Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 540 |
|
0.00% |
0 / 17 |
CRAP | |
0.00% |
0 / 1 |
| SecurityCheckPlugin | |
0.00% |
0 / 540 |
|
0.00% |
0 / 17 |
4830 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| assertRequiredConfig | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getMergeVariableInfoClosure | |
0.00% |
0 / 41 |
|
0.00% |
0 / 1 |
342 | |||
| analyzeStringLiteralStatement | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
72 | |||
| taintToString | |
0.00% |
0 / 52 |
|
0.00% |
0 / 1 |
30 | |||
| builtinFuncHasTaint | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getBuiltinFuncTaint | |
0.00% |
0 / 38 |
|
0.00% |
0 / 1 |
56 | |||
| assertFunctionTaintArrayWellFormed | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
72 | |||
| getCustomFuncTaints | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
0 | |||
| isFalsePositive | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| parseTaintLine | |
0.00% |
0 / 40 |
|
0.00% |
0 / 1 |
156 | |||
| modifyParamSinkTaint | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| modifyArgTaint | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| convertTaintNameToConstant | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
2 | |||
| getPHPFuncTaints | |
0.00% |
0 / 295 |
|
0.00% |
0 / 1 |
2 | |||
| getBeforeLoopBodyAnalysisVisitorClassName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| clearCaches | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php declare( strict_types=1 ); |
| 2 | |
| 3 | /** |
| 4 | * Base class for SecurityCheckPlugin. Extend if you want to customize. |
| 5 | * |
| 6 | * Copyright (C) 2017 Brian Wolff <bawolff@gmail.com> |
| 7 | * |
| 8 | * @license GPL-2.0-or-later |
| 9 | */ |
| 10 | |
| 11 | namespace SecurityCheckPlugin; |
| 12 | |
| 13 | use ast\Node; |
| 14 | use Closure; |
| 15 | use InvalidArgumentException; |
| 16 | use LogicException; |
| 17 | use Phan\CodeBase; |
| 18 | use Phan\Config; |
| 19 | use Phan\Language\Context; |
| 20 | use Phan\Language\Element\Comment\Builder; |
| 21 | use Phan\Language\Element\FunctionInterface; |
| 22 | use Phan\Language\Element\Variable; |
| 23 | use Phan\Language\FQSEN\FullyQualifiedFunctionLikeName; |
| 24 | use Phan\Language\FQSEN\FullyQualifiedMethodName; |
| 25 | use Phan\Language\Scope; |
| 26 | use Phan\PluginV3; |
| 27 | use Phan\PluginV3\AnalyzeLiteralStatementCapability; |
| 28 | use Phan\PluginV3\BeforeLoopBodyAnalysisCapability; |
| 29 | use Phan\PluginV3\MergeVariableInfoCapability; |
| 30 | use Phan\PluginV3\PostAnalyzeNodeCapability; |
| 31 | use Phan\PluginV3\PreAnalyzeNodeCapability; |
| 32 | use RuntimeException; |
| 33 | |
| 34 | /** |
| 35 | * Base class used by the Generic and MediaWiki flavours of the plugin. |
| 36 | */ |
| 37 | abstract class SecurityCheckPlugin extends PluginV3 implements |
| 38 | PostAnalyzeNodeCapability, |
| 39 | PreAnalyzeNodeCapability, |
| 40 | BeforeLoopBodyAnalysisCapability, |
| 41 | MergeVariableInfoCapability, |
| 42 | AnalyzeLiteralStatementCapability |
| 43 | { |
| 44 | use TaintednessAccessorsTrait; |
| 45 | |
| 46 | // Various taint flags. The _EXEC_ varieties mean |
| 47 | // that it is unsafe to assign that type of taint |
| 48 | // to the variable in question. |
| 49 | |
| 50 | public const NO_TAINT = 0; |
| 51 | |
| 52 | // Flag to denote that we don't know |
| 53 | public const UNKNOWN_TAINT = 1 << 0; |
| 54 | |
| 55 | // Flag for function parameters and the like, where it |
| 56 | // preserves whatever taint the function is given. |
| 57 | public const PRESERVE_TAINT = 1 << 1; |
| 58 | |
| 59 | // In future might separate out different types of html quoting. |
| 60 | // e.g. "<div data-foo='" . htmlspecialchars( $bar ) . "'>"; |
| 61 | // is unsafe. |
| 62 | public const HTML_TAINT = 1 << 2; |
| 63 | public const HTML_EXEC_TAINT = 1 << 3; |
| 64 | |
| 65 | public const SQL_TAINT = 1 << 4; |
| 66 | public const SQL_EXEC_TAINT = 1 << 5; |
| 67 | |
| 68 | public const SHELL_TAINT = 1 << 6; |
| 69 | public const SHELL_EXEC_TAINT = 1 << 7; |
| 70 | |
| 71 | public const SERIALIZE_TAINT = 1 << 8; |
| 72 | public const SERIALIZE_EXEC_TAINT = 1 << 9; |
| 73 | |
| 74 | // Tainted paths, as input to include(), require() and some FS functions (path traversal) |
| 75 | public const PATH_TAINT = 1 << 10; |
| 76 | public const PATH_EXEC_TAINT = 1 << 11; |
| 77 | |
| 78 | // User-controlled code, for RCE |
| 79 | public const CODE_TAINT = 1 << 12; |
| 80 | public const CODE_EXEC_TAINT = 1 << 13; |
| 81 | |
| 82 | // User-controlled regular expressions, for ReDoS |
| 83 | public const REGEX_TAINT = 1 << 14; |
| 84 | public const REGEX_EXEC_TAINT = 1 << 15; |
| 85 | |
| 86 | // To allow people to add other application specific taints. |
| 87 | public const CUSTOM1_TAINT = 1 << 16; |
| 88 | public const CUSTOM1_EXEC_TAINT = 1 << 17; |
| 89 | public const CUSTOM2_TAINT = 1 << 18; |
| 90 | public const CUSTOM2_EXEC_TAINT = 1 << 19; |
| 91 | |
| 92 | // Special purpose for supporting MediaWiki's IDatabase::select |
| 93 | // and friends. Like SQL_TAINT, but only applies to the numeric |
| 94 | // keys of an array. Note: These are not included in YES_TAINT/EXEC_TAINT. |
| 95 | // e.g. given $f = [ $_GET['foo'] ]; $f would have the flag, but |
| 96 | // $g = $_GET['foo']; or $h = [ 's' => $_GET['foo'] ] would not. |
| 97 | // The associative keys also have this flag if they are tainted. |
| 98 | // It is also assumed anything with this flag will also have |
| 99 | // the SQL_TAINT flag set. |
| 100 | public const SQL_NUMKEY_TAINT = 1 << 20; |
| 101 | public const SQL_NUMKEY_EXEC_TAINT = 1 << 21; |
| 102 | |
| 103 | // For double escaped variables |
| 104 | public const ESCAPED_TAINT = 1 << 22; |
| 105 | public const ESCAPED_EXEC_TAINT = 1 << 23; |
| 106 | |
| 107 | // Special purpose flags (Starting at 2^28) |
| 108 | // TODO Renumber these. Requires changing format of the hardcoded arrays |
| 109 | // Cancel's out all EXEC flags on a function arg if arg is array. |
| 110 | public const ARRAY_OK = 1 << 28; |
| 111 | |
| 112 | // Do not allow autodetected taint info override given taint. |
| 113 | // TODO Store this and other special flags somewhere else in the FunctionTaintedness object, not |
| 114 | // as normal taint flags. |
| 115 | public const NO_OVERRIDE = 1 << 29; |
| 116 | |
| 117 | public const VARIADIC_PARAM = 1 << 30; |
| 118 | |
| 119 | // *All* function flags |
| 120 | //TODO Add a structure test for this |
| 121 | public const FUNCTION_FLAGS = self::ARRAY_OK | self::NO_OVERRIDE; |
| 122 | |
| 123 | // Combination flags. |
| 124 | |
| 125 | // YES_TAINT denotes all taint a user controlled variable would have |
| 126 | public const YES_TAINT = self::HTML_TAINT | self::SQL_TAINT | self::SHELL_TAINT | self::SERIALIZE_TAINT | |
| 127 | self::PATH_TAINT | self::CODE_TAINT | self::REGEX_TAINT | self::CUSTOM1_TAINT | self::CUSTOM2_TAINT; |
| 128 | public const EXEC_TAINT = self::YES_TAINT << 1; |
| 129 | // @phan-suppress-next-line PhanUnreferencedPublicClassConstant |
| 130 | public const YES_EXEC_TAINT = self::YES_TAINT | self::EXEC_TAINT; |
| 131 | |
| 132 | // ALL taint is YES + special purpose taints, but not including special flags. |
| 133 | public const ALL_TAINT = self::YES_TAINT | self::SQL_NUMKEY_TAINT | self::ESCAPED_TAINT; |
| 134 | public const ALL_EXEC_TAINT = |
| 135 | self::EXEC_TAINT | self::SQL_NUMKEY_EXEC_TAINT | self::ESCAPED_EXEC_TAINT; |
| 136 | public const ALL_YES_EXEC_TAINT = self::ALL_TAINT | self::ALL_EXEC_TAINT; |
| 137 | |
| 138 | // Taints that support backpropagation. |
| 139 | public const BACKPROP_TAINTS = self::ALL_EXEC_TAINT; |
| 140 | |
| 141 | public const ESCAPES_HTML = ( self::YES_TAINT & ~self::HTML_TAINT ) | self::ESCAPED_EXEC_TAINT; |
| 142 | |
| 143 | // As the name would suggest, this must include *ALL* possible taint flags. |
| 144 | public const ALL_TAINT_FLAGS = self::ALL_YES_EXEC_TAINT | self::FUNCTION_FLAGS | |
| 145 | self::UNKNOWN_TAINT | self::PRESERVE_TAINT | self::VARIADIC_PARAM; |
| 146 | |
| 147 | /** |
| 148 | * Used to print taint debug data, see BlockAnalysisVisitor::PHAN_DEBUG_VAR_REGEX |
| 149 | */ |
| 150 | private const DEBUG_TAINTEDNESS_REGEXP = |
| 151 | '/@phan-debug-var-taintedness\s+\$(' . Builder::WORD_REGEX . '(,\s*\$' . Builder::WORD_REGEX . ')*)/'; |
| 152 | // @phan-suppress-previous-line PhanAccessClassConstantInternal It's just perfect for use here |
| 153 | |
| 154 | public const PARAM_ANNOTATION_REGEX = |
| 155 | '/@param-taint\s+&?(?P<variadic>\.\.\.)?\$(?P<paramname>\S+)\s+(?P<taint>.*)$/'; |
| 156 | |
| 157 | /** |
| 158 | * @var self Passed to the visitor for context |
| 159 | */ |
| 160 | public static $pluginInstance; |
| 161 | |
| 162 | /** |
| 163 | * @var array<array<FunctionTaintedness|MethodLinks>> Cache of parsed docblocks. This is declared here (as opposed |
| 164 | * to the BaseVisitor) so that PHPUnit can snapshot and restore it. |
| 165 | * @phan-var array<array{0:FunctionTaintedness,1:MethodLinks}> |
| 166 | */ |
| 167 | public static $docblockCache = []; |
| 168 | |
| 169 | /** @var FunctionTaintedness[] Cache of taintedness of builtin functions */ |
| 170 | private static $builtinFuncTaintCache = []; |
| 171 | |
| 172 | /** |
| 173 | * Save the subclass instance to make it accessible from the visitor |
| 174 | */ |
| 175 | public function __construct() { |
| 176 | $this->assertRequiredConfig(); |
| 177 | self::$pluginInstance = $this; |
| 178 | } |
| 179 | |
| 180 | /** |
| 181 | * Ensure that the options we need are enabled. |
| 182 | */ |
| 183 | private function assertRequiredConfig(): void { |
| 184 | if ( Config::get_quick_mode() ) { |
| 185 | throw new RuntimeException( 'Quick mode must be disabled to run taint-check' ); |
| 186 | } |
| 187 | } |
| 188 | |
| 189 | /** |
| 190 | * @inheritDoc |
| 191 | */ |
| 192 | public function getMergeVariableInfoClosure(): Closure { |
| 193 | /** |
| 194 | * For branches that are not guaranteed to be executed, merge taint info for any involved |
| 195 | * variable across all branches. |
| 196 | * |
| 197 | * @note This method is HOT, so keep it optimized |
| 198 | * |
| 199 | * @param Variable $variable |
| 200 | * @param Scope[] $scopeList |
| 201 | * @param bool $varExistsInAllScopes @phan-unused-param |
| 202 | * @suppress PhanUnreferencedClosure, PhanUndeclaredProperty, UnusedSuppression |
| 203 | */ |
| 204 | return static function ( Variable $variable, array $scopeList, bool $varExistsInAllScopes ): void { |
| 205 | $varName = $variable->getName(); |
| 206 | |
| 207 | $vars = []; |
| 208 | $firstVar = null; |
| 209 | foreach ( $scopeList as $scope ) { |
| 210 | $localVar = $scope->getVariableByNameOrNull( $varName ); |
| 211 | if ( $localVar ) { |
| 212 | if ( !$firstVar ) { |
| 213 | $firstVar = $localVar; |
| 214 | } else { |
| 215 | $vars[] = $localVar; |
| 216 | } |
| 217 | } |
| 218 | } |
| 219 | |
| 220 | if ( !$firstVar ) { |
| 221 | return; |
| 222 | } |
| 223 | |
| 224 | /** @var Taintedness $taintedness */ |
| 225 | $taintedness = $prevTaint = $firstVar->taintedness ?? null; |
| 226 | /** @var MethodLinks $methodLinks */ |
| 227 | $methodLinks = $prevLinks = $firstVar->taintedMethodLinks ?? null; |
| 228 | /** @var CausedByLines $error */ |
| 229 | $error = $prevErr = $firstVar->taintedOriginalError ?? null; |
| 230 | |
| 231 | foreach ( $vars as $localVar ) { |
| 232 | // Below we only merge data if it's non-null in the current scope and different from the previous |
| 233 | // branch. Using arrays to save all previous values and then in_array seems useless on MW core, |
| 234 | // since >99% cases of duplication are already covered by these simple checks. |
| 235 | |
| 236 | $taintOrNull = $localVar->taintedness ?? null; |
| 237 | if ( $taintOrNull && $taintOrNull !== $prevTaint ) { |
| 238 | $prevTaint = $taintOrNull; |
| 239 | if ( $taintedness ) { |
| 240 | $taintedness = $taintedness->asMergedWith( $taintOrNull ); |
| 241 | } else { |
| 242 | $taintedness = $taintOrNull; |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | $variableObjLinksOrNull = $localVar->taintedMethodLinks ?? null; |
| 247 | if ( $variableObjLinksOrNull && $variableObjLinksOrNull !== $prevLinks ) { |
| 248 | $prevLinks = $variableObjLinksOrNull; |
| 249 | if ( $methodLinks ) { |
| 250 | $methodLinks = $methodLinks->asMergedWith( $variableObjLinksOrNull ); |
| 251 | } else { |
| 252 | $methodLinks = $variableObjLinksOrNull; |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | $varErrorOrNull = $localVar->taintedOriginalError ?? null; |
| 257 | if ( $varErrorOrNull && $varErrorOrNull !== $prevErr ) { |
| 258 | $prevErr = $varErrorOrNull; |
| 259 | if ( $error ) { |
| 260 | $error = $error->asMergedWith( $varErrorOrNull ); |
| 261 | } else { |
| 262 | $error = $varErrorOrNull; |
| 263 | } |
| 264 | } |
| 265 | } |
| 266 | |
| 267 | if ( $taintedness ) { |
| 268 | self::setTaintednessRaw( $variable, $taintedness ); |
| 269 | } |
| 270 | if ( $methodLinks ) { |
| 271 | self::setMethodLinks( $variable, $methodLinks ); |
| 272 | } |
| 273 | if ( $error ) { |
| 274 | self::setCausedByRaw( $variable, $error ); |
| 275 | } |
| 276 | }; |
| 277 | } |
| 278 | |
| 279 | /** |
| 280 | * Print the taintedness of a variable, when requested |
| 281 | * @see BlockAnalysisVisitor::analyzeSubstituteVarAssert() |
| 282 | * @inheritDoc |
| 283 | * @suppress PhanUndeclaredProperty, UnusedSuppression |
| 284 | */ |
| 285 | public function analyzeStringLiteralStatement( CodeBase $codeBase, Context $context, string $statement ): bool { |
| 286 | $found = false; |
| 287 | if ( preg_match_all( self::DEBUG_TAINTEDNESS_REGEXP, $statement, $matches, PREG_SET_ORDER ) ) { |
| 288 | $scope = $context->getScope(); |
| 289 | foreach ( $matches as $group ) { |
| 290 | foreach ( explode( ',', $group[1] ) as $rawVar ) { |
| 291 | $varName = ltrim( trim( $rawVar ), '$' ); |
| 292 | if ( $scope->hasVariableWithName( $varName ) ) { |
| 293 | $var = $scope->getVariableByName( $varName ); |
| 294 | $taintOrNull = self::getTaintednessRaw( $var ); |
| 295 | $taint = $taintOrNull ? $taintOrNull->toShortString() : 'unset'; |
| 296 | $msg = "Variable {CODE} has taintedness: {DETAILS}"; |
| 297 | $params = [ "\$$varName", $taint ]; |
| 298 | } else { |
| 299 | $msg = "Variable {CODE} doesn't exist in scope"; |
| 300 | $params = [ "\$$varName" ]; |
| 301 | } |
| 302 | self::emitIssue( |
| 303 | $codeBase, |
| 304 | $context, |
| 305 | 'SecurityCheckDebugTaintedness', |
| 306 | $msg, |
| 307 | $params |
| 308 | ); |
| 309 | $found = true; |
| 310 | } |
| 311 | } |
| 312 | } elseif ( str_contains( $statement, '@taint-check-debug-method-first-arg' ) ) { |
| 313 | // FIXME This is a hack. The annotation is INTERNAL, for use only in the backpropoffsets-blowup |
| 314 | // test. We should either find a better way to test that, or maybe add a public annotation |
| 315 | // for debugging taintedness of a method (probably unreadable on a single line). |
| 316 | $funcName = preg_replace( '/@taint-check-debug-method-first-arg ([a-z0-9:]+)\b.*/i', '$1', $statement ); |
| 317 | // Let any exception bubble up here, the annotation is for internal use in testing |
| 318 | $fqsen = FullyQualifiedMethodName::fromStringInContext( $funcName, $context ); |
| 319 | $method = $codeBase->getMethodByFQSEN( $fqsen ); |
| 320 | /** @var FunctionTaintedness|null $fTaint */ |
| 321 | $fTaint = $method->funcTaint ?? null; |
| 322 | if ( !$fTaint ) { |
| 323 | return false; |
| 324 | } |
| 325 | self::emitIssue( |
| 326 | $codeBase, |
| 327 | $context, |
| 328 | 'SecurityCheckDebugTaintedness', |
| 329 | "Method {CODE} has first param with taintedness: {DETAILS}", |
| 330 | [ $funcName, $fTaint->getParamSinkTaint( 0 )->toShortString() ] |
| 331 | ); |
| 332 | return true; |
| 333 | } |
| 334 | return $found; |
| 335 | } |
| 336 | |
| 337 | /** |
| 338 | * Get a string representation of a taint integer |
| 339 | * |
| 340 | * The prefix ~ means all input taints except the letter given. |
| 341 | * The prefix * means the EXEC version of the taint. |
| 342 | */ |
| 343 | public static function taintToString( int $taint ): string { |
| 344 | if ( $taint === self::NO_TAINT ) { |
| 345 | return 'NONE'; |
| 346 | } |
| 347 | |
| 348 | // Note, order matters here. |
| 349 | static $mapping = [ |
| 350 | self::UNKNOWN_TAINT => 'UNKNOWN', |
| 351 | self::PRESERVE_TAINT => 'PRESERVE', |
| 352 | self::ALL_TAINT => 'ALL', |
| 353 | self::YES_TAINT => 'YES', |
| 354 | self::YES_TAINT & |
| 355 | ( ~self::HTML_TAINT ) => '~HTML', |
| 356 | self::YES_TAINT & |
| 357 | ( ~self::SQL_TAINT ) => '~SQL', |
| 358 | self::YES_TAINT & |
| 359 | ( ~self::SHELL_TAINT ) => '~SHELL', |
| 360 | self::YES_TAINT & |
| 361 | ( ~self::SERIALIZE_TAINT ) => '~SERIALIZE', |
| 362 | self::YES_TAINT & |
| 363 | ( ~self::CUSTOM1_TAINT ) => '~CUSTOM1', |
| 364 | self::YES_TAINT & |
| 365 | ( ~self::CUSTOM2_TAINT ) => '~CUSTOM2', |
| 366 | // We skip ~ versions of flags which shouldn't be possible. |
| 367 | self::HTML_TAINT => 'HTML', |
| 368 | self::SQL_TAINT => 'SQL', |
| 369 | self::SHELL_TAINT => 'SHELL', |
| 370 | self::ESCAPED_TAINT => 'ESCAPED', |
| 371 | self::SERIALIZE_TAINT => 'SERIALIZE', |
| 372 | self::CUSTOM1_TAINT => 'CUSTOM1', |
| 373 | self::CUSTOM2_TAINT => 'CUSTOM2', |
| 374 | self::CODE_TAINT => 'CODE', |
| 375 | self::PATH_TAINT => 'PATH', |
| 376 | self::REGEX_TAINT => 'REGEX', |
| 377 | self::SQL_NUMKEY_TAINT => 'SQL_NUMKEY', |
| 378 | self::ARRAY_OK => 'ARRAY_OK', |
| 379 | self::ALL_EXEC_TAINT => '*ALL', |
| 380 | self::HTML_EXEC_TAINT => '*HTML', |
| 381 | self::SQL_EXEC_TAINT => '*SQL', |
| 382 | self::SHELL_EXEC_TAINT => '*SHELL', |
| 383 | self::ESCAPED_EXEC_TAINT => '*ESCAPED', |
| 384 | self::SERIALIZE_EXEC_TAINT => '*SERIALIZE', |
| 385 | self::CUSTOM1_EXEC_TAINT => '*CUSTOM1', |
| 386 | self::CUSTOM2_EXEC_TAINT => '*CUSTOM2', |
| 387 | self::CODE_EXEC_TAINT => '*CODE', |
| 388 | self::PATH_EXEC_TAINT => '*PATH', |
| 389 | self::REGEX_EXEC_TAINT => '*REGEX', |
| 390 | self::SQL_NUMKEY_EXEC_TAINT => '*SQL_NUMKEY', |
| 391 | ]; |
| 392 | |
| 393 | $types = []; |
| 394 | foreach ( $mapping as $bitmap => $val ) { |
| 395 | if ( ( $bitmap & $taint ) === $bitmap ) { |
| 396 | $types[] = $val; |
| 397 | $taint &= ~$bitmap; |
| 398 | } |
| 399 | } |
| 400 | if ( $taint !== 0 ) { |
| 401 | $types[] = "Unrecognized: $taint"; |
| 402 | } |
| 403 | return implode( ', ', $types ); |
| 404 | } |
| 405 | |
| 406 | public function builtinFuncHasTaint( FullyQualifiedFunctionLikeName $fqsen ): bool { |
| 407 | return $this->getBuiltinFuncTaint( $fqsen ) !== null; |
| 408 | } |
| 409 | |
| 410 | /** |
| 411 | * Get the taintedness of a function |
| 412 | * |
| 413 | * This allows overriding the default taint of a function |
| 414 | * |
| 415 | * If you want to provide custom taint hints for your application, |
| 416 | * override the getCustomFuncTaints() |
| 417 | * |
| 418 | * @param FullyQualifiedFunctionLikeName $fqsen The function/method in question |
| 419 | * @return FunctionTaintedness|null Null to autodetect taintedness |
| 420 | */ |
| 421 | public function getBuiltinFuncTaint( FullyQualifiedFunctionLikeName $fqsen ): ?FunctionTaintedness { |
| 422 | $name = (string)$fqsen; |
| 423 | |
| 424 | if ( isset( self::$builtinFuncTaintCache[$name] ) ) { |
| 425 | return self::$builtinFuncTaintCache[$name]; |
| 426 | } |
| 427 | |
| 428 | static $funcTaints = null; |
| 429 | if ( $funcTaints === null ) { |
| 430 | $funcTaints = $this->getCustomFuncTaints() + $this->getPHPFuncTaints(); |
| 431 | } |
| 432 | |
| 433 | if ( isset( $funcTaints[$name] ) ) { |
| 434 | $rawFuncTaint = $funcTaints[$name]; |
| 435 | if ( $rawFuncTaint instanceof FunctionTaintedness ) { |
| 436 | $funcTaint = $rawFuncTaint; |
| 437 | } else { |
| 438 | self::assertFunctionTaintArrayWellFormed( $rawFuncTaint ); |
| 439 | // Note: for backcompat, we set NO_OVERRIDE everywhere. |
| 440 | $overallFlags = ( $rawFuncTaint['overall'] & self::FUNCTION_FLAGS ) | self::NO_OVERRIDE; |
| 441 | $funcTaint = new FunctionTaintedness( |
| 442 | new Taintedness( $rawFuncTaint['overall'] & ~$overallFlags ), |
| 443 | $overallFlags |
| 444 | ); |
| 445 | unset( $rawFuncTaint['overall'] ); |
| 446 | foreach ( $rawFuncTaint as $i => $val ) { |
| 447 | assert( ( $val & self::UNKNOWN_TAINT ) === 0, 'Cannot set UNKNOWN' ); |
| 448 | $paramFlags = ( $val & self::FUNCTION_FLAGS ) | self::NO_OVERRIDE; |
| 449 | // TODO Split sink and preserve in the hardcoded arrays |
| 450 | if ( $val & self::VARIADIC_PARAM ) { |
| 451 | $pTaint = new Taintedness( $val & ~( self::VARIADIC_PARAM | $paramFlags ) ); |
| 452 | $funcTaint = $funcTaint |
| 453 | ->withVariadicParamSinkTaint( $i, $pTaint->withOnly( self::ALL_EXEC_TAINT ), $paramFlags ) |
| 454 | ->withVariadicParamPreservedTaint( |
| 455 | $i, |
| 456 | $pTaint->without( self::ALL_EXEC_TAINT )->asPreservedTaintedness() |
| 457 | ); |
| 458 | } else { |
| 459 | $pTaint = new Taintedness( $val & ~$paramFlags ); |
| 460 | $funcTaint = $funcTaint |
| 461 | ->withParamSinkTaint( $i, $pTaint->withOnly( self::ALL_EXEC_TAINT ), $paramFlags ) |
| 462 | ->withParamPreservedTaint( |
| 463 | $i, |
| 464 | $pTaint->without( self::ALL_EXEC_TAINT )->asPreservedTaintedness() |
| 465 | ); |
| 466 | } |
| 467 | } |
| 468 | } |
| 469 | self::$builtinFuncTaintCache[$name] = $funcTaint; |
| 470 | return self::$builtinFuncTaintCache[$name]; |
| 471 | } |
| 472 | return null; |
| 473 | } |
| 474 | |
| 475 | /** |
| 476 | * Assert that a taintedness array is well-formed, and fail hard if it isn't. |
| 477 | * |
| 478 | * @param int[] $taint |
| 479 | */ |
| 480 | private static function assertFunctionTaintArrayWellFormed( array $taint ): void { |
| 481 | if ( !isset( $taint['overall'] ) ) { |
| 482 | throw new LogicException( 'Overall taint must be set' ); |
| 483 | } |
| 484 | |
| 485 | foreach ( $taint as $i => $t ) { |
| 486 | if ( !is_int( $i ) && $i !== 'overall' ) { |
| 487 | throw new LogicException( "Taint indexes must be int or 'overall', got '$i'" ); |
| 488 | } |
| 489 | if ( !is_int( $t ) || ( $t & ~self::ALL_TAINT_FLAGS ) ) { |
| 490 | throw new LogicException( "Wrong taint index $i, got: " . var_export( $t, true ) ); |
| 491 | } |
| 492 | if ( $t & ~self::ALL_TAINT_FLAGS ) { |
| 493 | throw new LogicException( "Taint index $i has unknown flags: " . decbin( $t ) ); |
| 494 | } |
| 495 | } |
| 496 | } |
| 497 | |
| 498 | /** |
| 499 | * Get an array of function taints custom for the application |
| 500 | * |
| 501 | * @return array<string,int[]|FunctionTaintedness> Array of function taints. The keys are FQSENs. The values can be |
| 502 | * either FunctionTaintedness objects, or arrays with 'overall' string key and numeric keys for parameters. |
| 503 | * |
| 504 | * For example: [ self::YES_TAINT, 'overall' => self::NO_TAINT ] |
| 505 | * means that the taint of the return value is the same as the taint |
| 506 | * of the first arg, and all other args are ignored. |
| 507 | * [ self::HTML_EXEC_TAINT, 'overall' => self::NO_TAINT ] |
| 508 | * Means that the first arg is output in an html context (e.g. like echo) |
| 509 | * [ self::YES_TAINT & ~self::HTML_TAINT, 'overall' => self::NO_TAINT ] |
| 510 | * Means that the function removes html taint (escapes) e.g. htmlspecialchars |
| 511 | * [ 'overall' => self::YES_TAINT ] |
| 512 | * Means that it returns a tainted value (e.g. return $_POST['foo']; ) |
| 513 | * @see FunctionTaintedness for more details |
| 514 | */ |
| 515 | abstract protected function getCustomFuncTaints(): array; |
| 516 | |
| 517 | /** |
| 518 | * Can be used to force specific issues to be marked false positives |
| 519 | * |
| 520 | * For example, a specific application might be able to recognize |
| 521 | * that we are in a CLI context, and thus the XSS is really a false positive. |
| 522 | * |
| 523 | * @param int $combinedTaint Combined and adjusted taint of LHS+RHS |
| 524 | * @param string &$msg Issue description (so plugin can modify to state why false) |
| 525 | * @param Context $context |
| 526 | * @param CodeBase $code_base |
| 527 | * @return bool Is this a false positive? |
| 528 | * @suppress PhanUnusedPublicMethodParameter No param is used |
| 529 | */ |
| 530 | public function isFalsePositive( |
| 531 | int $combinedTaint, |
| 532 | string &$msg, |
| 533 | Context $context, |
| 534 | CodeBase $code_base |
| 535 | ): bool { |
| 536 | return false; |
| 537 | } |
| 538 | |
| 539 | /** |
| 540 | * Given a param description line, extract taint |
| 541 | * |
| 542 | * This is to allow putting taint information in method docblocks. |
| 543 | * If a function has a docblock comment like: |
| 544 | * * @param-taint $foo escapes_html |
| 545 | * This converts that line into: |
| 546 | * ( self::YES_TAINT & ~self::SQL_TAINT ) |
| 547 | * Multiple taint types are separated by commas |
| 548 | * (which are interpreted as bitwise OR ( "|" ). Future versions |
| 549 | * might support more complex bitwise operators, but for now it |
| 550 | * doesn't seem needed. |
| 551 | * |
| 552 | * The following keywords are supported where {type} can be |
| 553 | * html, sql, shell, serialize, custom1, custom2, sql_numkey, |
| 554 | * escaped. |
| 555 | * * {type} - just set the flag. 99% you should only use 'none' or 'tainted' |
| 556 | * * exec_{type} - sets the exec flag. |
| 557 | * * escapes_{type} - self::YES_TAINT & ~self::{type}_TAINT. |
| 558 | * Note: escapes_html adds the exec_escaped flag, use |
| 559 | * escapes_htmlnoent if the value is safe to double encode. |
| 560 | * * onlysafefor_{type} |
| 561 | * Same as above, intended for return type declarations. |
| 562 | * Only difference is that onlysafefor_html sets ESCAPED_TAINT instead |
| 563 | * of ESCAPED_EXEC_TAINT |
| 564 | * * none - self::NO_TAINT |
| 565 | * * tainted - self::YES_TAINT |
| 566 | * * array_ok - sets self::ARRAY_OK |
| 567 | * * allow_override - Allow autodetected taints to override annotation |
| 568 | * |
| 569 | * @todo What about ~ operator? |
| 570 | * @note The special casing to have escapes_html always add exec_escaped |
| 571 | * (and having htmlnoent exist) is "experimental" and may change in |
| 572 | * future versions (Maybe all types should set exec_escaped. Maybe it |
| 573 | * should be explicit) |
| 574 | * @note Excluding UNKNOWN here on purpose, as if we're setting it, it's not unknown |
| 575 | * @param string $line A line from the docblock |
| 576 | * @return array|null Array of [taintedness, flags], or null on no info |
| 577 | * @phan-return array{0:Taintedness,1:int}|null |
| 578 | */ |
| 579 | public static function parseTaintLine( string $line ): ?array { |
| 580 | $types = '(?P<type>htmlnoent|html|sql|shell|serialize|custom1|' |
| 581 | . 'custom2|code|path|regex|sql_numkey|escaped|none|tainted)'; |
| 582 | $prefixes = '(?P<prefix>escapes|onlysafefor|exec)'; |
| 583 | $taintExpr = "(?P<taint>(?:{$prefixes}_)?$types|array_ok|allow_override)"; |
| 584 | |
| 585 | $filteredLine = preg_replace( "/((?:$taintExpr,? *)+)(?: .*)?$/", '$1', $line ); |
| 586 | $taints = explode( ',', strtolower( $filteredLine ) ); |
| 587 | $taints = array_map( 'trim', $taints ); |
| 588 | |
| 589 | $overallTaint = Taintedness::safeSingleton(); |
| 590 | $overallFlags = self::NO_OVERRIDE; |
| 591 | $numberOfTaintsProcessed = 0; |
| 592 | foreach ( $taints as $taint ) { |
| 593 | $taintParts = []; |
| 594 | if ( !preg_match( "/^$taintExpr$/", $taint, $taintParts ) ) { |
| 595 | continue; |
| 596 | } |
| 597 | $numberOfTaintsProcessed++; |
| 598 | if ( $taintParts['taint'] === 'array_ok' ) { |
| 599 | $overallFlags |= self::ARRAY_OK; |
| 600 | continue; |
| 601 | } |
| 602 | if ( $taintParts['taint'] === 'allow_override' ) { |
| 603 | $overallFlags &= ~self::NO_OVERRIDE; |
| 604 | continue; |
| 605 | } |
| 606 | $taintAsInt = self::convertTaintNameToConstant( $taintParts['type'] ); |
| 607 | switch ( $taintParts['prefix'] ) { |
| 608 | case '': |
| 609 | $overallTaint = $overallTaint->with( $taintAsInt ); |
| 610 | break; |
| 611 | case 'exec': |
| 612 | $overallTaint = $overallTaint->with( Taintedness::flagsAsYesToExecTaint( $taintAsInt ) ); |
| 613 | break; |
| 614 | case 'escapes': |
| 615 | case 'onlysafefor': |
| 616 | $overallTaint = $overallTaint->with( self::YES_TAINT & ~$taintAsInt ); |
| 617 | if ( $taintParts['type'] === 'html' ) { |
| 618 | if ( $taintParts['prefix'] === 'escapes' ) { |
| 619 | $overallTaint = $overallTaint->with( self::ESCAPED_EXEC_TAINT ); |
| 620 | } else { |
| 621 | $overallTaint = $overallTaint->with( self::ESCAPED_TAINT ); |
| 622 | } |
| 623 | } |
| 624 | break; |
| 625 | } |
| 626 | } |
| 627 | if ( $numberOfTaintsProcessed === 0 ) { |
| 628 | return null; |
| 629 | } |
| 630 | return [ $overallTaint, $overallFlags ]; |
| 631 | } |
| 632 | |
| 633 | /** |
| 634 | * Hook to override the sink taintedness of a method parameter depending on the current argument. |
| 635 | * |
| 636 | * @internal This method is unstable and may be removed without prior notice. |
| 637 | * |
| 638 | * @param Taintedness $paramSinkTaint |
| 639 | * @param Taintedness $curArgTaintedness |
| 640 | * @param Node $argument Note: This hook is not called on literals |
| 641 | * @param int $argIndex Which argument number is this |
| 642 | * @param FunctionInterface $func The function/method being called |
| 643 | * @param FunctionTaintedness $funcTaint Taint of method formal parameters |
| 644 | * @param CausedByLines $paramSinkError |
| 645 | * @param Context $context Context object |
| 646 | * @param CodeBase $code_base CodeBase object |
| 647 | * @return array<Taintedness|CausedByLines> The taint and caused-by lines to use for actual parameter |
| 648 | * @phan-return array{0:Taintedness,1:CausedByLines} |
| 649 | * @suppress PhanUnusedPublicMethodParameter |
| 650 | */ |
| 651 | public function modifyParamSinkTaint( |
| 652 | Taintedness $paramSinkTaint, |
| 653 | Taintedness $curArgTaintedness, |
| 654 | Node $argument, |
| 655 | int $argIndex, |
| 656 | FunctionInterface $func, |
| 657 | FunctionTaintedness $funcTaint, |
| 658 | CausedByLines $paramSinkError, |
| 659 | Context $context, |
| 660 | CodeBase $code_base |
| 661 | ): array { |
| 662 | // no-op |
| 663 | return [ $paramSinkTaint, $paramSinkError ]; |
| 664 | } |
| 665 | |
| 666 | /** |
| 667 | * Hook to override how taint of an argument to method call is calculated |
| 668 | * |
| 669 | * @param Taintedness $curArgTaintedness |
| 670 | * @param Node $argument Note: This hook is not called on literals |
| 671 | * @param int $argIndex Which argument number is this |
| 672 | * @param FunctionInterface $func The function/method being called |
| 673 | * @param FunctionTaintedness $funcTaint Taint of method formal parameters |
| 674 | * @param Context $context Context object |
| 675 | * @param CodeBase $code_base CodeBase object |
| 676 | * @return Taintedness The taint to use for actual parameter |
| 677 | * @suppress PhanUnusedPublicMethodParameter |
| 678 | */ |
| 679 | public function modifyArgTaint( |
| 680 | Taintedness $curArgTaintedness, |
| 681 | Node $argument, |
| 682 | int $argIndex, |
| 683 | FunctionInterface $func, |
| 684 | FunctionTaintedness $funcTaint, |
| 685 | Context $context, |
| 686 | CodeBase $code_base |
| 687 | ): Taintedness { |
| 688 | // no-op |
| 689 | return $curArgTaintedness; |
| 690 | } |
| 691 | |
| 692 | /** |
| 693 | * Convert a string like 'html' to self::HTML_TAINT. |
| 694 | * |
| 695 | * @note htmlnoent treated like self::HTML_TAINT. |
| 696 | * @param string $name one of: |
| 697 | * html, sql, shell, serialize, custom1, custom2, code, path, regex, sql_numkey, |
| 698 | * escaped, none (= self::NO_TAINT), tainted (= self::YES_TAINT) |
| 699 | * @return int One of the TAINT constants |
| 700 | */ |
| 701 | public static function convertTaintNameToConstant( string $name ): int { |
| 702 | return match ( $name ) { |
| 703 | 'html', 'htmlnoent' => self::HTML_TAINT, |
| 704 | 'sql' => self::SQL_TAINT, |
| 705 | 'shell' => self::SHELL_TAINT, |
| 706 | 'serialize' => self::SERIALIZE_TAINT, |
| 707 | 'custom1' => self::CUSTOM1_TAINT, |
| 708 | 'custom2' => self::CUSTOM2_TAINT, |
| 709 | 'code' => self::CODE_TAINT, |
| 710 | 'path' => self::PATH_TAINT, |
| 711 | 'regex' => self::REGEX_TAINT, |
| 712 | 'sql_numkey' => self::SQL_NUMKEY_TAINT, |
| 713 | 'escaped' => self::ESCAPED_TAINT, |
| 714 | 'tainted' => self::YES_TAINT, |
| 715 | 'none' => self::NO_TAINT, |
| 716 | // @codeCoverageIgnoreStart |
| 717 | default => throw new InvalidArgumentException( "$name not valid taint" ) |
| 718 | // @codeCoverageIgnoreEnd |
| 719 | }; |
| 720 | } |
| 721 | |
| 722 | /** |
| 723 | * Taints for builtin php functions |
| 724 | * |
| 725 | * @return int[][] List of func taints (See getBuiltinFuncTaint()) |
| 726 | * @phan-return array<string,int[]> |
| 727 | */ |
| 728 | private function getPHPFuncTaints(): array { |
| 729 | $pregMatchTaint = [ |
| 730 | self::REGEX_EXEC_TAINT, |
| 731 | self::YES_TAINT, |
| 732 | // TODO: Possibly unsafe pass-by-ref |
| 733 | self::NO_TAINT, |
| 734 | self::NO_TAINT, |
| 735 | self::NO_TAINT, |
| 736 | 'overall' => self::NO_TAINT, |
| 737 | ]; |
| 738 | $pregReplaceTaint = [ |
| 739 | self::REGEX_EXEC_TAINT, |
| 740 | // TODO: This is used for strings (in preg_replace) and callbacks (in preg_replace_callback) |
| 741 | self::YES_TAINT, |
| 742 | self::YES_TAINT, |
| 743 | self::NO_TAINT, |
| 744 | self::NO_TAINT, |
| 745 | 'overall' => self::NO_TAINT |
| 746 | ]; |
| 747 | return [ |
| 748 | '\htmlentities' => [ |
| 749 | self::ESCAPES_HTML, |
| 750 | 'overall' => self::ESCAPED_TAINT |
| 751 | ], |
| 752 | '\htmlspecialchars' => [ |
| 753 | self::ESCAPES_HTML, |
| 754 | 'overall' => self::ESCAPED_TAINT |
| 755 | ], |
| 756 | '\escapeshellarg' => [ |
| 757 | ~self::SHELL_TAINT & self::YES_TAINT, |
| 758 | 'overall' => self::NO_TAINT |
| 759 | ], |
| 760 | // TODO: Perhaps we should distinguish arguments escape vs command escape |
| 761 | '\escapeshellcmd' => [ |
| 762 | ~self::SHELL_TAINT & self::YES_TAINT, |
| 763 | 'overall' => self::NO_TAINT |
| 764 | ], |
| 765 | '\shell_exec' => [ |
| 766 | self::SHELL_EXEC_TAINT, |
| 767 | 'overall' => self::YES_TAINT |
| 768 | ], |
| 769 | '\passthru' => [ |
| 770 | self::SHELL_EXEC_TAINT, |
| 771 | self::NO_TAINT, |
| 772 | 'overall' => self::NO_TAINT |
| 773 | ], |
| 774 | '\exec' => [ |
| 775 | self::SHELL_EXEC_TAINT, |
| 776 | // TODO: This is an unsafe passbyref |
| 777 | self::NO_TAINT, |
| 778 | self::NO_TAINT, |
| 779 | 'overall' => self::YES_TAINT |
| 780 | ], |
| 781 | '\system' => [ |
| 782 | self::SHELL_EXEC_TAINT, |
| 783 | self::NO_TAINT, |
| 784 | 'overall' => self::YES_TAINT |
| 785 | ], |
| 786 | '\proc_open' => [ |
| 787 | self::SHELL_EXEC_TAINT, |
| 788 | self::NO_TAINT, |
| 789 | // TODO: Unsafe passbyref |
| 790 | self::NO_TAINT, |
| 791 | self::NO_TAINT, |
| 792 | self::NO_TAINT, |
| 793 | self::NO_TAINT, |
| 794 | // TODO: Perhaps not so safe |
| 795 | 'overall' => self::NO_TAINT |
| 796 | ], |
| 797 | '\popen' => [ |
| 798 | self::SHELL_EXEC_TAINT, |
| 799 | self::NO_TAINT, |
| 800 | // TODO: Perhaps not so safe |
| 801 | 'overall' => self::NO_TAINT |
| 802 | ], |
| 803 | // Or any time the serialized data comes from a trusted source. |
| 804 | '\serialize' => [ |
| 805 | 'overall' => self::YES_TAINT & ~self::SERIALIZE_TAINT, |
| 806 | ], |
| 807 | '\unserialize' => [ |
| 808 | self::SERIALIZE_EXEC_TAINT, |
| 809 | 'overall' => self::NO_TAINT, |
| 810 | ], |
| 811 | '\mysql_query' => [ |
| 812 | self::SQL_EXEC_TAINT, |
| 813 | 'overall' => self::UNKNOWN_TAINT |
| 814 | ], |
| 815 | '\mysqli_query' => [ |
| 816 | self::NO_TAINT, |
| 817 | self::SQL_EXEC_TAINT, |
| 818 | 'overall' => self::UNKNOWN_TAINT |
| 819 | ], |
| 820 | '\mysqli::query' => [ |
| 821 | self::SQL_EXEC_TAINT, |
| 822 | 'overall' => self::UNKNOWN_TAINT |
| 823 | ], |
| 824 | '\mysqli_real_query' => [ |
| 825 | self::NO_TAINT, |
| 826 | self::SQL_EXEC_TAINT, |
| 827 | 'overall' => self::UNKNOWN_TAINT |
| 828 | ], |
| 829 | '\mysqli::real_query' => [ |
| 830 | self::SQL_EXEC_TAINT, |
| 831 | 'overall' => self::UNKNOWN_TAINT |
| 832 | ], |
| 833 | '\mysqli_fetch_all' => [ |
| 834 | 'overall' => self::YES_TAINT, |
| 835 | ], |
| 836 | '\mysqli_result::fetch' => [ |
| 837 | 'overall' => self::YES_TAINT, |
| 838 | ], |
| 839 | '\mysqli_fetch_array' => [ |
| 840 | 'overall' => self::YES_TAINT, |
| 841 | ], |
| 842 | '\mysqli_result::fetch_array' => [ |
| 843 | 'overall' => self::YES_TAINT, |
| 844 | ], |
| 845 | '\mysqli_fetch_assoc' => [ |
| 846 | 'overall' => self::YES_TAINT, |
| 847 | ], |
| 848 | '\mysqli_result::fetch_assoc' => [ |
| 849 | 'overall' => self::YES_TAINT, |
| 850 | ], |
| 851 | '\mysqli_fetch_column' => [ |
| 852 | 'overall' => self::YES_TAINT, |
| 853 | ], |
| 854 | '\mysqli_result::fetch_column' => [ |
| 855 | 'overall' => self::YES_TAINT, |
| 856 | ], |
| 857 | '\mysqli_fetch_object' => [ |
| 858 | 'overall' => self::YES_TAINT, |
| 859 | ], |
| 860 | '\mysqli_result::fetch_object' => [ |
| 861 | 'overall' => self::YES_TAINT, |
| 862 | ], |
| 863 | '\mysqli_fetch_row' => [ |
| 864 | 'overall' => self::YES_TAINT, |
| 865 | ], |
| 866 | '\mysqli_result::fetch_row' => [ |
| 867 | 'overall' => self::YES_TAINT, |
| 868 | ], |
| 869 | '\sqlite_query' => [ |
| 870 | self::NO_TAINT, |
| 871 | self::SQL_EXEC_TAINT, |
| 872 | self::NO_TAINT, |
| 873 | self::NO_TAINT, |
| 874 | 'overall' => self::UNKNOWN_TAINT |
| 875 | ], |
| 876 | '\sqlite_single_query' => [ |
| 877 | self::NO_TAINT, |
| 878 | self::SQL_EXEC_TAINT, |
| 879 | self::NO_TAINT, |
| 880 | self::NO_TAINT, |
| 881 | 'overall' => self::UNKNOWN_TAINT |
| 882 | ], |
| 883 | // Note: addslashes, addcslashes etc. intentionally omitted because they're not |
| 884 | // enough to avoid SQLi. |
| 885 | '\mysqli_escape_string' => [ |
| 886 | self::NO_TAINT, |
| 887 | self::YES_TAINT & ~self::SQL_TAINT, |
| 888 | 'overall' => self::NO_TAINT |
| 889 | ], |
| 890 | '\mysqli_real_escape_string' => [ |
| 891 | self::NO_TAINT, |
| 892 | self::YES_TAINT & ~self::SQL_TAINT, |
| 893 | 'overall' => self::NO_TAINT |
| 894 | ], |
| 895 | '\mysqli::escape_string' => [ |
| 896 | self::YES_TAINT & ~self::SQL_TAINT, |
| 897 | 'overall' => self::NO_TAINT |
| 898 | ], |
| 899 | '\mysqli::real_escape_string' => [ |
| 900 | self::YES_TAINT & ~self::SQL_TAINT, |
| 901 | 'overall' => self::NO_TAINT |
| 902 | ], |
| 903 | '\sqlite_escape_string' => [ |
| 904 | self::YES_TAINT & ~self::SQL_TAINT, |
| 905 | 'overall' => self::NO_TAINT |
| 906 | ], |
| 907 | '\PDO::query' => [ |
| 908 | self::SQL_EXEC_TAINT, |
| 909 | self::NO_TAINT, |
| 910 | self::NO_TAINT, |
| 911 | self::NO_TAINT, |
| 912 | 'overall' => self::UNKNOWN_TAINT |
| 913 | ], |
| 914 | '\PDO::prepare' => [ |
| 915 | self::SQL_EXEC_TAINT, |
| 916 | self::NO_TAINT, |
| 917 | 'overall' => self::UNKNOWN_TAINT |
| 918 | ], |
| 919 | '\PDO::exec' => [ |
| 920 | self::SQL_EXEC_TAINT, |
| 921 | 'overall' => self::NO_TAINT |
| 922 | ], |
| 923 | '\base64_encode' => [ |
| 924 | self::YES_TAINT & ~self::HTML_TAINT, |
| 925 | 'overall' => self::NO_TAINT |
| 926 | ], |
| 927 | '\file_put_contents' => [ |
| 928 | self::PATH_EXEC_TAINT, |
| 929 | self::NO_TAINT, |
| 930 | self::NO_TAINT, |
| 931 | self::NO_TAINT, |
| 932 | 'overall' => self::NO_TAINT |
| 933 | ], |
| 934 | // TODO: What about file_get_contents() and file() ? |
| 935 | '\fopen' => [ |
| 936 | self::PATH_EXEC_TAINT, |
| 937 | self::NO_TAINT, |
| 938 | self::NO_TAINT, |
| 939 | self::NO_TAINT, |
| 940 | // TODO: Perhaps not so safe |
| 941 | 'overall' => self::NO_TAINT |
| 942 | ], |
| 943 | '\opendir' => [ |
| 944 | self::PATH_EXEC_TAINT, |
| 945 | self::NO_TAINT, |
| 946 | // TODO: Perhaps not so safe |
| 947 | 'overall' => self::NO_TAINT |
| 948 | ], |
| 949 | '\rawurlencode' => [ |
| 950 | self::YES_TAINT & ~self::PATH_TAINT, |
| 951 | 'overall' => self::NO_TAINT |
| 952 | ], |
| 953 | '\urlencode' => [ |
| 954 | self::YES_TAINT & ~self::PATH_TAINT, |
| 955 | 'overall' => self::NO_TAINT |
| 956 | ], |
| 957 | '\printf' => [ |
| 958 | self::HTML_EXEC_TAINT, |
| 959 | // TODO We could check if the respective specifiers are safe |
| 960 | self::HTML_EXEC_TAINT | self::VARIADIC_PARAM, |
| 961 | 'overall' => self::NO_TAINT |
| 962 | ], |
| 963 | '\preg_filter' => [ |
| 964 | self::REGEX_EXEC_TAINT, |
| 965 | self::YES_TAINT, |
| 966 | self::YES_TAINT, |
| 967 | self::NO_TAINT, |
| 968 | self::NO_TAINT, |
| 969 | 'overall' => self::NO_TAINT |
| 970 | ], |
| 971 | '\preg_grep' => [ |
| 972 | self::REGEX_EXEC_TAINT, |
| 973 | self::YES_TAINT, |
| 974 | self::NO_TAINT, |
| 975 | 'overall' => self::NO_TAINT |
| 976 | ], |
| 977 | '\preg_match_all' => $pregMatchTaint, |
| 978 | '\preg_match' => $pregMatchTaint, |
| 979 | '\preg_quote' => [ |
| 980 | self::YES_TAINT & ~self::REGEX_TAINT, |
| 981 | self::NO_TAINT, |
| 982 | 'overall' => self::NO_TAINT |
| 983 | ], |
| 984 | '\preg_replace' => $pregReplaceTaint, |
| 985 | '\preg_replace_callback' => $pregReplaceTaint, |
| 986 | '\preg_replace_callback_array' => [ |
| 987 | self::REGEX_EXEC_TAINT, |
| 988 | self::YES_TAINT, |
| 989 | self::NO_TAINT, |
| 990 | self::NO_TAINT, |
| 991 | self::NO_TAINT, |
| 992 | 'overall' => self::NO_TAINT |
| 993 | ], |
| 994 | '\preg_split' => [ |
| 995 | self::REGEX_EXEC_TAINT, |
| 996 | self::YES_TAINT, |
| 997 | self::NO_TAINT, |
| 998 | self::NO_TAINT, |
| 999 | 'overall' => self::NO_TAINT |
| 1000 | ], |
| 1001 | // We assume that hashing functions are safe, see T272492 |
| 1002 | '\md5' => [ |
| 1003 | self::NO_TAINT, |
| 1004 | self::NO_TAINT, |
| 1005 | 'overall' => self::NO_TAINT |
| 1006 | ], |
| 1007 | '\sha1' => [ |
| 1008 | self::NO_TAINT, |
| 1009 | self::NO_TAINT, |
| 1010 | 'overall' => self::NO_TAINT |
| 1011 | ], |
| 1012 | '\crc32' => [ |
| 1013 | self::NO_TAINT, |
| 1014 | 'overall' => self::NO_TAINT |
| 1015 | ], |
| 1016 | '\hash' => [ |
| 1017 | self::NO_TAINT, |
| 1018 | self::NO_TAINT, |
| 1019 | self::NO_TAINT, |
| 1020 | self::NO_TAINT, |
| 1021 | 'overall' => self::NO_TAINT |
| 1022 | ], |
| 1023 | '\hash_hmac' => [ |
| 1024 | self::NO_TAINT, |
| 1025 | self::NO_TAINT, |
| 1026 | self::NO_TAINT, |
| 1027 | self::NO_TAINT, |
| 1028 | 'overall' => self::NO_TAINT |
| 1029 | ], |
| 1030 | // exit() and die() became ordinary functions in PHP 8.4 |
| 1031 | '\exit' => [ |
| 1032 | self::HTML_EXEC_TAINT, |
| 1033 | 'overall' => self::NO_TAINT |
| 1034 | ], |
| 1035 | '\die' => [ |
| 1036 | self::HTML_EXEC_TAINT, |
| 1037 | 'overall' => self::NO_TAINT |
| 1038 | ], |
| 1039 | ]; |
| 1040 | } |
| 1041 | |
| 1042 | /** |
| 1043 | * @inheritDoc |
| 1044 | */ |
| 1045 | public static function getBeforeLoopBodyAnalysisVisitorClassName(): string { |
| 1046 | return TaintednessLoopVisitor::class; |
| 1047 | } |
| 1048 | |
| 1049 | /** |
| 1050 | * Clear caches for testing. |
| 1051 | * @suppress PhanUnreferencedPublicMethod Used in tests (not analyzed by phan) |
| 1052 | */ |
| 1053 | public static function clearCaches(): void { |
| 1054 | self::$docblockCache = []; |
| 1055 | TaintednessVisitor::$fqsensWithoutToStringCache = []; |
| 1056 | } |
| 1057 | } |