Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
81.07% |
257 / 317 |
|
68.97% |
40 / 58 |
CRAP | |
0.00% |
0 / 1 |
SiteConfig | |
81.07% |
257 / 317 |
|
68.97% |
40 / 58 |
189.56 | |
0.00% |
0 / 1 |
__construct | |
53.33% |
8 / 15 |
|
0.00% |
0 / 1 |
5.63 | |||
reset | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
combineRegexArrays | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
4.05 | |||
addNamespace | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
detectFeatures | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
hasVideoInfo | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
loadSiteData | |
97.67% |
84 / 86 |
|
0.00% |
0 / 1 |
21 | |||
galleryOptions | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
allowedExternalImagePrefixes | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
determineArticlePath | |
84.21% |
16 / 19 |
|
0.00% |
0 / 1 |
4.06 | |||
baseURI | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
relativeLinkPrefix | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
canonicalNamespaceId | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
namespaceId | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
namespaceName | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
namespaceHasSubpages | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
namespaceCase | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
specialPageLocalName | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
magicLinkEnabled | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
interwikiMagic | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
interwikiMap | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
iwp | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
legalTitleChars | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
linkPrefixRegex | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
linkTrail | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
langBcp47 | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
mainpage | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
mainPageLinkTarget | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
getMWConfigValue | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
4 | |||
rtl | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
langConverterEnabledBcp47 | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
script | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
scriptpath | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
server | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
exportMetadataToHeadBcp47 | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
2 | |||
redirectRegexp | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
categoryRegexp | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
bswRegexp | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
timezoneOffset | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
variantsFor | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
widthOption | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
getVariableIDs | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
haveComputedFunctionSynonyms | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
updateFunctionSynonym | |
96.55% |
28 / 29 |
|
0.00% |
0 / 1 |
6 | |||
getMagicWords | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
getMagicWordMatcher | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
getParameterizedAliasMatcher | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
5 | |||
ensureExtensionTag | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
getNonNativeExtensionTags | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
getMaxTemplateDepth | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
getSpecialNSAliases | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
getSpecialPageAliases | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
getProtocols | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
metrics | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
incrementCounter | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
observeTiming | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
getNoFollowConfig | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
getExternalLinkTarget | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 |
1 | <?php |
2 | |
3 | declare( strict_types = 1 ); |
4 | |
5 | namespace Wikimedia\Parsoid\Config\Api; |
6 | |
7 | use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface; |
8 | use Wikimedia\Bcp47Code\Bcp47Code; |
9 | use Wikimedia\Parsoid\Config\SiteConfig as ISiteConfig; |
10 | use Wikimedia\Parsoid\Config\StubMetadataCollector; |
11 | use Wikimedia\Parsoid\Core\ContentMetadataCollector; |
12 | use Wikimedia\Parsoid\DOM\Document; |
13 | use Wikimedia\Parsoid\Mocks\MockMetrics; |
14 | use Wikimedia\Parsoid\Utils\ConfigUtils; |
15 | use Wikimedia\Parsoid\Utils\PHPUtils; |
16 | use Wikimedia\Parsoid\Utils\Title; |
17 | use Wikimedia\Parsoid\Utils\UrlUtils; |
18 | use Wikimedia\Parsoid\Utils\Utils; |
19 | |
20 | /** |
21 | * SiteConfig via MediaWiki's Action API |
22 | * |
23 | * Note this is intended for testing, not performance. |
24 | */ |
25 | class SiteConfig extends ISiteConfig { |
26 | |
27 | /** @var ApiHelper */ |
28 | private $api; |
29 | |
30 | /** @var array|null */ |
31 | private $siteData; |
32 | |
33 | /** @var array|null */ |
34 | private $protocols; |
35 | |
36 | /** @var string|null */ |
37 | private $baseUri; |
38 | |
39 | /** @var string|null */ |
40 | private $relativeLinkPrefix; |
41 | |
42 | /** @var string */ |
43 | private $savedCategoryRegexp; |
44 | |
45 | /** @var string */ |
46 | private $savedRedirectRegexp; |
47 | |
48 | /** @var string */ |
49 | private $savedBswRegexp; |
50 | |
51 | /** @var array<int,string> */ |
52 | protected $nsNames = []; |
53 | |
54 | /** @var array<int,string> */ |
55 | protected $nsCase = []; |
56 | |
57 | /** @var array<string,int> */ |
58 | protected $nsIds = []; |
59 | |
60 | /** @var array<string,int> */ |
61 | protected $nsCanon = []; |
62 | |
63 | /** @var array<int,bool> */ |
64 | protected $nsWithSubpages = []; |
65 | |
66 | /** @var array<string,string> */ |
67 | private $specialPageNames = []; |
68 | |
69 | /** @var array */ |
70 | private $specialPageAliases = []; |
71 | |
72 | /** @var array|null */ |
73 | private $interwikiMap; |
74 | |
75 | /** @var array<string,array>|null Keys are stored as lowercased BCP-47 code strings */ |
76 | private $variants; |
77 | |
78 | /** @var array<string,bool>|null Keys are stored as lowercased BCP-47 code strings */ |
79 | private $langConverterEnabled; |
80 | |
81 | /** @var array|null */ |
82 | private $apiMagicWords; |
83 | |
84 | /** @var array|null */ |
85 | private $paramMWs; |
86 | |
87 | /** @var array|null */ |
88 | private $apiVariables; |
89 | |
90 | /** @var array|null */ |
91 | private $apiFunctionHooks; |
92 | |
93 | /** @var array|null */ |
94 | private $allMWs; |
95 | |
96 | /** @var array|null */ |
97 | private $extensionTags; |
98 | |
99 | /** @var int|null */ |
100 | private $widthOption; |
101 | |
102 | /** @var int */ |
103 | private $maxDepth = 40; |
104 | |
105 | private $featureDetectionDone = false; |
106 | private $hasVideoInfo = false; |
107 | |
108 | /** @var string[] Base parameters for a siteinfo query */ |
109 | public const SITE_CONFIG_QUERY_PARAMS = [ |
110 | 'action' => 'query', |
111 | 'meta' => 'siteinfo', |
112 | 'siprop' => 'general|protocols|namespaces|namespacealiases|magicwords|interwikimap|' |
113 | . 'languagevariants|defaultoptions|specialpagealiases|extensiontags|' |
114 | . 'functionhooks|variables', |
115 | ]; |
116 | |
117 | public function __construct( ApiHelper $api, array $opts ) { |
118 | parent::__construct(); |
119 | |
120 | $this->api = $api; |
121 | |
122 | $this->linterEnabled = (bool)( $opts['linting'] ?? false ); |
123 | $this->addHTMLTemplateParameters = (bool)( $opts['addHTMLTemplateParameters'] ?? false ); |
124 | |
125 | if ( isset( $opts['maxDepth'] ) ) { |
126 | $this->maxDepth = (int)$opts['maxDepth']; |
127 | } |
128 | |
129 | $this->setLogger( $opts['logger'] ?? self::createLogger() ); |
130 | |
131 | if ( isset( $opts['wt2htmlLimits'] ) ) { |
132 | $this->wt2htmlLimits = array_merge( |
133 | $this->wt2htmlLimits, $opts['wt2htmlLimits'] |
134 | ); |
135 | } |
136 | if ( isset( $opts['html2wtLimits'] ) ) { |
137 | $this->html2wtLimits = array_merge( |
138 | $this->html2wtLimits, $opts['html2wtLimits'] |
139 | ); |
140 | } |
141 | } |
142 | |
143 | protected function reset() { |
144 | $this->siteData = null; |
145 | $this->baseUri = null; |
146 | $this->relativeLinkPrefix = null; |
147 | // Superclass value reset since parsertests reuse SiteConfig objects |
148 | $this->linkTrailRegex = false; |
149 | $this->mwAliases = null; |
150 | $this->interwikiMapNoNamespaces = null; |
151 | $this->iwMatcher = null; |
152 | } |
153 | |
154 | /** |
155 | * Combine sets of regex fragments |
156 | * @param string[][] $res |
157 | * - $regexes[0] are case-insensitive regex fragments. Must not be empty. |
158 | * - $regexes[1] are case-sensitive regex fragments. Must not be empty. |
159 | * @return string Combined regex fragment. May be an alternation. Assumes |
160 | * the outer environment is case-sensitive. |
161 | */ |
162 | private function combineRegexArrays( array $res ): string { |
163 | if ( $res ) { |
164 | if ( isset( $res[0] ) ) { |
165 | $res[0] = '(?i:' . implode( '|', $res[0] ) . ')'; |
166 | } |
167 | if ( isset( $res[1] ) ) { |
168 | $res[1] = '(?:' . implode( '|', $res[1] ) . ')'; |
169 | } |
170 | return implode( '|', $res ); |
171 | } |
172 | // None? Return a failing regex |
173 | return '(?!)'; |
174 | } |
175 | |
176 | /** |
177 | * Add a new namespace to the config |
178 | * |
179 | * Protected access to let mocks and parser tests versions |
180 | * add new namespaces as required. |
181 | * |
182 | * @param array $ns Namespace info |
183 | */ |
184 | protected function addNamespace( array $ns ): void { |
185 | $id = (int)$ns['id']; |
186 | $this->nsNames[$id] = $ns['name']; |
187 | $this->nsIds[Utils::normalizeNamespaceName( $ns['name'] )] = $id; |
188 | $this->nsCanon[Utils::normalizeNamespaceName( $ns['canonical'] ?? $ns['name'] )] = $id; |
189 | if ( $ns['subpages'] ) { |
190 | $this->nsWithSubpages[$id] = true; |
191 | } |
192 | $this->nsCase[$id] = (string)$ns['case']; |
193 | } |
194 | |
195 | private function detectFeatures(): void { |
196 | if ( !$this->featureDetectionDone ) { |
197 | $this->featureDetectionDone = true; |
198 | $data = $this->api->makeRequest( [ 'action' => 'paraminfo', 'modules' => 'query' ] ); |
199 | $props = $data["paraminfo"]["modules"][0]["parameters"]["0"]["type"] ?? []; |
200 | $this->hasVideoInfo = in_array( 'videoinfo', $props, true ); |
201 | } |
202 | } |
203 | |
204 | public function hasVideoInfo(): bool { |
205 | $this->detectFeatures(); |
206 | return $this->hasVideoInfo; |
207 | } |
208 | |
209 | /** |
210 | * Load site data from the Action API, if necessary |
211 | */ |
212 | private function loadSiteData(): void { |
213 | if ( $this->siteData !== null ) { |
214 | return; |
215 | } |
216 | |
217 | $data = $this->api->makeRequest( self::SITE_CONFIG_QUERY_PARAMS )['query']; |
218 | |
219 | $this->siteData = $data['general']; |
220 | $this->widthOption = $data['general']['thumblimits'][$data['defaultoptions']['thumbsize']]; |
221 | $this->protocols = $data['protocols']; |
222 | $this->apiVariables = $data['variables']; |
223 | $this->apiFunctionHooks = PHPUtils::makeSet( $data['functionhooks'] ); |
224 | |
225 | // Process namespace data from API |
226 | $this->nsNames = []; |
227 | $this->nsCase = []; |
228 | $this->nsIds = []; |
229 | $this->nsCanon = []; |
230 | $this->nsWithSubpages = []; |
231 | foreach ( $data['namespaces'] as $ns ) { |
232 | $this->addNamespace( $ns ); |
233 | } |
234 | foreach ( $data['namespacealiases'] as $ns ) { |
235 | $this->nsIds[Utils::normalizeNamespaceName( $ns['alias'] )] = $ns['id']; |
236 | } |
237 | |
238 | // Process magic word data from API |
239 | $bsws = []; |
240 | $this->paramMWs = []; |
241 | $this->allMWs = []; |
242 | |
243 | // Recast the API results in the format that core MediaWiki returns internally |
244 | // This enables us to use the Production SiteConfig without changes and add the |
245 | // extra overhead to this developer API usage. |
246 | $this->apiMagicWords = []; |
247 | foreach ( $data['magicwords'] as $mw ) { |
248 | $cs = (int)$mw['case-sensitive']; |
249 | $mwName = $mw['name']; |
250 | $this->apiMagicWords[$mwName][] = $cs; |
251 | $pmws = []; |
252 | $allMWs = []; |
253 | foreach ( $mw['aliases'] as $alias ) { |
254 | $this->apiMagicWords[$mwName][] = $alias; |
255 | // Aliases for double underscore mws include the underscores |
256 | if ( substr( $alias, 0, 2 ) === '__' && substr( $alias, -2 ) === '__' ) { |
257 | $bsws[$cs][] = preg_quote( substr( $alias, 2, -2 ), '@' ); |
258 | } |
259 | if ( strpos( $alias, '$1' ) !== false ) { |
260 | $pmws[$cs][] = strtr( preg_quote( $alias, '/' ), [ '\\$1' => "(.*?)" ] ); |
261 | } |
262 | $allMWs[$cs][] = preg_quote( $alias, '/' ); |
263 | } |
264 | |
265 | if ( $pmws ) { |
266 | $this->paramMWs[$mwName] = '/^(?:' . $this->combineRegexArrays( $pmws ) . ')$/uDS'; |
267 | } |
268 | $this->allMWs[$mwName] = '/^(?:' . $this->combineRegexArrays( $allMWs ) . ')$/D'; |
269 | } |
270 | |
271 | $bswRegexp = $this->combineRegexArrays( $bsws ); |
272 | |
273 | // Parse interwiki map data from the API |
274 | $this->interwikiMap = ConfigUtils::computeInterwikiMap( $data['interwikimap'] ); |
275 | |
276 | // Parse variant data from the API |
277 | # T320662: API should return these in BCP-47 forms |
278 | $this->langConverterEnabled = []; |
279 | $this->variants = []; |
280 | foreach ( $data['languagevariants'] as $base => $variants ) { |
281 | $baseBcp47 = Utils::mwCodeToBcp47( $base ); |
282 | if ( $this->siteData['langconversion'] ) { |
283 | $baseKey = strtolower( $baseBcp47->toBcp47Code() ); |
284 | $this->langConverterEnabled[$baseKey] = true; |
285 | foreach ( $variants as $code => $vdata ) { |
286 | $variantKey = strtolower( Utils::mwCodeToBcp47( $code )->toBcp47Code() ); |
287 | $this->variants[$variantKey] = [ |
288 | 'base' => $baseBcp47, |
289 | 'fallbacks' => array_map( |
290 | [ Utils::class, 'mwCodeToBcp47' ], |
291 | $vdata['fallbacks'] |
292 | ), |
293 | ]; |
294 | } |
295 | } |
296 | } |
297 | |
298 | // Parse extension tag data from the API |
299 | $this->extensionTags = []; |
300 | foreach ( $data['extensiontags'] as $tag ) { |
301 | $tag = preg_replace( '/^<|>$/D', '', $tag ); |
302 | $this->ensureExtensionTag( $tag ); |
303 | } |
304 | |
305 | $this->specialPageAliases = $data['specialpagealiases']; |
306 | $this->specialPageNames = []; |
307 | foreach ( $this->specialPageAliases as $special ) { |
308 | $alias = strtr( mb_strtoupper( $special['realname'] ), ' ', '_' ); |
309 | $this->specialPageNames[$alias] = $special['aliases'][0]; |
310 | foreach ( $special['aliases'] as $alias ) { |
311 | $alias = strtr( mb_strtoupper( $alias ), ' ', '_' ); |
312 | $this->specialPageNames[$alias] = $special['aliases'][0]; |
313 | } |
314 | } |
315 | |
316 | $redirect = '(?i:\#REDIRECT)'; |
317 | $quote = static function ( $s ) { |
318 | $q = preg_quote( $s, '@' ); |
319 | # Note that PHP < 7.3 doesn't escape # in preg_quote. That means |
320 | # that the $redirect regexp will fail if used with the `x` flag. |
321 | # Manually hack around this for PHP 7.2; can remove this workaround |
322 | # once minimum PHP version >= 7.3 |
323 | if ( preg_quote( '#' ) === '#' ) { |
324 | $q = str_replace( '#', '\\#', $q ); |
325 | } |
326 | return $q; |
327 | }; |
328 | foreach ( $data['magicwords'] as $mw ) { |
329 | if ( $mw['name'] === 'redirect' ) { |
330 | $redirect = implode( '|', array_map( $quote, $mw['aliases'] ) ); |
331 | if ( !$mw['case-sensitive'] ) { |
332 | $redirect = '(?i:' . $redirect . ')'; |
333 | } |
334 | break; |
335 | } |
336 | } |
337 | // `$this->nsNames[14]` is set earlier by the calls to `$this->addNamespace( $ns )` |
338 | // @phan-suppress-next-line PhanCoalescingAlwaysNull |
339 | $category = $this->quoteTitleRe( $this->nsNames[14] ?? 'Category', '@' ); |
340 | if ( $category !== 'Category' ) { |
341 | $category = "(?:$category|Category)"; |
342 | } |
343 | |
344 | $this->savedCategoryRegexp = "@{$category}@"; |
345 | $this->savedRedirectRegexp = "@{$redirect}@"; |
346 | $this->savedBswRegexp = "@{$bswRegexp}@"; |
347 | } |
348 | |
349 | public function galleryOptions(): array { |
350 | $this->loadSiteData(); |
351 | return $this->siteData['galleryoptions']; |
352 | } |
353 | |
354 | public function allowedExternalImagePrefixes(): array { |
355 | $this->loadSiteData(); |
356 | return $this->siteData['externalimages'] ?? []; |
357 | } |
358 | |
359 | /** |
360 | * Determine the article base URI and relative prefix |
361 | */ |
362 | private function determineArticlePath(): void { |
363 | $this->loadSiteData(); |
364 | |
365 | $url = $this->siteData['server'] . $this->siteData['articlepath']; |
366 | |
367 | if ( substr( $url, -2 ) !== '$1' ) { |
368 | throw new \UnexpectedValueException( "Article path '$url' does not have '$1' at the end" ); |
369 | } |
370 | $url = substr( $url, 0, -2 ); |
371 | |
372 | $bits = UrlUtils::parseUrl( $url ); |
373 | if ( !$bits ) { |
374 | throw new \UnexpectedValueException( "Failed to parse article path '$url'" ); |
375 | } |
376 | |
377 | if ( empty( $bits['path'] ) ) { |
378 | $path = '/'; |
379 | } else { |
380 | $path = UrlUtils::removeDotSegments( $bits['path'] ); |
381 | } |
382 | |
383 | $relParts = [ 'query' => true, 'fragment' => true ]; |
384 | $base = array_diff_key( $bits, $relParts ); |
385 | $rel = array_intersect_key( $bits, $relParts ); |
386 | |
387 | $i = strrpos( $path, '/' ); |
388 | $base['path'] = substr( $path, 0, $i + 1 ); |
389 | $rel['path'] = '.' . substr( $path, $i ); |
390 | |
391 | $this->baseUri = UrlUtils::assembleUrl( $base ); |
392 | $this->relativeLinkPrefix = UrlUtils::assembleUrl( $rel ); |
393 | } |
394 | |
395 | public function baseURI(): string { |
396 | if ( $this->baseUri === null ) { |
397 | $this->determineArticlePath(); |
398 | } |
399 | return $this->baseUri; |
400 | } |
401 | |
402 | public function relativeLinkPrefix(): string { |
403 | if ( $this->relativeLinkPrefix === null ) { |
404 | $this->determineArticlePath(); |
405 | } |
406 | return $this->relativeLinkPrefix; |
407 | } |
408 | |
409 | /** @inheritDoc */ |
410 | public function canonicalNamespaceId( string $name ): ?int { |
411 | $this->loadSiteData(); |
412 | return $this->nsCanon[Utils::normalizeNamespaceName( $name )] ?? null; |
413 | } |
414 | |
415 | /** @inheritDoc */ |
416 | public function namespaceId( string $name ): ?int { |
417 | $this->loadSiteData(); |
418 | $name = Utils::normalizeNamespaceName( $name ); |
419 | return $this->nsCanon[$name] ?? $this->nsIds[$name] ?? null; |
420 | } |
421 | |
422 | /** @inheritDoc */ |
423 | public function namespaceName( int $ns ): ?string { |
424 | $this->loadSiteData(); |
425 | return $this->nsNames[$ns] ?? null; |
426 | } |
427 | |
428 | /** @inheritDoc */ |
429 | public function namespaceHasSubpages( int $ns ): bool { |
430 | $this->loadSiteData(); |
431 | return $this->nsWithSubpages[$ns] ?? false; |
432 | } |
433 | |
434 | /** @inheritDoc */ |
435 | public function namespaceCase( int $ns ): string { |
436 | $this->loadSiteData(); |
437 | return $this->nsCase[$ns] ?? 'first-letter'; |
438 | } |
439 | |
440 | /** @inheritDoc */ |
441 | public function specialPageLocalName( string $alias ): ?string { |
442 | $this->loadSiteData(); |
443 | $alias = strtr( mb_strtoupper( $alias ), ' ', '_' ); |
444 | return $this->specialPageNames[$alias] ?? null; |
445 | } |
446 | |
447 | /** @inheritDoc */ |
448 | public function magicLinkEnabled( string $which ): bool { |
449 | $this->loadSiteData(); |
450 | $magic = $this->siteData['magiclinks'] ?? []; |
451 | // Default to true, as wikis too old to export the 'magiclinks' |
452 | // property always had magic links enabled. |
453 | return $magic[$which] ?? true; |
454 | } |
455 | |
456 | public function interwikiMagic(): bool { |
457 | $this->loadSiteData(); |
458 | return $this->siteData['interwikimagic']; |
459 | } |
460 | |
461 | public function interwikiMap(): array { |
462 | $this->loadSiteData(); |
463 | return $this->interwikiMap; |
464 | } |
465 | |
466 | public function iwp(): string { |
467 | $this->loadSiteData(); |
468 | return $this->siteData['wikiid']; |
469 | } |
470 | |
471 | public function legalTitleChars(): string { |
472 | $this->loadSiteData(); |
473 | return $this->siteData['legaltitlechars']; |
474 | } |
475 | |
476 | public function linkPrefixRegex(): ?string { |
477 | $this->loadSiteData(); |
478 | |
479 | if ( !empty( $this->siteData['linkprefixcharset'] ) ) { |
480 | return '/[' . $this->siteData['linkprefixcharset'] . ']+$/Du'; |
481 | } else { |
482 | // We don't care about super-old MediaWiki, so don't try to parse 'linkprefix'. |
483 | return null; |
484 | } |
485 | } |
486 | |
487 | /** @inheritDoc */ |
488 | protected function linkTrail(): string { |
489 | $this->loadSiteData(); |
490 | return $this->siteData['linktrail']; |
491 | } |
492 | |
493 | public function langBcp47(): Bcp47Code { |
494 | $this->loadSiteData(); |
495 | return Utils::mwCodeToBcp47( $this->siteData['lang'] ); |
496 | } |
497 | |
498 | public function mainpage(): string { |
499 | $this->loadSiteData(); |
500 | return $this->siteData['mainpage']; |
501 | } |
502 | |
503 | public function mainPageLinkTarget(): Title { |
504 | $this->loadSiteData(); |
505 | return Title::newFromText( $this->siteData['mainpage'], $this ); |
506 | } |
507 | |
508 | /** @inheritDoc */ |
509 | public function getMWConfigValue( string $key ) { |
510 | $this->loadSiteData(); |
511 | switch ( $key ) { |
512 | // Hardcoded values for these 2 keys |
513 | case 'CiteResponsiveReferences': |
514 | return $this->siteData['citeresponsivereferences'] ?? false; |
515 | |
516 | case 'CiteResponsiveReferencesThreshold': |
517 | return 10; |
518 | |
519 | // We can add more hardcoded keys based on testing needs |
520 | // but null is the default for keys unsupported in this mode. |
521 | default: |
522 | return null; |
523 | } |
524 | } |
525 | |
526 | public function rtl(): bool { |
527 | $this->loadSiteData(); |
528 | return $this->siteData['rtl']; |
529 | } |
530 | |
531 | /** @inheritDoc */ |
532 | public function langConverterEnabledBcp47( Bcp47Code $lang ): bool { |
533 | $this->loadSiteData(); |
534 | return $this->langConverterEnabled[strtolower( $lang->toBcp47Code() )] ?? false; |
535 | } |
536 | |
537 | public function script(): string { |
538 | $this->loadSiteData(); |
539 | return $this->siteData['script']; |
540 | } |
541 | |
542 | public function scriptpath(): string { |
543 | $this->loadSiteData(); |
544 | return $this->siteData['scriptpath']; |
545 | } |
546 | |
547 | public function server(): string { |
548 | $this->loadSiteData(); |
549 | return $this->siteData['server']; |
550 | } |
551 | |
552 | /** |
553 | * @inheritDoc |
554 | */ |
555 | public function exportMetadataToHeadBcp47( |
556 | Document $document, |
557 | ContentMetadataCollector $metadata, |
558 | string $defaultTitle, |
559 | Bcp47Code $lang |
560 | ): void { |
561 | '@phan-var StubMetadataCollector $metadata'; // @var StubMetadataCollector $metadata |
562 | $moduleLoadURI = $this->server() . $this->scriptpath() . '/load.php'; |
563 | // Parsoid/JS always made this protocol-relative, so match |
564 | // that (for now at least) |
565 | $moduleLoadURI = preg_replace( '#^https?://#', '//', $moduleLoadURI ); |
566 | // Look for a displaytitle. |
567 | $displayTitle = $metadata->getPageProperty( 'displaytitle' ) ?? |
568 | // Use the default title, properly escaped |
569 | Utils::escapeHtml( $defaultTitle ); |
570 | $this->exportMetadataHelper( |
571 | $document, |
572 | $moduleLoadURI, |
573 | $metadata->getModules(), |
574 | $metadata->getModuleStyles(), |
575 | $metadata->getJsConfigVars(), |
576 | $displayTitle, |
577 | $lang |
578 | ); |
579 | } |
580 | |
581 | public function redirectRegexp(): string { |
582 | $this->loadSiteData(); |
583 | return $this->savedRedirectRegexp; |
584 | } |
585 | |
586 | public function categoryRegexp(): string { |
587 | $this->loadSiteData(); |
588 | return $this->savedCategoryRegexp; |
589 | } |
590 | |
591 | public function bswRegexp(): string { |
592 | $this->loadSiteData(); |
593 | return $this->savedBswRegexp; |
594 | } |
595 | |
596 | public function timezoneOffset(): int { |
597 | $this->loadSiteData(); |
598 | return $this->siteData['timeoffset']; |
599 | } |
600 | |
601 | /** @inheritDoc */ |
602 | public function variantsFor( Bcp47Code $lang ): ?array { |
603 | $this->loadSiteData(); |
604 | return $this->variants[strtolower( $lang->toBcp47Code() )] ?? null; |
605 | } |
606 | |
607 | public function widthOption(): int { |
608 | $this->loadSiteData(); |
609 | return $this->widthOption; |
610 | } |
611 | |
612 | /** @inheritDoc */ |
613 | protected function getVariableIDs(): array { |
614 | $this->loadSiteData(); |
615 | return $this->apiVariables; |
616 | } |
617 | |
618 | /** @inheritDoc */ |
619 | protected function haveComputedFunctionSynonyms(): bool { |
620 | return false; |
621 | } |
622 | |
623 | private static $noHashFunctions = null; |
624 | |
625 | /** @inheritDoc */ |
626 | protected function updateFunctionSynonym( string $func, string $magicword, bool $caseSensitive ): void { |
627 | if ( !$this->apiFunctionHooks ) { |
628 | $this->loadSiteData(); |
629 | } |
630 | if ( isset( $this->apiFunctionHooks[$magicword] ) ) { |
631 | if ( !self::$noHashFunctions ) { |
632 | // FIXME: This is an approximation only computed in non-integrated mode for |
633 | // commandline and developer testing. This set is probably not up to date |
634 | // and also doesn't reflect no-hash functions registered by extensions |
635 | // via setFunctionHook calls. As such, you might run into GOTCHAs during |
636 | // debugging of production issues in standalone / API config mode. |
637 | self::$noHashFunctions = PHPUtils::makeSet( [ |
638 | 'ns', 'nse', 'urlencode', 'lcfirst', 'ucfirst', 'lc', 'uc', |
639 | 'localurl', 'localurle', 'fullurl', 'fullurle', 'canonicalurl', |
640 | 'canonicalurle', 'formatnum', 'grammar', 'gender', 'plural', 'bidi', |
641 | 'numberofpages', 'numberofusers', 'numberofactiveusers', |
642 | 'numberofarticles', 'numberoffiles', 'numberofadmins', |
643 | 'numberingroup', 'numberofedits', 'language', |
644 | 'padleft', 'padright', 'anchorencode', 'defaultsort', 'filepath', |
645 | 'pagesincategory', 'pagesize', 'protectionlevel', 'protectionexpiry', |
646 | 'namespacee', 'namespacenumber', 'talkspace', 'talkspacee', |
647 | 'subjectspace', 'subjectspacee', 'pagename', 'pagenamee', |
648 | 'fullpagename', 'fullpagenamee', 'rootpagename', 'rootpagenamee', |
649 | 'basepagename', 'basepagenamee', 'subpagename', 'subpagenamee', |
650 | 'talkpagename', 'talkpagenamee', 'subjectpagename', |
651 | 'subjectpagenamee', 'pageid', 'revisionid', 'revisionday', |
652 | 'revisionday2', 'revisionmonth', 'revisionmonth1', 'revisionyear', |
653 | 'revisiontimestamp', 'revisionuser', 'cascadingsources', |
654 | // Special callbacks in core |
655 | 'namespace', 'int', 'displaytitle', 'pagesinnamespace', |
656 | ] ); |
657 | } |
658 | |
659 | $syn = $func; |
660 | if ( substr( $syn, -1 ) === ':' ) { |
661 | $syn = substr( $syn, 0, -1 ); |
662 | } |
663 | if ( !isset( self::$noHashFunctions[$magicword] ) ) { |
664 | $syn = '#' . $syn; |
665 | } |
666 | $this->functionSynonyms[intval( $caseSensitive )][$syn] = $magicword; |
667 | } |
668 | } |
669 | |
670 | /** @inheritDoc */ |
671 | protected function getMagicWords(): array { |
672 | $this->loadSiteData(); |
673 | return $this->apiMagicWords; |
674 | } |
675 | |
676 | /** @inheritDoc */ |
677 | public function getMagicWordMatcher( string $id ): string { |
678 | $this->loadSiteData(); |
679 | return $this->allMWs[$id] ?? '/^(?!)$/'; |
680 | } |
681 | |
682 | /** @inheritDoc */ |
683 | public function getParameterizedAliasMatcher( array $words ): callable { |
684 | $this->loadSiteData(); |
685 | $regexes = array_intersect_key( $this->paramMWs, array_flip( $words ) ); |
686 | return static function ( $text ) use ( $regexes ) { |
687 | /** |
688 | * $name is the canonical magic word name |
689 | * $re has patterns for matching aliases |
690 | */ |
691 | foreach ( $regexes as $name => $re ) { |
692 | if ( preg_match( $re, $text, $m ) ) { |
693 | unset( $m[0] ); |
694 | |
695 | // Ex. regexp here is, /^(?:(?:|vinculo\=(.*?)|enlace\=(.*?)|link\=(.*?)))$/uS |
696 | // Check all the capture groups for a value, if not, it's safe to return an |
697 | // empty string since we did get a match. |
698 | foreach ( $m as $v ) { |
699 | if ( $v !== '' ) { |
700 | return [ 'k' => $name, 'v' => $v ]; |
701 | } |
702 | } |
703 | return [ 'k' => $name, 'v' => '' ]; |
704 | } |
705 | } |
706 | return null; |
707 | }; |
708 | } |
709 | |
710 | /** |
711 | * This function is public so it can be used to synchronize env for |
712 | * hybrid parserTests. The parserTests setup includes the definition |
713 | * of a number of non-standard extension tags, whose names are passed |
714 | * over from the JS side in hybrid testing. |
715 | * @param string $tag Name of an extension tag assumed to be present |
716 | */ |
717 | public function ensureExtensionTag( string $tag ): void { |
718 | $this->loadSiteData(); |
719 | $this->extensionTags[mb_strtolower( $tag )] = true; |
720 | } |
721 | |
722 | /** @inheritDoc */ |
723 | protected function getNonNativeExtensionTags(): array { |
724 | $this->loadSiteData(); |
725 | return $this->extensionTags; |
726 | } |
727 | |
728 | /** @inheritDoc */ |
729 | public function getMaxTemplateDepth(): int { |
730 | // Not in the API result |
731 | return $this->maxDepth; |
732 | } |
733 | |
734 | /** @inheritDoc */ |
735 | protected function getSpecialNSAliases(): array { |
736 | $nsAliases = [ |
737 | 'Special', |
738 | ]; |
739 | foreach ( $this->nsIds as $name => $id ) { |
740 | if ( $id === -1 ) { |
741 | $nsAliases[] = $this->quoteTitleRe( $name, '!' ); |
742 | } |
743 | } |
744 | return $nsAliases; |
745 | } |
746 | |
747 | /** @inheritDoc */ |
748 | protected function getSpecialPageAliases( string $specialPage ): array { |
749 | $spAliases = [ $specialPage ]; |
750 | foreach ( $this->specialPageAliases as $special ) { |
751 | if ( $special['realname'] === $specialPage ) { |
752 | $spAliases = array_merge( $spAliases, $special['aliases'] ); |
753 | break; |
754 | } |
755 | } |
756 | return $spAliases; |
757 | } |
758 | |
759 | /** @inheritDoc */ |
760 | protected function getProtocols(): array { |
761 | $this->loadSiteData(); |
762 | return $this->protocols; |
763 | } |
764 | |
765 | /** @var ?MockMetrics */ |
766 | private $metrics; |
767 | |
768 | /** @inheritDoc */ |
769 | public function metrics(): ?StatsdDataFactoryInterface { |
770 | if ( $this->metrics === null ) { |
771 | $this->metrics = new MockMetrics(); |
772 | } |
773 | return $this->metrics; |
774 | } |
775 | |
776 | /** |
777 | * Increment a counter metric |
778 | * @param string $name |
779 | * @param array $labels |
780 | * @param float $amount |
781 | * @return void |
782 | */ |
783 | public function incrementCounter( string $name, array $labels, float $amount = 1 ): void { |
784 | // We don't use the labels for now, using MockMetrics instead |
785 | $this->metrics->increment( $name ); |
786 | } |
787 | |
788 | /** |
789 | * Record a timing metric |
790 | * @param string $name |
791 | * @param float $value |
792 | * @param array $labels |
793 | * @return void |
794 | */ |
795 | public function observeTiming( string $name, float $value, array $labels ): void { |
796 | // We don't use the labels for now, using MockMetrics instead |
797 | $this->metrics->timing( $name, $value ); |
798 | } |
799 | |
800 | /** @inheritDoc */ |
801 | public function getNoFollowConfig(): array { |
802 | $this->loadSiteData(); |
803 | return [ |
804 | 'nofollow' => $this->siteData['nofollowlinks'] ?? true, |
805 | 'nsexceptions' => $this->siteData['nofollownsexceptions'] ?? [], |
806 | 'domainexceptions' => $this->siteData['nofollowdomainexceptions'] ?? [ 'mediawiki.org' ] |
807 | ]; |
808 | } |
809 | |
810 | /** @inheritDoc */ |
811 | public function getExternalLinkTarget() { |
812 | $this->loadSiteData(); |
813 | return $this->siteData['externallinktarget'] ?? false; |
814 | } |
815 | } |