalidatedArchiveIdsSafeToDelete($archiveTable, $setGroupContentMaxLen = true) |
| 61 | { |
| 62 | if ($setGroupContentMaxLen) { |
| 63 | try { |
| 64 | Db::get()->query('SET SESSION group_concat_max_len=' . (128 * 1024)); |
| 65 | } catch (\Exception $ex) { |
| 66 | $this->logger->info("Could not set group_concat_max_len MySQL session variable."); |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | $sql = "SELECT idsite, date1, date2, period, name, |
| 71 | GROUP_CONCAT(idarchive, '.', value ORDER BY ts_archived DESC) as archives |
| 72 | FROM `$archiveTable` |
| 73 | WHERE name LIKE 'done%' |
| 74 | AND ts_archived IS NOT NULL |
| 75 | AND `value` NOT IN (" . ArchiveWriter::DONE_ERROR . ") |
| 76 | GROUP BY idsite, date1, date2, period, name HAVING count(*) > 1"; |
| 77 | |
| 78 | $archiveIds = array(); |
| 79 | |
| 80 | $rows = Db::fetchAll($sql); |
| 81 | foreach ($rows as $row) { |
| 82 | $duplicateArchives = explode(',', $row['archives']); |
| 83 | |
| 84 | // do not consider purging partial archives, if they are the latest archive, |
| 85 | // and we don't want to delete the latest archive if it is usable |
| 86 | while (!empty($duplicateArchives)) { |
| 87 | $pair = $duplicateArchives[0]; |
| 88 | if ($this->isCutOffGroupConcatResult($pair)) { // can occur if the GROUP_CONCAT value is cut off |
| 89 | break; |
| 90 | } |
| 91 | |
| 92 | list($idarchive, $value) = explode('.', $pair); |
| 93 | |
| 94 | array_shift($duplicateArchives); |
| 95 | |
| 96 | if ($value != ArchiveWriter::DONE_PARTIAL) { |
| 97 | break; |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | // if there is more than one archive, the older invalidated ones can be deleted |
| 102 | if (!empty($duplicateArchives)) { |
| 103 | foreach ($duplicateArchives as $pair) { |
| 104 | if ($this->isCutOffGroupConcatResult($pair)) { |
| 105 | $this->logger->info("GROUP_CONCAT cut off the query result, you may have to purge archives again."); |
| 106 | break; |
| 107 | } |
| 108 | |
| 109 | list($idarchive, $value) = explode('.', $pair); |
| 110 | $archiveIds[] = $idarchive; // does not matter what the value is, the latest is usable so older archives can be purged |
| 111 | } |
| 112 | } |
| 113 | } |
| 114 | |
| 115 | return $archiveIds; |
| 116 | } |
| 117 | |
| 118 | public function getPlaceholderArchiveIds($archiveTable) |
| 119 | { |
| 120 | $sql = "SELECT DISTINCT idarchive FROM `$archiveTable` WHERE ts_archived IS NULL"; |
| 121 | $result = Db::fetchAll($sql); |
| 122 | $result = array_column($result, 'idarchive'); |
| 123 | return $result; |
| 124 | } |
| 125 | |
| 126 | public function updateArchiveAsInvalidated($archiveTable, $idSites, $allPeriodsToInvalidate, Segment $segment = null, |
| 127 | $forceInvalidateNonexistantRanges = false, $name = null) |
| 128 | { |
| 129 | if (empty($idSites)) { |
| 130 | return 0; |
| 131 | } |
| 132 | |
| 133 | // select all idarchive/name pairs we want to invalidate |
| 134 | $sql = "SELECT idarchive, idsite, period, date1, date2, `name`, `value` |
| 135 | FROM `$archiveTable` |
| 136 | WHERE idsite IN (" . implode(',', $idSites) . ") AND value <> " . ArchiveWriter::DONE_PARTIAL; |
| 137 | |
| 138 | $periodCondition = ''; |
| 139 | if (!empty($allPeriodsToInvalidate)) { |
| 140 | $periodCondition .= " AND ("; |
| 141 | |
| 142 | $isFirst = true; |
| 143 | /** @var Period $period */ |
| 144 | foreach ($allPeriodsToInvalidate as $period) { |
| 145 | if ($isFirst) { |
| 146 | $isFirst = false; |
| 147 | } else { |
| 148 | $periodCondition .= " OR "; |
| 149 | } |
| 150 | |
| 151 | if ($period->getLabel() == 'range') { // for ranges, we delete all ranges that contain the given date(s) |
| 152 | $periodCondition .= "(period = " . (int)$period->getId() |
| 153 | . " AND date2 >= '" . $period->getDateStart()->getDatetime() |
| 154 | . "' AND date1 <= '" . $period->getDateEnd()->getDatetime() . "')"; |
| 155 | } else { |
| 156 | $periodCondition .= "(period = " . (int)$period->getId() |
| 157 | . " AND date1 = '" . $period->getDateStart()->getDatetime() . "'" |
| 158 | . " AND date2 = '" . $period->getDateEnd()->getDatetime() . "')"; |
| 159 | } |
| 160 | } |
| 161 | $periodCondition .= ")"; |
| 162 | } |
| 163 | $sql .= $periodCondition; |
| 164 | |
| 165 | if (!empty($name)) { |
| 166 | if (strpos($name, '.') !== false) { |
| 167 | list($plugin, $name) = explode('.', $name, 2); |
| 168 | } else { |
| 169 | $plugin = $name; |
| 170 | $name = null; |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | if (empty($plugin)) { |
| 175 | $doneFlag = Rules::getDoneFlagArchiveContainsAllPlugins($segment ?: new Segment('', [])); |
| 176 | } else { |
| 177 | $doneFlag = Rules::getDoneFlagArchiveContainsOnePlugin($segment ?: new Segment('', []), $plugin); |
| 178 | } |
| 179 | |
| 180 | $nameCondition = "name LIKE '$doneFlag%'"; |
| 181 | |
| 182 | $sql .= " AND $nameCondition"; |
| 183 | |
| 184 | $idArchives = []; |
| 185 | $archivesToInvalidate = []; |
| 186 | |
| 187 | // update each archive as invalidated (but only for full archives or plugin archives, not for partial archives. |
| 188 | // DONE_INVALIDATED also implies that an archive is whole and not partial, and we want to avoid that.) |
| 189 | if (empty($name)) { |
| 190 | $archivesToInvalidate = Db::fetchAll($sql); |
| 191 | $idArchives = array_column($archivesToInvalidate, 'idarchive'); |
| 192 | |
| 193 | if (!empty($idArchives)) { |
| 194 | $idArchives = array_map('intval', $idArchives); |
| 195 | |
| 196 | $sql = "UPDATE `$archiveTable` SET `value` = " . ArchiveWriter::DONE_INVALIDATED . " WHERE idarchive IN (" |
| 197 | . implode(',', $idArchives) . ") AND $nameCondition"; |
| 198 | |
| 199 | Db::query($sql); |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | // we add every archive we need to invalidate + the archives that do not already exist to archive_invalidations. |
| 204 | // except for archives that are DONE_IN_PROGRESS. |
| 205 | $archivesToCreateInvalidationRowsFor = []; |
| 206 | foreach ($archivesToInvalidate as $row) { |
| 207 | if ($row['name'] != $doneFlag) { // only look at done flags that equal the one we are explicitly adding |
| 208 | continue; |
| 209 | } |
| 210 | |
| 211 | $archivesToCreateInvalidationRowsFor[$row['idsite']][$row['period']][$row['date1']][$row['date2']] = $row['idarchive']; |
| 212 | } |
| 213 | |
| 214 | $now = Date::now()->getDatetime(); |
| 215 | |
| 216 | $existingInvalidations = $this->getExistingInvalidations($idSites, $periodCondition, $nameCondition); |
| 217 | |
| 218 | $dummyArchives = []; |
| 219 | foreach ($idSites as $idSite) { |
| 220 | try { |
| 221 | $siteCreationTime = Site::getCreationDateFor($idSite); |
| 222 | } catch (\Exception $ex) { |
| 223 | continue; |
| 224 | } |
| 225 | |
| 226 | $siteCreationTime = Date::factory($siteCreationTime); |
| 227 | foreach ($allPeriodsToInvalidate as $period) { |
| 228 | if ($period->getLabel() == 'range' |
| 229 | && !$forceInvalidateNonexistantRanges |
| 230 | ) { |
| 231 | continue; // range |
| 232 | } |
| 233 | |
| 234 | if ($period->getDateEnd()->isEarlier($siteCreationTime)) { |
| 235 | continue; // don't add entries if it is before the time the site was created |
| 236 | } |
| 237 | |
| 238 | $date1 = $period->getDateStart()->toString(); |
| 239 | $date2 = $period->getDateEnd()->toString(); |
| 240 | |
| 241 | $key = $this->makeExistingInvalidationArrayKey($idSite, $date1, $date2, $period->getId(), $doneFlag, $name); |
| 242 | if (!empty($existingInvalidations[$key])) { |
| 243 | continue; // avoid adding duplicates where possible |
| 244 | } |
| 245 | |
| 246 | $idArchive = $archivesToCreateInvalidationRowsFor[$idSite][$period->getId()][$date1][$date2] ?? null; |
| 247 | |
| 248 | $dummyArchives[] = [ |
| 249 | 'idarchive' => $idArchive, |
| 250 | 'name' => $doneFlag, |
| 251 | 'report' => $name, |
| 252 | 'idsite' => $idSite, |
| 253 | 'date1' => $period->getDateStart()->getDatetime(), |
| 254 | 'date2' => $period->getDateEnd()->getDatetime(), |
| 255 | 'period' => $period->getId(), |
| 256 | 'ts_invalidated' => $now, |
| 257 | ]; |
| 258 | } |
| 259 | } |
| 260 | |
| 261 | if (!empty($dummyArchives)) { |
| 262 | $fields = ['idarchive', 'name', 'report', 'idsite', 'date1', 'date2', 'period', 'ts_invalidated']; |
| 263 | Db\BatchInsert::tableInsertBatch(Common::prefixTable('archive_invalidations'), $fields, $dummyArchives); |
| 264 | } |
| 265 | |
| 266 | return count($idArchives); |
| 267 | } |
| 268 | |
| 269 | private function getExistingInvalidations($idSites, $periodCondition, $nameCondition) |
| 270 | { |
| 271 | $table = Common::prefixTable('archive_invalidations'); |
| 272 | |
| 273 | $idSites = array_map('intval', $idSites); |
| 274 | |
| 275 | $sql = "SELECT idsite, date1, date2, period, name, report, COUNT(*) as `count` FROM `$table` |
| 276 | WHERE idsite IN (" . implode(',', $idSites) . ") AND status = " . ArchiveInvalidator::INVALIDATION_STATUS_QUEUED . " |
| 277 | $periodCondition AND $nameCondition |
| 278 | GROUP BY idsite, date1, date2, period, name"; |
| 279 | $rows = Db::fetchAll($sql); |
| 280 | |
| 281 | $invalidations = []; |
| 282 | foreach ($rows as $row) { |
| 283 | $key = $this->makeExistingInvalidationArrayKey($row['idsite'], $row['date1'], $row['date2'], $row['period'], $row['name'], $row['report']); |
| 284 | $invalidations[$key] = $row['count']; |
| 285 | } |
| 286 | return $invalidations; |
| 287 | } |
| 288 | |
| 289 | private function makeExistingInvalidationArrayKey($idSite, $date1, $date2, $period, $name, $report) |
| 290 | { |
| 291 | return implode('.', [$idSite, $date1, $date2, $period, $name, $report]); |
| 292 | } |
| 293 | |
| 294 | /** |
| 295 | * @param string $archiveTable Prefixed table name |
| 296 | * @param int[] $idSites |
| 297 | * @param string[][] $datesByPeriodType |
| 298 | * @param Segment $segment |
| 299 | * @return \Zend_Db_Statement |
| 300 | * @throws Exception |
| 301 | */ |
| 302 | public function updateRangeArchiveAsInvalidated($archiveTable, $idSites, $allPeriodsToInvalidate, Segment $segment = null) |
| 303 | { |
| 304 | if (empty($idSites)) { |
| 305 | return; |
| 306 | } |
| 307 | |
| 308 | $bind = array(); |
| 309 | |
| 310 | $periodConditions = array(); |
| 311 | if (!empty($allPeriodsToInvalidate)) { |
| 312 | foreach ($allPeriodsToInvalidate as $period) { |
| 313 | $dateConditions = array(); |
| 314 | |
| 315 | /** @var Period $period */ |
| 316 | $dateConditions[] = "(date1 <= ? AND ? <= date2)"; |
| 317 | $bind[] = $period->getDateStart(); |
| 318 | $bind[] = $period->getDateEnd(); |
| 319 | |
| 320 | $dateConditionsSql = implode(" OR ", $dateConditions); |
| 321 | $periodConditions[] = "(period = 5 AND ($dateConditionsSql))"; |
| 322 | } |
| 323 | } |
| 324 | |
| 325 | if ($segment) { |
| 326 | $nameCondition = "name LIKE '" . Rules::getDoneFlagArchiveContainsAllPlugins($segment) . "%'"; |
| 327 | } else { |
| 328 | $nameCondition = "name LIKE 'done%'"; |
| 329 | } |
| 330 | |
| 331 | $sql = "UPDATE $archiveTable SET value = " . ArchiveWriter::DONE_INVALIDATED |
| 332 | . " WHERE $nameCondition |
| 333 | AND idsite IN (" . implode(", ", $idSites) . ") |
| 334 | AND (" . implode(" OR ", $periodConditions) . ")"; |
| 335 | |
| 336 | return Db::query($sql, $bind); |
| 337 | } |
| 338 | |
| 339 | public function getTemporaryArchivesOlderThan($archiveTable, $purgeArchivesOlderThan) |
| 340 | { |
| 341 | $query = "SELECT idarchive FROM " . $archiveTable . " |
| 342 | WHERE name LIKE 'done%' |
| 343 | AND (( value = " . ArchiveWriter::DONE_OK_TEMPORARY . " |
| 344 | AND ts_archived < ?) |
| 345 | OR value = " . ArchiveWriter::DONE_ERROR . ")"; |
| 346 | |
| 347 | return Db::fetchAll($query, array($purgeArchivesOlderThan)); |
| 348 | } |
| 349 | |
| 350 | public function deleteArchivesWithPeriod($numericTable, $blobTable, $period, $date) |
| 351 | { |
| 352 | if (SettingsServer::isArchivePhpTriggered()) { |
| 353 | StaticContainer::get(LoggerInterface::class)->info('deleteArchivesWithPeriod: ' . $numericTable . ' with period = ' . $period . ' and date = ' . $date); |
| 354 | } |
| 355 | |
| 356 | $query = "DELETE FROM %s WHERE period = ? AND ts_archived < ?"; |
| 357 | $bind = array($period, $date); |
| 358 | |
| 359 | $queryObj = Db::query(sprintf($query, $numericTable), $bind); |
| 360 | $deletedRows = $queryObj->rowCount(); |
| 361 | |
| 362 | try { |
| 363 | $queryObj = Db::query(sprintf($query, $blobTable), $bind); |
| 364 | $deletedRows += $queryObj->rowCount(); |
| 365 | } catch (Exception $e) { |
| 366 | // Individual blob tables could be missing |
| 367 | $this->logger->debug("Unable to delete archives by period from {blobTable}.", array( |
| 368 | 'blobTable' => $blobTable, |
| 369 | 'exception' => $e, |
| 370 | )); |
| 371 | } |
| 372 | |
| 373 | return $deletedRows; |
| 374 | } |
| 375 | |
| 376 | public function deleteArchiveIds($numericTable, $blobTable, $idsToDelete) |
| 377 | { |
| 378 | $idsToDelete = array_values($idsToDelete); |
| 379 | |
| 380 | $idsToDelete = array_map('intval', $idsToDelete); |
| 381 | $query = "DELETE FROM %s WHERE idarchive IN (" . implode(',', $idsToDelete) . ")"; |
| 382 | |
| 383 | $queryObj = Db::query(sprintf($query, $numericTable), array()); |
| 384 | $deletedRows = $queryObj->rowCount(); |
| 385 | |
| 386 | try { |
| 387 | $queryObj = Db::query(sprintf($query, $blobTable), array()); |
| 388 | $deletedRows += $queryObj->rowCount(); |
| 389 | } catch (Exception $e) { |
| 390 | // Individual blob tables could be missing |
| 391 | $this->logger->debug("Unable to delete archive IDs from {blobTable}.", array( |
| 392 | 'blobTable' => $blobTable, |
| 393 | 'exception' => $e, |
| 394 | )); |
| 395 | } |
| 396 | |
| 397 | return $deletedRows; |
| 398 | } |
| 399 | |
| 400 | public function deleteOlderArchives(Parameters $params, $name, $tsArchived, $idArchive) |
| 401 | { |
| 402 | $dateStart = $params->getPeriod()->getDateStart(); |
| 403 | $dateEnd = $params->getPeriod()->getDateEnd(); |
| 404 | |
| 405 | $numericTable = ArchiveTableCreator::getNumericTable($dateStart); |
| 406 | $blobTable = ArchiveTableCreator::getBlobTable($dateStart); |
| 407 | |
| 408 | $sql = "SELECT idarchive FROM `$numericTable` WHERE idsite = ? AND date1 = ? AND date2 = ? AND period = ? AND name = ? AND ts_archived < ? AND idarchive < ?"; |
| 409 | |
| 410 | $idArchives = Db::fetchAll($sql, [$params->getSite()->getId(), $dateStart->getDatetime(), $dateEnd->getDatetime(), $params->getPeriod()->getId(), $name, $tsArchived, $idArchive]); |
| 411 | $idArchives = array_column($idArchives, 'idarchive'); |
| 412 | if (empty($idArchives)) { |
| 413 | return; |
| 414 | } |
| 415 | |
| 416 | if (SettingsServer::isArchivePhpTriggered()) { |
| 417 | StaticContainer::get(LoggerInterface::class)->info('deleteOlderArchives with ' . $params . ', name = ' . $name . ', ts_archived < ' . $tsArchived . ', idarchive < ' . $idArchive); |
| 418 | } |
| 419 | |
| 420 | $this->deleteArchiveIds($numericTable, $blobTable, $idArchives); |
| 421 | } |
| 422 | |
| 423 | public function getArchiveIdAndVisits($numericTable, $idSite, $period, $dateStartIso, $dateEndIso, $minDatetimeIsoArchiveProcessedUTC, |
| 424 | $doneFlags, $doneFlagValues = null) |
| 425 | { |
| 426 | $bindSQL = array($idSite, |
| 427 | $dateStartIso, |
| 428 | $dateEndIso, |
| 429 | $period, |
| 430 | ); |
| 431 | |
| 432 | $sqlWhereArchiveName = self::getNameCondition($doneFlags, $doneFlagValues); |
| 433 | |
| 434 | $timeStampWhere = ''; |
| 435 | if ($minDatetimeIsoArchiveProcessedUTC) { |
| 436 | $timeStampWhere = " AND arc1.ts_archived >= ? "; |
| 437 | $bindSQL[] = $minDatetimeIsoArchiveProcessedUTC; |
| 438 | } |
| 439 | |
| 440 | // NOTE: we can't predict how many segments there will be so there could be lots of nb_visits/nb_visits_converted rows... have to select everything. |
| 441 | $sqlQuery = "SELECT arc1.idarchive, arc1.value, arc1.name, arc1.ts_archived, arc1.date1 as startDate, arc2.value as " . ArchiveSelector::NB_VISITS_RECORD_LOOKED_UP . ", arc3.value as " . ArchiveSelector::NB_VISITS_CONVERTED_RECORD_LOOKED_UP . " |
| 442 | FROM $numericTable arc1 |
| 443 | LEFT JOIN $numericTable arc2 on arc2.idarchive = arc1.idarchive and (arc2.name = '" . ArchiveSelector::NB_VISITS_RECORD_LOOKED_UP . "') |
| 444 | LEFT JOIN $numericTable arc3 on arc3.idarchive = arc1.idarchive and (arc3.name = '" . ArchiveSelector::NB_VISITS_CONVERTED_RECORD_LOOKED_UP . "') |
| 445 | WHERE arc1.idsite = ? |
| 446 | AND arc1.date1 = ? |
| 447 | AND arc1.date2 = ? |
| 448 | AND arc1.period = ? |
| 449 | AND ($sqlWhereArchiveName) |
| 450 | $timeStampWhere |
| 451 | AND arc1.ts_archived IS NOT NULL |
| 452 | ORDER BY arc1.ts_archived DESC, arc1.idarchive DESC"; |
| 453 | |
| 454 | $results = Db::fetchAll($sqlQuery, $bindSQL); |
| 455 | |
| 456 | return $results; |
| 457 | } |
| 458 | |
| 459 | public function createArchiveTable($tableName, $tableNamePrefix) |
| 460 | { |
| 461 | $db = Db::get(); |
| 462 | $sql = DbHelper::getTableCreateSql($tableNamePrefix); |
| 463 | |
| 464 | // replace table name template by real name |
| 465 | $tableNamePrefix = Common::prefixTable($tableNamePrefix); |
| 466 | $sql = str_replace($tableNamePrefix, $tableName, $sql); |
| 467 | |
| 468 | try { |
| 469 | $db->query($sql); |
| 470 | } catch (Exception $e) { |
| 471 | // accept mysql error 1050: table already exists, throw otherwise |
| 472 | if (!$db->isErrNo($e, '1050')) { |
| 473 | throw $e; |
| 474 | } |
| 475 | } |
| 476 | |
| 477 | try { |
| 478 | if (ArchiveTableCreator::NUMERIC_TABLE === ArchiveTableCreator::getTypeFromTableName($tableName)) { |
| 479 | $sequence = new Sequence($tableName); |
| 480 | $sequence->create(); |
| 481 | } |
| 482 | } catch (Exception $e) { |
| 483 | } |
| 484 | } |
| 485 | |
| 486 | public function getInstalledArchiveTables() |
| 487 | { |
| 488 | $allArchiveNumeric = Db::get()->fetchCol("SHOW TABLES LIKE '" . Common::prefixTable('archive_numeric%') . "'"); |
| 489 | $allArchiveBlob = Db::get()->fetchCol("SHOW TABLES LIKE '" . Common::prefixTable('archive_blob%') ."'"); |
| 490 | |
| 491 | return array_merge($allArchiveBlob, $allArchiveNumeric); |
| 492 | } |
| 493 | |
| 494 | public function allocateNewArchiveId($numericTable) |
| 495 | { |
| 496 | $sequence = new Sequence($numericTable); |
| 497 | |
| 498 | try { |
| 499 | $idarchive = $sequence->getNextId(); |
| 500 | } catch (Exception $e) { |
| 501 | // edge case: sequence was not found, create it now |
| 502 | $sequence->create(); |
| 503 | |
| 504 | $idarchive = $sequence->getNextId(); |
| 505 | } |
| 506 | |
| 507 | return $idarchive; |
| 508 | } |
| 509 | |
| 510 | public function updateArchiveStatus($numericTable, $archiveId, $doneFlag, $value) |
| 511 | { |
| 512 | Db::query("UPDATE $numericTable SET `value` = ? WHERE idarchive = ? and `name` = ?", |
| 513 | array($value, $archiveId, $doneFlag) |
| 514 | ); |
| 515 | } |
| 516 | |
| 517 | public function insertRecord($tableName, $fields, $record, $name, $value) |
| 518 | { |
| 519 | // duplicate idarchives are Ignored, see https://github.com/piwik/piwik/issues/987 |
| 520 | $query = "INSERT IGNORE INTO " . $tableName . " (" . implode(", ", $fields) . ") |
| 521 | VALUES (?,?,?,?,?,?,?,?) ON DUPLICATE KEY UPDATE " . end($fields) . " = ?"; |
| 522 | |
| 523 | $bindSql = $record; |
| 524 | $bindSql[] = $name; |
| 525 | $bindSql[] = $value; |
| 526 | $bindSql[] = $value; |
| 527 | |
| 528 | Db::query($query, $bindSql); |
| 529 | |
| 530 | return true; |
| 531 | } |
| 532 | |
| 533 | /** |
| 534 | * Returns the site IDs for invalidated archives in an archive table. |
| 535 | * |
| 536 | * @param string $numericTable The numeric table to search through. |
| 537 | * @return int[] |
| 538 | */ |
| 539 | public function getSitesWithInvalidatedArchive($numericTable) |
| 540 | { |
| 541 | $rows = Db::fetchAll("SELECT DISTINCT idsite FROM `$numericTable` WHERE `name` LIKE 'done%' AND `value` IN (" |
| 542 | . ArchiveWriter::DONE_INVALIDATED . ")"); |
| 543 | |
| 544 | $result = array(); |
| 545 | foreach ($rows as $row) { |
| 546 | $result[] = $row['idsite']; |
| 547 | } |
| 548 | return $result; |
| 549 | } |
| 550 | |
| 551 | /** |
| 552 | * Get a list of IDs of archives that don't have any matching rows in the site table. Excludes temporary archives |
| 553 | * that may still be in use, as specified by the $oldestToKeep passed in. |
| 554 | * @param string $archiveTableName |
| 555 | * @param string $oldestToKeep Datetime string |
| 556 | * @return array of IDs |
| 557 | */ |
| 558 | public function getArchiveIdsForDeletedSites($archiveTableName) |
| 559 | { |
| 560 | $sql = "SELECT DISTINCT idsite FROM " . $archiveTableName; |
| 561 | $rows = Db::getReader()->fetchAll($sql, array()); |
| 562 | |
| 563 | if (empty($rows)) { |
| 564 | return array(); // nothing to delete |
| 565 | } |
| 566 | |
| 567 | $idSitesUsed = array_column($rows, 'idsite'); |
| 568 | |
| 569 | $model = new \Piwik\Plugins\SitesManager\Model(); |
| 570 | $idSitesExisting = $model->getSitesId(); |
| 571 | |
| 572 | $deletedSites = array_diff($idSitesUsed, $idSitesExisting); |
| 573 | |
| 574 | if (empty($deletedSites)) { |
| 575 | return array(); |
| 576 | } |
| 577 | $deletedSites = array_values($deletedSites); |
| 578 | $deletedSites = array_map('intval', $deletedSites); |
| 579 | |
| 580 | $sql = "SELECT DISTINCT idarchive FROM " . $archiveTableName . " WHERE idsite IN (".implode(',',$deletedSites).")"; |
| 581 | |
| 582 | $rows = Db::getReader()->fetchAll($sql, array()); |
| 583 | |
| 584 | return array_column($rows, 'idarchive'); |
| 585 | } |
| 586 | |
| 587 | /** |
| 588 | * Get a list of IDs of archives with segments that no longer exist in the DB. Excludes temporary archives that |
| 589 | * may still be in use, as specified by the $oldestToKeep passed in. |
| 590 | * @param string $archiveTableName |
| 591 | * @param array $segments List of segments to match against |
| 592 | * @param string $oldestToKeep Datetime string |
| 593 | * @return array With keys idarchive, name, idsite |
| 594 | */ |
| 595 | public function getArchiveIdsForSegments($archiveTableName, array $segments, $oldestToKeep) |
| 596 | { |
| 597 | $segmentClauses = []; |
| 598 | foreach ($segments as $segment) { |
| 599 | if (!empty($segment['definition'])) { |
| 600 | $segmentClauses[] = $this->getDeletedSegmentWhereClause($segment); |
| 601 | } |
| 602 | } |
| 603 | |
| 604 | if (empty($segmentClauses)) { |
| 605 | return array(); |
| 606 | } |
| 607 | |
| 608 | $segmentClauses = implode(' OR ', $segmentClauses); |
| 609 | |
| 610 | $sql = 'SELECT idarchive FROM ' . $archiveTableName |
| 611 | . ' WHERE ts_archived < ?' |
| 612 | . ' AND (' . $segmentClauses . ')'; |
| 613 | |
| 614 | $rows = Db::fetchAll($sql, array($oldestToKeep)); |
| 615 | |
| 616 | return array_column($rows, 'idarchive'); |
| 617 | } |
| 618 | |
| 619 | private function getDeletedSegmentWhereClause(array $segment) |
| 620 | { |
| 621 | $idSite = (int)$segment['enable_only_idsite']; |
| 622 | $segmentHash = $segment['hash']; |
| 623 | // Valid segment hashes are md5 strings - just confirm that it is so it's safe for SQL injection |
| 624 | if (!ctype_xdigit($segmentHash)) { |
| 625 | throw new Exception($segment . ' expected to be an md5 hash'); |
| 626 | } |
| 627 | |
| 628 | $nameClause = 'name LIKE "done' . $segmentHash . '%"'; |
| 629 | $idSiteClause = ''; |
| 630 | if ($idSite > 0) { |
| 631 | $idSiteClause = ' AND idsite = ' . $idSite; |
| 632 | } elseif (! empty($segment['idsites_to_preserve'])) { |
| 633 | // A segment for all sites was deleted, but there are segments for a single site with the same definition |
| 634 | $idSitesToPreserve = array_map('intval', $segment['idsites_to_preserve']); |
| 635 | $idSiteClause = ' AND idsite NOT IN (' . implode(',', $idSitesToPreserve) . ')'; |
| 636 | } |
| 637 | |
| 638 | return "($nameClause $idSiteClause)"; |
| 639 | } |
| 640 | |
| 641 | /** |
| 642 | * Returns the SQL condition used to find successfully completed archives that |
| 643 | * this instance is querying for. |
| 644 | */ |
| 645 | private static function getNameCondition($doneFlags, $possibleValues) |
| 646 | { |
| 647 | $allDoneFlags = "'" . implode("','", $doneFlags) . "'"; |
| 648 | |
| 649 | // create the SQL to find archives that are DONE |
| 650 | $result = "((arc1.name IN ($allDoneFlags))"; |
| 651 | |
| 652 | if (!empty($possibleValues)) { |
| 653 | $result .= " AND (arc1.value IN (" . implode(',', $possibleValues) . ")))"; |
| 654 | } |
| 655 | $result .= ')'; |
| 656 | |
| 657 | return $result; |
| 658 | } |
| 659 | |
| 660 | /** |
| 661 | * Marks an archive as in progress if it has not been already. This method must be thread |
| 662 | * safe. |
| 663 | */ |
| 664 | public function startArchive($invalidation) |
| 665 | { |
| 666 | $table = Common::prefixTable('archive_invalidations'); |
| 667 | |
| 668 | // set archive value to in progress if not set already |
| 669 | $statement = Db::query("UPDATE `$table` SET `status` = ?, ts_started = NOW() WHERE idinvalidation = ? AND status = ?", [ |
| 670 | ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, |
| 671 | $invalidation['idinvalidation'], |
| 672 | ArchiveInvalidator::INVALIDATION_STATUS_QUEUED, |
| 673 | ]); |
| 674 | |
| 675 | if ($statement->rowCount() > 0) { // if we updated, then we've marked the archive as started |
| 676 | return true; |
| 677 | } |
| 678 | |
| 679 | // if we didn't get anything, some process either got there first, OR |
| 680 | // the archive was started previously and failed in a way that kept it's done value |
| 681 | // set to DONE_IN_PROGRESS. try to acquire the lock and if acquired, archiving isn' in process |
| 682 | // so we can claim it. |
| 683 | $lock = $this->archivingStatus->acquireArchiveInProgressLock($invalidation['idsite'], $invalidation['date1'], |
| 684 | $invalidation['date2'], $invalidation['period'], $invalidation['name']); |
| 685 | if (!$lock->isLocked()) { |
| 686 | return false; // we couldn't claim the lock, archive is in progress |
| 687 | } |
| 688 | |
| 689 | // remove similar invalidations w/ lesser idinvalidation values |
| 690 | $bind = [ |
| 691 | $invalidation['idsite'], |
| 692 | $invalidation['period'], |
| 693 | $invalidation['date1'], |
| 694 | $invalidation['date2'], |
| 695 | $invalidation['name'], |
| 696 | ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, |
| 697 | ]; |
| 698 | |
| 699 | if (empty($invalidation['report'])) { |
| 700 | $reportClause = "(report IS NULL OR report = '')"; |
| 701 | } else { |
| 702 | $reportClause = "report = ?"; |
| 703 | $bind[] = $invalidation['report']; |
| 704 | } |
| 705 | |
| 706 | $sql = "DELETE FROM " . Common::prefixTable('archive_invalidations') . " WHERE idinvalidation < ? AND idsite = ? AND " |
| 707 | . "date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND $reportClause"; |
| 708 | Db::query($sql, $bind); |
| 709 | |
| 710 | return true; |
| 711 | } |
| 712 | |
| 713 | public function isSimilarArchiveInProgress($invalidation) |
| 714 | { |
| 715 | $table = Common::prefixTable('archive_invalidations'); |
| 716 | |
| 717 | $bind = [ |
| 718 | $invalidation['idsite'], |
| 719 | $invalidation['period'], |
| 720 | $invalidation['date1'], |
| 721 | $invalidation['date2'], |
| 722 | $invalidation['name'], |
| 723 | ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, |
| 724 | ]; |
| 725 | |
| 726 | if (empty($invalidation['report'])) { |
| 727 | $reportClause = "(report IS NULL OR report = '')"; |
| 728 | } else { |
| 729 | $reportClause = "report = ?"; |
| 730 | $bind[] = $invalidation['report']; |
| 731 | } |
| 732 | |
| 733 | $sql = "SELECT idinvalidation FROM `$table` WHERE idsite = ? AND `period` = ? AND date1 = ? AND date2 = ? AND `name` = ? AND `status` = ? AND ts_started IS NOT NULL AND $reportClause LIMIT 1"; |
| 734 | $result = Db::fetchOne($sql, $bind); |
| 735 | |
| 736 | return !empty($result); |
| 737 | } |
| 738 | |
| 739 | /** |
| 740 | * Gets the next invalidated archive that should be archived in a table. |
| 741 | * |
| 742 | * @param int $idSite |
| 743 | * @param string $archivingStartTime |
| 744 | * @param int[]|null $idInvalidationsToExclude |
| 745 | * @param bool $useLimit Whether to limit the result set to one result or not. Used in tests only. |
| 746 | */ |
| 747 | public function getNextInvalidatedArchive($idSite, $archivingStartTime, $idInvalidationsToExclude = null, $useLimit = true) |
| 748 | { |
| 749 | $table = Common::prefixTable('archive_invalidations'); |
| 750 | $sql = "SELECT idinvalidation, idarchive, idsite, date1, date2, period, `name`, report, ts_invalidated |
| 751 | FROM `$table` |
| 752 | WHERE idsite = ? AND status != ? AND ts_invalidated <= ?"; |
| 753 | $bind = [ |
| 754 | $idSite, |
| 755 | ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, |
| 756 | $archivingStartTime, |
| 757 | ]; |
| 758 | |
| 759 | if (!empty($idInvalidationsToExclude)) { |
| 760 | $idInvalidationsToExclude = array_map('intval', $idInvalidationsToExclude); |
| 761 | $sql .= " AND idinvalidation NOT IN (" . implode(',', $idInvalidationsToExclude) . ')'; |
| 762 | } |
| 763 | |
| 764 | // NOTE: order here is very important to ensure we process lower period archives first, and general 'all' archives before |
| 765 | // segment archives, and so we use the latest idinvalidation |
| 766 | $sql .= " ORDER BY date1 DESC, period ASC, CHAR_LENGTH(name) ASC, idinvalidation DESC"; |
| 767 | |
| 768 | if ($useLimit) { |
| 769 | $sql .= " LIMIT 1"; |
| 770 | return Db::fetchRow($sql, $bind); |
| 771 | } else { |
| 772 | return Db::fetchAll($sql, $bind); |
| 773 | } |
| 774 | } |
| 775 | |
| 776 | public function deleteInvalidations($archiveInvalidations) |
| 777 | { |
| 778 | $ids = array_column($archiveInvalidations, 'idinvalidation'); |
| 779 | $ids = array_map('intval', $ids); |
| 780 | |
| 781 | $table = Common::prefixTable('archive_invalidations'); |
| 782 | $sql = "DELETE FROM `$table` WHERE idinvalidation IN (" . implode(', ', $ids) . ")"; |
| 783 | |
| 784 | Db::query($sql); |
| 785 | } |
| 786 | |
| 787 | public function removeInvalidationsLike($idSite, $start) |
| 788 | { |
| 789 | $idSitesClause = $this->getRemoveInvalidationsIdSitesClause($idSite); |
| 790 | |
| 791 | $table = Common::prefixTable('archive_invalidations'); |
| 792 | $sql = "DELETE FROM `$table` WHERE $idSitesClause `name` LIKE ?"; |
| 793 | |
| 794 | Db::query($sql, ['done%.' . str_replace('_', "\\_", $start)]); |
| 795 | } |
| 796 | |
| 797 | public function removeInvalidations($idSite, $plugin, $report) |
| 798 | { |
| 799 | $idSitesClause = $this->getRemoveInvalidationsIdSitesClause($idSite); |
| 800 | |
| 801 | $table = Common::prefixTable('archive_invalidations'); |
| 802 | $sql = "DELETE FROM `$table` WHERE $idSitesClause `name` LIKE ? AND report = ?"; |
| 803 | |
| 804 | Db::query($sql, ['done%.' . str_replace('_', "\\_", $plugin), $report]); |
| 805 | } |
| 806 | |
| 807 | public function isArchiveAlreadyInProgress($invalidatedArchive) |
| 808 | { |
| 809 | $table = Common::prefixTable('archive_invalidations'); |
| 810 | |
| 811 | $bind = [ |
| 812 | $invalidatedArchive['idsite'], |
| 813 | $invalidatedArchive['date1'], |
| 814 | $invalidatedArchive['date2'], |
| 815 | $invalidatedArchive['period'], |
| 816 | $invalidatedArchive['name'], |
| 817 | ]; |
| 818 | |
| 819 | $reportClause = "(report = '' OR report IS NULL)"; |
| 820 | if (!empty($invalidatedArchive['report'])) { |
| 821 | $reportClause = "report = ?"; |
| 822 | $bind[] = $invalidatedArchive['report']; |
| 823 | } |
| 824 | |
| 825 | $sql = "SELECT MAX(idinvalidation) FROM `$table` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND status = 1 AND $reportClause"; |
| 826 | |
| 827 | $inProgressInvalidation = Db::fetchOne($sql, $bind); |
| 828 | return $inProgressInvalidation; |
| 829 | } |
| 830 | |
| 831 | /** |
| 832 | * Returns true if there is an archive that exists that can be used when aggregating an archive for $period. |
| 833 | * |
| 834 | * @param $idSite |
| 835 | * @param Period $period |
| 836 | * @return bool |
| 837 | * @throws Exception |
| 838 | */ |
| 839 | public function hasChildArchivesInPeriod($idSite, Period $period) |
| 840 | { |
| 841 | $date = $period->getDateStart(); |
| 842 | while ($date->isEarlier($period->getDateEnd()->addPeriod(1, 'month'))) { |
| 843 | $archiveTable = ArchiveTableCreator::getNumericTable($date); |
| 844 | |
| 845 | // we look for any archive that can be used to compute this one. this includes invalidated archives, since it is possible |
| 846 | // under certain circumstances for them to exist, when archiving a higher period that includes them. the main example being |
| 847 | // the GoogleAnalyticsImporter which disallows the recomputation of invalidated archives for imported data, since that would |
| 848 | // essentially get rid of the imported data. |
| 849 | $usableDoneFlags = [ArchiveWriter::DONE_OK, ArchiveWriter::DONE_INVALIDATED, ArchiveWriter::DONE_PARTIAL, ArchiveWriter::DONE_OK_TEMPORARY]; |
| 850 | |
| 851 | $sql = "SELECT idarchive |
| 852 | FROM `$archiveTable` |
| 853 | WHERE idsite = ? AND date1 >= ? AND date2 <= ? AND period < ? AND `name` LIKE 'done%' AND `value` IN (" . implode(', ', $usableDoneFlags) . ") |
| 854 | LIMIT 1"; |
| 855 | $bind = [$idSite, $period->getDateStart()->getDatetime(), $period->getDateEnd()->getDatetime(), $period->getId()]; |
| 856 | |
| 857 | $result = (bool) Db::fetchOne($sql, $bind); |
| 858 | if ($result) { |
| 859 | return true; |
| 860 | } |
| 861 | |
| 862 | $date = $date->addPeriod(1, 'month'); // move to next archive table |
| 863 | } |
| 864 | return false; |
| 865 | } |
| 866 | |
| 867 | /** |
| 868 | * Returns true if any invalidations exists for the given |
| 869 | * $idsite and $doneFlag (name column) for the $period. |
| 870 | * |
| 871 | * @param mixed $idSite |
| 872 | * @param Period $period |
| 873 | * @param mixed $doneFlag |
| 874 | * @param mixed $report |
| 875 | * @return bool |
| 876 | * @throws Exception |
| 877 | */ |
| 878 | public function hasInvalidationForPeriodAndName($idSite, Period $period, $doneFlag, $report = null) |
| 879 | { |
| 880 | $table = Common::prefixTable('archive_invalidations'); |
| 881 | |
| 882 | if (empty($report)) { |
| 883 | $sql = "SELECT idinvalidation FROM `$table` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND `report` IS NULL LIMIT 1"; |
| 884 | } else { |
| 885 | $sql = "SELECT idinvalidation FROM `$table` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND `report` = ? LIMIT 1"; |
| 886 | } |
| 887 | |
| 888 | $bind = [ |
| 889 | $idSite, |
| 890 | $period->getDateStart()->toString(), |
| 891 | $period->getDateEnd()->toString(), |
| 892 | $period->getId(), |
| 893 | $doneFlag |
| 894 | ]; |
| 895 | |
| 896 | if (!empty($report)) { |
| 897 | $bind[] = $report; |
| 898 | } |
| 899 | |
| 900 | $idInvalidation = Db::fetchOne($sql, $bind); |
| 901 | |
| 902 | if (empty($idInvalidation)) { |
| 903 | return false; |
| 904 | } |
| 905 | |
| 906 | return true; |
| 907 | } |
| 908 | |
| 909 | public function deleteInvalidationsForSites(array $idSites) |
| 910 | { |
| 911 | $idSites = array_map('intval', $idSites); |
| 912 | |
| 913 | $table = Common::prefixTable('archive_invalidations'); |
| 914 | $sql = "DELETE FROM `$table` WHERE idsite IN (" . implode(',', $idSites) . ")"; |
| 915 | |
| 916 | Db::query($sql); |
| 917 | } |
| 918 | |
| 919 | public function deleteInvalidationsForDeletedSites() |
| 920 | { |
| 921 | $siteTable = Common::prefixTable('site'); |
| 922 | $table = Common::prefixTable('archive_invalidations'); |
| 923 | $sql = "DELETE a FROM `$table` a LEFT JOIN `$siteTable` s ON a.idsite = s.idsite WHERE s.idsite IS NULL"; |
| 924 | Db::query($sql); |
| 925 | } |
| 926 | |
| 927 | private function getRemoveInvalidationsIdSitesClause($idSite) |
| 928 | { |
| 929 | if ($idSite === 'all') { |
| 930 | return ''; |
| 931 | } |
| 932 | |
| 933 | $idSites = is_array($idSite) ? $idSite : [$idSite]; |
| 934 | $idSites = array_map('intval', $idSites); |
| 935 | $idSitesStr = implode(',', $idSites); |
| 936 | |
| 937 | return "idsite IN ($idSitesStr) AND"; |
| 938 | } |
| 939 | |
| 940 | public function releaseInProgressInvalidation($idinvalidation) |
| 941 | { |
| 942 | $table = Common::prefixTable('archive_invalidations'); |
| 943 | $sql = "UPDATE $table SET status = " . ArchiveInvalidator::INVALIDATION_STATUS_QUEUED . ", ts_started = NULL WHERE idinvalidation = ?"; |
| 944 | Db::query($sql, [$idinvalidation]); |
| 945 | } |
| 946 | |
| 947 | public function resetFailedArchivingJobs() |
| 948 | { |
| 949 | $table = Common::prefixTable('archive_invalidations'); |
| 950 | $sql = "UPDATE $table SET status = ? WHERE status = ? AND (ts_started IS NULL OR ts_started < ?)"; |
| 951 | |
| 952 | $bind = [ |
| 953 | ArchiveInvalidator::INVALIDATION_STATUS_QUEUED, |
| 954 | ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, |
| 955 | Date::now()->subDay(1)->getDatetime(), |
| 956 | ]; |
| 957 | |
| 958 | $query = Db::query($sql, $bind); |
| 959 | return $query->rowCount(); |
| 960 | } |
| 961 | |
| 962 | private function isCutOffGroupConcatResult($pair) |
| 963 | { |
| 964 | $position = strpos($pair, '.'); |
| 965 | return $position === false || $position === strlen($pair) - 1; |
| 966 | } |
| 967 | } |
| 968 |