patchouli/lib/FWIF/FWIF.php

<?php
namespace FWIF;

class FWIF {
    public const CONTENT_TYPE = 'text/plain; charset=us-ascii'; // TODO: come up with a mime type

    public const TYPE_NULL      = 0;    // NULL, no data
    public const TYPE_INTEGER   = 0x01; // LEB128, implicit length
    public const TYPE_FLOAT     = 0x02; // double precision IEEE 754, fixed length of 8 bytes
    public const TYPE_STRING    = 0x03; // UTF-8 string, terminated with TYPE_TRAILER
    public const TYPE_ARRAY     = 0x04; // List of values, terminated with TYPE_TRAILER
    public const TYPE_OBJECT    = 0x05; // List of values with ASCII names, terminated with TYPE_TRAILER
    public const TYPE_BUFFER    = 0x06; // Buffer with binary data, prefixed with a LEB128 length
    public const TYPE_DATE      = 0x07; // A gregorian year, month and day, fixed length of * bytes
    public const TYPE_DATETIME  = 0x08; // A gregorian year, month and day as well as an hour, minute and seconds component, fixed length of * bytes
    public const TYPE_PERIOD    = 0x09; // A time period, fixed length of * bytes
    public const TYPE_TRAILER   = 0xFF; // Termination byte

    private const CODECS = [
        self::TYPE_NULL => 'Null',
        self::TYPE_INTEGER => 'Integer',
        self::TYPE_FLOAT => 'Float',
        self::TYPE_STRING => 'String',
        self::TYPE_ARRAY => 'Array',
        self::TYPE_OBJECT => 'Object',
    ];

    private static function isAssocArray($array): bool {
        if(!is_array($array) || $array === [])
            return false;
        return array_keys($array) !== range(0, count($array) - 1);
    }

    private static function detectType($data): int {
        if(is_null($data))
            return self::TYPE_NULL;
        if(is_int($data))
            return self::TYPE_INTEGER;
        if(is_float($data))
            return self::TYPE_FLOAT;
        if(is_string($data)) // Should this check if a string is valid UTF-8 and swap over to TYPE_BUFFER?
            return self::TYPE_STRING;
        if(is_object($data) || self::isAssocArray($data))
            return self::TYPE_OBJECT;
        if(is_array($data))
            return self::TYPE_ARRAY;
        throw new FWIFUnsupportedTypeException(gettype($data));
    }

    public static function encode($data): string {
        if($data instanceof FWIFSerializable)
            $data = $data->fwifSerialize();
        $type = self::detectType($data);
        return chr($type) . self::{'encode' . self::CODECS[$type]}($data);
    }

    public static function decode(string $data) {
        return self::decodeInternal(new FWIFDecodeStream($data));
    }

    private static function decodeInternal(FWIFDecodeStream $data) {
        $type = $data->readByte();
        if(!array_key_exists($type, self::CODECS)) {
            $hexType = dechex($type); $hexPos = dechex($data->getPosition());
            throw new FWIFUnsupportedTypeException("Unsupported type {$type} (0x{$hexType}) at position {$data->getPosition()} (0x{$hexPos})");
        }
        return self::{'decode' . self::CODECS[$type]}($data);
    }

    private static function encodeNull($data): string { return ''; }
    private static function decodeNull(FWIFDecodeStream $data) { return null; }

    private static function encodeInteger(int $number): string {
        $packed = ''; $more = 1; $negative = $number < 0; $size = PHP_INT_SIZE * 8;
        while($more) {
            $byte = $number & 0x7F;
            $number >>= 7;
            if($negative)
                $number |= (~0 << ($size - 7));
            if((!$number && !($byte & 0x40)) || ($number === -1 && ($byte & 0x40)))
                $more = 0;
            else
                $byte |= 0x80;
            $packed .= chr($byte);
        }
        return $packed;
    }
    private static function decodeInteger(FWIFDecodeStream $data): int {
        $number = 0; $shift = 0; $o = 0; $size = PHP_INT_SIZE * 8;
        do {
            $byte = $data->readByte();
            $number |= ($byte & 0x7F) << $shift;
            $shift += 7;
        } while($byte & 0x80);
        if(($shift < $size) && ($byte & 0x40))
            $number |= (~0 << $shift);
        return $number;
    }

    private static function encodeFloat(float $number): string {
        return pack('E', $number);
    }
    private static function decodeFloat(FWIFDecodeStream $data): float {
        $packed = ''; for($i = 0; $i < 8; ++$i) $packed .= chr($data->readByte());
        return unpack('E', $packed)[1];
    }

    private static function encodeString(string $string): string {
        $packed = ''; $string = unpack('C*', mb_convert_encoding($string, 'utf-8'));
        foreach($string as $char)
            $packed .= chr($char);
        return $packed . chr(self::TYPE_TRAILER);
    }
    private static function decodeAsciiString(FWIFDecodeStream $data): string {
        $string = '';
        for(;;) {
            $byte = $data->readByte();
            if($byte === self::TYPE_TRAILER)
                break;
            $string .= chr($byte);
        }
        return $string;
    }
    private static function decodeString(FWIFDecodeStream $data): string {   // This should decode based on the utf-8 spec rather than just
        return mb_convert_encoding(self::decodeAsciiString($data), 'utf-8'); //  grabbing the FF terminated string representation.
    }

    private static function encodeArray(array $array): string {
        $packed = '';
        foreach($array as $value)
            $packed .= self::encode($value);
        return $packed . chr(self::TYPE_TRAILER);
    }
    private static function decodeArray(FWIFDecodeStream $data): array {
        $array = [];
        for(;;) {
            if($data->readByte() === self::TYPE_TRAILER)
                break;
            $data->stepBack();
            $array[] = self::decodeInternal($data);
        }
        return $array;
    }

    private static function encodeObject($object): string {
        $packed = ''; $array = (array)$object;
        foreach($array as $name => $value)
            $packed .= $name . chr(self::TYPE_TRAILER) . self::encode($value);
        return $packed . chr(self::TYPE_TRAILER);
    }
    private static function decodeObject(FWIFDecodeStream $data): object {
        $array = [];
        for(;;) {
            if($data->readByte() === self::TYPE_TRAILER)
                break;
            $data->stepBack();
            $name = self::decodeAsciiString($data);
            $array[$name] = self::decodeInternal($data);
        }
        return (object)$array;
    }
}