ArchivingStatus.php
5 years ago
Loader.php
5 years ago
Parameters.php
5 years ago
PluginsArchiver.php
5 years ago
PluginsArchiverException.php
5 years ago
Rules.php
5 years ago
Loader.php
454 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\ArchiveProcessor; |
| 10 | |
| 11 | use Piwik\Archive\ArchiveInvalidator; |
| 12 | use Piwik\Cache; |
| 13 | use Piwik\Common; |
| 14 | use Piwik\Config; |
| 15 | use Piwik\Container\StaticContainer; |
| 16 | use Piwik\Context; |
| 17 | use Piwik\DataAccess\ArchiveSelector; |
| 18 | use Piwik\DataAccess\ArchiveTableCreator; |
| 19 | use Piwik\DataAccess\Model; |
| 20 | use Piwik\DataAccess\RawLogDao; |
| 21 | use Piwik\Date; |
| 22 | use Piwik\Db; |
| 23 | use Piwik\Period; |
| 24 | use Piwik\Piwik; |
| 25 | use Piwik\SettingsServer; |
| 26 | use Piwik\Site; |
| 27 | use Psr\Log\LoggerInterface; |
| 28 | |
| 29 | /** |
| 30 | * This class uses PluginsArchiver class to trigger data aggregation and create archives. |
| 31 | */ |
| 32 | class Loader |
| 33 | { |
| 34 | const MIN_VISIT_TIME_TTL = 3600; |
| 35 | |
| 36 | /** |
| 37 | * @var Parameters |
| 38 | */ |
| 39 | protected $params; |
| 40 | |
| 41 | /** |
| 42 | * @var ArchiveInvalidator |
| 43 | */ |
| 44 | private $invalidator; |
| 45 | |
| 46 | /** |
| 47 | * @var \Matomo\Cache\Cache |
| 48 | */ |
| 49 | private $cache; |
| 50 | |
| 51 | /** |
| 52 | * @var LoggerInterface |
| 53 | */ |
| 54 | private $logger; |
| 55 | |
| 56 | /** |
| 57 | * @var RawLogDao |
| 58 | */ |
| 59 | private $rawLogDao; |
| 60 | |
| 61 | /** |
| 62 | * @var Model |
| 63 | */ |
| 64 | private $dataAccessModel; |
| 65 | |
| 66 | public function __construct(Parameters $params, $invalidateBeforeArchiving = false) |
| 67 | { |
| 68 | $this->params = $params; |
| 69 | $this->invalidateBeforeArchiving = $invalidateBeforeArchiving; |
| 70 | $this->invalidator = StaticContainer::get(ArchiveInvalidator::class); |
| 71 | $this->cache = Cache::getTransientCache(); |
| 72 | $this->logger = StaticContainer::get(LoggerInterface::class); |
| 73 | $this->rawLogDao = new RawLogDao(); |
| 74 | $this->dataAccessModel = new Model(); |
| 75 | } |
| 76 | |
| 77 | /** |
| 78 | * @return bool |
| 79 | */ |
| 80 | protected function isThereSomeVisits($visits) |
| 81 | { |
| 82 | return $visits > 0; |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * @return bool |
| 87 | */ |
| 88 | protected function mustProcessVisitCount($visits) |
| 89 | { |
| 90 | return $visits === false; |
| 91 | } |
| 92 | |
| 93 | public function prepareArchive($pluginName) |
| 94 | { |
| 95 | return Context::changeIdSite($this->params->getSite()->getId(), function () use ($pluginName) { |
| 96 | return $this->prepareArchiveImpl($pluginName); |
| 97 | }); |
| 98 | } |
| 99 | |
| 100 | private function prepareArchiveImpl($pluginName) |
| 101 | { |
| 102 | $this->params->setRequestedPlugin($pluginName); |
| 103 | |
| 104 | if (SettingsServer::isArchivePhpTriggered()) { |
| 105 | $requestedReport = Common::getRequestVar('requestedReport', '', 'string'); |
| 106 | if (!empty($requestedReport)) { |
| 107 | $this->params->setArchiveOnlyReport($requestedReport); |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | // NOTE: $idArchives will contain the latest DONE_OK/DONE_INVALIDATED archive as well as any partial archives |
| 112 | // with a ts_archived >= the DONE_OK/DONE_INVALIDATED date. |
| 113 | list($idArchives, $visits, $visitsConverted, $isAnyArchiveExists) = $this->loadExistingArchiveIdFromDb(); |
| 114 | if (!empty($idArchives) |
| 115 | && !$this->params->getArchiveOnlyReport() |
| 116 | && !Rules::isForceArchivingSinglePlugin() |
| 117 | ) { |
| 118 | // we have a usable idarchive (it's not invalidated and it's new enough), and we are not archiving |
| 119 | // a single report |
| 120 | return [$idArchives, $visits]; |
| 121 | } |
| 122 | |
| 123 | // NOTE: this optimization helps when archiving large periods. eg, if archiving a year w/ a segment where |
| 124 | // there are not visits in the entire year, we don't have to go through and do anything. but, w/o this |
| 125 | // code, we will end up launching archiving for each month, week and day, even though we don't have to. |
| 126 | // |
| 127 | // we don't create an archive in this case, because the archive may be in progress in some way, so a 0 |
| 128 | // visits archive can be inaccurate in the long run. |
| 129 | if ($this->canSkipThisArchive()) { |
| 130 | return [false, 0]; |
| 131 | } |
| 132 | |
| 133 | // if there is an archive, but we can't use it for some reason, invalidate existing archives before |
| 134 | // we start archiving. if the archive is made invalid, we will correctly re-archive below. |
| 135 | if ($this->invalidateBeforeArchiving |
| 136 | && $isAnyArchiveExists |
| 137 | ) { |
| 138 | $this->invalidatedReportsIfNeeded(); |
| 139 | } |
| 140 | |
| 141 | /** @var ArchivingStatus $archivingStatus */ |
| 142 | $archivingStatus = StaticContainer::get(ArchivingStatus::class); |
| 143 | $locked = $archivingStatus->archiveStarted($this->params); |
| 144 | |
| 145 | try { |
| 146 | list($visits, $visitsConverted) = $this->prepareCoreMetricsArchive($visits, $visitsConverted); |
| 147 | list($idArchive, $visits) = $this->prepareAllPluginsArchive($visits, $visitsConverted); |
| 148 | } finally { |
| 149 | if ($locked) { |
| 150 | $archivingStatus->archiveFinished(); |
| 151 | } |
| 152 | } |
| 153 | |
| 154 | if ($this->isThereSomeVisits($visits) || PluginsArchiver::doesAnyPluginArchiveWithoutVisits()) { |
| 155 | return [[$idArchive], $visits]; |
| 156 | } |
| 157 | |
| 158 | return [false, false]; |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * Prepares the core metrics if needed. |
| 163 | * |
| 164 | * @param $visits |
| 165 | * @return array |
| 166 | */ |
| 167 | protected function prepareCoreMetricsArchive($visits, $visitsConverted) |
| 168 | { |
| 169 | $createSeparateArchiveForCoreMetrics = $this->mustProcessVisitCount($visits) |
| 170 | && !$this->doesRequestedPluginIncludeVisitsSummary(); |
| 171 | |
| 172 | if ($createSeparateArchiveForCoreMetrics) { |
| 173 | $requestedPlugin = $this->params->getRequestedPlugin(); |
| 174 | $requestedReport = $this->params->getArchiveOnlyReport(); |
| 175 | |
| 176 | $this->params->setRequestedPlugin('VisitsSummary'); |
| 177 | $this->params->setArchiveOnlyReport(null); |
| 178 | |
| 179 | $pluginsArchiver = new PluginsArchiver($this->params); |
| 180 | $metrics = $pluginsArchiver->callAggregateCoreMetrics(); |
| 181 | $pluginsArchiver->finalizeArchive(); |
| 182 | |
| 183 | $this->params->setRequestedPlugin($requestedPlugin); |
| 184 | $this->params->setArchiveOnlyReport($requestedReport); |
| 185 | |
| 186 | $visits = $metrics['nb_visits']; |
| 187 | $visitsConverted = $metrics['nb_visits_converted']; |
| 188 | } |
| 189 | |
| 190 | return array($visits, $visitsConverted); |
| 191 | } |
| 192 | |
| 193 | protected function prepareAllPluginsArchive($visits, $visitsConverted) |
| 194 | { |
| 195 | $pluginsArchiver = new PluginsArchiver($this->params); |
| 196 | |
| 197 | if ($this->mustProcessVisitCount($visits) |
| 198 | || $this->doesRequestedPluginIncludeVisitsSummary() |
| 199 | ) { |
| 200 | $metrics = $pluginsArchiver->callAggregateCoreMetrics(); |
| 201 | $visits = $metrics['nb_visits']; |
| 202 | $visitsConverted = $metrics['nb_visits_converted']; |
| 203 | } |
| 204 | |
| 205 | $forceArchivingWithoutVisits = !$this->isThereSomeVisits($visits) && $this->shouldArchiveForSiteEvenWhenNoVisits(); |
| 206 | $pluginsArchiver->callAggregateAllPlugins($visits, $visitsConverted, $forceArchivingWithoutVisits); |
| 207 | |
| 208 | $idArchive = $pluginsArchiver->finalizeArchive(); |
| 209 | |
| 210 | return array($idArchive, $visits); |
| 211 | } |
| 212 | |
| 213 | protected function doesRequestedPluginIncludeVisitsSummary() |
| 214 | { |
| 215 | $processAllReportsIncludingVisitsSummary = |
| 216 | Rules::shouldProcessReportsAllPlugins(array($this->params->getSite()->getId()), $this->params->getSegment(), $this->params->getPeriod()->getLabel()); |
| 217 | $doesRequestedPluginIncludeVisitsSummary = $processAllReportsIncludingVisitsSummary |
| 218 | || $this->params->getRequestedPlugin() == 'VisitsSummary'; |
| 219 | return $doesRequestedPluginIncludeVisitsSummary; |
| 220 | } |
| 221 | |
| 222 | protected function isArchivingForcedToTrigger() |
| 223 | { |
| 224 | $period = $this->params->getPeriod()->getLabel(); |
| 225 | $debugSetting = 'always_archive_data_period'; // default |
| 226 | |
| 227 | if ($period == 'day') { |
| 228 | $debugSetting = 'always_archive_data_day'; |
| 229 | } elseif ($period == 'range') { |
| 230 | $debugSetting = 'always_archive_data_range'; |
| 231 | } |
| 232 | |
| 233 | return (bool) Config::getInstance()->Debug[$debugSetting]; |
| 234 | } |
| 235 | |
| 236 | /** |
| 237 | * Returns the idArchive if the archive is available in the database for the requested plugin. |
| 238 | * Returns false if the archive needs to be processed. |
| 239 | * |
| 240 | * (public for tests) |
| 241 | * |
| 242 | * @return array |
| 243 | */ |
| 244 | public function loadExistingArchiveIdFromDb() |
| 245 | { |
| 246 | if ($this->isArchivingForcedToTrigger()) { |
| 247 | $this->logger->debug("Archiving forced to trigger for {$this->params}."); |
| 248 | |
| 249 | // return no usable archive found, and no existing archive. this will skip invalidation, which should |
| 250 | // be fine since we just force archiving. |
| 251 | return [false, false, false, false]; |
| 252 | } |
| 253 | |
| 254 | $minDatetimeArchiveProcessedUTC = $this->getMinTimeArchiveProcessed(); |
| 255 | $result = ArchiveSelector::getArchiveIdAndVisits($this->params, $minDatetimeArchiveProcessedUTC); |
| 256 | return $result; |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * Returns the minimum archive processed datetime to look at. Only public for tests. |
| 261 | * |
| 262 | * @return int|bool Datetime timestamp, or false if must look at any archive available |
| 263 | */ |
| 264 | protected function getMinTimeArchiveProcessed() |
| 265 | { |
| 266 | $endDateTimestamp = self::determineIfArchivePermanent($this->params->getDateEnd()); |
| 267 | if ($endDateTimestamp) { |
| 268 | // past archive |
| 269 | return $endDateTimestamp; |
| 270 | } |
| 271 | $dateStart = $this->params->getDateStart(); |
| 272 | $period = $this->params->getPeriod(); |
| 273 | $segment = $this->params->getSegment(); |
| 274 | $site = $this->params->getSite(); |
| 275 | // in-progress archive |
| 276 | return Rules::getMinTimeProcessedForInProgressArchive($dateStart, $period, $segment, $site); |
| 277 | } |
| 278 | |
| 279 | protected static function determineIfArchivePermanent(Date $dateEnd) |
| 280 | { |
| 281 | $now = time(); |
| 282 | $endTimestampUTC = strtotime($dateEnd->getDateEndUTC()); |
| 283 | |
| 284 | if ($endTimestampUTC <= $now) { |
| 285 | // - if the period we are looking for is finished, we look for a ts_archived that |
| 286 | // is greater than the last day of the archive |
| 287 | return $endTimestampUTC; |
| 288 | } |
| 289 | |
| 290 | return false; |
| 291 | } |
| 292 | |
| 293 | private function shouldArchiveForSiteEvenWhenNoVisits() |
| 294 | { |
| 295 | $idSitesToArchive = $this->getIdSitesToArchiveWhenNoVisits(); |
| 296 | return in_array($this->params->getSite()->getId(), $idSitesToArchive); |
| 297 | } |
| 298 | |
| 299 | private function getIdSitesToArchiveWhenNoVisits() |
| 300 | { |
| 301 | $cache = Cache::getTransientCache(); |
| 302 | $cacheKey = 'Archiving.getIdSitesToArchiveWhenNoVisits'; |
| 303 | |
| 304 | if (!$cache->contains($cacheKey)) { |
| 305 | $idSites = array(); |
| 306 | |
| 307 | // leaving undocumented unless decided otherwise |
| 308 | Piwik::postEvent('Archiving.getIdSitesToArchiveWhenNoVisits', array(&$idSites)); |
| 309 | |
| 310 | $cache->save($cacheKey, $idSites); |
| 311 | } |
| 312 | |
| 313 | return $cache->fetch($cacheKey); |
| 314 | } |
| 315 | |
| 316 | // public for tests |
| 317 | public function getReportsToInvalidate() |
| 318 | { |
| 319 | $sitesPerDays = $this->invalidator->getRememberedArchivedReportsThatShouldBeInvalidated(); |
| 320 | |
| 321 | foreach ($sitesPerDays as $dateStr => $siteIds) { |
| 322 | if (empty($siteIds) |
| 323 | || !in_array($this->params->getSite()->getId(), $siteIds) |
| 324 | ) { |
| 325 | unset($sitesPerDays[$dateStr]); |
| 326 | } |
| 327 | |
| 328 | $date = Date::factory($dateStr); |
| 329 | if ($date->isEarlier($this->params->getPeriod()->getDateStart()) |
| 330 | || $date->isLater($this->params->getPeriod()->getDateEnd()) |
| 331 | ) { // date in list is not the current date, so ignore it |
| 332 | unset($sitesPerDays[$dateStr]); |
| 333 | } |
| 334 | } |
| 335 | |
| 336 | return $sitesPerDays; |
| 337 | } |
| 338 | |
| 339 | private function invalidatedReportsIfNeeded() |
| 340 | { |
| 341 | $sitesPerDays = $this->getReportsToInvalidate(); |
| 342 | if (empty($sitesPerDays)) { |
| 343 | return; |
| 344 | } |
| 345 | |
| 346 | foreach ($sitesPerDays as $date => $siteIds) { |
| 347 | try { |
| 348 | $this->invalidator->markArchivesAsInvalidated([$this->params->getSite()->getId()], array(Date::factory($date)), false, $this->params->getSegment()); |
| 349 | } catch (\Exception $e) { |
| 350 | Site::clearCache(); |
| 351 | throw $e; |
| 352 | } |
| 353 | } |
| 354 | |
| 355 | Site::clearCache(); |
| 356 | } |
| 357 | |
| 358 | public function canSkipThisArchive() |
| 359 | { |
| 360 | $params = $this->params; |
| 361 | $idSite = $params->getSite()->getId(); |
| 362 | |
| 363 | $isWebsiteUsingTracker = $this->isWebsiteUsingTheTracker($idSite); |
| 364 | $isArchivingForcedWhenNoVisits = $this->shouldArchiveForSiteEvenWhenNoVisits(); |
| 365 | $hasSiteVisitsBetweenTimeframe = $this->hasSiteVisitsBetweenTimeframe($idSite, $params->getPeriod()); |
| 366 | $hasChildArchivesInPeriod = $this->dataAccessModel->hasChildArchivesInPeriod($idSite, $params->getPeriod()); |
| 367 | |
| 368 | return $isWebsiteUsingTracker |
| 369 | && !$isArchivingForcedWhenNoVisits |
| 370 | && !$hasSiteVisitsBetweenTimeframe |
| 371 | && !$hasChildArchivesInPeriod; |
| 372 | } |
| 373 | |
| 374 | private function isWebsiteUsingTheTracker($idSite) |
| 375 | { |
| 376 | $idSitesNotUsingTracker = self::getSitesNotUsingTracker(); |
| 377 | |
| 378 | $isUsingTracker = !in_array($idSite, $idSitesNotUsingTracker); |
| 379 | |
| 380 | return $isUsingTracker; |
| 381 | } |
| 382 | |
| 383 | public static function getSitesNotUsingTracker() |
| 384 | { |
| 385 | $cache = Cache::getTransientCache(); |
| 386 | |
| 387 | $cacheKey = 'Archiving.isWebsiteUsingTheTracker'; |
| 388 | $idSitesNotUsingTracker = $cache->fetch($cacheKey); |
| 389 | if ($idSitesNotUsingTracker === false || !isset($idSitesNotUsingTracker)) { |
| 390 | // we want to trigger event only once |
| 391 | $idSitesNotUsingTracker = array(); |
| 392 | |
| 393 | /** |
| 394 | * This event is triggered when detecting whether there are sites that do not use the tracker. |
| 395 | * |
| 396 | * By default we only archive a site when there was actually any visit since the last archiving. |
| 397 | * However, some plugins do import data from another source instead of using the tracker and therefore |
| 398 | * will never have any visits for this site. To make sure we still archive data for such a site when |
| 399 | * archiving for this site is requested, you can listen to this event and add the idSite to the list of |
| 400 | * sites that do not use the tracker. |
| 401 | * |
| 402 | * @param bool $idSitesNotUsingTracker The list of idSites that rather import data instead of using the tracker |
| 403 | */ |
| 404 | Piwik::postEvent('CronArchive.getIdSitesNotUsingTracker', array(&$idSitesNotUsingTracker)); |
| 405 | |
| 406 | $cache->save($cacheKey, $idSitesNotUsingTracker); |
| 407 | } |
| 408 | return $idSitesNotUsingTracker; |
| 409 | } |
| 410 | |
| 411 | private function hasSiteVisitsBetweenTimeframe($idSite, Period $period) |
| 412 | { |
| 413 | $minVisitTimesPerSite = $this->getMinVisitTimesPerSite($idSite); |
| 414 | if (empty($minVisitTimesPerSite)) { |
| 415 | return false; |
| 416 | } |
| 417 | |
| 418 | $timezone = Site::getTimezoneFor($idSite); |
| 419 | list($date1, $date2) = $period->getBoundsInTimezone($timezone); |
| 420 | if ($date2->isEarlier($minVisitTimesPerSite)) { |
| 421 | return false; |
| 422 | } |
| 423 | |
| 424 | return $this->rawLogDao->hasSiteVisitsBetweenTimeframe($date1->getDatetime(), $date2->getDatetime(), $idSite); |
| 425 | } |
| 426 | |
| 427 | private function getMinVisitTimesPerSite($idSite) |
| 428 | { |
| 429 | $cache = Cache::getLazyCache(); |
| 430 | $cacheKey = 'Archiving.minVisitTime.' . $idSite; |
| 431 | |
| 432 | $value = $cache->fetch($cacheKey); |
| 433 | if ($value === false) { |
| 434 | $value = $this->rawLogDao->getMinimumVisitTimeForSite($idSite); |
| 435 | if (!empty($value)) { |
| 436 | $cache->save($cacheKey, $value, $ttl = self::MIN_VISIT_TIME_TTL); |
| 437 | } |
| 438 | } |
| 439 | |
| 440 | if (!empty($value)) { |
| 441 | $value = Date::factory($value); |
| 442 | } |
| 443 | |
| 444 | return $value; |
| 445 | } |
| 446 | |
| 447 | public static function invalidateMinVisitTimeCache($idSite) |
| 448 | { |
| 449 | $cache = Cache::getLazyCache(); |
| 450 | $cacheKey = 'Archiving.minVisitTime.' . $idSite; |
| 451 | $cache->delete($cacheKey); |
| 452 | } |
| 453 | } |
| 454 |