Simple LR(1) parser generator. Generally, build a parser by setting terminals and rules, then calling processGrammar(). For example, here is a simple grammar which accepts one or more "a" followed by exactly one "b":

$parser = id(new PhutilParserGenerator())
  ->setTerminals(array('a', 'b'))
  ->setStartRule('S')
  ->setRules(
    array(
      'S' => 'A b',
      'A' => array(
        'A a',
        'a',
      )))
  ->processGrammar();

To actually parse token streams, use parseTokens().

$tokens = get_tokens(); // Usually from PhutilLexer
$callback = 'some_callback';
$tree = $parser->parseTokens($tokens, $callback);

The callback is invoked when a grammar rule matches. It should have this signature:

function parser_callback($rule, $production, array $tokens) {
  // ...
}

The $rule is the matching rule; the $production is the matching production, and $tokens is the matching tokens (for terminal rules) or the return value of previous parse callbacks (for nonterminal rules).

You should either return a result of evaluation, or some sort of abstract representation of the parse tree (this is more likely to be useful for more complex grammars).