ArchivingStatus.php
6 months ago
BlobTableAggregator.php
2 weeks ago
Loader.php
2 weeks ago
LoaderLock.php
2 weeks ago
Parameters.php
2 weeks ago
PluginsArchiver.php
2 weeks ago
PluginsArchiverException.php
2 years ago
Record.php
2 weeks ago
RecordBuilder.php
2 weeks ago
Rules.php
2 weeks ago
Loader.php
575 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\ArchiveProcessor; |
| 10 | |
| 11 | use Piwik\Archive\ArchiveInvalidator; |
| 12 | use Piwik\ArchiveProcessor; |
| 13 | use Piwik\Cache; |
| 14 | use Piwik\CacheId; |
| 15 | use Piwik\Common; |
| 16 | use Piwik\Config; |
| 17 | use Piwik\Container\StaticContainer; |
| 18 | use Piwik\Context; |
| 19 | use Piwik\DataAccess\ArchiveSelector; |
| 20 | use Piwik\DataAccess\ArchiveWriter; |
| 21 | use Piwik\DataAccess\Model; |
| 22 | use Piwik\DataAccess\RawLogDao; |
| 23 | use Piwik\Date; |
| 24 | use Piwik\Period; |
| 25 | use Piwik\Piwik; |
| 26 | use Piwik\SettingsServer; |
| 27 | use Piwik\Site; |
| 28 | use Piwik\Log\LoggerInterface; |
| 29 | use Piwik\CronArchive\SegmentArchiving; |
| 30 | /** |
| 31 | * This class uses PluginsArchiver class to trigger data aggregation and create archives. |
| 32 | */ |
| 33 | class Loader |
| 34 | { |
| 35 | private static $archivingDepth = 0; |
| 36 | /** |
| 37 | * Tracks whether the current prepareArchive run reused an existing archive instead of processing. |
| 38 | * |
| 39 | * @var boolean |
| 40 | */ |
| 41 | private $didReuseArchive = \false; |
| 42 | /** |
| 43 | * @var Parameters |
| 44 | */ |
| 45 | protected $params; |
| 46 | /** |
| 47 | * @var ArchiveInvalidator |
| 48 | */ |
| 49 | private $invalidator; |
| 50 | /** |
| 51 | * @var \Matomo\Cache\Cache |
| 52 | */ |
| 53 | private $cache; |
| 54 | /** |
| 55 | * @var LoggerInterface |
| 56 | */ |
| 57 | private $logger; |
| 58 | /** |
| 59 | * @var RawLogDao |
| 60 | */ |
| 61 | private $rawLogDao; |
| 62 | /** |
| 63 | * @var Model |
| 64 | */ |
| 65 | private $dataAccessModel; |
| 66 | /** |
| 67 | * @var bool |
| 68 | */ |
| 69 | private $invalidateBeforeArchiving; |
| 70 | public function __construct(\Piwik\ArchiveProcessor\Parameters $params, $invalidateBeforeArchiving = \false) |
| 71 | { |
| 72 | $this->params = $params; |
| 73 | $this->invalidateBeforeArchiving = $invalidateBeforeArchiving; |
| 74 | $this->invalidator = StaticContainer::get(ArchiveInvalidator::class); |
| 75 | $this->cache = Cache::getTransientCache(); |
| 76 | $this->logger = StaticContainer::get(LoggerInterface::class); |
| 77 | $this->rawLogDao = new RawLogDao(); |
| 78 | $this->dataAccessModel = new Model(); |
| 79 | } |
| 80 | /** |
| 81 | * @return bool |
| 82 | */ |
| 83 | protected function isThereSomeVisits($visits) |
| 84 | { |
| 85 | return $visits > 0; |
| 86 | } |
| 87 | /** |
| 88 | * @return bool |
| 89 | */ |
| 90 | protected function mustProcessVisitCount($visits) |
| 91 | { |
| 92 | return $visits === \false; |
| 93 | } |
| 94 | public function prepareArchive($pluginName) |
| 95 | { |
| 96 | return Context::changeIdSite($this->params->getSite()->getId(), function () use($pluginName) { |
| 97 | try { |
| 98 | ++self::$archivingDepth; |
| 99 | if (self::$archivingDepth === 1) { |
| 100 | $this->didReuseArchive = \false; |
| 101 | } |
| 102 | return $this->prepareArchiveImpl($pluginName); |
| 103 | } finally { |
| 104 | --self::$archivingDepth; |
| 105 | } |
| 106 | }); |
| 107 | } |
| 108 | /** |
| 109 | * @throws \Exception |
| 110 | */ |
| 111 | private function prepareArchiveImpl($pluginName) |
| 112 | { |
| 113 | $this->params->setRequestedPlugin($pluginName); |
| 114 | if (SettingsServer::isArchivePhpTriggered()) { |
| 115 | $requestedReport = Common::getRequestVar('requestedReport', '', 'string'); |
| 116 | if (!empty($requestedReport)) { |
| 117 | $this->params->setArchiveOnlyReport($requestedReport); |
| 118 | } |
| 119 | } |
| 120 | // invalidate existing archives before we start archiving in case data was tracked in the past. if the archive is |
| 121 | // made invalid, we will correctly re-archive below. |
| 122 | if ($this->invalidateBeforeArchiving && \Piwik\ArchiveProcessor\Rules::isBrowserTriggerEnabled()) { |
| 123 | $this->invalidatedReportsIfNeeded(); |
| 124 | } |
| 125 | // load existing data from archive |
| 126 | $data = $this->loadArchiveData(); |
| 127 | if (sizeof($data) == 2) { |
| 128 | $this->didReuseArchive = \true; |
| 129 | return $data; |
| 130 | } |
| 131 | [$idArchives, $visits, $visitsConverted, $foundRecords] = $data; |
| 132 | // only lock meet those conditions |
| 133 | if (ArchiveProcessor::$isRootArchivingRequest && !SettingsServer::isArchivePhpTriggered()) { |
| 134 | $lockId = $this->makeArchivingLockId(); |
| 135 | //ini lock |
| 136 | $lock = new \Piwik\ArchiveProcessor\LoaderLock($lockId); |
| 137 | //set mysql lock the entire process if another process is running |
| 138 | $lock->setLock(); |
| 139 | try { |
| 140 | $data = $this->loadArchiveData(); |
| 141 | if (sizeof($data) == 2) { |
| 142 | $this->didReuseArchive = \true; |
| 143 | return $data; |
| 144 | } |
| 145 | [$idArchives, $visits, $visitsConverted, $foundRecords] = $data; |
| 146 | return $this->insertArchiveData($visits, $visitsConverted, $idArchives, $foundRecords); |
| 147 | } finally { |
| 148 | $lock->unlock(); |
| 149 | } |
| 150 | } else { |
| 151 | return $this->insertArchiveData($visits, $visitsConverted, $idArchives, $foundRecords); |
| 152 | } |
| 153 | } |
| 154 | /** |
| 155 | * @param bool|int|float $visits |
| 156 | * @param bool|int|float $visitsConverted |
| 157 | * @param array|false $existingArchives |
| 158 | * @param array|null $foundRecords |
| 159 | * @return array{0: array, 1: bool|int|float} |
| 160 | */ |
| 161 | protected function insertArchiveData($visits, $visitsConverted, $existingArchives, $foundRecords) |
| 162 | { |
| 163 | if (SettingsServer::isArchivePhpTriggered()) { |
| 164 | $this->logger->info("initiating archiving via core:archive for " . $this->params); |
| 165 | } |
| 166 | if (!empty($foundRecords)) { |
| 167 | $this->params->setFoundRequestedReports($foundRecords); |
| 168 | } |
| 169 | [$visits, $visitsConverted] = $this->prepareCoreMetricsArchive($visits, $visitsConverted); |
| 170 | [$idArchive, $visits] = $this->prepareAllPluginsArchive($visits, $visitsConverted); |
| 171 | $idArchivesToQuery = [$idArchive]; |
| 172 | if (!empty($foundRecords)) { |
| 173 | $idArchivesToQuery = array_merge($idArchivesToQuery, $existingArchives ?: []); |
| 174 | } |
| 175 | return [$idArchivesToQuery, $visits]; |
| 176 | } |
| 177 | /** |
| 178 | * @return string |
| 179 | * @throws \Exception |
| 180 | */ |
| 181 | private function makeArchivingLockId() |
| 182 | { |
| 183 | $doneFlag = \Piwik\ArchiveProcessor\Rules::getDoneStringFlagFor([$this->params->getSite()->getId()], $this->params->getSegment(), $this->params->getPeriod()->getLabel(), $this->params->getRequestedPlugin()); |
| 184 | return $this->params->getPeriod()->getDateStart()->toString() . $this->params->getPeriod()->getDateEnd()->toString() . '.' . $doneFlag; |
| 185 | } |
| 186 | /** |
| 187 | * @return array |
| 188 | */ |
| 189 | protected function loadArchiveData() |
| 190 | { |
| 191 | // this hack was used to check the main function goes to return or continue |
| 192 | // NOTE: $idArchives will contain the latest DONE_OK/DONE_INVALIDATED archive as well as any partial archives |
| 193 | // with a ts_archived >= the DONE_OK/DONE_INVALIDATED date. |
| 194 | $archiveInfo = $this->loadExistingArchiveIdFromDb(); |
| 195 | $idArchives = $archiveInfo['idArchives']; |
| 196 | $visits = $archiveInfo['visits']; |
| 197 | $visitsConverted = $archiveInfo['visitsConverted']; |
| 198 | $tsArchived = $archiveInfo['tsArchived']; |
| 199 | $doneFlagValue = $archiveInfo['doneFlagValue']; |
| 200 | $existingArchives = $archiveInfo['existingRecords']; |
| 201 | $requestedRecords = $this->params->getArchiveOnlyReportAsArray(); |
| 202 | $isMissingRequestedRecords = !empty($requestedRecords) && is_array($existingArchives) && count($requestedRecords) != count($existingArchives); |
| 203 | if (!empty($idArchives) && !\Piwik\ArchiveProcessor\Rules::isActuallyForceArchivingSinglePlugin() && !$this->shouldForceInvalidatedArchive($doneFlagValue, $tsArchived) && !$isMissingRequestedRecords) { |
| 204 | // we have a usable idarchive (it's not invalidated and it's new enough), and we are not archiving |
| 205 | // a single report |
| 206 | return [$idArchives, $visits]; |
| 207 | } |
| 208 | // NOTE: this optimization helps when archiving large periods. eg, if archiving a year w/ a segment where |
| 209 | // there are not visits in the entire year, we don't have to go through and do anything. but, w/o this |
| 210 | // code, we will end up launching archiving for each month, week and day, even though we don't have to. |
| 211 | // |
| 212 | // we don't create an archive in this case, because the archive may be in progress in some way, so a 0 |
| 213 | // visits archive can be inaccurate in the long run. |
| 214 | if ($this->canSkipThisArchive()) { |
| 215 | if (!empty($idArchives)) { |
| 216 | return [$idArchives, $visits]; |
| 217 | } else { |
| 218 | return [\false, 0]; |
| 219 | } |
| 220 | } |
| 221 | if (self::$archivingDepth > 1) { |
| 222 | $this->logger->debug(sprintf("Sub-period archive requires processing. Archiving depth: %d", self::$archivingDepth)); |
| 223 | $this->params->logStatusDebug(); |
| 224 | } |
| 225 | return [$idArchives, $visits, $visitsConverted, $existingArchives]; |
| 226 | } |
| 227 | /** |
| 228 | * Prepares the core metrics if needed. |
| 229 | * |
| 230 | * @param bool|int|float $visits |
| 231 | * @param bool|int|float $visitsConverted |
| 232 | * @return array |
| 233 | */ |
| 234 | protected function prepareCoreMetricsArchive($visits, $visitsConverted) |
| 235 | { |
| 236 | $createSeparateArchiveForCoreMetrics = $this->mustProcessVisitCount($visits) && !$this->doesRequestedPluginIncludeVisitsSummary(); |
| 237 | if ($createSeparateArchiveForCoreMetrics) { |
| 238 | $requestedPlugin = $this->params->getRequestedPlugin(); |
| 239 | $requestedReport = $this->params->getArchiveOnlyReport(); |
| 240 | $this->params->setRequestedPlugin('VisitsSummary'); |
| 241 | $this->params->setArchiveOnlyReport(null); |
| 242 | $metrics = Context::executeWithQueryParameters(['requestedReport' => ''], function () { |
| 243 | $pluginsArchiver = new \Piwik\ArchiveProcessor\PluginsArchiver($this->params); |
| 244 | $metrics = $pluginsArchiver->callAggregateCoreMetrics(); |
| 245 | $pluginsArchiver->finalizeArchive(); |
| 246 | return $metrics; |
| 247 | }); |
| 248 | $this->params->setRequestedPlugin($requestedPlugin); |
| 249 | $this->params->setArchiveOnlyReport($requestedReport); |
| 250 | $visits = $metrics['nb_visits']; |
| 251 | $visitsConverted = $metrics['nb_visits_converted']; |
| 252 | } |
| 253 | return array($visits, $visitsConverted); |
| 254 | } |
| 255 | protected function prepareAllPluginsArchive($visits, $visitsConverted) |
| 256 | { |
| 257 | $pluginsArchiver = new \Piwik\ArchiveProcessor\PluginsArchiver($this->params); |
| 258 | if ($this->mustProcessVisitCount($visits) || $this->doesRequestedPluginIncludeVisitsSummary()) { |
| 259 | $metrics = $pluginsArchiver->callAggregateCoreMetrics(); |
| 260 | $visits = $metrics['nb_visits']; |
| 261 | $visitsConverted = $metrics['nb_visits_converted']; |
| 262 | } |
| 263 | $forceArchivingWithoutVisits = !$this->isThereSomeVisits($visits) && $this->shouldArchiveForSiteEvenWhenNoVisits(); |
| 264 | $pluginsArchiver->callAggregateAllPlugins($visits, $visitsConverted, $forceArchivingWithoutVisits); |
| 265 | $idArchive = $pluginsArchiver->finalizeArchive(); |
| 266 | return array($idArchive, $visits); |
| 267 | } |
| 268 | protected function doesRequestedPluginIncludeVisitsSummary() |
| 269 | { |
| 270 | $processAllReportsIncludingVisitsSummary = \Piwik\ArchiveProcessor\Rules::shouldProcessReportsAllPlugins(array($this->params->getSite()->getId()), $this->params->getSegment(), $this->params->getPeriod()->getLabel()); |
| 271 | $doesRequestedPluginIncludeVisitsSummary = $processAllReportsIncludingVisitsSummary || $this->params->getRequestedPlugin() == 'VisitsSummary'; |
| 272 | return $doesRequestedPluginIncludeVisitsSummary; |
| 273 | } |
| 274 | protected function isArchivingForcedToTrigger() |
| 275 | { |
| 276 | $period = $this->params->getPeriod()->getLabel(); |
| 277 | $debugSetting = 'always_archive_data_period'; |
| 278 | // default |
| 279 | if ($period == 'day') { |
| 280 | $debugSetting = 'always_archive_data_day'; |
| 281 | } elseif ($period == 'range') { |
| 282 | $debugSetting = 'always_archive_data_range'; |
| 283 | } |
| 284 | return (bool) Config::getInstance()->Debug[$debugSetting]; |
| 285 | } |
| 286 | /** |
| 287 | * Returns the idArchive if the archive is available in the database for the requested plugin. |
| 288 | * Returns false if the archive needs to be processed. |
| 289 | * |
| 290 | * (public for tests) |
| 291 | * |
| 292 | * @return array |
| 293 | */ |
| 294 | public function loadExistingArchiveIdFromDb() |
| 295 | { |
| 296 | if ($this->isArchivingForcedToTrigger()) { |
| 297 | $this->logger->debug("Archiving forced to trigger for {$this->params}."); |
| 298 | // return no usable archive found, and no existing archive. this will skip invalidation, which should |
| 299 | // be fine since we just force archiving. |
| 300 | return ['idArchives' => \false, 'visits' => \false, 'visitsConverted' => \false, 'archiveExists' => \false, 'tsArchived' => \false, 'doneFlagValue' => \false, 'existingRecords' => null]; |
| 301 | } |
| 302 | $minDatetimeArchiveProcessedUTC = $this->getMinTimeArchiveProcessed(); |
| 303 | $result = ArchiveSelector::getArchiveIdAndVisits($this->params, $minDatetimeArchiveProcessedUTC); |
| 304 | return $result; |
| 305 | } |
| 306 | /** |
| 307 | * Returns the minimum archive processed datetime to look at. Only public for tests. |
| 308 | * |
| 309 | * @return int|bool Datetime timestamp, or false if must look at any archive available |
| 310 | */ |
| 311 | protected function getMinTimeArchiveProcessed() |
| 312 | { |
| 313 | // for range periods we can archive in a browser request request, make sure to check for the ttl no matter what |
| 314 | $isRangeArchiveAndArchivingEnabled = $this->params->getPeriod()->getLabel() == 'range' && \Piwik\ArchiveProcessor\Rules::isArchivingEnabledFor([$this->params->getSite()->getId()], $this->params->getSegment(), $this->params->getPeriod()->getLabel()); |
| 315 | if (!$isRangeArchiveAndArchivingEnabled) { |
| 316 | $endDateTimestamp = self::determineIfArchivePermanent($this->params->getDateEnd()); |
| 317 | if ($endDateTimestamp) { |
| 318 | // past archive |
| 319 | return $endDateTimestamp; |
| 320 | } |
| 321 | } |
| 322 | $dateStart = $this->params->getDateStart(); |
| 323 | $period = $this->params->getPeriod(); |
| 324 | $segment = $this->params->getSegment(); |
| 325 | $site = $this->params->getSite(); |
| 326 | // in-progress archive |
| 327 | return \Piwik\ArchiveProcessor\Rules::getMinTimeProcessedForInProgressArchive($dateStart, $period, $segment, $site); |
| 328 | } |
| 329 | protected static function determineIfArchivePermanent(Date $dateEnd) |
| 330 | { |
| 331 | $now = time(); |
| 332 | $endTimestampUTC = strtotime($dateEnd->getDateEndUTC()); |
| 333 | if ($endTimestampUTC <= $now) { |
| 334 | // - if the period we are looking for is finished, we look for a ts_archived that |
| 335 | // is greater than the last day of the archive |
| 336 | return $endTimestampUTC; |
| 337 | } |
| 338 | return \false; |
| 339 | } |
| 340 | private function shouldArchiveForSiteEvenWhenNoVisits() |
| 341 | { |
| 342 | $idSitesToArchive = $this->getIdSitesToArchiveWhenNoVisits(); |
| 343 | return in_array($this->params->getSite()->getId(), $idSitesToArchive); |
| 344 | } |
| 345 | private function getIdSitesToArchiveWhenNoVisits() |
| 346 | { |
| 347 | $cacheKey = 'Archiving.getIdSitesToArchiveWhenNoVisits'; |
| 348 | if (!$this->cache->contains($cacheKey)) { |
| 349 | $idSites = array(); |
| 350 | // leaving undocumented unless decided otherwise |
| 351 | Piwik::postEvent('Archiving.getIdSitesToArchiveWhenNoVisits', array(&$idSites)); |
| 352 | $this->cache->save($cacheKey, $idSites); |
| 353 | } |
| 354 | return $this->cache->fetch($cacheKey); |
| 355 | } |
| 356 | // public for tests |
| 357 | public function getReportsToInvalidate() |
| 358 | { |
| 359 | $sitesPerDays = $this->invalidator->getRememberedArchivedReportsThatShouldBeInvalidated(); |
| 360 | foreach ($sitesPerDays as $dateStr => $siteIds) { |
| 361 | if (empty($siteIds) || !in_array($this->params->getSite()->getId(), $siteIds)) { |
| 362 | unset($sitesPerDays[$dateStr]); |
| 363 | } |
| 364 | $date = Date::factory($dateStr); |
| 365 | if ($date->isEarlier($this->params->getPeriod()->getDateStart()) || $date->isLater($this->params->getPeriod()->getDateEnd())) { |
| 366 | // date in list is not the current date, so ignore it |
| 367 | unset($sitesPerDays[$dateStr]); |
| 368 | } |
| 369 | } |
| 370 | return $sitesPerDays; |
| 371 | } |
| 372 | private function invalidatedReportsIfNeeded() |
| 373 | { |
| 374 | $sitesPerDays = $this->getReportsToInvalidate(); |
| 375 | if (empty($sitesPerDays)) { |
| 376 | return; |
| 377 | } |
| 378 | foreach ($sitesPerDays as $date => $siteIds) { |
| 379 | try { |
| 380 | $this->invalidator->markArchivesAsInvalidated([$this->params->getSite()->getId()], array(Date::factory($date)), \false, $this->params->getSegment()); |
| 381 | } catch (\Exception $e) { |
| 382 | Site::clearCache(); |
| 383 | throw $e; |
| 384 | } |
| 385 | } |
| 386 | Site::clearCache(); |
| 387 | } |
| 388 | public function canSkipThisArchive() |
| 389 | { |
| 390 | return $this->canSkipThisArchiveWithReason()[0]; |
| 391 | } |
| 392 | /** |
| 393 | * @internal |
| 394 | * |
| 395 | * @return array{0: bool, 1: string} |
| 396 | */ |
| 397 | public function canSkipThisArchiveWithReason() : array |
| 398 | { |
| 399 | $params = $this->params; |
| 400 | $idSite = $params->getSite()->getId(); |
| 401 | $isWebsiteUsingTracker = $this->isWebsiteUsingTheTracker($idSite); |
| 402 | $isArchivingForcedWhenNoVisits = $this->shouldArchiveForSiteEvenWhenNoVisits(); |
| 403 | $hasSiteVisitsBetweenTimeframe = $this->hasSiteVisitsBetweenTimeframe($idSite, $params->getPeriod()); |
| 404 | $hasChildArchivesInPeriod = $this->hasChildArchivesInPeriod($idSite, $params->getPeriod()); |
| 405 | $canSkipArchiveForSegment = $this->canSkipArchiveForSegmentWithReason(); |
| 406 | if ($canSkipArchiveForSegment[0]) { |
| 407 | return [\true, 'Skip archive for segment: ' . $canSkipArchiveForSegment[1]]; |
| 408 | } |
| 409 | if (!$isWebsiteUsingTracker) { |
| 410 | return [\false, 'Site is not using the JavaScript tracker']; |
| 411 | } |
| 412 | if ($isArchivingForcedWhenNoVisits) { |
| 413 | return [\false, 'Archiving is forced when no visits']; |
| 414 | } |
| 415 | if ($hasSiteVisitsBetweenTimeframe) { |
| 416 | return [\false, 'Site has visits between start and end date']; |
| 417 | } |
| 418 | if ($hasChildArchivesInPeriod) { |
| 419 | return [\false, 'There are child archives in the period']; |
| 420 | } |
| 421 | return [\true, 'Site is using tracker & archiving is not forced when no visits & site has has no visits between start and end date & there are no child archives in the period']; |
| 422 | } |
| 423 | public function didReuseArchive() : bool |
| 424 | { |
| 425 | return $this->didReuseArchive; |
| 426 | } |
| 427 | private function hasChildArchivesInPeriod($idSite, Period $period) : bool |
| 428 | { |
| 429 | $cacheKey = CacheId::siteAware('Archiving.hasChildArchivesInPeriod.' . $period->getRangeString(), [$idSite]); |
| 430 | if ($this->cache->contains($cacheKey)) { |
| 431 | $hasChildArchivesInPeriod = $this->cache->fetch($cacheKey); |
| 432 | } else { |
| 433 | $hasChildArchivesInPeriod = $this->dataAccessModel->hasChildArchivesInPeriod($idSite, $period); |
| 434 | $this->cache->save($cacheKey, $hasChildArchivesInPeriod); |
| 435 | } |
| 436 | return $hasChildArchivesInPeriod; |
| 437 | } |
| 438 | /** |
| 439 | * @return array{0: bool, 1: string} |
| 440 | */ |
| 441 | private function canSkipArchiveForSegmentWithReason() : array |
| 442 | { |
| 443 | $params = $this->params; |
| 444 | if ($params->getSegment()->isEmpty()) { |
| 445 | return [\false, 'Segment is empty']; |
| 446 | } |
| 447 | if (!empty($params->getRequestedPlugin()) && \Piwik\ArchiveProcessor\Rules::isSegmentPluginArchivingDisabled($params->getRequestedPlugin(), $params->getSite()->getId())) { |
| 448 | return [\true, 'Plugin provided and segment plugin archiving disabled']; |
| 449 | } |
| 450 | // For better understanding of the next check please have a look at Rules::shouldProcessReportsAllPlugins implementation |
| 451 | // and what conditions it returns false on. For our use here, we need to ensure that: |
| 452 | // - we are not running CLI archiving |
| 453 | // - we are not dealing with a range period |
| 454 | // - we don't have an empty segment |
| 455 | // - we don't have a segment that should be preprocessed |
| 456 | // - we are not forcing a single plugin archiving |
| 457 | if (!\Piwik\ArchiveProcessor\Rules::shouldProcessReportsAllPlugins($params->getIdSites(), $params->getSegment(), $params->getPeriod()->getLabel())) { |
| 458 | return [\false, 'shouldProcessReportsAllPlugins reported false']; |
| 459 | } |
| 460 | /** @var SegmentArchiving */ |
| 461 | $segmentArchiving = StaticContainer::get(SegmentArchiving::class); |
| 462 | $segmentInfo = $segmentArchiving->findSegmentForHash($params->getSegment()->getHash(), $params->getSite()->getId()); |
| 463 | if (!$segmentInfo) { |
| 464 | return [\false, 'segment not found for hash']; |
| 465 | } |
| 466 | $segmentArchiveStartDate = $segmentArchiving->getReArchiveSegmentStartDate($segmentInfo); |
| 467 | if ($segmentArchiveStartDate !== null && $segmentArchiveStartDate->isLater($params->getPeriod()->getDateEnd()->getEndOfDay())) { |
| 468 | $doneFlag = \Piwik\ArchiveProcessor\Rules::getDoneStringFlagFor([$params->getSite()->getId()], $params->getSegment(), $params->getPeriod()->getLabel(), $params->getRequestedPlugin()); |
| 469 | // if there is no invalidation where the report is null, we can skip |
| 470 | // if we have invalidations for the period and name, but only for a specific reports, we can skip |
| 471 | // if the report is not null we only want to rearchive if we have invalidation for that report |
| 472 | // if we don't find invalidation for that report, we can skip |
| 473 | $hasInvalidationsForPeriodAndName = $this->dataAccessModel->hasInvalidationForPeriodAndName($params->getSite()->getId(), $params->getPeriod(), $doneFlag, $params->getArchiveOnlyReport()); |
| 474 | if ($hasInvalidationsForPeriodAndName) { |
| 475 | return [\false, 'Has invalidations for period and name']; |
| 476 | } else { |
| 477 | return [\true, 'No invalidations for period and name']; |
| 478 | } |
| 479 | } |
| 480 | return [\false, 'Segment archive date set or segment archive start date is earlier than period end of day']; |
| 481 | } |
| 482 | public function canSkipArchiveForSegment() |
| 483 | { |
| 484 | return $this->canSkipArchiveForSegmentWithReason()[0]; |
| 485 | } |
| 486 | private function isWebsiteUsingTheTracker($idSite) |
| 487 | { |
| 488 | $idSitesNotUsingTracker = self::getSitesNotUsingTracker(); |
| 489 | $isUsingTracker = !in_array($idSite, $idSitesNotUsingTracker); |
| 490 | return $isUsingTracker; |
| 491 | } |
| 492 | public static function getSitesNotUsingTracker() |
| 493 | { |
| 494 | $cache = Cache::getTransientCache(); |
| 495 | $cacheKey = 'Archiving.isWebsiteUsingTheTracker'; |
| 496 | $idSitesNotUsingTracker = $cache->fetch($cacheKey); |
| 497 | if ($idSitesNotUsingTracker === \false || !isset($idSitesNotUsingTracker)) { |
| 498 | // we want to trigger event only once |
| 499 | $idSitesNotUsingTracker = array(); |
| 500 | /** |
| 501 | * This event is triggered when detecting whether there are sites that do not use the tracker. |
| 502 | * |
| 503 | * By default we only archive a site when there was actually any visit since the last archiving. |
| 504 | * However, some plugins do import data from another source instead of using the tracker and therefore |
| 505 | * will never have any visits for this site. To make sure we still archive data for such a site when |
| 506 | * archiving for this site is requested, you can listen to this event and add the idSite to the list of |
| 507 | * sites that do not use the tracker. |
| 508 | * |
| 509 | * @param bool $idSitesNotUsingTracker The list of idSites that rather import data instead of using the tracker |
| 510 | */ |
| 511 | Piwik::postEvent('CronArchive.getIdSitesNotUsingTracker', array(&$idSitesNotUsingTracker)); |
| 512 | $cache->save($cacheKey, $idSitesNotUsingTracker); |
| 513 | } |
| 514 | return $idSitesNotUsingTracker; |
| 515 | } |
| 516 | private function hasSiteVisitsBetweenTimeframe($idSite, Period $period) : bool |
| 517 | { |
| 518 | $cacheKeyStr = 'Archiving.hasSiteVisitsBetweenTimeframe.%s.%s'; |
| 519 | $cacheKey = CacheId::siteAware(sprintf($cacheKeyStr, $period->getLabel(), $period->getRangeString()), [$idSite]); |
| 520 | if ($this->cache->contains($cacheKey)) { |
| 521 | return $this->cache->fetch($cacheKey); |
| 522 | } |
| 523 | $timezone = Site::getTimezoneFor($idSite); |
| 524 | /** @var Date $date1 */ |
| 525 | /** @var Date $date2 */ |
| 526 | [$date1, $date2] = $period->getBoundsInTimezone($timezone); |
| 527 | $hasSiteVisitsBetweenTimeframe = $this->rawLogDao->hasSiteVisitsBetweenTimeframe($date1->getDatetime(), $date2->getDatetime(), $idSite); |
| 528 | $this->cache->save($cacheKey, $hasSiteVisitsBetweenTimeframe); |
| 529 | if ($hasSiteVisitsBetweenTimeframe) { |
| 530 | $currentPeriod = $period; |
| 531 | do { |
| 532 | $parentPeriodLabel = $currentPeriod->getParentPeriodLabel(); |
| 533 | if (!Period\Factory::isPeriodEnabledForAPI($parentPeriodLabel)) { |
| 534 | $parentPeriodLabel = null; |
| 535 | } |
| 536 | if ($parentPeriodLabel) { |
| 537 | $parentPeriod = Period\Factory::build($parentPeriodLabel, $date1); |
| 538 | $cacheKey = CacheId::siteAware(sprintf($cacheKeyStr, $parentPeriod->getLabel(), $parentPeriod->getRangeString()), [$idSite]); |
| 539 | $this->cache->save($cacheKey, \true); |
| 540 | $currentPeriod = $parentPeriod; |
| 541 | } |
| 542 | } while ($parentPeriodLabel); |
| 543 | } |
| 544 | return $hasSiteVisitsBetweenTimeframe; |
| 545 | } |
| 546 | public static function getArchivingDepth() |
| 547 | { |
| 548 | return self::$archivingDepth; |
| 549 | } |
| 550 | private function shouldForceInvalidatedArchive($value, $tsArchived) |
| 551 | { |
| 552 | $params = $this->params; |
| 553 | // the archive is invalidated and we are in a browser request that is allowed archive it |
| 554 | if ($value == ArchiveWriter::DONE_INVALIDATED && \Piwik\ArchiveProcessor\Rules::isArchivingEnabledFor([$params->getSite()->getId()], $params->getSegment(), $params->getPeriod()->getLabel())) { |
| 555 | // if coming from core:archive, force rearchiving, since if we don't the entry will be removed from archive_invalidations |
| 556 | // w/o being rearchived |
| 557 | if (SettingsServer::isArchivePhpTriggered()) { |
| 558 | return \true; |
| 559 | } |
| 560 | // if coming from a browser request, and period does not contain today, force rearchiving |
| 561 | $timezone = $params->getSite()->getTimezone(); |
| 562 | if (!$params->getPeriod()->isDateInPeriod(Date::factoryInTimezone('today', $timezone))) { |
| 563 | return \true; |
| 564 | } |
| 565 | // if coming from a browser request, and period does contain today, check the ttl for the period (done just below this) |
| 566 | $minDatetimeArchiveProcessedUTC = \Piwik\ArchiveProcessor\Rules::getMinTimeProcessedForInProgressArchive($params->getDateStart(), $params->getPeriod(), $params->getSegment(), $params->getSite()); |
| 567 | $minDatetimeArchiveProcessedUTC = Date::factory($minDatetimeArchiveProcessedUTC); |
| 568 | if ($minDatetimeArchiveProcessedUTC && Date::factory($tsArchived)->isEarlier($minDatetimeArchiveProcessedUTC)) { |
| 569 | return \true; |
| 570 | } |
| 571 | } |
| 572 | return \false; |
| 573 | } |
| 574 | } |
| 575 |