Skip to content
Parser and lexer for PHP
Branch: master
Clone or download
Latest commit 7458a74 Jun 1, 2019

Build Status Build status

Parle provides lexing and parsing facilities for PHP

Lexing and parsing is used widely in the PHP core and extensions. Usually such a functionality is packed into a piece of C/C++ and depends on tools like flex, re2c, Bison, LEMON or similar. With Parle, it is possible to implement lexing and parsing in PHP while relying on features and principles of the parser/lexer generator tools for C/C++. The Lexer and Parser classes are there in the Parle namespace. The implementation bases on the work of Ben Hanson

The lexer is based on the pattern matching similar to flex. The parser is LALR(1).

Supported is PHP 7.0 and above. A C++14 capable compiler is required. As of version 0.7.3 parle can optionally be compiled with internal UTF-32 support, making it possible to use Unicode character classes in patterns.

The full extension documentation is available in the PHP Manual.


Read the documentation.

Example tokenizing comma separated integer list

use Parle\Token;
use Parle\Lexer;
use Parle\LexerException;

/* name => id */
$token = array(
        "COMMA" => 1,
        "CRLF" => 2,
        "DECIMAL" => 3,
/* id => name */
$tokenIdToName = array_flip($token);

$lex = new Lexer;
$lex->push("[\x2c]", $token["COMMA"]);
$lex->push("[\r][\n]", $token["CRLF"]);
$lex->push("[\d]+", $token["DECIMAL"]);

$in = "0,1,2\r\n3,42,5\r\n6,77,8\r\n";


do {
        $tok = $lex->getToken();

        if (Token::UNKNOWN == $tok->id) {
                throw new LexerException("Unknown token '{$tok->value}' at offset {$lex->marker}.");

        echo "TOKEN: ", $tokenIdToName[$tok->id], PHP_EOL;
} while (Token::EOI != $tok->id);

Example parsing comma separated number list

use Parle\Lexer;
use Parle\Parser;
use Parle\ParserException;

$p = new Parser;
$p->push("START", "RECORDS");
$prod_record_0 = $p->push("RECORDS", "RECORD CRLF");
$prod_record_1 = $p->push("RECORDS", "RECORDS RECORD CRLF");
$prod_int_0 = $p->push("RECORD", "INTEGER");
$prod_int_1 = $p->push("RECORD", "RECORD COMMA INTEGER");
$p->push("DECIMAL", "INTEGER COMMA INTEGER"); /* Production index unused. */
$prod_dec_0 = $p->push("RECORD", "'\"' DECIMAL '\"'");
$prod_dec_1 = $p->push("RECORD", "RECORD COMMA '\"' DECIMAL '\"'");

$lex = new Lexer;
$lex->push("[\x2c]", $p->tokenId("COMMA"));
$lex->push("[\r][\n]", $p->tokenId("CRLF"));
$lex->push("[\d]+", $p->tokenId("INTEGER"));
$lex->push("[\x22]", $p->tokenId("'\"'"));

/* Specifically using comma as both list separator and as a decimal mark. */
$in = "000,111,222\r\n\"333,3\",444,555\r\n666,777,\"888,8\"\r\n";

$p->consume($in, $lex);

do {
	switch ($p->action) {
		case Parser::ACTION_ERROR:
			$err = $p->errorInfo();
			if (Parser::ERROR_UNKNOWN_TOKEN == $err->id) {
				$tok = $err->token;
				$msg = "Unknown token '{$tok->value}' at offset {$err->position}";
			} else if (Parser::ERROR_NON_ASSOCIATIVE == $err->id) {
				$tok = $err->token;
				$msg = "Token '{$tok->id}' at offset {$lex->marker} is not associative";
			} else if (Parser::ERROR_SYNTAX == $err->id) {
				$tok = $err->token;
				$msg = "Syntax error at offset {$lex->marker}";
			} else {
				$msg = "Parse error";
			throw new ParserException($msg);
		case Parser::ACTION_SHIFT:
		case Parser::ACTION_GOTO:
		case Parser::ACTION_ACCEPT:
		case Parser::ACTION_REDUCE:
			switch ($p->reduceId) {
				case $prod_int_0:
					/* INTEGER */
					echo $p->sigil(), PHP_EOL;
				case $prod_int_1:
					echo $p->sigil(2), PHP_EOL;
				case $prod_dec_0:
					/* '\"' DECIMAL '\"' */
					echo $p->sigil(1), PHP_EOL;
				case $prod_dec_1:
					/* RECORD COMMA '\"' DECIMAL '\"' */
					echo $p->sigil(3), PHP_EOL;
				case $prod_record_0:
				case $prod_record_1:
					echo "=====", PHP_EOL;
} while (Parser::ACTION_ACCEPT != $p->action);
You can’t perform that action at this time.