LogQueryBuilder
2 years ago
Actions.php
2 years ago
ArchiveSelector.php
2 years ago
ArchiveTableCreator.php
2 years ago
ArchiveTableDao.php
2 years ago
ArchiveWriter.php
2 years ago
ArchivingDbAdapter.php
2 years ago
LogAggregator.php
2 years ago
LogQueryBuilder.php
2 years ago
LogTableTemporary.php
2 years ago
Model.php
2 years ago
RawLogDao.php
2 years ago
TableMetadata.php
2 years ago
Model.php
745 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | */ |
| 9 | namespace Piwik\DataAccess; |
| 10 | |
| 11 | use Exception; |
| 12 | use Piwik\Archive\ArchiveInvalidator; |
| 13 | use Piwik\ArchiveProcessor\Parameters; |
| 14 | use Piwik\ArchiveProcessor\Rules; |
| 15 | use Piwik\Common; |
| 16 | use Piwik\Container\StaticContainer; |
| 17 | use Piwik\Date; |
| 18 | use Piwik\Db; |
| 19 | use Piwik\DbHelper; |
| 20 | use Piwik\Period; |
| 21 | use Piwik\Segment; |
| 22 | use Piwik\Sequence; |
| 23 | use Piwik\SettingsServer; |
| 24 | use Piwik\Site; |
| 25 | use Piwik\Log\LoggerInterface; |
| 26 | /** |
| 27 | * Cleans up outdated archives |
| 28 | */ |
| 29 | class Model |
| 30 | { |
| 31 | /** |
| 32 | * @var LoggerInterface |
| 33 | */ |
| 34 | private $logger; |
| 35 | public function __construct(LoggerInterface $logger = null) |
| 36 | { |
| 37 | $this->logger = $logger ?: StaticContainer::get(LoggerInterface::class); |
| 38 | } |
| 39 | /** |
| 40 | * Returns the archives IDs that have already been invalidated and have been since re-processed. |
| 41 | * |
| 42 | * These archives { archive name (includes segment hash) , idsite, date, period } will be deleted. |
| 43 | * |
| 44 | * @param string $archiveTable |
| 45 | * @param array $idSites |
| 46 | * @param bool $setGroupContentMaxLen for tests only |
| 47 | * @return array |
| 48 | * @throws Exception |
| 49 | */ |
| 50 | public function getInvalidatedArchiveIdsSafeToDelete($archiveTable, $setGroupContentMaxLen = true) |
| 51 | { |
| 52 | if ($setGroupContentMaxLen) { |
| 53 | try { |
| 54 | Db::get()->query('SET SESSION group_concat_max_len=' . 128 * 1024); |
| 55 | } catch (\Exception $ex) { |
| 56 | $this->logger->info("Could not set group_concat_max_len MySQL session variable."); |
| 57 | } |
| 58 | } |
| 59 | $sql = "SELECT idsite, date1, date2, period, name,\n GROUP_CONCAT(idarchive, '.', value ORDER BY ts_archived DESC) as archives\n FROM `{$archiveTable}`\n WHERE name LIKE 'done%'\n AND `value` NOT IN (" . \Piwik\DataAccess\ArchiveWriter::DONE_ERROR . ")\n GROUP BY idsite, date1, date2, period, name HAVING count(*) > 1"; |
| 60 | $archiveIds = array(); |
| 61 | $rows = Db::fetchAll($sql); |
| 62 | foreach ($rows as $row) { |
| 63 | $duplicateArchives = explode(',', $row['archives']); |
| 64 | // do not consider purging partial archives, if they are the latest archive, |
| 65 | // and we don't want to delete the latest archive if it is usable |
| 66 | while (!empty($duplicateArchives)) { |
| 67 | $pair = $duplicateArchives[0]; |
| 68 | if ($this->isCutOffGroupConcatResult($pair)) { |
| 69 | // can occur if the GROUP_CONCAT value is cut off |
| 70 | break; |
| 71 | } |
| 72 | [$idarchive, $value] = explode('.', $pair); |
| 73 | array_shift($duplicateArchives); |
| 74 | if ($value != \Piwik\DataAccess\ArchiveWriter::DONE_PARTIAL) { |
| 75 | break; |
| 76 | } |
| 77 | } |
| 78 | // if there is more than one archive, the older invalidated ones can be deleted |
| 79 | if (!empty($duplicateArchives)) { |
| 80 | foreach ($duplicateArchives as $pair) { |
| 81 | if ($this->isCutOffGroupConcatResult($pair)) { |
| 82 | $this->logger->info("GROUP_CONCAT cut off the query result, you may have to purge archives again."); |
| 83 | break; |
| 84 | } |
| 85 | [$idarchive, $value] = explode('.', $pair); |
| 86 | $archiveIds[] = $idarchive; |
| 87 | // does not matter what the value is, the latest is usable so older archives can be purged |
| 88 | } |
| 89 | } |
| 90 | } |
| 91 | return $archiveIds; |
| 92 | } |
| 93 | public function updateArchiveAsInvalidated($archiveTable, $idSites, $allPeriodsToInvalidate, Segment $segment = null, $forceInvalidateNonexistentRanges = false, $name = null) |
| 94 | { |
| 95 | if (empty($idSites)) { |
| 96 | return 0; |
| 97 | } |
| 98 | // select all idarchive/name pairs we want to invalidate |
| 99 | $sql = "SELECT idarchive, idsite, period, date1, date2, `name`, `value`\n FROM `{$archiveTable}`\n WHERE idsite IN (" . implode(',', $idSites) . ") AND value <> " . \Piwik\DataAccess\ArchiveWriter::DONE_PARTIAL; |
| 100 | $periodCondition = ''; |
| 101 | if (!empty($allPeriodsToInvalidate)) { |
| 102 | $periodCondition .= " AND ("; |
| 103 | $isFirst = true; |
| 104 | /** @var Period $period */ |
| 105 | foreach ($allPeriodsToInvalidate as $period) { |
| 106 | if ($isFirst) { |
| 107 | $isFirst = false; |
| 108 | } else { |
| 109 | $periodCondition .= " OR "; |
| 110 | } |
| 111 | if ($period->getLabel() == 'range') { |
| 112 | // for ranges, we delete all ranges that contain the given date(s) |
| 113 | $periodCondition .= "(period = " . (int) $period->getId() . " AND date2 >= '" . $period->getDateStart()->getDatetime() . "' AND date1 <= '" . $period->getDateEnd()->getDatetime() . "')"; |
| 114 | } else { |
| 115 | $periodCondition .= "(period = " . (int) $period->getId() . " AND date1 = '" . $period->getDateStart()->getDatetime() . "'" . " AND date2 = '" . $period->getDateEnd()->getDatetime() . "')"; |
| 116 | } |
| 117 | } |
| 118 | $periodCondition .= ")"; |
| 119 | } |
| 120 | $sql .= $periodCondition; |
| 121 | if (!empty($name)) { |
| 122 | if (strpos($name, '.') !== false) { |
| 123 | [$plugin, $name] = explode('.', $name, 2); |
| 124 | } else { |
| 125 | $plugin = $name; |
| 126 | $name = null; |
| 127 | } |
| 128 | } |
| 129 | if (empty($plugin)) { |
| 130 | $doneFlag = Rules::getDoneFlagArchiveContainsAllPlugins($segment ?: new Segment('', [])); |
| 131 | } else { |
| 132 | $doneFlag = Rules::getDoneFlagArchiveContainsOnePlugin($segment ?: new Segment('', []), $plugin); |
| 133 | } |
| 134 | $nameCondition = "name LIKE '{$doneFlag}%'"; |
| 135 | $sql .= " AND {$nameCondition}"; |
| 136 | $idArchives = []; |
| 137 | $archivesToInvalidate = []; |
| 138 | // update each archive as invalidated (but only for full archives or plugin archives, not for partial archives. |
| 139 | // DONE_INVALIDATED also implies that an archive is whole and not partial, and we want to avoid that.) |
| 140 | if (empty($name)) { |
| 141 | $archivesToInvalidate = Db::fetchAll($sql); |
| 142 | $idArchives = array_column($archivesToInvalidate, 'idarchive'); |
| 143 | if (!empty($idArchives)) { |
| 144 | $idArchives = array_map('intval', $idArchives); |
| 145 | $sql = "UPDATE `{$archiveTable}` SET `value` = " . \Piwik\DataAccess\ArchiveWriter::DONE_INVALIDATED . " WHERE idarchive IN (" . implode(',', $idArchives) . ") AND {$nameCondition}"; |
| 146 | Db::query($sql); |
| 147 | } |
| 148 | } |
| 149 | // we add every archive we need to invalidate + the archives that do not already exist to archive_invalidations. |
| 150 | // except for archives that are DONE_IN_PROGRESS. |
| 151 | $archivesToCreateInvalidationRowsFor = []; |
| 152 | foreach ($archivesToInvalidate as $row) { |
| 153 | $archivesToCreateInvalidationRowsFor[$row['idsite']][$row['period']][$row['date1']][$row['date2']][$row['name']] = $row['idarchive']; |
| 154 | } |
| 155 | $now = Date::now()->getDatetime(); |
| 156 | $existingInvalidations = $this->getExistingInvalidations($idSites, $periodCondition, $nameCondition); |
| 157 | $hashesOfAllSegmentsToArchiveInCoreArchive = Rules::getSegmentsToProcess($idSites); |
| 158 | $hashesOfAllSegmentsToArchiveInCoreArchive = array_map(function ($definition) { |
| 159 | return Segment::getSegmentHash($definition); |
| 160 | }, $hashesOfAllSegmentsToArchiveInCoreArchive); |
| 161 | $dummyArchives = []; |
| 162 | foreach ($idSites as $idSite) { |
| 163 | try { |
| 164 | $siteCreationTime = Site::getCreationDateFor($idSite); |
| 165 | } catch (\Exception $ex) { |
| 166 | continue; |
| 167 | } |
| 168 | $siteCreationTime = Date::factory($siteCreationTime); |
| 169 | foreach ($allPeriodsToInvalidate as $period) { |
| 170 | if ($period->getLabel() == 'range' && !$forceInvalidateNonexistentRanges) { |
| 171 | continue; |
| 172 | // range |
| 173 | } |
| 174 | if ($period->getDateEnd()->isEarlier($siteCreationTime)) { |
| 175 | continue; |
| 176 | // don't add entries if it is before the time the site was created |
| 177 | } |
| 178 | $date1 = $period->getDateStart()->toString(); |
| 179 | $date2 = $period->getDateEnd()->toString(); |
| 180 | // we insert rows for the doneFlag we want to invalidate + any others we invalidated when doing the LIKE above. |
| 181 | // if we invalidated something in the archive tables, we want to make sure it appears in the invalidation queue, |
| 182 | // so we'll eventually reprocess it. |
| 183 | $doneFlagsFound = $archivesToCreateInvalidationRowsFor[$idSite][$period->getId()][$date1][$date2] ?? []; |
| 184 | $doneFlagsFound = array_keys($doneFlagsFound); |
| 185 | $doneFlagsToCheck = array_merge([$doneFlag], $doneFlagsFound); |
| 186 | $doneFlagsToCheck = array_unique($doneFlagsToCheck); |
| 187 | foreach ($doneFlagsToCheck as $doneFlagToCheck) { |
| 188 | $key = $this->makeExistingInvalidationArrayKey($idSite, $date1, $date2, $period->getId(), $doneFlagToCheck, $name); |
| 189 | if (!empty($existingInvalidations[$key])) { |
| 190 | continue; |
| 191 | // avoid adding duplicates where possible |
| 192 | } |
| 193 | $hash = $this->getHashFromDoneFlag($doneFlagToCheck); |
| 194 | if ($doneFlagToCheck != $doneFlag && (empty($hash) || !in_array($hash, $hashesOfAllSegmentsToArchiveInCoreArchive) || strpos($doneFlagToCheck, '.') !== false)) { |
| 195 | continue; |
| 196 | // the done flag is for a segment that is not auto archive or a plugin specific archive, so we don't want to process it. |
| 197 | } |
| 198 | $idArchive = $archivesToCreateInvalidationRowsFor[$idSite][$period->getId()][$date1][$date2][$doneFlagToCheck] ?? null; |
| 199 | $dummyArchives[] = ['idarchive' => $idArchive, 'name' => $doneFlagToCheck, 'report' => $name, 'idsite' => $idSite, 'date1' => $period->getDateStart()->getDatetime(), 'date2' => $period->getDateEnd()->getDatetime(), 'period' => $period->getId(), 'ts_invalidated' => $now]; |
| 200 | } |
| 201 | } |
| 202 | } |
| 203 | if (!empty($dummyArchives)) { |
| 204 | $fields = ['idarchive', 'name', 'report', 'idsite', 'date1', 'date2', 'period', 'ts_invalidated']; |
| 205 | Db\BatchInsert::tableInsertBatch(Common::prefixTable('archive_invalidations'), $fields, $dummyArchives); |
| 206 | } |
| 207 | return count($idArchives); |
| 208 | } |
| 209 | private function getExistingInvalidations($idSites, $periodCondition, $nameCondition) |
| 210 | { |
| 211 | $table = Common::prefixTable('archive_invalidations'); |
| 212 | $idSites = array_map('intval', $idSites); |
| 213 | $sql = "SELECT idsite, date1, date2, period, name, report, COUNT(*) as `count` FROM `{$table}`\n WHERE idsite IN (" . implode(',', $idSites) . ") AND status = " . ArchiveInvalidator::INVALIDATION_STATUS_QUEUED . "\n {$periodCondition} AND {$nameCondition}\n GROUP BY idsite, date1, date2, period, name"; |
| 214 | $rows = Db::fetchAll($sql); |
| 215 | $invalidations = []; |
| 216 | foreach ($rows as $row) { |
| 217 | $key = $this->makeExistingInvalidationArrayKey($row['idsite'], $row['date1'], $row['date2'], $row['period'], $row['name'], $row['report']); |
| 218 | $invalidations[$key] = $row['count']; |
| 219 | } |
| 220 | return $invalidations; |
| 221 | } |
| 222 | private function makeExistingInvalidationArrayKey($idSite, $date1, $date2, $period, $name, $report) |
| 223 | { |
| 224 | return implode('.', [$idSite, $date1, $date2, $period, $name, $report]); |
| 225 | } |
| 226 | /** |
| 227 | * @param string $archiveTable Prefixed table name |
| 228 | * @param int[] $idSites |
| 229 | * @param string[][] $datesByPeriodType |
| 230 | * @param Segment $segment |
| 231 | * @return \Zend_Db_Statement |
| 232 | * @throws Exception |
| 233 | */ |
| 234 | public function updateRangeArchiveAsInvalidated($archiveTable, $idSites, $allPeriodsToInvalidate, Segment $segment = null) |
| 235 | { |
| 236 | if (empty($idSites)) { |
| 237 | return; |
| 238 | } |
| 239 | $bind = array(); |
| 240 | $periodConditions = array(); |
| 241 | if (!empty($allPeriodsToInvalidate)) { |
| 242 | foreach ($allPeriodsToInvalidate as $period) { |
| 243 | $dateConditions = array(); |
| 244 | /** @var Period $period */ |
| 245 | $dateConditions[] = "(date1 <= ? AND ? <= date2)"; |
| 246 | $bind[] = $period->getDateStart()->getDatetime(); |
| 247 | $bind[] = $period->getDateEnd()->getDatetime(); |
| 248 | $dateConditionsSql = implode(" OR ", $dateConditions); |
| 249 | $periodConditions[] = "(period = 5 AND ({$dateConditionsSql}))"; |
| 250 | } |
| 251 | } |
| 252 | if ($segment) { |
| 253 | $nameCondition = "name LIKE '" . Rules::getDoneFlagArchiveContainsAllPlugins($segment) . "%'"; |
| 254 | } else { |
| 255 | $nameCondition = "name LIKE 'done%'"; |
| 256 | } |
| 257 | $sql = "UPDATE {$archiveTable} SET value = " . \Piwik\DataAccess\ArchiveWriter::DONE_INVALIDATED . " WHERE {$nameCondition}\n AND idsite IN (" . implode(", ", $idSites) . ")\n AND (" . implode(" OR ", $periodConditions) . ")"; |
| 258 | return Db::query($sql, $bind); |
| 259 | } |
| 260 | public function getTemporaryArchivesOlderThan($archiveTable, $purgeArchivesOlderThan) |
| 261 | { |
| 262 | $query = "SELECT idarchive FROM " . $archiveTable . "\n WHERE name LIKE 'done%'\n AND (( value = " . \Piwik\DataAccess\ArchiveWriter::DONE_OK_TEMPORARY . "\n AND ts_archived < ?)\n OR value = " . \Piwik\DataAccess\ArchiveWriter::DONE_ERROR . ")"; |
| 263 | return Db::fetchAll($query, array($purgeArchivesOlderThan)); |
| 264 | } |
| 265 | public function deleteArchivesWithPeriod($numericTable, $blobTable, $period, $date) |
| 266 | { |
| 267 | if (SettingsServer::isArchivePhpTriggered()) { |
| 268 | StaticContainer::get(LoggerInterface::class)->info('deleteArchivesWithPeriod: ' . $numericTable . ' with period = ' . $period . ' and date = ' . $date); |
| 269 | } |
| 270 | $query = "DELETE FROM %s WHERE period = ? AND ts_archived < ?"; |
| 271 | $bind = array($period, $date); |
| 272 | $queryObj = Db::query(sprintf($query, $numericTable), $bind); |
| 273 | $deletedRows = $queryObj->rowCount(); |
| 274 | try { |
| 275 | $queryObj = Db::query(sprintf($query, $blobTable), $bind); |
| 276 | $deletedRows += $queryObj->rowCount(); |
| 277 | } catch (Exception $e) { |
| 278 | // Individual blob tables could be missing |
| 279 | $this->logger->debug("Unable to delete archives by period from {blobTable}.", array('blobTable' => $blobTable, 'exception' => $e)); |
| 280 | } |
| 281 | return $deletedRows; |
| 282 | } |
| 283 | public function deleteArchiveIds($numericTable, $blobTable, $idsToDelete) |
| 284 | { |
| 285 | $idsToDelete = array_values($idsToDelete); |
| 286 | $idsToDelete = array_map('intval', $idsToDelete); |
| 287 | $query = "DELETE FROM %s WHERE idarchive IN (" . implode(',', $idsToDelete) . ")"; |
| 288 | $queryObj = Db::query(sprintf($query, $numericTable), array()); |
| 289 | $deletedRows = $queryObj->rowCount(); |
| 290 | try { |
| 291 | $queryObj = Db::query(sprintf($query, $blobTable), array()); |
| 292 | $deletedRows += $queryObj->rowCount(); |
| 293 | } catch (Exception $e) { |
| 294 | // Individual blob tables could be missing |
| 295 | $this->logger->debug("Unable to delete archive IDs from {blobTable}.", array('blobTable' => $blobTable, 'exception' => $e)); |
| 296 | } |
| 297 | return $deletedRows; |
| 298 | } |
| 299 | public function deleteOlderArchives(Parameters $params, $name, $tsArchived, $idArchive) |
| 300 | { |
| 301 | $dateStart = $params->getPeriod()->getDateStart(); |
| 302 | $dateEnd = $params->getPeriod()->getDateEnd(); |
| 303 | $numericTable = \Piwik\DataAccess\ArchiveTableCreator::getNumericTable($dateStart); |
| 304 | $blobTable = \Piwik\DataAccess\ArchiveTableCreator::getBlobTable($dateStart); |
| 305 | $sql = "SELECT idarchive FROM `{$numericTable}` WHERE idsite = ? AND date1 = ? AND date2 = ? AND period = ? AND name = ? AND ts_archived < ? AND idarchive < ?"; |
| 306 | $idArchives = Db::fetchAll($sql, [$params->getSite()->getId(), $dateStart->getDatetime(), $dateEnd->getDatetime(), $params->getPeriod()->getId(), $name, $tsArchived, $idArchive]); |
| 307 | $idArchives = array_column($idArchives, 'idarchive'); |
| 308 | if (empty($idArchives)) { |
| 309 | return; |
| 310 | } |
| 311 | if (SettingsServer::isArchivePhpTriggered()) { |
| 312 | StaticContainer::get(LoggerInterface::class)->info('deleteOlderArchives with ' . $params . ', name = ' . $name . ', ts_archived < ' . $tsArchived . ', idarchive < ' . $idArchive); |
| 313 | } |
| 314 | $this->deleteArchiveIds($numericTable, $blobTable, $idArchives); |
| 315 | } |
| 316 | public function getArchiveIdAndVisits($numericTable, $idSite, $period, $dateStartIso, $dateEndIso, $minDatetimeIsoArchiveProcessedUTC, $doneFlags, $doneFlagValues = null) |
| 317 | { |
| 318 | $bindSQL = array($idSite, $dateStartIso, $dateEndIso, $period); |
| 319 | $sqlWhereArchiveName = self::getNameCondition($doneFlags, $doneFlagValues); |
| 320 | $timeStampWhere = ''; |
| 321 | if ($minDatetimeIsoArchiveProcessedUTC) { |
| 322 | $timeStampWhere = " AND arc1.ts_archived >= ? "; |
| 323 | $bindSQL[] = $minDatetimeIsoArchiveProcessedUTC; |
| 324 | } |
| 325 | // 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. |
| 326 | $sqlQuery = "SELECT arc1.idarchive, arc1.value, arc1.name, arc1.ts_archived, arc1.date1 as startDate, arc2.value as " . \Piwik\DataAccess\ArchiveSelector::NB_VISITS_RECORD_LOOKED_UP . ", arc3.value as " . \Piwik\DataAccess\ArchiveSelector::NB_VISITS_CONVERTED_RECORD_LOOKED_UP . "\n FROM {$numericTable} arc1\n LEFT JOIN {$numericTable} arc2 on arc2.idarchive = arc1.idarchive and (arc2.name = '" . \Piwik\DataAccess\ArchiveSelector::NB_VISITS_RECORD_LOOKED_UP . "')\n LEFT JOIN {$numericTable} arc3 on arc3.idarchive = arc1.idarchive and (arc3.name = '" . \Piwik\DataAccess\ArchiveSelector::NB_VISITS_CONVERTED_RECORD_LOOKED_UP . "')\n WHERE arc1.idsite = ?\n AND arc1.date1 = ?\n AND arc1.date2 = ?\n AND arc1.period = ?\n AND ({$sqlWhereArchiveName})\n {$timeStampWhere}\n ORDER BY arc1.ts_archived DESC, arc1.idarchive DESC"; |
| 327 | $results = Db::fetchAll($sqlQuery, $bindSQL); |
| 328 | return $results; |
| 329 | } |
| 330 | public function createArchiveTable($tableName, $tableNamePrefix) |
| 331 | { |
| 332 | $db = Db::get(); |
| 333 | $sql = DbHelper::getTableCreateSql($tableNamePrefix); |
| 334 | // replace table name template by real name |
| 335 | $tableNamePrefix = Common::prefixTable($tableNamePrefix); |
| 336 | $sql = str_replace($tableNamePrefix, $tableName, $sql); |
| 337 | try { |
| 338 | $db->query($sql); |
| 339 | } catch (Exception $e) { |
| 340 | // accept mysql error 1050: table already exists, throw otherwise |
| 341 | if (!$db->isErrNo($e, '1050')) { |
| 342 | throw $e; |
| 343 | } |
| 344 | } |
| 345 | try { |
| 346 | if (\Piwik\DataAccess\ArchiveTableCreator::NUMERIC_TABLE === \Piwik\DataAccess\ArchiveTableCreator::getTypeFromTableName($tableName)) { |
| 347 | $sequence = new Sequence($tableName); |
| 348 | $sequence->create(); |
| 349 | } |
| 350 | } catch (Exception $e) { |
| 351 | } |
| 352 | } |
| 353 | public function getInstalledArchiveTables() |
| 354 | { |
| 355 | $allArchiveNumeric = Db::get()->fetchCol("SHOW TABLES LIKE '" . Common::prefixTable('archive_numeric%') . "'"); |
| 356 | $allArchiveBlob = Db::get()->fetchCol("SHOW TABLES LIKE '" . Common::prefixTable('archive_blob%') . "'"); |
| 357 | return array_merge($allArchiveBlob, $allArchiveNumeric); |
| 358 | } |
| 359 | public function allocateNewArchiveId($numericTable) |
| 360 | { |
| 361 | $sequence = new Sequence($numericTable); |
| 362 | try { |
| 363 | $idarchive = $sequence->getNextId(); |
| 364 | } catch (Exception $e) { |
| 365 | // edge case: sequence was not found, create it now |
| 366 | try { |
| 367 | $sequence->create(); |
| 368 | } catch (Exception $ex) { |
| 369 | // Ignore duplicate entry error, as that means another request might have already created the sequence |
| 370 | if (!Db::get()->isErrNo($ex, \Piwik\Updater\Migration\Db::ERROR_CODE_DUPLICATE_ENTRY)) { |
| 371 | throw $ex; |
| 372 | } |
| 373 | } |
| 374 | $idarchive = $sequence->getNextId(); |
| 375 | } |
| 376 | return $idarchive; |
| 377 | } |
| 378 | public function updateArchiveStatus($numericTable, $archiveId, $doneFlag, $value) |
| 379 | { |
| 380 | Db::query("UPDATE {$numericTable} SET `value` = ? WHERE idarchive = ? and `name` = ?", array($value, $archiveId, $doneFlag)); |
| 381 | } |
| 382 | public function insertRecord($tableName, $fields, $record, $name, $value) |
| 383 | { |
| 384 | // duplicate idarchives are Ignored, see https://github.com/piwik/piwik/issues/987 |
| 385 | $query = "INSERT IGNORE INTO " . $tableName . " (" . implode(", ", $fields) . ")\n VALUES (?,?,?,?,?,?,?,?) ON DUPLICATE KEY UPDATE " . end($fields) . " = ?"; |
| 386 | $bindSql = $record; |
| 387 | $bindSql[] = $name; |
| 388 | $bindSql[] = $value; |
| 389 | $bindSql[] = $value; |
| 390 | Db::query($query, $bindSql); |
| 391 | return true; |
| 392 | } |
| 393 | /** |
| 394 | * Returns the site IDs for invalidated archives in an archive table. |
| 395 | * |
| 396 | * @param string $numericTable The numeric table to search through. |
| 397 | * @return int[] |
| 398 | */ |
| 399 | public function getSitesWithInvalidatedArchive($numericTable) |
| 400 | { |
| 401 | $rows = Db::fetchAll("SELECT DISTINCT idsite FROM `{$numericTable}` WHERE `name` LIKE 'done%' AND `value` IN (" . \Piwik\DataAccess\ArchiveWriter::DONE_INVALIDATED . ")"); |
| 402 | $result = array(); |
| 403 | foreach ($rows as $row) { |
| 404 | $result[] = $row['idsite']; |
| 405 | } |
| 406 | return $result; |
| 407 | } |
| 408 | /** |
| 409 | * Get a list of IDs of archives that don't have any matching rows in the site table. Excludes temporary archives |
| 410 | * that may still be in use, as specified by the $oldestToKeep passed in. |
| 411 | * @param string $archiveTableName |
| 412 | * @param string $oldestToKeep Datetime string |
| 413 | * @return array of IDs |
| 414 | */ |
| 415 | public function getArchiveIdsForDeletedSites($archiveTableName) |
| 416 | { |
| 417 | $sql = "SELECT DISTINCT idsite FROM " . $archiveTableName; |
| 418 | $rows = Db::getReader()->fetchAll($sql, array()); |
| 419 | if (empty($rows)) { |
| 420 | return array(); |
| 421 | // nothing to delete |
| 422 | } |
| 423 | $idSitesUsed = array_column($rows, 'idsite'); |
| 424 | $model = new \Piwik\Plugins\SitesManager\Model(); |
| 425 | $idSitesExisting = $model->getSitesId(); |
| 426 | $deletedSites = array_diff($idSitesUsed, $idSitesExisting); |
| 427 | if (empty($deletedSites)) { |
| 428 | return array(); |
| 429 | } |
| 430 | $deletedSites = array_values($deletedSites); |
| 431 | $deletedSites = array_map('intval', $deletedSites); |
| 432 | $sql = "SELECT DISTINCT idarchive FROM " . $archiveTableName . " WHERE idsite IN (" . implode(',', $deletedSites) . ")"; |
| 433 | $rows = Db::getReader()->fetchAll($sql, array()); |
| 434 | return array_column($rows, 'idarchive'); |
| 435 | } |
| 436 | /** |
| 437 | * Get a list of IDs of archives with segments that no longer exist in the DB. Excludes temporary archives that |
| 438 | * may still be in use, as specified by the $oldestToKeep passed in. |
| 439 | * @param string $archiveTableName |
| 440 | * @param array $segments List of segments to match against |
| 441 | * @param string $oldestToKeep Datetime string |
| 442 | * @return array With keys idarchive, name, idsite |
| 443 | */ |
| 444 | public function getArchiveIdsForSegments($archiveTableName, array $segments, $oldestToKeep) |
| 445 | { |
| 446 | $segmentClauses = []; |
| 447 | foreach ($segments as $segment) { |
| 448 | if (!empty($segment['definition'])) { |
| 449 | $segmentClauses[] = $this->getDeletedSegmentWhereClause($segment); |
| 450 | } |
| 451 | } |
| 452 | if (empty($segmentClauses)) { |
| 453 | return array(); |
| 454 | } |
| 455 | $segmentClauses = implode(' OR ', $segmentClauses); |
| 456 | $sql = 'SELECT idarchive FROM ' . $archiveTableName . ' WHERE ts_archived < ?' . ' AND (' . $segmentClauses . ')'; |
| 457 | $rows = Db::fetchAll($sql, array($oldestToKeep)); |
| 458 | return array_column($rows, 'idarchive'); |
| 459 | } |
| 460 | private function getDeletedSegmentWhereClause(array $segment) |
| 461 | { |
| 462 | $idSite = (int) $segment['enable_only_idsite']; |
| 463 | $segmentHash = $segment['hash'] ?? ''; |
| 464 | // Valid segment hashes are md5 strings - just confirm that it is so it's safe for SQL injection |
| 465 | if (!ctype_xdigit($segmentHash)) { |
| 466 | throw new Exception($segmentHash . ' expected to be an md5 hash'); |
| 467 | } |
| 468 | $nameClause = 'name LIKE "done' . $segmentHash . '%"'; |
| 469 | $idSiteClause = ''; |
| 470 | if ($idSite > 0) { |
| 471 | $idSiteClause = ' AND idsite = ' . $idSite; |
| 472 | } elseif (!empty($segment['idsites_to_preserve'])) { |
| 473 | // A segment for all sites was deleted, but there are segments for a single site with the same definition |
| 474 | $idSitesToPreserve = array_map('intval', $segment['idsites_to_preserve']); |
| 475 | $idSiteClause = ' AND idsite NOT IN (' . implode(',', $idSitesToPreserve) . ')'; |
| 476 | } |
| 477 | return "({$nameClause} {$idSiteClause})"; |
| 478 | } |
| 479 | /** |
| 480 | * Returns the SQL condition used to find successfully completed archives that |
| 481 | * this instance is querying for. |
| 482 | */ |
| 483 | private static function getNameCondition($doneFlags, $possibleValues) |
| 484 | { |
| 485 | $allDoneFlags = "'" . implode("','", $doneFlags) . "'"; |
| 486 | // create the SQL to find archives that are DONE |
| 487 | $result = "((arc1.name IN ({$allDoneFlags}))"; |
| 488 | if (!empty($possibleValues)) { |
| 489 | $result .= " AND (arc1.value IN (" . implode(',', $possibleValues) . ")))"; |
| 490 | } |
| 491 | $result .= ')'; |
| 492 | return $result; |
| 493 | } |
| 494 | /** |
| 495 | * Marks an archive as in progress if it has not been already. This method must be thread |
| 496 | * safe. |
| 497 | */ |
| 498 | public function startArchive($invalidation) |
| 499 | { |
| 500 | $table = Common::prefixTable('archive_invalidations'); |
| 501 | // set archive value to in progress if not set already |
| 502 | $statement = Db::query("UPDATE `{$table}` SET `status` = ?, ts_started = NOW() WHERE idinvalidation = ? AND status = ?", [ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, $invalidation['idinvalidation'], ArchiveInvalidator::INVALIDATION_STATUS_QUEUED]); |
| 503 | if ($statement->rowCount() > 0) { |
| 504 | // if we updated, then we've marked the archive as started |
| 505 | return true; |
| 506 | } |
| 507 | // archive was not originally started or was started within 24 hours, we assume it's ongoing and another process |
| 508 | // (on this machine or another) is actively archiving it. |
| 509 | if (empty($invalidation['ts_started']) || $invalidation['ts_started'] > Date::now()->subDay(1)->getTimestamp()) { |
| 510 | return false; |
| 511 | } |
| 512 | // archive was started over 24 hours ago, we assume it failed and take it over |
| 513 | Db::query("UPDATE `{$table}` SET `status` = ?, ts_started = NOW() WHERE idinvalidation = ?", [ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, $invalidation['idinvalidation']]); |
| 514 | // remove similar invalidations w/ lesser idinvalidation values |
| 515 | $bind = [$invalidation['idsite'], $invalidation['period'], $invalidation['date1'], $invalidation['date2'], $invalidation['name'], ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS]; |
| 516 | if (empty($invalidation['report'])) { |
| 517 | $reportClause = "(report IS NULL OR report = '')"; |
| 518 | } else { |
| 519 | $reportClause = "report = ?"; |
| 520 | $bind[] = $invalidation['report']; |
| 521 | } |
| 522 | $sql = "DELETE FROM " . Common::prefixTable('archive_invalidations') . " WHERE idinvalidation < ? AND idsite = ? AND " . "date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND {$reportClause}"; |
| 523 | Db::query($sql, $bind); |
| 524 | return true; |
| 525 | } |
| 526 | public function isSimilarArchiveInProgress($invalidation) |
| 527 | { |
| 528 | $table = Common::prefixTable('archive_invalidations'); |
| 529 | $bind = [$invalidation['idsite'], $invalidation['period'], $invalidation['date1'], $invalidation['date2'], $invalidation['name'], ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS]; |
| 530 | if (empty($invalidation['report'])) { |
| 531 | $reportClause = "(report IS NULL OR report = '')"; |
| 532 | } else { |
| 533 | $reportClause = "report = ?"; |
| 534 | $bind[] = $invalidation['report']; |
| 535 | } |
| 536 | $sql = "SELECT idinvalidation FROM `{$table}` WHERE idsite = ? AND `period` = ? AND date1 = ? AND date2 = ? AND `name` = ? AND `status` = ? AND ts_started IS NOT NULL AND {$reportClause} LIMIT 1"; |
| 537 | $result = Db::fetchOne($sql, $bind); |
| 538 | return !empty($result); |
| 539 | } |
| 540 | /** |
| 541 | * Gets the next invalidated archive that should be archived in a table. |
| 542 | * |
| 543 | * @param int $idSite |
| 544 | * @param string $archivingStartTime |
| 545 | * @param int[]|null $idInvalidationsToExclude |
| 546 | * @param bool $useLimit Whether to limit the result set to one result or not. Used in tests only. |
| 547 | */ |
| 548 | public function getNextInvalidatedArchive($idSite, $archivingStartTime, $idInvalidationsToExclude = null, $useLimit = true) |
| 549 | { |
| 550 | $table = Common::prefixTable('archive_invalidations'); |
| 551 | $sql = "SELECT *\n FROM `{$table}`\n WHERE idsite = ? AND status != ? AND ts_invalidated <= ?"; |
| 552 | $bind = [$idSite, ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, $archivingStartTime]; |
| 553 | if (!empty($idInvalidationsToExclude)) { |
| 554 | $idInvalidationsToExclude = array_map('intval', $idInvalidationsToExclude); |
| 555 | $sql .= " AND idinvalidation NOT IN (" . implode(',', $idInvalidationsToExclude) . ')'; |
| 556 | } |
| 557 | // NOTE: order here is very important to ensure we process lower period archives first, and general 'all' archives before |
| 558 | // segment archives, and so we use the latest idinvalidation |
| 559 | $sql .= " ORDER BY date1 DESC, period ASC, CHAR_LENGTH(name) ASC, idinvalidation DESC"; |
| 560 | if ($useLimit) { |
| 561 | $sql .= " LIMIT 1"; |
| 562 | return Db::fetchRow($sql, $bind); |
| 563 | } else { |
| 564 | return Db::fetchAll($sql, $bind); |
| 565 | } |
| 566 | } |
| 567 | public function deleteInvalidations($archiveInvalidations) |
| 568 | { |
| 569 | $ids = array_column($archiveInvalidations, 'idinvalidation'); |
| 570 | $ids = array_map('intval', $ids); |
| 571 | $table = Common::prefixTable('archive_invalidations'); |
| 572 | $sql = "DELETE FROM `{$table}` WHERE idinvalidation IN (" . implode(', ', $ids) . ")"; |
| 573 | Db::query($sql); |
| 574 | } |
| 575 | public function removeInvalidationsLike($idSite, $start) |
| 576 | { |
| 577 | $idSitesClause = $this->getRemoveInvalidationsIdSitesClause($idSite); |
| 578 | $table = Common::prefixTable('archive_invalidations'); |
| 579 | $sql = "DELETE FROM `{$table}` WHERE {$idSitesClause} `name` LIKE ?"; |
| 580 | Db::query($sql, ['done%.' . str_replace('_', "\\_", $start)]); |
| 581 | } |
| 582 | public function removeInvalidations($idSite, $plugin, $report) |
| 583 | { |
| 584 | $idSitesClause = $this->getRemoveInvalidationsIdSitesClause($idSite); |
| 585 | $table = Common::prefixTable('archive_invalidations'); |
| 586 | $sql = "DELETE FROM `{$table}` WHERE {$idSitesClause} `name` LIKE ? AND report = ?"; |
| 587 | Db::query($sql, ['done%.' . str_replace('_', "\\_", $plugin), $report]); |
| 588 | } |
| 589 | public function isArchiveAlreadyInProgress($invalidatedArchive) |
| 590 | { |
| 591 | $table = Common::prefixTable('archive_invalidations'); |
| 592 | $bind = [$invalidatedArchive['idsite'], $invalidatedArchive['date1'], $invalidatedArchive['date2'], $invalidatedArchive['period'], $invalidatedArchive['name']]; |
| 593 | $reportClause = "(report = '' OR report IS NULL)"; |
| 594 | if (!empty($invalidatedArchive['report'])) { |
| 595 | $reportClause = "report = ?"; |
| 596 | $bind[] = $invalidatedArchive['report']; |
| 597 | } |
| 598 | $sql = "SELECT MAX(idinvalidation) FROM `{$table}` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND status = 1 AND {$reportClause}"; |
| 599 | $inProgressInvalidation = Db::fetchOne($sql, $bind); |
| 600 | return $inProgressInvalidation; |
| 601 | } |
| 602 | /** |
| 603 | * Returns true if there is an archive that exists that can be used when aggregating an archive for $period. |
| 604 | * |
| 605 | * @param $idSite |
| 606 | * @param Period $period |
| 607 | * @return bool |
| 608 | * @throws Exception |
| 609 | */ |
| 610 | public function hasChildArchivesInPeriod($idSite, Period $period) |
| 611 | { |
| 612 | $date = $period->getDateStart(); |
| 613 | while ($date->isEarlier($period->getDateEnd()->addPeriod(1, 'month'))) { |
| 614 | $archiveTable = \Piwik\DataAccess\ArchiveTableCreator::getNumericTable($date); |
| 615 | // we look for any archive that can be used to compute this one. this includes invalidated archives, since it is possible |
| 616 | // under certain circumstances for them to exist, when archiving a higher period that includes them. the main example being |
| 617 | // the GoogleAnalyticsImporter which disallows the recomputation of invalidated archives for imported data, since that would |
| 618 | // essentially get rid of the imported data. |
| 619 | $usableDoneFlags = [\Piwik\DataAccess\ArchiveWriter::DONE_OK, \Piwik\DataAccess\ArchiveWriter::DONE_INVALIDATED, \Piwik\DataAccess\ArchiveWriter::DONE_PARTIAL, \Piwik\DataAccess\ArchiveWriter::DONE_OK_TEMPORARY]; |
| 620 | $sql = "SELECT idarchive\n FROM `{$archiveTable}`\n WHERE idsite = ? AND date1 >= ? AND date2 <= ? AND period < ? AND `name` LIKE 'done%' AND `value` IN (" . implode(', ', $usableDoneFlags) . ")\n LIMIT 1"; |
| 621 | $bind = [$idSite, $period->getDateStart()->getDatetime(), $period->getDateEnd()->getDatetime(), $period->getId()]; |
| 622 | $result = (bool) Db::fetchOne($sql, $bind); |
| 623 | if ($result) { |
| 624 | return true; |
| 625 | } |
| 626 | $date = $date->addPeriod(1, 'month'); |
| 627 | // move to next archive table |
| 628 | } |
| 629 | return false; |
| 630 | } |
| 631 | /** |
| 632 | * Returns true if any invalidations exists for the given |
| 633 | * $idsite and $doneFlag (name column) for the $period. |
| 634 | * |
| 635 | * @param mixed $idSite |
| 636 | * @param Period $period |
| 637 | * @param mixed $doneFlag |
| 638 | * @param mixed $report |
| 639 | * @return bool |
| 640 | * @throws Exception |
| 641 | */ |
| 642 | public function hasInvalidationForPeriodAndName($idSite, Period $period, $doneFlag, $report = null) |
| 643 | { |
| 644 | $table = Common::prefixTable('archive_invalidations'); |
| 645 | if (empty($report)) { |
| 646 | $sql = "SELECT idinvalidation FROM `{$table}` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND `report` IS NULL LIMIT 1"; |
| 647 | } else { |
| 648 | $sql = "SELECT idinvalidation FROM `{$table}` WHERE idsite = ? AND date1 = ? AND date2 = ? AND `period` = ? AND `name` = ? AND `report` = ? LIMIT 1"; |
| 649 | } |
| 650 | $bind = [$idSite, $period->getDateStart()->toString(), $period->getDateEnd()->toString(), $period->getId(), $doneFlag]; |
| 651 | if (!empty($report)) { |
| 652 | $bind[] = $report; |
| 653 | } |
| 654 | $idInvalidation = Db::fetchOne($sql, $bind); |
| 655 | if (empty($idInvalidation)) { |
| 656 | return false; |
| 657 | } |
| 658 | return true; |
| 659 | } |
| 660 | public function deleteInvalidationsForSites(array $idSites) |
| 661 | { |
| 662 | $idSites = array_map('intval', $idSites); |
| 663 | $table = Common::prefixTable('archive_invalidations'); |
| 664 | $sql = "DELETE FROM `{$table}` WHERE idsite IN (" . implode(',', $idSites) . ")"; |
| 665 | Db::query($sql); |
| 666 | } |
| 667 | public function deleteInvalidationsForDeletedSites() |
| 668 | { |
| 669 | $siteTable = Common::prefixTable('site'); |
| 670 | $table = Common::prefixTable('archive_invalidations'); |
| 671 | $sql = "DELETE a FROM `{$table}` a LEFT JOIN `{$siteTable}` s ON a.idsite = s.idsite WHERE s.idsite IS NULL"; |
| 672 | Db::query($sql); |
| 673 | } |
| 674 | private function getRemoveInvalidationsIdSitesClause($idSite) |
| 675 | { |
| 676 | if ($idSite === 'all') { |
| 677 | return ''; |
| 678 | } |
| 679 | $idSites = is_array($idSite) ? $idSite : [$idSite]; |
| 680 | $idSites = array_map('intval', $idSites); |
| 681 | $idSitesStr = implode(',', $idSites); |
| 682 | return "idsite IN ({$idSitesStr}) AND"; |
| 683 | } |
| 684 | public function releaseInProgressInvalidation($idinvalidation) |
| 685 | { |
| 686 | $table = Common::prefixTable('archive_invalidations'); |
| 687 | $sql = "UPDATE {$table} SET status = " . ArchiveInvalidator::INVALIDATION_STATUS_QUEUED . ", ts_started = NULL WHERE idinvalidation = ?"; |
| 688 | Db::query($sql, [$idinvalidation]); |
| 689 | } |
| 690 | public function resetFailedArchivingJobs() |
| 691 | { |
| 692 | $table = Common::prefixTable('archive_invalidations'); |
| 693 | $sql = "UPDATE {$table} SET status = ? WHERE status = ? AND (ts_started IS NULL OR ts_started < ?)"; |
| 694 | $bind = [ArchiveInvalidator::INVALIDATION_STATUS_QUEUED, ArchiveInvalidator::INVALIDATION_STATUS_IN_PROGRESS, Date::now()->subDay(1)->getDatetime()]; |
| 695 | $query = Db::query($sql, $bind); |
| 696 | return $query->rowCount(); |
| 697 | } |
| 698 | public function getRecordsContainedInArchives(Date $archiveStartDate, array $idArchives, $requestedRecords) : array |
| 699 | { |
| 700 | $idArchives = array_map('intval', $idArchives); |
| 701 | $idArchives = implode(',', $idArchives); |
| 702 | $requestedRecords = is_string($requestedRecords) ? [$requestedRecords] : $requestedRecords; |
| 703 | $placeholders = Common::getSqlStringFieldsArray($requestedRecords); |
| 704 | $countSql = "SELECT DISTINCT name FROM %s WHERE idarchive IN ({$idArchives}) AND name IN ({$placeholders}) LIMIT " . count($requestedRecords); |
| 705 | $numericTable = \Piwik\DataAccess\ArchiveTableCreator::getNumericTable($archiveStartDate); |
| 706 | $blobTable = \Piwik\DataAccess\ArchiveTableCreator::getBlobTable($archiveStartDate); |
| 707 | // if the requested metrics look numeric, prioritize the numeric table, otherwise the blob table. this way, if all the metrics are |
| 708 | // found in this table (which will be most of the time), we don't have to query the other table |
| 709 | if ($this->doRequestedRecordsLookNumeric($requestedRecords)) { |
| 710 | $tablesToSearch = [$numericTable, $blobTable]; |
| 711 | } else { |
| 712 | $tablesToSearch = [$blobTable, $numericTable]; |
| 713 | } |
| 714 | $existingRecords = []; |
| 715 | foreach ($tablesToSearch as $tableName) { |
| 716 | $sql = sprintf($countSql, $tableName); |
| 717 | $rows = Db::fetchAll($sql, $requestedRecords); |
| 718 | $existingRecords = array_merge($existingRecords, array_column($rows, 'name')); |
| 719 | if (count($existingRecords) == count($requestedRecords)) { |
| 720 | break; |
| 721 | } |
| 722 | } |
| 723 | return $existingRecords; |
| 724 | } |
| 725 | private function isCutOffGroupConcatResult($pair) |
| 726 | { |
| 727 | $position = strpos($pair, '.'); |
| 728 | return $position === false || $position === strlen($pair) - 1; |
| 729 | } |
| 730 | private function getHashFromDoneFlag($doneFlag) |
| 731 | { |
| 732 | preg_match('/^done([a-zA-Z0-9]+)/', $doneFlag, $matches); |
| 733 | return $matches[1] ?? ''; |
| 734 | } |
| 735 | private function doRequestedRecordsLookNumeric(array $requestedRecords) : bool |
| 736 | { |
| 737 | foreach ($requestedRecords as $record) { |
| 738 | if (preg_match('/^nb_/', $record)) { |
| 739 | return true; |
| 740 | } |
| 741 | } |
| 742 | return false; |
| 743 | } |
| 744 | } |
| 745 |