AOEpeople /
crawler
| Total Complexity | 45 |
| Total Lines | 450 |
| Duplicated Lines | 0 % |
| Coverage | 38.02% |
| Changes | 1 | ||
| Bugs | 0 | Features | 0 |
Complex classes like CrawlerApi 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 CrawlerApi, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 46 | class CrawlerApi |
||
| 47 | { |
||
| 48 | /** |
||
| 49 | * @var QueueRepository |
||
| 50 | * @deprecated |
||
| 51 | */ |
||
| 52 | protected $queueRepository; |
||
| 53 | |||
| 54 | /** |
||
| 55 | * @var array |
||
| 56 | * @deprecated |
||
| 57 | */ |
||
| 58 | protected $allowedConfigurations = []; |
||
| 59 | |||
| 60 | /** |
||
| 61 | * @var QueryBuilder |
||
| 62 | * @deprecated |
||
| 63 | */ |
||
| 64 | protected $queryBuilder; |
||
| 65 | |||
| 66 | /** |
||
| 67 | * @var string |
||
| 68 | * @deprecated |
||
| 69 | */ |
||
| 70 | protected $tableName = 'tx_crawler_queue'; |
||
| 71 | |||
| 72 | /** |
||
| 73 | * @var CrawlerController |
||
| 74 | * @deprecated |
||
| 75 | */ |
||
| 76 | protected $crawlerController; |
||
| 77 | |||
| 78 | /** |
||
| 79 | * CrawlerApi constructor. |
||
| 80 | * @deprecated |
||
| 81 | */ |
||
| 82 | 9 | public function __construct() |
|
| 83 | { |
||
| 84 | 9 | trigger_error('The ' . __CLASS__ . ' is deprecated an will be removed in v10.x, most of functionality is moved to the respective Domain Model and Repository, and some to the CrawlerController.'); |
|
| 85 | /** @var ObjectManager $objectManager */ |
||
| 86 | 9 | $objectManager = GeneralUtility::makeInstance(ObjectManager::class); |
|
| 87 | 9 | $this->queueRepository = $objectManager->get(QueueRepository::class); |
|
|
Deprecated Code
introduced
by
|
|||
| 88 | 9 | } |
|
| 89 | |||
| 90 | /** |
||
| 91 | * Each crawler run has a setid, this facade method delegates |
||
| 92 | * the it to the crawler object |
||
| 93 | * |
||
| 94 | * @param int $id |
||
| 95 | * @throws \Exception |
||
| 96 | */ |
||
| 97 | 1 | public function overwriteSetId(int $id): void |
|
| 98 | { |
||
| 99 | 1 | $this->findCrawler()->setID = $id; |
|
| 100 | 1 | } |
|
| 101 | |||
| 102 | /** |
||
| 103 | * This method is used to limit the configuration selection to |
||
| 104 | * a set of configurations. |
||
| 105 | * |
||
| 106 | * @param array $allowedConfigurations |
||
| 107 | */ |
||
| 108 | 1 | public function setAllowedConfigurations(array $allowedConfigurations): void |
|
| 109 | { |
||
| 110 | 1 | $this->allowedConfigurations = $allowedConfigurations; |
|
| 111 | 1 | } |
|
| 112 | |||
| 113 | /** |
||
| 114 | * @return array |
||
| 115 | */ |
||
| 116 | 1 | public function getAllowedConfigurations() |
|
| 117 | { |
||
| 118 | 1 | return $this->allowedConfigurations; |
|
| 119 | } |
||
| 120 | |||
| 121 | /** |
||
| 122 | * Returns the setID of the crawler |
||
| 123 | * |
||
| 124 | * @return int |
||
| 125 | */ |
||
| 126 | 1 | public function getSetId() |
|
| 127 | { |
||
| 128 | 1 | return $this->findCrawler()->setID; |
|
| 129 | } |
||
| 130 | |||
| 131 | /** |
||
| 132 | * Method to get an instance of the internal crawler singleton |
||
| 133 | * |
||
| 134 | * @return CrawlerController Instance of the crawler lib |
||
| 135 | * |
||
| 136 | * @throws \Exception |
||
| 137 | */ |
||
| 138 | 2 | protected function findCrawler() |
|
| 139 | { |
||
| 140 | 2 | if (!is_object($this->crawlerController)) { |
|
| 141 | 2 | $this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
|
| 142 | 2 | $this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
|
| 143 | } |
||
| 144 | |||
| 145 | 2 | if (is_object($this->crawlerController)) { |
|
| 146 | 2 | return $this->crawlerController; |
|
| 147 | } else { |
||
| 148 | throw new \Exception('no crawler object', 1512659759); |
||
| 149 | } |
||
| 150 | } |
||
| 151 | |||
| 152 | /** |
||
| 153 | * Adds a page to the crawlerqueue by uid |
||
| 154 | * |
||
| 155 | * @param int $uid uid |
||
| 156 | */ |
||
| 157 | public function addPageToQueue($uid): void |
||
| 158 | { |
||
| 159 | $uid = intval($uid); |
||
| 160 | //non timed elements will be added with timestamp 0 |
||
| 161 | $this->addPageToQueueTimed($uid, 0); |
||
| 162 | } |
||
| 163 | |||
| 164 | /** |
||
| 165 | * This method is used to limit the processing instructions to the processing instructions |
||
| 166 | * that are allowed. |
||
| 167 | */ |
||
| 168 | 2 | protected function filterUnallowedConfigurations(array $configurations): array |
|
| 169 | { |
||
| 170 | 2 | if (count($this->allowedConfigurations) > 0) { |
|
| 171 | // remove configuration that does not match the current selection |
||
| 172 | foreach ($configurations as $confKey => $confArray) { |
||
| 173 | if (!in_array($confKey, $this->allowedConfigurations)) { |
||
| 174 | unset($configurations[$confKey]); |
||
| 175 | } |
||
| 176 | } |
||
| 177 | } |
||
| 178 | |||
| 179 | 2 | return $configurations; |
|
| 180 | } |
||
| 181 | |||
| 182 | /** |
||
| 183 | * Adds a page to the crawlerqueue by uid and sets a |
||
| 184 | * timestamp when the page should be crawled. |
||
| 185 | * |
||
| 186 | * @param int $uid pageid |
||
| 187 | * @param int $time timestamp |
||
| 188 | * |
||
| 189 | * @return void |
||
| 190 | * @throws \Exception |
||
| 191 | */ |
||
| 192 | 2 | public function addPageToQueueTimed($uid, $time): void |
|
| 193 | { |
||
| 194 | 2 | $uid = intval($uid); |
|
| 195 | 2 | $time = intval($time); |
|
| 196 | |||
| 197 | 2 | $crawler = $this->findCrawler(); |
|
| 198 | 2 | $pageData = GeneralUtility::makeInstance(PageRepository::class)->getPage($uid); |
|
| 199 | 2 | $configurations = $crawler->getUrlsForPageRow($pageData); |
|
| 200 | 2 | $configurations = $this->filterUnallowedConfigurations($configurations); |
|
| 201 | 2 | $downloadUrls = []; |
|
| 202 | 2 | $duplicateTrack = []; |
|
| 203 | |||
| 204 | 2 | if (is_array($configurations)) { |
|
| 205 | 2 | foreach ($configurations as $cv) { |
|
| 206 | //enable inserting of entries |
||
| 207 | 2 | $crawler->registerQueueEntriesInternallyOnly = false; |
|
| 208 | 2 | $crawler->urlListFromUrlArray( |
|
| 209 | 2 | $cv, |
|
| 210 | 2 | $pageData, |
|
| 211 | 2 | $time, |
|
| 212 | 2 | 300, |
|
| 213 | 2 | true, |
|
| 214 | 2 | false, |
|
| 215 | 2 | $duplicateTrack, |
|
| 216 | 2 | $downloadUrls, |
|
| 217 | 2 | array_keys($this->getCrawlerProcInstructions()) |
|
| 218 | ); |
||
| 219 | |||
| 220 | //reset the queue because the entries have been written to the db |
||
| 221 | 2 | unset($crawler->queueEntries); |
|
| 222 | } |
||
| 223 | } |
||
| 224 | 2 | } |
|
| 225 | |||
| 226 | /** |
||
| 227 | * Counts all entries in the database which are scheduled for a given page id and a schedule timestamp. |
||
| 228 | * |
||
| 229 | * @param int $page_uid |
||
| 230 | * @param int $schedule_timestamp |
||
| 231 | * |
||
| 232 | * @return int |
||
| 233 | */ |
||
| 234 | 1 | protected function countEntriesInQueueForPageByScheduleTime($page_uid, $schedule_timestamp) |
|
| 235 | { |
||
| 236 | 1 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName); |
|
| 237 | $count = $queryBuilder |
||
| 238 | 1 | ->count('*') |
|
| 239 | 1 | ->from($this->tableName); |
|
| 240 | |||
| 241 | //if the same page is scheduled for the same time and has not be executed? |
||
| 242 | //un-timed elements need an exec_time with 0 because they can occur multiple times |
||
| 243 | 1 | if ($schedule_timestamp == 0) { |
|
| 244 | 1 | $count->where( |
|
| 245 | 1 | $queryBuilder->expr()->eq('page_id', $page_uid), |
|
| 246 | 1 | $queryBuilder->expr()->eq('exec_time', 0), |
|
| 247 | 1 | $queryBuilder->expr()->eq('scheduled', $schedule_timestamp) |
|
| 248 | ); |
||
| 249 | } else { |
||
| 250 | //timed elements have got a fixed schedule time, if a record with this time |
||
| 251 | //exists it is maybe queued for the future, or is has been queue for the past and therefore |
||
| 252 | //also been processed. |
||
| 253 | 1 | $count->where( |
|
| 254 | 1 | $queryBuilder->expr()->eq('page_id', $page_uid), |
|
| 255 | 1 | $queryBuilder->expr()->eq('scheduled', $schedule_timestamp) |
|
| 256 | ); |
||
| 257 | } |
||
| 258 | |||
| 259 | 1 | return $count->execute()->rowCount(); |
|
| 260 | } |
||
| 261 | |||
| 262 | /** |
||
| 263 | * Method to return the latest Crawle Timestamp for a page. |
||
| 264 | * |
||
| 265 | * @param int $uid uid id of the page |
||
| 266 | * @param bool $future_crawldates_only |
||
| 267 | * @param bool $unprocessed_only |
||
| 268 | * |
||
| 269 | * @return int |
||
| 270 | */ |
||
| 271 | 1 | public function getLatestCrawlTimestampForPage($uid, $future_crawldates_only = false, $unprocessed_only = false) |
|
| 272 | { |
||
| 273 | 1 | $uid = intval($uid); |
|
| 274 | |||
| 275 | 1 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName); |
|
| 276 | $query = $queryBuilder |
||
| 277 | 1 | ->from($this->tableName) |
|
| 278 | 1 | ->selectLiteral('max(scheduled) as latest') |
|
| 279 | 1 | ->where( |
|
| 280 | 1 | $queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid)) |
|
| 281 | ); |
||
| 282 | |||
| 283 | 1 | if ($future_crawldates_only) { |
|
| 284 | $query->andWhere( |
||
| 285 | $queryBuilder->expr()->gt('scheduled', time()) |
||
| 286 | ); |
||
| 287 | } |
||
| 288 | |||
| 289 | 1 | if ($unprocessed_only) { |
|
| 290 | $query->andWhere( |
||
| 291 | $queryBuilder->expr()->eq('exec_time', 0) |
||
| 292 | ); |
||
| 293 | } |
||
| 294 | |||
| 295 | 1 | $row = $query->execute()->fetch(0); |
|
| 296 | 1 | if ($row['latest']) { |
|
| 297 | 1 | $res = $row['latest']; |
|
| 298 | } else { |
||
| 299 | $res = 0; |
||
| 300 | } |
||
| 301 | |||
| 302 | 1 | return intval($res); |
|
| 303 | } |
||
| 304 | |||
| 305 | /** |
||
| 306 | * Returns an array with timestamps when the page has been scheduled for crawling and |
||
| 307 | * at what time the scheduled crawl has been executed. The array also contains items that are |
||
| 308 | * scheduled but have note been crawled yet. |
||
| 309 | * |
||
| 310 | * @return array array with the crawl-history of a page => 0 : scheduled time , 1 : executed_time, 2 : set_id |
||
| 311 | */ |
||
| 312 | 1 | public function getCrawlHistoryForPage(int $uid, int $limit = 0) |
|
| 313 | { |
||
| 314 | 1 | $uid = intval($uid); |
|
| 315 | 1 | $limit = intval($limit); |
|
| 316 | |||
| 317 | 1 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName); |
|
| 318 | $statement = $queryBuilder |
||
| 319 | 1 | ->from($this->tableName) |
|
| 320 | 1 | ->select('scheduled', 'exec_time', 'set_id') |
|
| 321 | 1 | ->where( |
|
| 322 | 1 | $queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid, \PDO::PARAM_INT)) |
|
| 323 | ); |
||
| 324 | 1 | if ($limit) { |
|
| 325 | 1 | $statement->setMaxResults($limit); |
|
| 326 | } |
||
| 327 | |||
| 328 | 1 | return $statement->execute()->fetchAll(); |
|
| 329 | } |
||
| 330 | |||
| 331 | /** |
||
| 332 | * Reads the registered processingInstructions of the crawler |
||
| 333 | * |
||
| 334 | * @return array |
||
| 335 | */ |
||
| 336 | 2 | private function getCrawlerProcInstructions(): array |
|
| 337 | { |
||
| 338 | 2 | $crawlerProcInstructions = []; |
|
| 339 | 2 | if (!empty($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'])) { |
|
| 340 | foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'] as $configuration) { |
||
| 341 | $crawlerProcInstructions[$configuration['key']] = $configuration['value']; |
||
| 342 | } |
||
| 343 | } |
||
| 344 | |||
| 345 | 2 | return $crawlerProcInstructions; |
|
| 346 | } |
||
| 347 | |||
| 348 | /** |
||
| 349 | * Get queue statistics |
||
| 350 | */ |
||
| 351 | 1 | public function getQueueStatistics(): array |
|
| 352 | { |
||
| 353 | return [ |
||
| 354 | 1 | 'assignedButUnprocessed' => $this->queueRepository->countAllAssignedPendingItems(), |
|
| 355 | 1 | 'unprocessed' => $this->queueRepository->countAllPendingItems(), |
|
| 356 | ]; |
||
| 357 | } |
||
| 358 | |||
| 359 | /** |
||
| 360 | * Get queue statistics by configuration |
||
| 361 | * |
||
| 362 | * @return array array of array('configuration' => <>, 'assignedButUnprocessed' => <>, 'unprocessed' => <>) |
||
| 363 | */ |
||
| 364 | public function getQueueStatisticsByConfiguration() |
||
| 365 | { |
||
| 366 | $statistics = $this->queueRepository->countPendingItemsGroupedByConfigurationKey(); |
||
| 367 | $setIds = $this->queueRepository->getSetIdWithUnprocessedEntries(); |
||
| 368 | $totals = $this->queueRepository->getTotalQueueEntriesByConfiguration($setIds); |
||
| 369 | |||
| 370 | // "merge" arrays |
||
| 371 | foreach ($statistics as &$value) { |
||
| 372 | $value['total'] = $totals[$value['configuration']]; |
||
| 373 | } |
||
| 374 | |||
| 375 | return $statistics; |
||
| 376 | } |
||
| 377 | |||
| 378 | /** |
||
| 379 | * Get active processes count |
||
| 380 | */ |
||
| 381 | public function getActiveProcessesCount(): int |
||
| 382 | { |
||
| 383 | $processRepository = new ProcessRepository(); |
||
| 384 | |||
| 385 | return $processRepository->countActive(); |
||
| 386 | } |
||
| 387 | |||
| 388 | /** |
||
| 389 | * @param int $limit |
||
| 390 | * @return array |
||
| 391 | */ |
||
| 392 | public function getLastProcessedQueueEntries(int $limit): array |
||
| 395 | } |
||
| 396 | |||
| 397 | /** |
||
| 398 | * Get current crawling speed |
||
| 399 | * |
||
| 400 | * @return int|float|bool |
||
| 401 | */ |
||
| 402 | public function getCurrentCrawlingSpeed() |
||
| 435 | } |
||
| 436 | |||
| 437 | /** |
||
| 438 | * Get some performance data |
||
| 439 | * |
||
| 440 | * @param integer $start |
||
| 441 | * @param integer $end |
||
| 442 | * @param integer $resolution |
||
| 443 | * |
||
| 444 | * @return array data |
||
| 445 | * |
||
| 446 | * @throws \Exception |
||
| 447 | */ |
||
| 448 | public function getPerformanceData($start, $end, $resolution) |
||
| 496 | } |
||
| 497 | } |
||
| 498 |
Loading history...
