Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 99 |
|
0.00% |
0 / 35 |
CRAP | |
0.00% |
0 / 1 |
| StubMetadataCollector | |
0.00% |
0 / 99 |
|
0.00% |
0 / 35 |
3192 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addCategory | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| addWarningMsg | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addExternalLink | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setOutputFlag | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| appendOutputStrings | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| setPageProperty | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setExtensionData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setJsConfigVar | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| appendExtensionData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| appendJsConfigVar | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addModules | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| addModuleStyles | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| setLimitReportData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setTOCData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addLink | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addImage | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addLanguageLink | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getLanguageLinks | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| collect | |
0.00% |
0 / 31 |
|
0.00% |
0 / 1 |
110 | |||
| get | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
20 | |||
| getModules | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getModuleStyles | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getJsConfigVars | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| getCategoryNames | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getCategorySortKey | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getPageProperty | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getExtensionData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getOutputFlags | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| getTOCData | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setIndicator | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getIndicators | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getImages | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| linkToString | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| stringToLink | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | declare( strict_types = 1 ); |
| 4 | |
| 5 | namespace Wikimedia\Parsoid\Config; |
| 6 | |
| 7 | use Psr\Log\LoggerInterface; |
| 8 | use Psr\Log\LogLevel; |
| 9 | use Wikimedia\Parsoid\Core\ContentMetadataCollector; |
| 10 | use Wikimedia\Parsoid\Core\ContentMetadataCollectorCompat; |
| 11 | use Wikimedia\Parsoid\Core\LinkTarget; |
| 12 | use Wikimedia\Parsoid\Core\TOCData; |
| 13 | use Wikimedia\Parsoid\Utils\Title; |
| 14 | |
| 15 | /** |
| 16 | * Minimal implementation of a ContentMetadataCollector which just |
| 17 | * records all metadata in an array. Used for testing or operation |
| 18 | * in API mode. |
| 19 | */ |
| 20 | class StubMetadataCollector implements ContentMetadataCollector { |
| 21 | use ContentMetadataCollectorCompat; |
| 22 | |
| 23 | /** @var SiteConfig */ |
| 24 | private $siteConfig; |
| 25 | |
| 26 | /** @var LoggerInterface */ |
| 27 | private $logger; |
| 28 | |
| 29 | /** @var array<string,array> */ |
| 30 | private $mWarningMsgs = []; |
| 31 | |
| 32 | /** @var array */ |
| 33 | private $storage = []; |
| 34 | |
| 35 | /** @var string */ |
| 36 | private const MERGE_STRATEGY_KEY = '_parsoid-strategy_'; |
| 37 | |
| 38 | /** |
| 39 | * Non-standard merge strategy to use for properties which are *not* |
| 40 | * accumulators: "write-once" means that the property should be set |
| 41 | * once (although subsequently resetting it to the same value is ok) |
| 42 | * and an error will be thrown if there is an attempt to combine |
| 43 | * multiple values. |
| 44 | * |
| 45 | * This strategy is internal to the StubMetadataCollector for now; |
| 46 | * ParserOutput implements similar semantics for many of its properties, |
| 47 | * but not (yet) in a principled or uniform way. |
| 48 | */ |
| 49 | private const MERGE_STRATEGY_WRITE_ONCE = 'write-once'; |
| 50 | |
| 51 | /** |
| 52 | * @param SiteConfig $siteConfig Used to resolve title namespaces |
| 53 | * and to log warnings for unsafe metadata updates |
| 54 | */ |
| 55 | public function __construct( |
| 56 | SiteConfig $siteConfig |
| 57 | ) { |
| 58 | $this->siteConfig = $siteConfig; |
| 59 | $this->logger = $siteConfig->getLogger(); |
| 60 | } |
| 61 | |
| 62 | /** @inheritDoc */ |
| 63 | public function addCategory( $c, $sort = '' ): void { |
| 64 | if ( $c instanceof LinkTarget ) { |
| 65 | $c = $c->getDBkey(); |
| 66 | } |
| 67 | // Numeric strings often become an `int` when passed to addCategory() |
| 68 | $this->collect( 'categories', (string)$c, $sort, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 69 | } |
| 70 | |
| 71 | /** @inheritDoc */ |
| 72 | public function addWarningMsg( string $msg, ...$args ): void { |
| 73 | $this->mWarningMsgs[$msg] = $args; |
| 74 | } |
| 75 | |
| 76 | /** @inheritDoc */ |
| 77 | public function addExternalLink( string $url ): void { |
| 78 | $this->collect( 'externallinks', '', $url ); |
| 79 | } |
| 80 | |
| 81 | /** @inheritDoc */ |
| 82 | public function setOutputFlag( string $name, bool $value = true ): void { |
| 83 | $this->collect( 'outputflags', $name, (string)$value, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 84 | } |
| 85 | |
| 86 | /** @inheritDoc */ |
| 87 | public function appendOutputStrings( string $name, array $value ): void { |
| 88 | foreach ( $value as $v ) { |
| 89 | $this->collect( 'outputstrings', $name, $v ); |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | /** @inheritDoc */ |
| 94 | public function setPageProperty( string $name, $value ): void { |
| 95 | $this->collect( 'properties', $name, $value, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 96 | } |
| 97 | |
| 98 | /** @inheritDoc */ |
| 99 | public function setExtensionData( string $key, $value ): void { |
| 100 | $this->collect( 'extensiondata', $key, $value, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 101 | } |
| 102 | |
| 103 | /** @inheritDoc */ |
| 104 | public function setJsConfigVar( string $key, $value ): void { |
| 105 | $this->collect( 'jsconfigvars', $key, $value, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 106 | } |
| 107 | |
| 108 | /** @inheritDoc */ |
| 109 | public function appendExtensionData( |
| 110 | string $key, |
| 111 | $value, |
| 112 | string $strategy = self::MERGE_STRATEGY_UNION |
| 113 | ): void { |
| 114 | $this->collect( 'extensiondata', $key, $value, $strategy ); |
| 115 | } |
| 116 | |
| 117 | /** @inheritDoc */ |
| 118 | public function appendJsConfigVar( |
| 119 | string $key, |
| 120 | string $value, |
| 121 | string $strategy = self::MERGE_STRATEGY_UNION |
| 122 | ): void { |
| 123 | $this->collect( 'jsconfigvars', $key, $value, $strategy ); |
| 124 | } |
| 125 | |
| 126 | /** @inheritDoc */ |
| 127 | public function addModules( array $modules ): void { |
| 128 | foreach ( $modules as $module ) { |
| 129 | $this->collect( 'modules', '', $module ); |
| 130 | } |
| 131 | } |
| 132 | |
| 133 | /** @inheritDoc */ |
| 134 | public function addModuleStyles( array $moduleStyles ): void { |
| 135 | foreach ( $moduleStyles as $style ) { |
| 136 | $this->collect( 'modulestyles', '', $style ); |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | /** @inheritDoc */ |
| 141 | public function setLimitReportData( string $key, $value ): void { |
| 142 | // XXX maybe need to JSON-encode $value |
| 143 | $this->collect( 'limitreportdata', $key, $value, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 144 | } |
| 145 | |
| 146 | /** @inheritDoc */ |
| 147 | public function setTOCData( TOCData $tocData ): void { |
| 148 | $this->collect( 'tocdata', '', $tocData, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 149 | } |
| 150 | |
| 151 | /** @inheritDoc */ |
| 152 | public function addLink( LinkTarget $link, $id = null ): void { |
| 153 | $this->collect( 'links', '', $this->linkToString( $link ) ); |
| 154 | } |
| 155 | |
| 156 | /** @inheritDoc */ |
| 157 | public function addImage( LinkTarget $name, $timestamp = null, $sha1 = null ): void { |
| 158 | $title = Title::newFromLinkTarget( $name, $this->siteConfig ); |
| 159 | $this->collect( 'images', '', $title->getDBkey() ); |
| 160 | } |
| 161 | |
| 162 | /** @inheritDoc */ |
| 163 | public function addLanguageLink( LinkTarget $lt ): void { |
| 164 | $this->collect( 'language-link', '', $this->linkToString( $lt ) ); |
| 165 | } |
| 166 | |
| 167 | /** @return LinkTarget[] */ |
| 168 | public function getLanguageLinks(): array { |
| 169 | return array_map( function ( $v ) { |
| 170 | return $this->stringToLink( $v ); |
| 171 | }, $this->get( 'language-link', '' ) ); |
| 172 | } |
| 173 | |
| 174 | /** |
| 175 | * Unified internal implementation of metadata collection. |
| 176 | * @param string $which Internal string identifying the type of metadata. |
| 177 | * @param string $key Key for storage (or '' if this is not relevant) |
| 178 | * @param mixed $value Value to store |
| 179 | * @param string $strategy "union" or "write-once" |
| 180 | */ |
| 181 | private function collect( |
| 182 | string $which, string $key, $value, |
| 183 | string $strategy = self::MERGE_STRATEGY_UNION |
| 184 | ): void { |
| 185 | if ( !array_key_exists( $which, $this->storage ) ) { |
| 186 | $this->storage[$which] = []; |
| 187 | } |
| 188 | if ( !array_key_exists( $key, $this->storage[$which] ) ) { |
| 189 | $this->storage[$which][$key] = [ self::MERGE_STRATEGY_KEY => $strategy ]; |
| 190 | if ( $strategy === self::MERGE_STRATEGY_WRITE_ONCE ) { |
| 191 | $this->storage[$which][$key]['value'] = $value; |
| 192 | return; |
| 193 | } |
| 194 | } |
| 195 | if ( $this->storage[$which][$key][self::MERGE_STRATEGY_KEY] !== $strategy ) { |
| 196 | $this->logger->log( |
| 197 | LogLevel::WARNING, |
| 198 | "Conflicting strategies for $which $key" |
| 199 | ); |
| 200 | // Destructive update for compatibility; this is deprecated! |
| 201 | unset( $this->storage[$which][$key] ); |
| 202 | $this->collect( $which, $key, $value, $strategy ); |
| 203 | return; |
| 204 | } |
| 205 | if ( $strategy === self::MERGE_STRATEGY_WRITE_ONCE ) { |
| 206 | if ( ( $this->storage[$which][$key]['value'] ?? null ) === $value ) { |
| 207 | return; // already exists with the desired value |
| 208 | } |
| 209 | $this->logger->log( |
| 210 | LogLevel::WARNING, |
| 211 | "Multiple writes to a write-once: $which $key" |
| 212 | ); |
| 213 | // Destructive update for compatibility; this is deprecated! |
| 214 | unset( $this->storage[$which][$key] ); |
| 215 | $this->collect( $which, $key, $value, $strategy ); |
| 216 | return; |
| 217 | } elseif ( $strategy === self::MERGE_STRATEGY_UNION ) { |
| 218 | if ( !( is_string( $value ) || is_int( $value ) ) ) { |
| 219 | throw new \InvalidArgumentException( "Bad value type for $key: " . gettype( $value ) ); |
| 220 | } |
| 221 | $this->storage[$which][$key][$value] = true; |
| 222 | return; |
| 223 | } else { |
| 224 | throw new \InvalidArgumentException( "Unknown strategy: $strategy" ); |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | /** |
| 229 | * Retrieve values from the collector. |
| 230 | * @param string $which Internal string identifying the type of metadata. |
| 231 | * @param string|null $key Key for storage (or '' if this is not relevant) |
| 232 | * @param string $defaultStrategy Determines whether to return an empty |
| 233 | * array or null for a missing $key |
| 234 | * @return mixed |
| 235 | */ |
| 236 | private function get( string $which, ?string $key = null, string $defaultStrategy = self::MERGE_STRATEGY_UNION ) { |
| 237 | if ( $key !== null ) { |
| 238 | $result = ( $this->storage[$which] ?? [] )[$key] ?? []; |
| 239 | $strategy = $result[self::MERGE_STRATEGY_KEY] ?? $defaultStrategy; |
| 240 | unset( $result[self::MERGE_STRATEGY_KEY] ); |
| 241 | if ( $strategy === self::MERGE_STRATEGY_WRITE_ONCE ) { |
| 242 | return $result['value'] ?? null; |
| 243 | } else { |
| 244 | return array_keys( $result ); |
| 245 | } |
| 246 | } |
| 247 | $result = []; |
| 248 | foreach ( ( $this->storage[$which] ?? [] ) as $key => $ignore ) { |
| 249 | $result[$key] = $this->get( $which, (string)$key ); |
| 250 | } |
| 251 | return $result; |
| 252 | } |
| 253 | |
| 254 | // @internal introspection methods |
| 255 | |
| 256 | /** @return string[] */ |
| 257 | public function getModules(): array { |
| 258 | return $this->get( 'modules', '' ); |
| 259 | } |
| 260 | |
| 261 | /** @return string[] */ |
| 262 | public function getModuleStyles(): array { |
| 263 | return $this->get( 'modulestyles', '' ); |
| 264 | } |
| 265 | |
| 266 | /** @return string[] */ |
| 267 | public function getJsConfigVars(): array { |
| 268 | // This is somewhat unusual, in that we expose the 'set' represenation |
| 269 | // as $key => true, instead of just returning array_keys(). |
| 270 | $result = $this->storage['jsconfigvars'] ?? []; |
| 271 | foreach ( $result as $key => &$value ) { |
| 272 | $strategy = $value[self::MERGE_STRATEGY_KEY] ?? null; |
| 273 | unset( $value[self::MERGE_STRATEGY_KEY] ); |
| 274 | if ( $strategy === self::MERGE_STRATEGY_WRITE_ONCE ) { |
| 275 | $value = array_keys( $value )[0]; |
| 276 | } |
| 277 | } |
| 278 | return $result; |
| 279 | } |
| 280 | |
| 281 | /** @return list<string> */ |
| 282 | public function getCategoryNames(): array { |
| 283 | // array keys can get converted to int if numeric, so ensure |
| 284 | // return value is all strings. |
| 285 | return array_map( 'strval', array_keys( $this->get( 'categories' ) ) ); |
| 286 | } |
| 287 | |
| 288 | /** |
| 289 | * @param string $name Category name |
| 290 | * @return ?string Sort key |
| 291 | */ |
| 292 | public function getCategorySortKey( string $name ): ?string { |
| 293 | return $this->get( 'categories', $name ); |
| 294 | } |
| 295 | |
| 296 | /** |
| 297 | * @param string $name |
| 298 | * @return ?string |
| 299 | */ |
| 300 | public function getPageProperty( string $name ): ?string { |
| 301 | return $this->get( 'properties', $name, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 302 | } |
| 303 | |
| 304 | /** |
| 305 | * Return the collected extension data under the given key. |
| 306 | * @param string $key |
| 307 | * @return mixed|null |
| 308 | */ |
| 309 | public function getExtensionData( string $key ) { |
| 310 | return $this->get( 'extensiondata', $key, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 311 | } |
| 312 | |
| 313 | /** |
| 314 | * Return the active output flags. |
| 315 | * @return string[] |
| 316 | */ |
| 317 | public function getOutputFlags() { |
| 318 | $result = []; |
| 319 | foreach ( $this->get( 'outputflags', null ) as $key => $value ) { |
| 320 | if ( $value ) { |
| 321 | $result[] = $key; |
| 322 | } |
| 323 | } |
| 324 | return $result; |
| 325 | } |
| 326 | |
| 327 | /** |
| 328 | * Return the collected TOC data, or null if no TOC data was collected. |
| 329 | * @return ?TOCData |
| 330 | */ |
| 331 | public function getTOCData(): ?TOCData { |
| 332 | return $this->get( 'tocdata', '', self::MERGE_STRATEGY_WRITE_ONCE ); |
| 333 | } |
| 334 | |
| 335 | /** |
| 336 | * Set the content for an indicator. |
| 337 | * @param string $name |
| 338 | * @param string $content |
| 339 | */ |
| 340 | public function setIndicator( $name, $content ): void { |
| 341 | $this->collect( 'indicators', $name, $content, self::MERGE_STRATEGY_WRITE_ONCE ); |
| 342 | } |
| 343 | |
| 344 | /** |
| 345 | * Return a "name" => "content-id" mapping of recorded indicators |
| 346 | * @return array |
| 347 | */ |
| 348 | public function getIndicators(): array { |
| 349 | return $this->get( 'indicators' ); |
| 350 | } |
| 351 | |
| 352 | /** |
| 353 | * @return array |
| 354 | */ |
| 355 | public function getImages(): array { |
| 356 | return $this->get( 'images', '' ); |
| 357 | } |
| 358 | |
| 359 | // helper functions for recording LinkTarget objects |
| 360 | |
| 361 | /** |
| 362 | * Convert a LinkTarget to a string for storing in the collected metadata. |
| 363 | * @param LinkTarget $lt |
| 364 | * @return string |
| 365 | */ |
| 366 | private function linkToString( LinkTarget $lt ): string { |
| 367 | $title = Title::newFromLinkTarget( $lt, $this->siteConfig ); |
| 368 | $text = $title->getPrefixedText(); |
| 369 | $fragment = $title->getFragment(); |
| 370 | if ( $fragment !== '' ) { |
| 371 | $text .= '#' . $fragment; |
| 372 | } |
| 373 | return $text; |
| 374 | } |
| 375 | |
| 376 | /** |
| 377 | * Convert a string back into a LinkTarget for retrieval from the |
| 378 | * collected metadata. |
| 379 | * @param string $s |
| 380 | * @return LinkTarget |
| 381 | */ |
| 382 | private function stringToLink( string $s ): LinkTarget { |
| 383 | return Title::newFromText( $s, $this->siteConfig ); |
| 384 | } |
| 385 | } |