Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
81.82% |
18 / 22 |
|
0.00% |
0 / 2 |
CRAP | |
0.00% |
0 / 1 |
| DedupeHelper | |
81.82% |
18 / 22 |
|
0.00% |
0 / 2 |
7.29 | |
0.00% |
0 / 1 |
| getDedupeHash | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| hasRecentlyDeliveredDuplicate | |
94.74% |
18 / 19 |
|
0.00% |
0 / 1 |
4.00 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\MassMessage; |
| 4 | |
| 5 | use MediaWiki\Json\FormatJson; |
| 6 | use MediaWiki\MediaWikiServices; |
| 7 | use MediaWiki\Title\Title; |
| 8 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 9 | |
| 10 | /** |
| 11 | * Compute a dedupe hash based on the message subject and contents for |
| 12 | * each message delivery job, and store it in ct_params alongside the |
| 13 | * massmessage-delivery change tag. When delivering a message, check |
| 14 | * whether there is another MassMessage delivery within the past 5 page |
| 15 | * revisions with an identical dedupe hash, and skip the delivery if |
| 16 | * there is. |
| 17 | * |
| 18 | * Notes: |
| 19 | * - This relies on direct database queries since ct_params is not exposed |
| 20 | * through the API. |
| 21 | * - This only works for wikitext talk pages since we don't attach the |
| 22 | * change tag for either Flow or LQT. |
| 23 | */ |
| 24 | class DedupeHelper { |
| 25 | |
| 26 | private const RECENT_REVISIONS_LIMIT = 5; |
| 27 | |
| 28 | /** |
| 29 | * Get the dedupe hash corresponding to a MassMessageJob |
| 30 | * |
| 31 | * @param string $subject |
| 32 | * @param string $message |
| 33 | * @param ?LanguageAwareText $pageSubject |
| 34 | * @param ?LanguageAwareText $pageMessage |
| 35 | * @return string |
| 36 | */ |
| 37 | public static function getDedupeHash( |
| 38 | string $subject, |
| 39 | string $message, |
| 40 | ?LanguageAwareText $pageSubject, |
| 41 | ?LanguageAwareText $pageMessage |
| 42 | ): string { |
| 43 | $pageSubjectText = $pageSubject !== null ? $pageSubject->getWikitext() : ''; |
| 44 | $pageMessageText = $pageMessage !== null ? $pageMessage->getWikitext() : ''; |
| 45 | return md5( $subject . $message . $pageSubjectText . $pageMessageText ); |
| 46 | } |
| 47 | |
| 48 | /** |
| 49 | * For the given title, check if any of the most recent RECENT_REVISIONS_LIMIT revisions is a |
| 50 | * MassMessage delivery for the same message. |
| 51 | * |
| 52 | * @param Title $title |
| 53 | * @param string $dedupeHash |
| 54 | * @return bool |
| 55 | */ |
| 56 | public static function hasRecentlyDeliveredDuplicate( Title $title, string $dedupeHash ): bool { |
| 57 | $services = MediaWikiServices::getInstance(); |
| 58 | |
| 59 | $changeTagId = $services->getChangeTagDefStore()->acquireId( 'massmessage-delivery' ); |
| 60 | |
| 61 | // Connect to the primary to avoid issues with replication lag. |
| 62 | $dbw = $services->getDBLoadBalancerFactory()->getPrimaryDatabase(); |
| 63 | $res = $dbw->newSelectQueryBuilder() |
| 64 | ->select( 'ct_params' ) |
| 65 | ->from( 'revision' ) |
| 66 | ->leftJoin( 'change_tag', null, [ 'ct_rev_id = rev_id', 'ct_tag_id' => $changeTagId ] ) |
| 67 | ->where( [ 'rev_page' => $title->getArticleID() ] ) |
| 68 | ->orderBy( 'rev_timestamp', SelectQueryBuilder::SORT_DESC ) |
| 69 | ->limit( self::RECENT_REVISIONS_LIMIT ) |
| 70 | ->caller( __METHOD__ ) |
| 71 | ->fetchResultSet(); |
| 72 | |
| 73 | foreach ( $res as $row ) { |
| 74 | if ( $row->ct_params === null ) { |
| 75 | continue; |
| 76 | } |
| 77 | $params = FormatJson::decode( $row->ct_params, true ); |
| 78 | if ( $dedupeHash === ( $params['dedupe_hash'] ?? null ) ) { |
| 79 | return true; |
| 80 | } |
| 81 | } |
| 82 | return false; |
| 83 | } |
| 84 | } |