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
LogAggregator.php
1216 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 Piwik\ArchiveProcessor\ArchivingStatus; |
| 12 | use Piwik\ArchiveProcessor\Parameters; |
| 13 | use Piwik\Common; |
| 14 | use Piwik\Config; |
| 15 | use Piwik\Container\StaticContainer; |
| 16 | use Piwik\DataArray; |
| 17 | use Piwik\Date; |
| 18 | use Piwik\Db; |
| 19 | use Piwik\DbHelper; |
| 20 | use Piwik\Metrics; |
| 21 | use Piwik\Plugin\LogTablesProvider; |
| 22 | use Piwik\Segment; |
| 23 | use Piwik\Tracker\GoalManager; |
| 24 | use Psr\Log\LoggerInterface; |
| 25 | |
| 26 | /** |
| 27 | * Contains methods that calculate metrics by aggregating log data (visits, actions, conversions, |
| 28 | * ecommerce items). |
| 29 | * |
| 30 | * You can use the methods in this class within {@link Piwik\Plugin\Archiver Archiver} descendants |
| 31 | * to aggregate log data without having to write SQL queries. |
| 32 | * |
| 33 | * ### Aggregation Dimension |
| 34 | * |
| 35 | * All aggregation methods accept a **dimension** parameter. These parameters are important as |
| 36 | * they control how rows in a table are aggregated together. |
| 37 | * |
| 38 | * A **_dimension_** is just a table column. Rows that have the same values for these columns are |
| 39 | * aggregated together. The result of these aggregations is a set of metrics for every recorded value |
| 40 | * of a **dimension**. |
| 41 | * |
| 42 | * _Note: A dimension is essentially the same as a **GROUP BY** field._ |
| 43 | * |
| 44 | * ### Examples |
| 45 | * |
| 46 | * **Aggregating visit data** |
| 47 | * |
| 48 | * $archiveProcessor = // ... |
| 49 | * $logAggregator = $archiveProcessor->getLogAggregator(); |
| 50 | * |
| 51 | * // get metrics for every used browser language of all visits by returning visitors |
| 52 | * $query = $logAggregator->queryVisitsByDimension( |
| 53 | * $dimensions = array('log_visit.location_browser_lang'), |
| 54 | * $where = 'log_visit.visitor_returning = 1', |
| 55 | * |
| 56 | * // also count visits for each browser language that are not located in the US |
| 57 | * $additionalSelects = array('sum(case when log_visit.location_country <> 'us' then 1 else 0 end) as nonus'), |
| 58 | * |
| 59 | * // we're only interested in visits, unique visitors & actions, so don't waste time calculating anything else |
| 60 | * $metrics = array(Metrics::INDEX_NB_UNIQ_VISITORS, Metrics::INDEX_NB_VISITS, Metrics::INDEX_NB_ACTIONS), |
| 61 | * ); |
| 62 | * if ($query === false) { |
| 63 | * return; |
| 64 | * } |
| 65 | * |
| 66 | * while ($row = $query->fetch()) { |
| 67 | * $uniqueVisitors = $row[Metrics::INDEX_NB_UNIQ_VISITORS]; |
| 68 | * $visits = $row[Metrics::INDEX_NB_VISITS]; |
| 69 | * $actions = $row[Metrics::INDEX_NB_ACTIONS]; |
| 70 | * |
| 71 | * // ... do something w/ calculated metrics ... |
| 72 | * } |
| 73 | * |
| 74 | * **Aggregating conversion data** |
| 75 | * |
| 76 | * $archiveProcessor = // ... |
| 77 | * $logAggregator = $archiveProcessor->getLogAggregator(); |
| 78 | * |
| 79 | * // get metrics for ecommerce conversions for each country |
| 80 | * $query = $logAggregator->queryConversionsByDimension( |
| 81 | * $dimensions = array('log_conversion.location_country'), |
| 82 | * $where = 'log_conversion.idgoal = 0', // 0 is the special ecommerceOrder idGoal value in the table |
| 83 | * |
| 84 | * // also calculate average tax and max shipping per country |
| 85 | * $additionalSelects = array( |
| 86 | * 'AVG(log_conversion.revenue_tax) as avg_tax', |
| 87 | * 'MAX(log_conversion.revenue_shipping) as max_shipping' |
| 88 | * ) |
| 89 | * ); |
| 90 | * if ($query === false) { |
| 91 | * return; |
| 92 | * } |
| 93 | * |
| 94 | * while ($row = $query->fetch()) { |
| 95 | * $country = $row['location_country']; |
| 96 | * $numEcommerceSales = $row[Metrics::INDEX_GOAL_NB_CONVERSIONS]; |
| 97 | * $numVisitsWithEcommerceSales = $row[Metrics::INDEX_GOAL_NB_VISITS_CONVERTED]; |
| 98 | * $avgTaxForCountry = $row['avg_tax']; |
| 99 | * $maxShippingForCountry = $row['max_shipping']; |
| 100 | * |
| 101 | * // ... do something with aggregated data ... |
| 102 | * } |
| 103 | */ |
| 104 | class LogAggregator |
| 105 | { |
| 106 | const LOG_VISIT_TABLE = 'log_visit'; |
| 107 | |
| 108 | const LOG_ACTIONS_TABLE = 'log_link_visit_action'; |
| 109 | |
| 110 | const LOG_CONVERSION_TABLE = "log_conversion"; |
| 111 | |
| 112 | const REVENUE_SUBTOTAL_FIELD = 'revenue_subtotal'; |
| 113 | |
| 114 | const REVENUE_TAX_FIELD = 'revenue_tax'; |
| 115 | |
| 116 | const REVENUE_SHIPPING_FIELD = 'revenue_shipping'; |
| 117 | |
| 118 | const REVENUE_DISCOUNT_FIELD = 'revenue_discount'; |
| 119 | |
| 120 | const TOTAL_REVENUE_FIELD = 'revenue'; |
| 121 | |
| 122 | const ITEMS_COUNT_FIELD = "items"; |
| 123 | |
| 124 | const CONVERSION_DATETIME_FIELD = "server_time"; |
| 125 | |
| 126 | const ACTION_DATETIME_FIELD = "server_time"; |
| 127 | |
| 128 | const VISIT_DATETIME_FIELD = 'visit_last_action_time'; |
| 129 | |
| 130 | const IDGOAL_FIELD = 'idgoal'; |
| 131 | |
| 132 | const FIELDS_SEPARATOR = ", \n\t\t\t"; |
| 133 | |
| 134 | const LOG_TABLE_SEGMENT_TEMPORARY_PREFIX = 'logtmpsegment'; |
| 135 | |
| 136 | /** @var \Piwik\Date */ |
| 137 | protected $dateStart; |
| 138 | |
| 139 | /** @var \Piwik\Date */ |
| 140 | protected $dateEnd; |
| 141 | |
| 142 | /** @var int[] */ |
| 143 | protected $sites; |
| 144 | |
| 145 | /** @var \Piwik\Segment */ |
| 146 | protected $segment; |
| 147 | |
| 148 | /** |
| 149 | * @var string |
| 150 | */ |
| 151 | private $queryOriginHint = ''; |
| 152 | |
| 153 | /** |
| 154 | * @var LoggerInterface |
| 155 | */ |
| 156 | private $logger; |
| 157 | |
| 158 | /** |
| 159 | * @var bool |
| 160 | */ |
| 161 | private $isRootArchiveRequest; |
| 162 | |
| 163 | /** |
| 164 | * @var bool |
| 165 | */ |
| 166 | private $allowUsageSegmentCache = false; |
| 167 | |
| 168 | /** |
| 169 | * @var Parameters |
| 170 | */ |
| 171 | private $params; |
| 172 | |
| 173 | /** |
| 174 | * Constructor. |
| 175 | * |
| 176 | * @param \Piwik\ArchiveProcessor\Parameters $params |
| 177 | */ |
| 178 | public function __construct(Parameters $params, LoggerInterface $logger = null) |
| 179 | { |
| 180 | $this->dateStart = $params->getDateTimeStart(); |
| 181 | $this->dateEnd = $params->getDateTimeEnd(); |
| 182 | $this->segment = $params->getSegment(); |
| 183 | $this->sites = $params->getIdSites(); |
| 184 | $this->isRootArchiveRequest = $params->isRootArchiveRequest(); |
| 185 | $this->logger = $logger ?: StaticContainer::get('Psr\Log\LoggerInterface'); |
| 186 | $this->params = $params; |
| 187 | } |
| 188 | |
| 189 | public function setSites($sites) |
| 190 | { |
| 191 | $this->sites = array_map('intval', $sites); |
| 192 | } |
| 193 | |
| 194 | public function getSites() |
| 195 | { |
| 196 | return $this->sites; |
| 197 | } |
| 198 | |
| 199 | public function getSegment() |
| 200 | { |
| 201 | return $this->segment; |
| 202 | } |
| 203 | |
| 204 | public function setQueryOriginHint($nameOfOrigiin) |
| 205 | { |
| 206 | $this->queryOriginHint = $nameOfOrigiin; |
| 207 | } |
| 208 | |
| 209 | public function getSegmentTmpTableName() |
| 210 | { |
| 211 | $bind = $this->getGeneralQueryBindParams(); |
| 212 | $tableName = self::LOG_TABLE_SEGMENT_TEMPORARY_PREFIX . md5(json_encode($bind) . $this->segment->getString()); |
| 213 | |
| 214 | $lengthPrefix = Common::mb_strlen(Common::prefixTable('')); |
| 215 | $maxLength = Db\Schema\Mysql::MAX_TABLE_NAME_LENGTH - $lengthPrefix; |
| 216 | |
| 217 | return Common::mb_substr($tableName, 0, $maxLength); |
| 218 | } |
| 219 | |
| 220 | public function cleanup() |
| 221 | { |
| 222 | if (!$this->segment->isEmpty() && $this->isSegmentCacheEnabled()) { |
| 223 | $segmentTable = $this->getSegmentTmpTableName(); |
| 224 | $segmentTable = Common::prefixTable($segmentTable); |
| 225 | |
| 226 | if ($this->doesSegmentTableExist($segmentTable)) { |
| 227 | // safety in case an older MySQL version is used that does not drop table at the end of the connection |
| 228 | // automatically. also helps us release disk space/memory earlier when multiple segments are archived |
| 229 | $this->getDb()->query('DROP TEMPORARY TABLE IF EXISTS ' . $segmentTable); |
| 230 | } |
| 231 | |
| 232 | $logTablesProvider = $this->getLogTableProvider(); |
| 233 | if ($logTablesProvider->getLogTable($segmentTable)) { |
| 234 | $logTablesProvider->setTempTable(null); // no longer available |
| 235 | } |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | private function doesSegmentTableExist($segmentTablePrefixed) |
| 240 | { |
| 241 | try { |
| 242 | // using DROP TABLE IF EXISTS would not work on a DB reader if the table doesn't exist... |
| 243 | $this->getDb()->fetchOne('SELECT 1 FROM ' . $segmentTablePrefixed . ' LIMIT 1'); |
| 244 | $tableExists = true; |
| 245 | } catch (\Exception $e) { |
| 246 | $tableExists = false; |
| 247 | } |
| 248 | |
| 249 | return $tableExists; |
| 250 | } |
| 251 | |
| 252 | private function isSegmentCacheEnabled() |
| 253 | { |
| 254 | if (!$this->allowUsageSegmentCache) { |
| 255 | return false; |
| 256 | } |
| 257 | |
| 258 | $config = Config::getInstance(); |
| 259 | $general = $config->General; |
| 260 | return !empty($general['enable_segments_cache']); |
| 261 | } |
| 262 | |
| 263 | public function allowUsageSegmentCache() |
| 264 | { |
| 265 | $this->allowUsageSegmentCache = true; |
| 266 | } |
| 267 | |
| 268 | private function getLogTableProvider() |
| 269 | { |
| 270 | return StaticContainer::get(LogTablesProvider::class); |
| 271 | } |
| 272 | |
| 273 | private function createTemporaryTable($unprefixedSegmentTableName, $segmentSelectSql, $segmentSelectBind) |
| 274 | { |
| 275 | $table = Common::prefixTable($unprefixedSegmentTableName); |
| 276 | |
| 277 | if ($this->doesSegmentTableExist($table)) { |
| 278 | return; // no need to create the table, it was already created... better to have a select vs unneeded create table |
| 279 | } |
| 280 | |
| 281 | $engine = ''; |
| 282 | if (defined('PIWIK_TEST_MODE') && PIWIK_TEST_MODE) { |
| 283 | $engine = 'ENGINE=MEMORY'; |
| 284 | } |
| 285 | $tempTableIdVisitColumn = 'idvisit BIGINT(10) UNSIGNED NOT NULL'; |
| 286 | $createTableSql = 'CREATE TEMPORARY TABLE ' . $table . ' (' . $tempTableIdVisitColumn . ') ' . $engine; |
| 287 | // we do not insert the data right away using create temporary table ... select ... |
| 288 | // to avoid metadata lock see eg https://www.percona.com/blog/2018/01/10/why-avoid-create-table-as-select-statement/ |
| 289 | |
| 290 | $readerDb = Db::getReader(); |
| 291 | try { |
| 292 | $readerDb->query($createTableSql); |
| 293 | } catch (\Exception $e) { |
| 294 | if ($readerDb->isErrNo($e, \Piwik\Updater\Migration\Db::ERROR_CODE_TABLE_EXISTS)) { |
| 295 | return; |
| 296 | } elseif ($readerDb->isErrNo($e, \Piwik\Updater\Migration\Db::ERROR_CODE_REQUIRES_PRIMARY_KEY) |
| 297 | || $readerDb->isErrNo($e, \Piwik\Updater\Migration\Db::ERROR_CODE_UNABLE_CREATE_TABLE_WITHOUT_PRIMARY_KEY |
| 298 | || stripos($e->getMessage(), 'requires a primary key') !== false |
| 299 | || stripos($e->getMessage(), 'table without a primary key') !== false) |
| 300 | ) { |
| 301 | $createTableSql = str_replace($tempTableIdVisitColumn, $tempTableIdVisitColumn . ', PRIMARY KEY (`idvisit`)', $createTableSql); |
| 302 | |
| 303 | try { |
| 304 | $readerDb->query($createTableSql); |
| 305 | } catch (\Exception $e) { |
| 306 | if ($readerDb->isErrNo($e, \Piwik\Updater\Migration\Db::ERROR_CODE_TABLE_EXISTS)) { |
| 307 | return; |
| 308 | } else { |
| 309 | throw $e; |
| 310 | } |
| 311 | } |
| 312 | } else { |
| 313 | throw $e; |
| 314 | } |
| 315 | } |
| 316 | |
| 317 | $transactionLevel = new Db\TransactionLevel($readerDb); |
| 318 | $canSetTransactionLevel = $transactionLevel->canLikelySetTransactionLevel(); |
| 319 | |
| 320 | if ($canSetTransactionLevel) { |
| 321 | // i know this could be shortened to one if or one line but I want to make sure this line where we |
| 322 | // set uncommitted is easily noticeable in the code as it could be missed quite easily otherwise |
| 323 | // we set uncommitted so we don't make the INSERT INTO... SELECT... locking ... we do not want to lock |
| 324 | // eg the visits table |
| 325 | if (!$transactionLevel->setUncommitted()) { |
| 326 | $canSetTransactionLevel = false; |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | if (!$canSetTransactionLevel) { |
| 331 | // transaction level doesn't work... we're instead executing the select individually and then insert the data |
| 332 | // this uses more memory but at least is not locking |
| 333 | $all = $readerDb->fetchAll($segmentSelectSql, $segmentSelectBind); |
| 334 | if (!empty($all)) { |
| 335 | // we're not using batchinsert since this would not support the reader DB. |
| 336 | $readerDb->query('INSERT INTO ' . $table . ' VALUES ('.implode('),(', array_column($all, 'idvisit')).')'); |
| 337 | } |
| 338 | return; |
| 339 | } |
| 340 | |
| 341 | $insertIntoStatement = 'INSERT INTO ' . $table . ' (idvisit) ' . $segmentSelectSql; |
| 342 | $readerDb->query($insertIntoStatement, $segmentSelectBind); |
| 343 | |
| 344 | $transactionLevel->restorePreviousStatus(); |
| 345 | } |
| 346 | |
| 347 | public function generateQuery($select, $from, $where, $groupBy, $orderBy, $limit = 0, $offset = 0) |
| 348 | { |
| 349 | $segment = $this->segment; |
| 350 | $bind = $this->getGeneralQueryBindParams(); |
| 351 | |
| 352 | if (!$this->segment->isEmpty() && $this->isSegmentCacheEnabled()) { |
| 353 | // here we create the TMP table and apply the segment including the datetime and the requested idsite |
| 354 | // at the end we generated query will no longer need to apply the datetime/idsite and segment |
| 355 | $segment = new Segment('', $this->sites, $this->params->getPeriod()->getDateTimeStart(), $this->params->getPeriod()->getDateTimeEnd()); |
| 356 | |
| 357 | $segmentTable = $this->getSegmentTmpTableName(); |
| 358 | |
| 359 | $segmentWhere = $this->getWhereStatement('log_visit', 'visit_last_action_time'); |
| 360 | $segmentBind = $this->getGeneralQueryBindParams(); |
| 361 | |
| 362 | $logQueryBuilder = StaticContainer::get('Piwik\DataAccess\LogQueryBuilder'); |
| 363 | $forceGroupByBackup = $logQueryBuilder->getForcedInnerGroupBySubselect(); |
| 364 | $logQueryBuilder->forceInnerGroupBySubselect(LogQueryBuilder::FORCE_INNER_GROUP_BY_NO_SUBSELECT); |
| 365 | $segmentSql = $this->segment->getSelectQuery('distinct log_visit.idvisit as idvisit', 'log_visit', $segmentWhere, $segmentBind, 'log_visit.idvisit ASC'); |
| 366 | $logQueryBuilder->forceInnerGroupBySubselect($forceGroupByBackup); |
| 367 | |
| 368 | $this->createTemporaryTable($segmentTable, $segmentSql['sql'], $segmentSql['bind']); |
| 369 | |
| 370 | if (!is_array($from)) { |
| 371 | $from = array($segmentTable, $from); |
| 372 | } else { |
| 373 | array_unshift($from, $segmentTable); |
| 374 | } |
| 375 | |
| 376 | $logTablesProvider = $this->getLogTableProvider(); |
| 377 | $logTablesProvider->setTempTable(new LogTableTemporary($segmentTable)); |
| 378 | |
| 379 | foreach ($logTablesProvider->getAllLogTables() as $logTable) { |
| 380 | if ($logTable->getDateTimeColumn()) { |
| 381 | $whereTest = $this->getWhereStatement($logTable->getName(), $logTable->getDateTimeColumn()); |
| 382 | if (strpos($where, $whereTest) === 0) { |
| 383 | // we don't need to apply the where statement again as it would have been applied already |
| 384 | // in the temporary table... instead it should join the tables through the idvisit index |
| 385 | $where = ltrim(str_replace($whereTest, '', $where)); |
| 386 | if (stripos($where, 'and ') === 0) { |
| 387 | $where = substr($where, strlen('and ')); |
| 388 | } |
| 389 | $bind = array(); |
| 390 | break; |
| 391 | } |
| 392 | } |
| 393 | |
| 394 | } |
| 395 | |
| 396 | } |
| 397 | |
| 398 | $query = $segment->getSelectQuery($select, $from, $where, $bind, $orderBy, $groupBy, $limit, $offset); |
| 399 | |
| 400 | $select = 'SELECT'; |
| 401 | if ($this->queryOriginHint && is_array($query) && 0 === strpos(trim($query['sql']), $select)) { |
| 402 | $query['sql'] = trim($query['sql']); |
| 403 | $query['sql'] = 'SELECT /* ' . $this->queryOriginHint . ' */' . substr($query['sql'], strlen($select)); |
| 404 | } |
| 405 | |
| 406 | if (!$this->getSegment()->isEmpty() && is_array($query) && 0 === strpos(trim($query['sql']), $select)) { |
| 407 | $query['sql'] = trim($query['sql']); |
| 408 | $query['sql'] = 'SELECT /* ' . $this->dateStart->toString() . ',' . $this->dateEnd->toString() . ' sites ' . implode(',', array_map('intval', $this->sites)) . ' segmenthash ' . $this->getSegment()->getHash(). ' */' . substr($query['sql'], strlen($select)); |
| 409 | } |
| 410 | |
| 411 | return $query; |
| 412 | } |
| 413 | |
| 414 | protected function getVisitsMetricFields() |
| 415 | { |
| 416 | return array( |
| 417 | Metrics::INDEX_NB_UNIQ_VISITORS => "count(distinct " . self::LOG_VISIT_TABLE . ".idvisitor)", |
| 418 | Metrics::INDEX_NB_UNIQ_FINGERPRINTS => "count(distinct " . self::LOG_VISIT_TABLE . ".config_id)", |
| 419 | Metrics::INDEX_NB_VISITS => "count(*)", |
| 420 | Metrics::INDEX_NB_ACTIONS => "sum(" . self::LOG_VISIT_TABLE . ".visit_total_actions)", |
| 421 | Metrics::INDEX_MAX_ACTIONS => "max(" . self::LOG_VISIT_TABLE . ".visit_total_actions)", |
| 422 | Metrics::INDEX_SUM_VISIT_LENGTH => "sum(" . self::LOG_VISIT_TABLE . ".visit_total_time)", |
| 423 | Metrics::INDEX_BOUNCE_COUNT => "sum(case " . self::LOG_VISIT_TABLE . ".visit_total_actions when 1 then 1 when 0 then 1 else 0 end)", |
| 424 | Metrics::INDEX_NB_VISITS_CONVERTED => "sum(case " . self::LOG_VISIT_TABLE . ".visit_goal_converted when 1 then 1 else 0 end)", |
| 425 | Metrics::INDEX_NB_USERS => "count(distinct " . self::LOG_VISIT_TABLE . ".user_id)", |
| 426 | ); |
| 427 | } |
| 428 | |
| 429 | public static function getConversionsMetricFields() |
| 430 | { |
| 431 | return array( |
| 432 | Metrics::INDEX_GOAL_NB_CONVERSIONS => "count(*)", |
| 433 | Metrics::INDEX_GOAL_NB_VISITS_CONVERTED => "count(distinct " . self::LOG_CONVERSION_TABLE . ".idvisit)", |
| 434 | Metrics::INDEX_GOAL_REVENUE => self::getSqlConversionRevenueSum(self::TOTAL_REVENUE_FIELD), |
| 435 | Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_SUBTOTAL => self::getSqlConversionRevenueSum(self::REVENUE_SUBTOTAL_FIELD), |
| 436 | Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_TAX => self::getSqlConversionRevenueSum(self::REVENUE_TAX_FIELD), |
| 437 | Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_SHIPPING => self::getSqlConversionRevenueSum(self::REVENUE_SHIPPING_FIELD), |
| 438 | Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_DISCOUNT => self::getSqlConversionRevenueSum(self::REVENUE_DISCOUNT_FIELD), |
| 439 | Metrics::INDEX_GOAL_ECOMMERCE_ITEMS => "SUM(" . self::LOG_CONVERSION_TABLE . "." . self::ITEMS_COUNT_FIELD . ")", |
| 440 | ); |
| 441 | } |
| 442 | |
| 443 | private static function getSqlConversionRevenueSum($field) |
| 444 | { |
| 445 | return self::getSqlRevenue('SUM(' . self::LOG_CONVERSION_TABLE . '.' . $field . ')'); |
| 446 | } |
| 447 | |
| 448 | public static function getSqlRevenue($field) |
| 449 | { |
| 450 | return "ROUND(" . $field . "," . GoalManager::REVENUE_PRECISION . ")"; |
| 451 | } |
| 452 | |
| 453 | /** |
| 454 | * Helper function that returns an array with common metrics for a given log_visit field distinct values. |
| 455 | * |
| 456 | * The statistics returned are: |
| 457 | * - number of unique visitors |
| 458 | * - number of visits |
| 459 | * - number of actions |
| 460 | * - maximum number of action for a visit |
| 461 | * - sum of the visits' length in sec |
| 462 | * - count of bouncing visits (visits with one page view) |
| 463 | * |
| 464 | * For example if $dimension = 'config_os' it will return the statistics for every distinct Operating systems |
| 465 | * The returned array will have a row per distinct operating systems, |
| 466 | * and a column per stat (nb of visits, max actions, etc) |
| 467 | * |
| 468 | * 'label' Metrics::INDEX_NB_UNIQ_VISITORS Metrics::INDEX_NB_VISITS etc. |
| 469 | * Linux 27 66 ... |
| 470 | * Windows XP 12 ... |
| 471 | * Mac OS 15 36 ... |
| 472 | * |
| 473 | * @param string $dimension Table log_visit field name to be use to compute common stats |
| 474 | * @return DataArray |
| 475 | */ |
| 476 | public function getMetricsFromVisitByDimension($dimension) |
| 477 | { |
| 478 | if (!is_array($dimension)) { |
| 479 | $dimension = array($dimension); |
| 480 | } |
| 481 | if (count($dimension) == 1) { |
| 482 | $dimension = array("label" => reset($dimension)); |
| 483 | } |
| 484 | $query = $this->queryVisitsByDimension($dimension); |
| 485 | $metrics = new DataArray(); |
| 486 | while ($row = $query->fetch()) { |
| 487 | $metrics->sumMetricsVisits($row["label"], $row); |
| 488 | } |
| 489 | return $metrics; |
| 490 | } |
| 491 | |
| 492 | /** |
| 493 | * Executes and returns a query aggregating visit logs, optionally grouping by some dimension. Returns |
| 494 | * a DB statement that can be used to iterate over the result |
| 495 | * |
| 496 | * **Result Set** |
| 497 | * |
| 498 | * The following columns are in each row of the result set: |
| 499 | * |
| 500 | * - **{@link \Piwik\Metrics::INDEX_NB_UNIQ_VISITORS}**: The total number of unique visitors in this group |
| 501 | * of aggregated visits. |
| 502 | * - **{@link \Piwik\Metrics::INDEX_NB_VISITS}**: The total number of visits aggregated. |
| 503 | * - **{@link \Piwik\Metrics::INDEX_NB_ACTIONS}**: The total number of actions performed in this group of |
| 504 | * aggregated visits. |
| 505 | * - **{@link \Piwik\Metrics::INDEX_MAX_ACTIONS}**: The maximum actions performed in one visit for this group of |
| 506 | * visits. |
| 507 | * - **{@link \Piwik\Metrics::INDEX_SUM_VISIT_LENGTH}**: The total amount of time spent on the site for this |
| 508 | * group of visits. |
| 509 | * - **{@link \Piwik\Metrics::INDEX_BOUNCE_COUNT}**: The total number of bounced visits in this group of |
| 510 | * visits. |
| 511 | * - **{@link \Piwik\Metrics::INDEX_NB_VISITS_CONVERTED}**: The total number of visits for which at least one |
| 512 | * conversion occurred, for this group of visits. |
| 513 | * |
| 514 | * Additional data can be selected by setting the `$additionalSelects` parameter. |
| 515 | * |
| 516 | * _Note: The metrics returned by this query can be customized by the `$metrics` parameter._ |
| 517 | * |
| 518 | * @param array|string $dimensions `SELECT` fields (or just one field) that will be grouped by, |
| 519 | * eg, `'referrer_name'` or `array('referrer_name', 'referrer_keyword')`. |
| 520 | * The metrics retrieved from the query will be specific to combinations |
| 521 | * of these fields. So if `array('referrer_name', 'referrer_keyword')` |
| 522 | * is supplied, the query will aggregate visits for each referrer/keyword |
| 523 | * combination. |
| 524 | * @param bool|string $where Additional condition for the `WHERE` clause. Can be used to filter |
| 525 | * the set of visits that are considered for aggregation. |
| 526 | * @param array $additionalSelects Additional `SELECT` fields that are not included in the group by |
| 527 | * clause. These can be aggregate expressions, eg, `SUM(somecol)`. |
| 528 | * @param bool|array $metrics The set of metrics to calculate and return. If false, the query will select |
| 529 | * all of them. The following values can be used: |
| 530 | * |
| 531 | * - {@link \Piwik\Metrics::INDEX_NB_UNIQ_VISITORS} |
| 532 | * - {@link \Piwik\Metrics::INDEX_NB_VISITS} |
| 533 | * - {@link \Piwik\Metrics::INDEX_NB_ACTIONS} |
| 534 | * - {@link \Piwik\Metrics::INDEX_MAX_ACTIONS} |
| 535 | * - {@link \Piwik\Metrics::INDEX_SUM_VISIT_LENGTH} |
| 536 | * - {@link \Piwik\Metrics::INDEX_BOUNCE_COUNT} |
| 537 | * - {@link \Piwik\Metrics::INDEX_NB_VISITS_CONVERTED} |
| 538 | * @param bool|\Piwik\RankingQuery $rankingQuery |
| 539 | * A pre-configured ranking query instance that will be used to limit the result. |
| 540 | * If set, the return value is the array returned by {@link \Piwik\RankingQuery::execute()}. |
| 541 | * @param bool|string $orderBy Order By clause to add (e.g. user_id ASC) |
| 542 | * @param int $timeLimitInMs Adds a MAX_EXECUTION_TIME query hint to the query if $timeLimitInMs > 0 |
| 543 | * |
| 544 | * @return mixed A Zend_Db_Statement if `$rankingQuery` isn't supplied, otherwise the result of |
| 545 | * {@link \Piwik\RankingQuery::execute()}. Read {@link queryVisitsByDimension() this} |
| 546 | * to see what aggregate data is calculated by the query. |
| 547 | * @api |
| 548 | */ |
| 549 | public function queryVisitsByDimension(array $dimensions = array(), $where = false, array $additionalSelects = array(), |
| 550 | $metrics = false, $rankingQuery = false, $orderBy = false, $timeLimitInMs = -1) |
| 551 | { |
| 552 | $tableName = self::LOG_VISIT_TABLE; |
| 553 | $availableMetrics = $this->getVisitsMetricFields(); |
| 554 | |
| 555 | $select = $this->getSelectStatement($dimensions, $tableName, $additionalSelects, $availableMetrics, $metrics); |
| 556 | $from = array($tableName); |
| 557 | $where = $this->getWhereStatement($tableName, self::VISIT_DATETIME_FIELD, $where); |
| 558 | $groupBy = $this->getGroupByStatement($dimensions, $tableName); |
| 559 | $orderBys = $orderBy ? [$orderBy] : []; |
| 560 | |
| 561 | if ($rankingQuery) { |
| 562 | $orderBys[] = '`' . Metrics::INDEX_NB_VISITS . '` DESC'; |
| 563 | } |
| 564 | |
| 565 | $query = $this->generateQuery($select, $from, $where, $groupBy, implode(', ', $orderBys)); |
| 566 | |
| 567 | if ($rankingQuery) { |
| 568 | unset($availableMetrics[Metrics::INDEX_MAX_ACTIONS]); |
| 569 | |
| 570 | // INDEX_NB_UNIQ_FINGERPRINTS is only processed if specifically asked for |
| 571 | if (!$this->isMetricRequested(Metrics::INDEX_NB_UNIQ_FINGERPRINTS, $metrics)) { |
| 572 | unset($availableMetrics[Metrics::INDEX_NB_UNIQ_FINGERPRINTS]); |
| 573 | } |
| 574 | |
| 575 | $sumColumns = array_keys($availableMetrics); |
| 576 | |
| 577 | if ($metrics) { |
| 578 | $sumColumns = array_intersect($sumColumns, $metrics); |
| 579 | } |
| 580 | |
| 581 | $rankingQuery->addColumn($sumColumns, 'sum'); |
| 582 | if ($this->isMetricRequested(Metrics::INDEX_MAX_ACTIONS, $metrics)) { |
| 583 | $rankingQuery->addColumn(Metrics::INDEX_MAX_ACTIONS, 'max'); |
| 584 | } |
| 585 | |
| 586 | return $rankingQuery->execute($query['sql'], $query['bind'], $timeLimitInMs); |
| 587 | } |
| 588 | |
| 589 | $query['sql'] = DbHelper::addMaxExecutionTimeHintToQuery($query['sql'], $timeLimitInMs); |
| 590 | |
| 591 | return $this->getDb()->query($query['sql'], $query['bind']); |
| 592 | } |
| 593 | |
| 594 | protected function getSelectsMetrics($metricsAvailable, $metricsRequested = false) |
| 595 | { |
| 596 | $selects = array(); |
| 597 | |
| 598 | foreach ($metricsAvailable as $metricId => $statement) { |
| 599 | if ($this->isMetricRequested($metricId, $metricsRequested)) { |
| 600 | $aliasAs = $this->getSelectAliasAs($metricId); |
| 601 | $selects[] = $statement . $aliasAs; |
| 602 | } |
| 603 | } |
| 604 | |
| 605 | return $selects; |
| 606 | } |
| 607 | |
| 608 | protected function getSelectStatement($dimensions, $tableName, $additionalSelects, array $availableMetrics, $requestedMetrics = false) |
| 609 | { |
| 610 | $dimensionsToSelect = $this->getDimensionsToSelect($dimensions, $additionalSelects); |
| 611 | |
| 612 | $selects = array_merge( |
| 613 | $this->getSelectDimensions($dimensionsToSelect, $tableName), |
| 614 | $this->getSelectsMetrics($availableMetrics, $requestedMetrics), |
| 615 | !empty($additionalSelects) ? $additionalSelects : array() |
| 616 | ); |
| 617 | |
| 618 | $select = implode(self::FIELDS_SEPARATOR, $selects); |
| 619 | return $select; |
| 620 | } |
| 621 | |
| 622 | /** |
| 623 | * Will return the subset of $dimensions that are not found in $additionalSelects |
| 624 | * |
| 625 | * @param $dimensions |
| 626 | * @param array $additionalSelects |
| 627 | * @return array |
| 628 | */ |
| 629 | protected function getDimensionsToSelect($dimensions, $additionalSelects) |
| 630 | { |
| 631 | if (empty($additionalSelects)) { |
| 632 | return $dimensions; |
| 633 | } |
| 634 | |
| 635 | $dimensionsToSelect = array(); |
| 636 | foreach ($dimensions as $selectAs => $dimension) { |
| 637 | $asAlias = $this->getSelectAliasAs($dimension); |
| 638 | foreach ($additionalSelects as $additionalSelect) { |
| 639 | if (strpos($additionalSelect, $asAlias) === false) { |
| 640 | $dimensionsToSelect[$selectAs] = $dimension; |
| 641 | } |
| 642 | } |
| 643 | } |
| 644 | |
| 645 | $dimensionsToSelect = array_unique($dimensionsToSelect); |
| 646 | return $dimensionsToSelect; |
| 647 | } |
| 648 | |
| 649 | /** |
| 650 | * Returns the dimensions array, where |
| 651 | * (1) the table name is prepended to the field |
| 652 | * (2) the "AS `label` " is appended to the field |
| 653 | * |
| 654 | * @param $dimensions |
| 655 | * @param $tableName |
| 656 | * @param bool $appendSelectAs |
| 657 | * @param bool $parseSelectAs |
| 658 | * @return mixed |
| 659 | */ |
| 660 | protected function getSelectDimensions($dimensions, $tableName, $appendSelectAs = true) |
| 661 | { |
| 662 | foreach ($dimensions as $selectAs => &$field) { |
| 663 | $selectAsString = $field; |
| 664 | |
| 665 | if (!is_numeric($selectAs)) { |
| 666 | $selectAsString = $selectAs; |
| 667 | } else if ($this->isFieldFunctionOrComplexExpression($field)) { |
| 668 | // if complex expression has a select as, use it |
| 669 | if (!$appendSelectAs && preg_match('/\s+AS\s+(.*?)\s*$/', $field, $matches)) { |
| 670 | $field = $matches[1]; |
| 671 | continue; |
| 672 | } |
| 673 | |
| 674 | // if function w/o select as, do not alias or prefix |
| 675 | $selectAsString = $appendSelectAs = false; |
| 676 | } |
| 677 | |
| 678 | $isKnownField = !in_array($field, array('referrer_data')); |
| 679 | |
| 680 | if ($selectAsString == $field && $isKnownField) { |
| 681 | $field = $this->prefixColumn($field, $tableName); |
| 682 | } |
| 683 | |
| 684 | if ($appendSelectAs && $selectAsString) { |
| 685 | $field = $this->prefixColumn($field, $tableName) . $this->getSelectAliasAs($selectAsString); |
| 686 | } |
| 687 | } |
| 688 | |
| 689 | return $dimensions; |
| 690 | } |
| 691 | |
| 692 | /** |
| 693 | * Prefixes a column name with a table name if not already done. |
| 694 | * |
| 695 | * @param string $column eg, 'location_provider' |
| 696 | * @param string $tableName eg, 'log_visit' |
| 697 | * @return string eg, 'log_visit.location_provider' |
| 698 | */ |
| 699 | private function prefixColumn($column, $tableName) |
| 700 | { |
| 701 | if (strpos($column, '.') === false) { |
| 702 | return $tableName . '.' . $column; |
| 703 | } else { |
| 704 | return $column; |
| 705 | } |
| 706 | } |
| 707 | |
| 708 | protected function isFieldFunctionOrComplexExpression($field) |
| 709 | { |
| 710 | return strpos($field, "(") !== false |
| 711 | || strpos($field, "CASE") !== false; |
| 712 | } |
| 713 | |
| 714 | protected function getSelectAliasAs($metricId) |
| 715 | { |
| 716 | return " AS `" . $metricId . "`"; |
| 717 | } |
| 718 | |
| 719 | protected function isMetricRequested($metricId, $metricsRequested) |
| 720 | { |
| 721 | // do not process INDEX_NB_UNIQ_FINGERPRINTS unless specifically asked for |
| 722 | if ($metricsRequested === false) { |
| 723 | if ($metricId == Metrics::INDEX_NB_UNIQ_FINGERPRINTS) { |
| 724 | return false; |
| 725 | } |
| 726 | return true; |
| 727 | } |
| 728 | return in_array($metricId, $metricsRequested); |
| 729 | } |
| 730 | |
| 731 | public function getWhereStatement($tableName, $datetimeField, $extraWhere = false) |
| 732 | { |
| 733 | $where = "$tableName.$datetimeField >= ? |
| 734 | AND $tableName.$datetimeField <= ? |
| 735 | AND $tableName.idsite IN (". Common::getSqlStringFieldsArray($this->sites) . ")"; |
| 736 | |
| 737 | if (!empty($extraWhere)) { |
| 738 | $extraWhere = sprintf($extraWhere, $tableName, $tableName); |
| 739 | $where .= ' AND ' . $extraWhere; |
| 740 | } |
| 741 | |
| 742 | return $where; |
| 743 | } |
| 744 | |
| 745 | protected function getGroupByStatement($dimensions, $tableName) |
| 746 | { |
| 747 | $dimensions = $this->getSelectDimensions($dimensions, $tableName, $appendSelectAs = false); |
| 748 | $groupBy = implode(", ", $dimensions); |
| 749 | |
| 750 | return $groupBy; |
| 751 | } |
| 752 | |
| 753 | /** |
| 754 | * Returns general bind parameters for all log aggregation queries. This includes the datetime |
| 755 | * start of entities, datetime end of entities and IDs of all sites. |
| 756 | * |
| 757 | * @return array |
| 758 | */ |
| 759 | public function getGeneralQueryBindParams() |
| 760 | { |
| 761 | $bind = array($this->dateStart->toString(Date::DATE_TIME_FORMAT), $this->dateEnd->toString(Date::DATE_TIME_FORMAT)); |
| 762 | $bind = array_merge($bind, $this->sites); |
| 763 | |
| 764 | return $bind; |
| 765 | } |
| 766 | |
| 767 | /** |
| 768 | * Executes and returns a query aggregating ecommerce item data (everything stored in the |
| 769 | * **log\_conversion\_item** table) and returns a DB statement that can be used to iterate over the result |
| 770 | * |
| 771 | * <a name="queryEcommerceItems-result-set"></a> |
| 772 | * **Result Set** |
| 773 | * |
| 774 | * Each row of the result set represents an aggregated group of ecommerce items. The following |
| 775 | * columns are in each row of the result set: |
| 776 | * |
| 777 | * - **{@link Piwik\Metrics::INDEX_ECOMMERCE_ITEM_REVENUE}**: The total revenue for the group of items. |
| 778 | * - **{@link Piwik\Metrics::INDEX_ECOMMERCE_ITEM_QUANTITY}**: The total number of items in this group. |
| 779 | * - **{@link Piwik\Metrics::INDEX_ECOMMERCE_ITEM_PRICE}**: The total price for the group of items. |
| 780 | * - **{@link Piwik\Metrics::INDEX_ECOMMERCE_ORDERS}**: The total number of orders this group of items |
| 781 | * belongs to. This will be <= to the total number |
| 782 | * of items in this group. |
| 783 | * - **{@link Piwik\Metrics::INDEX_NB_VISITS}**: The total number of visits that caused these items to be logged. |
| 784 | * - **ecommerceType**: Either {@link Piwik\Tracker\GoalManager::IDGOAL_CART} if the items in this group were |
| 785 | * abandoned by a visitor, or {@link Piwik\Tracker\GoalManager::IDGOAL_ORDER} if they |
| 786 | * were ordered by a visitor. |
| 787 | * |
| 788 | * **Limitations** |
| 789 | * |
| 790 | * Segmentation is not yet supported for this aggregation method. |
| 791 | * |
| 792 | * @param string $dimension One or more **log\_conversion\_item** columns to group aggregated data by. |
| 793 | * Eg, `'idaction_sku'` or `'idaction_sku, idaction_category'`. |
| 794 | * @return \Zend_Db_Statement A statement object that can be used to iterate through the query's |
| 795 | * result set. See [above](#queryEcommerceItems-result-set) to learn more |
| 796 | * about what this query selects. |
| 797 | * @api |
| 798 | */ |
| 799 | public function queryEcommerceItems($dimension) |
| 800 | { |
| 801 | $query = $this->generateQuery( |
| 802 | // SELECT ... |
| 803 | implode( |
| 804 | ', ', |
| 805 | array( |
| 806 | "log_action.name AS label", |
| 807 | sprintf("log_conversion_item.%s AS labelIdAction", $dimension), |
| 808 | sprintf( |
| 809 | '%s AS `%d`', |
| 810 | self::getSqlRevenue('SUM(log_conversion_item.quantity * log_conversion_item.price)'), |
| 811 | Metrics::INDEX_ECOMMERCE_ITEM_REVENUE |
| 812 | ), |
| 813 | sprintf( |
| 814 | '%s AS `%d`', |
| 815 | self::getSqlRevenue('SUM(log_conversion_item.quantity)'), |
| 816 | Metrics::INDEX_ECOMMERCE_ITEM_QUANTITY |
| 817 | ), |
| 818 | sprintf( |
| 819 | '%s AS `%d`', |
| 820 | self::getSqlRevenue('SUM(log_conversion_item.price)'), |
| 821 | Metrics::INDEX_ECOMMERCE_ITEM_PRICE |
| 822 | ), |
| 823 | sprintf( |
| 824 | 'COUNT(distinct log_conversion_item.idorder) AS `%d`', |
| 825 | Metrics::INDEX_ECOMMERCE_ORDERS |
| 826 | ), |
| 827 | sprintf( |
| 828 | 'COUNT(distinct log_conversion_item.idvisit) AS `%d`', |
| 829 | Metrics::INDEX_NB_VISITS |
| 830 | ), |
| 831 | sprintf( |
| 832 | 'CASE log_conversion_item.idorder WHEN \'0\' THEN %d ELSE %d END AS ecommerceType', |
| 833 | GoalManager::IDGOAL_CART, |
| 834 | GoalManager::IDGOAL_ORDER |
| 835 | ) |
| 836 | ) |
| 837 | ), |
| 838 | |
| 839 | // FROM ... |
| 840 | array( |
| 841 | "log_conversion_item", |
| 842 | array( |
| 843 | "table" => "log_action", |
| 844 | "joinOn" => sprintf("log_conversion_item.%s = log_action.idaction", $dimension) |
| 845 | ) |
| 846 | ), |
| 847 | |
| 848 | // WHERE ... AND ... |
| 849 | implode( |
| 850 | ' AND ', |
| 851 | array( |
| 852 | 'log_conversion_item.server_time >= ?', |
| 853 | 'log_conversion_item.server_time <= ?', |
| 854 | 'log_conversion_item.idsite IN (' . Common::getSqlStringFieldsArray($this->sites) . ')', |
| 855 | 'log_conversion_item.deleted = 0' |
| 856 | ) |
| 857 | ), |
| 858 | |
| 859 | // GROUP BY ... |
| 860 | sprintf( |
| 861 | "ecommerceType, log_conversion_item.%s", |
| 862 | $dimension |
| 863 | ), |
| 864 | |
| 865 | // ORDER ... |
| 866 | false |
| 867 | ); |
| 868 | |
| 869 | return $this->getDb()->query($query['sql'], $query['bind']); |
| 870 | } |
| 871 | |
| 872 | /** |
| 873 | * Executes and returns a query aggregating action data (everything in the log_action table) and returns |
| 874 | * a DB statement that can be used to iterate over the result |
| 875 | * |
| 876 | * <a name="queryActionsByDimension-result-set"></a> |
| 877 | * **Result Set** |
| 878 | * |
| 879 | * Each row of the result set represents an aggregated group of actions. The following columns |
| 880 | * are in each aggregate row: |
| 881 | * |
| 882 | * - **{@link Piwik\Metrics::INDEX_NB_UNIQ_VISITORS}**: The total number of unique visitors that performed |
| 883 | * the actions in this group. |
| 884 | * - **{@link Piwik\Metrics::INDEX_NB_VISITS}**: The total number of visits these actions belong to. |
| 885 | * - **{@link Piwik\Metrics::INDEX_NB_ACTIONS}**: The total number of actions in this aggregate group. |
| 886 | * |
| 887 | * Additional data can be selected through the `$additionalSelects` parameter. |
| 888 | * |
| 889 | * _Note: The metrics calculated by this query can be customized by the `$metrics` parameter._ |
| 890 | * |
| 891 | * @param array|string $dimensions One or more SELECT fields that will be used to group the log_action |
| 892 | * rows by. This parameter determines which log_action rows will be |
| 893 | * aggregated together. |
| 894 | * @param bool|string $where Additional condition for the WHERE clause. Can be used to filter |
| 895 | * the set of visits that are considered for aggregation. |
| 896 | * @param array $additionalSelects Additional SELECT fields that are not included in the group by |
| 897 | * clause. These can be aggregate expressions, eg, `SUM(somecol)`. |
| 898 | * @param bool|array $metrics The set of metrics to calculate and return. If `false`, the query will select |
| 899 | * all of them. The following values can be used: |
| 900 | * |
| 901 | * - {@link Piwik\Metrics::INDEX_NB_UNIQ_VISITORS} |
| 902 | * - {@link Piwik\Metrics::INDEX_NB_VISITS} |
| 903 | * - {@link Piwik\Metrics::INDEX_NB_ACTIONS} |
| 904 | * @param bool|\Piwik\RankingQuery $rankingQuery |
| 905 | * A pre-configured ranking query instance that will be used to limit the result. |
| 906 | * If set, the return value is the array returned by {@link Piwik\RankingQuery::execute()}. |
| 907 | * @param bool|string $joinLogActionOnColumn One or more columns from the **log_link_visit_action** table that |
| 908 | * log_action should be joined on. The table alias used for each join |
| 909 | * is `"log_action$i"` where `$i` is the index of the column in this |
| 910 | * array. |
| 911 | * |
| 912 | * If a string is used for this parameter, the table alias is not |
| 913 | * suffixed (since there is only one column). |
| 914 | * @param string $secondaryOrderBy A secondary order by clause for the ranking query |
| 915 | * @param int $timeLimitInMs Adds a MAX_EXECUTION_TIME hint to the query if $timeLimitInMs > 0 |
| 916 | * @return mixed A Zend_Db_Statement if `$rankingQuery` isn't supplied, otherwise the result of |
| 917 | * {@link Piwik\RankingQuery::execute()}. Read [this](#queryEcommerceItems-result-set) |
| 918 | * to see what aggregate data is calculated by the query. |
| 919 | * @api |
| 920 | */ |
| 921 | public function queryActionsByDimension( |
| 922 | $dimensions, |
| 923 | $where = '', |
| 924 | $additionalSelects = array(), |
| 925 | $metrics = false, |
| 926 | $rankingQuery = null, |
| 927 | $joinLogActionOnColumn = false, |
| 928 | $secondaryOrderBy = null, |
| 929 | $timeLimitInMs = -1 |
| 930 | ) { |
| 931 | $tableName = self::LOG_ACTIONS_TABLE; |
| 932 | $availableMetrics = $this->getActionsMetricFields(); |
| 933 | |
| 934 | $select = $this->getSelectStatement($dimensions, $tableName, $additionalSelects, $availableMetrics, $metrics); |
| 935 | $from = array($tableName); |
| 936 | $where = $this->getWhereStatement($tableName, self::ACTION_DATETIME_FIELD, $where); |
| 937 | $groupBy = $this->getGroupByStatement($dimensions, $tableName); |
| 938 | |
| 939 | if ($joinLogActionOnColumn !== false) { |
| 940 | $multiJoin = is_array($joinLogActionOnColumn); |
| 941 | if (!$multiJoin) { |
| 942 | $joinLogActionOnColumn = array($joinLogActionOnColumn); |
| 943 | } |
| 944 | |
| 945 | foreach ($joinLogActionOnColumn as $i => $joinColumn) { |
| 946 | $tableAlias = 'log_action' . ($multiJoin ? $i + 1 : ''); |
| 947 | |
| 948 | if (strpos($joinColumn, ' ') === false) { |
| 949 | $joinOn = $tableAlias . '.idaction = ' . $tableName . '.' . $joinColumn; |
| 950 | } else { |
| 951 | // more complex join column like if (...) |
| 952 | $joinOn = $tableAlias . '.idaction = ' . $joinColumn; |
| 953 | } |
| 954 | |
| 955 | $from[] = array( |
| 956 | 'table' => 'log_action', |
| 957 | 'tableAlias' => $tableAlias, |
| 958 | 'joinOn' => $joinOn |
| 959 | ); |
| 960 | } |
| 961 | } |
| 962 | |
| 963 | $orderBy = false; |
| 964 | if ($rankingQuery) { |
| 965 | $orderBy = '`' . Metrics::INDEX_NB_ACTIONS . '` DESC'; |
| 966 | if ($secondaryOrderBy) { |
| 967 | $orderBy .= ', ' . $secondaryOrderBy; |
| 968 | } |
| 969 | } |
| 970 | |
| 971 | $query = $this->generateQuery($select, $from, $where, $groupBy, $orderBy); |
| 972 | |
| 973 | if ($rankingQuery !== null) { |
| 974 | $sumColumns = array_keys($availableMetrics); |
| 975 | if ($metrics) { |
| 976 | $sumColumns = array_intersect($sumColumns, $metrics); |
| 977 | } |
| 978 | |
| 979 | $rankingQuery->addColumn($sumColumns, 'sum'); |
| 980 | |
| 981 | return $rankingQuery->execute($query['sql'], $query['bind'], $timeLimitInMs); |
| 982 | } |
| 983 | |
| 984 | $query['sql'] = DbHelper::addMaxExecutionTimeHintToQuery($query['sql'], $timeLimitInMs); |
| 985 | |
| 986 | return $this->getDb()->query($query['sql'], $query['bind']); |
| 987 | } |
| 988 | |
| 989 | protected function getActionsMetricFields() |
| 990 | { |
| 991 | return array( |
| 992 | Metrics::INDEX_NB_VISITS => "count(distinct " . self::LOG_ACTIONS_TABLE . ".idvisit)", |
| 993 | Metrics::INDEX_NB_UNIQ_VISITORS => "count(distinct " . self::LOG_ACTIONS_TABLE . ".idvisitor)", |
| 994 | Metrics::INDEX_NB_ACTIONS => "count(*)", |
| 995 | ); |
| 996 | } |
| 997 | |
| 998 | /** |
| 999 | * Executes a query aggregating conversion data (everything in the **log_conversion** table) and returns |
| 1000 | * a DB statement that can be used to iterate over the result. |
| 1001 | * |
| 1002 | * <a name="queryConversionsByDimension-result-set"></a> |
| 1003 | * **Result Set** |
| 1004 | * |
| 1005 | * Each row of the result set represents an aggregated group of conversions. The |
| 1006 | * following columns are in each aggregate row: |
| 1007 | * |
| 1008 | * - **{@link Piwik\Metrics::INDEX_GOAL_NB_CONVERSIONS}**: The total number of conversions in this aggregate |
| 1009 | * group. |
| 1010 | * - **{@link Piwik\Metrics::INDEX_GOAL_NB_VISITS_CONVERTED}**: The total number of visits during which these |
| 1011 | * conversions were converted. |
| 1012 | * - **{@link Piwik\Metrics::INDEX_GOAL_REVENUE}**: The total revenue generated by these conversions. This value |
| 1013 | * includes the revenue from individual ecommerce items. |
| 1014 | * - **{@link Piwik\Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_SUBTOTAL}**: The total cost of all ecommerce items sold |
| 1015 | * within these conversions. This value does not |
| 1016 | * include tax, shipping or any applied discount. |
| 1017 | * |
| 1018 | * _This metric is only applicable to the special |
| 1019 | * **ecommerce** goal (where `idGoal == 'ecommerceOrder'`)._ |
| 1020 | * - **{@link Piwik\Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_TAX}**: The total tax applied to every transaction in these |
| 1021 | * conversions. |
| 1022 | * |
| 1023 | * _This metric is only applicable to the special |
| 1024 | * **ecommerce** goal (where `idGoal == 'ecommerceOrder'`)._ |
| 1025 | * - **{@link Piwik\Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_SHIPPING}**: The total shipping cost for every transaction |
| 1026 | * in these conversions. |
| 1027 | * |
| 1028 | * _This metric is only applicable to the special |
| 1029 | * **ecommerce** goal (where `idGoal == 'ecommerceOrder'`)._ |
| 1030 | * - **{@link Piwik\Metrics::INDEX_GOAL_ECOMMERCE_REVENUE_DISCOUNT}**: The total discount applied to every transaction |
| 1031 | * in these conversions. |
| 1032 | * |
| 1033 | * _This metric is only applicable to the special |
| 1034 | * **ecommerce** goal (where `idGoal == 'ecommerceOrder'`)._ |
| 1035 | * - **{@link Piwik\Metrics::INDEX_GOAL_ECOMMERCE_ITEMS}**: The total number of ecommerce items sold in each transaction |
| 1036 | * in these conversions. |
| 1037 | * |
| 1038 | * _This metric is only applicable to the special |
| 1039 | * **ecommerce** goal (where `idGoal == 'ecommerceOrder'`)._ |
| 1040 | * |
| 1041 | * Additional data can be selected through the `$additionalSelects` parameter. |
| 1042 | * |
| 1043 | * _Note: This method will only query the **log_conversion** table. Other tables cannot be joined |
| 1044 | * using this method._ |
| 1045 | * |
| 1046 | * @param array|string $dimensions One or more **SELECT** fields that will be used to group the log_conversion |
| 1047 | * rows by. This parameter determines which **log_conversion** rows will be |
| 1048 | * aggregated together. |
| 1049 | * @param bool|string $where An optional SQL expression used in the SQL's **WHERE** clause. |
| 1050 | * @param array $additionalSelects Additional SELECT fields that are not included in the group by |
| 1051 | * clause. These can be aggregate expressions, eg, `SUM(somecol)`. |
| 1052 | * @return \Zend_Db_Statement |
| 1053 | */ |
| 1054 | public function queryConversionsByDimension($dimensions = array(), $where = false, $additionalSelects = array(), $extraFrom = []) |
| 1055 | { |
| 1056 | $dimensions = array_merge(array(self::IDGOAL_FIELD), $dimensions); |
| 1057 | $tableName = self::LOG_CONVERSION_TABLE; |
| 1058 | $availableMetrics = $this->getConversionsMetricFields(); |
| 1059 | |
| 1060 | $select = $this->getSelectStatement($dimensions, $tableName, $additionalSelects, $availableMetrics); |
| 1061 | |
| 1062 | $from = array_merge([$tableName], $extraFrom); |
| 1063 | $where = $this->getWhereStatement($tableName, self::CONVERSION_DATETIME_FIELD, $where); |
| 1064 | $groupBy = $this->getGroupByStatement($dimensions, $tableName); |
| 1065 | $orderBy = false; |
| 1066 | $query = $this->generateQuery($select, $from, $where, $groupBy, $orderBy); |
| 1067 | |
| 1068 | return $this->getDb()->query($query['sql'], $query['bind']); |
| 1069 | } |
| 1070 | |
| 1071 | /** |
| 1072 | * Creates and returns an array of SQL `SELECT` expressions that will each count how |
| 1073 | * many rows have a column whose value is within a certain range. |
| 1074 | * |
| 1075 | * **Note:** The result of this function is meant for use in the `$additionalSelects` parameter |
| 1076 | * in one of the query... methods (for example {@link queryVisitsByDimension()}). |
| 1077 | * |
| 1078 | * **Example** |
| 1079 | * |
| 1080 | * // summarize one column |
| 1081 | * $visitTotalActionsRanges = array( |
| 1082 | * array(1, 1), |
| 1083 | * array(2, 10), |
| 1084 | * array(10) |
| 1085 | * ); |
| 1086 | * $selects = LogAggregator::getSelectsFromRangedColumn('visit_total_actions', $visitTotalActionsRanges, 'log_visit', 'vta'); |
| 1087 | * |
| 1088 | * // summarize another column in the same request |
| 1089 | * $visitCountVisitsRanges = array( |
| 1090 | * array(1, 1), |
| 1091 | * array(2, 20), |
| 1092 | * array(20) |
| 1093 | * ); |
| 1094 | * $selects = array_merge( |
| 1095 | * $selects, |
| 1096 | * LogAggregator::getSelectsFromRangedColumn('visitor_count_visits', $visitCountVisitsRanges, 'log_visit', 'vcv') |
| 1097 | * ); |
| 1098 | * |
| 1099 | * // perform the query |
| 1100 | * $logAggregator = // get the LogAggregator somehow |
| 1101 | * $query = $logAggregator->queryVisitsByDimension($dimensions = array(), $where = false, $selects); |
| 1102 | * $tableSummary = $query->fetch(); |
| 1103 | * |
| 1104 | * $numberOfVisitsWithOneAction = $tableSummary['vta0']; |
| 1105 | * $numberOfVisitsBetweenTwoAnd10 = $tableSummary['vta1']; |
| 1106 | * |
| 1107 | * $numberOfVisitsWithVisitCountOfOne = $tableSummary['vcv0']; |
| 1108 | * |
| 1109 | * @param string $column The name of a column in `$table` that will be summarized. |
| 1110 | * @param array $ranges The array of ranges over which the data in the table |
| 1111 | * will be summarized. For example, |
| 1112 | * ``` |
| 1113 | * array( |
| 1114 | * array(1, 1), |
| 1115 | * array(2, 2), |
| 1116 | * array(3, 8), |
| 1117 | * array(8) // everything over 8 |
| 1118 | * ) |
| 1119 | * ``` |
| 1120 | * @param string $table The unprefixed name of the table whose rows will be summarized. |
| 1121 | * @param string $selectColumnPrefix The prefix to prepend to each SELECT expression. This |
| 1122 | * prefix is used to differentiate different sets of |
| 1123 | * range summarization SELECTs. You can supply different |
| 1124 | * values to this argument to summarize several columns |
| 1125 | * in one query (see above for an example). |
| 1126 | * @param bool $restrictToReturningVisitors Whether to only summarize rows that belong to |
| 1127 | * visits of returning visitors or not. If this |
| 1128 | * argument is true, then the SELECT expressions |
| 1129 | * returned can only be used with the |
| 1130 | * {@link queryVisitsByDimension()} method. |
| 1131 | * @return array An array of SQL SELECT expressions, for example, |
| 1132 | * ``` |
| 1133 | * array( |
| 1134 | * 'sum(case when log_visit.visit_total_actions between 0 and 2 then 1 else 0 end) as vta0', |
| 1135 | * 'sum(case when log_visit.visit_total_actions > 2 then 1 else 0 end) as vta1' |
| 1136 | * ) |
| 1137 | * ``` |
| 1138 | * @api |
| 1139 | */ |
| 1140 | public static function getSelectsFromRangedColumn($column, $ranges, $table, $selectColumnPrefix, $restrictToReturningVisitors = false) |
| 1141 | { |
| 1142 | $selects = array(); |
| 1143 | $extraCondition = ''; |
| 1144 | |
| 1145 | $tableColumn = $column; |
| 1146 | if (strpos($tableColumn, $table) === false) { |
| 1147 | $tableColumn = "$table.$column"; |
| 1148 | } |
| 1149 | |
| 1150 | if ($restrictToReturningVisitors) { |
| 1151 | // extra condition for the SQL SELECT that makes sure only returning visits are counted |
| 1152 | // when creating the 'days since last visit' report |
| 1153 | $extraCondition = 'and log_visit.visitor_returning = 1'; |
| 1154 | $extraSelect = "sum(case when log_visit.visitor_returning = 0 then 1 else 0 end) " |
| 1155 | . " as `" . $selectColumnPrefix . 'General_NewVisits' . "`"; |
| 1156 | $selects[] = $extraSelect; |
| 1157 | } |
| 1158 | |
| 1159 | foreach ($ranges as $gap) { |
| 1160 | if (count($gap) == 2) { |
| 1161 | $lowerBound = $gap[0]; |
| 1162 | $upperBound = $gap[1]; |
| 1163 | |
| 1164 | $selectAs = "$selectColumnPrefix$lowerBound-$upperBound"; |
| 1165 | |
| 1166 | $selects[] = "sum(case when $tableColumn between $lowerBound and $upperBound $extraCondition" . |
| 1167 | " then 1 else 0 end) as `$selectAs`"; |
| 1168 | } else { |
| 1169 | $lowerBound = $gap[0]; |
| 1170 | |
| 1171 | $selectAs = $selectColumnPrefix . ($lowerBound + 1) . urlencode('+'); |
| 1172 | $selects[] = "sum(case when $tableColumn > $lowerBound $extraCondition then 1 else 0 end) as `$selectAs`"; |
| 1173 | } |
| 1174 | } |
| 1175 | |
| 1176 | return $selects; |
| 1177 | } |
| 1178 | |
| 1179 | /** |
| 1180 | * Clean up the row data and return values. |
| 1181 | * $lookForThisPrefix can be used to make sure only SOME of the data in $row is used. |
| 1182 | * |
| 1183 | * The array will have one column $columnName |
| 1184 | * |
| 1185 | * @param $row |
| 1186 | * @param $columnName |
| 1187 | * @param bool $lookForThisPrefix A string that identifies which elements of $row to use |
| 1188 | * in the result. Every key of $row that starts with this |
| 1189 | * value is used. |
| 1190 | * @return array |
| 1191 | */ |
| 1192 | public static function makeArrayOneColumn($row, $columnName, $lookForThisPrefix = false) |
| 1193 | { |
| 1194 | $cleanRow = array(); |
| 1195 | |
| 1196 | foreach ($row as $label => $count) { |
| 1197 | if (empty($lookForThisPrefix) |
| 1198 | || strpos($label, $lookForThisPrefix) === 0 |
| 1199 | ) { |
| 1200 | $cleanLabel = substr($label, strlen($lookForThisPrefix)); |
| 1201 | $cleanRow[$cleanLabel] = array($columnName => $count); |
| 1202 | } |
| 1203 | } |
| 1204 | |
| 1205 | return $cleanRow; |
| 1206 | } |
| 1207 | |
| 1208 | public function getDb() |
| 1209 | { |
| 1210 | /** @var ArchivingStatus $archivingStatus */ |
| 1211 | $archivingStatus = StaticContainer::get(ArchivingStatus::class); |
| 1212 | $archivingLock = $archivingStatus->getCurrentArchivingLock(); |
| 1213 | return new ArchivingDbAdapter(Db::getReader(), $archivingLock, $this->logger); |
| 1214 | } |
| 1215 | } |
| 1216 |