Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
92.37% |
533 / 577 |
|
34.78% |
8 / 23 |
CRAP | |
0.00% |
0 / 1 |
| MWVisitor | |
92.37% |
533 / 577 |
|
34.78% |
8 / 23 |
231.93 | |
0.00% |
0 / 1 |
| analyzeCallNode | |
95.83% |
23 / 24 |
|
0.00% |
0 / 1 |
11 | |||
| checkExternalLink | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
3 | |||
| doSelectWrapperSpecialHandling | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
6 | |||
| triggerHook | |
79.49% |
31 / 39 |
|
0.00% |
0 / 1 |
10.86 | |||
| hookArgsContainReference | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
4 | |||
| extractHookArgs | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
3.04 | |||
| getHookTypeForRegistrationMethod | |
71.43% |
5 / 7 |
|
0.00% |
0 / 1 |
4.37 | |||
| handleNormalHookRegistration | |
58.33% |
7 / 12 |
|
0.00% |
0 / 1 |
5.16 | |||
| handleParserHookRegistration | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
3.04 | |||
| registerHook | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| visitReturn | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
6 | |||
| handleGetQueryInfoReturn | |
96.67% |
29 / 30 |
|
0.00% |
0 / 1 |
9 | |||
| checkMakeList | |
97.73% |
43 / 44 |
|
0.00% |
0 / 1 |
12 | |||
| literalListConstToName | |
66.67% |
10 / 15 |
|
0.00% |
0 / 1 |
8.81 | |||
| checkSQLOptions | |
96.55% |
28 / 29 |
|
0.00% |
0 / 1 |
7 | |||
| checkJoinCond | |
80.00% |
36 / 45 |
|
0.00% |
0 / 1 |
14.35 | |||
| visitReturnOfFunctionHook | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
12 | |||
| getCallableFromHookRegistration | |
90.00% |
18 / 20 |
|
0.00% |
0 / 1 |
12.14 | |||
| getSingleCallable | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| getCallbackForVar | |
81.82% |
9 / 11 |
|
0.00% |
0 / 1 |
5.15 | |||
| visitAssign | |
100.00% |
27 / 27 |
|
100.00% |
1 / 1 |
16 | |||
| detectHTMLForm | |
97.66% |
167 / 171 |
|
0.00% |
0 / 1 |
56 | |||
| visitArray | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace SecurityCheckPlugin; |
| 4 | |
| 5 | use ast\Node; |
| 6 | use Phan\Analysis\PostOrderAnalysisVisitor; |
| 7 | use Phan\AST\ContextNode; |
| 8 | use Phan\Exception\CodeBaseException; |
| 9 | use Phan\Exception\InvalidFQSENException; |
| 10 | use Phan\Exception\IssueException; |
| 11 | use Phan\Language\Element\FunctionInterface; |
| 12 | use Phan\Language\Element\Method; |
| 13 | use Phan\Language\FQSEN\FullyQualifiedClassName; |
| 14 | use Phan\Language\FQSEN\FullyQualifiedFunctionLikeName; |
| 15 | use Phan\Language\FQSEN\FullyQualifiedFunctionName; |
| 16 | use Phan\Language\FQSEN\FullyQualifiedMethodName; |
| 17 | use Phan\Language\UnionType; |
| 18 | |
| 19 | /** |
| 20 | * MediaWiki specific node visitor |
| 21 | * |
| 22 | * Copyright (C) 2017 Brian Wolff <bawolff@gmail.com> |
| 23 | * |
| 24 | * This program is free software; you can redistribute it and/or modify |
| 25 | * it under the terms of the GNU General Public License as published by |
| 26 | * the Free Software Foundation; either version 2 of the License, or |
| 27 | * (at your option) any later version. |
| 28 | * |
| 29 | * This program is distributed in the hope that it will be useful, |
| 30 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 31 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 32 | * GNU General Public License for more details. |
| 33 | * |
| 34 | * You should have received a copy of the GNU General Public License along |
| 35 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 36 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
| 37 | */ |
| 38 | class MWVisitor extends TaintednessVisitor { |
| 39 | /** |
| 40 | * @todo This is a temporary hack. Proper solution is refactoring/avoiding overrideContext |
| 41 | * @var bool|null |
| 42 | * @suppress PhanWriteOnlyProtectedProperty |
| 43 | */ |
| 44 | protected $isHook; |
| 45 | |
| 46 | /** |
| 47 | * Try and recognize hook registration |
| 48 | * @inheritDoc |
| 49 | */ |
| 50 | protected function analyzeCallNode( Node $node, iterable $funcs ): void { |
| 51 | parent::analyzeCallNode( $node, $funcs ); |
| 52 | if ( !isset( $node->children['method'] ) ) { |
| 53 | // Called by visitCall |
| 54 | return; |
| 55 | } |
| 56 | |
| 57 | assert( is_array( $funcs ) && count( $funcs ) === 1 ); |
| 58 | $method = $funcs[0]; |
| 59 | assert( $method instanceof Method ); |
| 60 | |
| 61 | // Should this be getDefiningFQSEN() instead? |
| 62 | $methodName = (string)$method->getFQSEN(); |
| 63 | $parserFQSEN = MediaWikiHooksHelper::getInstance()->getMwParserClassFQSEN( $this->code_base )->__toString(); |
| 64 | // $this->debug( __METHOD__, "Checking to see if we should register $methodName" ); |
| 65 | switch ( $methodName ) { |
| 66 | case "$parserFQSEN::setFunctionHook": |
| 67 | case "$parserFQSEN::setHook": |
| 68 | $type = $this->getHookTypeForRegistrationMethod( $methodName ); |
| 69 | if ( $type === null ) { |
| 70 | break; |
| 71 | } |
| 72 | // $this->debug( __METHOD__, "registering $methodName as $type" ); |
| 73 | $this->handleParserHookRegistration( $node, $type ); |
| 74 | break; |
| 75 | case '\Hooks::register': |
| 76 | $this->handleNormalHookRegistration( $node ); |
| 77 | break; |
| 78 | case '\Hooks::run': |
| 79 | case '\Hooks::runWithoutAbort': |
| 80 | $this->triggerHook( $node ); |
| 81 | break; |
| 82 | case '\Linker::makeExternalLink': |
| 83 | $this->checkExternalLink( $node ); |
| 84 | break; |
| 85 | default: |
| 86 | $this->doSelectWrapperSpecialHandling( $node, $method ); |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * Linker::makeExternalLink escaping depends on third argument |
| 92 | * |
| 93 | * @param Node $node |
| 94 | */ |
| 95 | private function checkExternalLink( Node $node ): void { |
| 96 | $escapeArg = $this->resolveValue( $node->children['args']->children[2] ?? true ); |
| 97 | $text = $node->children['args']->children[1] ?? null; |
| 98 | if ( !$escapeArg && $text instanceof Node ) { |
| 99 | $this->maybeEmitIssueSimplified( |
| 100 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 101 | $text, |
| 102 | "Calling Linker::makeExternalLink with user controlled text " . |
| 103 | "and third argument set to false" |
| 104 | ); |
| 105 | } |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * Special casing for complex format of IDatabase::select |
| 110 | * |
| 111 | * This handles the $options, and $join_cond. Other args are |
| 112 | * handled through normal means |
| 113 | * |
| 114 | * @param Node $node Either an AST_METHOD_CALL or AST_STATIC_CALL |
| 115 | * @param Method $method |
| 116 | */ |
| 117 | private function doSelectWrapperSpecialHandling( Node $node, Method $method ): void { |
| 118 | $relevantMethods = [ |
| 119 | 'makeList' => true, |
| 120 | 'select' => true, |
| 121 | 'selectField' => true, |
| 122 | 'selectFieldValues' => true, |
| 123 | 'selectSQLText' => true, |
| 124 | 'selectRowCount' => true, |
| 125 | 'selectRow' => true, |
| 126 | ]; |
| 127 | |
| 128 | if ( !isset( $relevantMethods[$method->getName()] ) ) { |
| 129 | return; |
| 130 | } |
| 131 | |
| 132 | $idbFQSEN = FullyQualifiedClassName::fromFullyQualifiedString( '\\Wikimedia\\Rdbms\\IDatabase' ); |
| 133 | if ( !self::isSubclassOf( $method->getClassFQSEN(), $idbFQSEN, $this->code_base ) ) { |
| 134 | return; |
| 135 | } |
| 136 | |
| 137 | if ( $method->getName() === 'makeList' ) { |
| 138 | $this->checkMakeList( $node ); |
| 139 | return; |
| 140 | } |
| 141 | |
| 142 | $args = $node->children['args']->children; |
| 143 | if ( isset( $args[4] ) ) { |
| 144 | $this->checkSQLOptions( $args[4] ); |
| 145 | } |
| 146 | if ( isset( $args[5] ) ) { |
| 147 | $this->checkJoinCond( $args[5] ); |
| 148 | } |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Dispatch a hook (i.e. Handle Hooks::run) |
| 153 | * |
| 154 | * @param Node $node The Hooks::run AST_STATIC_CALL |
| 155 | */ |
| 156 | private function triggerHook( Node $node ): void { |
| 157 | $argList = $node->children['args']->children; |
| 158 | if ( count( $argList ) === 0 ) { |
| 159 | $this->debug( __METHOD__, "Too few args to Hooks::run" ); |
| 160 | return; |
| 161 | } |
| 162 | if ( !is_string( $argList[0] ) ) { |
| 163 | $this->debug( __METHOD__, "Cannot determine hook name" ); |
| 164 | return; |
| 165 | } |
| 166 | '@phan-var array{0:string,1?:Node} $argList'; |
| 167 | $hookName = $argList[0]; |
| 168 | if ( |
| 169 | count( $argList ) < 2 |
| 170 | || $argList[1]->kind !== \ast\AST_ARRAY |
| 171 | ) { |
| 172 | // @todo There are definitely cases where this |
| 173 | // will prevent us from running hooks |
| 174 | // e.g. EditPageGetPreviewContent |
| 175 | $this->debug( __METHOD__, "Could not run hook $hookName due to complex args" ); |
| 176 | return; |
| 177 | } |
| 178 | $args = $this->extractHookArgs( $argList[1] ); |
| 179 | $hasPassByRef = self::hookArgsContainReference( $argList[1] ); |
| 180 | $analyzer = new PostOrderAnalysisVisitor( $this->code_base, $this->context, [] ); |
| 181 | $argumentTypes = array_fill( 0, count( $args ), UnionType::empty() ); |
| 182 | |
| 183 | $subscribers = MediaWikiHooksHelper::getInstance()->getHookSubscribers( $hookName ); |
| 184 | foreach ( $subscribers as $subscriber ) { |
| 185 | if ( $subscriber instanceof FullyQualifiedMethodName ) { |
| 186 | if ( !$this->code_base->hasMethodWithFQSEN( $subscriber ) ) { |
| 187 | $this->debug( __METHOD__, "Hook subscriber $subscriber not found!" ); |
| 188 | continue; |
| 189 | } |
| 190 | $func = $this->code_base->getMethodByFQSEN( $subscriber ); |
| 191 | } else { |
| 192 | assert( $subscriber instanceof FullyQualifiedFunctionName ); |
| 193 | if ( !$this->code_base->hasFunctionWithFQSEN( $subscriber ) ) { |
| 194 | $this->debug( __METHOD__, "Hook subscriber $subscriber not found!" ); |
| 195 | continue; |
| 196 | } |
| 197 | $func = $this->code_base->getFunctionByFQSEN( $subscriber ); |
| 198 | } |
| 199 | |
| 200 | // $this->debug( __METHOD__, "Dispatching $hookName to $subscriber" ); |
| 201 | // This is hacky, but try to ensure that the associated line |
| 202 | // number for any issues is in the extension, and not the |
| 203 | // line where the Hooks::register() is in MW core. |
| 204 | // FIXME: In the case of reference parameters, this is |
| 205 | // still reporting things being in MW core instead of extension. |
| 206 | $oldContext = $this->overrideContext; |
| 207 | $fContext = $func->getContext(); |
| 208 | $newContext = clone $this->context; |
| 209 | $newContext = $newContext->withFile( $fContext->getFile() ) |
| 210 | ->withLineNumberStart( $fContext->getLineNumberStart() ); |
| 211 | $this->overrideContext = $newContext; |
| 212 | $this->isHook = true; |
| 213 | |
| 214 | if ( $hasPassByRef ) { |
| 215 | // Trigger an analysis of the function call (see e.g. ClosureReturnTypeOverridePlugin's |
| 216 | // handling of call_user_func_array). Note that it's not enough to use our |
| 217 | // handleMethodCall, because that doesn't handle references correctly. |
| 218 | |
| 219 | // NOTE: This is only known to be necessary with references, hence the check above |
| 220 | // (for performance). There might be other edge cases, though... |
| 221 | |
| 222 | // TODO We don't care about types, so we use an empty union type. However this looks |
| 223 | // very very fragile. |
| 224 | // TODO 2: Someday we could write a generic-purpose MW plugin, which could (among other |
| 225 | // things) understand hook. It could share some code with taint-check, and at that |
| 226 | // point we'd likely want to use the correct types here (note that phan alone isn't |
| 227 | // able to analyze hooks at all). |
| 228 | $analyzer->analyzeCallableWithArgumentTypes( $argumentTypes, $func, $args ); |
| 229 | } |
| 230 | $this->handleMethodCall( $func, $subscriber, $args, false, true ); |
| 231 | |
| 232 | $this->overrideContext = $oldContext; |
| 233 | $this->isHook = false; |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | /** |
| 238 | * Check whether any argument to (inside an array) is a reference. |
| 239 | * |
| 240 | * @param Node $argArrayNode |
| 241 | * @return bool |
| 242 | */ |
| 243 | private static function hookArgsContainReference( Node $argArrayNode ): bool { |
| 244 | foreach ( $argArrayNode->children as $child ) { |
| 245 | if ( $child instanceof Node && ( $child->flags & \ast\flags\ARRAY_ELEM_REF ) ) { |
| 246 | return true; |
| 247 | } |
| 248 | } |
| 249 | return false; |
| 250 | } |
| 251 | |
| 252 | /** |
| 253 | * Convenience methods for extracting hooks arguments. Copied from |
| 254 | * ClosureReturnTypeOverridePlugin::extractArrayArgs (which is private) |
| 255 | * and simplified for our use case. |
| 256 | * |
| 257 | * @param Node $argArrayNode |
| 258 | * @return Node[] |
| 259 | */ |
| 260 | private function extractHookArgs( Node $argArrayNode ): array { |
| 261 | assert( $argArrayNode->kind === \ast\AST_ARRAY ); |
| 262 | $arguments = []; |
| 263 | foreach ( $argArrayNode->children as $child ) { |
| 264 | if ( !( $child instanceof Node ) ) { |
| 265 | continue; |
| 266 | } |
| 267 | $arguments[] = $child->children['value']; |
| 268 | } |
| 269 | return $arguments; |
| 270 | } |
| 271 | |
| 272 | /** |
| 273 | * @param string $method The method name of the registration function |
| 274 | * @return string|null The name of the hook that gets registered |
| 275 | */ |
| 276 | private function getHookTypeForRegistrationMethod( string $method ): ?string { |
| 277 | $parserFQSEN = MediaWikiHooksHelper::getInstance()->getMwParserClassFQSEN( $this->code_base )->__toString(); |
| 278 | switch ( $method ) { |
| 279 | case "$parserFQSEN::setFunctionHook": |
| 280 | return '!ParserFunctionHook'; |
| 281 | case "$parserFQSEN::setHook": |
| 282 | return '!ParserHook'; |
| 283 | default: |
| 284 | $this->debug( __METHOD__, "$method not a hook registerer" ); |
| 285 | return null; |
| 286 | } |
| 287 | } |
| 288 | |
| 289 | /** |
| 290 | * Handle registering a normal hook from Hooks::register (Not from $wgHooks) |
| 291 | * |
| 292 | * @param Node $node The node representing the AST_STATIC_CALL |
| 293 | */ |
| 294 | private function handleNormalHookRegistration( Node $node ): void { |
| 295 | assert( $node->kind === \ast\AST_STATIC_CALL ); |
| 296 | $params = $node->children['args']->children; |
| 297 | if ( count( $params ) < 2 ) { |
| 298 | $this->debug( __METHOD__, "Could not understand Hooks::register" ); |
| 299 | return; |
| 300 | } |
| 301 | $hookName = $params[0]; |
| 302 | if ( !is_string( $hookName ) ) { |
| 303 | $this->debug( __METHOD__, "Could not register hook. Name is complex" ); |
| 304 | return; |
| 305 | } |
| 306 | $cb = $this->getCallableFromHookRegistration( $params[1], $hookName ); |
| 307 | if ( $cb ) { |
| 308 | $this->registerHook( $hookName, $cb ); |
| 309 | } else { |
| 310 | $this->debug( __METHOD__, "Could not register $hookName hook due to complex callback" ); |
| 311 | } |
| 312 | } |
| 313 | |
| 314 | /** |
| 315 | * When someone calls $parser->setFunctionHook() or setTagHook() |
| 316 | * |
| 317 | * @note Causes phan to error out if given non-existent class |
| 318 | * @param Node $node The AST_METHOD_CALL node |
| 319 | * @param string $hookType The name of the hook |
| 320 | */ |
| 321 | private function handleParserHookRegistration( Node $node, string $hookType ): void { |
| 322 | $args = $node->children['args']->children; |
| 323 | if ( count( $args ) < 2 ) { |
| 324 | return; |
| 325 | } |
| 326 | $callback = $this->getCallableFromNode( $args[1] ); |
| 327 | if ( $callback ) { |
| 328 | $this->registerHook( $hookType, $callback ); |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | /** |
| 333 | * @param string $hookType |
| 334 | * @param FunctionInterface $callback |
| 335 | */ |
| 336 | private function registerHook( string $hookType, FunctionInterface $callback ): void { |
| 337 | $fqsen = $callback->getFQSEN(); |
| 338 | $alreadyRegistered = MediaWikiHooksHelper::getInstance()->registerHook( $hookType, $fqsen ); |
| 339 | if ( !$alreadyRegistered ) { |
| 340 | // $this->debug( __METHOD__, "registering $fqsen for hook $hookType" ); |
| 341 | // If this is the first time seeing this, make sure we reanalyze the hook function now that |
| 342 | // we know what it is, in case it's already been analyzed. |
| 343 | $this->analyzeFunc( $callback ); |
| 344 | } |
| 345 | } |
| 346 | |
| 347 | /** |
| 348 | * For special hooks, check their return value |
| 349 | * |
| 350 | * e.g. A tag hook's return value is output as html. |
| 351 | * @param Node $node |
| 352 | */ |
| 353 | public function visitReturn( Node $node ): void { |
| 354 | parent::visitReturn( $node ); |
| 355 | if ( |
| 356 | !$node->children['expr'] instanceof Node || |
| 357 | !$this->context->isInFunctionLikeScope() |
| 358 | ) { |
| 359 | return; |
| 360 | } |
| 361 | $funcFQSEN = $this->context->getFunctionLikeFQSEN(); |
| 362 | |
| 363 | if ( strpos( (string)$funcFQSEN, '::getQueryInfo' ) !== false ) { |
| 364 | $this->handleGetQueryInfoReturn( $node->children['expr'] ); |
| 365 | } |
| 366 | |
| 367 | $hookType = MediaWikiHooksHelper::getInstance()->isSpecialHookSubscriber( $funcFQSEN ); |
| 368 | switch ( $hookType ) { |
| 369 | case '!ParserFunctionHook': |
| 370 | $this->visitReturnOfFunctionHook( $node->children['expr'], $funcFQSEN ); |
| 371 | break; |
| 372 | case '!ParserHook': |
| 373 | $ret = $node->children['expr']; |
| 374 | $this->maybeEmitIssueSimplified( |
| 375 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 376 | $ret, |
| 377 | "Outputting user controlled HTML from Parser tag hook {FUNCTIONLIKE}", |
| 378 | [ $funcFQSEN ] |
| 379 | ); |
| 380 | break; |
| 381 | } |
| 382 | } |
| 383 | |
| 384 | /** |
| 385 | * Methods named getQueryInfo() in MediaWiki usually |
| 386 | * return an array that is later fed to select |
| 387 | * |
| 388 | * @note This will only work where the return |
| 389 | * statement is an array literal. |
| 390 | * @param Node|mixed $node Node from ast tree |
| 391 | */ |
| 392 | private function handleGetQueryInfoReturn( $node ): void { |
| 393 | if ( |
| 394 | !( $node instanceof Node ) || |
| 395 | $node->kind !== \ast\AST_ARRAY |
| 396 | ) { |
| 397 | return; |
| 398 | } |
| 399 | // The argument order is |
| 400 | // $table, $vars, $conds = '', $fname = __METHOD__, |
| 401 | // $options = [], $join_conds = [] |
| 402 | $keysToArg = [ |
| 403 | 'tables' => 0, |
| 404 | 'fields' => 1, |
| 405 | 'conds' => 2, |
| 406 | 'options' => 4, |
| 407 | 'join_conds' => 5, |
| 408 | ]; |
| 409 | $args = [ '', '', '', '' ]; |
| 410 | foreach ( $node->children as $child ) { |
| 411 | assert( $child->kind === \ast\AST_ARRAY_ELEM ); |
| 412 | $key = $child->children['key']; |
| 413 | if ( $key instanceof Node ) { |
| 414 | // Dynamic name, skip (T268055). |
| 415 | continue; |
| 416 | } |
| 417 | if ( !isset( $keysToArg[$key] ) ) { |
| 418 | continue; |
| 419 | } |
| 420 | $args[$keysToArg[$key]] = $child->children['value']; |
| 421 | } |
| 422 | $selectFQSEN = FullyQualifiedMethodName::fromFullyQualifiedString( |
| 423 | '\Wikimedia\Rdbms\IDatabase::select' |
| 424 | ); |
| 425 | if ( !$this->code_base->hasMethodWithFQSEN( $selectFQSEN ) ) { |
| 426 | // Huh. Core wasn't parsed. That's bad, but don't fail hard. |
| 427 | $this->debug( __METHOD__, 'Database::select does not exist.' ); |
| 428 | return; |
| 429 | } |
| 430 | $select = $this->code_base->getMethodByFQSEN( $selectFQSEN ); |
| 431 | // TODO: The message about calling Database::select here is not very clear. |
| 432 | $this->handleMethodCall( $select, $selectFQSEN, $args, false ); |
| 433 | if ( isset( $args[4] ) ) { |
| 434 | $this->checkSQLOptions( $args[4] ); |
| 435 | } |
| 436 | if ( isset( $args[5] ) ) { |
| 437 | $this->checkJoinCond( $args[5] ); |
| 438 | } |
| 439 | } |
| 440 | |
| 441 | /** |
| 442 | * Check IDatabase::makeList |
| 443 | * |
| 444 | * Special cased because the second arg totally changes |
| 445 | * how this function is interpreted. |
| 446 | * @param Node $node |
| 447 | */ |
| 448 | private function checkMakeList( Node $node ): void { |
| 449 | $args = $node->children['args']; |
| 450 | // First determine which IDatabase::LIST_* |
| 451 | // 0 = IDatabase::LIST_COMMA is default value. |
| 452 | $typeArg = $args->children[1] ?? 0; |
| 453 | if ( $typeArg instanceof Node ) { |
| 454 | $typeArg = $this->getCtxN( $typeArg )->getEquivalentPHPValueForNode( |
| 455 | $typeArg, |
| 456 | ContextNode::RESOLVE_SCALAR_DEFAULT & ~ContextNode::RESOLVE_CONSTANTS |
| 457 | ); |
| 458 | } |
| 459 | if ( $typeArg instanceof Node ) { |
| 460 | if ( $typeArg->kind === \ast\AST_CLASS_CONST ) { |
| 461 | // Probably IDatabase::LIST_*. Note that non-class constants are resolved |
| 462 | $typeArg = $typeArg->children['const']; |
| 463 | } elseif ( $typeArg->kind === \ast\AST_CONST ) { |
| 464 | $typeArg = $typeArg->children['name']->children['name']; |
| 465 | } else { |
| 466 | // Something that cannot be resolved statically. Since LIST_NAMES is very rare, and LIST_COMMA is |
| 467 | // default, assume its LIST_AND or LIST_OR |
| 468 | $this->debug( __METHOD__, "Could not determine 2nd arg makeList()" ); |
| 469 | $this->maybeEmitIssueSimplified( |
| 470 | new Taintedness( SecurityCheckPlugin::SQL_NUMKEY_EXEC_TAINT ), |
| 471 | $args->children[0], |
| 472 | "IDatabase::makeList with unknown type arg is " . |
| 473 | "given an array with unescaped keynames or " . |
| 474 | "values for numeric keys (May be false positive)" |
| 475 | ); |
| 476 | |
| 477 | return; |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | // Make sure not to mix strings and ints in switch cases, as that will break horribly |
| 482 | if ( is_int( $typeArg ) ) { |
| 483 | $typeArg = $this->literalListConstToName( $typeArg ); |
| 484 | } |
| 485 | switch ( $typeArg ) { |
| 486 | case 'LIST_COMMA': |
| 487 | // String keys ignored. Everything escaped. So nothing to worry about. |
| 488 | break; |
| 489 | case 'LIST_AND': |
| 490 | case 'LIST_SET': |
| 491 | case 'LIST_OR': |
| 492 | // exec_sql_numkey |
| 493 | $this->maybeEmitIssueSimplified( |
| 494 | new Taintedness( SecurityCheckPlugin::SQL_NUMKEY_EXEC_TAINT ), |
| 495 | $args->children[0], |
| 496 | "IDatabase::makeList with LIST_AND, LIST_OR or " |
| 497 | . "LIST_SET must sql escape string key names and values of numeric keys" |
| 498 | ); |
| 499 | break; |
| 500 | case 'LIST_NAMES': |
| 501 | // Like comma but with no escaping. |
| 502 | $this->maybeEmitIssueSimplified( |
| 503 | new Taintedness( SecurityCheckPlugin::SQL_EXEC_TAINT ), |
| 504 | $args->children[0], |
| 505 | "IDatabase::makeList with LIST_NAMES needs " |
| 506 | . "to escape for SQL" |
| 507 | ); |
| 508 | break; |
| 509 | default: |
| 510 | $this->debug( __METHOD__, "Unrecognized 2nd arg " . "to IDatabase::makeList: '$typeArg'" ); |
| 511 | } |
| 512 | } |
| 513 | |
| 514 | /** |
| 515 | * Convert a literal int value for a LIST_* constant to its name. This is a horrible hack for crappy code |
| 516 | * that uses the constants literally rather than by name. Such code shouldn't deserve taint analysis. |
| 517 | * This method can obviously break very easily if the values are changed. |
| 518 | * |
| 519 | * @param int $value |
| 520 | * @return string |
| 521 | */ |
| 522 | private function literalListConstToName( int $value ): string { |
| 523 | switch ( $value ) { |
| 524 | case 0: |
| 525 | return 'LIST_COMMA'; |
| 526 | case 1: |
| 527 | return 'LIST_AND'; |
| 528 | case 2: |
| 529 | return 'LIST_SET'; |
| 530 | case 3: |
| 531 | return 'LIST_NAMES'; |
| 532 | case 4: |
| 533 | return 'LIST_OR'; |
| 534 | default: |
| 535 | // Oh boy, what the heck are you doing? Well, DWIM |
| 536 | $this->debug( |
| 537 | __METHOD__, |
| 538 | 'Someone specified a LIST_* constant literally but it is not a valid value. Wow.' |
| 539 | ); |
| 540 | return 'LIST_AND'; |
| 541 | } |
| 542 | } |
| 543 | |
| 544 | /** |
| 545 | * Check the options parameter to IDatabase::select |
| 546 | * |
| 547 | * This only works if its specified as an array literal. |
| 548 | * |
| 549 | * Relevant options: |
| 550 | * GROUP BY is put directly in the query (array gets imploded) |
| 551 | * HAVING is treated like a WHERE clause |
| 552 | * ORDER BY is put directly in the query (array gets imploded) |
| 553 | * USE INDEX is directly put in string (both array and string version) |
| 554 | * IGNORE INDEX ditto |
| 555 | * @param Node|mixed $node The node from the AST tree |
| 556 | */ |
| 557 | private function checkSQLOptions( $node ): void { |
| 558 | if ( !( $node instanceof Node ) || $node->kind !== \ast\AST_ARRAY ) { |
| 559 | return; |
| 560 | } |
| 561 | $relevant = [ |
| 562 | 'GROUP BY' => true, |
| 563 | 'ORDER BY' => true, |
| 564 | 'HAVING' => true, |
| 565 | 'USE INDEX' => true, |
| 566 | 'IGNORE INDEX' => true, |
| 567 | ]; |
| 568 | foreach ( $node->children as $arrayElm ) { |
| 569 | assert( $arrayElm->kind === \ast\AST_ARRAY_ELEM ); |
| 570 | $val = $arrayElm->children['value']; |
| 571 | $key = $arrayElm->children['key']; |
| 572 | |
| 573 | if ( isset( $relevant[$key] ) ) { |
| 574 | $taintType = ( $key === 'HAVING' && $this->nodeIsArray( $val ) ) ? |
| 575 | SecurityCheckPlugin::SQL_NUMKEY_EXEC_TAINT : |
| 576 | SecurityCheckPlugin::SQL_EXEC_TAINT; |
| 577 | $taintType = new Taintedness( $taintType ); |
| 578 | |
| 579 | $this->backpropagateArgTaint( $node, $taintType ); |
| 580 | $ctx = clone $this->context; |
| 581 | $this->overrideContext = $ctx->withLineNumberStart( |
| 582 | $val->lineno ?? $ctx->getLineNumberStart() |
| 583 | ); |
| 584 | $this->maybeEmitIssueSimplified( |
| 585 | $taintType, |
| 586 | $val, |
| 587 | "{STRING_LITERAL} clause is user controlled", |
| 588 | [ $key ] |
| 589 | ); |
| 590 | $this->overrideContext = null; |
| 591 | } |
| 592 | } |
| 593 | } |
| 594 | |
| 595 | /** |
| 596 | * Check a join_cond structure. |
| 597 | * |
| 598 | * Syntax is like |
| 599 | * |
| 600 | * [ 'aliasOfTable' => [ 'JOIN TYPE', $onConditions ], ... ] |
| 601 | * join type is usually something safe like INNER JOIN, but it is not |
| 602 | * validated or escaped. $onConditions is the same form as a WHERE clause. |
| 603 | * |
| 604 | * @param Node|mixed $node |
| 605 | */ |
| 606 | private function checkJoinCond( $node ): void { |
| 607 | if ( !( $node instanceof Node ) || $node->kind !== \ast\AST_ARRAY ) { |
| 608 | return; |
| 609 | } |
| 610 | |
| 611 | foreach ( $node->children as $table ) { |
| 612 | assert( $table->kind === \ast\AST_ARRAY_ELEM ); |
| 613 | |
| 614 | $tableName = is_string( $table->children['key'] ) ? |
| 615 | $table->children['key'] : |
| 616 | '[UNKNOWN TABLE]'; |
| 617 | $joinInfo = $table->children['value']; |
| 618 | if ( $joinInfo instanceof Node && $joinInfo->kind === \ast\AST_ARRAY ) { |
| 619 | if ( |
| 620 | count( $joinInfo->children ) === 0 || |
| 621 | $joinInfo->children[0]->children['key'] !== null |
| 622 | ) { |
| 623 | $this->debug( __METHOD__, "join info has named key??" ); |
| 624 | continue; |
| 625 | } |
| 626 | $joinType = $joinInfo->children[0]->children['value']; |
| 627 | // join type does not get escaped. |
| 628 | $this->maybeEmitIssueSimplified( |
| 629 | new Taintedness( SecurityCheckPlugin::SQL_EXEC_TAINT ), |
| 630 | $joinType, |
| 631 | "Join type for {STRING_LITERAL} is user controlled", |
| 632 | [ $tableName ] |
| 633 | ); |
| 634 | if ( $joinType instanceof Node ) { |
| 635 | $this->backpropagateArgTaint( |
| 636 | $joinType, |
| 637 | new Taintedness( SecurityCheckPlugin::SQL_EXEC_TAINT ) |
| 638 | ); |
| 639 | } |
| 640 | // On to the join ON conditions. |
| 641 | if ( |
| 642 | count( $joinInfo->children ) === 1 || |
| 643 | $joinInfo->children[1]->children['key'] !== null |
| 644 | ) { |
| 645 | $this->debug( __METHOD__, "join info has named key??" ); |
| 646 | continue; |
| 647 | } |
| 648 | $onCond = $joinInfo->children[1]->children['value']; |
| 649 | $ctx = clone $this->context; |
| 650 | $this->overrideContext = $ctx->withLineNumberStart( |
| 651 | $onCond->lineno ?? $ctx->getLineNumberStart() |
| 652 | ); |
| 653 | $this->maybeEmitIssueSimplified( |
| 654 | new Taintedness( SecurityCheckPlugin::SQL_NUMKEY_EXEC_TAINT ), |
| 655 | $onCond, |
| 656 | "The ON conditions are not properly escaped for the join to `{STRING_LITERAL}`", |
| 657 | [ $tableName ] |
| 658 | ); |
| 659 | if ( $onCond instanceof Node ) { |
| 660 | $this->backpropagateArgTaint( |
| 661 | $onCond, |
| 662 | new Taintedness( SecurityCheckPlugin::SQL_NUMKEY_EXEC_TAINT ) |
| 663 | ); |
| 664 | } |
| 665 | $this->overrideContext = null; |
| 666 | } |
| 667 | } |
| 668 | } |
| 669 | |
| 670 | /** |
| 671 | * Check to see if isHTML => true and is tainted. |
| 672 | * |
| 673 | * @param Node $node The expr child of the return. NOT the return itself |
| 674 | * @param FullyQualifiedFunctionLikeName $funcName |
| 675 | */ |
| 676 | private function visitReturnOfFunctionHook( Node $node, FullyQualifiedFunctionLikeName $funcName ): void { |
| 677 | if ( $node->kind !== \ast\AST_ARRAY || count( $node->children ) < 2 ) { |
| 678 | return; |
| 679 | } |
| 680 | $isHTML = false; |
| 681 | foreach ( $node->children as $child ) { |
| 682 | assert( $child instanceof Node && $child->kind === \ast\AST_ARRAY_ELEM ); |
| 683 | |
| 684 | if ( |
| 685 | $child->children['key'] === 'isHTML' && |
| 686 | $child->children['value'] instanceof Node && |
| 687 | $child->children['value']->kind === \ast\AST_CONST && |
| 688 | $child->children['value']->children['name'] instanceof Node && |
| 689 | $child->children['value']->children['name']->children['name'] === 'true' |
| 690 | ) { |
| 691 | $isHTML = true; |
| 692 | break; |
| 693 | } |
| 694 | } |
| 695 | if ( !$isHTML ) { |
| 696 | return; |
| 697 | } |
| 698 | |
| 699 | $arg = $node->children[0]; |
| 700 | assert( $arg instanceof Node && $arg->kind === \ast\AST_ARRAY_ELEM ); |
| 701 | $this->maybeEmitIssueSimplified( |
| 702 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 703 | $arg->children['value'], |
| 704 | "Outputting user controlled HTML from Parser function hook {FUNCTIONLIKE}", |
| 705 | [ $funcName ] |
| 706 | ); |
| 707 | } |
| 708 | |
| 709 | /** |
| 710 | * Given a MediaWiki hook registration, find the callback |
| 711 | * |
| 712 | * @note This is a different format than Parser hooks use. |
| 713 | * |
| 714 | * Valid examples of callbacks: |
| 715 | * "wfSomeFunction" |
| 716 | * "SomeClass::SomeStaticMethod" |
| 717 | * A Closure |
| 718 | * $instanceOfSomeObject (With implied method name based on hook name) |
| 719 | * new SomeClass |
| 720 | * [ <one of the above>, $extraArgsForCallback, ...] |
| 721 | * [ [<one of the above>], $extraArgsForCallback, ...] |
| 722 | * [ $instanceOfObj, 'methodName', $optionalArgForCallback, ... ] |
| 723 | * [ [ $instanceOfObj, 'methodName' ], $optionalArgForCallback, ...] |
| 724 | * |
| 725 | * Oddly enough, [ 'NameOfClass', 'NameOfStaticMethod' ] does not appear |
| 726 | * to be valid, despite that being a valid callable. |
| 727 | * |
| 728 | * @param Node|mixed $node |
| 729 | * @param string $hookName |
| 730 | * @return FunctionInterface|null |
| 731 | */ |
| 732 | private function getCallableFromHookRegistration( $node, string $hookName ): ?FunctionInterface { |
| 733 | // "wfSomething", "Class::Method", closure |
| 734 | if ( !$node instanceof Node || $node->kind === \ast\AST_CLOSURE ) { |
| 735 | return $this->getCallableFromNode( $node ); |
| 736 | } |
| 737 | |
| 738 | $cb = $this->getSingleCallable( $node, 'on' . $hookName ); |
| 739 | if ( $cb ) { |
| 740 | return $cb; |
| 741 | } |
| 742 | |
| 743 | if ( $node->kind === \ast\AST_ARRAY ) { |
| 744 | if ( count( $node->children ) === 0 ) { |
| 745 | return null; |
| 746 | } |
| 747 | $firstChild = $node->children[0]->children['value']; |
| 748 | if ( |
| 749 | ( $firstChild instanceof Node && $firstChild->kind === \ast\AST_ARRAY ) || |
| 750 | !( $firstChild instanceof Node ) || |
| 751 | count( $node->children ) === 1 |
| 752 | ) { |
| 753 | // One of: |
| 754 | // [ [ <callback> ], $optionalArgs, ... ] |
| 755 | // [ 'SomeClass::method', $optionalArgs, ... ] |
| 756 | // [ <callback> ] |
| 757 | // Important to note, this is safe because the |
| 758 | // [ 'SomeClass', 'MethodToCallStatically' ] |
| 759 | // syntax isn't supported by hooks. |
| 760 | return $this->getCallableFromHookRegistration( $firstChild, $hookName ); |
| 761 | } |
| 762 | // Remaining case is: [ $someObject, 'methodToCall', 'arg', ... ] |
| 763 | $methodName = $this->resolveValue( $node->children[1]->children['value'] ); |
| 764 | if ( !is_string( $methodName ) ) { |
| 765 | return null; |
| 766 | } |
| 767 | $cb = $this->getSingleCallable( $firstChild, $methodName ); |
| 768 | if ( $cb ) { |
| 769 | return $cb; |
| 770 | } |
| 771 | } |
| 772 | return null; |
| 773 | } |
| 774 | |
| 775 | /** |
| 776 | * @param Node $node |
| 777 | * @param string $methodName |
| 778 | * @return FunctionInterface|null |
| 779 | */ |
| 780 | private function getSingleCallable( Node $node, string $methodName ): ?FunctionInterface { |
| 781 | if ( $node->kind === \ast\AST_VAR && is_string( $node->children['name'] ) ) { |
| 782 | return $this->getCallbackForVar( $node, $methodName ); |
| 783 | } |
| 784 | if ( $node->kind === \ast\AST_NEW ) { |
| 785 | $cxn = $this->getCtxN( $node ); |
| 786 | try { |
| 787 | $ctor = $cxn->getMethod( '__construct', false, false, true ); |
| 788 | return $ctor->getClass( $this->code_base )->getMethodByName( $this->code_base, $methodName ); |
| 789 | } catch ( CodeBaseException $e ) { |
| 790 | // @todo Should probably emit a non-security issue |
| 791 | $this->debug( __METHOD__, "Missing hook handle: " . $this->getDebugInfo( $e ) ); |
| 792 | } |
| 793 | } |
| 794 | return null; |
| 795 | } |
| 796 | |
| 797 | /** |
| 798 | * Given an AST_VAR node, figure out what it represents as callback |
| 799 | * |
| 800 | * @param Node $node The variable |
| 801 | * @param string $defaultMethod If the var is an object, what method to use |
| 802 | * @return FunctionInterface|null |
| 803 | */ |
| 804 | private function getCallbackForVar( Node $node, $defaultMethod = '' ): ?FunctionInterface { |
| 805 | assert( $node->kind === \ast\AST_VAR ); |
| 806 | $cnode = $this->getCtxN( $node ); |
| 807 | // Try the class case first, because the callable case might emit issues (about missing __invoke) if executed |
| 808 | // for a variable holding just a class instance. |
| 809 | try { |
| 810 | // Don't warn if it's the wrong type, for it might be a callable and not a class. |
| 811 | $classes = $cnode->getClassList( true, ContextNode::CLASS_LIST_ACCEPT_ANY, null, false ); |
| 812 | } catch ( CodeBaseException | IssueException $_ ) { |
| 813 | $classes = []; |
| 814 | } |
| 815 | foreach ( $classes as $class ) { |
| 816 | if ( $class->getFQSEN()->__toString() === '\Closure' ) { |
| 817 | // This means callable case, done below. |
| 818 | continue; |
| 819 | } |
| 820 | try { |
| 821 | return $class->getMethodByName( $this->code_base, $defaultMethod ); |
| 822 | } catch ( CodeBaseException $_ ) { |
| 823 | return null; |
| 824 | } |
| 825 | } |
| 826 | |
| 827 | return $this->getCallableFromNode( $node ); |
| 828 | } |
| 829 | |
| 830 | /** |
| 831 | * Check for $wgHooks registration |
| 832 | * |
| 833 | * @param Node $node |
| 834 | * @note This assumes $wgHooks is always the global |
| 835 | * even if there is no globals declaration. |
| 836 | */ |
| 837 | public function visitAssign( Node $node ): void { |
| 838 | parent::visitAssign( $node ); |
| 839 | |
| 840 | $var = $node->children['var']; |
| 841 | if ( !$var instanceof Node ) { |
| 842 | // Syntax error |
| 843 | return; |
| 844 | } |
| 845 | $hookName = null; |
| 846 | $expr = $node->children['expr']; |
| 847 | // The $wgHooks['foo'][] case |
| 848 | if ( |
| 849 | $var->kind === \ast\AST_DIM && |
| 850 | $var->children['dim'] === null && |
| 851 | $var->children['expr'] instanceof Node && |
| 852 | $var->children['expr']->kind === \ast\AST_DIM && |
| 853 | $var->children['expr']->children['expr'] instanceof Node && |
| 854 | is_string( $var->children['expr']->children['dim'] ) && |
| 855 | /* The $wgHooks['SomeHook'][] case */ |
| 856 | ( ( $var->children['expr']->children['expr']->kind === \ast\AST_VAR && |
| 857 | $var->children['expr']->children['expr']->children['name'] === 'wgHooks' ) || |
| 858 | /* The $_GLOBALS['wgHooks']['SomeHook'][] case */ |
| 859 | ( $var->children['expr']->children['expr']->kind === \ast\AST_DIM && |
| 860 | $var->children['expr']->children['expr']->children['expr'] instanceof Node && |
| 861 | $var->children['expr']->children['expr']->children['expr']->kind === \ast\AST_VAR && |
| 862 | $var->children['expr']->children['expr']->children['expr']->children['name'] === '_GLOBALS' ) ) |
| 863 | ) { |
| 864 | $hookName = $var->children['expr']->children['dim']; |
| 865 | } |
| 866 | |
| 867 | if ( $hookName !== null ) { |
| 868 | $cb = $this->getCallableFromHookRegistration( $expr, $hookName ); |
| 869 | if ( $cb ) { |
| 870 | $this->registerHook( $hookName, $cb ); |
| 871 | } else { |
| 872 | $this->debug( __METHOD__, "Could not register hook " . |
| 873 | "$hookName due to complex callback" |
| 874 | ); |
| 875 | } |
| 876 | } |
| 877 | } |
| 878 | |
| 879 | /** |
| 880 | * Special implementation of visitArray to detect HTMLForm specifiers |
| 881 | * |
| 882 | * @param Node $node |
| 883 | */ |
| 884 | private function detectHTMLForm( Node $node ): void { |
| 885 | // Try to immediately filter out things that certainly aren't HTMLForms |
| 886 | $maybeHTMLForm = false; |
| 887 | foreach ( $node->children as $child ) { |
| 888 | if ( $child instanceof Node && $child->kind === \ast\AST_ARRAY_ELEM ) { |
| 889 | $key = $child->children['key']; |
| 890 | if ( $key instanceof Node || $key === 'class' || $key === 'type' ) { |
| 891 | $maybeHTMLForm = true; |
| 892 | break; |
| 893 | } |
| 894 | } |
| 895 | } |
| 896 | if ( !$maybeHTMLForm ) { |
| 897 | return; |
| 898 | } |
| 899 | |
| 900 | $authReqFQSEN = FullyQualifiedClassName::fromFullyQualifiedString( |
| 901 | 'MediaWiki\Auth\AuthenticationRequest' |
| 902 | ); |
| 903 | |
| 904 | if ( |
| 905 | $this->code_base->hasClassWithFQSEN( $authReqFQSEN ) && |
| 906 | $this->context->isInClassScope() && |
| 907 | self::isSubclassOf( $this->context->getClassFQSEN(), $authReqFQSEN, $this->code_base ) |
| 908 | ) { |
| 909 | // AuthenticationRequest::getFieldInfo() defines a very |
| 910 | // similar array but with different rules. T202112 |
| 911 | return; |
| 912 | } |
| 913 | |
| 914 | // This is a rather superficial check. There |
| 915 | // are many ways to construct htmlform specifiers this |
| 916 | // won't catch, and it may also have some false positives. |
| 917 | |
| 918 | static $validHTMLFormTypes = [ |
| 919 | 'api', |
| 920 | 'text', |
| 921 | 'textwithbutton', |
| 922 | 'textarea', |
| 923 | 'select', |
| 924 | 'combobox', |
| 925 | 'radio', |
| 926 | 'multiselect', |
| 927 | 'limitselect', |
| 928 | 'check', |
| 929 | 'toggle', |
| 930 | 'int', |
| 931 | 'float', |
| 932 | 'info', |
| 933 | 'selectorother', |
| 934 | 'selectandother', |
| 935 | 'namespaceselect', |
| 936 | 'namespaceselectwithbutton', |
| 937 | 'tagfilter', |
| 938 | 'sizefilter', |
| 939 | 'submit', |
| 940 | 'hidden', |
| 941 | 'edittools', |
| 942 | 'checkmatrix', |
| 943 | 'cloner', |
| 944 | 'autocompleteselect', |
| 945 | 'date', |
| 946 | 'time', |
| 947 | 'datetime', |
| 948 | 'email', |
| 949 | 'password', |
| 950 | 'url', |
| 951 | 'title', |
| 952 | 'user', |
| 953 | 'usersmultiselect', |
| 954 | ]; |
| 955 | |
| 956 | $type = null; |
| 957 | $raw = null; |
| 958 | $class = null; |
| 959 | $rawLabel = null; |
| 960 | $help = null; |
| 961 | $label = null; |
| 962 | $default = null; |
| 963 | $options = null; |
| 964 | $isInfo = false; |
| 965 | // options key is really messed up with escaping. |
| 966 | $isOptionsSafe = true; |
| 967 | foreach ( $node->children as $child ) { |
| 968 | if ( $child === null || $child->kind === \ast\AST_UNPACK ) { |
| 969 | // If we have list( , $x ) = foo(), or an in-place unpack, chances are this is not an HTMLForm. |
| 970 | return; |
| 971 | } |
| 972 | assert( $child->kind === \ast\AST_ARRAY_ELEM ); |
| 973 | if ( $child->children['key'] === null ) { |
| 974 | // Implicit offset, hence most certainly not an HTMLForm. |
| 975 | return; |
| 976 | } |
| 977 | $key = $this->resolveOffset( $child->children['key'] ); |
| 978 | if ( !is_string( $key ) ) { |
| 979 | // Either not resolvable (so nothing we can say) or a non-string literal, skip. |
| 980 | return; |
| 981 | } |
| 982 | switch ( $key ) { |
| 983 | case 'type': |
| 984 | $type = $this->resolveValue( $child->children['value'] ); |
| 985 | break; |
| 986 | case 'class': |
| 987 | $class = $this->resolveValue( $child->children['value'] ); |
| 988 | break; |
| 989 | case 'label': |
| 990 | $label = $this->resolveValue( $child->children['value'] ); |
| 991 | break; |
| 992 | case 'options': |
| 993 | $options = $this->resolveValue( $child->children['value'] ); |
| 994 | break; |
| 995 | case 'default': |
| 996 | $default = $this->resolveValue( $child->children['value'] ); |
| 997 | break; |
| 998 | case 'label-raw': |
| 999 | $rawLabel = $this->resolveValue( $child->children['value'] ); |
| 1000 | break; |
| 1001 | case 'raw': |
| 1002 | case 'rawrow': |
| 1003 | $raw = $this->resolveValue( $child->children['value'] ); |
| 1004 | break; |
| 1005 | case 'help': |
| 1006 | $help = $this->resolveValue( $child->children['value'] ); |
| 1007 | break; |
| 1008 | } |
| 1009 | } |
| 1010 | |
| 1011 | if ( !$class && !$type ) { |
| 1012 | // Definitely not an HTMLForm |
| 1013 | // Also important to reject empty string, not just |
| 1014 | // null, otherwise 9e409c781015 of Wikibase causes |
| 1015 | // this to fatal |
| 1016 | return; |
| 1017 | } |
| 1018 | |
| 1019 | if ( |
| 1020 | $raw === null && $label === null && $rawLabel === null && $help === null |
| 1021 | && $default === null && $options === null |
| 1022 | ) { |
| 1023 | // e.g. [ 'class' => 'someCssClass' ] appears a lot |
| 1024 | // in the code base. If we don't have any of the html |
| 1025 | // fields, skip out early. |
| 1026 | return; |
| 1027 | } |
| 1028 | |
| 1029 | if ( $type !== null && !in_array( $type, $validHTMLFormTypes, true ) ) { |
| 1030 | // Not a valid HTMLForm field |
| 1031 | // (Or someone just added a new field type) |
| 1032 | return; |
| 1033 | } |
| 1034 | |
| 1035 | if ( $type === 'info' ) { |
| 1036 | $isInfo = true; |
| 1037 | } |
| 1038 | |
| 1039 | if ( in_array( $type, [ 'radio', 'multiselect' ], true ) ) { |
| 1040 | $isOptionsSafe = false; |
| 1041 | } |
| 1042 | |
| 1043 | if ( $class !== null ) { |
| 1044 | if ( !is_string( $class ) ) { |
| 1045 | return; |
| 1046 | } |
| 1047 | try { |
| 1048 | $fqsen = FullyQualifiedClassName::fromStringInContext( |
| 1049 | $class, |
| 1050 | $this->context |
| 1051 | ); |
| 1052 | } catch ( InvalidFQSENException $_ ) { |
| 1053 | // 'class' refers to something which is not a class, and this is probably not |
| 1054 | // an HTMLForm |
| 1055 | return; |
| 1056 | } |
| 1057 | if ( !$this->code_base->hasClassWithFQSEN( $fqsen ) ) { |
| 1058 | return; |
| 1059 | } |
| 1060 | if ( (string)$fqsen === '\HTMLInfoField' ) { |
| 1061 | $isInfo = true; |
| 1062 | } |
| 1063 | if ( |
| 1064 | (string)$fqsen === '\HTMLMultiSelectField' || |
| 1065 | (string)$fqsen === '\HTMLRadioField' |
| 1066 | ) { |
| 1067 | $isOptionsSafe = false; |
| 1068 | } |
| 1069 | |
| 1070 | $fqsenBase = FullyQualifiedClassName::fromFullyQualifiedString( |
| 1071 | '\HTMLFormField' |
| 1072 | ); |
| 1073 | if ( !$this->code_base->hasClassWithFQSEN( $fqsenBase ) ) { |
| 1074 | $this->debug( __METHOD__, "Missing HTMLFormField base class?!" ); |
| 1075 | return; |
| 1076 | } |
| 1077 | |
| 1078 | $isAField = self::isSubclassOf( $fqsen, $fqsenBase, $this->code_base ); |
| 1079 | |
| 1080 | if ( !$isAField ) { |
| 1081 | return; |
| 1082 | } |
| 1083 | } |
| 1084 | |
| 1085 | if ( $label !== null ) { |
| 1086 | // double escape check for label. |
| 1087 | $this->maybeEmitIssueSimplified( |
| 1088 | new Taintedness( SecurityCheckPlugin::ESCAPED_EXEC_TAINT ), |
| 1089 | $label, |
| 1090 | 'HTMLForm label key escapes its input' |
| 1091 | ); |
| 1092 | } |
| 1093 | if ( $rawLabel !== null ) { |
| 1094 | $this->maybeEmitIssueSimplified( |
| 1095 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 1096 | $rawLabel, |
| 1097 | 'HTMLForm label-raw needs to escape input' |
| 1098 | ); |
| 1099 | } |
| 1100 | if ( $help !== null ) { |
| 1101 | $this->maybeEmitIssueSimplified( |
| 1102 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 1103 | $help, |
| 1104 | 'HTMLForm help needs to escape input' |
| 1105 | ); |
| 1106 | } |
| 1107 | if ( $isInfo && $raw === true ) { |
| 1108 | $this->maybeEmitIssueSimplified( |
| 1109 | new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ), |
| 1110 | $default, |
| 1111 | 'HTMLForm info field in raw mode needs to escape default key' |
| 1112 | ); |
| 1113 | } |
| 1114 | if ( $isInfo && ( $raw === false || $raw === null ) ) { |
| 1115 | $this->maybeEmitIssueSimplified( |
| 1116 | new Taintedness( SecurityCheckPlugin::ESCAPED_EXEC_TAINT ), |
| 1117 | $default, |
| 1118 | 'HTMLForm info field (non-raw) escapes default key already' |
| 1119 | ); |
| 1120 | } |
| 1121 | if ( !$isOptionsSafe && $options instanceof Node ) { |
| 1122 | $htmlExecTaint = new Taintedness( SecurityCheckPlugin::HTML_EXEC_TAINT ); |
| 1123 | $optTaint = $this->getTaintedness( $options ); |
| 1124 | $this->maybeEmitIssue( |
| 1125 | $htmlExecTaint, |
| 1126 | $optTaint->getTaintedness()->asKeyForForeach(), |
| 1127 | 'HTMLForm option label needs escaping{DETAILS}', |
| 1128 | [ $optTaint->getError() ] |
| 1129 | ); |
| 1130 | } |
| 1131 | } |
| 1132 | |
| 1133 | /** |
| 1134 | * Try to detect HTMLForm specifiers |
| 1135 | * |
| 1136 | * @param Node $node |
| 1137 | */ |
| 1138 | public function visitArray( Node $node ): void { |
| 1139 | parent::visitArray( $node ); |
| 1140 | // Performance: use isset(), not property_exists |
| 1141 | // @phan-suppress-next-line PhanUndeclaredProperty |
| 1142 | if ( !isset( $node->skipHTMLFormAnalysis ) ) { |
| 1143 | $this->detectHTMLForm( $node ); |
| 1144 | } |
| 1145 | } |
| 1146 | } |