Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
25 / 25 |
|
100.00% |
3 / 3 |
CRAP | |
100.00% |
1 / 1 |
| ParserObserver | |
100.00% |
25 / 25 |
|
100.00% |
3 / 3 |
6 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| notifyParse | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
4 | |||
| getParseId | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Observer to detect parser behaviors such as duplicate parses |
| 5 | * |
| 6 | * @license GPL-2.0-or-later |
| 7 | * @since 1.38 |
| 8 | * |
| 9 | * @file |
| 10 | * @ingroup Parser |
| 11 | * |
| 12 | * @author Cindy Cicalese |
| 13 | */ |
| 14 | |
| 15 | namespace MediaWiki\Parser; |
| 16 | |
| 17 | use MediaWiki\Content\Content; |
| 18 | use MediaWiki\Page\CacheKeyHelper; |
| 19 | use MediaWiki\Page\PageReference; |
| 20 | use MediaWiki\Title\Title; |
| 21 | use Psr\Log\LoggerInterface; |
| 22 | use RuntimeException; |
| 23 | use Wikimedia\ObjectCache\MapCacheLRU; |
| 24 | |
| 25 | /** |
| 26 | * For observing and detecting parser behaviors, such as duplicate parses |
| 27 | * @internal |
| 28 | */ |
| 29 | class ParserObserver { |
| 30 | private readonly MapCacheLRU $previousParseStackTraces; |
| 31 | |
| 32 | public function __construct( private readonly LoggerInterface $logger ) { |
| 33 | $this->previousParseStackTraces = new MapCacheLRU( 10 ); |
| 34 | } |
| 35 | |
| 36 | /** |
| 37 | * @param PageReference $page |
| 38 | * @param int|null $revId |
| 39 | * @param ParserOptions $options |
| 40 | * @param Content $content |
| 41 | * @param ParserOutput $output |
| 42 | */ |
| 43 | public function notifyParse( |
| 44 | PageReference $page, ?int $revId, ParserOptions $options, Content $content, ParserOutput $output |
| 45 | ) { |
| 46 | $pageKey = CacheKeyHelper::getKeyForPage( $page ); |
| 47 | |
| 48 | $optionsHash = $options->optionsHash( |
| 49 | $output->getUsedOptions(), |
| 50 | Title::newFromPageReference( $page ) |
| 51 | ); |
| 52 | |
| 53 | $contentStr = $content->isValid() ? $content->serialize() : null; |
| 54 | // $contentStr may be null if the content could not be serialized |
| 55 | $contentSha1 = $contentStr ? sha1( $contentStr ) : 'INVALID'; |
| 56 | |
| 57 | $index = $this->getParseId( $pageKey, $revId, $optionsHash, $contentSha1 ); |
| 58 | |
| 59 | $stackTrace = ( new RuntimeException() )->getTraceAsString(); |
| 60 | if ( $this->previousParseStackTraces->has( $index ) ) { |
| 61 | |
| 62 | // NOTE: there may be legitimate changes to re-parse the same WikiText content, |
| 63 | // e.g. if predicted revision ID for the REVISIONID magic word mismatched. |
| 64 | // But that should be rare. |
| 65 | $this->logger->debug( |
| 66 | __METHOD__ . ': Possibly redundant parse!', |
| 67 | [ |
| 68 | 'page' => $pageKey, |
| 69 | 'rev' => $revId, |
| 70 | 'options-hash' => $optionsHash, |
| 71 | 'contentSha1' => $contentSha1, |
| 72 | 'trace' => $stackTrace, |
| 73 | 'previous-trace' => $this->previousParseStackTraces->get( $index ), |
| 74 | ] |
| 75 | ); |
| 76 | } |
| 77 | $this->previousParseStackTraces->set( $index, $stackTrace ); |
| 78 | } |
| 79 | |
| 80 | /** |
| 81 | * @param string $titleStr |
| 82 | * @param int|null $revId |
| 83 | * @param string $optionsHash |
| 84 | * @param string $contentSha1 |
| 85 | * @return string |
| 86 | */ |
| 87 | private function getParseId( string $titleStr, ?int $revId, string $optionsHash, string $contentSha1 ): string { |
| 88 | // $revId may be null when previewing a new page |
| 89 | $revIdStr = $revId ?? ""; |
| 90 | |
| 91 | return "$titleStr.$revIdStr.$optionsHash.$contentSha1"; |
| 92 | } |
| 93 | |
| 94 | } |