NNTmux /
newznab-tmux
| Total Complexity | 70 |
| Total Lines | 603 |
| Duplicated Lines | 0 % |
| Changes | 7 | ||
| Bugs | 3 | Features | 0 |
Complex classes like NntmuxPopulateSearchIndexes often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use NntmuxPopulateSearchIndexes, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 13 | class NntmuxPopulateSearchIndexes extends Command |
||
| 14 | { |
||
| 15 | /** |
||
| 16 | * The name and signature of the console command. |
||
| 17 | * |
||
| 18 | * @var string |
||
| 19 | */ |
||
| 20 | protected $signature = 'nntmux:populate |
||
| 21 | {--manticore : Use ManticoreSearch} |
||
| 22 | {--elastic : Use ElasticSearch} |
||
| 23 | {--releases : Populates the releases index} |
||
| 24 | {--predb : Populates the predb index} |
||
| 25 | {--count=50000 : Sets the chunk size} |
||
| 26 | {--parallel=4 : Number of parallel processes} |
||
| 27 | {--batch-size=5000 : Batch size for bulk operations} |
||
| 28 | {--disable-keys : Disable database keys during population} |
||
| 29 | {--optimize : Optimize ManticoreSearch indexes}'; |
||
| 30 | |||
| 31 | /** |
||
| 32 | * The console command description. |
||
| 33 | * |
||
| 34 | * @var string |
||
| 35 | */ |
||
| 36 | protected $description = 'Populate Manticore/Elasticsearch indexes with either releases or predb'; |
||
| 37 | |||
| 38 | private const SUPPORTED_ENGINES = ['manticore', 'elastic']; |
||
| 39 | |||
| 40 | private const SUPPORTED_INDEXES = ['releases', 'predb']; |
||
| 41 | |||
| 42 | private const GROUP_CONCAT_MAX_LEN = 16384; |
||
| 43 | |||
| 44 | private const DEFAULT_CHUNK_SIZE = 50000; |
||
| 45 | |||
| 46 | private const DEFAULT_PARALLEL_PROCESSES = 4; |
||
| 47 | |||
| 48 | private const DEFAULT_BATCH_SIZE = 5000; |
||
| 49 | |||
| 50 | /** |
||
| 51 | * Execute the console command. |
||
| 52 | */ |
||
| 53 | public function handle(): int |
||
| 54 | { |
||
| 55 | try { |
||
| 56 | if ($this->option('optimize')) { |
||
| 57 | return $this->handleOptimize(); |
||
| 58 | } |
||
| 59 | |||
| 60 | $engine = $this->getSelectedEngine(); |
||
| 61 | $index = $this->getSelectedIndex(); |
||
| 62 | |||
| 63 | if (! $engine || ! $index) { |
||
| 64 | $this->error('You must specify both an engine (--manticore or --elastic) and an index (--releases or --predb).'); |
||
| 65 | $this->info('Use --help to see all available options.'); |
||
| 66 | |||
| 67 | return Command::FAILURE; |
||
| 68 | } |
||
| 69 | |||
| 70 | return $this->populateIndex($engine, $index); |
||
| 71 | |||
| 72 | } catch (Exception $e) { |
||
| 73 | $this->error("An error occurred: {$e->getMessage()}"); |
||
| 74 | |||
| 75 | if ($this->output->isVerbose()) { |
||
| 76 | $this->error($e->getTraceAsString()); |
||
| 77 | } |
||
| 78 | |||
| 79 | return Command::FAILURE; |
||
| 80 | } |
||
| 81 | } |
||
| 82 | |||
| 83 | /** |
||
| 84 | * Get the selected search engine from options |
||
| 85 | */ |
||
| 86 | private function getSelectedEngine(): ?string |
||
| 87 | { |
||
| 88 | foreach (self::SUPPORTED_ENGINES as $engine) { |
||
| 89 | if ($this->option($engine)) { |
||
| 90 | return $engine; |
||
| 91 | } |
||
| 92 | } |
||
| 93 | |||
| 94 | return null; |
||
| 95 | } |
||
| 96 | |||
| 97 | /** |
||
| 98 | * Get the selected index from options |
||
| 99 | */ |
||
| 100 | private function getSelectedIndex(): ?string |
||
| 101 | { |
||
| 102 | foreach (self::SUPPORTED_INDEXES as $index) { |
||
| 103 | if ($this->option($index)) { |
||
| 104 | return $index; |
||
| 105 | } |
||
| 106 | } |
||
| 107 | |||
| 108 | return null; |
||
| 109 | } |
||
| 110 | |||
| 111 | /** |
||
| 112 | * Handle the optimize command |
||
| 113 | */ |
||
| 114 | private function handleOptimize(): int |
||
| 115 | { |
||
| 116 | $this->info('Optimizing ManticoreSearch indexes...'); |
||
| 117 | |||
| 118 | try { |
||
| 119 | (new ManticoreSearch)->optimizeRTIndex(); |
||
| 120 | $this->info('Optimization completed successfully!'); |
||
| 121 | |||
| 122 | return Command::SUCCESS; |
||
| 123 | } catch (Exception $e) { |
||
| 124 | $this->error("Optimization failed: {$e->getMessage()}"); |
||
| 125 | |||
| 126 | return Command::FAILURE; |
||
| 127 | } |
||
| 128 | } |
||
| 129 | |||
| 130 | /** |
||
| 131 | * Populate the specified index with the specified engine |
||
| 132 | */ |
||
| 133 | private function populateIndex(string $engine, string $index): int |
||
| 134 | { |
||
| 135 | $methodName = "{$engine}".ucfirst($index); |
||
| 136 | |||
| 137 | if (! method_exists($this, $methodName)) { |
||
| 138 | $this->error("Method {$methodName} not implemented."); |
||
| 139 | |||
| 140 | return Command::FAILURE; |
||
| 141 | } |
||
| 142 | |||
| 143 | $this->info("Starting {$engine} {$index} population..."); |
||
| 144 | |||
| 145 | $startTime = microtime(true); |
||
| 146 | $result = $this->{$methodName}(); |
||
| 147 | $executionTime = round(microtime(true) - $startTime, 2); |
||
| 148 | |||
| 149 | if ($result === Command::SUCCESS) { |
||
| 150 | $this->info("Population completed in {$executionTime} seconds."); |
||
| 151 | } |
||
| 152 | |||
| 153 | return $result; |
||
| 154 | } |
||
| 155 | |||
| 156 | private function manticoreReleases(): int |
||
| 157 | { |
||
| 158 | $manticore = new ManticoreSearch; |
||
| 159 | $indexName = 'releases_rt'; |
||
| 160 | |||
| 161 | $manticore->truncateRTIndex(Arr::wrap($indexName)); |
||
| 162 | |||
| 163 | $total = Release::count(); |
||
| 164 | if (! $total) { |
||
| 165 | $this->warn('Releases table is empty. Nothing to do.'); |
||
| 166 | |||
| 167 | return Command::SUCCESS; |
||
| 168 | } |
||
| 169 | |||
| 170 | $query = Release::query() |
||
| 171 | ->orderByDesc('releases.id') |
||
| 172 | ->leftJoin('release_files', 'releases.id', '=', 'release_files.releases_id') |
||
| 173 | ->select([ |
||
| 174 | 'releases.id', |
||
| 175 | 'releases.name', |
||
| 176 | 'releases.searchname', |
||
| 177 | 'releases.fromname', |
||
| 178 | 'releases.categories_id', |
||
| 179 | ]) |
||
| 180 | ->selectRaw('IFNULL(GROUP_CONCAT(release_files.name SEPARATOR " "),"") AS filename') |
||
| 181 | ->groupBy([ |
||
| 182 | 'releases.id', |
||
| 183 | 'releases.name', |
||
| 184 | 'releases.searchname', |
||
| 185 | 'releases.fromname', |
||
| 186 | 'releases.categories_id', |
||
| 187 | ]); |
||
| 188 | |||
| 189 | return $this->processManticoreData( |
||
| 190 | $indexName, |
||
| 191 | $total, |
||
|
Bug
introduced
by
|
|||
| 192 | $query, |
||
| 193 | function ($item) { |
||
| 194 | return [ |
||
| 195 | 'id' => (string) $item->id, |
||
| 196 | 'name' => (string) ($item->name ?: ''), |
||
| 197 | 'searchname' => (string) ($item->searchname ?: ''), |
||
| 198 | 'fromname' => (string) ($item->fromname ?: ''), |
||
| 199 | 'categories_id' => (string) ($item->categories_id ?: '0'), |
||
| 200 | 'filename' => (string) ($item->filename ?: ''), |
||
| 201 | 'dummy' => '1', |
||
| 202 | ]; |
||
| 203 | } |
||
| 204 | ); |
||
| 205 | } |
||
| 206 | |||
| 207 | /** |
||
| 208 | * Populate ManticoreSearch predb index |
||
| 209 | */ |
||
| 210 | private function manticorePredb(): int |
||
| 211 | { |
||
| 212 | $manticore = new ManticoreSearch; |
||
| 213 | $indexName = 'predb_rt'; |
||
| 214 | |||
| 215 | $manticore->truncateRTIndex([$indexName]); |
||
| 216 | |||
| 217 | $total = Predb::count(); |
||
| 218 | if (! $total) { |
||
| 219 | $this->warn('PreDB table is empty. Nothing to do.'); |
||
| 220 | |||
| 221 | return Command::SUCCESS; |
||
| 222 | } |
||
| 223 | |||
| 224 | $query = Predb::query() |
||
| 225 | ->select(['id', 'title', 'filename', 'source']) |
||
| 226 | ->orderBy('id'); |
||
| 227 | |||
| 228 | return $this->processManticoreData( |
||
| 229 | $indexName, |
||
| 230 | $total, |
||
| 231 | $query, |
||
| 232 | function ($item) { |
||
| 233 | return [ |
||
| 234 | 'id' => $item->id, |
||
| 235 | 'title' => (string) ($item->title ?? ''), |
||
| 236 | 'filename' => (string) ($item->filename ?? ''), |
||
| 237 | 'source' => (string) ($item->source ?? ''), |
||
| 238 | 'dummy' => 1, |
||
| 239 | ]; |
||
| 240 | } |
||
| 241 | ); |
||
| 242 | } |
||
| 243 | |||
| 244 | /** |
||
| 245 | * Process data for ManticoreSearch with optimizations |
||
| 246 | */ |
||
| 247 | private function processManticoreData(string $indexName, int $total, $query, callable $transformer): int |
||
| 248 | { |
||
| 249 | $manticore = new ManticoreSearch; |
||
| 250 | $chunkSize = $this->getChunkSize(); |
||
| 251 | $batchSize = $this->getBatchSize(); |
||
| 252 | |||
| 253 | $this->optimizeDatabase(); |
||
| 254 | $this->setGroupConcatMaxLen(); |
||
| 255 | |||
| 256 | $this->info(sprintf( |
||
| 257 | "Populating ManticoreSearch index '%s' with %s rows using chunks of %s and batch size of %s.", |
||
| 258 | $indexName, |
||
| 259 | number_format($total), |
||
| 260 | number_format($chunkSize), |
||
| 261 | number_format($batchSize) |
||
| 262 | )); |
||
| 263 | |||
| 264 | $bar = $this->output->createProgressBar($total); |
||
| 265 | $bar->setFormat('verbose'); |
||
| 266 | $bar->start(); |
||
| 267 | |||
| 268 | $processedCount = 0; |
||
| 269 | $errorCount = 0; |
||
| 270 | $batchData = []; |
||
| 271 | |||
| 272 | try { |
||
| 273 | $query->chunk($chunkSize, function ($items) use ($manticore, $indexName, $transformer, $bar, &$processedCount, &$errorCount, $batchSize, &$batchData) { |
||
| 274 | foreach ($items as $item) { |
||
| 275 | try { |
||
| 276 | $batchData[] = $transformer($item); |
||
| 277 | $processedCount++; |
||
| 278 | |||
| 279 | // Process in optimized batch sizes |
||
| 280 | if (count($batchData) >= $batchSize) { |
||
| 281 | $this->processBatch($manticore, $indexName, $batchData); |
||
| 282 | $batchData = []; |
||
| 283 | } |
||
| 284 | } catch (Exception $e) { |
||
| 285 | $errorCount++; |
||
| 286 | if ($this->output->isVerbose()) { |
||
| 287 | $this->error("Error processing item {$item->id}: {$e->getMessage()}"); |
||
| 288 | } |
||
| 289 | } |
||
| 290 | $bar->advance(); |
||
| 291 | } |
||
| 292 | }); |
||
| 293 | |||
| 294 | // Process remaining items |
||
| 295 | if (! empty($batchData)) { |
||
| 296 | $this->processBatch($manticore, $indexName, $batchData); |
||
| 297 | } |
||
| 298 | |||
| 299 | $bar->finish(); |
||
| 300 | $this->newLine(); |
||
| 301 | |||
| 302 | if ($errorCount > 0) { |
||
| 303 | $this->warn("Completed with {$errorCount} errors out of {$processedCount} processed items."); |
||
| 304 | } else { |
||
| 305 | $this->info('ManticoreSearch population completed successfully!'); |
||
| 306 | } |
||
| 307 | |||
| 308 | return Command::SUCCESS; |
||
| 309 | |||
| 310 | } catch (Exception $e) { |
||
| 311 | $bar->finish(); |
||
| 312 | $this->newLine(); |
||
| 313 | $this->error("Failed to populate ManticoreSearch: {$e->getMessage()}"); |
||
| 314 | |||
| 315 | return Command::FAILURE; |
||
| 316 | } finally { |
||
| 317 | $this->restoreDatabase(); |
||
| 318 | } |
||
| 319 | } |
||
| 320 | |||
| 321 | /** |
||
| 322 | * Populate ElasticSearch releases index |
||
| 323 | */ |
||
| 324 | private function elasticReleases(): int |
||
| 325 | { |
||
| 326 | $total = Release::count(); |
||
| 327 | if (! $total) { |
||
| 328 | $this->warn('Releases table is empty. Nothing to do.'); |
||
| 329 | |||
| 330 | return Command::SUCCESS; |
||
| 331 | } |
||
| 332 | |||
| 333 | $query = Release::query() |
||
| 334 | ->orderByDesc('releases.id') |
||
| 335 | ->leftJoin('release_files', 'releases.id', '=', 'release_files.releases_id') |
||
| 336 | ->select([ |
||
| 337 | 'releases.id', |
||
| 338 | 'releases.name', |
||
| 339 | 'releases.searchname', |
||
| 340 | 'releases.fromname', |
||
| 341 | 'releases.categories_id', |
||
| 342 | 'releases.postdate', |
||
| 343 | ]) |
||
| 344 | ->selectRaw('IFNULL(GROUP_CONCAT(release_files.name SEPARATOR " "),"") AS filename') |
||
| 345 | ->groupBy([ |
||
| 346 | 'releases.id', |
||
| 347 | 'releases.name', |
||
| 348 | 'releases.searchname', |
||
| 349 | 'releases.fromname', |
||
| 350 | 'releases.categories_id', |
||
| 351 | 'releases.postdate', |
||
| 352 | ]); |
||
| 353 | |||
| 354 | return $this->processElasticData( |
||
| 355 | 'releases', |
||
| 356 | $total, |
||
| 357 | $query, |
||
| 358 | function ($item) { |
||
| 359 | $searchName = str_replace(['.', '-'], ' ', $item->searchname ?? ''); |
||
| 360 | |||
| 361 | return [ |
||
| 362 | 'id' => $item->id, |
||
| 363 | 'name' => $item->name, |
||
| 364 | 'searchname' => $item->searchname, |
||
| 365 | 'plainsearchname' => $searchName, |
||
| 366 | 'fromname' => $item->fromname, |
||
| 367 | 'categories_id' => $item->categories_id, |
||
| 368 | 'filename' => $item->filename ?? '', |
||
| 369 | 'postdate' => $item->postdate, |
||
| 370 | ]; |
||
| 371 | } |
||
| 372 | ); |
||
| 373 | } |
||
| 374 | |||
| 375 | /** |
||
| 376 | * Populate ElasticSearch predb index |
||
| 377 | */ |
||
| 378 | private function elasticPredb(): int |
||
| 379 | { |
||
| 380 | $total = Predb::count(); |
||
| 381 | if (! $total) { |
||
| 382 | $this->warn('PreDB table is empty. Nothing to do.'); |
||
| 383 | |||
| 384 | return Command::SUCCESS; |
||
| 385 | } |
||
| 386 | |||
| 387 | $query = Predb::query() |
||
| 388 | ->select(['id', 'title', 'filename', 'source']) |
||
| 389 | ->orderBy('id'); |
||
| 390 | |||
| 391 | return $this->processElasticData( |
||
| 392 | 'predb', |
||
| 393 | $total, |
||
| 394 | $query, |
||
| 395 | function ($item) { |
||
| 396 | return [ |
||
| 397 | 'id' => $item->id, |
||
| 398 | 'title' => $item->title, |
||
| 399 | 'filename' => $item->filename, |
||
| 400 | 'source' => $item->source, |
||
| 401 | ]; |
||
| 402 | } |
||
| 403 | ); |
||
| 404 | } |
||
| 405 | |||
| 406 | /** |
||
| 407 | * Process data for ElasticSearch with optimizations |
||
| 408 | */ |
||
| 409 | private function processElasticData(string $indexName, int $total, $query, callable $transformer): int |
||
| 486 | } |
||
| 487 | } |
||
| 488 | |||
| 489 | /** |
||
| 490 | * Process ManticoreSearch batch with retry logic |
||
| 491 | */ |
||
| 492 | private function processBatch(ManticoreSearch $manticore, string $indexName, array $data): void |
||
| 493 | { |
||
| 494 | $retries = 3; |
||
| 495 | $attempt = 0; |
||
| 496 | |||
| 497 | while ($attempt < $retries) { |
||
| 498 | try { |
||
| 499 | $manticore->manticoreSearch->table($indexName)->replaceDocuments($data); |
||
| 500 | break; |
||
| 501 | } catch (Exception $e) { |
||
| 502 | $attempt++; |
||
| 503 | if ($attempt >= $retries) { |
||
| 504 | throw $e; |
||
| 505 | } |
||
| 506 | usleep(100000); // 100ms delay before retry |
||
| 507 | } |
||
| 508 | } |
||
| 509 | } |
||
| 510 | |||
| 511 | /** |
||
| 512 | * Process ElasticSearch batch with retry logic |
||
| 513 | */ |
||
| 514 | private function processElasticBatch(array $data, int &$errorCount): void |
||
| 515 | { |
||
| 516 | $retries = 3; |
||
| 517 | $attempt = 0; |
||
| 518 | |||
| 519 | while ($attempt < $retries) { |
||
| 520 | try { |
||
| 521 | $response = \Elasticsearch::bulk($data); |
||
| 522 | |||
| 523 | // Check for errors in bulk response |
||
| 524 | if (isset($response['errors']) && $response['errors']) { |
||
| 525 | foreach ($response['items'] as $item) { |
||
| 526 | if (isset($item['index']['error'])) { |
||
| 527 | $errorCount++; |
||
| 528 | if ($this->output->isVerbose()) { |
||
| 529 | $this->error('ElasticSearch error: '.json_encode($item['index']['error'])); |
||
| 530 | } |
||
| 531 | } |
||
| 532 | } |
||
| 533 | } |
||
| 534 | break; |
||
| 535 | } catch (Exception $e) { |
||
| 536 | $attempt++; |
||
| 537 | if ($attempt >= $retries) { |
||
| 538 | throw $e; |
||
| 539 | } |
||
| 540 | usleep(100000); // 100ms delay before retry |
||
| 541 | } |
||
| 542 | } |
||
| 543 | } |
||
| 544 | |||
| 545 | /** |
||
| 546 | * Optimize database settings for bulk operations |
||
| 547 | */ |
||
| 548 | private function optimizeDatabase(): void |
||
| 549 | { |
||
| 550 | if ($this->option('disable-keys')) { |
||
| 551 | $this->info('Disabling database keys for faster bulk operations...'); |
||
| 552 | |||
| 553 | try { |
||
| 554 | // Disable foreign key checks |
||
| 555 | DB::statement('SET FOREIGN_KEY_CHECKS = 0'); |
||
| 556 | DB::statement('SET UNIQUE_CHECKS = 0'); |
||
| 557 | DB::statement('SET AUTOCOMMIT = 0'); |
||
| 558 | |||
| 559 | // Increase buffer sizes |
||
| 560 | DB::statement('SET SESSION innodb_buffer_pool_size = 1073741824'); // 1GB |
||
| 561 | DB::statement('SET SESSION bulk_insert_buffer_size = 268435456'); // 256MB |
||
| 562 | DB::statement('SET SESSION read_buffer_size = 2097152'); // 2MB |
||
| 563 | DB::statement('SET SESSION sort_buffer_size = 16777216'); // 16MB |
||
| 564 | |||
| 565 | } catch (Exception $e) { |
||
| 566 | $this->warn("Could not optimize database settings: {$e->getMessage()}"); |
||
| 567 | } |
||
| 568 | } |
||
| 569 | } |
||
| 570 | |||
| 571 | /** |
||
| 572 | * Restore database settings after bulk operations |
||
| 573 | */ |
||
| 574 | private function restoreDatabase(): void |
||
| 575 | { |
||
| 576 | if ($this->option('disable-keys')) { |
||
| 577 | $this->info('Restoring database settings...'); |
||
| 578 | |||
| 579 | try { |
||
| 580 | DB::statement('SET FOREIGN_KEY_CHECKS = 1'); |
||
| 581 | DB::statement('SET UNIQUE_CHECKS = 1'); |
||
| 582 | DB::statement('SET AUTOCOMMIT = 1'); |
||
| 583 | DB::statement('COMMIT'); |
||
| 584 | } catch (Exception $e) { |
||
| 585 | $this->warn("Could not restore database settings: {$e->getMessage()}"); |
||
| 586 | } |
||
| 587 | } |
||
| 588 | } |
||
| 589 | |||
| 590 | /** |
||
| 591 | * Get the chunk size from options |
||
| 592 | */ |
||
| 593 | private function getChunkSize(): int |
||
| 598 | } |
||
| 599 | |||
| 600 | /** |
||
| 601 | * Get the batch size from options |
||
| 602 | */ |
||
| 603 | private function getBatchSize(): int |
||
| 608 | } |
||
| 609 | |||
| 610 | /** |
||
| 611 | * Set the GROUP_CONCAT max length for the session |
||
| 612 | */ |
||
| 613 | private function setGroupConcatMaxLen(): void |
||
| 616 | } |
||
| 617 | } |
||
| 618 |
Loading history...