Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
13.33% |
8 / 60 |
|
40.00% |
2 / 5 |
CRAP | |
0.00% |
0 / 1 |
| ApiTrait | |
13.33% |
8 / 60 |
|
40.00% |
2 / 5 |
431.85 | |
0.00% |
0 / 1 |
| getCirrusConnection | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getSearchConfig | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| loadDocuments | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
72 | |||
| determineCirrusDocId | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
72 | |||
| hasRedirect | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
30 | |||
| getUser | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Api; |
| 4 | |
| 5 | use CirrusSearch\Connection; |
| 6 | use CirrusSearch\SearchConfig; |
| 7 | use CirrusSearch\Searcher; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use MediaWiki\Revision\RevisionRecord; |
| 10 | use MediaWiki\Revision\SlotRecord; |
| 11 | use PageArchive; |
| 12 | use Title; |
| 13 | use User; |
| 14 | |
| 15 | trait ApiTrait { |
| 16 | /** @var Connection */ |
| 17 | private $connection; |
| 18 | /** @var SearchConfig */ |
| 19 | private $searchConfig; |
| 20 | |
| 21 | /** |
| 22 | * @return Connection |
| 23 | */ |
| 24 | public function getCirrusConnection() { |
| 25 | if ( $this->connection === null ) { |
| 26 | $this->connection = new Connection( $this->getSearchConfig() ); |
| 27 | } |
| 28 | return $this->connection; |
| 29 | } |
| 30 | |
| 31 | /** |
| 32 | * @return SearchConfig |
| 33 | */ |
| 34 | protected function getSearchConfig() { |
| 35 | if ( $this->searchConfig === null ) { |
| 36 | $this->searchConfig = MediaWikiServices::getInstance() |
| 37 | ->getConfigFactory() |
| 38 | ->makeConfig( 'CirrusSearch' ); |
| 39 | } |
| 40 | return $this->searchConfig; |
| 41 | } |
| 42 | |
| 43 | /** |
| 44 | * @param Title $title |
| 45 | * @param string[]|bool $sourceFiltering source filtering to apply |
| 46 | * @return array |
| 47 | */ |
| 48 | public function loadDocuments( Title $title, $sourceFiltering = true ) { |
| 49 | list( $docId, $hasRedirects ) = $this->determineCirrusDocId( $title ); |
| 50 | if ( $docId === null ) { |
| 51 | return []; |
| 52 | } |
| 53 | // could be optimized by implementing multi-get but not |
| 54 | // expecting much usage except debugging/tests. |
| 55 | $searcher = new Searcher( $this->getCirrusConnection(), 0, 0, $this->getSearchConfig(), [], $this->getUser() ); |
| 56 | $esSources = $searcher->get( [ $docId ], $sourceFiltering ); |
| 57 | $result = []; |
| 58 | if ( $esSources->isOK() ) { |
| 59 | foreach ( $esSources->getValue() as $esSource ) { |
| 60 | // If we have followed redirects only report the |
| 61 | // article dump if the redirect has been indexed. If it |
| 62 | // hasn't been indexed this document does not represent |
| 63 | // the original title. |
| 64 | if ( $hasRedirects && |
| 65 | !$this->hasRedirect( $esSource->getData(), $title ) |
| 66 | ) { |
| 67 | continue; |
| 68 | } |
| 69 | |
| 70 | // If this was not a redirect and the title doesn't match that |
| 71 | // means a page was moved, but elasticsearch has not yet been |
| 72 | // updated. Don't return the document that doesn't actually |
| 73 | // represent the page (yet). |
| 74 | if ( !$hasRedirects && $esSource->getData()['title'] != $title->getText() ) { |
| 75 | continue; |
| 76 | } |
| 77 | |
| 78 | $result[] = [ |
| 79 | 'index' => $esSource->getIndex(), |
| 80 | 'type' => $esSource->getType(), |
| 81 | 'id' => $esSource->getId(), |
| 82 | 'version' => $esSource->getVersion(), |
| 83 | 'source' => $esSource->getData(), |
| 84 | ]; |
| 85 | } |
| 86 | } |
| 87 | return $result; |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * Trace redirects to find the page id the title should be indexed to in |
| 92 | * cirrussearch. Differs from Updater::traceRedirects in that this also |
| 93 | * supports archived pages. Archive support is important for integration |
| 94 | * tests that need to know when a page that was deleted from SQL was |
| 95 | * finally removed from elasticsearch. |
| 96 | * |
| 97 | * This still fails to find the correct page id if something was moved, as |
| 98 | * that page is renamed rather than being moved to the archive. We could |
| 99 | * further complicate things by looking into move logs but not sure that |
| 100 | * is worth the complication. |
| 101 | * |
| 102 | * @param Title $title |
| 103 | * @return array Two element array containing first the cirrus doc id |
| 104 | * the title should have been indexed into elasticsearch and second a |
| 105 | * boolean indicating if redirects were followed. If the page would |
| 106 | * not be indexed (for example a redirect loop, or redirect to |
| 107 | * invalid page) the first array element will be null. |
| 108 | */ |
| 109 | private function determineCirrusDocId( Title $title ) { |
| 110 | $hasRedirects = false; |
| 111 | $seen = []; |
| 112 | $now = wfTimestamp( TS_MW ); |
| 113 | $contentHandlerFactory = MediaWikiServices::getInstance()->getContentHandlerFactory(); |
| 114 | while ( true ) { |
| 115 | if ( isset( $seen[$title->getPrefixedText()] ) || count( $seen ) > 10 ) { |
| 116 | return [ null, $hasRedirects ]; |
| 117 | } |
| 118 | $seen[$title->getPrefixedText()] = true; |
| 119 | |
| 120 | // To help the integration tests figure out when a deleted page has |
| 121 | // been removed from the elasticsearch index we lookup the page in |
| 122 | // the archive to get it's page id. getPreviousRevisionRecord will |
| 123 | // check both the archive and live content to return the most recent. |
| 124 | $revRecord = ( new PageArchive( $title ) )->getPreviousRevisionRecord( $now ); |
| 125 | if ( !$revRecord ) { |
| 126 | return [ null, $hasRedirects ]; |
| 127 | } |
| 128 | |
| 129 | $pageId = $revRecord->getPageId(); |
| 130 | $mainSlot = $revRecord->getSlot( SlotRecord::MAIN, RevisionRecord::RAW ); |
| 131 | $handler = $contentHandlerFactory->getContentHandler( $mainSlot->getModel() ); |
| 132 | if ( !$handler->supportsRedirects() ) { |
| 133 | return [ $pageId, $hasRedirects ]; |
| 134 | } |
| 135 | $content = $mainSlot->getContent(); |
| 136 | // getUltimateRedirectTarget() would be prefered, but it wont find |
| 137 | // archive pages... |
| 138 | if ( !$content->isRedirect() ) { |
| 139 | return [ $this->getSearchConfig()->makeId( $pageId ), $hasRedirects ]; |
| 140 | } |
| 141 | $redirect = $content->getRedirectTarget(); |
| 142 | if ( !$redirect ) { |
| 143 | // TODO: Can this happen? |
| 144 | return [ $pageId, $hasRedirects ]; |
| 145 | } |
| 146 | |
| 147 | $hasRedirects = true; |
| 148 | $title = $redirect; |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * @param array $source _source document from elasticsearch |
| 154 | * @param Title $title Title to check for redirect |
| 155 | * @return bool True when $title is stored as a redirect in $source |
| 156 | */ |
| 157 | private function hasRedirect( array $source, Title $title ) { |
| 158 | if ( !isset( $source['redirect'] ) ) { |
| 159 | return false; |
| 160 | } |
| 161 | foreach ( $source['redirect'] as $redirect ) { |
| 162 | if ( $redirect['namespace'] === $title->getNamespace() |
| 163 | && $redirect['title'] === $title->getText() |
| 164 | ) { |
| 165 | return true; |
| 166 | } |
| 167 | } |
| 168 | return false; |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * @return User |
| 173 | */ |
| 174 | abstract public function getUser(); |
| 175 | |
| 176 | } |