PHP Unserializer

Unserializer.php

<?php

/*
 * Custom PHP unserializer with better error reporting
 *
 * Does not support objects.
 */
class Unserializer {
	private $serialized;
	private $len;
	private $i;
	
	public function __construct($serialized) {
		$this->serialized = $serialized;
		$this->len = strlen($this->serialized);
		$this->i = 0;
	}
	
	private function read() {
		if ($this->i >= $this->len) {
			$this->err("Unexpected end");
		}
		
		$b = substr($this->serialized, $this->i, 1);
		$this->i++;
		return $b;
	}
	
	private function err($msg) {
		$offset = $this->i - 1;
		throw new Exception("{$msg} at {$offset}");
	}
	
	private function arr() {
		$len = $this->num(true);
		$this->byte(':');
		$this->byte('{');
		$result = array();
		while ($len-- > 0) {
			$key = $this->any();
			if (!is_scalar($key)) {
				$this->err("Non-scalar key");
			}
			if (isset($result[$key])) {
				$this->err("Duplicate key '{$key}'");
			}
			$value = $this->any();
			$result[$key] = $value;
		}
		$this->byte('}');
		return $result;
	}
	
	private function num($int_only = false) {
		$result = 0;
		$b = $this->byte('0123456789');
		$result *= 10;
		$result += $b;
		while (true) {
			$b = $this->byte($int_only ? '0123456789' : '0123456789.', false);
			if ($b === null) {
				return $result;
			}
			if ($b === '.') {
				break;
			}
			$result *= 10;
			$result += $b;
		}
		$denom = 1;
		while (true) {
			$b = $this->byte('0123456789', false);
			if ($b === null) {
				return $result;
			}
			$denom *= 10;
			$result += $b / $denom;
		}
	}
	
	private function bool() {
		$b = $this->byte('01');
		if ($b == '1') {
			$result = true;
		} else if ($b == '0') {
			$result = false;
		}
		$this->byte(';');
		return $result;
	}
	
	private function str() {
		$len = $this->num(true);
		$this->byte(':');
		$this->byte('"');
		$result = '';
		while ($len-- > 0) {
			$result .= $this->read();
		}
		$this->byte('"');
		$this->byte(';');
		return $result;
	}
	
	private function byte($exp, $err = true) {
		$exp = str_split($exp);
		$b = $this->read();
		if (!in_array($b, $exp, true)) {
			if ($err) {
				$exp = implode(' or ', array_map(function($c) { return "'{$c}'"; }, $exp));
				$this->err("Unexpected '{$b}' (expecting {$exp})");
			} else {
				$this->i--;
				return null;
			}
		}
		return $b;
	}
	
	private function any() {
		$b = $this->byte('bidsaN');
		if (in_array($b, array('b', 'i', 'd', 's', 'a'))) {
			$this->byte(':');
			if ($b === 'b') {
				$result = $this->bool();
			} else if ($b === 'i' || $b === 'd') {
				$result = $this->num();
				$this->byte(';');
			} else if ($b === 's') {
				$result = $this->str();
			} else if ($b === 'a') {
				$result = $this->arr();
			}
		} else if ($b === 'N') {
			$result = null;
			$this->byte(';');
		}
		return $result;
	}
	
	public function unserialize() {
		$result = $this->any();
		if ($this->len !== $this->i) {
			$this->err("Unexpected more bytes");
		}
		return $result;
	}
}