Table Of Contents


You can obtain free community support for example through stackoverflow.

If you think you found a bug, please create a ticket in the bug tracker.

Continuous Inspections

If you take code quality seriously, try out the new continuous inspection service.


Parser Library

This library allows you to easily implement recursive-descent parsers.


You can install this library through composer:

composer require jms/parser-lib

or add it to your composer.json file directly.


Let’s assume that you would like to write a parser for a calculator. For simplicity sake, we will assume that the parser would already return the result of the calculation. Inputs could look like this 1 + 1 and we would expect 2 as a result.

The first step, is to create a lexer which breaks the input string up into individual tokens which can then be consumed by the parser. This library provides a convenient class for simple problems which we will use:

$lexer = new \JMS\Parser\SimpleLexer(
        # Numbers

        # Do not surround with () because whitespace is not meaningful for
        # our purposes.

        # Operators; we support only + and -
    /x', // The x modifier tells PCRE to ignore whitespace in the regex above.

    // This maps token types to a human readable name.
    array(0 => 'T_UNKNOWN', 1 => 'T_INT', 2 => 'T_PLUS', 3 => 'T_MINUS'),

    // This function tells the lexer which type a token has. The first element is
    // an integer from the map above, the second element the normalized value.
    function($value) {
        if ('+' === $value) {
            return array(2, '+');
        if ('-' === $value) {
            return array(3, '-');
        if (is_numeric($value)) {
            return array(1, (integer) $value);

        return array(0, $value);

Now the second step, is to create the parser which can consume the tokens once the lexer has split them:

class MyParser extends \JMS\Parser\AbstractParser
    const T_UNKNOWN = 0;
    const T_INT = 1;
    const T_PLUS = 2;
    const T_MINUS = 3;

    public function parseInternal()
        $result = $this->match(self::T_INT);

        while ($this->lexer->isNextAny(array(self::T_PLUS, self::T_MINUS))) {
            if ($this->lexer->isNext(self::T_PLUS)) {
                $result += $this->match(self::T_INT);
            } else if ($this->lexer->isNext(self::T_MINUS)) {
                $result -= $this->match(self::T_INT);
            } else {
                throw new \LogicException('Previous ifs were exhaustive.');

        return $result;

$parser = new MyParser($lexer);
$parser->parse('1 + 1'); // int(2)
$parser->parse('5 + 10 - 4'); // int(11)

That’s it. Now you can perform basic operations already. If you like you can now also replace the hard-coded integers in the lexer with the class constants of the parser.


The code is released under the business-friendly Apache2 license.

Documentation is subject to the Attribution-NonCommercial-NoDerivs 3.0 Unported license.