Skip to content
This repository has been archived by the owner on May 20, 2019. It is now read-only.
/ lexer Public archive

[DEPRECATED] Please use phplrt/lexer instead

License

Notifications You must be signed in to change notification settings

railt/lexer

Repository files navigation

Railt

Travis CI

PHP 7.1+ railt.org Discord Latest Stable Version Total Downloads License MIT

Lexer

Note: All questions and issues please send to https://github.com/railt/railt/issues

Note: Tests can not always pass correctly. This may be due to the inaccessibility of PPA servers for updating gcc and g++. The lexertl build requires the support of a modern compiler inside Travis CI. In this case, a gray badge will be displayed with the message "Build Error".

In order to quickly understand how it works - just write ~4 lines of code:

$lexer = Railt\Component\Lexer\Factory::create(['T_WHITESPACE' => '\s+', 'T_DIGIT' => '\d+'], ['T_WHITESPACE']);

foreach ($lexer->lex(Railt\Component\Io\File::fromSources('23 42')) as $token) {
    echo $token . "\n";
}

This example will read the source text and return the set of tokens from which it is composed:

  1. T_DIGIT with value "23"
  2. T_DIGIT with value "42"

The second argument to the Factory class is the list of token names that are ignored in the lex method result. That's why we only got two significant tokens T_DIGIT. Although this is not entirely true, the answer contains a T_EOI (End Of Input) token which can also be removed from the output by adding an array of the second argument of Factory class.

...and now let's try to understand more!

The lexer contains two types of runtime:

  1. Basic - Set of algorithms with one state.
  2. Multistate - Set of algorithms with the possibility of state transition between tokens.

In connection with the fact that there were almost no differences in speed between several implementations (Stateful vs Stateless) of the same algorithm, it was decided to abandon the immutable stateful lexers.

use Railt\Component\Lexer\Factory;

/**
 * List of available tokens in format "name => pcre"
 */
$tokens = ['T_DIGIT' => '\d+', 'T_WHITESPACE' => '\s+'];

/**
 * List of skipped tokens
 */
$skip   = ['T_WHITESPACE'];

/**
 * Options:
 *   0 - Nothing.
 *   2 - With PCRE lookahead support.
 *   4 - With multistate support.
 */
$flags = Factory::LOOKAHEAD | Factory::MULTISTATE;

/**
 * Create lexer and tokenize sources. 
 */
$lexer = Factory::create($tokens, $skip, $flags);

In order to tokenize the source text, you must use the method ->lex(...), which returns iterator of the TokenInterface objects.

foreach ($lexer->lex(File::fromSources('23 42')) as $token) {
    echo $token . "\n";
}

A TokenInterface provides a convenient API to obtain information about a token:

interface TokenInterface
{
    public function getName(): string;
    public function getOffset(): int;
    public function getValue(int $group = 0): ?string;
    public function getGroups(): iterable;
    public function getBytes(): int;
    public function getLength(): int;
}

Drivers

The factory returns one of the available implementations, however you can create it yourself.

Basic

NativeRegex

NativeRegex implementation is based on the built-in php PCRE functions.

use Railt\Component\Lexer\Driver\NativeRegex;
use Railt\Component\Io\File;

$lexer = new NativeRegex(['T_WHITESPACE' => '\s+', 'T_DIGIT' => '\d+'], ['T_WHITESPACE', 'T_EOI']);

foreach ($lexer->lex(File::fromSources('23 42')) as $token) {
    echo $token->getName() . ' -> ' . $token->getValue() . ' at ' . $token->getOffset() . "\n";
}

// Outputs:
// T_DIGIT -> 23 at 0
// T_DIGIT -> 42 at 3

Lexertl

Experimental lexer based on the C++ lexertl library. To use it, you need support for Parle extension.

use Railt\Component\Lexer\Driver\ParleLexer;
use Railt\Component\Io\File;

$lexer = new ParleLexer(['T_WHITESPACE' => '\s+', 'T_DIGIT' => '\d+'], ['T_WHITESPACE', 'T_EOI']);

foreach ($lexer->lex(File::fromSources('23 42')) as $token) {
    echo $token->getName() . ' -> ' . $token->getValue() . ' at ' . $token->getOffset() . "\n";
}

// Outputs:
// T_DIGIT -> 23 at 0
// T_DIGIT -> 42 at 3

Be careful: The library is not fully compatible with the PCRE regex syntax. See the official documentation.

Multistate

This functionality is not yet implemented.