php-parser/lib/PhpParser/Lexer.php

407 lines
16 KiB
PHP
Raw Normal View History

2017-08-18 22:57:27 +02:00
<?php declare(strict_types=1);
2011-04-18 19:02:30 +02:00
namespace PhpParser;
use PhpParser\Parser\Tokens;
class Lexer
2011-04-18 19:02:30 +02:00
{
2011-06-03 22:02:02 +02:00
protected $code;
2011-04-18 19:02:30 +02:00
protected $tokens;
protected $pos;
protected $line;
protected $filePos;
protected $prevCloseTagHasNewline;
2011-04-18 19:02:30 +02:00
protected $tokenMap;
protected $dropTokens;
2011-04-18 19:02:30 +02:00
2019-05-12 15:26:26 +02:00
private $attributeStartLineUsed;
private $attributeEndLineUsed;
private $attributeStartTokenPosUsed;
private $attributeEndTokenPosUsed;
private $attributeStartFilePosUsed;
private $attributeEndFilePosUsed;
private $attributeCommentsUsed;
/**
* Creates a Lexer.
*
* @param array $options Options array. Currently only the 'usedAttributes' option is supported,
* which is an array of attributes to add to the AST nodes. Possible
* attributes are: 'comments', 'startLine', 'endLine', 'startTokenPos',
* 'endTokenPos', 'startFilePos', 'endFilePos'. The option defaults to the
* first three. For more info see getNextToken() docs.
*/
public function __construct(array $options = []) {
2014-02-12 17:47:34 +01:00
// map from internal tokens to PhpParser tokens
$this->tokenMap = $this->createTokenMap();
// Compatibility define for PHP < 7.4
if (!defined('T_BAD_CHARACTER')) {
\define('T_BAD_CHARACTER', -1);
}
// map of tokens to drop while lexing (the map is only used for isset lookup,
// that's why the value is simply set to 1; the value is never actually used.)
2016-04-02 00:54:01 +02:00
$this->dropTokens = array_fill_keys(
[\T_WHITESPACE, \T_OPEN_TAG, \T_COMMENT, \T_DOC_COMMENT, \T_BAD_CHARACTER], 1
2016-04-02 00:54:01 +02:00
);
2019-05-12 15:26:26 +02:00
$defaultAttributes = ['comments', 'startLine', 'endLine'];
$usedAttributes = array_fill_keys($options['usedAttributes'] ?? $defaultAttributes, true);
// Create individual boolean properties to make these checks faster.
$this->attributeStartLineUsed = isset($usedAttributes['startLine']);
$this->attributeEndLineUsed = isset($usedAttributes['endLine']);
$this->attributeStartTokenPosUsed = isset($usedAttributes['startTokenPos']);
$this->attributeEndTokenPosUsed = isset($usedAttributes['endTokenPos']);
$this->attributeStartFilePosUsed = isset($usedAttributes['startFilePos']);
$this->attributeEndFilePosUsed = isset($usedAttributes['endFilePos']);
$this->attributeCommentsUsed = isset($usedAttributes['comments']);
}
/**
* Initializes the lexer for lexing the provided source code.
*
* This function does not throw if lexing errors occur. Instead, errors may be retrieved using
* the getErrors() method.
2011-07-13 23:07:05 +02:00
*
* @param string $code The source code to lex
* @param ErrorHandler|null $errorHandler Error handler to use for lexing errors. Defaults to
* ErrorHandler\Throwing
*/
2017-04-28 21:40:59 +02:00
public function startLexing(string $code, ErrorHandler $errorHandler = null) {
if (null === $errorHandler) {
$errorHandler = new ErrorHandler\Throwing();
}
$this->code = $code; // keep the code around for __halt_compiler() handling
$this->pos = -1;
$this->line = 1;
$this->filePos = 0;
// If inline HTML occurs without preceding code, treat it as if it had a leading newline.
// This ensures proper composability, because having a newline is the "safe" assumption.
$this->prevCloseTagHasNewline = true;
$scream = ini_set('xdebug.scream', '0');
2014-04-19 22:26:05 +02:00
error_clear_last();
$this->tokens = @token_get_all($code);
$this->handleErrors($errorHandler);
if (false !== $scream) {
ini_set('xdebug.scream', $scream);
}
}
private function handleInvalidCharacterRange($start, $end, $line, ErrorHandler $errorHandler) {
$tokens = [];
for ($i = $start; $i < $end; $i++) {
$chr = $this->code[$i];
if ($chr === "\0") {
// PHP cuts error message after null byte, so need special case
$errorMsg = 'Unexpected null byte';
} else {
$errorMsg = sprintf(
'Unexpected character "%s" (ASCII %d)', $chr, ord($chr)
);
}
2016-09-30 20:33:56 +02:00
$tokens[] = [\T_BAD_CHARACTER, $chr, $line];
$errorHandler->handleError(new Error($errorMsg, [
'startLine' => $line,
'endLine' => $line,
'startFilePos' => $i,
'endFilePos' => $i,
]));
}
return $tokens;
}
2017-01-24 08:38:55 +01:00
/**
* Check whether comment token is unterminated.
*
2017-01-24 08:38:55 +01:00
* @return bool
*/
2017-04-28 21:40:59 +02:00
private function isUnterminatedComment($token) : bool {
return ($token[0] === \T_COMMENT || $token[0] === \T_DOC_COMMENT)
&& substr($token[1], 0, 2) === '/*'
&& substr($token[1], -2) !== '*/';
}
2017-01-24 08:38:55 +01:00
/**
* Check whether an error *may* have occurred during tokenization.
*
2017-01-24 08:38:55 +01:00
* @return bool
*/
2017-04-28 21:40:59 +02:00
private function errorMayHaveOccurred() : bool {
if (defined('HHVM_VERSION')) {
// In HHVM token_get_all() does not throw warnings, so we need to conservatively
// assume that an error occurred
return true;
}
return null !== error_get_last();
}
protected function handleErrors(ErrorHandler $errorHandler) {
if (!$this->errorMayHaveOccurred()) {
return;
}
// PHP's error handling for token_get_all() is rather bad, so if we want detailed
// error information we need to compute it ourselves. Invalid character errors are
// detected by finding "gaps" in the token array. Unterminated comments are detected
// by checking if a trailing comment has a "*/" at the end.
$filePos = 0;
$line = 1;
$numTokens = \count($this->tokens);
for ($i = 0; $i < $numTokens; $i++) {
$token = $this->tokens[$i];
// Since PHP 7.4 invalid characters are represented by a T_BAD_CHARACTER token.
// In this case we only need to emit an error.
if ($token[0] === \T_BAD_CHARACTER) {
$this->handleInvalidCharacterRange($filePos, $filePos + 1, $line, $errorHandler);
}
$tokenValue = \is_string($token) ? $token : $token[1];
$tokenLen = \strlen($tokenValue);
if (substr($this->code, $filePos, $tokenLen) !== $tokenValue) {
// Something is missing, must be an invalid character
$nextFilePos = strpos($this->code, $tokenValue, $filePos);
$badCharTokens = $this->handleInvalidCharacterRange(
$filePos, $nextFilePos, $line, $errorHandler);
2018-01-10 17:31:33 +01:00
$filePos = (int) $nextFilePos;
array_splice($this->tokens, $i, 0, $badCharTokens);
$numTokens += \count($badCharTokens);
$i += \count($badCharTokens);
}
$filePos += $tokenLen;
$line += substr_count($tokenValue, "\n");
}
if ($filePos !== \strlen($this->code)) {
2016-09-30 20:57:21 +02:00
if (substr($this->code, $filePos, 2) === '/*') {
// Unlike PHP, HHVM will drop unterminated comments entirely
$comment = substr($this->code, $filePos);
$errorHandler->handleError(new Error('Unterminated comment', [
2016-09-30 20:57:21 +02:00
'startLine' => $line,
'endLine' => $line + substr_count($comment, "\n"),
'startFilePos' => $filePos,
'endFilePos' => $filePos + \strlen($comment),
]));
2016-09-30 20:57:21 +02:00
// Emulate the PHP behavior
$isDocComment = isset($comment[3]) && $comment[3] === '*';
$this->tokens[] = [$isDocComment ? \T_DOC_COMMENT : \T_COMMENT, $comment, $line];
2016-09-30 20:57:21 +02:00
} else {
// Invalid characters at the end of the input
$badCharTokens = $this->handleInvalidCharacterRange(
$filePos, \strlen($this->code), $line, $errorHandler);
$this->tokens = array_merge($this->tokens, $badCharTokens);
2016-09-30 20:57:21 +02:00
}
2016-09-30 21:05:44 +02:00
return;
2011-07-13 12:24:10 +02:00
}
if (count($this->tokens) > 0) {
// Check for unterminated comment
$lastToken = $this->tokens[count($this->tokens) - 1];
if ($this->isUnterminatedComment($lastToken)) {
$errorHandler->handleError(new Error('Unterminated comment', [
'startLine' => $line - substr_count($lastToken[1], "\n"),
'endLine' => $line,
'startFilePos' => $filePos - \strlen($lastToken[1]),
'endFilePos' => $filePos,
]));
}
}
2011-04-18 19:02:30 +02:00
}
/**
* Fetches the next token.
*
* The available attributes are determined by the 'usedAttributes' option, which can
* be specified in the constructor. The following attributes are supported:
*
* * 'comments' => Array of PhpParser\Comment or PhpParser\Comment\Doc instances,
* representing all comments that occurred between the previous
* non-discarded token and the current one.
* * 'startLine' => Line in which the node starts.
* * 'endLine' => Line in which the node ends.
* * 'startTokenPos' => Offset into the token array of the first token in the node.
* * 'endTokenPos' => Offset into the token array of the last token in the node.
* * 'startFilePos' => Offset into the code string of the first character that is part of the node.
* * 'endFilePos' => Offset into the code string of the last character that is part of the node.
*
* @param mixed $value Variable to store token content in
* @param mixed $startAttributes Variable to store start attributes in
* @param mixed $endAttributes Variable to store end attributes in
*
* @return int Token id
*/
2017-04-28 21:40:59 +02:00
public function getNextToken(&$value = null, &$startAttributes = null, &$endAttributes = null) : int {
$startAttributes = [];
$endAttributes = [];
while (1) {
if (isset($this->tokens[++$this->pos])) {
$token = $this->tokens[$this->pos];
} else {
// EOF token with ID 0
$token = "\0";
}
2011-06-03 22:02:02 +02:00
2019-05-12 15:26:26 +02:00
if ($this->attributeStartLineUsed) {
$startAttributes['startLine'] = $this->line;
}
2019-05-12 15:26:26 +02:00
if ($this->attributeStartTokenPosUsed) {
$startAttributes['startTokenPos'] = $this->pos;
}
2019-05-12 15:26:26 +02:00
if ($this->attributeStartFilePosUsed) {
$startAttributes['startFilePos'] = $this->filePos;
}
if (\is_string($token)) {
$value = $token;
if (isset($token[1])) {
// bug in token_get_all
$this->filePos += 2;
$id = ord('"');
} else {
$this->filePos += 1;
$id = ord($token);
}
2016-04-02 00:54:01 +02:00
} elseif (!isset($this->dropTokens[$token[0]])) {
$value = $token[1];
$id = $this->tokenMap[$token[0]];
if (\T_CLOSE_TAG === $token[0]) {
$this->prevCloseTagHasNewline = false !== strpos($token[1], "\n");
} elseif (\T_INLINE_HTML === $token[0]) {
$startAttributes['hasLeadingNewline'] = $this->prevCloseTagHasNewline;
}
2011-07-13 13:27:14 +02:00
2016-04-02 00:54:01 +02:00
$this->line += substr_count($value, "\n");
$this->filePos += \strlen($value);
2016-04-02 00:54:01 +02:00
} else {
if (\T_COMMENT === $token[0] || \T_DOC_COMMENT === $token[0]) {
2019-05-12 15:26:26 +02:00
if ($this->attributeCommentsUsed) {
$comment = \T_DOC_COMMENT === $token[0]
2017-11-04 17:45:14 +01:00
? new Comment\Doc($token[1], $this->line, $this->filePos, $this->pos)
: new Comment($token[1], $this->line, $this->filePos, $this->pos);
2016-04-02 00:54:01 +02:00
$startAttributes['comments'][] = $comment;
}
2011-07-13 13:27:14 +02:00
}
2016-04-02 00:54:01 +02:00
$this->line += substr_count($token[1], "\n");
$this->filePos += \strlen($token[1]);
continue;
2011-04-18 19:02:30 +02:00
}
2019-05-12 15:26:26 +02:00
if ($this->attributeEndLineUsed) {
$endAttributes['endLine'] = $this->line;
}
2019-05-12 15:26:26 +02:00
if ($this->attributeEndTokenPosUsed) {
$endAttributes['endTokenPos'] = $this->pos;
}
2019-05-12 15:26:26 +02:00
if ($this->attributeEndFilePosUsed) {
$endAttributes['endFilePos'] = $this->filePos - 1;
}
return $id;
2011-04-18 19:02:30 +02:00
}
throw new \RuntimeException('Reached end of lexer loop');
2011-04-18 19:02:30 +02:00
}
/**
* Returns the token array for current code.
*
* The token array is in the same format as provided by the
* token_get_all() function and does not discard tokens (i.e.
* whitespace and comments are included). The token position
* attributes are against this token array.
*
* @return array Array of tokens in token_get_all() format
*/
2017-04-28 21:40:59 +02:00
public function getTokens() : array {
return $this->tokens;
}
2011-06-03 22:02:02 +02:00
/**
* Handles __halt_compiler() by returning the text after it.
*
* @return string Remaining text
*/
2017-04-28 21:40:59 +02:00
public function handleHaltCompiler() : string {
2011-06-03 22:02:02 +02:00
// text after T_HALT_COMPILER, still including ();
$textAfter = substr($this->code, $this->filePos);
2011-06-03 22:02:02 +02:00
// ensure that it is followed by ();
// this simplifies the situation, by not allowing any comments
// in between of the tokens.
2015-06-20 12:09:03 +02:00
if (!preg_match('~^\s*\(\s*\)\s*(?:;|\?>\r?\n?)~', $textAfter, $matches)) {
throw new Error('__HALT_COMPILER must be followed by "();"');
2011-06-03 22:02:02 +02:00
}
// prevent the lexer from returning any further tokens
$this->pos = count($this->tokens);
// return with (); removed
return substr($textAfter, strlen($matches[0]));
2011-06-03 22:02:02 +02:00
}
/**
* Creates the token map.
*
* The token map maps the PHP internal token identifiers
* to the identifiers used by the Parser. Additionally it
* maps T_OPEN_TAG_WITH_ECHO to T_ECHO and T_CLOSE_TAG to ';'.
*
* @return array The token map
*/
2017-04-28 21:40:59 +02:00
protected function createTokenMap() : array {
$tokenMap = [];
// 256 is the minimum possible token number, as everything below
// it is an ASCII value
for ($i = 256; $i < 1000; ++$i) {
if (\T_DOUBLE_COLON === $i) {
2015-03-14 19:52:37 +01:00
// T_DOUBLE_COLON is equivalent to T_PAAMAYIM_NEKUDOTAYIM
$tokenMap[$i] = Tokens::T_PAAMAYIM_NEKUDOTAYIM;
} elseif(\T_OPEN_TAG_WITH_ECHO === $i) {
2015-03-14 19:52:37 +01:00
// T_OPEN_TAG_WITH_ECHO with dropped T_OPEN_TAG results in T_ECHO
$tokenMap[$i] = Tokens::T_ECHO;
} elseif(\T_CLOSE_TAG === $i) {
2015-03-14 19:52:37 +01:00
// T_CLOSE_TAG is equivalent to ';'
$tokenMap[$i] = ord(';');
2015-03-14 19:52:37 +01:00
} elseif ('UNKNOWN' !== $name = token_name($i)) {
if ('T_HASHBANG' === $name) {
// HHVM uses a special token for #! hashbang lines
$tokenMap[$i] = Tokens::T_INLINE_HTML;
} elseif (defined($name = Tokens::class . '::' . $name)) {
2015-03-14 19:52:37 +01:00
// Other tokens can be mapped directly
$tokenMap[$i] = constant($name);
}
2011-04-18 19:02:30 +02:00
}
}
2014-09-30 20:55:43 +02:00
// HHVM uses a special token for numbers that overflow to double
if (defined('T_ONUMBER')) {
$tokenMap[\T_ONUMBER] = Tokens::T_DNUMBER;
2014-09-30 20:55:43 +02:00
}
// HHVM also has a separate token for the __COMPILER_HALT_OFFSET__ constant
if (defined('T_COMPILER_HALT_OFFSET')) {
$tokenMap[\T_COMPILER_HALT_OFFSET] = Tokens::T_STRING;
}
2014-09-30 20:55:43 +02:00
return $tokenMap;
2011-04-18 19:02:30 +02:00
}
}