Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 123 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| CountOrphanedFiles | |
0.00% |
0 / 117 |
|
0.00% |
0 / 7 |
756 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
20 | |||
| getSites | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
42 | |||
| isExtensionInstalled | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
42 | |||
| fetchUsedFiles | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| reportUnusedFiles | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
30 | |||
| deleteFiles | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | |
| 3 | use MediaWiki\Extension\Phonos\Engine\Engine; |
| 4 | use MediaWiki\Http\HttpRequestFactory; |
| 5 | use MediaWiki\Maintenance\Maintenance; |
| 6 | use MediaWiki\Site\MediaWikiSite; |
| 7 | use MediaWiki\Site\SiteList; |
| 8 | use MediaWiki\Site\SiteStore; |
| 9 | use MediaWiki\Status\Status; |
| 10 | use MediaWiki\WikiMap\WikiMap; |
| 11 | use Wikimedia\FileBackend\FileBackend; |
| 12 | use Wikimedia\Rdbms\LBFactory; |
| 13 | |
| 14 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 15 | if ( $IP === false ) { |
| 16 | $IP = __DIR__ . '/../../..'; |
| 17 | } |
| 18 | require_once "$IP/maintenance/Maintenance.php"; |
| 19 | |
| 20 | /** |
| 21 | * Maintenance script to find and optionally delete orphaned Phonos files. |
| 22 | * |
| 23 | * On wiki farms, you can use the '--wikis' flag to specify which wikis to process, passing |
| 24 | * in the global IDs (database names). If not provided, the script will loop through all |
| 25 | * wikis as specified in the 'sites' table, and process any where Phonos is installed. |
| 26 | * If the 'sites' table is not set up, the script will act only on the current wiki. |
| 27 | * |
| 28 | * @see https://www.mediawiki.org/wiki/Manual:AddSite.php |
| 29 | * |
| 30 | * @ingroup Maintenance |
| 31 | */ |
| 32 | class CountOrphanedFiles extends Maintenance { |
| 33 | |
| 34 | private HttpRequestFactory $requestFactory; |
| 35 | private LBFactory $lbFactory; |
| 36 | private SiteStore $siteStore; |
| 37 | private FileBackend $backend; |
| 38 | |
| 39 | /** @var string|false */ |
| 40 | private $apiProxy; |
| 41 | |
| 42 | public function __construct() { |
| 43 | parent::__construct(); |
| 44 | $this->addDescription( 'Find and optionally delete orphaned Phonos files across all wikis.' ); |
| 45 | $this->addOption( 'delete', 'Delete the orphaned files in addition to reporting how many there are.' ); |
| 46 | $this->addOption( |
| 47 | 'wikis', |
| 48 | 'Comma-separated list of db names. Only these wikis will be processed.', |
| 49 | false, |
| 50 | true |
| 51 | ); |
| 52 | $this->requireExtension( 'Phonos' ); |
| 53 | } |
| 54 | |
| 55 | public function execute(): void { |
| 56 | $services = $this->getServiceContainer(); |
| 57 | $config = $services->getMainConfig(); |
| 58 | $this->requestFactory = $services->getHttpRequestFactory(); |
| 59 | $this->apiProxy = $config->get( 'PhonosApiProxy' ); |
| 60 | $this->lbFactory = $services->getDBLoadBalancerFactory(); |
| 61 | $this->siteStore = $services->getSiteStore(); |
| 62 | $this->backend = Engine::getFileBackend( |
| 63 | $services->getFileBackendGroup(), |
| 64 | $config |
| 65 | ); |
| 66 | |
| 67 | $usedFiles = []; |
| 68 | $skippedSites = 0; |
| 69 | /** @var MediaWikiSite $site */ |
| 70 | foreach ( $this->getSites() as $site ) { |
| 71 | try { |
| 72 | $usedFiles = array_unique( array_merge( $usedFiles, $this->fetchUsedFiles( $site ) ) ); |
| 73 | } catch ( Throwable $e ) { |
| 74 | $skippedSites++; |
| 75 | $this->error( $e->getMessage() . "\n" ); |
| 76 | continue; |
| 77 | } |
| 78 | } |
| 79 | |
| 80 | $msg = count( $usedFiles ) . ' in-use files found.' . |
| 81 | ( $skippedSites > 0 ? " $skippedSites sites skipped due to errors." : '' ); |
| 82 | $this->output( "$msg\n" ); |
| 83 | |
| 84 | $this->reportUnusedFiles( array_unique( $usedFiles ) ); |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * Get an array of all the sites we need to query. |
| 89 | * |
| 90 | * @return SiteList |
| 91 | */ |
| 92 | private function getSites(): SiteList { |
| 93 | $wikisOption = $this->getOption( 'wikis' ); |
| 94 | if ( $wikisOption ) { |
| 95 | $wikis = explode( ',', $wikisOption ); |
| 96 | $sites = new SiteList(); |
| 97 | foreach ( $wikis as $wiki ) { |
| 98 | /** @var MediaWikiSite $site */ |
| 99 | $site = $this->siteStore->getSite( $wiki ); |
| 100 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
| 101 | if ( $site && $this->isExtensionInstalled( $site ) ) { |
| 102 | $sites->setSite( $site ); |
| 103 | } else { |
| 104 | $this->output( "Wiki '$wiki' not found or Phonos isn't installed, skipping...\n" ); |
| 105 | } |
| 106 | } |
| 107 | } else { |
| 108 | $sites = $this->siteStore->getSites(); |
| 109 | } |
| 110 | |
| 111 | if ( $sites->isEmpty() ) { |
| 112 | // 'sites' table is probably not set up. |
| 113 | // Assume this is a MW installation and act only on the current wiki. |
| 114 | $id = WikiMap::getCurrentWikiId(); |
| 115 | $this->output( "sites table is empty, processing only $id...\n" ); |
| 116 | $site = new MediaWikiSite(); |
| 117 | $site->setGlobalId( $id ); |
| 118 | $sites->setSite( $site ); |
| 119 | } |
| 120 | |
| 121 | return $sites; |
| 122 | } |
| 123 | |
| 124 | /** |
| 125 | * Query API:Siteinfo to determine if Phonos is installed on the given Site. |
| 126 | * |
| 127 | * @param MediaWikiSite $site |
| 128 | * @return bool |
| 129 | */ |
| 130 | private function isExtensionInstalled( MediaWikiSite $site ): bool { |
| 131 | $wiki = $site->getGlobalId(); |
| 132 | if ( WikiMap::isCurrentWikiId( $wiki ) ) { |
| 133 | // The API code will error out for local installations since MediaWiki-Docker |
| 134 | // can't talk to localhost as if it were public. Phonos has to be installed |
| 135 | // for the script to be ran anyway, so there's no need to check for the current wiki. |
| 136 | return true; |
| 137 | } |
| 138 | |
| 139 | try { |
| 140 | $apiRoot = $site->getFileUrl( 'api.php' ); |
| 141 | } catch ( RuntimeException $e ) { |
| 142 | $this->fatalError( "file_path not specified in the sites table for wiki '$wiki'.\n" ); |
| 143 | } |
| 144 | |
| 145 | $request = $this->requestFactory->create( |
| 146 | $apiRoot . '?' . http_build_query( [ |
| 147 | 'action' => 'query', |
| 148 | 'meta' => 'siteinfo', |
| 149 | 'siprop' => 'extensions', |
| 150 | 'format' => 'json' |
| 151 | ] ), |
| 152 | [ |
| 153 | 'proxy' => $this->apiProxy, |
| 154 | 'followRedirects' => true |
| 155 | ], |
| 156 | __METHOD__ |
| 157 | ); |
| 158 | $status = $request->execute(); |
| 159 | if ( !$status->isOK() ) { |
| 160 | $msg = $status->getMessage(); |
| 161 | $this->fatalError( "Could not fetch siteinfo for wiki '$wiki': $msg\n" ); |
| 162 | } |
| 163 | |
| 164 | $extensions = json_decode( $request->getContent(), true )['query']['extensions'] ?? []; |
| 165 | foreach ( $extensions as $extension ) { |
| 166 | if ( $extension['name'] === 'Phonos' ) { |
| 167 | return true; |
| 168 | } |
| 169 | } |
| 170 | |
| 171 | return false; |
| 172 | } |
| 173 | |
| 174 | /** |
| 175 | * Query for the 'phonos-files' page property to find all Phonos files that are in-use. |
| 176 | * |
| 177 | * @param MediaWikiSite $site |
| 178 | * @return string[] Paths to the files relative to root storage path with Engine::STORAGE_PREFIX. |
| 179 | */ |
| 180 | private function fetchUsedFiles( MediaWikiSite $site ): array { |
| 181 | $dbr = $this->lbFactory->getReplicaDatabase( $site->getGlobalId() ); |
| 182 | $queryBuilder = $dbr->newSelectQueryBuilder(); |
| 183 | $queryBuilder->select( 'pp_value' ) |
| 184 | ->from( 'page_props' ) |
| 185 | ->where( [ 'pp_propname' => 'phonos-files' ] ); |
| 186 | $props = $queryBuilder->caller( __METHOD__ )->fetchFieldValues(); |
| 187 | return array_unique( array_merge( ...array_map( 'json_decode', $props ) ) ); |
| 188 | } |
| 189 | |
| 190 | /** |
| 191 | * Reports the number of unused files in storage, optionally deleting them as well. |
| 192 | * |
| 193 | * @param array $usedFiles |
| 194 | */ |
| 195 | private function reportUnusedFiles( array $usedFiles ): void { |
| 196 | $this->output( "Finding unused files in storage...\n" ); |
| 197 | $dir = $this->backend->getRootStoragePath() . '/' . Engine::STORAGE_PREFIX; |
| 198 | $filesToDelete = []; |
| 199 | |
| 200 | foreach ( |
| 201 | $this->backend->getFileList( [ 'dir' => $dir, 'adviseStat' => true ] ) as $file |
| 202 | ) { |
| 203 | $slug = basename( $file, '.mp3' ); |
| 204 | if ( !in_array( $slug, $usedFiles ) ) { |
| 205 | $fullPath = $dir . '/' . $file; |
| 206 | $filesToDelete[] = [ 'op' => 'delete', 'src' => $fullPath ]; |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | $count = count( $filesToDelete ); |
| 211 | |
| 212 | if ( $count ) { |
| 213 | $this->output( $count . " unused files found.\n" ); |
| 214 | if ( $this->getOption( 'delete' ) ) { |
| 215 | $this->deleteFiles( $dir, $filesToDelete ); |
| 216 | } |
| 217 | } else { |
| 218 | $this->output( "No unused files found!\n" ); |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | /** |
| 223 | * Delete the given files within the given directory. |
| 224 | * This operation is batched for performance reasons. |
| 225 | * |
| 226 | * @param string $dir |
| 227 | * @param array $files |
| 228 | */ |
| 229 | private function deleteFiles( string $dir, array $files ): void { |
| 230 | $this->output( "Deleting files from storage...\n" ); |
| 231 | $deletedCount = 0; |
| 232 | foreach ( array_chunk( $files, 1000 ) as $chunk ) { |
| 233 | $ret = $this->backend->doQuickOperations( $chunk ); |
| 234 | |
| 235 | if ( $ret->isOK() ) { |
| 236 | $deletedCount += count( $chunk ); |
| 237 | $this->output( "$deletedCount...\n" ); |
| 238 | } else { |
| 239 | $status = Status::wrap( $ret ); |
| 240 | $this->output( "Deleting unused Phonos files errored.\n" ); |
| 241 | $this->fatalError( $status->getWikiText( false, false, 'en' ) ); |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | $this->output( "$deletedCount orphaned Phonos files deleted.\n" ); |
| 246 | |
| 247 | // Remove empty directories. |
| 248 | $ret = $this->backend->clean( [ |
| 249 | 'dir' => $dir, |
| 250 | 'recursive' => true, |
| 251 | ] ); |
| 252 | if ( !$ret->isOK() ) { |
| 253 | $status = Status::wrap( $ret ); |
| 254 | $this->output( "Cleaning empty directories errored.\n" ); |
| 255 | $this->fatalError( $status->getWikiText( false, false, 'en' ) ); |
| 256 | } |
| 257 | } |
| 258 | } |
| 259 | |
| 260 | $maintClass = CountOrphanedFiles::class; |
| 261 | require_once RUN_MAINTENANCE_IF_MAIN; |