| 1 |
<?php |
| 2 |
// phpcs:ignoreFile -- Bundled third-party (Mozart) dependency; exempt from plugin coding standards. |
| 3 |
|
| 4 |
namespace WPDeveloper\BetterDocs\Dependencies\PhpParser; |
| 5 |
|
| 6 |
use WPDeveloper\BetterDocs\Dependencies\PhpParser\Parser\Tokens; |
| 7 |
|
| 8 |
class Lexer |
| 9 |
{ |
| 10 |
protected $code; |
| 11 |
protected $tokens; |
| 12 |
protected $pos; |
| 13 |
protected $line; |
| 14 |
protected $filePos; |
| 15 |
protected $prevCloseTagHasNewline; |
| 16 |
|
| 17 |
protected $tokenMap; |
| 18 |
protected $dropTokens; |
| 19 |
|
| 20 |
protected $usedAttributes; |
| 21 |
|
| 22 |
/** |
| 23 |
* Creates a Lexer. |
| 24 |
* |
| 25 |
* @param array $options Options array. Currently only the 'usedAttributes' option is supported, |
| 26 |
* which is an array of attributes to add to the AST nodes. Possible |
| 27 |
* attributes are: 'comments', 'startLine', 'endLine', 'startTokenPos', |
| 28 |
* 'endTokenPos', 'startFilePos', 'endFilePos'. The option defaults to the |
| 29 |
* first three. For more info see getNextToken() docs. |
| 30 |
*/ |
| 31 |
public function __construct(array $options = array()) { |
| 32 |
// map from internal tokens to WPDeveloper\BetterDocs\Dependencies\PhpParser tokens |
| 33 |
$this->tokenMap = $this->createTokenMap(); |
| 34 |
|
| 35 |
// map of tokens to drop while lexing (the map is only used for isset lookup, |
| 36 |
// that's why the value is simply set to 1; the value is never actually used.) |
| 37 |
$this->dropTokens = array_fill_keys( |
| 38 |
array(T_WHITESPACE, T_OPEN_TAG, T_COMMENT, T_DOC_COMMENT), 1 |
| 39 |
); |
| 40 |
|
| 41 |
// the usedAttributes member is a map of the used attribute names to a dummy |
| 42 |
// value (here "true") |
| 43 |
$options += array( |
| 44 |
'usedAttributes' => array('comments', 'startLine', 'endLine'), |
| 45 |
); |
| 46 |
$this->usedAttributes = array_fill_keys($options['usedAttributes'], true); |
| 47 |
} |
| 48 |
|
| 49 |
/** |
| 50 |
* Initializes the lexer for lexing the provided source code. |
| 51 |
* |
| 52 |
* This function does not throw if lexing errors occur. Instead, errors may be retrieved using |
| 53 |
* the getErrors() method. |
| 54 |
* |
| 55 |
* @param string $code The source code to lex |
| 56 |
* @param ErrorHandler|null $errorHandler Error handler to use for lexing errors. Defaults to |
| 57 |
* ErrorHandler\Throwing |
| 58 |
*/ |
| 59 |
public function startLexing($code, ?ErrorHandler $errorHandler = null) { |
| 60 |
if (null === $errorHandler) { |
| 61 |
$errorHandler = new ErrorHandler\Throwing(); |
| 62 |
} |
| 63 |
|
| 64 |
$this->code = $code; // keep the code around for __halt_compiler() handling |
| 65 |
$this->pos = -1; |
| 66 |
$this->line = 1; |
| 67 |
$this->filePos = 0; |
| 68 |
|
| 69 |
// If inline HTML occurs without preceding code, treat it as if it had a leading newline. |
| 70 |
// This ensures proper composability, because having a newline is the "safe" assumption. |
| 71 |
$this->prevCloseTagHasNewline = true; |
| 72 |
|
| 73 |
$scream = ini_set('xdebug.scream', '0'); |
| 74 |
|
| 75 |
$this->resetErrors(); |
| 76 |
$this->tokens = @token_get_all($code); |
| 77 |
$this->handleErrors($errorHandler); |
| 78 |
|
| 79 |
if (false !== $scream) { |
| 80 |
ini_set('xdebug.scream', $scream); |
| 81 |
} |
| 82 |
} |
| 83 |
|
| 84 |
protected function resetErrors() { |
| 85 |
if (function_exists('error_clear_last')) { |
| 86 |
error_clear_last(); |
| 87 |
} else { |
| 88 |
// set error_get_last() to defined state by forcing an undefined variable error |
| 89 |
set_error_handler(function() { return false; }, 0); |
| 90 |
@$undefinedVariable; |
| 91 |
restore_error_handler(); |
| 92 |
} |
| 93 |
} |
| 94 |
|
| 95 |
private function handleInvalidCharacterRange($start, $end, $line, ErrorHandler $errorHandler) { |
| 96 |
for ($i = $start; $i < $end; $i++) { |
| 97 |
$chr = $this->code[$i]; |
| 98 |
if ($chr === 'b' || $chr === 'B') { |
| 99 |
// HHVM does not treat b" tokens correctly, so ignore these |
| 100 |
continue; |
| 101 |
} |
| 102 |
|
| 103 |
if ($chr === "\0") { |
| 104 |
// PHP cuts error message after null byte, so need special case |
| 105 |
$errorMsg = 'Unexpected null byte'; |
| 106 |
} else { |
| 107 |
$errorMsg = sprintf( |
| 108 |
'Unexpected character "%s" (ASCII %d)', $chr, ord($chr) |
| 109 |
); |
| 110 |
} |
| 111 |
|
| 112 |
$errorHandler->handleError(new Error($errorMsg, [ |
| 113 |
'startLine' => $line, |
| 114 |
'endLine' => $line, |
| 115 |
'startFilePos' => $i, |
| 116 |
'endFilePos' => $i, |
| 117 |
])); |
| 118 |
} |
| 119 |
} |
| 120 |
|
| 121 |
private function isUnterminatedComment($token) { |
| 122 |
return ($token[0] === T_COMMENT || $token[0] === T_DOC_COMMENT) |
| 123 |
&& substr($token[1], 0, 2) === '/*' |
| 124 |
&& substr($token[1], -2) !== '*/'; |
| 125 |
} |
| 126 |
|
| 127 |
private function errorMayHaveOccurred() { |
| 128 |
if (defined('HHVM_VERSION')) { |
| 129 |
// In HHVM token_get_all() does not throw warnings, so we need to conservatively |
| 130 |
// assume that an error occurred |
| 131 |
return true; |
| 132 |
} |
| 133 |
|
| 134 |
$error = error_get_last(); |
| 135 |
return null !== $error |
| 136 |
&& false === strpos($error['message'], 'Undefined variable'); |
| 137 |
} |
| 138 |
|
| 139 |
protected function handleErrors(ErrorHandler $errorHandler) { |
| 140 |
if (!$this->errorMayHaveOccurred()) { |
| 141 |
return; |
| 142 |
} |
| 143 |
|
| 144 |
// PHP's error handling for token_get_all() is rather bad, so if we want detailed |
| 145 |
// error information we need to compute it ourselves. Invalid character errors are |
| 146 |
// detected by finding "gaps" in the token array. Unterminated comments are detected |
| 147 |
// by checking if a trailing comment has a "*/" at the end. |
| 148 |
|
| 149 |
$filePos = 0; |
| 150 |
$line = 1; |
| 151 |
foreach ($this->tokens as $i => $token) { |
| 152 |
$tokenValue = \is_string($token) ? $token : $token[1]; |
| 153 |
$tokenLen = \strlen($tokenValue); |
| 154 |
|
| 155 |
if (substr($this->code, $filePos, $tokenLen) !== $tokenValue) { |
| 156 |
// Something is missing, must be an invalid character |
| 157 |
$nextFilePos = strpos($this->code, $tokenValue, $filePos); |
| 158 |
$this->handleInvalidCharacterRange( |
| 159 |
$filePos, $nextFilePos, $line, $errorHandler); |
| 160 |
$filePos = $nextFilePos; |
| 161 |
} |
| 162 |
|
| 163 |
$filePos += $tokenLen; |
| 164 |
$line += substr_count($tokenValue, "\n"); |
| 165 |
} |
| 166 |
|
| 167 |
if ($filePos !== \strlen($this->code)) { |
| 168 |
if (substr($this->code, $filePos, 2) === '/*') { |
| 169 |
// Unlike PHP, HHVM will drop unterminated comments entirely |
| 170 |
$comment = substr($this->code, $filePos); |
| 171 |
$errorHandler->handleError(new Error('Unterminated comment', [ |
| 172 |
'startLine' => $line, |
| 173 |
'endLine' => $line + substr_count($comment, "\n"), |
| 174 |
'startFilePos' => $filePos, |
| 175 |
'endFilePos' => $filePos + \strlen($comment), |
| 176 |
])); |
| 177 |
|
| 178 |
// Emulate the PHP behavior |
| 179 |
$isDocComment = isset($comment[3]) && $comment[3] === '*'; |
| 180 |
$this->tokens[] = [$isDocComment ? T_DOC_COMMENT : T_COMMENT, $comment, $line]; |
| 181 |
} else { |
| 182 |
// Invalid characters at the end of the input |
| 183 |
$this->handleInvalidCharacterRange( |
| 184 |
$filePos, \strlen($this->code), $line, $errorHandler); |
| 185 |
} |
| 186 |
return; |
| 187 |
} |
| 188 |
|
| 189 |
if (count($this->tokens) > 0) { |
| 190 |
// Check for unterminated comment |
| 191 |
$lastToken = $this->tokens[count($this->tokens) - 1]; |
| 192 |
if ($this->isUnterminatedComment($lastToken)) { |
| 193 |
$errorHandler->handleError(new Error('Unterminated comment', [ |
| 194 |
'startLine' => $line - substr_count($lastToken[1], "\n"), |
| 195 |
'endLine' => $line, |
| 196 |
'startFilePos' => $filePos - \strlen($lastToken[1]), |
| 197 |
'endFilePos' => $filePos, |
| 198 |
])); |
| 199 |
} |
| 200 |
} |
| 201 |
} |
| 202 |
|
| 203 |
/** |
| 204 |
* Fetches the next token. |
| 205 |
* |
| 206 |
* The available attributes are determined by the 'usedAttributes' option, which can |
| 207 |
* be specified in the constructor. The following attributes are supported: |
| 208 |
* |
| 209 |
* * 'comments' => Array of WPDeveloper\BetterDocs\Dependencies\PhpParser\Comment or WPDeveloper\BetterDocs\Dependencies\PhpParser\Comment\Doc instances, |
| 210 |
* representing all comments that occurred between the previous |
| 211 |
* non-discarded token and the current one. |
| 212 |
* * 'startLine' => Line in which the node starts. |
| 213 |
* * 'endLine' => Line in which the node ends. |
| 214 |
* * 'startTokenPos' => Offset into the token array of the first token in the node. |
| 215 |
* * 'endTokenPos' => Offset into the token array of the last token in the node. |
| 216 |
* * 'startFilePos' => Offset into the code string of the first character that is part of the node. |
| 217 |
* * 'endFilePos' => Offset into the code string of the last character that is part of the node. |
| 218 |
* |
| 219 |
* @param mixed $value Variable to store token content in |
| 220 |
* @param mixed $startAttributes Variable to store start attributes in |
| 221 |
* @param mixed $endAttributes Variable to store end attributes in |
| 222 |
* |
| 223 |
* @return int Token id |
| 224 |
*/ |
| 225 |
public function getNextToken(&$value = null, &$startAttributes = null, &$endAttributes = null) { |
| 226 |
$startAttributes = array(); |
| 227 |
$endAttributes = array(); |
| 228 |
|
| 229 |
while (1) { |
| 230 |
if (isset($this->tokens[++$this->pos])) { |
| 231 |
$token = $this->tokens[$this->pos]; |
| 232 |
} else { |
| 233 |
// EOF token with ID 0 |
| 234 |
$token = "\0"; |
| 235 |
} |
| 236 |
|
| 237 |
if (isset($this->usedAttributes['startLine'])) { |
| 238 |
$startAttributes['startLine'] = $this->line; |
| 239 |
} |
| 240 |
if (isset($this->usedAttributes['startTokenPos'])) { |
| 241 |
$startAttributes['startTokenPos'] = $this->pos; |
| 242 |
} |
| 243 |
if (isset($this->usedAttributes['startFilePos'])) { |
| 244 |
$startAttributes['startFilePos'] = $this->filePos; |
| 245 |
} |
| 246 |
|
| 247 |
if (\is_string($token)) { |
| 248 |
$value = $token; |
| 249 |
if (isset($token[1])) { |
| 250 |
// bug in token_get_all |
| 251 |
$this->filePos += 2; |
| 252 |
$id = ord('"'); |
| 253 |
} else { |
| 254 |
$this->filePos += 1; |
| 255 |
$id = ord($token); |
| 256 |
} |
| 257 |
} elseif (!isset($this->dropTokens[$token[0]])) { |
| 258 |
$value = $token[1]; |
| 259 |
$id = $this->tokenMap[$token[0]]; |
| 260 |
if (T_CLOSE_TAG === $token[0]) { |
| 261 |
$this->prevCloseTagHasNewline = false !== strpos($token[1], "\n"); |
| 262 |
} else if (T_INLINE_HTML === $token[0]) { |
| 263 |
$startAttributes['hasLeadingNewline'] = $this->prevCloseTagHasNewline; |
| 264 |
} |
| 265 |
|
| 266 |
$this->line += substr_count($value, "\n"); |
| 267 |
$this->filePos += \strlen($value); |
| 268 |
} else { |
| 269 |
if (T_COMMENT === $token[0] || T_DOC_COMMENT === $token[0]) { |
| 270 |
if (isset($this->usedAttributes['comments'])) { |
| 271 |
$comment = T_DOC_COMMENT === $token[0] |
| 272 |
? new Comment\Doc($token[1], $this->line, $this->filePos) |
| 273 |
: new Comment($token[1], $this->line, $this->filePos); |
| 274 |
$startAttributes['comments'][] = $comment; |
| 275 |
} |
| 276 |
} |
| 277 |
|
| 278 |
$this->line += substr_count($token[1], "\n"); |
| 279 |
$this->filePos += \strlen($token[1]); |
| 280 |
continue; |
| 281 |
} |
| 282 |
|
| 283 |
if (isset($this->usedAttributes['endLine'])) { |
| 284 |
$endAttributes['endLine'] = $this->line; |
| 285 |
} |
| 286 |
if (isset($this->usedAttributes['endTokenPos'])) { |
| 287 |
$endAttributes['endTokenPos'] = $this->pos; |
| 288 |
} |
| 289 |
if (isset($this->usedAttributes['endFilePos'])) { |
| 290 |
$endAttributes['endFilePos'] = $this->filePos - 1; |
| 291 |
} |
| 292 |
|
| 293 |
return $id; |
| 294 |
} |
| 295 |
|
| 296 |
throw new \RuntimeException('Reached end of lexer loop'); |
| 297 |
} |
| 298 |
|
| 299 |
/** |
| 300 |
* Returns the token array for current code. |
| 301 |
* |
| 302 |
* The token array is in the same format as provided by the |
| 303 |
* token_get_all() function and does not discard tokens (i.e. |
| 304 |
* whitespace and comments are included). The token position |
| 305 |
* attributes are against this token array. |
| 306 |
* |
| 307 |
* @return array Array of tokens in token_get_all() format |
| 308 |
*/ |
| 309 |
public function getTokens() { |
| 310 |
return $this->tokens; |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* Handles __halt_compiler() by returning the text after it. |
| 315 |
* |
| 316 |
* @return string Remaining text |
| 317 |
*/ |
| 318 |
public function handleHaltCompiler() { |
| 319 |
// text after T_HALT_COMPILER, still including (); |
| 320 |
$textAfter = substr($this->code, $this->filePos); |
| 321 |
|
| 322 |
// ensure that it is followed by (); |
| 323 |
// this simplifies the situation, by not allowing any comments |
| 324 |
// in between of the tokens. |
| 325 |
if (!preg_match('~^\s*\(\s*\)\s*(?:;|\?>\r?\n?)~', $textAfter, $matches)) { |
| 326 |
throw new Error('__HALT_COMPILER must be followed by "();"'); |
| 327 |
} |
| 328 |
|
| 329 |
// prevent the lexer from returning any further tokens |
| 330 |
$this->pos = count($this->tokens); |
| 331 |
|
| 332 |
// return with (); removed |
| 333 |
return (string) substr($textAfter, strlen($matches[0])); // (string) converts false to '' |
| 334 |
} |
| 335 |
|
| 336 |
/** |
| 337 |
* Creates the token map. |
| 338 |
* |
| 339 |
* The token map maps the PHP internal token identifiers |
| 340 |
* to the identifiers used by the Parser. Additionally it |
| 341 |
* maps T_OPEN_TAG_WITH_ECHO to T_ECHO and T_CLOSE_TAG to ';'. |
| 342 |
* |
| 343 |
* @return array The token map |
| 344 |
*/ |
| 345 |
protected function createTokenMap() { |
| 346 |
$tokenMap = array(); |
| 347 |
|
| 348 |
// 256 is the minimum possible token number, as everything below |
| 349 |
// it is an ASCII value |
| 350 |
for ($i = 256; $i < 1000; ++$i) { |
| 351 |
if (T_DOUBLE_COLON === $i) { |
| 352 |
// T_DOUBLE_COLON is equivalent to T_PAAMAYIM_NEKUDOTAYIM |
| 353 |
$tokenMap[$i] = Tokens::T_PAAMAYIM_NEKUDOTAYIM; |
| 354 |
} elseif(T_OPEN_TAG_WITH_ECHO === $i) { |
| 355 |
// T_OPEN_TAG_WITH_ECHO with dropped T_OPEN_TAG results in T_ECHO |
| 356 |
$tokenMap[$i] = Tokens::T_ECHO; |
| 357 |
} elseif(T_CLOSE_TAG === $i) { |
| 358 |
// T_CLOSE_TAG is equivalent to ';' |
| 359 |
$tokenMap[$i] = ord(';'); |
| 360 |
} elseif ('UNKNOWN' !== $name = token_name($i)) { |
| 361 |
if ('T_HASHBANG' === $name) { |
| 362 |
// HHVM uses a special token for #! hashbang lines |
| 363 |
$tokenMap[$i] = Tokens::T_INLINE_HTML; |
| 364 |
} else if (defined($name = Tokens::class . '::' . $name)) { |
| 365 |
// Other tokens can be mapped directly |
| 366 |
$tokenMap[$i] = constant($name); |
| 367 |
} |
| 368 |
} |
| 369 |
} |
| 370 |
|
| 371 |
// HHVM uses a special token for numbers that overflow to double |
| 372 |
if (defined('T_ONUMBER')) { |
| 373 |
$tokenMap[T_ONUMBER] = Tokens::T_DNUMBER; |
| 374 |
} |
| 375 |
// HHVM also has a separate token for the __COMPILER_HALT_OFFSET__ constant |
| 376 |
if (defined('T_COMPILER_HALT_OFFSET')) { |
| 377 |
$tokenMap[T_COMPILER_HALT_OFFSET] = Tokens::T_STRING; |
| 378 |
} |
| 379 |
|
| 380 |
return $tokenMap; |
| 381 |
} |
| 382 |
} |
| 383 |
|
