wikimedia /
eventmetrics
| Total Complexity | 57 |
| Total Lines | 680 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like EventWikiRepository 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 EventWikiRepository, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 20 | class EventWikiRepository extends Repository |
||
| 21 | { |
||
| 22 | /** Max number of pages to be returned when generating page IDs. */ |
||
| 23 | public const MAX_PAGES = 50000; |
||
| 24 | |||
| 25 | /** |
||
| 26 | * Class name of associated entity. |
||
| 27 | * Implements Repository::getEntityClass |
||
| 28 | * @return string |
||
| 29 | */ |
||
| 30 | public function getEntityClass(): string |
||
| 31 | { |
||
| 32 | return EventWiki::class; |
||
| 33 | } |
||
| 34 | |||
| 35 | /** |
||
| 36 | * Get the wiki's domain name without the .org given a database name or domain. |
||
| 37 | * @param string $value |
||
| 38 | * @return string|null Null if no wiki was found. |
||
| 39 | */ |
||
| 40 | public function getDomainFromEventWikiInput(string $value): ?string |
||
| 41 | { |
||
| 42 | if ('*.' === substr($value, 0, 2)) { |
||
| 43 | $ret = $this->getWikiFamilyName(substr($value, 2)); |
||
| 44 | return null !== $ret ? '*.'.$ret : null; |
||
| 45 | } |
||
| 46 | |||
| 47 | $conn = $this->getMetaConnection(); |
||
| 48 | $rqb = $conn->createQueryBuilder(); |
||
| 49 | $rqb->select(['dbname, url']) |
||
| 50 | ->from('wiki') |
||
| 51 | ->where($rqb->expr()->eq('dbname', ':project')) |
||
| 52 | ->orWhere($rqb->expr()->like('url', ':projectUrl')) |
||
| 53 | ->orWhere($rqb->expr()->like('url', ':projectUrl2')) |
||
| 54 | ->setParameter('project', str_replace('_p', '', $value)) |
||
| 55 | ->setParameter('projectUrl', "https://$value") |
||
| 56 | ->setParameter('projectUrl2', "https://$value.org"); |
||
| 57 | $ret = $this->executeQueryBuilder($rqb)->fetch(); |
||
| 58 | |||
| 59 | // No matches found. |
||
| 60 | if (!$ret) { |
||
| 61 | return null; |
||
| 62 | } |
||
| 63 | |||
| 64 | // Extract and return just the domain name without '.org' suffix. |
||
| 65 | $matches = []; |
||
| 66 | preg_match('/^https?\:\/\/(.*)\.org$/', $ret['url'], $matches); |
||
| 67 | if (isset($matches[1])) { |
||
| 68 | return $matches[1]; |
||
| 69 | } else { |
||
| 70 | return null; |
||
| 71 | } |
||
| 72 | } |
||
| 73 | |||
| 74 | /** |
||
| 75 | * This effectively validates the given name as a wiki family |
||
| 76 | * (wikipedia, wiktionary, etc). Null is returned if invalid. |
||
| 77 | * @param string $value The wiki family name. |
||
| 78 | * @return string|null The wiki family name, or null if invalid. |
||
| 79 | */ |
||
| 80 | public function getWikiFamilyName(string $value): ?string |
||
| 81 | { |
||
| 82 | $conn = $this->getMetaConnection(); |
||
| 83 | $rqb = $conn->createQueryBuilder(); |
||
| 84 | $rqb->select(['family']) |
||
| 85 | ->from('wiki') |
||
| 86 | ->where($rqb->expr()->eq('family', ':family')) |
||
| 87 | ->setParameter('family', $value); |
||
| 88 | return $this->executeQueryBuilder($rqb)->fetch()['family']; |
||
| 89 | } |
||
| 90 | |||
| 91 | /** |
||
| 92 | * Get the database name of the given (partial) domain name. |
||
| 93 | * @param string $domain The domain name, without trailing '.org'. |
||
| 94 | * @return string Null if not found. |
||
| 95 | */ |
||
| 96 | public function getDbNameFromDomain(string $domain): string |
||
| 97 | { |
||
| 98 | $projectUrl = "https://$domain.org"; |
||
| 99 | |||
| 100 | $conn = $this->getMetaConnection(); |
||
| 101 | $rqb = $conn->createQueryBuilder(); |
||
| 102 | $rqb->select(["CONCAT(dbname, '_p') AS dbname"]) |
||
| 103 | ->from('wiki') |
||
| 104 | ->where('url = :projectUrl') |
||
| 105 | ->setParameter('projectUrl', $projectUrl); |
||
| 106 | |||
| 107 | $row = $this->executeQueryBuilder($rqb)->fetch(); |
||
| 108 | if (!isset($row['dbname'])) { |
||
| 109 | throw new Exception("Unable to determine database name for domain '$domain'."); |
||
| 110 | } |
||
| 111 | return $row['dbname']; |
||
| 112 | } |
||
| 113 | |||
| 114 | /** |
||
| 115 | * Public static method to convert wikitext to HTML, can be used on any arbitrary string. |
||
| 116 | * Does NOT support section links unless you specify a page. |
||
| 117 | * @param string $wikitext |
||
| 118 | * @param string $domain The project domain such as en.wikipedia |
||
| 119 | * @param string $pageTitle The title of the page, including namespace. |
||
| 120 | * @return string |
||
| 121 | * @static |
||
| 122 | */ |
||
| 123 | public static function wikifyString(string $wikitext, string $domain, ?string $pageTitle = null): string |
||
| 124 | { |
||
| 125 | $wikitext = htmlspecialchars(html_entity_decode($wikitext), ENT_NOQUOTES); |
||
| 126 | $sectionMatch = null; |
||
| 127 | $isSection = preg_match_all("/^\/\* (.*?) \*\//", $wikitext, $sectionMatch); |
||
| 128 | $pagePath = "https://$domain.org/wiki/"; |
||
| 129 | |||
| 130 | if ($isSection && isset($pageTitle)) { |
||
| 131 | $pageUrl = $pagePath.ucfirst(str_replace(' ', '_', $pageTitle)); |
||
| 132 | $sectionTitle = $sectionMatch[1][0]; |
||
| 133 | |||
| 134 | // Must have underscores for the link to properly go to the section. |
||
| 135 | $sectionTitleLink = htmlspecialchars(str_replace(' ', '_', $sectionTitle)); |
||
| 136 | |||
| 137 | $sectionWikitext = "<a target='_blank' href='$pageUrl#$sectionTitleLink'>→</a>". |
||
| 138 | "<em class='text-muted'>".htmlspecialchars($sectionTitle).":</em> "; |
||
| 139 | $wikitext = str_replace($sectionMatch[0][0], trim($sectionWikitext), $wikitext); |
||
| 140 | } |
||
| 141 | |||
| 142 | return self::wikifyInternalLinks($wikitext, $domain); |
||
| 143 | } |
||
| 144 | |||
| 145 | /** |
||
| 146 | * Converts internal links in wikitext to HTML. |
||
| 147 | * @param string $wikitext |
||
| 148 | * @param string $domain The project domain such as en.wikipedia |
||
| 149 | * @return string Updated wikitext. |
||
| 150 | * @static |
||
| 151 | */ |
||
| 152 | private static function wikifyInternalLinks(string $wikitext, string $domain): string |
||
| 153 | { |
||
| 154 | $pagePath = "https://$domain.org/wiki/"; |
||
| 155 | $linkMatch = null; |
||
| 156 | |||
| 157 | while (preg_match_all("/\[\[:?(.*?)\]\]/", $wikitext, $linkMatch)) { |
||
| 158 | $wikiLinkParts = explode('|', $linkMatch[1][0]); |
||
| 159 | $wikiLinkPath = htmlspecialchars($wikiLinkParts[0]); |
||
| 160 | $wikiLinkText = htmlspecialchars( |
||
| 161 | $wikiLinkParts[1] ?? $wikiLinkPath |
||
| 162 | ); |
||
| 163 | |||
| 164 | // Use normalized page title (underscored, capitalized). |
||
| 165 | $pageUrl = $pagePath.ucfirst(str_replace(' ', '_', $wikiLinkPath)); |
||
| 166 | $link = "<a target='_blank' href='$pageUrl'>$wikiLinkText</a>"; |
||
| 167 | $wikitext = str_replace($linkMatch[0][0], $link, $wikitext); |
||
| 168 | } |
||
| 169 | |||
| 170 | return $wikitext; |
||
| 171 | } |
||
| 172 | |||
| 173 | /** |
||
| 174 | * Get all available wikis on the replicas, as defined by EventWiki::VALID_WIKI_PATTERN. |
||
| 175 | * @return string[] With domain as the keys, database name as the values. |
||
| 176 | */ |
||
| 177 | public function getAvailableWikis(): array |
||
| 197 | } |
||
| 198 | |||
| 199 | /** |
||
| 200 | * Get all unique page IDs edited/created within the Event for the given wiki. If you need to do this for pages |
||
| 201 | * within specific categories, without participants, use EventCategoryRepository::getPagesInCategories(). |
||
| 202 | * @param string $dbName |
||
| 203 | * @param DateTime $start |
||
| 204 | * @param DateTime $end |
||
| 205 | * @param int[] $actors |
||
| 206 | * @param string[] $categoryTitles |
||
| 207 | * @param string $type Whether only pages 'created' or 'edited' should be returned. Default is to return both. |
||
| 208 | * To get pages improved, first get edited then use array_diff against created. |
||
| 209 | * @return int[] |
||
| 210 | */ |
||
| 211 | public function getPageIds( |
||
| 270 | } |
||
| 271 | |||
| 272 | /** |
||
| 273 | * Get the total pageviews count for a set of pages, from a given date until today. Optionally reduce to an average |
||
| 274 | * of the last N days, where N is Event::AVAILABLE_METRICS['pages-improved-pageviews-avg']. |
||
| 275 | * @param string $dbName |
||
| 276 | * @param string $domain |
||
| 277 | * @param DateTime $start |
||
| 278 | * @param int[] $pageIds |
||
| 279 | * @param bool $getDailyAverage |
||
| 280 | * @return int |
||
| 281 | */ |
||
| 282 | public function getPageviews( |
||
| 319 | } |
||
| 320 | |||
| 321 | /** |
||
| 322 | * Get the page titles of the pages with the given IDs. |
||
| 323 | * @param string $dbName |
||
| 324 | * @param int[] $pageIds |
||
| 325 | * @param bool $stmt Whether to get only the statement, so that the calling method can use fetch(). |
||
| 326 | * @param bool $includePageIds Whether to include page IDs in the result. |
||
| 327 | * @return mixed[]|\Doctrine\DBAL\Driver\ResultStatement |
||
| 328 | */ |
||
| 329 | public function getPageTitles(string $dbName, array $pageIds, bool $stmt = false, bool $includePageIds = false) |
||
| 340 | } |
||
| 341 | |||
| 342 | /** |
||
| 343 | * Calculates the number of bytes changed during an event |
||
| 344 | * |
||
| 345 | * @param Event $event |
||
| 346 | * @param string $dbName |
||
| 347 | * @param int[] $pageIds |
||
| 348 | * @param int[] $actors |
||
| 349 | * @return int |
||
| 350 | */ |
||
| 351 | public function getBytesChanged(Event $event, string $dbName, array $pageIds, array $actors): int |
||
| 400 | } |
||
| 401 | |||
| 402 | /** |
||
| 403 | * Get the list of users participating in an event with no predefined user list |
||
| 404 | * |
||
| 405 | * @param string $dbName |
||
| 406 | * @param int[] $pageIds |
||
| 407 | * @param Event $event |
||
| 408 | * @return string[] |
||
| 409 | */ |
||
| 410 | public function getUsersFromPageIDs(string $dbName, array $pageIds, Event $event): array |
||
| 425 | } |
||
| 426 | |||
| 427 | /** |
||
| 428 | * Get data for a single page, to be included in the Pages Created report. |
||
| 429 | * @param string $dbName |
||
| 430 | * @param int $pageId |
||
| 431 | * @param string $pageTitle |
||
| 432 | * @param string[] $usernames |
||
| 433 | * @param DateTime $end |
||
| 434 | * @return string[] |
||
| 435 | */ |
||
| 436 | public function getSinglePageCreatedData( |
||
| 437 | string $dbName, |
||
| 438 | int $pageId, |
||
| 439 | string $pageTitle, |
||
| 440 | array $usernames, |
||
| 441 | DateTime $end |
||
| 442 | ): array { |
||
| 443 | // Use cache if it exists. |
||
| 444 | $cacheKey = $this->getCacheKey(func_get_args(), 'pages_created_info'); |
||
| 445 | if ($this->cache->hasItem($cacheKey)) { |
||
| 446 | return $this->cache->getItem($cacheKey)->get(); |
||
| 447 | } |
||
| 448 | |||
| 449 | $end = $end->format('YmdHis'); |
||
| 450 | $usernamesSql = empty($usernames) ? '' : 'AND rev_user_text IN (:usernames)'; |
||
| 451 | |||
| 452 | // Only use revision_userindex when filtering by user. |
||
| 453 | $userRevisionTable = empty($usernames) ? 'revision' : 'revision_userindex'; |
||
| 454 | |||
| 455 | $sql = "SELECT `metric`, `value` FROM ( |
||
| 456 | ( |
||
| 457 | SELECT 'creator' AS `metric`, rev_user_text AS `value` |
||
| 458 | FROM $dbName.revision |
||
| 459 | WHERE rev_page = :pageId |
||
| 460 | LIMIT 1 |
||
| 461 | ) UNION ( |
||
| 462 | SELECT 'edits' AS `metric`, COUNT(*) AS `value` |
||
| 463 | FROM $dbName.$userRevisionTable |
||
| 464 | WHERE rev_page = :pageId |
||
| 465 | AND rev_timestamp <= :end |
||
| 466 | $usernamesSql |
||
| 467 | ) UNION ( |
||
| 468 | SELECT 'bytes' AS `metric`, rev_len AS `value` |
||
| 469 | FROM $dbName.$userRevisionTable |
||
| 470 | WHERE rev_page = :pageId |
||
| 471 | AND rev_timestamp <= :end |
||
| 472 | $usernamesSql |
||
| 473 | ORDER BY rev_timestamp DESC |
||
| 474 | LIMIT 1 |
||
| 475 | ) UNION ( |
||
| 476 | SELECT 'links' AS `metric`, COUNT(*) AS `value` |
||
| 477 | FROM $dbName.pagelinks |
||
| 478 | JOIN $dbName.page ON page_id = pl_from |
||
| 479 | WHERE pl_from_namespace = 0 |
||
| 480 | AND pl_namespace = 0 |
||
| 481 | AND pl_title = :pageTitle |
||
| 482 | AND page_is_redirect = 0 |
||
| 483 | ) |
||
| 484 | ) t1"; |
||
| 485 | |||
| 486 | $ret = $this->executeReplicaQueryWithTypes( |
||
| 487 | $sql, |
||
| 488 | [ |
||
| 489 | 'pageId' => $pageId, |
||
| 490 | 'pageTitle' => $pageTitle, |
||
| 491 | 'usernames' => $usernames, |
||
| 492 | 'end' => $end, |
||
| 493 | ], |
||
| 494 | [ |
||
| 495 | 'usernames' => Connection::PARAM_STR_ARRAY, |
||
| 496 | ] |
||
| 497 | )->fetchAll(\PDO::FETCH_KEY_PAIR); |
||
| 498 | |||
| 499 | // Cache for 10 minutes. |
||
| 500 | return $this->setCache($cacheKey, $ret, 'PT10M'); |
||
| 501 | } |
||
| 502 | |||
| 503 | /** |
||
| 504 | * Get the data needed for the Pages Created report, for a single EventWiki. |
||
| 505 | * @param EventWiki $wiki |
||
| 506 | * @param string[] $usernames |
||
| 507 | * @return mixed[] |
||
| 508 | */ |
||
| 509 | public function getPagesCreatedData(EventWiki $wiki, array $usernames): array |
||
| 510 | { |
||
| 511 | if ($wiki->isFamilyWiki()) { |
||
| 512 | return []; |
||
| 513 | } |
||
| 514 | |||
| 515 | $dbName = $this->getDbNameFromDomain($wiki->getDomain()); |
||
| 516 | $pageviewsRepo = $this->getPageviewsRepository(); |
||
| 517 | $avgPageviewsOffset = Event::AVAILABLE_METRICS['pages-improved-pageviews-avg']; |
||
| 518 | $pages = $this->getPageTitles($dbName, $wiki->getPagesCreated(), true, true); |
||
| 519 | $start = $wiki->getEvent()->getStartUTC(); |
||
| 520 | $end = $wiki->getEvent()->getEndUTC(); |
||
| 521 | $now = new DateTime('yesterday midnight'); |
||
| 522 | $data = []; |
||
| 523 | |||
| 524 | while ($page = $pages->fetch()) { |
||
| 525 | // FIXME: async? |
||
| 526 | [$pageviews, $avgPageviews] = $pageviewsRepo->getPageviews( |
||
| 527 | $wiki->getDomain(), |
||
| 528 | [$page['page_title']], |
||
| 529 | $start, |
||
| 530 | $now, |
||
| 531 | $avgPageviewsOffset |
||
| 532 | ); |
||
| 533 | |||
| 534 | $pageInfo = $this->getSinglePageCreatedData( |
||
| 535 | $dbName, |
||
| 536 | (int)$page['page_id'], |
||
| 537 | $page['page_title'], |
||
| 538 | $usernames, |
||
| 539 | $end |
||
| 540 | ); |
||
| 541 | |||
| 542 | $data[] = array_merge($pageInfo, [ |
||
| 543 | 'pageTitle' => $page['page_title'], |
||
| 544 | 'wiki' => $wiki->getDomain(), |
||
| 545 | 'pageviews' => (int)$pageviews, |
||
| 546 | 'avgPageviews' => (int)$avgPageviews, |
||
| 547 | ]); |
||
| 548 | } |
||
| 549 | |||
| 550 | return $data; |
||
| 551 | } |
||
| 552 | |||
| 553 | /** |
||
| 554 | * Get data for a single page, to be included in the Pages Improved report. |
||
| 555 | * @param string $dbName |
||
| 556 | * @param int $pageId |
||
| 557 | * @param string $pageTitle |
||
| 558 | * @param string[] $usernames |
||
| 559 | * @param DateTime $start |
||
| 560 | * @param DateTime $end |
||
| 561 | * @return string[] |
||
| 562 | * @throws \Psr\Cache\InvalidArgumentException |
||
| 563 | */ |
||
| 564 | public function getSinglePageImprovedData( |
||
| 565 | string $dbName, |
||
| 566 | int $pageId, |
||
| 567 | string $pageTitle, |
||
| 568 | array $usernames, |
||
| 569 | DateTime $start, |
||
| 570 | DateTime $end |
||
| 571 | ): array { |
||
| 572 | // Use cache if it exists. |
||
| 573 | $cacheKey = $this->getCacheKey(func_get_args(), 'pages_improved_info'); |
||
| 574 | if ($this->cache->hasItem($cacheKey)) { |
||
| 575 | return $this->cache->getItem($cacheKey)->get(); |
||
| 576 | } |
||
| 577 | |||
| 578 | $start = $start->format('YmdHis'); |
||
| 579 | $end = $end->format('YmdHis'); |
||
| 580 | $usernamesSql = empty($usernames) ? '' : 'AND rev.rev_user_text IN (:usernames)'; |
||
| 581 | |||
| 582 | // Only use revision_userindex when filtering by user. |
||
| 583 | $userRevisionTable = empty($usernames) ? 'revision' : 'revision_userindex'; |
||
| 584 | |||
| 585 | $sql = "SELECT `metric`, `value` FROM ( |
||
| 586 | ( |
||
| 587 | SELECT 'edits' AS `metric`, COUNT(*) AS `value` |
||
| 588 | FROM $dbName.$userRevisionTable rev |
||
| 589 | WHERE rev_page = :pageId |
||
| 590 | AND rev_timestamp BETWEEN :start AND :end |
||
| 591 | $usernamesSql |
||
| 592 | ) UNION ( |
||
| 593 | SELECT 'start_bytes' AS `metric`, COALESCE(prev.rev_len, 0) AS `value` |
||
| 594 | FROM $dbName.$userRevisionTable rev |
||
| 595 | LEFT JOIN $dbName.$userRevisionTable prev ON rev.rev_parent_id=prev.rev_id |
||
| 596 | WHERE rev.rev_page=:pageId |
||
| 597 | AND rev.rev_timestamp BETWEEN :start AND :end |
||
| 598 | {$usernamesSql} |
||
| 599 | ORDER BY rev.rev_timestamp ASC |
||
| 600 | LIMIT 1 |
||
| 601 | ) UNION ( |
||
| 602 | SELECT 'end_bytes' AS `metric`, COALESCE(rev_len, 0) AS `value` |
||
| 603 | FROM $dbName.$userRevisionTable rev |
||
| 604 | WHERE rev_page=:pageId |
||
| 605 | AND rev_timestamp BETWEEN :start AND :end |
||
| 606 | {$usernamesSql} |
||
| 607 | ORDER BY rev_timestamp DESC |
||
| 608 | LIMIT 1 |
||
| 609 | ) UNION ( |
||
| 610 | SELECT 'links' AS `metric`, COUNT(*) AS `value` |
||
| 611 | FROM $dbName.pagelinks |
||
| 612 | JOIN $dbName.page ON page_id = pl_from |
||
| 613 | WHERE pl_from_namespace = 0 |
||
| 614 | AND pl_namespace = 0 |
||
| 615 | AND pl_title = :pageTitle |
||
| 616 | AND page_is_redirect = 0 |
||
| 617 | ) |
||
| 618 | ) t1"; |
||
| 619 | |||
| 620 | $rows = $this->executeReplicaQueryWithTypes( |
||
| 621 | $sql, |
||
| 622 | [ |
||
| 623 | 'pageId' => $pageId, |
||
| 624 | 'pageTitle' => $pageTitle, |
||
| 625 | 'usernames' => $usernames, |
||
| 626 | 'start' => $start, |
||
| 627 | 'end' => $end, |
||
| 628 | ], |
||
| 629 | [ |
||
| 630 | 'usernames' => Connection::PARAM_STR_ARRAY, |
||
| 631 | ] |
||
| 632 | )->fetchAll(\PDO::FETCH_KEY_PAIR); |
||
| 633 | |||
| 634 | $ret = [ |
||
| 635 | 'edits' => $rows['edits'], |
||
| 636 | 'links' => $rows['links'], |
||
| 637 | 'bytes' => $rows['end_bytes'] - $rows['start_bytes'], |
||
| 638 | ]; |
||
| 639 | |||
| 640 | // Cache for 10 minutes. |
||
| 641 | return $this->setCache($cacheKey, $ret, 'PT10M'); |
||
| 642 | } |
||
| 643 | |||
| 644 | /** |
||
| 645 | * Get the data needed for the Pages Created report, for a single EventWiki. |
||
| 646 | * @param EventWiki $wiki |
||
| 647 | * @param string[] $usernames |
||
| 648 | * @return mixed[] |
||
| 649 | */ |
||
| 650 | public function getPagesImprovedData(EventWiki $wiki, array $usernames): array |
||
| 689 | } |
||
| 690 | |||
| 691 | /** |
||
| 692 | * Creates and initializes a pageviews repository |
||
| 693 | * @return PageviewsRepository |
||
| 694 | */ |
||
| 695 | private function getPageviewsRepository(): PageviewsRepository |
||
| 702 |
This check marks implicit conversions of arrays to boolean values in a comparison. While in PHP an empty array is considered to be equal (but not identical) to false, this is not always apparent.
Consider making the comparison explicit by using
empty(..)or! empty(...)instead.