Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
95.97% |
119 / 124 |
|
70.00% |
7 / 10 |
CRAP | |
0.00% |
0 / 1 |
BaseInterwikiResolver | |
95.97% |
119 / 124 |
|
70.00% |
7 / 10 |
29 | |
0.00% |
0 / 1 |
__construct | |
100.00% |
5 / 5 |
|
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 MediaWiki\Interwiki\InterwikiLookup; |
6 | use MediaWiki\Logger\LoggerFactory; |
7 | use Wikimedia\Http\MultiHttpClient; |
8 | use Wikimedia\ObjectCache\WANObjectCache; |
9 | |
10 | /** |
11 | * Base InterwikiResolver class. |
12 | * Subclasses just need to provide the full matrix array |
13 | * by implementing loadMatrix(), the resulting matrix will |
14 | * be stored by this base class. |
15 | */ |
16 | abstract class BaseInterwikiResolver implements InterwikiResolver { |
17 | private const CONFIG_CACHE_TTL = 600; |
18 | |
19 | /** @var array[]|null full IW matrix (@see loadMatrix()) */ |
20 | private ?array $matrix = null; |
21 | |
22 | /** @var SearchConfig main wiki config */ |
23 | protected SearchConfig $config; |
24 | |
25 | /** @var bool use cirrus config dump API */ |
26 | private $useConfigDumpApi; |
27 | |
28 | /** |
29 | * @var MultiHttpClient http client to fetch config of other wikis |
30 | */ |
31 | private MultiHttpClient $httpClient; |
32 | |
33 | protected InterwikiLookup $interwikiLookup; |
34 | |
35 | protected WANObjectCache $wanCache; |
36 | |
37 | /** |
38 | * @param SearchConfig $config |
39 | * @param MultiHttpClient $client http client to fetch cirrus config |
40 | * @param WANObjectCache $wanCache Cache object for caching repeated requests |
41 | * @param InterwikiLookup $iwLookup |
42 | */ |
43 | public function __construct( |
44 | SearchConfig $config, |
45 | MultiHttpClient $client, |
46 | WANObjectCache $wanCache, |
47 | InterwikiLookup $iwLookup |
48 | ) { |
49 | $this->config = $config; |
50 | $this->useConfigDumpApi = $this->config->get( 'CirrusSearchFetchConfigFromApi' ); |
51 | $this->httpClient = $client; |
52 | $this->interwikiLookup = $iwLookup; |
53 | $this->wanCache = $wanCache; |
54 | } |
55 | |
56 | /** |
57 | * @return string[] |
58 | */ |
59 | public function getSisterProjectPrefixes() { |
60 | $matrix = $this->getMatrix(); |
61 | return $matrix['sister_projects'] ?? []; |
62 | } |
63 | |
64 | /** |
65 | * @return SearchConfig[] configs of sister project indexed by interwiki prefix |
66 | */ |
67 | public function getSisterProjectConfigs() { |
68 | $prefixes = $this->getSisterProjectPrefixes(); |
69 | return $this->loadConfigFromAPI( $prefixes, [], [ $this, 'minimalSearchConfig' ] ); |
70 | } |
71 | |
72 | /** |
73 | * @param string $wikiId |
74 | * @return string|null |
75 | */ |
76 | public function getInterwikiPrefix( $wikiId ) { |
77 | $matrix = $this->getMatrix(); |
78 | return $matrix['prefixes_by_wiki'][$wikiId] ?? null; |
79 | } |
80 | |
81 | /** |
82 | * @param string $lang |
83 | * @return string[] a two elements array [ 'prefix', 'language' ] |
84 | */ |
85 | public function getSameProjectWikiByLang( $lang ) { |
86 | $matrix = $this->getMatrix(); |
87 | // Most of the time the language is equal to the interwiki prefix. |
88 | // But it's not always the case, use the language_map to identify the interwiki prefix first. |
89 | $lang = $matrix['language_map'][$lang] ?? $lang; |
90 | return isset( $matrix['cross_language'][$lang] ) ? [ $matrix['cross_language'][$lang], $lang ] : []; |
91 | } |
92 | |
93 | /** |
94 | * @param string $lang |
95 | * @return SearchConfig[] single element array: [ interwiki => SearchConfig ] |
96 | */ |
97 | public function getSameProjectConfigByLang( $lang ) { |
98 | $wikiAndPrefix = $this->getSameProjectWikiByLang( $lang ); |
99 | if ( !$wikiAndPrefix ) { |
100 | return []; |
101 | } |
102 | [ $wiki, $prefix ] = $wikiAndPrefix; |
103 | return $this->loadConfigFromAPI( |
104 | [ $prefix => $wiki ], |
105 | [], |
106 | [ $this, 'minimalSearchConfig' ] ); |
107 | } |
108 | |
109 | /** @return array[] */ |
110 | private function getMatrix() { |
111 | if ( $this->matrix === null ) { |
112 | $this->matrix = $this->loadMatrix(); |
113 | |
114 | } |
115 | return $this->matrix; |
116 | } |
117 | |
118 | /** |
119 | * Load the interwiki matric information |
120 | * The returned array must include the following keys: |
121 | * - sister_projects: an array with the list of sister wikis indexed by |
122 | * interwiki prefix |
123 | * - cross_language: an array with the list of wikis running the same |
124 | * project/site indexed by interwiki prefix |
125 | * - language_map: an array with the list of interwiki prefixes where |
126 | * where the language code of the wiki does not match the prefix |
127 | * - prefixes_by_wiki: an array with the list of interwiki indexed |
128 | * by wikiID |
129 | * |
130 | * The result of this method is stored in the current InterwikiResolver instance |
131 | * so it can be called only once per request. |
132 | * |
133 | * return array[] |
134 | */ |
135 | abstract protected function loadMatrix(); |
136 | |
137 | /** |
138 | * @param string[] $wikis |
139 | * @param string[] $hashConfigFlags constructor flags for SearchConfig |
140 | * @param callable $fallbackConfig function to load the config if the |
141 | * api is not usable or if a failure occurs |
142 | * @return SearchConfig[] config indexed by iw prefix |
143 | */ |
144 | private function loadConfigFromAPI( $wikis, array $hashConfigFlags, $fallbackConfig ) { |
145 | $endpoints = []; |
146 | foreach ( $wikis as $prefix => $wiki ) { |
147 | $iw = $this->interwikiLookup->fetch( $prefix ); |
148 | if ( !$iw || !$this->useConfigDumpApi || !$iw->isLocal() ) { |
149 | continue; |
150 | } |
151 | $api = $iw->getAPI(); |
152 | if ( !$api ) { |
153 | $parts = parse_url( $iw->getURL() ); |
154 | if ( !isset( $parts['host'] ) ) { |
155 | continue; |
156 | } |
157 | $api = $parts['scheme'] ?? 'http'; |
158 | $api .= '://' . $parts['host']; |
159 | $api .= isset( $parts['port'] ) ? ':' . $parts['port'] : ''; |
160 | $api .= '/w/api.php'; |
161 | } |
162 | $endpoints[$prefix] = [ 'url' => $api, 'wiki' => $wiki ]; |
163 | } |
164 | |
165 | if ( $endpoints ) { |
166 | $prefixes = array_keys( $endpoints ); |
167 | asort( $prefixes ); |
168 | $cacheKey = implode( '-', $prefixes ); |
169 | $configs = $this->wanCache->getWithSetCallback( |
170 | $this->wanCache->makeKey( 'cirrussearch-load-iw-config', $cacheKey ), |
171 | self::CONFIG_CACHE_TTL, |
172 | function () use ( $endpoints ) { |
173 | return $this->sendConfigDumpRequest( $endpoints ); |
174 | } |
175 | ); |
176 | } else { |
177 | $configs = []; |
178 | } |
179 | $retValue = []; |
180 | foreach ( $wikis as $prefix => $wiki ) { |
181 | if ( isset( $configs[$prefix] ) ) { |
182 | $config = $configs[$prefix]; |
183 | $config['_wikiID'] = $wiki; |
184 | |
185 | $retValue[$prefix] = new HashSearchConfig( |
186 | $config, |
187 | array_merge( $hashConfigFlags, [ HashSearchConfig::FLAG_INHERIT ] ) |
188 | ); |
189 | } else { |
190 | $retValue[$prefix] = $fallbackConfig( $wiki, $hashConfigFlags ); |
191 | } |
192 | } |
193 | return $retValue; |
194 | } |
195 | |
196 | /** |
197 | * @param array[] $endpoints list of arrays containing 'url' and 'wiki', indexed by iw prefix |
198 | * @return array[] list of array containing extracted config vars, failed wikis |
199 | * are not returned. |
200 | */ |
201 | private function sendConfigDumpRequest( $endpoints ) { |
202 | $logger = LoggerFactory::getInstance( 'CirrusSearch' ); |
203 | $reqs = []; |
204 | foreach ( $endpoints as $prefix => $info ) { |
205 | $reqs[$prefix] = [ |
206 | 'method' => 'GET', |
207 | 'url' => $info['url'], |
208 | 'query' => [ |
209 | 'action' => 'cirrus-config-dump', |
210 | 'format' => 'json', |
211 | 'formatversion' => '2', |
212 | ] |
213 | ]; |
214 | } |
215 | if ( !$reqs ) { |
216 | return []; |
217 | } |
218 | $responses = $this->httpClient->runMulti( $reqs ); |
219 | $configs = []; |
220 | foreach ( $responses as $prefix => $response ) { |
221 | if ( $response['response']['code'] !== 200 ) { |
222 | $logger->warning( |
223 | 'Failed to fetch config for {wiki} at {url}: ' . |
224 | 'http status {httpstatus} : {clienterror}', |
225 | [ |
226 | 'wiki' => $endpoints[$prefix]['wiki'], |
227 | 'url' => $endpoints[$prefix]['url'], |
228 | 'httpstatus' => $response['response']['code'], |
229 | 'clienterror' => $response['response']['error'] |
230 | ] |
231 | ); |
232 | continue; |
233 | } |
234 | |
235 | $data = json_decode( $response['response']['body'], true ); |
236 | if ( json_last_error() !== JSON_ERROR_NONE ) { |
237 | $logger->warning( |
238 | 'Failed to fetch config for {wiki} at {url}: ' . |
239 | 'json error code {jsonerrorcode}', |
240 | [ |
241 | 'wiki' => $endpoints[$prefix]['wiki'], |
242 | 'url' => $endpoints[$prefix]['url'], |
243 | 'jsonerrorcode' => json_last_error() |
244 | ] |
245 | ); |
246 | continue; |
247 | } |
248 | |
249 | if ( isset( $data['error'] ) ) { |
250 | $logger->warning( |
251 | 'Failed to fetch config for {wiki} at {url}: {apierrormessage}', |
252 | [ |
253 | 'wiki' => $endpoints[$prefix]['wiki'], |
254 | 'url' => $endpoints[$prefix]['url'], |
255 | 'apierrormessage' => $data['error'] |
256 | ] |
257 | ); |
258 | continue; |
259 | } |
260 | unset( $data['warnings'] ); |
261 | $configs[$prefix] = $data; |
262 | } |
263 | return $configs; |
264 | } |
265 | |
266 | /** |
267 | * Minimal config needed to run a search on a target wiki |
268 | * living on the same cluster as the host wiki |
269 | * |
270 | * @param string $wiki |
271 | * @param string[] $hashConfigFlags constructor flags for HashSearchConfig |
272 | * @return SearchConfig |
273 | */ |
274 | protected function minimalSearchConfig( $wiki, array $hashConfigFlags ) { |
275 | return new HashSearchConfig( |
276 | [ |
277 | '_wikiID' => $wiki, |
278 | 'CirrusSearchIndexBaseName' => $wiki, |
279 | ], |
280 | array_merge( [ HashSearchConfig::FLAG_INHERIT ], $hashConfigFlags ) |
281 | ); |
282 | } |
283 | } |