[Data Liberation] EBNF processor #1981
Description
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;
}
}