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
RawLogDao.php
456 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\Common; |
| 12 | use Piwik\Config as PiwikConfig; |
| 13 | use Piwik\Container\StaticContainer; |
| 14 | use Piwik\Db; |
| 15 | use Piwik\Plugin\Dimension\DimensionMetadataProvider; |
| 16 | use Piwik\Plugin\LogTablesProvider; |
| 17 | |
| 18 | /** |
| 19 | * DAO that queries log tables. |
| 20 | */ |
| 21 | class RawLogDao |
| 22 | { |
| 23 | const DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME = 'tmp_log_actions_to_keep'; |
| 24 | |
| 25 | /** |
| 26 | * @var DimensionMetadataProvider |
| 27 | */ |
| 28 | private $dimensionMetadataProvider; |
| 29 | |
| 30 | /** |
| 31 | * @var LogTablesProvider |
| 32 | */ |
| 33 | private $logTablesProvider; |
| 34 | |
| 35 | public function __construct(DimensionMetadataProvider $provider = null, LogTablesProvider $logTablesProvider = null) |
| 36 | { |
| 37 | $this->dimensionMetadataProvider = $provider ?: StaticContainer::get('Piwik\Plugin\Dimension\DimensionMetadataProvider'); |
| 38 | $this->logTablesProvider = $logTablesProvider ?: StaticContainer::get('Piwik\Plugin\LogTablesProvider'); |
| 39 | } |
| 40 | |
| 41 | /** |
| 42 | * @param array $values |
| 43 | * @param string $idVisit |
| 44 | */ |
| 45 | public function updateVisits(array $values, $idVisit) |
| 46 | { |
| 47 | $sql = "UPDATE " . Common::prefixTable('log_visit') |
| 48 | . " SET " . $this->getColumnSetExpressions(array_keys($values)) |
| 49 | . " WHERE idvisit = ?"; |
| 50 | |
| 51 | $this->update($sql, $values, $idVisit); |
| 52 | } |
| 53 | |
| 54 | /** |
| 55 | * @param array $values |
| 56 | * @param string $idVisit |
| 57 | */ |
| 58 | public function updateConversions(array $values, $idVisit) |
| 59 | { |
| 60 | $sql = "UPDATE " . Common::prefixTable('log_conversion') |
| 61 | . " SET " . $this->getColumnSetExpressions(array_keys($values)) |
| 62 | . " WHERE idvisit = ?"; |
| 63 | |
| 64 | $this->update($sql, $values, $idVisit); |
| 65 | } |
| 66 | |
| 67 | /** |
| 68 | * @param string $from |
| 69 | * @param string $to |
| 70 | * @return int |
| 71 | */ |
| 72 | public function countVisitsWithDatesLimit($from, $to) |
| 73 | { |
| 74 | $sql = "SELECT COUNT(*) AS num_rows" |
| 75 | . " FROM " . Common::prefixTable('log_visit') |
| 76 | . " WHERE visit_last_action_time >= ? AND visit_last_action_time < ?"; |
| 77 | |
| 78 | $bind = array($from, $to); |
| 79 | |
| 80 | return (int) Db::fetchOne($sql, $bind); |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * Iterates over logs in a log table in chunks. Parameters to this function are as backend agnostic |
| 85 | * as possible w/o dramatically increasing code complexity. |
| 86 | * |
| 87 | * @param string $logTable The log table name. Unprefixed, eg, `log_visit`. |
| 88 | * @param array[] $conditions An array describing the conditions logs must match in the query. Translates to |
| 89 | * the WHERE part of a SELECT statement. Each element must contain three elements: |
| 90 | * |
| 91 | * * the column name |
| 92 | * * the operator (ie, '=', '<>', '<', etc.) |
| 93 | * * the operand (ie, a value) |
| 94 | * |
| 95 | * The elements are AND-ed together. |
| 96 | * |
| 97 | * Example: |
| 98 | * |
| 99 | * ``` |
| 100 | * array( |
| 101 | * array('visit_first_action_time', '>=', ...), |
| 102 | * array('visit_first_action_time', '<', ...) |
| 103 | * ) |
| 104 | * ``` |
| 105 | * @param int $iterationStep The number of rows to query at a time. |
| 106 | * @param callable $callback The callback that processes each chunk of rows. |
| 107 | * @param string $willDelete Set to true if you will make sure to delete all rows that were fetched. If you are in |
| 108 | * doubt and not sure if to set true or false, use "false". Setting it to true will |
| 109 | * enable an internal performance improvement but it can result in an endless loop if not |
| 110 | * used properly. |
| 111 | */ |
| 112 | public function forAllLogs($logTable, $fields, $conditions, $iterationStep, $callback, $willDelete) |
| 113 | { |
| 114 | $lastId = 0; |
| 115 | |
| 116 | if ($willDelete) { |
| 117 | // we don't want to look at eg idvisit so the query will be mostly index covered as the |
| 118 | // "where idvisit > 0 ... ORDER BY idvisit ASC" will be gone... meaning we don't need to look at a huge range |
| 119 | // of visits... |
| 120 | $idField = null; |
| 121 | $bindFunction = function ($bind, $lastId) { |
| 122 | return $bind; |
| 123 | }; |
| 124 | } else { |
| 125 | // when we are not deleting, we need to ensure to iterate over each visitor step by step... meaning we |
| 126 | // need to remember which visit we have already looked at and which one not. Therefore we need to apply |
| 127 | // "where idvisit > $lastId" in the query and "order by idvisit ASC" |
| 128 | $idField = $this->getIdFieldForLogTable($logTable); |
| 129 | $bindFunction = function ($bind, $lastId) { |
| 130 | return array_merge(array($lastId), $bind); |
| 131 | }; |
| 132 | } |
| 133 | |
| 134 | list($query, $bind) = $this->createLogIterationQuery($logTable, $idField, $fields, $conditions, $iterationStep); |
| 135 | |
| 136 | do { |
| 137 | $rows = Db::fetchAll($query, call_user_func($bindFunction, $bind, $lastId)); |
| 138 | if (!empty($rows)) { |
| 139 | if ($idField) { |
| 140 | $lastId = $rows[count($rows) - 1][$idField]; |
| 141 | } |
| 142 | $callback($rows); |
| 143 | } |
| 144 | } while (count($rows) == $iterationStep); |
| 145 | } |
| 146 | |
| 147 | /** |
| 148 | * Deletes conversions for the supplied visit IDs from log_conversion. This method does not cascade, so |
| 149 | * conversion items will not be deleted. |
| 150 | * |
| 151 | * @param int[] $visitIds |
| 152 | * @return int The number of deleted rows. |
| 153 | */ |
| 154 | public function deleteFromLogTable($tableName, $visitIds) |
| 155 | { |
| 156 | $sql = "DELETE FROM `" . Common::prefixTable($tableName) . "` WHERE idvisit IN " |
| 157 | . $this->getInFieldExpressionWithInts($visitIds); |
| 158 | |
| 159 | $statement = Db::query($sql); |
| 160 | return $statement->rowCount(); |
| 161 | } |
| 162 | |
| 163 | /** |
| 164 | * Deletes conversion items for the supplied visit IDs from log_conversion_item. |
| 165 | * |
| 166 | * @param int[] $visitIds |
| 167 | * @return int The number of deleted rows. |
| 168 | */ |
| 169 | public function deleteConversionItems($visitIds) |
| 170 | { |
| 171 | $sql = "DELETE FROM `" . Common::prefixTable('log_conversion_item') . "` WHERE idvisit IN " |
| 172 | . $this->getInFieldExpressionWithInts($visitIds); |
| 173 | |
| 174 | $statement = Db::query($sql); |
| 175 | return $statement->rowCount(); |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * Deletes all unused entries from the log_action table. This method uses a temporary table to store used |
| 180 | * actions, and then deletes rows from log_action that are not in this temporary table. |
| 181 | * |
| 182 | * Table locking is required to avoid concurrency issues. |
| 183 | * |
| 184 | * @throws \Exception If table locking permission is not granted to the current MySQL user. |
| 185 | */ |
| 186 | public function deleteUnusedLogActions() |
| 187 | { |
| 188 | if (!Db::isLockPrivilegeGranted()) { |
| 189 | throw new \Exception("RawLogDao.deleteUnusedLogActions() requires table locking permission in order to complete without error."); |
| 190 | } |
| 191 | |
| 192 | // get current max ID in log tables w/ idaction references. |
| 193 | $maxIds = $this->getMaxIdsInLogTables(); |
| 194 | |
| 195 | // get max rows to analyze |
| 196 | $max_rows_per_query = PiwikConfig::getInstance()->Deletelogs['delete_logs_unused_actions_max_rows_per_query']; |
| 197 | |
| 198 | $this->createTempTableForStoringUsedActions(); |
| 199 | |
| 200 | // do large insert (inserting everything before maxIds) w/o locking tables... |
| 201 | $this->insertActionsToKeep($maxIds, $deleteOlderThanMax = true, $max_rows_per_query); |
| 202 | |
| 203 | // ... then do small insert w/ locked tables to minimize the amount of time tables are locked. |
| 204 | $this->lockLogTables(); |
| 205 | $this->insertActionsToKeep($maxIds, $deleteOlderThanMax = false, $max_rows_per_query); |
| 206 | |
| 207 | // delete before unlocking tables so there's no chance a new log row that references an |
| 208 | // unused action will be inserted. |
| 209 | $this->deleteUnusedActions(); |
| 210 | |
| 211 | Db::unlockAllTables(); |
| 212 | |
| 213 | $this->dropTempTableForStoringUsedActions(); |
| 214 | } |
| 215 | |
| 216 | /** |
| 217 | * Returns the list of the website IDs that received some visits between the specified timestamp. The |
| 218 | * start date and the end date is included in the time frame. |
| 219 | * |
| 220 | * @param string $fromDateTime |
| 221 | * @param string $toDateTime |
| 222 | * @return bool true if there are visits for this site between the given timeframe, false if not |
| 223 | */ |
| 224 | public function hasSiteVisitsBetweenTimeframe($fromDateTime, $toDateTime, $idSite) |
| 225 | { |
| 226 | $sites = Db::fetchOne("SELECT 1 |
| 227 | FROM " . Common::prefixTable('log_visit') . " |
| 228 | WHERE idsite = ? |
| 229 | AND visit_last_action_time >= ? |
| 230 | AND visit_last_action_time <= ? |
| 231 | LIMIT 1", array($idSite, $fromDateTime, $toDateTime)); |
| 232 | |
| 233 | return (bool) $sites; |
| 234 | } |
| 235 | |
| 236 | /** |
| 237 | * @param array $columnsToSet |
| 238 | * @return string |
| 239 | */ |
| 240 | protected function getColumnSetExpressions(array $columnsToSet) |
| 241 | { |
| 242 | $columnsToSet = array_map( |
| 243 | function ($column) { |
| 244 | return $column . ' = ?'; |
| 245 | }, |
| 246 | $columnsToSet |
| 247 | ); |
| 248 | |
| 249 | return implode(', ', $columnsToSet); |
| 250 | } |
| 251 | |
| 252 | /** |
| 253 | * @param array $values |
| 254 | * @param $idVisit |
| 255 | * @param $sql |
| 256 | * @return \Zend_Db_Statement |
| 257 | * @throws \Exception |
| 258 | */ |
| 259 | protected function update($sql, array $values, $idVisit) |
| 260 | { |
| 261 | return Db::query($sql, array_merge(array_values($values), array($idVisit))); |
| 262 | } |
| 263 | |
| 264 | protected function getIdFieldForLogTable($logTable) |
| 265 | { |
| 266 | $idColumns = $this->getTableIdColumns(); |
| 267 | |
| 268 | if (isset($idColumns[$logTable])) { |
| 269 | return $idColumns[$logTable]; |
| 270 | } |
| 271 | |
| 272 | throw new \InvalidArgumentException("Unknown log table '$logTable'."); |
| 273 | } |
| 274 | |
| 275 | // TODO: instead of creating a log query like this, we should re-use segments. to do this, however, there must be a 1-1 |
| 276 | // mapping for dimensions => segments, and each dimension should automatically have a segment. |
| 277 | private function createLogIterationQuery($logTable, $idField, $fields, $conditions, $iterationStep) |
| 278 | { |
| 279 | $bind = array(); |
| 280 | |
| 281 | $sql = "SELECT " . implode(', ', $fields) . " FROM `" . Common::prefixTable($logTable) . "` WHERE "; |
| 282 | |
| 283 | $parts = array(); |
| 284 | |
| 285 | if ($idField) { |
| 286 | $parts[] = "$idField > ?"; |
| 287 | } |
| 288 | |
| 289 | foreach ($conditions as $condition) { |
| 290 | list($column, $operator, $value) = $condition; |
| 291 | |
| 292 | if (is_array($value)) { |
| 293 | $parts[] = "$column IN (" . Common::getSqlStringFieldsArray($value) . ")"; |
| 294 | |
| 295 | $bind = array_merge($bind, $value); |
| 296 | } else { |
| 297 | $parts[]= "$column $operator ?"; |
| 298 | |
| 299 | $bind[] = $value; |
| 300 | } |
| 301 | } |
| 302 | $sql .= implode(' AND ', $parts); |
| 303 | |
| 304 | if ($idField) { |
| 305 | $sql .= " ORDER BY $idField ASC"; |
| 306 | } |
| 307 | |
| 308 | $sql .= " LIMIT " . (int)$iterationStep; |
| 309 | |
| 310 | return array($sql, $bind); |
| 311 | } |
| 312 | |
| 313 | private function getInFieldExpressionWithInts($idVisits) |
| 314 | { |
| 315 | $sql = "("; |
| 316 | |
| 317 | $isFirst = true; |
| 318 | foreach ($idVisits as $idVisit) { |
| 319 | if ($isFirst) { |
| 320 | $isFirst = false; |
| 321 | } else { |
| 322 | $sql .= ', '; |
| 323 | } |
| 324 | |
| 325 | $sql .= (int)$idVisit; |
| 326 | } |
| 327 | |
| 328 | $sql .= ")"; |
| 329 | |
| 330 | return $sql; |
| 331 | } |
| 332 | |
| 333 | protected function getMaxIdsInLogTables() |
| 334 | { |
| 335 | $idColumns = $this->getTableIdColumns(); |
| 336 | $tables = array_keys($idColumns); |
| 337 | |
| 338 | $result = array(); |
| 339 | foreach ($tables as $table) { |
| 340 | $idCol = $idColumns[$table]; |
| 341 | $result[$table] = Db::fetchOne("SELECT MAX($idCol) FROM " . Common::prefixTable($table)); |
| 342 | } |
| 343 | |
| 344 | return $result; |
| 345 | } |
| 346 | |
| 347 | private function createTempTableForStoringUsedActions() |
| 348 | { |
| 349 | $sql = "CREATE TEMPORARY TABLE " . Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME) . " ( |
| 350 | idaction INT(11), |
| 351 | PRIMARY KEY (idaction) |
| 352 | )"; |
| 353 | Db::query($sql); |
| 354 | } |
| 355 | |
| 356 | private function dropTempTableForStoringUsedActions() |
| 357 | { |
| 358 | $sql = "DROP TABLE " . Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 359 | Db::query($sql); |
| 360 | } |
| 361 | |
| 362 | // protected for testing purposes |
| 363 | protected function insertActionsToKeep($maxIds, $olderThan = true, $insertIntoTempIterationStep = 100000) |
| 364 | { |
| 365 | $tempTableName = Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 366 | |
| 367 | $idColumns = $this->getTableIdColumns(); |
| 368 | foreach ($this->dimensionMetadataProvider->getActionReferenceColumnsByTable() as $table => $columns) { |
| 369 | $idCol = $idColumns[$table]; |
| 370 | // Create select query for requesting ALL needed fields at once |
| 371 | $sql = "SELECT " . implode(',' ,$columns) . " FROM " . Common::prefixTable($table) . " WHERE $idCol >= ? AND $idCol < ?"; |
| 372 | |
| 373 | if ($olderThan) { |
| 374 | // Why start on zero? When running for a couple of months, this will generate about 10000+ queries with zero result. Use the lowest value instead.... saves a LOT of waiting time! |
| 375 | $start = (int) Db::fetchOne("SELECT MIN($idCol) FROM " . Common::prefixTable($table));; |
| 376 | $finish = $maxIds[$table]; |
| 377 | } else { |
| 378 | $start = $maxIds[$table]; |
| 379 | $finish = (int) Db::fetchOne("SELECT MAX($idCol) FROM " . Common::prefixTable($table)); |
| 380 | } |
| 381 | // Borrowed from Db::segmentedFetchAll |
| 382 | // Request records per $insertIntoTempIterationStep amount |
| 383 | // Loop over the result set, mapping all numeric fields in a single insert query |
| 384 | |
| 385 | // Insert query would be: INSERT IGNORE INTO [temp_table] VALUES (X),(Y),(Z) depending on the amount of fields requested per row |
| 386 | for ($i = $start; $i <= $finish; $i += $insertIntoTempIterationStep) { |
| 387 | $currentParams = array($i, $i + $insertIntoTempIterationStep); |
| 388 | $result = Db::fetchAll($sql, $currentParams); |
| 389 | // Now we loop over the result set of max $insertIntoTempIterationStep rows and create insert queries |
| 390 | $keepValues = []; |
| 391 | foreach ($result as $row) { |
| 392 | $keepValues = array_merge($keepValues, array_filter(array_values($row), "is_numeric")); |
| 393 | if (count($keepValues) >= 1000) { |
| 394 | $insert = 'INSERT IGNORE INTO ' . $tempTableName .' VALUES ('; |
| 395 | $insert .= implode('),(', $keepValues); |
| 396 | $insert .= ')'; |
| 397 | |
| 398 | Db::exec($insert); |
| 399 | $keepValues = []; |
| 400 | } |
| 401 | } |
| 402 | |
| 403 | $insert = 'INSERT IGNORE INTO ' . $tempTableName .' VALUES ('; |
| 404 | $insert .= implode('),(', $keepValues); |
| 405 | $insert .= ')'; |
| 406 | |
| 407 | Db::exec($insert); |
| 408 | } |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | private function lockLogTables() |
| 413 | { |
| 414 | $tables = $this->getTableIdColumns(); |
| 415 | unset($tables['log_action']); // we write lock it |
| 416 | $tableNames = array_keys($tables); |
| 417 | |
| 418 | $readLocks = array(); |
| 419 | foreach ($tableNames as $tableName) { |
| 420 | $readLocks[] = Common::prefixTable($tableName); |
| 421 | } |
| 422 | |
| 423 | Db::lockTables( |
| 424 | $readLocks, |
| 425 | $writeLocks = Common::prefixTables('log_action') |
| 426 | ); |
| 427 | } |
| 428 | |
| 429 | private function deleteUnusedActions() |
| 430 | { |
| 431 | list($logActionTable, $tempTableName) = Common::prefixTables("log_action", self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 432 | |
| 433 | $deleteSql = "DELETE LOW_PRIORITY QUICK IGNORE $logActionTable |
| 434 | FROM $logActionTable |
| 435 | LEFT JOIN $tempTableName tmp ON tmp.idaction = $logActionTable.idaction |
| 436 | WHERE tmp.idaction IS NULL"; |
| 437 | |
| 438 | Db::query($deleteSql); |
| 439 | } |
| 440 | |
| 441 | protected function getTableIdColumns() |
| 442 | { |
| 443 | $columns = array(); |
| 444 | |
| 445 | foreach ($this->logTablesProvider->getAllLogTables() as $logTable) { |
| 446 | $idColumn = $logTable->getIdColumn(); |
| 447 | |
| 448 | if (!empty($idColumn)) { |
| 449 | $columns[$logTable->getName()] = $idColumn; |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | return $columns; |
| 454 | } |
| 455 | } |
| 456 |