Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
78.05% |
64 / 82 |
|
50.00% |
5 / 10 |
CRAP | |
0.00% |
0 / 1 |
| MessageParser | |
78.05% |
64 / 82 |
|
50.00% |
5 / 10 |
30.09 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| transform | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
3.01 | |||
| parse | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
2 | |||
| parseWithoutPostprocessing | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| getOptions | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
3.01 | |||
| createOptions | |
66.67% |
12 / 18 |
|
0.00% |
0 / 1 |
4.59 | |||
| normalizeTargetLanguage | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| getPlaceholderTitle | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| acquireParser | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
3 | |||
| releaseParser | |
40.00% |
2 / 5 |
|
0.00% |
0 / 1 |
2.86 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Language; |
| 4 | |
| 5 | use LogicException; |
| 6 | use MediaWiki\Context\RequestContext; |
| 7 | use MediaWiki\MainConfigNames; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use MediaWiki\OutputTransform\OutputTransformPipeline; |
| 10 | use MediaWiki\Page\PageReference; |
| 11 | use MediaWiki\Page\PageReferenceValue; |
| 12 | use MediaWiki\Parser\Parser; |
| 13 | use MediaWiki\Parser\ParserFactory; |
| 14 | use MediaWiki\Parser\ParserOptions; |
| 15 | use MediaWiki\Parser\ParserOutput; |
| 16 | use MediaWiki\StubObject\StubObject; |
| 17 | use MediaWiki\StubObject\StubUserLang; |
| 18 | use Psr\Log\LoggerInterface; |
| 19 | |
| 20 | /** |
| 21 | * Service for transformation of interface message text. |
| 22 | * |
| 23 | * @since 1.44 |
| 24 | */ |
| 25 | class MessageParser { |
| 26 | private const DEPTH_EXCEEDED_MESSAGE = |
| 27 | '<span class="error">Message parse depth limit exceeded</span>'; |
| 28 | |
| 29 | private ParserFactory $parserFactory; |
| 30 | private OutputTransformPipeline $outputPipeline; |
| 31 | private LanguageFactory $langFactory; |
| 32 | private LoggerInterface $logger; |
| 33 | |
| 34 | /** @var Parser[] Cached Parser objects */ |
| 35 | private array $parsers = []; |
| 36 | /** @var ParserOptions[] Parser options associated with each Parser */ |
| 37 | private array $parserOptions = []; |
| 38 | /** @var int Index into $this->parsers for the active Parser */ |
| 39 | private int $curParser = -1; |
| 40 | |
| 41 | /** |
| 42 | * Parsing some messages may require parsing another message first, due to special page |
| 43 | * transclusion and some hooks (T372891). This constant is the limit of nesting depth where |
| 44 | * we'll display an error instead of the other message. |
| 45 | */ |
| 46 | private const MAX_PARSER_DEPTH = 5; |
| 47 | |
| 48 | public function __construct( |
| 49 | ParserFactory $parserFactory, |
| 50 | OutputTransformPipeline $outputPipeline, |
| 51 | LanguageFactory $languageFactory, |
| 52 | LoggerInterface $logger |
| 53 | ) { |
| 54 | $this->parserFactory = $parserFactory; |
| 55 | $this->outputPipeline = $outputPipeline; |
| 56 | $this->langFactory = $languageFactory; |
| 57 | $this->logger = $logger; |
| 58 | } |
| 59 | |
| 60 | /** |
| 61 | * Run message text through the preprocessor, expanding parser functions |
| 62 | * |
| 63 | * @param string $message |
| 64 | * @param bool $interface |
| 65 | * @param Language|string|null $language Language for {{PLURAL:}} etc., or |
| 66 | * null for the user or content language depending on $interface |
| 67 | * @param PageReference|null $page |
| 68 | * @return string |
| 69 | */ |
| 70 | public function transform( |
| 71 | $message, |
| 72 | $interface = false, |
| 73 | $language = null, |
| 74 | ?PageReference $page = null |
| 75 | ) { |
| 76 | // Avoid creating parser if nothing to transform |
| 77 | if ( !str_contains( $message, '{{' ) ) { |
| 78 | return $message; |
| 79 | } |
| 80 | |
| 81 | $parser = $this->acquireParser(); |
| 82 | if ( !$parser ) { |
| 83 | return self::DEPTH_EXCEEDED_MESSAGE; |
| 84 | } |
| 85 | |
| 86 | $popts = $this->getOptions( $interface, $language ); |
| 87 | $page ??= $this->getPlaceholderTitle(); |
| 88 | |
| 89 | try { |
| 90 | return $parser->transformMsg( $message, $popts, $page ); |
| 91 | } finally { |
| 92 | $this->releaseParser( $parser ); |
| 93 | } |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * @param string $text |
| 98 | * @param ?PageReference $contextPage The context page, or null to use a placeholder |
| 99 | * @param bool $lineStart Whether this should be parsed in start-of-line context |
| 100 | * @param bool $interface Whether this is an interface message |
| 101 | * @param Language|StubUserLang|string|null $language Language for {{PLURAL:}} etc., or |
| 102 | * null for the user or content language depending on $interface |
| 103 | * @return ParserOutput |
| 104 | */ |
| 105 | public function parse( |
| 106 | string $text, |
| 107 | ?PageReference $contextPage = null, |
| 108 | bool $lineStart = true, |
| 109 | bool $interface = false, |
| 110 | $language = null |
| 111 | ): ParserOutput { |
| 112 | $options = [ |
| 113 | 'allowTOC' => false, |
| 114 | // Wrapping messages in an extra <div> is probably not expected. If |
| 115 | // they're outside the content area they probably shouldn't be |
| 116 | // targeted by CSS that's targeting the parser output, and if |
| 117 | // they're inside they already are from the outer div. |
| 118 | 'unwrap' => true, |
| 119 | 'userLang' => $language, |
| 120 | ]; |
| 121 | $parser = $this->acquireParser(); |
| 122 | if ( !$parser ) { |
| 123 | return new ParserOutput( self::DEPTH_EXCEEDED_MESSAGE ); |
| 124 | } |
| 125 | $popts = $this->getOptions( $interface, $language ); |
| 126 | $contextPage ??= $this->getPlaceholderTitle(); |
| 127 | |
| 128 | try { |
| 129 | $po = $parser->parse( $text, $contextPage, $popts, $lineStart ); |
| 130 | // Run the post-processing pipeline |
| 131 | return $this->outputPipeline->run( $po, $popts, $options ); |
| 132 | } finally { |
| 133 | $this->releaseParser( $parser ); |
| 134 | } |
| 135 | } |
| 136 | |
| 137 | /** |
| 138 | * @param string $text |
| 139 | * @param ?PageReference $page The context title, or null to use a placeholder |
| 140 | * @param bool $lineStart Whether this is at the start of a line |
| 141 | * @param bool $interface Whether this is an interface message |
| 142 | * @param Language|StubUserLang|string|null $language Target language |
| 143 | * @return ParserOutput |
| 144 | */ |
| 145 | public function parseWithoutPostprocessing( |
| 146 | $text, |
| 147 | ?PageReference $page = null, |
| 148 | $lineStart = true, |
| 149 | $interface = false, |
| 150 | $language = null |
| 151 | ): ParserOutput { |
| 152 | $parser = $this->acquireParser(); |
| 153 | if ( !$parser ) { |
| 154 | return new ParserOutput( self::DEPTH_EXCEEDED_MESSAGE ); |
| 155 | } |
| 156 | $popts = $this->getOptions( $interface, $language ); |
| 157 | $page ??= $this->getPlaceholderTitle(); |
| 158 | |
| 159 | try { |
| 160 | return $parser->parse( $text, $page, $popts, $lineStart ); |
| 161 | } finally { |
| 162 | $this->releaseParser( $parser ); |
| 163 | } |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Configure parser options |
| 168 | * |
| 169 | * @param bool $interface |
| 170 | * @param Language|StubUserLang|string|null $targetLanguage |
| 171 | * @return ParserOptions |
| 172 | */ |
| 173 | private function getOptions( $interface, $targetLanguage ): ParserOptions { |
| 174 | if ( $this->curParser < 0 ) { |
| 175 | throw new LogicException( 'getOptions must be called after acquireParser' ); |
| 176 | } |
| 177 | if ( isset( $this->parserOptions[$this->curParser] ) ) { |
| 178 | $popts = $this->parserOptions[$this->curParser]; |
| 179 | } else { |
| 180 | $popts = $this->createOptions( $this->curParser ); |
| 181 | } |
| 182 | $popts->setInterfaceMessage( $interface ); |
| 183 | $popts->setTargetLanguage( $this->normalizeTargetLanguage( $targetLanguage ) ); |
| 184 | $popts->setSuppressSectionEditLinks(); |
| 185 | return $popts; |
| 186 | } |
| 187 | |
| 188 | /** |
| 189 | * Create and maybe cache a ParserOptions object |
| 190 | * |
| 191 | * @param int $cacheIndex |
| 192 | * @return ParserOptions |
| 193 | */ |
| 194 | private function createOptions( $cacheIndex ): ParserOptions { |
| 195 | $context = RequestContext::getMain(); |
| 196 | $user = $context->getUser(); |
| 197 | $useParsoidMessages = MediaWikiServices::getInstance()->getMainConfig()->get( |
| 198 | MainConfigNames::UseParsoidMessages |
| 199 | ); |
| 200 | if ( !$user->isSafeToLoad() ) { |
| 201 | // It isn't safe to use the context user yet, so don't try to get a |
| 202 | // ParserOptions for it. And don't cache this ParserOptions |
| 203 | // either. |
| 204 | $po = ParserOptions::newFromAnon(); |
| 205 | $po->setAllowUnsafeRawHtml( false ); |
| 206 | if ( $useParsoidMessages !== null ) { |
| 207 | $po->setUseParsoid( $useParsoidMessages ); |
| 208 | } |
| 209 | return $po; |
| 210 | } |
| 211 | |
| 212 | $po = ParserOptions::newFromContext( $context ); |
| 213 | $po->setIsMessage( true ); |
| 214 | // Messages may take parameters that could come |
| 215 | // from malicious sources. As a precaution, disable |
| 216 | // the <html> parser tag when parsing messages. |
| 217 | $po->setAllowUnsafeRawHtml( false ); |
| 218 | if ( $useParsoidMessages !== null ) { |
| 219 | $po->setUseParsoid( $useParsoidMessages ); |
| 220 | } |
| 221 | $this->parserOptions[$cacheIndex] = $po; |
| 222 | return $po; |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * @param Language|string|null $lang |
| 227 | * @return Language|null |
| 228 | */ |
| 229 | private function normalizeTargetLanguage( $lang ): ?Language { |
| 230 | if ( is_string( $lang ) ) { |
| 231 | return $this->langFactory->getLanguage( $lang ); |
| 232 | } elseif ( $lang !== null ) { |
| 233 | StubObject::unstub( $lang ); |
| 234 | } |
| 235 | return $lang; |
| 236 | } |
| 237 | |
| 238 | private function getPlaceholderTitle(): PageReference { |
| 239 | return PageReferenceValue::localReference( NS_SPECIAL, 'Badtitle/MessageParser' ); |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Attempt to get a free parser from the cache. If none exists, create one, |
| 244 | * up to a limit of MAX_PARSER_DEPTH. If the limit is exceeded, return null. |
| 245 | * |
| 246 | * If a parser is returned, it must be released with releaseParser(). |
| 247 | * |
| 248 | * @return Parser|null |
| 249 | */ |
| 250 | private function acquireParser(): ?Parser { |
| 251 | $index = $this->curParser + 1; |
| 252 | if ( $index >= self::MAX_PARSER_DEPTH ) { |
| 253 | $this->logger->debug( __METHOD__ . ": Refusing to create a new parser with index {$index}" ); |
| 254 | return null; |
| 255 | } |
| 256 | $parser = $this->parsers[ $index ] ?? null; |
| 257 | if ( !$parser ) { |
| 258 | $this->logger->debug( __METHOD__ . ": Creating a new parser with index {$index}" ); |
| 259 | $parser = $this->parserFactory->create(); |
| 260 | } |
| 261 | $this->parsers[ $index ] = $parser; |
| 262 | $this->curParser = $index; |
| 263 | return $parser; |
| 264 | } |
| 265 | |
| 266 | /** |
| 267 | * Release a parser previously acquired by acquireParser(). |
| 268 | * |
| 269 | * @param Parser $parser |
| 270 | */ |
| 271 | private function releaseParser( Parser $parser ) { |
| 272 | if ( $this->parsers[$this->curParser] !== $parser ) { |
| 273 | throw new LogicException( 'releaseParser called with the wrong ' . |
| 274 | "parser instance: #{$this->curParser} = " . |
| 275 | gettype( $this->parsers[$this->curParser] ) ); |
| 276 | } |
| 277 | $this->curParser--; |
| 278 | } |
| 279 | |
| 280 | } |