ArchivingStatus.php
6 months ago
BlobTableAggregator.php
2 weeks ago
Loader.php
2 weeks ago
LoaderLock.php
2 weeks ago
Parameters.php
2 weeks ago
PluginsArchiver.php
2 weeks ago
PluginsArchiverException.php
2 years ago
Record.php
2 weeks ago
RecordBuilder.php
2 weeks ago
Rules.php
2 weeks ago
RecordBuilder.php
530 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | */ |
| 9 | namespace Piwik\ArchiveProcessor; |
| 10 | |
| 11 | use Piwik\Archive; |
| 12 | use Piwik\ArchiveProcessor; |
| 13 | use Piwik\Common; |
| 14 | use Piwik\DataTable; |
| 15 | use Piwik\DataTable\Row; |
| 16 | use Piwik\Piwik; |
| 17 | /** |
| 18 | * Inherit from this class to define archiving logic for one or more records. |
| 19 | */ |
| 20 | abstract class RecordBuilder |
| 21 | { |
| 22 | /** |
| 23 | * @var int|null |
| 24 | */ |
| 25 | protected $maxRowsInTable; |
| 26 | /** |
| 27 | * @var int|null |
| 28 | */ |
| 29 | protected $maxRowsInSubtable; |
| 30 | /** |
| 31 | * @var string|int|null |
| 32 | */ |
| 33 | protected $columnToSortByBeforeTruncation; |
| 34 | /** |
| 35 | * @var array|null |
| 36 | */ |
| 37 | protected $columnAggregationOps; |
| 38 | /** |
| 39 | * @var array<string|int,string|int>|null |
| 40 | */ |
| 41 | protected $columnToRenameAfterAggregation; |
| 42 | /** |
| 43 | * @param array|null $columnAggregationOps |
| 44 | * @param array<string|int,string|int>|null $columnToRenameAfterAggregation |
| 45 | */ |
| 46 | public function __construct(?int $maxRowsInTable = null, ?int $maxRowsInSubtable = null, ?string $columnToSortByBeforeTruncation = null, ?array $columnAggregationOps = null, ?array $columnToRenameAfterAggregation = null) |
| 47 | { |
| 48 | $this->maxRowsInTable = $maxRowsInTable; |
| 49 | $this->maxRowsInSubtable = $maxRowsInSubtable; |
| 50 | $this->columnToSortByBeforeTruncation = $columnToSortByBeforeTruncation; |
| 51 | $this->columnAggregationOps = $columnAggregationOps; |
| 52 | $this->columnToRenameAfterAggregation = $columnToRenameAfterAggregation; |
| 53 | } |
| 54 | public function isEnabled(ArchiveProcessor $archiveProcessor) : bool |
| 55 | { |
| 56 | return \true; |
| 57 | } |
| 58 | /** |
| 59 | * Uses the protected `aggregate()` function to build records by aggregating log table data directly, then |
| 60 | * inserts them as archive data. |
| 61 | */ |
| 62 | public function buildFromLogs(ArchiveProcessor $archiveProcessor) : void |
| 63 | { |
| 64 | if (!$this->isEnabled($archiveProcessor)) { |
| 65 | return; |
| 66 | } |
| 67 | $recordsBuilt = $this->getRecordMetadata($archiveProcessor); |
| 68 | $recordMetadataByName = []; |
| 69 | foreach ($recordsBuilt as $recordMetadata) { |
| 70 | if (!$recordMetadata instanceof \Piwik\ArchiveProcessor\Record) { |
| 71 | continue; |
| 72 | } |
| 73 | $recordMetadataByName[$recordMetadata->getName()] = $recordMetadata; |
| 74 | } |
| 75 | $numericRecords = []; |
| 76 | $records = $this->aggregate($archiveProcessor); |
| 77 | foreach ($records as $recordName => $recordValue) { |
| 78 | if (empty($recordMetadataByName[$recordName])) { |
| 79 | if ($recordValue instanceof DataTable) { |
| 80 | Common::destroy($recordValue); |
| 81 | } |
| 82 | continue; |
| 83 | } |
| 84 | if ($recordValue instanceof DataTable) { |
| 85 | $record = $recordMetadataByName[$recordName]; |
| 86 | $maxRowsInTable = $record->getMaxRowsInTable() ?? $this->maxRowsInTable; |
| 87 | $maxRowsInSubtable = $record->getMaxRowsInSubtable() ?? $this->maxRowsInSubtable; |
| 88 | $columnToSortByBeforeTruncation = $record->getColumnToSortByBeforeTruncation() ?? $this->columnToSortByBeforeTruncation; |
| 89 | $this->insertBlobRecord($archiveProcessor, $recordName, $recordValue, $maxRowsInTable, $maxRowsInSubtable, $columnToSortByBeforeTruncation); |
| 90 | Common::destroy($recordValue); |
| 91 | } else { |
| 92 | // collect numeric records so we can insert them all at once |
| 93 | $numericRecords[$recordName] = $recordValue; |
| 94 | } |
| 95 | } |
| 96 | unset($records); |
| 97 | if (!empty($numericRecords)) { |
| 98 | $archiveProcessor->insertNumericRecords($numericRecords); |
| 99 | } |
| 100 | } |
| 101 | /** |
| 102 | * Builds records for non-day periods by aggregating day records together, then inserts |
| 103 | * them as archive data. |
| 104 | */ |
| 105 | public function buildForNonDayPeriod(ArchiveProcessor $archiveProcessor) : void |
| 106 | { |
| 107 | if (!$this->isEnabled($archiveProcessor)) { |
| 108 | return; |
| 109 | } |
| 110 | $requestedReports = $archiveProcessor->getParams()->getArchiveOnlyReportAsArray(); |
| 111 | $foundRequestedReports = $archiveProcessor->getParams()->getFoundRequestedReports(); |
| 112 | $recordsBuilt = $this->getRecordMetadata($archiveProcessor); |
| 113 | $numericRecords = array_filter($recordsBuilt, function (\Piwik\ArchiveProcessor\Record $r) { |
| 114 | return $r->getType() == \Piwik\ArchiveProcessor\Record::TYPE_NUMERIC; |
| 115 | }); |
| 116 | $blobRecords = array_filter($recordsBuilt, function (\Piwik\ArchiveProcessor\Record $r) { |
| 117 | return $r->getType() == \Piwik\ArchiveProcessor\Record::TYPE_BLOB; |
| 118 | }); |
| 119 | $blobRecordsByName = []; |
| 120 | foreach ($blobRecords as $blobRecord) { |
| 121 | $blobRecordsByName[$blobRecord->getName()] = $blobRecord; |
| 122 | } |
| 123 | $aggregatedCounts = []; |
| 124 | foreach ($blobRecords as $record) { |
| 125 | $flatRecordName = $record->getBuiltFromFlatRecord(); |
| 126 | if (empty($flatRecordName) || !in_array($flatRecordName, $requestedReports)) { |
| 127 | continue; |
| 128 | } |
| 129 | // If the flat record is requested directly, also force aggregation of the corresponding |
| 130 | // hierarchical record so the API can still read the expected hierarchical blob. |
| 131 | if (!in_array($record->getName(), $requestedReports)) { |
| 132 | $requestedReports[] = $record->getName(); |
| 133 | } |
| 134 | // We are about to rebuild this record from flat data, so treat it as not-found and |
| 135 | // make sure it is re-aggregated even if a previous archive row exists. |
| 136 | $indexInFoundRecords = array_search($record->getName(), $foundRequestedReports); |
| 137 | if ($indexInFoundRecords !== \false) { |
| 138 | unset($foundRequestedReports[$indexInFoundRecords]); |
| 139 | } |
| 140 | } |
| 141 | // make sure if there are requested numeric records that depend on blob records, that the blob records will be archived first |
| 142 | foreach ($numericRecords as $record) { |
| 143 | if (empty($record->getCountOfRecordName()) || !in_array($record->getName(), $requestedReports)) { |
| 144 | continue; |
| 145 | } |
| 146 | $dependentRecordName = $record->getCountOfRecordName(); |
| 147 | if (!in_array($dependentRecordName, $requestedReports)) { |
| 148 | $requestedReports[] = $dependentRecordName; |
| 149 | } |
| 150 | // we need to aggregate the blob record to get the count, so even if it's found, we must re-aggregate it |
| 151 | // TODO: this could potentially be optimized away, but it would be non-trivial given the current ArchiveProcessor API |
| 152 | $indexInFoundRecords = array_search($dependentRecordName, $foundRequestedReports); |
| 153 | if ($indexInFoundRecords !== \false) { |
| 154 | unset($foundRequestedReports[$indexInFoundRecords]); |
| 155 | } |
| 156 | } |
| 157 | $processedFlatRecords = []; |
| 158 | foreach ($blobRecords as $record) { |
| 159 | if (!empty($requestedReports) && (!in_array($record->getName(), $requestedReports) || in_array($record->getName(), $foundRequestedReports))) { |
| 160 | continue; |
| 161 | } |
| 162 | if (isset($processedFlatRecords[$record->getName()])) { |
| 163 | continue; |
| 164 | } |
| 165 | $maxRowsInTable = $record->getMaxRowsInTable() ?? $this->maxRowsInTable; |
| 166 | $maxRowsInSubtable = $record->getMaxRowsInSubtable() ?? $this->maxRowsInSubtable; |
| 167 | $columnToSortByBeforeTruncation = $record->getColumnToSortByBeforeTruncation() ?? $this->columnToSortByBeforeTruncation; |
| 168 | $columnToRenameAfterAggregation = $record->getColumnToRenameAfterAggregation() ?? $this->columnToRenameAfterAggregation; |
| 169 | $columnAggregationOps = $record->getBlobColumnAggregationOps() ?? $this->columnAggregationOps; |
| 170 | if ($this->aggregateBuiltFromFlatRecordForNonDay($archiveProcessor, $record, $blobRecordsByName, $columnAggregationOps, $columnToRenameAfterAggregation, $columnToSortByBeforeTruncation, $processedFlatRecords)) { |
| 171 | continue; |
| 172 | } |
| 173 | // only do recursive row counts if there is a numeric record that depends on it |
| 174 | $countRecursiveRows = $countLeafRows = []; |
| 175 | foreach ($numericRecords as $numeric) { |
| 176 | if ($numeric->getCountOfRecordName() == $record->getName()) { |
| 177 | if ($numeric->getCountOfRecordNameIsRecursive()) { |
| 178 | $countRecursiveRows[] = $numeric->getCountOfRecordName(); |
| 179 | } |
| 180 | if ($numeric->getCountOfRecordNameIsForLeafs()) { |
| 181 | $countLeafRows[] = $numeric->getCountOfRecordName(); |
| 182 | } |
| 183 | } |
| 184 | } |
| 185 | $recordTransform = $record->getAggregatedRecordTransform(); |
| 186 | $postAggregationTransform = $recordTransform === null ? null : function (DataTable $table) use($recordTransform, $archiveProcessor, $record) : void { |
| 187 | $recordTransform($table, $archiveProcessor, $record); |
| 188 | }; |
| 189 | $counts = $archiveProcessor->aggregateDataTableRecords($record->getName(), $maxRowsInTable, $maxRowsInSubtable, $columnToSortByBeforeTruncation, $columnAggregationOps, $columnToRenameAfterAggregation, $countRecursiveRows, $countLeafRows, $postAggregationTransform); |
| 190 | $aggregatedCounts = array_merge($aggregatedCounts, $counts); |
| 191 | } |
| 192 | if (!empty($numericRecords)) { |
| 193 | // handle metrics that are aggregated using metric values from child periods |
| 194 | $autoAggregateMetrics = array_filter($numericRecords, function (\Piwik\ArchiveProcessor\Record $r) { |
| 195 | return empty($r->getCountOfRecordName()); |
| 196 | }); |
| 197 | $autoAggregateMetrics = array_map(function (\Piwik\ArchiveProcessor\Record $r) { |
| 198 | return $r->getName(); |
| 199 | }, $autoAggregateMetrics); |
| 200 | if (!empty($requestedReports)) { |
| 201 | $autoAggregateMetrics = array_filter($autoAggregateMetrics, function ($name) use($requestedReports, $foundRequestedReports) { |
| 202 | return in_array($name, $requestedReports) && !in_array($name, $foundRequestedReports); |
| 203 | }); |
| 204 | } |
| 205 | $autoAggregateMetrics = array_values($autoAggregateMetrics); |
| 206 | if (!empty($autoAggregateMetrics)) { |
| 207 | $archiveProcessor->aggregateNumericMetrics($autoAggregateMetrics, $this->columnAggregationOps); |
| 208 | } |
| 209 | // handle metrics that are set to counts of blob records |
| 210 | $recordCountMetricValues = []; |
| 211 | $recordCountMetrics = array_filter($numericRecords, function (\Piwik\ArchiveProcessor\Record $r) { |
| 212 | return !empty($r->getCountOfRecordName()); |
| 213 | }); |
| 214 | foreach ($recordCountMetrics as $record) { |
| 215 | $dependentRecordName = $record->getCountOfRecordName(); |
| 216 | if (empty($aggregatedCounts[$dependentRecordName])) { |
| 217 | continue; |
| 218 | // dependent record not archived, so skip this metric |
| 219 | } |
| 220 | $count = $aggregatedCounts[$dependentRecordName]; |
| 221 | if ($record->getCountOfRecordNameIsForLeafs()) { |
| 222 | $recordCountMetricValues[$record->getName()] = $count['leafs']; |
| 223 | } elseif ($record->getCountOfRecordNameIsRecursive()) { |
| 224 | $recordCountMetricValues[$record->getName()] = $count['recursive']; |
| 225 | } else { |
| 226 | $recordCountMetricValues[$record->getName()] = $count['level0']; |
| 227 | } |
| 228 | $transform = $record->getMultiplePeriodTransform(); |
| 229 | if (!empty($transform)) { |
| 230 | $recordCountMetricValues[$record->getName()] = $transform($recordCountMetricValues[$record->getName()], $count); |
| 231 | } |
| 232 | } |
| 233 | if (!empty($recordCountMetricValues)) { |
| 234 | $archiveProcessor->insertNumericRecords($recordCountMetricValues); |
| 235 | } |
| 236 | } |
| 237 | } |
| 238 | protected function aggregateBuiltFromFlatRecordForNonDay(ArchiveProcessor $archiveProcessor, \Piwik\ArchiveProcessor\Record $hierarchicalRecord, array $blobRecordsByName, ?array $columnAggregationOps, ?array $columnToRenameAfterAggregation, ?string $columnToSortByBeforeTruncation, array &$processedFlatRecords) : bool |
| 239 | { |
| 240 | $flatRecordName = $hierarchicalRecord->getBuiltFromFlatRecord(); |
| 241 | if (empty($flatRecordName)) { |
| 242 | return \false; |
| 243 | } |
| 244 | $flatToHierarchyPathCallback = $hierarchicalRecord->getFlatToHierarchyPathCallback(); |
| 245 | if (!is_callable($flatToHierarchyPathCallback)) { |
| 246 | return \false; |
| 247 | } |
| 248 | $flatRecord = $blobRecordsByName[$flatRecordName] ?? null; |
| 249 | if (empty($flatRecord)) { |
| 250 | return \false; |
| 251 | } |
| 252 | $flatColumnAggregationOps = $flatRecord->getBlobColumnAggregationOps() ?? $this->columnAggregationOps; |
| 253 | $flatColumnToRenameAfterAggregation = $flatRecord->getColumnToRenameAfterAggregation() ?? $this->columnToRenameAfterAggregation; |
| 254 | $flatColumnToSortByBeforeTruncation = $flatRecord->getColumnToSortByBeforeTruncation() ?? $this->columnToSortByBeforeTruncation; |
| 255 | $flatMaxRowsInTable = $flatRecord->getMaxRowsInTable() ?? $this->maxRowsInTable; |
| 256 | [$flatTable, $hasFlatSourceData, $periodsWithFlatRecord] = $this->aggregateRootDataTableFromBlobs($archiveProcessor, $flatRecordName, $flatColumnAggregationOps, $flatColumnToRenameAfterAggregation); |
| 257 | $allSubperiodKeys = $this->getAllSubperiodKeys($archiveProcessor); |
| 258 | $periodsWithoutFlatRecord = array_diff_key($allSubperiodKeys, $periodsWithFlatRecord); |
| 259 | $hasLegacyFallbackData = \false; |
| 260 | $legacyReducerCallback = $hierarchicalRecord->getLegacyHierarchyToFlatReducerCallback(); |
| 261 | if (!empty($periodsWithoutFlatRecord) && is_callable($legacyReducerCallback)) { |
| 262 | $hasLegacyFallbackData = $this->aggregateLegacyHierarchyPeriodsIntoFlatTable($archiveProcessor, $hierarchicalRecord->getName(), $flatTable, $legacyReducerCallback, $hierarchicalRecord, $columnAggregationOps, $columnToRenameAfterAggregation, $periodsWithoutFlatRecord); |
| 263 | } |
| 264 | if (!$hasFlatSourceData && !$hasLegacyFallbackData) { |
| 265 | Common::destroy($flatTable); |
| 266 | return \false; |
| 267 | } |
| 268 | $flatTransform = $flatRecord->getAggregatedRecordTransform(); |
| 269 | if (null !== $flatTransform) { |
| 270 | $flatTransform($flatTable, $archiveProcessor, $flatRecord); |
| 271 | } |
| 272 | $flatSerialized = $flatTable->getSerialized($flatMaxRowsInTable, null, $flatColumnToSortByBeforeTruncation); |
| 273 | $archiveProcessor->insertBlobRecord($flatRecordName, $flatSerialized); |
| 274 | unset($flatSerialized); |
| 275 | $processedFlatRecords[$flatRecordName] = \true; |
| 276 | $hierarchicalTable = $this->buildHierarchicalTableFromFlatTableAndConsumeRows($flatTable, $columnAggregationOps, function (Row $flatRow) use($flatToHierarchyPathCallback, $archiveProcessor, $hierarchicalRecord) { |
| 277 | return call_user_func($flatToHierarchyPathCallback, $flatRow, $archiveProcessor, $hierarchicalRecord); |
| 278 | }); |
| 279 | $this->beforeInsertBuiltFromFlatHierarchyRecord($archiveProcessor, $hierarchicalRecord, $hierarchicalTable, $flatTable); |
| 280 | $hierarchicalTransform = $hierarchicalRecord->getAggregatedRecordTransform(); |
| 281 | if (null !== $hierarchicalTransform) { |
| 282 | $hierarchicalTransform($hierarchicalTable, $archiveProcessor, $hierarchicalRecord); |
| 283 | } |
| 284 | $hierarchicalSerialized = $hierarchicalTable->getSerialized(null, null, $columnToSortByBeforeTruncation); |
| 285 | $archiveProcessor->insertBlobRecord($hierarchicalRecord->getName(), $hierarchicalSerialized); |
| 286 | unset($hierarchicalSerialized); |
| 287 | Common::destroy($hierarchicalTable); |
| 288 | Common::destroy($flatTable); |
| 289 | return \true; |
| 290 | } |
| 291 | protected function aggregateLegacyHierarchyPeriodsIntoFlatTable(ArchiveProcessor $archiveProcessor, string $recordName, DataTable $flatTable, callable $legacyReducerCallback, \Piwik\ArchiveProcessor\Record $hierarchicalRecord, ?array $columnsAggregationOperation, ?array $columnsToRenameAfterAggregation, ?array $periodsToInclude) : bool |
| 292 | { |
| 293 | $currentPeriod = null; |
| 294 | $currentPeriodRows = []; |
| 295 | $hasRows = \false; |
| 296 | foreach ($this->querySingleBlobRows($archiveProcessor, $recordName) as $archiveDataRow) { |
| 297 | $period = $archiveDataRow['date1'] . ',' . $archiveDataRow['date2']; |
| 298 | if ($periodsToInclude !== null && !isset($periodsToInclude[$period])) { |
| 299 | continue; |
| 300 | } |
| 301 | if ($currentPeriod !== null && $period !== $currentPeriod) { |
| 302 | $hasRows = $this->reduceLegacyHierarchyPeriodRowsIntoFlatTable($currentPeriodRows, $recordName, $flatTable, $legacyReducerCallback, $archiveProcessor, $hierarchicalRecord, $columnsAggregationOperation, $columnsToRenameAfterAggregation) || $hasRows; |
| 303 | $currentPeriodRows = []; |
| 304 | } |
| 305 | $currentPeriod = $period; |
| 306 | $currentPeriodRows[] = $archiveDataRow; |
| 307 | } |
| 308 | if (!empty($currentPeriodRows)) { |
| 309 | $hasRows = $this->reduceLegacyHierarchyPeriodRowsIntoFlatTable($currentPeriodRows, $recordName, $flatTable, $legacyReducerCallback, $archiveProcessor, $hierarchicalRecord, $columnsAggregationOperation, $columnsToRenameAfterAggregation) || $hasRows; |
| 310 | } |
| 311 | return $hasRows; |
| 312 | } |
| 313 | protected function reduceLegacyHierarchyPeriodRowsIntoFlatTable(array $periodRows, string $recordName, DataTable $flatTable, callable $legacyReducerCallback, ArchiveProcessor $archiveProcessor, \Piwik\ArchiveProcessor\Record $hierarchicalRecord, ?array $columnsAggregationOperation, ?array $columnsToRenameAfterAggregation) : bool |
| 314 | { |
| 315 | [$legacyHierarchicalTable, $hasRows] = \Piwik\ArchiveProcessor\BlobTableAggregator::aggregateBlobRows($periodRows, $recordName, $columnsAggregationOperation, function (DataTable $table) use($archiveProcessor, $columnsToRenameAfterAggregation) : void { |
| 316 | $archiveProcessor->renameColumnsAfterAggregation($table, $columnsToRenameAfterAggregation); |
| 317 | }); |
| 318 | if ($hasRows) { |
| 319 | call_user_func($legacyReducerCallback, $legacyHierarchicalTable, $flatTable, $archiveProcessor, $hierarchicalRecord); |
| 320 | } |
| 321 | Common::destroy($legacyHierarchicalTable); |
| 322 | return $hasRows; |
| 323 | } |
| 324 | /** |
| 325 | * Hook executed after the hierarchy table has been rebuilt from the flat table and before |
| 326 | * the hierarchical blob record is serialized and inserted. |
| 327 | * |
| 328 | * Intended for plugin-specific finalization (for example, metadata or column cleanup) when |
| 329 | * using setBuiltFromFlatRecord(). The flat table has already been serialized at this point |
| 330 | * and may have been fully consumed while rebuilding the hierarchy. |
| 331 | */ |
| 332 | protected function beforeInsertBuiltFromFlatHierarchyRecord(ArchiveProcessor $archiveProcessor, \Piwik\ArchiveProcessor\Record $hierarchicalRecord, DataTable $hierarchicalTable, DataTable $flatTable) : void |
| 333 | { |
| 334 | } |
| 335 | protected function buildHierarchicalTableFromFlatTable(DataTable $flatTable, ?array $columnAggregationOps, callable $flatToHierarchyPathCallback, array $defaultHierarchyRowColumns = []) : DataTable |
| 336 | { |
| 337 | $hierarchicalTable = new DataTable(); |
| 338 | if (!empty($columnAggregationOps)) { |
| 339 | $hierarchicalTable->setMetadata(DataTable::COLUMN_AGGREGATION_OPS_METADATA_NAME, $columnAggregationOps); |
| 340 | } |
| 341 | foreach ($flatTable->getRows() as $flatRow) { |
| 342 | if ($flatRow->isSummaryRow()) { |
| 343 | if ($this->isSummaryRowEmpty($flatRow)) { |
| 344 | continue; |
| 345 | } |
| 346 | $summaryRow = $hierarchicalTable->getRowFromId(DataTable::ID_SUMMARY_ROW); |
| 347 | if ($summaryRow === \false) { |
| 348 | $summaryRow = clone $flatRow; |
| 349 | $summaryRow->setIsSummaryRow(); |
| 350 | $hierarchicalTable->addSummaryRow($summaryRow); |
| 351 | continue; |
| 352 | } |
| 353 | $this->sumRowIntoDestination($flatRow, $summaryRow, $columnAggregationOps); |
| 354 | continue; |
| 355 | } |
| 356 | $path = call_user_func($flatToHierarchyPathCallback, $flatRow); |
| 357 | if (!is_array($path) || empty($path)) { |
| 358 | continue; |
| 359 | } |
| 360 | [$destinationRow, $level] = $hierarchicalTable->walkPath($path, $defaultHierarchyRowColumns, 0); |
| 361 | if (!$destinationRow instanceof Row) { |
| 362 | continue; |
| 363 | } |
| 364 | $this->sumRowIntoDestination($flatRow, $destinationRow, $columnAggregationOps); |
| 365 | } |
| 366 | return $hierarchicalTable; |
| 367 | } |
| 368 | protected function buildHierarchicalTableFromFlatTableAndConsumeRows(DataTable $flatTable, ?array $columnAggregationOps, callable $flatToHierarchyPathCallback, array $defaultHierarchyRowColumns = []) : DataTable |
| 369 | { |
| 370 | $hierarchicalTable = new DataTable(); |
| 371 | if (!empty($columnAggregationOps)) { |
| 372 | $hierarchicalTable->setMetadata(DataTable::COLUMN_AGGREGATION_OPS_METADATA_NAME, $columnAggregationOps); |
| 373 | } |
| 374 | while (($flatRow = $flatTable->shiftRow()) instanceof Row) { |
| 375 | $path = call_user_func($flatToHierarchyPathCallback, $flatRow); |
| 376 | if (is_array($path) && !empty($path)) { |
| 377 | [$destinationRow, $level] = $hierarchicalTable->walkPath($path, $defaultHierarchyRowColumns, 0); |
| 378 | if ($destinationRow instanceof Row) { |
| 379 | $this->sumRowIntoDestination($flatRow, $destinationRow, $columnAggregationOps); |
| 380 | } |
| 381 | } |
| 382 | Common::destroy($flatRow); |
| 383 | } |
| 384 | $summaryRow = $flatTable->getSummaryRow(); |
| 385 | if ($summaryRow instanceof Row && !$this->isSummaryRowEmpty($summaryRow)) { |
| 386 | $destinationSummaryRow = $hierarchicalTable->getRowFromId(DataTable::ID_SUMMARY_ROW); |
| 387 | if ($destinationSummaryRow === \false) { |
| 388 | $destinationSummaryRow = clone $summaryRow; |
| 389 | $destinationSummaryRow->setIsSummaryRow(); |
| 390 | $hierarchicalTable->addSummaryRow($destinationSummaryRow); |
| 391 | } else { |
| 392 | $this->sumRowIntoDestination($summaryRow, $destinationSummaryRow, $columnAggregationOps); |
| 393 | } |
| 394 | } |
| 395 | $flatTable->deleteRow(DataTable::ID_SUMMARY_ROW); |
| 396 | Common::destroy($summaryRow); |
| 397 | return $hierarchicalTable; |
| 398 | } |
| 399 | protected function sumRowIntoDestination(Row $source, Row $destination, ?array $columnAggregationOps) : void |
| 400 | { |
| 401 | $sourceCopy = clone $source; |
| 402 | // Preserve original column representation (eg "0.0620" strings) when |
| 403 | // destination does not have a value yet. This keeps day flat-first |
| 404 | // output consistent with legacy day archiving. |
| 405 | foreach ($sourceCopy->getColumns() as $columnName => $columnValue) { |
| 406 | if ($columnName === 'label') { |
| 407 | continue; |
| 408 | } |
| 409 | if ($destination->getColumn($columnName) !== \false) { |
| 410 | continue; |
| 411 | } |
| 412 | $destination->setColumn($columnName, $columnValue); |
| 413 | $sourceCopy->deleteColumn($columnName); |
| 414 | } |
| 415 | $destination->sumRow($sourceCopy, \true, $columnAggregationOps ?? []); |
| 416 | } |
| 417 | protected function isSummaryRowEmpty(Row $summaryRow) : bool |
| 418 | { |
| 419 | foreach ($summaryRow->getColumns() as $name => $value) { |
| 420 | if ($name === 'label') { |
| 421 | continue; |
| 422 | } |
| 423 | if (!empty($value)) { |
| 424 | return \false; |
| 425 | } |
| 426 | } |
| 427 | return \true; |
| 428 | } |
| 429 | /** |
| 430 | * Aggregates a root blob record while discovering periods that contain the root record in a single pass. |
| 431 | * |
| 432 | * @return array{0: DataTable, 1: bool, 2: array<string, bool>} |
| 433 | */ |
| 434 | protected function aggregateRootDataTableFromBlobs(ArchiveProcessor $archiveProcessor, string $recordName, ?array $columnsAggregationOperation, ?array $columnsToRenameAfterAggregation) : array |
| 435 | { |
| 436 | $periodsWithRootRecord = []; |
| 437 | [$result, $hasRows] = \Piwik\ArchiveProcessor\BlobTableAggregator::aggregateBlobRows($this->querySingleBlobRows($archiveProcessor, $recordName), $recordName, $columnsAggregationOperation, function (DataTable $table) use($archiveProcessor, $columnsToRenameAfterAggregation) : void { |
| 438 | $archiveProcessor->renameColumnsAfterAggregation($table, $columnsToRenameAfterAggregation); |
| 439 | }, function (array $archiveDataRow) use(&$periodsWithRootRecord, $recordName) : bool { |
| 440 | $period = $archiveDataRow['date1'] . ',' . $archiveDataRow['date2']; |
| 441 | if ($archiveDataRow['name'] === $recordName) { |
| 442 | $periodsWithRootRecord[$period] = \true; |
| 443 | return \true; |
| 444 | } |
| 445 | return isset($periodsWithRootRecord[$period]); |
| 446 | }); |
| 447 | return [$result, $hasRows, $periodsWithRootRecord]; |
| 448 | } |
| 449 | protected function querySingleBlobRows(ArchiveProcessor $archiveProcessor, string $recordName) : iterable |
| 450 | { |
| 451 | $archive = Archive::factory($archiveProcessor->getParams()->getSegment(), $archiveProcessor->getParams()->getPeriod()->getSubperiods(), [$archiveProcessor->getParams()->getSite()->getId()]); |
| 452 | if (!method_exists($archive, 'querySingleBlob')) { |
| 453 | return []; |
| 454 | } |
| 455 | return $archive->querySingleBlob($recordName); |
| 456 | } |
| 457 | protected function getAllSubperiodKeys(ArchiveProcessor $archiveProcessor) : array |
| 458 | { |
| 459 | $result = []; |
| 460 | foreach ($archiveProcessor->getParams()->getPeriod()->getSubperiods() as $period) { |
| 461 | $result[$period->getDateStart()->toString() . ',' . $period->getDateEnd()->toString()] = \true; |
| 462 | } |
| 463 | return $result; |
| 464 | } |
| 465 | /** |
| 466 | * Returns metadata for records primarily used when aggregating over non-day periods. Every numeric/blob |
| 467 | * record your RecordBuilder creates should have an associated piece of record metadata. |
| 468 | * |
| 469 | * @return Record[] |
| 470 | */ |
| 471 | public abstract function getRecordMetadata(ArchiveProcessor $archiveProcessor) : array; |
| 472 | /** |
| 473 | * Derived classes should define this method to aggregate log data for a single day and return the records |
| 474 | * to store indexed by record names. |
| 475 | * |
| 476 | * @return array<string, DataTable|int|float|string> Record values indexed by their record name, eg, `['MyPlugin_MyRecord' => new DataTable()]` |
| 477 | */ |
| 478 | protected abstract function aggregate(ArchiveProcessor $archiveProcessor) : array; |
| 479 | protected function insertBlobRecord(ArchiveProcessor $archiveProcessor, string $recordName, DataTable $record, ?int $maxRowsInTable, ?int $maxRowsInSubtable, ?string $columnToSortByBeforeTruncation) : void |
| 480 | { |
| 481 | $serialized = $record->getSerialized($maxRowsInTable ?? $this->maxRowsInTable, $maxRowsInSubtable ?? $this->maxRowsInSubtable, $columnToSortByBeforeTruncation ?? $this->columnToSortByBeforeTruncation); |
| 482 | $archiveProcessor->insertBlobRecord($recordName, $serialized); |
| 483 | unset($serialized); |
| 484 | } |
| 485 | public function getMaxRowsInTable() : ?int |
| 486 | { |
| 487 | return $this->maxRowsInTable; |
| 488 | } |
| 489 | public function getMaxRowsInSubtable() : ?int |
| 490 | { |
| 491 | return $this->maxRowsInSubtable; |
| 492 | } |
| 493 | public function getColumnToSortByBeforeTruncation() : ?string |
| 494 | { |
| 495 | return $this->columnToSortByBeforeTruncation; |
| 496 | } |
| 497 | public function getPluginName() : string |
| 498 | { |
| 499 | return Piwik::getPluginNameOfMatomoClass(get_class($this)); |
| 500 | } |
| 501 | /** |
| 502 | * Returns an extra hint for LogAggregator to add to log aggregation SQL. Can be overridden if you'd |
| 503 | * like the origin hint to have more information. |
| 504 | */ |
| 505 | public function getQueryOriginHint() : string |
| 506 | { |
| 507 | $recordBuilderName = get_class($this); |
| 508 | $recordBuilderName = explode('\\', $recordBuilderName); |
| 509 | return end($recordBuilderName); |
| 510 | } |
| 511 | /** |
| 512 | * Returns true if at least one of the given reports is handled by this RecordBuilder instance |
| 513 | * when invoked with the given ArchiveProcessor. |
| 514 | * |
| 515 | * @param ArchiveProcessor $archiveProcessor Archiving parameters, like idSite, can influence the list of |
| 516 | * all records a RecordBuilder produces, so it is required here. |
| 517 | * @param string[] $requestedReports The list of requested reports to check for. |
| 518 | */ |
| 519 | public function isBuilderForAtLeastOneOf(ArchiveProcessor $archiveProcessor, array $requestedReports) : bool |
| 520 | { |
| 521 | $recordMetadata = $this->getRecordMetadata($archiveProcessor); |
| 522 | foreach ($recordMetadata as $record) { |
| 523 | if (in_array($record->getName(), $requestedReports)) { |
| 524 | return \true; |
| 525 | } |
| 526 | } |
| 527 | return \false; |
| 528 | } |
| 529 | } |
| 530 |