[Data Liberation] EBNF processor

I started drafting an EBNF notation processor to potentially enable building parsers based on a grammar file, similarly to the MySQL parser (although that one is based on ANTLR grammar). I don't have any specific action for this issue, I only wanted to dump the code somewhere it would be searchable in the future. CC @janjakes

<?php

class EBNF_Processor {
    private $ebnf;
    private $bytes_already_parsed = 0;
    private $last_error;

    private $rule_name;
    private $rules = [];

    private $state = self::STATE_READY;

    const STATE_READY = 0;
    const STATE_FINISHED = 1;

    public function __construct($ebnf) {
        $this->ebnf = $ebnf;
    }

    public function get_last_error() {
        return $this->last_error;
    }

    public function is_finished() {
        return $this->state === self::STATE_FINISHED;
    }

    public function next_rule() {
        if($this->is_finished() || $this->get_last_error()) {
            return false;
        }

        while(true) {
            if(false === $this->skip_comments()) {
                return false;
            }
            if(false === $this->parse_rule_name()) {
                return false;
            }
         
            die();
        }

        $this->state = self::STATE_FINISHED;
        return false;
    }

    private function parse_branches() {
        $at = $this->bytes_already_parsed;
        if($at >= strlen($this->ebnf)) {
            $this->last_error = 'Unexpected end of input';
            return false;
        }
        switch($this->ebnf[$at]) {
            case '(':
                break;
            case '"':
            case "'":
                break;
        }
    }

    private function skip_whitespace($at) {
        return strspn($this->ebnf, " \t", $at);
    }

    private function skip_comments() {
        while(true) {
            $at = $this->bytes_already_parsed;
            if($at + 2 >= strlen($this->ebnf)) {
                return false;
            }

            if(
                $this->ebnf[$at] === "/" &&
                $this->ebnf[$at + 1] === "/"
            ) {
                // Skip comment
                $this->bytes_already_parsed = strpos($this->ebnf, "\n", $at) + 1;
            } else {
                break;
            }
        }
    }

    private function parse_rule_name() {
        $at = $this->bytes_already_parsed;
        // Skip whitespace at the beginning of the rule
        $at += strspn($this->ebnf, " \t", $at);

        $rule_name_length = strspn($this->ebnf, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789", $at);
        if( 0 === $rule_name_length ) {
            $this->last_error = 'Invalid rule name';
            return false;
        }
        $this->rule_name = substr($this->ebnf, $at, $rule_name_length);
        $at += $rule_name_length;
        $at += $this->skip_whitespace($at);

        // Skip past the '::='
        if( strlen($this->ebnf) <= $at + 3 ) {
            $this->last_error = 'Unexpected end of input';
            return false;
        }
        if( $this->ebnf[$at] !== ':' || $this->ebnf[$at + 1] !== ':' || $this->ebnf[$at + 2] !== '=' ) {
            $this->last_error = 'Expected "::="';
            return false;
        }
        $at += 3;

        $this->bytes_already_parsed = $at;
    }
}

The issue contains only an unfinished EBNF_Processor draft and names no repository file, test, or entry point. Review the draft and the linked MySQL parser context first; work cannot be considered done until the project defines a concrete scope and acceptance criteria.