Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
61.11% |
297 / 486 |
|
15.38% |
4 / 26 |
CRAP | |
0.00% |
0 / 1 |
| UstringLibrary | |
61.11% |
297 / 486 |
|
15.38% |
4 / 26 |
1997.81 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| register | |
39.39% |
13 / 33 |
|
0.00% |
0 / 1 |
2.89 | |||
| checkString | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
5.03 | |||
| ustringIsUtf8 | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| ustringByteoffset | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
90 | |||
| ustringCodepoint | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
20 | |||
| ustringGcodepointInit | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| ustringToNFC | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| ustringToNFD | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| ustringToNFKC | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| ustringToNFKD | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| ustringChar | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
56 | |||
| ustringLen | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| ustringSub | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
20 | |||
| ustringUpper | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| ustringLower | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| checkPattern | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
4.02 | |||
| patternToRegex | |
70.90% |
95 / 134 |
|
0.00% |
0 / 1 |
79.45 | |||
| bracketedCharSetToRegex | |
96.43% |
27 / 28 |
|
0.00% |
0 / 1 |
16 | |||
| addCapturesFromMatch | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| ustringFind | |
96.15% |
25 / 26 |
|
0.00% |
0 / 1 |
8 | |||
| ustringMatch | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
5.01 | |||
| ustringGmatchInit | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| ustringGmatchCallback | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| ustringGsub | |
79.38% |
77 / 97 |
|
0.00% |
0 / 1 |
52.33 | |||
| handlePCREError | |
37.04% |
10 / 27 |
|
0.00% |
0 / 1 |
34.96 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\Scribunto\Engines\LuaCommon; |
| 4 | |
| 5 | use LogicException; |
| 6 | use MapCacheLRU; |
| 7 | use MediaWiki\MainConfigNames; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use UtfNormal\Validator; |
| 10 | |
| 11 | class UstringLibrary extends LibraryBase { |
| 12 | /** |
| 13 | * Limit on pattern lengths, in bytes not characters |
| 14 | * @var int |
| 15 | */ |
| 16 | private $patternLengthLimit = 10000; |
| 17 | |
| 18 | /** |
| 19 | * Limit on string lengths, in bytes not characters |
| 20 | * @var int |
| 21 | */ |
| 22 | private $stringLengthLimit; |
| 23 | |
| 24 | /** |
| 25 | * PHP until 5.6.9 are buggy when the regex in preg_replace an |
| 26 | * preg_match_all matches the empty string. |
| 27 | * @var bool |
| 28 | */ |
| 29 | private $phpBug53823; |
| 30 | |
| 31 | /** |
| 32 | * A cache of patterns and the regexes they generate. |
| 33 | * @var MapCacheLRU |
| 34 | */ |
| 35 | private $patternRegexCache; |
| 36 | |
| 37 | /** @inheritDoc */ |
| 38 | public function __construct( $engine ) { |
| 39 | $maxArticleSize = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::MaxArticleSize ); |
| 40 | $this->stringLengthLimit = $maxArticleSize * 1024; |
| 41 | $this->phpBug53823 = preg_replace( '//us', 'x', "\xc3\xa1" ) === "x\xc3x\xa1x"; |
| 42 | $this->patternRegexCache = new MapCacheLRU( 100 ); |
| 43 | |
| 44 | parent::__construct( $engine ); |
| 45 | } |
| 46 | |
| 47 | public function register() { |
| 48 | $perf = $this->getEngine()->getPerformanceCharacteristics(); |
| 49 | |
| 50 | if ( $perf['phpCallsRequireSerialization'] ) { |
| 51 | $lib = [ |
| 52 | // Pattern matching is still much faster in PHP, even with the |
| 53 | // overhead of serialization |
| 54 | 'find' => [ $this, 'ustringFind' ], |
| 55 | 'match' => [ $this, 'ustringMatch' ], |
| 56 | 'gmatch_init' => [ $this, 'ustringGmatchInit' ], |
| 57 | 'gmatch_callback' => [ $this, 'ustringGmatchCallback' ], |
| 58 | 'gsub' => [ $this, 'ustringGsub' ], |
| 59 | ]; |
| 60 | } else { |
| 61 | $lib = [ |
| 62 | 'isutf8' => [ $this, 'ustringIsUtf8' ], |
| 63 | 'byteoffset' => [ $this, 'ustringByteoffset' ], |
| 64 | 'codepoint' => [ $this, 'ustringCodepoint' ], |
| 65 | 'gcodepoint_init' => [ $this, 'ustringGcodepointInit' ], |
| 66 | 'toNFC' => [ $this, 'ustringToNFC' ], |
| 67 | 'toNFD' => [ $this, 'ustringToNFD' ], |
| 68 | 'toNFKC' => [ $this, 'ustringToNFKC' ], |
| 69 | 'toNFKD' => [ $this, 'ustringToNFKD' ], |
| 70 | 'char' => [ $this, 'ustringChar' ], |
| 71 | 'len' => [ $this, 'ustringLen' ], |
| 72 | 'sub' => [ $this, 'ustringSub' ], |
| 73 | 'upper' => [ $this, 'ustringUpper' ], |
| 74 | 'lower' => [ $this, 'ustringLower' ], |
| 75 | 'find' => [ $this, 'ustringFind' ], |
| 76 | 'match' => [ $this, 'ustringMatch' ], |
| 77 | 'gmatch_init' => [ $this, 'ustringGmatchInit' ], |
| 78 | 'gmatch_callback' => [ $this, 'ustringGmatchCallback' ], |
| 79 | 'gsub' => [ $this, 'ustringGsub' ], |
| 80 | ]; |
| 81 | } |
| 82 | return $this->getEngine()->registerInterface( 'mw.ustring.lua', $lib, [ |
| 83 | 'stringLengthLimit' => $this->stringLengthLimit, |
| 84 | 'patternLengthLimit' => $this->patternLengthLimit, |
| 85 | ] ); |
| 86 | } |
| 87 | |
| 88 | /** |
| 89 | * Check a string first parameter |
| 90 | * @param string $name Function name, for errors |
| 91 | * @param mixed &$s Value to check |
| 92 | * @param bool $checkEncoding Whether to validate UTF-8 encoding. |
| 93 | */ |
| 94 | private function checkString( $name, &$s, $checkEncoding = true ) { |
| 95 | if ( $this->getLuaType( $s ) == 'number' ) { |
| 96 | $s = (string)$s; |
| 97 | } else { |
| 98 | $this->checkType( $name, 1, $s, 'string' ); |
| 99 | if ( $checkEncoding && !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 100 | throw new LuaError( "bad argument #1 to '$name' (string is not UTF-8)" ); |
| 101 | } |
| 102 | if ( strlen( $s ) > $this->stringLengthLimit ) { |
| 103 | throw new LuaError( |
| 104 | "bad argument #1 to '$name' (string is longer than $this->stringLengthLimit bytes)" |
| 105 | ); |
| 106 | } |
| 107 | } |
| 108 | } |
| 109 | |
| 110 | /** |
| 111 | * Handler for isUtf8 |
| 112 | * @internal |
| 113 | * @param string $s |
| 114 | * @return bool[] |
| 115 | */ |
| 116 | public function ustringIsUtf8( $s ) { |
| 117 | $this->checkString( 'isutf8', $s, false ); |
| 118 | return [ mb_check_encoding( $s, 'UTF-8' ) ]; |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * Handler for byteoffset |
| 123 | * @internal |
| 124 | * @param string $s |
| 125 | * @param int $l |
| 126 | * @param int $i |
| 127 | * @return int[]|null[] |
| 128 | */ |
| 129 | public function ustringByteoffset( $s, $l = 1, $i = 1 ) { |
| 130 | $this->checkString( 'byteoffset', $s ); |
| 131 | $this->checkTypeOptional( 'byteoffset', 2, $l, 'number', 1 ); |
| 132 | $this->checkTypeOptional( 'byteoffset', 3, $i, 'number', 1 ); |
| 133 | |
| 134 | $bytelen = strlen( $s ); |
| 135 | if ( $i < 0 ) { |
| 136 | $i = $bytelen + $i + 1; |
| 137 | } |
| 138 | if ( $i < 1 || $i > $bytelen ) { |
| 139 | return [ null ]; |
| 140 | } |
| 141 | $i--; |
| 142 | $j = $i; |
| 143 | while ( ( ord( $s[$i] ) & 0xc0 ) === 0x80 ) { |
| 144 | $i--; |
| 145 | } |
| 146 | if ( $l > 0 && $j === $i ) { |
| 147 | $l--; |
| 148 | } |
| 149 | $char = mb_strlen( substr( $s, 0, $i ), 'UTF-8' ) + $l; |
| 150 | if ( $char < 0 || $char >= mb_strlen( $s, 'UTF-8' ) ) { |
| 151 | return [ null ]; |
| 152 | } else { |
| 153 | return [ strlen( mb_substr( $s, 0, $char, 'UTF-8' ) ) + 1 ]; |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | /** |
| 158 | * Handler for codepoint |
| 159 | * @internal |
| 160 | * @param string $s |
| 161 | * @param int $i |
| 162 | * @param int|null $j |
| 163 | * @return int[] |
| 164 | */ |
| 165 | public function ustringCodepoint( $s, $i = 1, $j = null ) { |
| 166 | $this->checkString( 'codepoint', $s ); |
| 167 | $this->checkTypeOptional( 'codepoint', 2, $i, 'number', 1 ); |
| 168 | $this->checkTypeOptional( 'codepoint', 3, $j, 'number', $i ); |
| 169 | |
| 170 | $l = mb_strlen( $s, 'UTF-8' ); |
| 171 | if ( $i < 0 ) { |
| 172 | $i = $l + $i + 1; |
| 173 | } |
| 174 | if ( $j < 0 ) { |
| 175 | $j = $l + $j + 1; |
| 176 | } |
| 177 | if ( $j < $i ) { |
| 178 | return []; |
| 179 | } |
| 180 | $i = max( 1, min( $i, $l + 1 ) ); |
| 181 | $j = max( 1, min( $j, $l + 1 ) ); |
| 182 | $s = mb_substr( $s, $i - 1, $j - $i + 1, 'UTF-8' ); |
| 183 | return unpack( 'N*', mb_convert_encoding( $s, 'UTF-32BE', 'UTF-8' ) ); |
| 184 | } |
| 185 | |
| 186 | /** |
| 187 | * Handler for gcodepointInit |
| 188 | * @internal |
| 189 | * @param string $s |
| 190 | * @param int $i |
| 191 | * @param int|null $j |
| 192 | * @return int[][] |
| 193 | */ |
| 194 | public function ustringGcodepointInit( $s, $i = 1, $j = null ) { |
| 195 | return [ $this->ustringCodepoint( $s, $i, $j ) ]; |
| 196 | } |
| 197 | |
| 198 | /** |
| 199 | * Handler for toNFC |
| 200 | * @internal |
| 201 | * @param string $s |
| 202 | * @return string[]|null[] |
| 203 | */ |
| 204 | public function ustringToNFC( $s ) { |
| 205 | $this->checkString( 'toNFC', $s, false ); |
| 206 | if ( !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 207 | return [ null ]; |
| 208 | } |
| 209 | return [ Validator::toNFC( $s ) ]; |
| 210 | } |
| 211 | |
| 212 | /** |
| 213 | * Handler for toNFD |
| 214 | * @internal |
| 215 | * @param string $s |
| 216 | * @return string[]|null[] |
| 217 | */ |
| 218 | public function ustringToNFD( $s ) { |
| 219 | $this->checkString( 'toNFD', $s, false ); |
| 220 | if ( !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 221 | return [ null ]; |
| 222 | } |
| 223 | return [ Validator::toNFD( $s ) ]; |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * Handler for toNFKC |
| 228 | * @internal |
| 229 | * @param string $s |
| 230 | * @return string[]|null[] |
| 231 | */ |
| 232 | public function ustringToNFKC( $s ) { |
| 233 | $this->checkString( 'toNFKC', $s, false ); |
| 234 | if ( !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 235 | return [ null ]; |
| 236 | } |
| 237 | return [ Validator::toNFKC( $s ) ]; |
| 238 | } |
| 239 | |
| 240 | /** |
| 241 | * Handler for toNFKD |
| 242 | * @internal |
| 243 | * @param string $s |
| 244 | * @return string[]|null[] |
| 245 | */ |
| 246 | public function ustringToNFKD( $s ) { |
| 247 | $this->checkString( 'toNFKD', $s, false ); |
| 248 | if ( !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 249 | return [ null ]; |
| 250 | } |
| 251 | return [ Validator::toNFKD( $s ) ]; |
| 252 | } |
| 253 | |
| 254 | /** |
| 255 | * Handler for char |
| 256 | * @internal |
| 257 | * @return string[] |
| 258 | */ |
| 259 | public function ustringChar() { |
| 260 | $args = func_get_args(); |
| 261 | if ( count( $args ) > $this->stringLengthLimit ) { |
| 262 | throw new LuaError( "too many arguments to 'char'" ); |
| 263 | } |
| 264 | foreach ( $args as $k => &$v ) { |
| 265 | if ( !is_numeric( $v ) ) { |
| 266 | $this->checkType( 'char', $k + 1, $v, 'number' ); |
| 267 | } |
| 268 | $v = (int)floor( $v ); |
| 269 | if ( $v < 0 || $v > 0x10ffff ) { |
| 270 | $k++; |
| 271 | throw new LuaError( "bad argument #$k to 'char' (value out of range)" ); |
| 272 | } |
| 273 | } |
| 274 | $s = pack( 'N*', ...$args ); |
| 275 | $s = mb_convert_encoding( $s, 'UTF-8', 'UTF-32BE' ); |
| 276 | if ( strlen( $s ) > $this->stringLengthLimit ) { |
| 277 | throw new LuaError( "result to long for 'char'" ); |
| 278 | } |
| 279 | return [ $s ]; |
| 280 | } |
| 281 | |
| 282 | /** |
| 283 | * Handler for len |
| 284 | * @internal |
| 285 | * @param string $s |
| 286 | * @return int[]|null[] |
| 287 | */ |
| 288 | public function ustringLen( $s ) { |
| 289 | $this->checkString( 'len', $s, false ); |
| 290 | if ( !mb_check_encoding( $s, 'UTF-8' ) ) { |
| 291 | return [ null ]; |
| 292 | } |
| 293 | return [ mb_strlen( $s, 'UTF-8' ) ]; |
| 294 | } |
| 295 | |
| 296 | /** |
| 297 | * Handler for sub |
| 298 | * @internal |
| 299 | * @param string $s |
| 300 | * @param int $i |
| 301 | * @param int $j |
| 302 | * @return string[] |
| 303 | */ |
| 304 | public function ustringSub( $s, $i = 1, $j = -1 ) { |
| 305 | $this->checkString( 'sub', $s ); |
| 306 | $this->checkTypeOptional( 'sub', 2, $i, 'number', 1 ); |
| 307 | $this->checkTypeOptional( 'sub', 3, $j, 'number', -1 ); |
| 308 | |
| 309 | $len = mb_strlen( $s, 'UTF-8' ); |
| 310 | if ( $i < 0 ) { |
| 311 | $i = $len + $i + 1; |
| 312 | } |
| 313 | if ( $j < 0 ) { |
| 314 | $j = $len + $j + 1; |
| 315 | } |
| 316 | if ( $j < $i ) { |
| 317 | return [ '' ]; |
| 318 | } |
| 319 | $i = max( 1, min( (int)$i, $len + 1 ) ); |
| 320 | $j = max( 1, min( (int)$j, $len + 1 ) ); |
| 321 | $s = mb_substr( $s, $i - 1, $j - $i + 1, 'UTF-8' ); |
| 322 | return [ $s ]; |
| 323 | } |
| 324 | |
| 325 | /** |
| 326 | * Handler for upper |
| 327 | * @internal |
| 328 | * @param string $s |
| 329 | * @return string[] |
| 330 | */ |
| 331 | public function ustringUpper( $s ) { |
| 332 | $this->checkString( 'upper', $s ); |
| 333 | return [ mb_strtoupper( $s, 'UTF-8' ) ]; |
| 334 | } |
| 335 | |
| 336 | /** |
| 337 | * Handler for lower |
| 338 | * @internal |
| 339 | * @param string $s |
| 340 | * @return string[] |
| 341 | */ |
| 342 | public function ustringLower( $s ) { |
| 343 | $this->checkString( 'lower', $s ); |
| 344 | return [ mb_strtolower( $s, 'UTF-8' ) ]; |
| 345 | } |
| 346 | |
| 347 | /** |
| 348 | * Check a pattern as the second argument |
| 349 | * @param string $name Lua function name, for errors |
| 350 | * @param mixed $pattern Lua pattern |
| 351 | */ |
| 352 | private function checkPattern( $name, $pattern ) { |
| 353 | if ( $this->getLuaType( $pattern ) == 'number' ) { |
| 354 | $pattern = (string)$pattern; |
| 355 | } |
| 356 | $this->checkType( $name, 2, $pattern, 'string' ); |
| 357 | if ( !mb_check_encoding( $pattern, 'UTF-8' ) ) { |
| 358 | throw new LuaError( "bad argument #2 to '$name' (string is not UTF-8)" ); |
| 359 | } |
| 360 | if ( strlen( $pattern ) > $this->patternLengthLimit ) { |
| 361 | throw new LuaError( |
| 362 | "bad argument #2 to '$name' (pattern is longer than $this->patternLengthLimit bytes)" |
| 363 | ); |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | /** |
| 368 | * Convert a Lua pattern into a PCRE regex |
| 369 | * @param string $pattern Lua pattern to convert |
| 370 | * @param string|false $anchor Regex fragment (`^` or `\G`) to use |
| 371 | * when anchoring the start of the regex, or false to disable start-anchoring. |
| 372 | * @param string $name Lua function name, for errors |
| 373 | * @return array [ string $re, array $capt, bool $anypos ] |
| 374 | * - $re: The regular expression |
| 375 | * - $capt: Definition of capturing groups, see addCapturesFromMatch() |
| 376 | * - $anypos: Whether any positional captures were encountered in the pattern. |
| 377 | * @return-taint none |
| 378 | */ |
| 379 | private function patternToRegex( $pattern, $anchor, $name ) { |
| 380 | $cacheKey = serialize( [ $pattern, $anchor ] ); |
| 381 | if ( !$this->patternRegexCache->has( $cacheKey ) ) { |
| 382 | $this->checkPattern( $name, $pattern ); |
| 383 | $pat = preg_split( '//us', $pattern, -1, PREG_SPLIT_NO_EMPTY ); |
| 384 | |
| 385 | static $charsets = null, $brcharsets = null; |
| 386 | if ( $charsets === null ) { |
| 387 | $charsets = [ |
| 388 | // If you change these, also change lualib/ustring/make-tables.php |
| 389 | // (and run it to regenerate charsets.lua) |
| 390 | 'a' => '\p{L}', |
| 391 | 'c' => '\p{Cc}', |
| 392 | 'd' => '\p{Nd}', |
| 393 | 'l' => '\p{Ll}', |
| 394 | 'p' => '\p{P}', |
| 395 | 's' => '\p{Xps}', |
| 396 | 'u' => '\p{Lu}', |
| 397 | 'w' => '[\p{L}\p{Nd}]', |
| 398 | 'x' => '[0-9A-Fa-f０-９Ａ-Ｆａ-ｆ]', |
| 399 | 'z' => '\0', |
| 400 | |
| 401 | // These *must* be the inverse of the above |
| 402 | 'A' => '\P{L}', |
| 403 | 'C' => '\P{Cc}', |
| 404 | 'D' => '\P{Nd}', |
| 405 | 'L' => '\P{Ll}', |
| 406 | 'P' => '\P{P}', |
| 407 | 'S' => '\P{Xps}', |
| 408 | 'U' => '\P{Lu}', |
| 409 | 'W' => '[^\p{L}\p{Nd}]', |
| 410 | 'X' => '[^0-9A-Fa-f０-９Ａ-Ｆａ-ｆ]', |
| 411 | 'Z' => '[^\0]', |
| 412 | ]; |
| 413 | $brcharsets = [ |
| 414 | 'w' => '\p{L}\p{Nd}', |
| 415 | 'x' => '0-9A-Fa-f０-９Ａ-Ｆａ-ｆ', |
| 416 | |
| 417 | // Negated sets that are not expressable as a simple \P{} are |
| 418 | // unfortunately complicated. |
| 419 | |
| 420 | // Xan is L plus N, so ^Xan plus Nl plus No is anything that's not L or Nd |
| 421 | 'W' => '\P{Xan}\p{Nl}\p{No}', |
| 422 | |
| 423 | // Manually constructed. Fun. |
| 424 | 'X' => '\x00-\x2f\x3a-\x40\x47-\x60\x67-\x{ff0f}' |
| 425 | . '\x{ff1a}-\x{ff20}\x{ff27}-\x{ff40}\x{ff47}-\x{10ffff}', |
| 426 | |
| 427 | // Ha! |
| 428 | 'Z' => '\x01-\x{10ffff}', |
| 429 | ] + $charsets; |
| 430 | } |
| 431 | |
| 432 | $re = '/'; |
| 433 | $len = count( $pat ); |
| 434 | $capt = []; |
| 435 | $anypos = false; |
| 436 | $captparen = []; |
| 437 | $opencapt = []; |
| 438 | $bct = 0; |
| 439 | |
| 440 | for ( $i = 0; $i < $len; $i++ ) { |
| 441 | $ii = $i + 1; |
| 442 | $q = false; |
| 443 | switch ( $pat[$i] ) { |
| 444 | case '^': |
| 445 | $q = $i; |
| 446 | $re .= ( $anchor === false || $q ) ? '\\^' : $anchor; |
| 447 | break; |
| 448 | |
| 449 | case '$': |
| 450 | $q = ( $i < $len - 1 ); |
| 451 | $re .= $q ? '\\$' : '$'; |
| 452 | break; |
| 453 | |
| 454 | case '(': |
| 455 | if ( $i + 1 >= $len ) { |
| 456 | throw new LuaError( "Unmatched open-paren at pattern character $ii" ); |
| 457 | } |
| 458 | $n = count( $capt ) + 1; |
| 459 | $capt[$n] = ( $pat[$i + 1] === ')' ); |
| 460 | if ( $capt[$n] ) { |
| 461 | $anypos = true; |
| 462 | } |
| 463 | $re .= "(?<m$n>"; |
| 464 | $opencapt[] = $n; |
| 465 | $captparen[$n] = $ii; |
| 466 | break; |
| 467 | |
| 468 | case ')': |
| 469 | if ( count( $opencapt ) <= 0 ) { |
| 470 | throw new LuaError( "Unmatched close-paren at pattern character $ii" ); |
| 471 | } |
| 472 | array_pop( $opencapt ); |
| 473 | $re .= $pat[$i]; |
| 474 | break; |
| 475 | |
| 476 | case '%': |
| 477 | $i++; |
| 478 | if ( $i >= $len ) { |
| 479 | throw new LuaError( "malformed pattern (ends with '%')" ); |
| 480 | } |
| 481 | if ( isset( $charsets[$pat[$i]] ) ) { |
| 482 | $re .= $charsets[$pat[$i]]; |
| 483 | $q = true; |
| 484 | } elseif ( $pat[$i] === 'b' ) { |
| 485 | if ( $i + 2 >= $len ) { |
| 486 | throw new LuaError( "malformed pattern (missing arguments to \'%b\')" ); |
| 487 | } |
| 488 | $d1 = preg_quote( $pat[++$i], '/' ); |
| 489 | $d2 = preg_quote( $pat[++$i], '/' ); |
| 490 | if ( $d1 === $d2 ) { |
| 491 | $re .= "{$d1}[^$d1]*$d1"; |
| 492 | } else { |
| 493 | $bct++; |
| 494 | $re .= "(?<b$bct>$d1(?:(?>[^$d1$d2]+)|(?P>b$bct))*$d2)"; |
| 495 | } |
| 496 | } elseif ( $pat[$i] === 'f' ) { |
| 497 | if ( $i + 1 >= $len || $pat[++$i] !== '[' ) { |
| 498 | throw new LuaError( "missing '[' after %f in pattern at pattern character $ii" ); |
| 499 | } |
| 500 | [ $i, $re2 ] = $this->bracketedCharSetToRegex( $pat, $i, $len, $brcharsets ); |
| 501 | // Because %f considers the beginning and end of the string |
| 502 | // to be \0, determine if $re2 matches that and take it |
| 503 | // into account with "^" and "$". |
| 504 | // @phan-suppress-next-line PhanParamSuspiciousOrder |
| 505 | if ( preg_match( "/$re2/us", "\0" ) ) { |
| 506 | $re .= "(?<!^)(?<!$re2)(?=$re2|$)"; |
| 507 | } else { |
| 508 | $re .= "(?<!$re2)(?=$re2)"; |
| 509 | } |
| 510 | } elseif ( $pat[$i] >= '0' && $pat[$i] <= '9' ) { |
| 511 | $n = ord( $pat[$i] ) - 0x30; |
| 512 | if ( $n === 0 || $n > count( $capt ) || in_array( $n, $opencapt ) ) { |
| 513 | throw new LuaError( "invalid capture index %$n at pattern character $ii" ); |
| 514 | } |
| 515 | $re .= "\\g{m$n}"; |
| 516 | } else { |
| 517 | $re .= preg_quote( $pat[$i], '/' ); |
| 518 | $q = true; |
| 519 | } |
| 520 | break; |
| 521 | |
| 522 | case '[': |
| 523 | [ $i, $re2 ] = $this->bracketedCharSetToRegex( $pat, $i, $len, $brcharsets ); |
| 524 | $re .= $re2; |
| 525 | $q = true; |
| 526 | break; |
| 527 | |
| 528 | case ']': |
| 529 | throw new LuaError( "Unmatched close-bracket at pattern character $ii" ); |
| 530 | |
| 531 | case '.': |
| 532 | $re .= $pat[$i]; |
| 533 | $q = true; |
| 534 | break; |
| 535 | |
| 536 | default: |
| 537 | $re .= preg_quote( $pat[$i], '/' ); |
| 538 | $q = true; |
| 539 | break; |
| 540 | } |
| 541 | if ( $q && $i + 1 < $len ) { |
| 542 | switch ( $pat[$i + 1] ) { |
| 543 | case '*': |
| 544 | case '+': |
| 545 | case '?': |
| 546 | $re .= $pat[++$i]; |
| 547 | break; |
| 548 | case '-': |
| 549 | $re .= '*?'; |
| 550 | $i++; |
| 551 | break; |
| 552 | } |
| 553 | } |
| 554 | } |
| 555 | if ( count( $opencapt ) ) { |
| 556 | $ii = $captparen[$opencapt[0]]; |
| 557 | throw new LuaError( "Unclosed capture beginning at pattern character $ii" ); |
| 558 | } |
| 559 | $re .= '/us'; |
| 560 | |
| 561 | $this->patternRegexCache->set( $cacheKey, [ $re, $capt, $anypos ] ); |
| 562 | } |
| 563 | return $this->patternRegexCache->get( $cacheKey ); |
| 564 | } |
| 565 | |
| 566 | /** |
| 567 | * Convert a Lua pattern bracketed character set to a PCRE regex fragment |
| 568 | * @param string[] $pat Pattern being processed, split into individual characters. |
| 569 | * @param int $i Offset of the start of the bracketed character set in $pat. |
| 570 | * @param int $len Length of $pat. |
| 571 | * @param array $brcharsets Mapping from Lua pattern percent escapes to |
| 572 | * regex-style character ranges. |
| 573 | * @return array [ int $new_i, string $re_fragment ] |
| 574 | */ |
| 575 | private function bracketedCharSetToRegex( $pat, $i, $len, $brcharsets ) { |
| 576 | $ii = $i + 1; |
| 577 | $re = '['; |
| 578 | $i++; |
| 579 | if ( $i < $len && $pat[$i] === '^' ) { |
| 580 | $re .= '^'; |
| 581 | $i++; |
| 582 | } |
| 583 | for ( $j = $i; $i < $len && ( $j == $i || $pat[$i] !== ']' ); $i++ ) { |
| 584 | if ( $pat[$i] === '%' ) { |
| 585 | $i++; |
| 586 | if ( $i >= $len ) { |
| 587 | break; |
| 588 | } |
| 589 | $re .= $brcharsets[$pat[$i]] ?? preg_quote( $pat[$i], '/' ); |
| 590 | } elseif ( $i + 2 < $len && |
| 591 | $pat[$i + 1] === '-' && $pat[$i + 2] !== ']' && $pat[$i + 2] !== '%' |
| 592 | ) { |
| 593 | if ( $pat[$i] <= $pat[$i + 2] ) { |
| 594 | $re .= preg_quote( $pat[$i], '/' ) . '-' . preg_quote( $pat[$i + 2], '/' ); |
| 595 | } |
| 596 | $i += 2; |
| 597 | } else { |
| 598 | $re .= preg_quote( $pat[$i], '/' ); |
| 599 | } |
| 600 | } |
| 601 | if ( $i >= $len ) { |
| 602 | throw new LuaError( |
| 603 | "Missing close-bracket for character set beginning at pattern character $ii" |
| 604 | ); |
| 605 | } |
| 606 | $re .= ']'; |
| 607 | |
| 608 | // Lua just ignores invalid ranges, while pcre throws an error. |
| 609 | // We filter them out above, but then we need to special-case empty sets |
| 610 | if ( $re === '[]' ) { |
| 611 | // Can't directly quantify (*FAIL), so wrap it. |
| 612 | // "(?!)" would be simpler and could be quantified if not for a bug in PCRE 8.13 to 8.33 |
| 613 | $re = '(?:(*FAIL))'; |
| 614 | } elseif ( $re === '[^]' ) { |
| 615 | // 's' modifier is always used, so this works |
| 616 | $re = '.'; |
| 617 | } |
| 618 | |
| 619 | return [ $i, $re ]; |
| 620 | } |
| 621 | |
| 622 | /** |
| 623 | * Append captured groups to a result array |
| 624 | * @param array $arr Result array to append to. |
| 625 | * @param string $s String matched against. |
| 626 | * @param array $m Matches, from preg_match with PREG_OFFSET_CAPTURE. |
| 627 | * @param array $capt Capture groups (in $m) to process, see patternToRegex() |
| 628 | * @param bool $m0_if_no_captures Whether to append "$0" if $capt is empty. |
| 629 | * @return array |
| 630 | */ |
| 631 | private function addCapturesFromMatch( $arr, $s, $m, $capt, $m0_if_no_captures ) { |
| 632 | if ( count( $capt ) ) { |
| 633 | foreach ( $capt as $n => $pos ) { |
| 634 | if ( $pos ) { |
| 635 | $o = mb_strlen( substr( $s, 0, $m["m$n"][1] ), 'UTF-8' ) + 1; |
| 636 | $arr[] = $o; |
| 637 | } else { |
| 638 | $arr[] = $m["m$n"][0]; |
| 639 | } |
| 640 | } |
| 641 | } elseif ( $m0_if_no_captures ) { |
| 642 | $arr[] = $m[0][0]; |
| 643 | } |
| 644 | return $arr; |
| 645 | } |
| 646 | |
| 647 | /** |
| 648 | * Handler for find |
| 649 | * @internal |
| 650 | * @param string $s |
| 651 | * @param string $pattern |
| 652 | * @param int $init |
| 653 | * @param bool $plain |
| 654 | * @return array Format is [ null ], or [ int, int ], or [ int, int, (string|int)... ] |
| 655 | */ |
| 656 | public function ustringFind( $s, $pattern, $init = 1, $plain = false ) { |
| 657 | $this->checkString( 'find', $s ); |
| 658 | $this->checkTypeOptional( 'find', 3, $init, 'number', 1 ); |
| 659 | $this->checkTypeOptional( 'find', 4, $plain, 'boolean', false ); |
| 660 | |
| 661 | $len = mb_strlen( $s, 'UTF-8' ); |
| 662 | if ( $init < 0 ) { |
| 663 | $init = $len + $init + 1; |
| 664 | } elseif ( $init > $len + 1 ) { |
| 665 | $init = $len + 1; |
| 666 | } |
| 667 | |
| 668 | if ( $init > 1 ) { |
| 669 | $offset = strlen( mb_substr( $s, 0, $init - 1, 'UTF-8' ) ); |
| 670 | } else { |
| 671 | $init = 1; |
| 672 | $offset = 0; |
| 673 | } |
| 674 | |
| 675 | if ( $plain ) { |
| 676 | $this->checkPattern( 'find', $pattern ); |
| 677 | if ( $pattern !== '' ) { |
| 678 | $ret = mb_strpos( $s, $pattern, $init - 1, 'UTF-8' ); |
| 679 | } else { |
| 680 | $ret = $init - 1; |
| 681 | } |
| 682 | if ( $ret === false ) { |
| 683 | return [ null ]; |
| 684 | } else { |
| 685 | return [ $ret + 1, $ret + mb_strlen( $pattern ) ]; |
| 686 | } |
| 687 | } else { |
| 688 | [ $re, $capt ] = $this->patternToRegex( $pattern, '\G', 'find' ); |
| 689 | if ( !preg_match( $re, $s, $m, PREG_OFFSET_CAPTURE, $offset ) ) { |
| 690 | return [ null ]; |
| 691 | } |
| 692 | $o = mb_strlen( substr( $s, 0, $m[0][1] ), 'UTF-8' ); |
| 693 | $ret = [ $o + 1, $o + mb_strlen( $m[0][0], 'UTF-8' ) ]; |
| 694 | return $this->addCapturesFromMatch( $ret, $s, $m, $capt, false ); |
| 695 | } |
| 696 | } |
| 697 | |
| 698 | /** |
| 699 | * Handler for match |
| 700 | * @internal |
| 701 | * @param string $s |
| 702 | * @param string $pattern |
| 703 | * @param int $init |
| 704 | * @return array Format is [ null ] or [ (string|int)... ] |
| 705 | */ |
| 706 | public function ustringMatch( $s, $pattern, $init = 1 ) { |
| 707 | $this->checkString( 'match', $s ); |
| 708 | $this->checkTypeOptional( 'match', 3, $init, 'number', 1 ); |
| 709 | |
| 710 | $len = mb_strlen( $s, 'UTF-8' ); |
| 711 | if ( $init < 0 ) { |
| 712 | $init = $len + $init + 1; |
| 713 | } elseif ( $init > $len + 1 ) { |
| 714 | $init = $len + 1; |
| 715 | } |
| 716 | if ( $init > 1 ) { |
| 717 | $offset = strlen( mb_substr( $s, 0, $init - 1, 'UTF-8' ) ); |
| 718 | } else { |
| 719 | $offset = 0; |
| 720 | } |
| 721 | |
| 722 | [ $re, $capt ] = $this->patternToRegex( $pattern, '\G', 'match' ); |
| 723 | if ( !preg_match( $re, $s, $m, PREG_OFFSET_CAPTURE, $offset ) ) { |
| 724 | return [ null ]; |
| 725 | } |
| 726 | return $this->addCapturesFromMatch( [], $s, $m, $capt, true ); |
| 727 | } |
| 728 | |
| 729 | /** |
| 730 | * Handler for gmatchInit |
| 731 | * @internal |
| 732 | * @param string $s |
| 733 | * @param string $pattern |
| 734 | * @return array Format is [ string, bool[] ] |
| 735 | */ |
| 736 | public function ustringGmatchInit( $s, $pattern ) { |
| 737 | $this->checkString( 'gmatch', $s ); |
| 738 | |
| 739 | [ $re, $capt ] = $this->patternToRegex( $pattern, false, 'gmatch' ); |
| 740 | return [ $re, $capt ]; |
| 741 | } |
| 742 | |
| 743 | /** |
| 744 | * Handler for gmatchCallback |
| 745 | * @internal |
| 746 | * @param string $s |
| 747 | * @param string $re |
| 748 | * @param bool[] $capt |
| 749 | * @param int $pos |
| 750 | * @return array Format is [ int, [ null, (string|int)... ] ] |
| 751 | */ |
| 752 | public function ustringGmatchCallback( $s, $re, $capt, $pos ) { |
| 753 | if ( !preg_match( $re, $s, $m, PREG_OFFSET_CAPTURE, $pos ) ) { |
| 754 | return [ $pos, [] ]; |
| 755 | } |
| 756 | $pos = $m[0][1] + strlen( $m[0][0] ); |
| 757 | return [ $pos, $this->addCapturesFromMatch( [ null ], $s, $m, $capt, true ) ]; |
| 758 | } |
| 759 | |
| 760 | /** |
| 761 | * Handler for gsub |
| 762 | * @internal |
| 763 | * @param string $s |
| 764 | * @param string $pattern |
| 765 | * @param mixed $repl |
| 766 | * @param string|int|null $n |
| 767 | * @return array Format is [ string, int ] |
| 768 | */ |
| 769 | public function ustringGsub( $s, $pattern, $repl, $n = null ) { |
| 770 | $this->checkString( 'gsub', $s ); |
| 771 | $this->checkTypeOptional( 'gsub', 4, $n, 'number', null ); |
| 772 | |
| 773 | if ( $n === null ) { |
| 774 | $n = -1; |
| 775 | } elseif ( $n < 1 ) { |
| 776 | return [ $s, 0 ]; |
| 777 | } |
| 778 | |
| 779 | [ $re, $capt, $anypos ] = $this->patternToRegex( $pattern, '^', 'gsub' ); |
| 780 | $captures = []; |
| 781 | |
| 782 | if ( $this->phpBug53823 ) { |
| 783 | // PHP bug 53823 means that a zero-length match before a UTF-8 |
| 784 | // character will match again before every byte of that character. |
| 785 | // The workaround is to capture the first "character" of/after the |
| 786 | // match and verify that its first byte is legal to start a UTF-8 |
| 787 | // character. |
| 788 | $re = '/(?=(?<phpBug53823>.|$))' . substr( $re, 1 ); |
| 789 | } |
| 790 | |
| 791 | if ( $anypos ) { |
| 792 | // preg_replace_callback doesn't take a "flags" argument, so we |
| 793 | // can't pass PREG_OFFSET_CAPTURE to it, which is needed to handle |
| 794 | // position captures. So instead we have to do a preg_match_all and |
| 795 | // handle the captures ourself. |
| 796 | $ct = preg_match_all( $re, $s, $mm, PREG_OFFSET_CAPTURE | PREG_SET_ORDER ); |
| 797 | for ( $i = 0; $i < $ct; $i++ ) { |
| 798 | $m = $mm[$i]; |
| 799 | if ( $this->phpBug53823 ) { |
| 800 | $c = ord( $m['phpBug53823'][0] ); |
| 801 | if ( $c >= 0x80 && $c <= 0xbf ) { |
| 802 | continue; |
| 803 | } |
| 804 | } |
| 805 | $c = [ $m[0][0] ]; |
| 806 | foreach ( $this->addCapturesFromMatch( [], $s, $m, $capt, false ) as $k => $v ) { |
| 807 | $k++; |
| 808 | $c["m$k"] = $v; |
| 809 | } |
| 810 | $captures[] = $c; |
| 811 | if ( $n >= 0 && count( $captures ) >= $n ) { |
| 812 | break; |
| 813 | } |
| 814 | } |
| 815 | } |
| 816 | |
| 817 | switch ( $this->getLuaType( $repl ) ) { |
| 818 | case 'string': |
| 819 | case 'number': |
| 820 | $cb = static function ( $m ) use ( $repl, $anypos, &$captures ) { |
| 821 | if ( $anypos ) { |
| 822 | $m = array_shift( $captures ); |
| 823 | } |
| 824 | return preg_replace_callback( '/%([%0-9])/', static function ( $m2 ) use ( $m ) { |
| 825 | $x = $m2[1]; |
| 826 | if ( $x === '%' ) { |
| 827 | return '%'; |
| 828 | } elseif ( $x === '0' ) { |
| 829 | return $m[0]; |
| 830 | } elseif ( isset( $m["m$x"] ) ) { |
| 831 | return $m["m$x"]; |
| 832 | } elseif ( $x === '1' ) { |
| 833 | // Match undocumented Lua string.gsub behavior |
| 834 | return $m[0]; |
| 835 | } else { |
| 836 | throw new LuaError( "invalid capture index %$x in replacement string" ); |
| 837 | } |
| 838 | }, $repl ); |
| 839 | }; |
| 840 | break; |
| 841 | |
| 842 | case 'table': |
| 843 | $cb = function ( $m ) use ( $repl, $anypos, &$captures ) { |
| 844 | if ( $anypos ) { |
| 845 | $m = array_shift( $captures ); |
| 846 | } |
| 847 | $x = $m['m1'] ?? $m[0]; |
| 848 | if ( !isset( $repl[$x] ) ) { |
| 849 | return $m[0]; |
| 850 | } |
| 851 | $type = $this->getLuaType( $repl[$x] ); |
| 852 | if ( $type !== 'string' && $type !== 'number' ) { |
| 853 | throw new LuaError( "invalid replacement value (a $type)" ); |
| 854 | } |
| 855 | return $repl[$x]; |
| 856 | }; |
| 857 | break; |
| 858 | |
| 859 | case 'function': |
| 860 | $interpreter = $this->getInterpreter(); |
| 861 | $cb = function ( $m ) use ( $interpreter, $capt, $repl, $anypos, &$captures ) { |
| 862 | if ( $anypos ) { |
| 863 | $m = array_shift( $captures ); |
| 864 | } |
| 865 | $args = []; |
| 866 | if ( count( $capt ) ) { |
| 867 | foreach ( $capt as $i => $pos ) { |
| 868 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 869 | $args[] = $m["m$i"]; |
| 870 | } |
| 871 | } else { |
| 872 | $args[] = $m[0]; |
| 873 | } |
| 874 | $ret = $interpreter->callFunction( $repl, ...$args ); |
| 875 | if ( count( $ret ) === 0 || $ret[0] === null ) { |
| 876 | return $m[0]; |
| 877 | } |
| 878 | $type = $this->getLuaType( $ret[0] ); |
| 879 | if ( $type !== 'string' && $type !== 'number' ) { |
| 880 | throw new LuaError( "invalid replacement value (a $type)" ); |
| 881 | } |
| 882 | return $ret[0]; |
| 883 | }; |
| 884 | break; |
| 885 | |
| 886 | default: |
| 887 | $this->checkType( 'gsub', 3, $repl, 'function or table or string' ); |
| 888 | throw new LogicException( 'checkType above should have failed' ); |
| 889 | } |
| 890 | |
| 891 | $skippedMatches = 0; |
| 892 | if ( $this->phpBug53823 ) { |
| 893 | // Since we're having bogus matches, we need to keep track of the |
| 894 | // necessary adjustment and stop manually once we hit the limit. |
| 895 | $maxMatches = $n < 0 ? INF : $n; |
| 896 | $n = -1; |
| 897 | $realCallback = $cb; |
| 898 | $cb = static function ( $m ) use ( $realCallback, &$skippedMatches, &$maxMatches ) { |
| 899 | $c = ord( $m['phpBug53823'] ); |
| 900 | if ( ( $c >= 0x80 && $c <= 0xbf ) || $maxMatches <= 0 ) { |
| 901 | $skippedMatches++; |
| 902 | return $m[0]; |
| 903 | } else { |
| 904 | $maxMatches--; |
| 905 | return $realCallback( $m ); |
| 906 | } |
| 907 | }; |
| 908 | } |
| 909 | |
| 910 | $count = 0; |
| 911 | $s2 = preg_replace_callback( $re, $cb, $s, $n, $count ); |
| 912 | if ( $s2 === null ) { |
| 913 | $this->handlePCREError( preg_last_error(), $pattern ); |
| 914 | } |
| 915 | return [ $s2, $count - $skippedMatches ]; |
| 916 | } |
| 917 | |
| 918 | /** |
| 919 | * Handle a PCRE error |
| 920 | * @param int $error From preg_last_error() |
| 921 | * @param string $pattern Pattern being matched |
| 922 | * @throws LuaError |
| 923 | */ |
| 924 | private function handlePCREError( $error, $pattern ) { |
| 925 | $PREG_JIT_STACKLIMIT_ERROR = defined( 'PREG_JIT_STACKLIMIT_ERROR' ) |
| 926 | ? PREG_JIT_STACKLIMIT_ERROR |
| 927 | : 'PREG_JIT_STACKLIMIT_ERROR'; |
| 928 | |
| 929 | $error = preg_last_error(); |
| 930 | switch ( $error ) { |
| 931 | case PREG_NO_ERROR: |
| 932 | // Huh? |
| 933 | break; |
| 934 | case PREG_INTERNAL_ERROR: |
| 935 | throw new LuaError( "PCRE internal error" ); |
| 936 | case PREG_BACKTRACK_LIMIT_ERROR: |
| 937 | throw new LuaError( |
| 938 | "PCRE backtrack limit reached while matching pattern '$pattern'" |
| 939 | ); |
| 940 | case PREG_RECURSION_LIMIT_ERROR: |
| 941 | throw new LuaError( |
| 942 | "PCRE recursion limit reached while matching pattern '$pattern'" |
| 943 | ); |
| 944 | case PREG_BAD_UTF8_ERROR: |
| 945 | // Should have alreay been caught, but just in case |
| 946 | throw new LuaError( "PCRE bad UTF-8 error" ); |
| 947 | case PREG_BAD_UTF8_OFFSET_ERROR: |
| 948 | // Shouldn't happen, but just in case |
| 949 | throw new LuaError( "PCRE bad UTF-8 offset error" ); |
| 950 | case $PREG_JIT_STACKLIMIT_ERROR: |
| 951 | throw new LuaError( |
| 952 | "PCRE JIT stack limit reached while matching pattern '$pattern'" |
| 953 | ); |
| 954 | default: |
| 955 | throw new LuaError( |
| 956 | "PCRE error code $error while matching pattern '$pattern'" |
| 957 | ); |
| 958 | } |
| 959 | } |
| 960 | } |