Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 61 |
|
0.00% |
0 / 4 |
CRAP | |
0.00% |
0 / 1 |
| EditPageShowEditFormInitialHandler | |
0.00% |
0 / 61 |
|
0.00% |
0 / 4 |
182 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| onEditPage__showEditForm_initial | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
42 | |||
| getLangs | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| getLangsForEngine | |
0.00% |
0 / 35 |
|
0.00% |
0 / 1 |
30 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\Wikisource\HookHandler; |
| 4 | |
| 5 | use MediaWiki\Config\Config; |
| 6 | use MediaWiki\Config\ConfigException; |
| 7 | use MediaWiki\EditPage\EditPage; |
| 8 | use MediaWiki\Hook\EditPage__showEditForm_initialHook; |
| 9 | use MediaWiki\Logger\LoggerFactory; |
| 10 | use MediaWiki\MediaWikiServices; |
| 11 | use MediaWiki\Output\OutputPage; |
| 12 | use MediaWiki\ResourceLoader\Context; |
| 13 | use MediaWiki\User\Options\UserOptionsLookup; |
| 14 | |
| 15 | // phpcs:disable MediaWiki.NamingConventions.LowerCamelFunctionsName.FunctionName |
| 16 | class EditPageShowEditFormInitialHandler implements EditPage__showEditForm_initialHook { |
| 17 | |
| 18 | /** @var bool */ |
| 19 | private $enabled; |
| 20 | |
| 21 | /** @var string */ |
| 22 | private $toolUrl; |
| 23 | |
| 24 | /** @var array */ |
| 25 | private $WikisourceTranskribusModels; |
| 26 | |
| 27 | private UserOptionsLookup $userOptionsLookup; |
| 28 | |
| 29 | public function __construct( |
| 30 | Config $config, |
| 31 | UserOptionsLookup $userOptionsLookup |
| 32 | ) { |
| 33 | $this->enabled = (bool)$config->get( 'WikisourceEnableOcr' ); |
| 34 | $this->toolUrl = rtrim( $config->get( 'WikisourceOcrUrl' ), '/' ); |
| 35 | $this->WikisourceTranskribusModels = $config->get( 'WikisourceTranskribusModels' ); |
| 36 | $this->userOptionsLookup = $userOptionsLookup; |
| 37 | } |
| 38 | |
| 39 | /** |
| 40 | * @param EditPage $editor |
| 41 | * @param OutputPage $out OutputPage instance to write to |
| 42 | * @return bool|void True or no return value to continue or false to abort |
| 43 | */ |
| 44 | public function onEditPage__showEditForm_initial( $editor, $out ) { |
| 45 | if ( !$this->enabled ) { |
| 46 | return; |
| 47 | } |
| 48 | // Make sure we're editing a page of the right content type (and that PRP is available). |
| 49 | if ( !defined( 'CONTENT_MODEL_PROOFREAD_PAGE' ) |
| 50 | || $editor->contentModel !== CONTENT_MODEL_PROOFREAD_PAGE ) { |
| 51 | return; |
| 52 | } |
| 53 | // Make sure there's a tool URL defined. |
| 54 | if ( !$this->toolUrl ) { |
| 55 | throw new ConfigException( 'Please set tool URL with $wgWikisourceOcrUrl' ); |
| 56 | } |
| 57 | // Require the WikiEditor toolbar to be enabled. |
| 58 | $useBetaToolbar = $this->userOptionsLookup |
| 59 | ->getOption( $out->getUser(), 'usebetatoolbar' ); |
| 60 | if ( !$useBetaToolbar ) { |
| 61 | return; |
| 62 | } |
| 63 | |
| 64 | // Add tool's URL to Content Security Policy. |
| 65 | $out->getCSP()->addDefaultSrc( $this->toolUrl ); |
| 66 | // Add OCR modules. |
| 67 | $out->addModules( 'ext.wikisource.OCR' ); |
| 68 | $out->addJsConfigVars( [ |
| 69 | 'WikisourceOcrUrl' => $this->toolUrl, |
| 70 | 'WikisourceTranskribusModels' => $this->WikisourceTranskribusModels, |
| 71 | ] ); |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * Get all languages/models. |
| 76 | * @param Context $context |
| 77 | * @param Config $config |
| 78 | * @return array |
| 79 | */ |
| 80 | public static function getLangs( Context $context, Config $config ) { |
| 81 | return [ |
| 82 | 'google' => self::getLangsForEngine( 'google', $config ), |
| 83 | 'tesseract' => self::getLangsForEngine( 'tesseract', $config ), |
| 84 | 'transkribus' => self::getLangsForEngine( 'transkribus', $config ), |
| 85 | ]; |
| 86 | } |
| 87 | |
| 88 | /** |
| 89 | * Get available languages/models for a selected engine. |
| 90 | * @param string $engine |
| 91 | * @param Config $config |
| 92 | * @return array |
| 93 | */ |
| 94 | protected static function getLangsForEngine( $engine, $config ) { |
| 95 | if ( defined( 'MW_PHPUNIT_TEST' ) ) { |
| 96 | return []; |
| 97 | } |
| 98 | $http = MediaWikiServices::getInstance()->getHttpRequestFactory(); |
| 99 | $cache = MediaWikiServices::getInstance()->getMainWANObjectCache(); |
| 100 | $logger = LoggerFactory::getInstance( 'Wikisource' ); |
| 101 | $toolUrl = rtrim( $config->get( 'WikisourceOcrUrl' ), '/' ); |
| 102 | $proxy = $config->get( 'WikisourceHttpProxy' ); |
| 103 | $url = $toolUrl . '/api/available_langs?engine=' . $engine; |
| 104 | $fname = __METHOD__; |
| 105 | $langs = $cache->getWithSetCallback( |
| 106 | $cache->makeGlobalKey( 'wikisource-ocr-langs', $engine ), |
| 107 | $cache::TTL_DAY, |
| 108 | static function () use ( $url, $http, $proxy, $logger, $engine, $fname ) { |
| 109 | $logger->debug( 'Language list not cached for {engine}, fetching now', [ 'engine' => $engine ] ); |
| 110 | $options = []; |
| 111 | if ( $proxy ) { |
| 112 | $options[ 'proxy' ] = $proxy; |
| 113 | } |
| 114 | $startTime = microtime( true ); |
| 115 | $response = $http->get( $url, $options, $fname ); |
| 116 | $logger->info( |
| 117 | 'OCR tool responded with {response_size} bytes after {response_time}ms', |
| 118 | [ |
| 119 | 'response_size' => strlen( (string)$response ), |
| 120 | 'response_time' => ( microtime( true ) - $startTime ) * 1000, |
| 121 | ] |
| 122 | ); |
| 123 | if ( $response === null ) { |
| 124 | $logger->warning( 'OCR empty response from tool', [ 'url' => $url ] ); |
| 125 | return false; |
| 126 | } |
| 127 | $contents = json_decode( $response ); |
| 128 | return $contents->available_langs ?? false; |
| 129 | }, |
| 130 | [ 'staleTTL' => $cache::TTL_WEEK ] |
| 131 | ); |
| 132 | return $langs === false ? [] : $langs; |
| 133 | } |
| 134 | } |