Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.09% |
123 / 128 |
|
70.00% |
7 / 10 |
CRAP | |
0.00% |
0 / 1 |
| BaseInterwikiResolver | |
96.09% |
123 / 128 |
|
70.00% |
7 / 10 |
29 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
1 | |||
| getSisterProjectPrefixes | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getSisterProjectConfigs | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getInterwikiPrefix | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getSameProjectWikiByLang | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getSameProjectConfigByLang | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
2.01 | |||
| getMatrix | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| loadMatrix | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| loadConfigFromAPI | |
92.11% |
35 / 38 |
|
0.00% |
0 / 1 |
11.06 | |||
| sendConfigDumpRequest | |
98.15% |
53 / 54 |
|
0.00% |
0 / 1 |
7 | |||
| minimalSearchConfig | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch; |
| 4 | |
| 5 | use BagOStuff; |
| 6 | use MediaWiki\Interwiki\InterwikiLookup; |
| 7 | use MediaWiki\Logger\LoggerFactory; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use MultiHttpClient; |
| 10 | use WANObjectCache; |
| 11 | |
| 12 | /** |
| 13 | * Base InterwikiResolver class. |
| 14 | * Subclasses just need to provide the full matrix array |
| 15 | * by implementing loadMatrix(), the resulting matrix will |
| 16 | * be stored by this base class. |
| 17 | */ |
| 18 | abstract class BaseInterwikiResolver implements InterwikiResolver { |
| 19 | private const CONFIG_CACHE_TTL = 600; |
| 20 | |
| 21 | /** @var array[]|null full IW matrix (@see loadMatrix()) */ |
| 22 | private $matrix; |
| 23 | |
| 24 | /** @var SearchConfig main wiki config */ |
| 25 | protected $config; |
| 26 | |
| 27 | /** @var bool use cirrus config dump API */ |
| 28 | private $useConfigDumpApi; |
| 29 | |
| 30 | /** |
| 31 | * @var MultiHttpClient http client to fetch config of other wikis |
| 32 | */ |
| 33 | private $httpClient; |
| 34 | |
| 35 | /** |
| 36 | * @var InterwikiLookup |
| 37 | */ |
| 38 | private $interwikiLookup; |
| 39 | |
| 40 | /** |
| 41 | * @var BagOStuff |
| 42 | */ |
| 43 | private $srvCache; |
| 44 | |
| 45 | /** |
| 46 | * @param SearchConfig $config |
| 47 | * @param \MultiHttpClient|null $client http client to fetch cirrus config |
| 48 | * @param WANObjectCache|null $wanCache Cache object for caching repeated requests |
| 49 | * @param BagOStuff|null $srvCache Local server cache object for caching repeated requests |
| 50 | * @param InterwikiLookup|null $iwLookup |
| 51 | * @throws \Exception |
| 52 | */ |
| 53 | public function __construct( |
| 54 | SearchConfig $config, |
| 55 | MultiHttpClient $client = null, |
| 56 | WANObjectCache $wanCache = null, |
| 57 | BagOStuff $srvCache = null, |
| 58 | InterwikiLookup $iwLookup = null |
| 59 | ) { |
| 60 | $this->config = $config; |
| 61 | $this->useConfigDumpApi = $this->config->get( 'CirrusSearchFetchConfigFromApi' ); |
| 62 | $this->httpClient = $client ?? |
| 63 | MediaWikiServices::getInstance()->getHttpRequestFactory()->createMultiClient( [ |
| 64 | 'connTimeout' => $this->config->get( 'CirrusSearchInterwikiHTTPConnectTimeout' ), |
| 65 | 'reqTimeout' => $this->config->get( 'CirrusSearchInterwikiHTTPTimeout' ) |
| 66 | ] ); |
| 67 | $this->interwikiLookup = $iwLookup ?? MediaWikiServices::getInstance()->getInterwikiLookup(); |
| 68 | $this->srvCache = $srvCache ?? MediaWikiServices::getInstance()->getLocalServerObjectCache(); |
| 69 | } |
| 70 | |
| 71 | /** |
| 72 | * @return string[] |
| 73 | */ |
| 74 | public function getSisterProjectPrefixes() { |
| 75 | $matrix = $this->getMatrix(); |
| 76 | return $matrix['sister_projects'] ?? []; |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * @return SearchConfig[] configs of sister project indexed by interwiki prefix |
| 81 | */ |
| 82 | public function getSisterProjectConfigs() { |
| 83 | $prefixes = $this->getSisterProjectPrefixes(); |
| 84 | return $this->loadConfigFromAPI( $prefixes, [], [ $this, 'minimalSearchConfig' ] ); |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * @param string $wikiId |
| 89 | * @return string|null |
| 90 | */ |
| 91 | public function getInterwikiPrefix( $wikiId ) { |
| 92 | $matrix = $this->getMatrix(); |
| 93 | return $matrix['prefixes_by_wiki'][$wikiId] ?? null; |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * @param string $lang |
| 98 | * @return string[] a two elements array [ 'prefix', 'language' ] |
| 99 | */ |
| 100 | public function getSameProjectWikiByLang( $lang ) { |
| 101 | $matrix = $this->getMatrix(); |
| 102 | // Most of the time the language is equal to the interwiki prefix. |
| 103 | // But it's not always the case, use the language_map to identify the interwiki prefix first. |
| 104 | $lang = $matrix['language_map'][$lang] ?? $lang; |
| 105 | return isset( $matrix['cross_language'][$lang] ) ? [ $matrix['cross_language'][$lang], $lang ] : []; |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * @param string $lang |
| 110 | * @return SearchConfig[] single element array: [ interwiki => SearchConfig ] |
| 111 | */ |
| 112 | public function getSameProjectConfigByLang( $lang ) { |
| 113 | $wikiAndPrefix = $this->getSameProjectWikiByLang( $lang ); |
| 114 | if ( !$wikiAndPrefix ) { |
| 115 | return []; |
| 116 | } |
| 117 | [ $wiki, $prefix ] = $wikiAndPrefix; |
| 118 | return $this->loadConfigFromAPI( |
| 119 | [ $prefix => $wiki ], |
| 120 | [], |
| 121 | [ $this, 'minimalSearchConfig' ] ); |
| 122 | } |
| 123 | |
| 124 | /** @return array[] */ |
| 125 | private function getMatrix() { |
| 126 | if ( $this->matrix === null ) { |
| 127 | $this->matrix = $this->loadMatrix(); |
| 128 | |
| 129 | } |
| 130 | return $this->matrix; |
| 131 | } |
| 132 | |
| 133 | /** |
| 134 | * Load the interwiki matric information |
| 135 | * The returned array must include the following keys: |
| 136 | * - sister_projects: an array with the list of sister wikis indexed by |
| 137 | * interwiki prefix |
| 138 | * - cross_language: an array with the list of wikis running the same |
| 139 | * project/site indexed by interwiki prefix |
| 140 | * - language_map: an array with the list of interwiki prefixes where |
| 141 | * where the language code of the wiki does not match the prefix |
| 142 | * - prefixes_by_wiki: an array with the list of interwiki indexed |
| 143 | * by wikiID |
| 144 | * |
| 145 | * The result of this method is stored in the current InterwikiResolver instance |
| 146 | * so it can be called only once per request. |
| 147 | * |
| 148 | * return array[] |
| 149 | */ |
| 150 | abstract protected function loadMatrix(); |
| 151 | |
| 152 | /** |
| 153 | * @param string[] $wikis |
| 154 | * @param string[] $hashConfigFlags constructor flags for SearchConfig |
| 155 | * @param callable $fallbackConfig function to load the config if the |
| 156 | * api is not usable or if a failure occurs |
| 157 | * @return SearchConfig[] config indexed by iw prefix |
| 158 | */ |
| 159 | private function loadConfigFromAPI( $wikis, array $hashConfigFlags, $fallbackConfig ) { |
| 160 | $endpoints = []; |
| 161 | foreach ( $wikis as $prefix => $wiki ) { |
| 162 | $iw = $this->interwikiLookup->fetch( $prefix ); |
| 163 | if ( !$iw || !$this->useConfigDumpApi || !$iw->isLocal() ) { |
| 164 | continue; |
| 165 | } |
| 166 | $api = $iw->getAPI(); |
| 167 | if ( !$api ) { |
| 168 | $parts = parse_url( $iw->getURL() ); |
| 169 | if ( !isset( $parts['host'] ) ) { |
| 170 | continue; |
| 171 | } |
| 172 | $api = $parts['scheme'] ?? 'http'; |
| 173 | $api .= '://' . $parts['host']; |
| 174 | $api .= isset( $parts['port'] ) ? ':' . $parts['port'] : ''; |
| 175 | $api .= '/w/api.php'; |
| 176 | } |
| 177 | $endpoints[$prefix] = [ 'url' => $api, 'wiki' => $wiki ]; |
| 178 | } |
| 179 | |
| 180 | if ( $endpoints ) { |
| 181 | $prefixes = array_keys( $endpoints ); |
| 182 | asort( $prefixes ); |
| 183 | $cacheKey = implode( '-', $prefixes ); |
| 184 | $configs = $this->srvCache->getWithSetCallback( |
| 185 | $this->srvCache->makeKey( 'cirrussearch-load-iw-config', $cacheKey ), |
| 186 | self::CONFIG_CACHE_TTL, |
| 187 | function () use ( $endpoints ) { |
| 188 | return $this->sendConfigDumpRequest( $endpoints ); |
| 189 | } |
| 190 | ); |
| 191 | } else { |
| 192 | $configs = []; |
| 193 | } |
| 194 | $retValue = []; |
| 195 | foreach ( $wikis as $prefix => $wiki ) { |
| 196 | if ( isset( $configs[$prefix] ) ) { |
| 197 | $config = $configs[$prefix]; |
| 198 | $config['_wikiID'] = $wiki; |
| 199 | |
| 200 | $retValue[$prefix] = new HashSearchConfig( |
| 201 | $config, |
| 202 | array_merge( $hashConfigFlags, [ HashSearchConfig::FLAG_INHERIT ] ) |
| 203 | ); |
| 204 | } else { |
| 205 | $retValue[$prefix] = $fallbackConfig( $wiki, $hashConfigFlags ); |
| 206 | } |
| 207 | } |
| 208 | return $retValue; |
| 209 | } |
| 210 | |
| 211 | /** |
| 212 | * @param array[] $endpoints list of arrays containing 'url' and 'wiki', indexed by iw prefix |
| 213 | * @return array[] list of array containing extracted config vars, failed wikis |
| 214 | * are not returned. |
| 215 | */ |
| 216 | private function sendConfigDumpRequest( $endpoints ) { |
| 217 | $logger = LoggerFactory::getInstance( 'CirrusSearch' ); |
| 218 | $reqs = []; |
| 219 | foreach ( $endpoints as $prefix => $info ) { |
| 220 | $reqs[$prefix] = [ |
| 221 | 'method' => 'GET', |
| 222 | 'url' => $info['url'], |
| 223 | 'query' => [ |
| 224 | 'action' => 'cirrus-config-dump', |
| 225 | 'format' => 'json', |
| 226 | 'formatversion' => '2', |
| 227 | ] |
| 228 | ]; |
| 229 | } |
| 230 | if ( !$reqs ) { |
| 231 | return []; |
| 232 | } |
| 233 | $responses = $this->httpClient->runMulti( $reqs ); |
| 234 | $configs = []; |
| 235 | foreach ( $responses as $prefix => $response ) { |
| 236 | if ( $response['response']['code'] !== 200 ) { |
| 237 | $logger->warning( |
| 238 | 'Failed to fetch config for {wiki} at {url}: ' . |
| 239 | 'http status {httpstatus} : {clienterror}', |
| 240 | [ |
| 241 | 'wiki' => $endpoints[$prefix]['wiki'], |
| 242 | 'url' => $endpoints[$prefix]['url'], |
| 243 | 'httpstatus' => $response['response']['code'], |
| 244 | 'clienterror' => $response['response']['error'] |
| 245 | ] |
| 246 | ); |
| 247 | continue; |
| 248 | } |
| 249 | |
| 250 | $data = json_decode( $response['response']['body'], true ); |
| 251 | if ( json_last_error() !== JSON_ERROR_NONE ) { |
| 252 | $logger->warning( |
| 253 | 'Failed to fetch config for {wiki} at {url}: ' . |
| 254 | 'json error code {jsonerrorcode}', |
| 255 | [ |
| 256 | 'wiki' => $endpoints[$prefix]['wiki'], |
| 257 | 'url' => $endpoints[$prefix]['url'], |
| 258 | 'jsonerrorcode' => json_last_error() |
| 259 | ] |
| 260 | ); |
| 261 | continue; |
| 262 | } |
| 263 | |
| 264 | if ( isset( $data['error'] ) ) { |
| 265 | $logger->warning( |
| 266 | 'Failed to fetch config for {wiki} at {url}: {apierrormessage}', |
| 267 | [ |
| 268 | 'wiki' => $endpoints[$prefix]['wiki'], |
| 269 | 'url' => $endpoints[$prefix]['url'], |
| 270 | 'apierrormessage' => $data['error'] |
| 271 | ] |
| 272 | ); |
| 273 | continue; |
| 274 | } |
| 275 | unset( $data['warnings'] ); |
| 276 | $configs[$prefix] = $data; |
| 277 | } |
| 278 | return $configs; |
| 279 | } |
| 280 | |
| 281 | /** |
| 282 | * Minimal config needed to run a search on a target wiki |
| 283 | * living on the same cluster as the host wiki |
| 284 | * |
| 285 | * @param string $wiki |
| 286 | * @param string[] $hashConfigFlags constructor flags for HashSearchConfig |
| 287 | * @return SearchConfig |
| 288 | */ |
| 289 | protected function minimalSearchConfig( $wiki, array $hashConfigFlags ) { |
| 290 | return new HashSearchConfig( |
| 291 | [ |
| 292 | '_wikiID' => $wiki, |
| 293 | 'CirrusSearchIndexBaseName' => $wiki, |
| 294 | ], |
| 295 | array_merge( [ HashSearchConfig::FLAG_INHERIT ], $hashConfigFlags ) |
| 296 | ); |
| 297 | } |
| 298 | } |