LogQueryBuilder
3 months ago
Actions.php
6 months ago
ArchiveSelector.php
1 month ago
ArchiveTableCreator.php
1 month ago
ArchiveTableDao.php
1 month ago
ArchiveWriter.php
1 month ago
ArchivingDbAdapter.php
1 year ago
LogAggregator.php
1 month ago
LogQueryBuilder.php
6 months ago
LogTableTemporary.php
2 years ago
Model.php
1 month ago
RawLogDao.php
6 months ago
TableMetadata.php
1 year ago
RawLogDao.php
345 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | */ |
| 9 | namespace Piwik\DataAccess; |
| 10 | |
| 11 | use 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 | * DAO that queries log tables. |
| 19 | */ |
| 20 | class RawLogDao |
| 21 | { |
| 22 | public const DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME = 'tmp_log_actions_to_keep'; |
| 23 | /** |
| 24 | * @var DimensionMetadataProvider |
| 25 | */ |
| 26 | private $dimensionMetadataProvider; |
| 27 | /** |
| 28 | * @var LogTablesProvider |
| 29 | */ |
| 30 | private $logTablesProvider; |
| 31 | public function __construct(?DimensionMetadataProvider $provider = null, ?LogTablesProvider $logTablesProvider = null) |
| 32 | { |
| 33 | $this->dimensionMetadataProvider = $provider ?: StaticContainer::get('Piwik\\Plugin\\Dimension\\DimensionMetadataProvider'); |
| 34 | $this->logTablesProvider = $logTablesProvider ?: StaticContainer::get('Piwik\\Plugin\\LogTablesProvider'); |
| 35 | } |
| 36 | /** |
| 37 | * @param array $values |
| 38 | * @param string $idVisit |
| 39 | */ |
| 40 | public function updateVisits(array $values, $idVisit) |
| 41 | { |
| 42 | $sql = "UPDATE " . Common::prefixTable('log_visit') . " SET " . $this->getColumnSetExpressions(array_keys($values)) . " WHERE idvisit = ?"; |
| 43 | $this->update($sql, $values, $idVisit); |
| 44 | } |
| 45 | /** |
| 46 | * @param array $values |
| 47 | * @param string $idVisit |
| 48 | */ |
| 49 | public function updateConversions(array $values, $idVisit) |
| 50 | { |
| 51 | $sql = "UPDATE " . Common::prefixTable('log_conversion') . " SET " . $this->getColumnSetExpressions(array_keys($values)) . " WHERE idvisit = ?"; |
| 52 | $this->update($sql, $values, $idVisit); |
| 53 | } |
| 54 | /** |
| 55 | * @param string $from |
| 56 | * @param string $to |
| 57 | * @return int |
| 58 | */ |
| 59 | public function countVisitsWithDatesLimit($from, $to) |
| 60 | { |
| 61 | $sql = "SELECT COUNT(*) AS num_rows" . " FROM `" . Common::prefixTable('log_visit') . "`" . " WHERE visit_last_action_time >= ? AND visit_last_action_time < ?"; |
| 62 | $bind = array($from, $to); |
| 63 | return (int) Db::fetchOne($sql, $bind); |
| 64 | } |
| 65 | /** |
| 66 | * Iterates over logs in a log table in chunks. Parameters to this function are as backend agnostic |
| 67 | * as possible w/o dramatically increasing code complexity. |
| 68 | * |
| 69 | * @param string $logTable The log table name. Unprefixed, eg, `log_visit`. |
| 70 | * @param array[] $conditions An array describing the conditions logs must match in the query. Translates to |
| 71 | * the WHERE part of a SELECT statement. Each element must contain three elements: |
| 72 | * |
| 73 | * * the column name |
| 74 | * * the operator (ie, '=', '<>', '<', etc.) |
| 75 | * * the operand (ie, a value) |
| 76 | * |
| 77 | * The elements are AND-ed together. |
| 78 | * |
| 79 | * Example: |
| 80 | * |
| 81 | * ``` |
| 82 | * array( |
| 83 | * array('visit_first_action_time', '>=', ...), |
| 84 | * array('visit_first_action_time', '<', ...) |
| 85 | * ) |
| 86 | * ``` |
| 87 | * @param int $iterationStep The number of rows to query at a time. |
| 88 | * @param callable $callback The callback that processes each chunk of rows. |
| 89 | * @param string $willDelete Set to true if you will make sure to delete all rows that were fetched. If you are in |
| 90 | * doubt and not sure if to set true or false, use "false". Setting it to true will |
| 91 | * enable an internal performance improvement but it can result in an endless loop if not |
| 92 | * used properly. |
| 93 | */ |
| 94 | public function forAllLogs($logTable, $fields, $conditions, $iterationStep, $callback, $willDelete) |
| 95 | { |
| 96 | $lastId = 0; |
| 97 | if ($willDelete) { |
| 98 | // we don't want to look at eg idvisit so the query will be mostly index covered as the |
| 99 | // "where idvisit > 0 ... ORDER BY idvisit ASC" will be gone... meaning we don't need to look at a huge range |
| 100 | // of visits... |
| 101 | $idField = null; |
| 102 | $bindFunction = function ($bind, $lastId) { |
| 103 | return $bind; |
| 104 | }; |
| 105 | } else { |
| 106 | // when we are not deleting, we need to ensure to iterate over each visitor step by step... meaning we |
| 107 | // need to remember which visit we have already looked at and which one not. Therefore we need to apply |
| 108 | // "where idvisit > $lastId" in the query and "order by idvisit ASC" |
| 109 | $idField = $this->getIdFieldForLogTable($logTable); |
| 110 | $bindFunction = function ($bind, $lastId) { |
| 111 | return array_merge(array($lastId), $bind); |
| 112 | }; |
| 113 | } |
| 114 | [$query, $bind] = $this->createLogIterationQuery($logTable, $idField, $fields, $conditions, $iterationStep); |
| 115 | do { |
| 116 | $rows = Db::fetchAll($query, call_user_func($bindFunction, $bind, $lastId)); |
| 117 | if (!empty($rows)) { |
| 118 | if ($idField) { |
| 119 | $lastId = $rows[count($rows) - 1][$idField]; |
| 120 | } |
| 121 | $callback($rows); |
| 122 | } |
| 123 | } while (count($rows) == $iterationStep); |
| 124 | } |
| 125 | /** |
| 126 | * Deletes conversion items for the supplied visit IDs from log_conversion_item. |
| 127 | * |
| 128 | * @param int[] $visitIds |
| 129 | * @return int The number of deleted rows. |
| 130 | */ |
| 131 | public function deleteConversionItems($visitIds) |
| 132 | { |
| 133 | $sql = "DELETE FROM `" . Common::prefixTable('log_conversion_item') . "` WHERE idvisit IN " . $this->getInFieldExpressionWithInts($visitIds); |
| 134 | $statement = Db::query($sql); |
| 135 | return $statement->rowCount(); |
| 136 | } |
| 137 | /** |
| 138 | * Deletes all unused entries from the log_action table. This method uses a temporary table to store used |
| 139 | * actions, and then deletes rows from log_action that are not in this temporary table. |
| 140 | * |
| 141 | * Table locking is required to avoid concurrency issues. |
| 142 | * |
| 143 | * @throws \Exception If table locking permission is not granted to the current MySQL user. |
| 144 | */ |
| 145 | public function deleteUnusedLogActions() |
| 146 | { |
| 147 | if (!Db::isLockPrivilegeGranted()) { |
| 148 | throw new \Exception("RawLogDao.deleteUnusedLogActions() requires table locking permission in order to complete without error."); |
| 149 | } |
| 150 | // get current max ID in log tables w/ idaction references. |
| 151 | $maxIds = $this->getMaxIdsInLogTables(); |
| 152 | // get max rows to analyze |
| 153 | $max_rows_per_query = PiwikConfig::getInstance()->Deletelogs['delete_logs_unused_actions_max_rows_per_query']; |
| 154 | $this->createTempTableForStoringUsedActions(); |
| 155 | // do large insert (inserting everything before maxIds) w/o locking tables... |
| 156 | $this->insertActionsToKeep($maxIds, $deleteOlderThanMax = \true, $max_rows_per_query); |
| 157 | // ... then do small insert w/ locked tables to minimize the amount of time tables are locked. |
| 158 | $this->lockLogTables(); |
| 159 | $this->insertActionsToKeep($maxIds, $deleteOlderThanMax = \false, $max_rows_per_query); |
| 160 | // delete before unlocking tables so there's no chance a new log row that references an |
| 161 | // unused action will be inserted. |
| 162 | $this->deleteUnusedActions(); |
| 163 | Db::unlockAllTables(); |
| 164 | $this->dropTempTableForStoringUsedActions(); |
| 165 | } |
| 166 | /** |
| 167 | * Returns the list of the website IDs that received some visits between the specified timestamp. The |
| 168 | * start date and the end date is included in the time frame. |
| 169 | * |
| 170 | * @param string $fromDateTime |
| 171 | * @param string $toDateTime |
| 172 | * @return bool true if there are visits for this site between the given timeframe, false if not |
| 173 | */ |
| 174 | public function hasSiteVisitsBetweenTimeframe($fromDateTime, $toDateTime, $idSite) |
| 175 | { |
| 176 | $sites = Db::fetchOne("SELECT 1\n FROM `" . Common::prefixTable('log_visit') . "`\n WHERE idsite = ?\n AND visit_last_action_time >= ?\n AND visit_last_action_time <= ?\n LIMIT 1", array($idSite, $fromDateTime, $toDateTime)); |
| 177 | return (bool) $sites; |
| 178 | } |
| 179 | /** |
| 180 | * @param array $columnsToSet |
| 181 | * @return string |
| 182 | */ |
| 183 | protected function getColumnSetExpressions(array $columnsToSet) |
| 184 | { |
| 185 | $columnsToSet = array_map(function ($column) { |
| 186 | return $column . ' = ?'; |
| 187 | }, $columnsToSet); |
| 188 | return implode(', ', $columnsToSet); |
| 189 | } |
| 190 | /** |
| 191 | * @param array $values |
| 192 | * @param $idVisit |
| 193 | * @param $sql |
| 194 | * @return \Zend_Db_Statement |
| 195 | * @throws \Exception |
| 196 | */ |
| 197 | protected function update($sql, array $values, $idVisit) |
| 198 | { |
| 199 | return Db::query($sql, array_merge(array_values($values), array($idVisit))); |
| 200 | } |
| 201 | protected function getIdFieldForLogTable($logTable) |
| 202 | { |
| 203 | $idColumns = $this->getTableIdColumns(); |
| 204 | if (isset($idColumns[$logTable])) { |
| 205 | return $idColumns[$logTable]; |
| 206 | } |
| 207 | throw new \InvalidArgumentException("Unknown log table '{$logTable}'."); |
| 208 | } |
| 209 | // TODO: instead of creating a log query like this, we should re-use segments. to do this, however, there must be a 1-1 |
| 210 | // mapping for dimensions => segments, and each dimension should automatically have a segment. |
| 211 | private function createLogIterationQuery($logTable, $idField, $fields, $conditions, $iterationStep) |
| 212 | { |
| 213 | $bind = array(); |
| 214 | $sql = "SELECT " . implode(', ', $fields) . " FROM `" . Common::prefixTable($logTable) . "` WHERE "; |
| 215 | $parts = array(); |
| 216 | if ($idField) { |
| 217 | $parts[] = "{$idField} > ?"; |
| 218 | } |
| 219 | foreach ($conditions as $condition) { |
| 220 | [$column, $operator, $value] = $condition; |
| 221 | if (is_array($value)) { |
| 222 | $parts[] = "{$column} IN (" . Common::getSqlStringFieldsArray($value) . ")"; |
| 223 | $bind = array_merge($bind, $value); |
| 224 | } else { |
| 225 | $parts[] = "{$column} {$operator} ?"; |
| 226 | $bind[] = $value; |
| 227 | } |
| 228 | } |
| 229 | $sql .= implode(' AND ', $parts); |
| 230 | if ($idField) { |
| 231 | $sql .= " ORDER BY {$idField} ASC"; |
| 232 | } |
| 233 | $sql .= " LIMIT " . (int) $iterationStep; |
| 234 | return array($sql, $bind); |
| 235 | } |
| 236 | private function getInFieldExpressionWithInts($idVisits) |
| 237 | { |
| 238 | $sql = "("; |
| 239 | $isFirst = \true; |
| 240 | foreach ($idVisits as $idVisit) { |
| 241 | if ($isFirst) { |
| 242 | $isFirst = \false; |
| 243 | } else { |
| 244 | $sql .= ', '; |
| 245 | } |
| 246 | $sql .= (int) $idVisit; |
| 247 | } |
| 248 | $sql .= ")"; |
| 249 | return $sql; |
| 250 | } |
| 251 | protected function getMaxIdsInLogTables() |
| 252 | { |
| 253 | $idColumns = $this->getTableIdColumns(); |
| 254 | $tables = array_keys($idColumns); |
| 255 | $result = array(); |
| 256 | foreach ($tables as $table) { |
| 257 | $idCol = $idColumns[$table]; |
| 258 | $result[$table] = Db::fetchOne("SELECT MAX({$idCol}) FROM `" . Common::prefixTable($table) . "`"); |
| 259 | } |
| 260 | return $result; |
| 261 | } |
| 262 | private function createTempTableForStoringUsedActions() |
| 263 | { |
| 264 | $sql = "CREATE TEMPORARY TABLE " . Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME) . " (\n\t\t\t\t\tidaction INTEGER(10) UNSIGNED NOT NULL,\n\t\t\t\t\tPRIMARY KEY (idaction)\n\t\t\t\t)"; |
| 265 | Db::query($sql); |
| 266 | } |
| 267 | private function dropTempTableForStoringUsedActions() |
| 268 | { |
| 269 | $sql = "DROP TABLE " . Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 270 | Db::query($sql); |
| 271 | } |
| 272 | // protected for testing purposes |
| 273 | protected function insertActionsToKeep($maxIds, $olderThan = \true, $insertIntoTempIterationStep = 100000) |
| 274 | { |
| 275 | $tempTableName = Common::prefixTable(self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 276 | $idColumns = $this->getTableIdColumns(); |
| 277 | foreach ($this->dimensionMetadataProvider->getActionReferenceColumnsByTable() as $table => $columns) { |
| 278 | $idCol = $idColumns[$table]; |
| 279 | // Create select query for requesting ALL needed fields at once |
| 280 | $sql = "SELECT " . implode(',', $columns) . " FROM `" . Common::prefixTable($table) . "` WHERE {$idCol} >= ? AND {$idCol} < ?"; |
| 281 | if ($olderThan) { |
| 282 | // 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! |
| 283 | $start = (int) Db::fetchOne("SELECT MIN({$idCol}) FROM `" . Common::prefixTable($table) . "`"); |
| 284 | $finish = $maxIds[$table]; |
| 285 | } else { |
| 286 | $start = $maxIds[$table]; |
| 287 | $finish = (int) Db::fetchOne("SELECT MAX({$idCol}) FROM `" . Common::prefixTable($table) . "`"); |
| 288 | } |
| 289 | // Borrowed from Db::segmentedFetchAll |
| 290 | // Request records per $insertIntoTempIterationStep amount |
| 291 | // Loop over the result set, mapping all numeric fields in a single insert query |
| 292 | // Insert query would be: INSERT IGNORE INTO [temp_table] VALUES (X),(Y),(Z) depending on the amount of fields requested per row |
| 293 | for ($i = $start; $i <= $finish; $i += $insertIntoTempIterationStep) { |
| 294 | $currentParams = array($i, $i + $insertIntoTempIterationStep); |
| 295 | $result = Db::fetchAll($sql, $currentParams); |
| 296 | // Now we loop over the result set of max $insertIntoTempIterationStep rows and create insert queries |
| 297 | $keepValues = []; |
| 298 | foreach ($result as $row) { |
| 299 | $keepValues = array_merge($keepValues, array_filter(array_values($row), "is_numeric")); |
| 300 | if (count($keepValues) >= 1000) { |
| 301 | $insert = 'INSERT IGNORE INTO ' . $tempTableName . ' VALUES ('; |
| 302 | $insert .= implode('),(', $keepValues); |
| 303 | $insert .= ')'; |
| 304 | Db::exec($insert); |
| 305 | $keepValues = []; |
| 306 | } |
| 307 | } |
| 308 | $insert = 'INSERT IGNORE INTO ' . $tempTableName . ' VALUES ('; |
| 309 | $insert .= implode('),(', $keepValues); |
| 310 | $insert .= ')'; |
| 311 | Db::exec($insert); |
| 312 | } |
| 313 | } |
| 314 | } |
| 315 | private function lockLogTables() |
| 316 | { |
| 317 | $tables = $this->getTableIdColumns(); |
| 318 | unset($tables['log_action']); |
| 319 | // we write lock it |
| 320 | $tableNames = array_keys($tables); |
| 321 | $readLocks = array(); |
| 322 | foreach ($tableNames as $tableName) { |
| 323 | $readLocks[] = Common::prefixTable($tableName); |
| 324 | } |
| 325 | Db::lockTables($readLocks, $writeLocks = Common::prefixTables('log_action')); |
| 326 | } |
| 327 | private function deleteUnusedActions() |
| 328 | { |
| 329 | [$logActionTable, $tempTableName] = Common::prefixTables("log_action", self::DELETE_UNUSED_ACTIONS_TEMP_TABLE_NAME); |
| 330 | $deleteSql = "DELETE LOW_PRIORITY QUICK IGNORE `{$logActionTable}`\n\t\t\t\t\t\tFROM `{$logActionTable}`\n\t\t\t\t LEFT JOIN `{$tempTableName}` tmp ON tmp.idaction = `{$logActionTable}`.idaction\n\t\t\t\t\t WHERE tmp.idaction IS NULL"; |
| 331 | Db::query($deleteSql); |
| 332 | } |
| 333 | protected function getTableIdColumns() |
| 334 | { |
| 335 | $columns = array(); |
| 336 | foreach ($this->logTablesProvider->getAllLogTables() as $logTable) { |
| 337 | $idColumn = $logTable->getIdColumn(); |
| 338 | if (!empty($idColumn)) { |
| 339 | $columns[$logTable->getName()] = $idColumn; |
| 340 | } |
| 341 | } |
| 342 | return $columns; |
| 343 | } |
| 344 | } |
| 345 |