patchouli/lib/FWIF/FWIF.php

<?php
namespace FWIF;

use DateInterval;
use DateTime;
use DateTimeInterface;
use DateTimeImmutable;
use DateTimeZone;
use InvalidArgumentException;

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

    public const DEFAULT = 0;
    public const DISCARD_MILLISECONDS = 0x01; // Always exclude the millisecond component from DateTime
    public const EXCLUDE_VERSION = 0x02; // Exclude version byte at the start of the stream

    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 TRAILER
    public const TYPE_BUFFER    = 0x04; // Buffer with binary data, prefixed with a LEB128 length
    public const TYPE_ARRAY     = 0x05; // List of values, terminated with TRAILER
    public const TYPE_OBJECT    = 0x06; // List of values with ASCII names, terminated with TRAILER
    public const TYPE_DATETIME  = 0x07; // A gregorian year, month and day as well as an hour, minute, seconds and millisecond component, variable ranging from 4 to 7 bytes
    
    public const TRAILER = 0xFF; // Termination byte

    public const VERSION = 0x01; // min 1, max 254

    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',
        self::TYPE_BUFFER => 'Buffer',
        self::TYPE_DATETIME => 'DateTime',
    ];

    private const UTF8 = '%^(?:' // https://www.w3.org/International/questions/qa-forms-utf-8.en
                        . '[\x09\x0A\x0D\x20-\x7E]'             // ASCII
                        . '|[\xC2-\xDF][\x80-\xBF]'             // non-overlong 2-byte
                        . '|\xE0[\xA0-\xBF][\x80-\xBF]'         // excluding overlongs
                        . '|[\xE1-\xEC\xEE\xEF][\x80-\xBF]{2}'  // straight 3-byte
                        . '|\xED[\x80-\x9F][\x80-\xBF]'         // excluding surrogates
                        . '|\xF0[\x90-\xBF][\x80-\xBF]{2}'      // planes 1-3
                        . '|[\xF1-\xF3][\x80-\xBF]{3}'          // planes 4-15
                        . '|\xF4[\x80-\x8F][\x80-\xBF]{2}'      // plane 16 
                        . ')*$%xs';

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

    // apparently this is faster than mb_check_encoding($string, 'utf-8');
    // on PHP 7.1 on Windows at least, perhaps investigate this later
    // UPDATE TODO: does this even make any sense with other internal encodings?
    private static function isUTF8String(string $string): bool {
        return preg_match(self::UTF8, $string) === 1;
    }

    private static function detectType($data, int $flags): int {
        if(is_null($data))
            return self::TYPE_NULL;
        if(is_int($data))
            return self::TYPE_INTEGER;
        if(is_float($data))
            return self::TYPE_FLOAT;
        if(is_string($data))
            return self::isUTF8String($data) ? self::TYPE_STRING : self::TYPE_BUFFER;
        if(is_object($data) || self::isAssocArray($data)) {
            if($data instanceof DateTimeInterface)
                return self::TYPE_DATETIME;
            return self::TYPE_OBJECT;
        }
        if(is_array($data))
            return self::TYPE_ARRAY;
        throw new FWIFUnsupportedTypeException(gettype($data));
    }

    public static function encode($data, int $flags = self::DEFAULT): string {
        $encoded = self::encodeInternal($data, $flags);
        if(!($flags & self::EXCLUDE_VERSION))
            $encoded = chr(self::VERSION) . $encoded;
        return $encoded;
    }
    private static function encodeInternal($data, int $flags): string {
        if($data instanceof FWIFSerializable)
            $data = $data->fwifSerialize();
        $type = self::detectType($data, $flags);
        return chr($type) . self::{'encode' . self::CODECS[$type]}($data, $flags);
    }

    public static function decode($data, int $flags = self::DEFAULT) {
        if(is_string($data)) {
            $fd = fopen('php://memory', 'rb+');
            fwrite($fd, $data);
            fseek($fd, 0);
            $data = $fd;
        }
        if(!is_resource($data))
            throw new InvalidArgumentException('$data must be either a string or a file handle.');
        if(!($flags & self::EXCLUDE_VERSION)) {
            $version = ord(fgetc($data));
            if($version < 1 || $version > 254)
                throw new InvalidArgumentException('$data is not a valid FWIF serialized stream.');
            if($version > self::VERSION)
                throw new FWIFUnsupportedVersionException;
        }
        $decoded = self::decodeInternal($data, $flags);
        if(isset($fd))
            fclose($fd);
        return $decoded;
    }
    private static function decodeInternal($data, int $flags) {
        $type = ord(fgetc($data));
        if(!array_key_exists($type, self::CODECS)) {
            $hexType = dechex($type); $pos = ftell($data); $hexPos = dechex($pos);
            throw new FWIFUnsupportedTypeException("Unsupported type {$type} (0x{$hexType}) at position {$pos} (0x{$hexPos})");
        }
        return self::{'decode' . self::CODECS[$type]}($data, $flags);
    }

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

    private static function encodeInteger(int $number, int $flags): 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 $packed;
    }
    private static function decodeInteger($data, int $flags): int {
        $number = 0; $shift = 0; $o = 0; $size = PHP_INT_SIZE * 8;
        do {
            $byte = ord(fgetc($data));
            $number |= ($byte & 0x7F) << $shift;
            $shift += 7;
        } while($byte & 0x80);
        if(($shift < $size) && ($byte & 0x40))
            $number |= (~0 << $shift);
        return $number;
    }

    // I still don't like these
    private static function encodeFloat(float $number, int $flags): string {
        return pack('E', $number);
    }
    private static function decodeFloat($data, int $flags): float {
        return unpack('E', fread($data, 8))[1];
    }

    private static function encodeString(string $string, int $flags): string {
        $packed = '';
        $string = unpack('C*', mb_convert_encoding($string, 'utf-8', mb_internal_encoding()));
        foreach($string as $char)
            $packed .= chr($char);
        return $packed . chr(self::TRAILER);
    }
    private static function decodeString($data, int $flags): string {
        $packed = '';
        for(;;) {
            $char = fgetc($data); $byte = ord($char);
            if($byte == self::TRAILER)
                break;
            $packed .= $char;
            if(($byte & 0xF8) == 0xF0)
                $packed .= fread($data, 3);
            elseif(($byte & 0xF0) == 0xE0)
                $packed .= fread($data, 2);
            elseif(($byte & 0xE0) == 0xC0)
                $packed .= fgetc($data);
        }
        return mb_convert_encoding($packed, mb_internal_encoding(), 'utf-8');
    }

    private static function encodeArray(array $array, int $flags): string {
        $packed = '';
        foreach($array as $value)
            $packed .= self::encodeInternal($value, $flags);
        return $packed . chr(self::TRAILER);
    }
    private static function decodeArray($data, int $flags): array {
        $array = [];
        for(;;) {
            if(ord(fgetc($data)) === self::TRAILER)
                break;
            fseek($data, -1, SEEK_CUR);
            $array[] = self::decodeInternal($data, $flags);
        }
        return $array;
    }

    private static function encodeObject($object, int $flags): string {
        $packed = ''; $array = (array)$object;
        foreach($array as $name => $value)
            $packed .= mb_convert_encoding($name, 'us-ascii', mb_internal_encoding()) . chr(self::TRAILER) . self::encodeInternal($value, $flags);
        return $packed . chr(self::TRAILER);
    }
    private static function decodeObjectKey($data, int $flags): string {
        $packed = '';
        for(;;) {
            $char = fgetc($data);
            if(ord($char) === self::TRAILER)
                break;
            $packed .= $char;
        }
        return mb_convert_encoding($packed, mb_internal_encoding(), 'us-ascii');
    }
    private static function decodeObject($data, int $flags): object {
        $array = [];
        for(;;) {
            if(ord(fgetc($data)) === self::TRAILER)
                break;
            fseek($data, -1, SEEK_CUR);
            $array[self::decodeObjectKey($data, $flags)] = self::decodeInternal($data, $flags);
        }
        return (object)$array;
    }

    private static function encodeBuffer(string $buffer, int $flags): string {
        return self::encodeInteger(strlen($buffer), $flags) . $buffer;
    }
    private static function decodeBuffer($data, int $flags): string {
        return fread($data, self::decodeInteger($data, $flags));
    }

    private const DATETIME_FLAG_TIME   = 0x40;
    private const DATETIME_FLAG_MILLI  = 0x4000;

    private const DATETIME_YEAR_SIGN  = 0x40000000;
    private const DATETIME_YEAR_MASK  = 0x3FFF;
    private const DATETIME_YEAR_SHIFT = 16; // <<

    private const DATETIME_MONTH_MASK  = 0x0F;
    private const DATETIME_MONTH_SHIFT = 12; // <<

    private const DATETIME_DAY_MASK = 0x1F;
    private const DATETIME_DAY_SHIFT = 7; // <<

    private const DATETIME_HOUR_MASK  = 0x1F;

    private const DATETIME_MINS_MASK  = 0x3F;
    private const DATETIME_MINS_SHIFT = 8; // <<

    private const DATETIME_SECS_MASK = 0x3F;
    private const DATETIME_SECS_SHIFT = 2; // <<

    private const DATETIME_MILLI_HI_MASK  = 0x300;
    private const DATETIME_MILLI_HI_SHIFT = 8; // >>
    private const DATETIME_MILLI_LO_MASK  = 0x0FF;

    /* +--------+--------+
     * |.YYYYYYY|YYYYYYYY|
     * |MMMMDDDD|DT.HHHHH|
     * |.Wmmmmmm|SSSSSSww|
     * |wwwwwwww|        |
     * +--------+--------+
     */

    private static function encodeDateTime(DateTimeInterface $dt, int $flags): string {
        static $utc = null;
        if($utc === null)
            $utc = new DateTimeZone('utc');

        if($dt->getTimezone()->getOffset($dt) !== 0)
            $dt = DateTime::createFromInterface($dt)->setTimezone($utc);

        $year  = (int)$dt->format('Y');
        $month = (int)$dt->format('n');
        $day   = (int)$dt->format('j');
        $hours = (int)$dt->format('G');

        $subYear = $year < 0;
        if($subYear)
            $year = ~$year;
        
        $ymdfh  = $subYear ? self::DATETIME_YEAR_SIGN : 0;
        $ymdfh |= ($year  & self::DATETIME_YEAR_MASK)  << self::DATETIME_YEAR_SHIFT;
        $ymdfh |= ($month & self::DATETIME_MONTH_MASK) << self::DATETIME_MONTH_SHIFT;
        $ymdfh |= ($day   & self::DATETIME_DAY_MASK)   << self::DATETIME_DAY_SHIFT;
        $ymdfh |= ($hours & self::DATETIME_HOUR_MASK);

        $mins   = (int)$dt->format('i');
        $secs   = (int)$dt->format('s');
        $millis = ($flags & self::DISCARD_MILLISECONDS) ? 0 : (int)$dt->format('v');

        if($mins > 0 || $secs > 0 || $millis > 0) {
            $ymdfh |= self::DATETIME_FLAG_TIME;
            $msw    = 0;
            $msw   |= ($mins  & self::DATETIME_MINS_MASK) << self::DATETIME_MINS_SHIFT;
            $msw   |= ($secs  & self::DATETIME_SECS_MASK) << self::DATETIME_SECS_SHIFT;

            if($millis > 0) {
                $msw |=  self::DATETIME_FLAG_MILLI;
                $msw |= ($millis & self::DATETIME_MILLI_HI_MASK) >> self::DATETIME_MILLI_HI_SHIFT;
                $w    =  $millis & self::DATETIME_MILLI_LO_MASK;
            }
        }

        $packed = pack('N', $ymdfh);
        if($ymdfh & self::DATETIME_FLAG_TIME) {
            $packed .= pack('n', $msw);
            if($msw & self::DATETIME_FLAG_MILLI)
                $packed .= chr($w);
        }

        return $packed;
    }
    private static function decodeDateTime($data, int $flags): DateTimeInterface {
        $ymdfh  = unpack('N', fread($data, 4))[1];
        $hasMsw = $ymdfh & self::DATETIME_FLAG_TIME;
        $msw    = $hasMsw ? unpack('n', fread($data, 2))[1] : 0;
        $hasW   = $hasMsw && ($msw & self::DATETIME_FLAG_MILLI);
        $w      = $hasW   ? ord(fgetc($data)) : 0;

        $year  = ($ymdfh >> self::DATETIME_YEAR_SHIFT)  & self::DATETIME_YEAR_MASK;
        $month = ($ymdfh >> self::DATETIME_MONTH_SHIFT) & self::DATETIME_MONTH_MASK;
        $day   = ($ymdfh >> self::DATETIME_DAY_SHIFT)   & self::DATETIME_DAY_MASK;
        $hour  =  $ymdfh                                & self::DATETIME_HOUR_MASK;

        if($ymdfh & self::DATETIME_YEAR_SIGN)
            $year = ~$year;
        $dt = sprintf('%04d-%02d-%02dT%02d:', $year, $month, $day, $hour);

        if($hasMsw) {
            $mins = ($msw >> self::DATETIME_MINS_SHIFT) & self::DATETIME_MINS_MASK;
            $secs = ($msw >> self::DATETIME_SECS_SHIFT) & self::DATETIME_SECS_MASK;
            $dt  .= sprintf('%02d:%02d', $mins, $secs);
            if($hasV) {
                $millis  = ($msw << self::DATETIME_MILLI_HI_SHIFT) & self::DATETIME_MILLI_HI_MASK;
                $millis |=  $w;
                $dt .= sprintf('.%03d', $millis);
            }
        } else $dt .= '00:00';

        return new DateTimeImmutable($dt);
    }
}