Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 49 |
|
0.00% |
0 / 5 |
CRAP | |
0.00% |
0 / 1 |
| Hooks | |
0.00% |
0 / 49 |
|
0.00% |
0 / 5 |
342 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| trimExtract | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
12 | |||
| getExtractsData | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
6 | |||
| onApiOpenSearchSuggest | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
42 | |||
| onSearchResultProvideDescription | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
42 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\TextExtracts; |
| 4 | |
| 5 | use Generator; |
| 6 | use MediaWiki\Api\ApiMain; |
| 7 | use MediaWiki\Api\ApiResult; |
| 8 | use MediaWiki\Api\Hook\ApiOpenSearchSuggestHook; |
| 9 | use MediaWiki\Config\Config; |
| 10 | use MediaWiki\Config\ConfigFactory; |
| 11 | use MediaWiki\Request\FauxRequest; |
| 12 | use MediaWiki\Rest\Hook\SearchResultProvideDescriptionHook; |
| 13 | |
| 14 | /** |
| 15 | * @license GPL-2.0-or-later |
| 16 | */ |
| 17 | class Hooks implements |
| 18 | ApiOpenSearchSuggestHook, |
| 19 | SearchResultProvideDescriptionHook |
| 20 | { |
| 21 | |
| 22 | private Config $config; |
| 23 | |
| 24 | public function __construct( |
| 25 | ConfigFactory $configFactory |
| 26 | ) { |
| 27 | $this->config = $configFactory->makeConfig( 'textextracts' ); |
| 28 | } |
| 29 | |
| 30 | /** |
| 31 | * Trim an extract to a sensible length. |
| 32 | * |
| 33 | * Adapted from Extension:OpenSearchXml, which adapted it from |
| 34 | * Extension:ActiveAbstract. |
| 35 | * |
| 36 | * @param string $text |
| 37 | * @param int $length Target length; actual result will continue to the end of a sentence. |
| 38 | * @return string |
| 39 | */ |
| 40 | private static function trimExtract( $text, $length ) { |
| 41 | static $regex = null; |
| 42 | if ( $regex === null ) { |
| 43 | $endchars = [ |
| 44 | // regular ASCII |
| 45 | '([^\d])\.\s', '\!\s', '\?\s', |
| 46 | // full-width ideographic full-stop |
| 47 | '。', |
| 48 | // double-width roman forms |
| 49 | '．', '！', '？', |
| 50 | // half-width ideographic full stop |
| 51 | '｡', |
| 52 | ]; |
| 53 | $endgroup = implode( '|', $endchars ); |
| 54 | $end = "(?:$endgroup)"; |
| 55 | $sentence = ".{{$length},}?$end+"; |
| 56 | $regex = "/^($sentence)/u"; |
| 57 | } |
| 58 | $matches = []; |
| 59 | if ( preg_match( $regex, $text, $matches ) ) { |
| 60 | return trim( $matches[1] ); |
| 61 | } else { |
| 62 | // Just return the first line |
| 63 | return trim( explode( "\n", $text )[0] ); |
| 64 | } |
| 65 | } |
| 66 | |
| 67 | /** |
| 68 | * Retrieves extracts data for the given page IDs from the TextExtract API. |
| 69 | * The page IDs are chunked into the max limit of exlimit of the TextExtract API |
| 70 | * |
| 71 | * @param array $pageIds An array of page IDs to retrieve extracts for |
| 72 | * @return Generator Yields the result data from the API request |
| 73 | * $data = [ |
| 74 | * 'pageId' => [ |
| 75 | * 'ns' => int of the namespace |
| 76 | * 'title' => string of the title of the page |
| 77 | * 'extract' => string of the text extracts of the page |
| 78 | * ] |
| 79 | * ] |
| 80 | */ |
| 81 | private function getExtractsData( array $pageIds ) { |
| 82 | foreach ( array_chunk( $pageIds, 20 ) as $chunkedPageIds ) { |
| 83 | $api = new ApiMain( new FauxRequest( |
| 84 | [ |
| 85 | 'action' => 'query', |
| 86 | 'prop' => 'extracts', |
| 87 | 'explaintext' => true, |
| 88 | 'exintro' => true, |
| 89 | 'exlimit' => count( $chunkedPageIds ), |
| 90 | 'pageids' => implode( '|', $chunkedPageIds ), |
| 91 | ] |
| 92 | ) ); |
| 93 | $api->execute(); |
| 94 | yield $api->getResult()->getResultData( [ 'query', 'pages' ] ); |
| 95 | } |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * ApiOpenSearchSuggest hook handler |
| 100 | * @param array &$results Array of search results |
| 101 | */ |
| 102 | public function onApiOpenSearchSuggest( &$results ) { |
| 103 | if ( !$this->config->get( 'ExtractsExtendOpenSearchXml' ) || $results === [] ) { |
| 104 | return; |
| 105 | } |
| 106 | |
| 107 | $pageIds = array_keys( $results ); |
| 108 | foreach ( $this->getExtractsData( $pageIds ) as $data ) { |
| 109 | foreach ( $pageIds as $id ) { |
| 110 | $contentKey = $data[$id]['extract'][ApiResult::META_CONTENT] ?? '*'; |
| 111 | if ( isset( $data[$id]['extract'][$contentKey] ) ) { |
| 112 | $results[$id]['extract'] = $data[$id]['extract'][$contentKey]; |
| 113 | $results[$id]['extract trimmed'] = false; |
| 114 | } |
| 115 | } |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | /** |
| 120 | * Used to update Search Results with descriptions for Search Engine. |
| 121 | * @param array $pageIdentities Array (string=>SearchResultPageIdentity) where key is pageId |
| 122 | * @param array &$descriptions Output array (string=>string|null) |
| 123 | * where key is pageId and value is either a description for given page or null |
| 124 | */ |
| 125 | public function onSearchResultProvideDescription( |
| 126 | array $pageIdentities, |
| 127 | &$descriptions |
| 128 | ): void { |
| 129 | if ( !$this->config->get( 'ExtractsExtendRestSearch' ) || $pageIdentities === [] ) { |
| 130 | return; |
| 131 | } |
| 132 | |
| 133 | $pageIds = array_map( static function ( $identity ) { |
| 134 | return $identity->getId(); |
| 135 | }, $pageIdentities ); |
| 136 | foreach ( $this->getExtractsData( $pageIds ) as $data ) { |
| 137 | foreach ( $pageIds as $id ) { |
| 138 | $contentKey = $data[$id]['extract'][ApiResult::META_CONTENT] ?? '*'; |
| 139 | if ( isset( $data[$id]['extract'][$contentKey] ) ) { |
| 140 | $descriptions[$id] = self::trimExtract( $data[$id]['extract'][$contentKey], 150 ); |
| 141 | } |
| 142 | } |
| 143 | } |
| 144 | } |
| 145 | } |