Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
340 / 340 |
|
100.00% |
26 / 26 |
CRAP | |
100.00% |
1 / 1 |
| AFPTreeParser | |
100.00% |
340 / 340 |
|
100.00% |
26 / 26 |
111 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setFilter | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| resetState | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| move | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNextToken | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getState | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setState | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| parse | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| buildSyntaxTree | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| doLevelEntry | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| doLevelSemicolon | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
7 | |||
| doLevelSet | |
100.00% |
34 / 34 |
|
100.00% |
1 / 1 |
13 | |||
| doLevelConditions | |
100.00% |
56 / 56 |
|
100.00% |
1 / 1 |
13 | |||
| doLevelBoolOps | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelCompares | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
4 | |||
| doLevelSumRels | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelMulRels | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelPow | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelBoolInvert | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelKeywordOperators | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| doLevelUnarys | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| doLevelArrayElements | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
5 | |||
| doLevelParenthesis | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
7 | |||
| doLevelFunction | |
100.00% |
38 / 38 |
|
100.00% |
1 / 1 |
11 | |||
| doLevelAtom | |
100.00% |
46 / 46 |
|
100.00% |
1 / 1 |
16 | |||
| checkLogDeprecatedVar | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * A version of the abuse filter parser that separates parsing the filter and |
| 5 | * evaluating it into different passes, allowing the parse tree to be cached. |
| 6 | * |
| 7 | * @file |
| 8 | * @phan-file-suppress PhanPossiblyInfiniteRecursionSameParams Recursion controlled by class props |
| 9 | */ |
| 10 | |
| 11 | namespace MediaWiki\Extension\AbuseFilter\Parser; |
| 12 | |
| 13 | use MediaWiki\Extension\AbuseFilter\KeywordsManager; |
| 14 | use MediaWiki\Extension\AbuseFilter\Parser\Exception\UserVisibleException; |
| 15 | use Psr\Log\LoggerInterface; |
| 16 | use Wikimedia\Stats\StatsFactory; |
| 17 | use Wikimedia\Timestamp\ConvertibleTimestamp; |
| 18 | |
| 19 | /** |
| 20 | * A parser that transforms the text of the filter into a parse tree. |
| 21 | */ |
| 22 | class AFPTreeParser { |
| 23 | /** |
| 24 | * @var array<int,array{0:AFPToken,1:int}> Contains the AFPTokens for the code being parsed |
| 25 | */ |
| 26 | private $mTokens; |
| 27 | /** |
| 28 | * @var AFPToken The current token |
| 29 | */ |
| 30 | private $mCur; |
| 31 | /** @var int The position of the current token */ |
| 32 | private $mPos; |
| 33 | |
| 34 | /** |
| 35 | * @var string|null The ID of the filter being parsed, if available. Can also be "global-$ID" |
| 36 | */ |
| 37 | private $mFilter; |
| 38 | |
| 39 | public const CACHE_VERSION = 2; |
| 40 | |
| 41 | public function __construct( |
| 42 | private readonly LoggerInterface $logger, |
| 43 | private readonly StatsFactory $statsFactory, |
| 44 | private readonly KeywordsManager $keywordsManager |
| 45 | ) { |
| 46 | $this->resetState(); |
| 47 | } |
| 48 | |
| 49 | /** |
| 50 | * @param string $filter |
| 51 | */ |
| 52 | public function setFilter( $filter ) { |
| 53 | $this->mFilter = $filter; |
| 54 | } |
| 55 | |
| 56 | /** |
| 57 | * Resets the state |
| 58 | */ |
| 59 | private function resetState() { |
| 60 | $this->mTokens = []; |
| 61 | $this->mPos = 0; |
| 62 | $this->mFilter = null; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * Advances the parser to the next token in the filter code. |
| 67 | */ |
| 68 | private function move() { |
| 69 | [ $this->mCur, $this->mPos ] = $this->mTokens[$this->mPos]; |
| 70 | } |
| 71 | |
| 72 | /** |
| 73 | * Get the next token. This is similar to move() but doesn't change class members, |
| 74 | * allowing to look ahead without rolling back the state. |
| 75 | * |
| 76 | * @return AFPToken |
| 77 | */ |
| 78 | private function getNextToken() { |
| 79 | return $this->mTokens[$this->mPos][0]; |
| 80 | } |
| 81 | |
| 82 | /** |
| 83 | * getState() function allows parser state to be rollbacked to several tokens |
| 84 | * back. |
| 85 | * |
| 86 | * @return AFPParserState |
| 87 | */ |
| 88 | private function getState() { |
| 89 | return new AFPParserState( $this->mCur, $this->mPos ); |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * setState() function allows parser state to be rollbacked to several tokens |
| 94 | * back. |
| 95 | */ |
| 96 | private function setState( AFPParserState $state ) { |
| 97 | $this->mCur = $state->token; |
| 98 | $this->mPos = $state->pos; |
| 99 | } |
| 100 | |
| 101 | /** |
| 102 | * Parse the supplied filter source code into a tree. |
| 103 | * |
| 104 | * @param array<int,array{0:AFPToken,1:int}> $tokens |
| 105 | * @return AFPSyntaxTree |
| 106 | * @throws UserVisibleException |
| 107 | */ |
| 108 | public function parse( array $tokens ): AFPSyntaxTree { |
| 109 | $this->mTokens = $tokens; |
| 110 | $this->mPos = 0; |
| 111 | |
| 112 | return $this->buildSyntaxTree(); |
| 113 | } |
| 114 | |
| 115 | private function buildSyntaxTree(): AFPSyntaxTree { |
| 116 | $startTime = ConvertibleTimestamp::hrtime(); |
| 117 | $root = $this->doLevelEntry(); |
| 118 | $this->statsFactory->withComponent( 'AbuseFilter' ) |
| 119 | ->getTiming( 'parser_duration_seconds' ) |
| 120 | ->setLabel( 'phase', 'buildtree' ) |
| 121 | ->observeNanoseconds( ConvertibleTimestamp::hrtime() - $startTime ); |
| 122 | return new AFPSyntaxTree( $root ); |
| 123 | } |
| 124 | |
| 125 | /* Levels */ |
| 126 | |
| 127 | /** |
| 128 | * Handles unexpected characters after the expression. |
| 129 | * @return AFPTreeNode|null Null only if no statements |
| 130 | * @throws UserVisibleException |
| 131 | */ |
| 132 | private function doLevelEntry() { |
| 133 | $result = $this->doLevelSemicolon(); |
| 134 | |
| 135 | if ( $this->mCur->type !== AFPToken::TNONE ) { |
| 136 | throw new UserVisibleException( |
| 137 | 'unexpectedatend', |
| 138 | $this->mPos, [ $this->mCur->type ] |
| 139 | ); |
| 140 | } |
| 141 | |
| 142 | return $result; |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * Handles the semicolon operator. |
| 147 | * |
| 148 | * @return AFPTreeNode|null |
| 149 | */ |
| 150 | private function doLevelSemicolon() { |
| 151 | $statements = []; |
| 152 | |
| 153 | do { |
| 154 | $this->move(); |
| 155 | $position = $this->mPos; |
| 156 | |
| 157 | if ( |
| 158 | $this->mCur->type === AFPToken::TNONE || |
| 159 | ( $this->mCur->type === AFPToken::TBRACE && $this->mCur->value == ')' ) |
| 160 | ) { |
| 161 | // Handle special cases which the other parser handled in doLevelAtom |
| 162 | break; |
| 163 | } |
| 164 | |
| 165 | // Allow empty statements. |
| 166 | if ( $this->mCur->type === AFPToken::TSTATEMENTSEPARATOR ) { |
| 167 | continue; |
| 168 | } |
| 169 | |
| 170 | $statements[] = $this->doLevelSet(); |
| 171 | $position = $this->mPos; |
| 172 | } while ( $this->mCur->type === AFPToken::TSTATEMENTSEPARATOR ); |
| 173 | |
| 174 | // Flatten the tree if possible. |
| 175 | if ( count( $statements ) === 0 ) { |
| 176 | return null; |
| 177 | } elseif ( count( $statements ) === 1 ) { |
| 178 | return $statements[0]; |
| 179 | } else { |
| 180 | return new AFPTreeNode( AFPTreeNode::SEMICOLON, $statements, $position ); |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Handles variable assignment. |
| 186 | * |
| 187 | * @return AFPTreeNode |
| 188 | * @throws UserVisibleException |
| 189 | */ |
| 190 | private function doLevelSet() { |
| 191 | if ( $this->mCur->type === AFPToken::TID ) { |
| 192 | $varname = (string)$this->mCur->value; |
| 193 | |
| 194 | // Speculatively parse the assignment statement assuming it can |
| 195 | // potentially be an assignment, but roll back if it isn't. |
| 196 | // @todo Use $this->getNextToken for clearer code |
| 197 | $initialState = $this->getState(); |
| 198 | $this->move(); |
| 199 | |
| 200 | if ( $this->mCur->type === AFPToken::TOP && $this->mCur->value === ':=' ) { |
| 201 | $position = $this->mPos; |
| 202 | $this->move(); |
| 203 | $value = $this->doLevelSet(); |
| 204 | |
| 205 | return new AFPTreeNode( AFPTreeNode::ASSIGNMENT, [ $varname, $value ], $position ); |
| 206 | } |
| 207 | |
| 208 | if ( $this->mCur->type === AFPToken::TSQUAREBRACKET && $this->mCur->value === '[' ) { |
| 209 | $this->move(); |
| 210 | |
| 211 | if ( $this->mCur->type === AFPToken::TSQUAREBRACKET && $this->mCur->value === ']' ) { |
| 212 | $index = 'append'; |
| 213 | } else { |
| 214 | // Parse index offset. |
| 215 | $this->setState( $initialState ); |
| 216 | $this->move(); |
| 217 | $index = $this->doLevelSemicolon(); |
| 218 | if ( !( $this->mCur->type === AFPToken::TSQUAREBRACKET && $this->mCur->value === ']' ) ) { |
| 219 | throw new UserVisibleException( 'expectednotfound', $this->mPos, |
| 220 | [ ']', $this->mCur->type, $this->mCur->value ] ); |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | $this->move(); |
| 225 | if ( $this->mCur->type === AFPToken::TOP && $this->mCur->value === ':=' ) { |
| 226 | $position = $this->mPos; |
| 227 | $this->move(); |
| 228 | $value = $this->doLevelSet(); |
| 229 | if ( $index === 'append' ) { |
| 230 | return new AFPTreeNode( |
| 231 | AFPTreeNode::ARRAY_APPEND, [ $varname, $value ], $position ); |
| 232 | } else { |
| 233 | return new AFPTreeNode( |
| 234 | AFPTreeNode::INDEX_ASSIGNMENT, |
| 235 | [ $varname, $index, $value ], |
| 236 | $position |
| 237 | ); |
| 238 | } |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | // If we reached this point, we did not find an assignment. Roll back |
| 243 | // and assume this was just a literal. |
| 244 | $this->setState( $initialState ); |
| 245 | } |
| 246 | |
| 247 | return $this->doLevelConditions(); |
| 248 | } |
| 249 | |
| 250 | /** |
| 251 | * Handles ternary operator and if-then-else-end. |
| 252 | * |
| 253 | * @return AFPTreeNode |
| 254 | * @throws UserVisibleException |
| 255 | */ |
| 256 | private function doLevelConditions() { |
| 257 | if ( $this->mCur->type === AFPToken::TKEYWORD && $this->mCur->value === 'if' ) { |
| 258 | $position = $this->mPos; |
| 259 | $this->move(); |
| 260 | $condition = $this->doLevelBoolOps(); |
| 261 | |
| 262 | if ( !( $this->mCur->type === AFPToken::TKEYWORD && $this->mCur->value === 'then' ) ) { |
| 263 | throw new UserVisibleException( 'expectednotfound', |
| 264 | $this->mPos, |
| 265 | [ |
| 266 | 'then', |
| 267 | $this->mCur->type, |
| 268 | $this->mCur->value |
| 269 | ] |
| 270 | ); |
| 271 | } |
| 272 | $this->move(); |
| 273 | |
| 274 | $valueIfTrue = $this->doLevelConditions(); |
| 275 | |
| 276 | if ( $this->mCur->type === AFPToken::TKEYWORD && $this->mCur->value === 'else' ) { |
| 277 | $this->move(); |
| 278 | $valueIfFalse = $this->doLevelConditions(); |
| 279 | } else { |
| 280 | $valueIfFalse = null; |
| 281 | } |
| 282 | |
| 283 | if ( !( $this->mCur->type === AFPToken::TKEYWORD && $this->mCur->value === 'end' ) ) { |
| 284 | throw new UserVisibleException( 'expectednotfound', |
| 285 | $this->mPos, |
| 286 | [ |
| 287 | 'end', |
| 288 | $this->mCur->type, |
| 289 | $this->mCur->value |
| 290 | ] |
| 291 | ); |
| 292 | } |
| 293 | $this->move(); |
| 294 | |
| 295 | return new AFPTreeNode( |
| 296 | AFPTreeNode::CONDITIONAL, |
| 297 | [ $condition, $valueIfTrue, $valueIfFalse ], |
| 298 | $position |
| 299 | ); |
| 300 | } |
| 301 | |
| 302 | $condition = $this->doLevelBoolOps(); |
| 303 | if ( $this->mCur->type === AFPToken::TOP && $this->mCur->value === '?' ) { |
| 304 | $position = $this->mPos; |
| 305 | $this->move(); |
| 306 | |
| 307 | $valueIfTrue = $this->doLevelConditions(); |
| 308 | if ( !( $this->mCur->type === AFPToken::TOP && $this->mCur->value === ':' ) ) { |
| 309 | throw new UserVisibleException( 'expectednotfound', |
| 310 | $this->mPos, |
| 311 | [ |
| 312 | ':', |
| 313 | $this->mCur->type, |
| 314 | $this->mCur->value |
| 315 | ] |
| 316 | ); |
| 317 | } |
| 318 | $this->move(); |
| 319 | |
| 320 | $valueIfFalse = $this->doLevelConditions(); |
| 321 | return new AFPTreeNode( |
| 322 | AFPTreeNode::CONDITIONAL, |
| 323 | [ $condition, $valueIfTrue, $valueIfFalse ], |
| 324 | $position |
| 325 | ); |
| 326 | } |
| 327 | |
| 328 | return $condition; |
| 329 | } |
| 330 | |
| 331 | /** |
| 332 | * Handles logic operators. |
| 333 | * |
| 334 | * @return AFPTreeNode |
| 335 | */ |
| 336 | private function doLevelBoolOps() { |
| 337 | $leftOperand = $this->doLevelCompares(); |
| 338 | $ops = [ '&', '|', '^' ]; |
| 339 | while ( $this->mCur->type === AFPToken::TOP && in_array( $this->mCur->value, $ops ) ) { |
| 340 | $op = $this->mCur->value; |
| 341 | $position = $this->mPos; |
| 342 | $this->move(); |
| 343 | |
| 344 | $rightOperand = $this->doLevelCompares(); |
| 345 | |
| 346 | $leftOperand = new AFPTreeNode( |
| 347 | AFPTreeNode::LOGIC, |
| 348 | [ $op, $leftOperand, $rightOperand ], |
| 349 | $position |
| 350 | ); |
| 351 | } |
| 352 | return $leftOperand; |
| 353 | } |
| 354 | |
| 355 | /** |
| 356 | * Handles comparison operators. |
| 357 | * |
| 358 | * @return AFPTreeNode |
| 359 | */ |
| 360 | private function doLevelCompares() { |
| 361 | $leftOperand = $this->doLevelSumRels(); |
| 362 | $equalityOps = [ '==', '===', '!=', '!==', '=' ]; |
| 363 | $orderOps = [ '<', '>', '<=', '>=' ]; |
| 364 | // Only allow either a single operation, or a combination of a single equalityOps and a single |
| 365 | // orderOps. This resembles what PHP does, and allows `a < b == c` while rejecting `a < b < c` |
| 366 | $allowedOps = array_merge( $equalityOps, $orderOps ); |
| 367 | while ( $this->mCur->type === AFPToken::TOP && in_array( $this->mCur->value, $allowedOps ) ) { |
| 368 | $op = $this->mCur->value; |
| 369 | $allowedOps = in_array( $op, $equalityOps ) ? |