misuzu/devel/MarkovDictionary.php

<?php
class MarkovDictionary {
    private const MAGIC = 'FMkD';
    private const VERSION = 1;

    private $handle;
    private int $segmentSize;
    private int $totalSegments;
    private int $startSegments;

    public function __construct(string $path) {
        if(!is_file($path))
            throw new InvalidArgumentException('$path does not exist.');

        $this->handle = $handle = fopen($path, 'rb');

        $magic = fread($handle, 4);
        if($magic !== self::MAGIC)
            throw new InvalidArgumentException('$path is not a valid markov dictionary.');

        $header = fread($handle, 12);
        if(strlen($header) !== 12)
            throw new InvalidArgumentException('$path is missing header data.');

        extract(unpack('Cversion/Cunused1/Cunused2/CsegmentSize/VtotalSegments/VstartSegments', $header));

        if($version !== self::VERSION)
            throw new InvalidArgumentException('$path version is incompatible.');

        $this->segmentSize = $segmentSize;
        $this->totalSegments = $totalSegments;
        $this->startSegments = $startSegments;
    }

    public function close(): void {
        if($this->handle !== null) {
            fclose($this->handle);
            $this->handle = null;
        }
    }

    public function __destruct() {
        $this->close();
    }

    private function reset(): void {
        fseek($this->handle, 16, SEEK_SET);
    }

    public function getStartPosition(): int {
        $randomStart = mt_rand(0, $this->startSegments) - 2;
        if($randomStart > 0) {
            for(;;) {
                fseek($this->handle, 4 * $this->segmentSize, SEEK_CUR);
                $isStart = fgetc($this->handle) !== "\0";

                if($isStart) {
                    if($randomStart < 1)
                        break;
                    --$randomStart;
                }

                extract(unpack('vnextSegments', fread($this->handle, 2)));

                fseek($this->handle, 6 * $nextSegments, SEEK_CUR);
            }

            fseek($this->handle, -(4 * $this->segmentSize) - 1, SEEK_CUR);
        }

        $startPos = ftell($this->handle);
        $this->reset();

        return $startPos;
    }

    public function generate(int $safety = 2000, int $start = -1): string {
        if($start < 0)
            $start = $this->getStartPosition();

        fseek($this->handle, $start, SEEK_SET);

        $string = '';

        for($s = 0; $s < $safety; ++$s) {
            $string .= fread($this->handle, 4 * $this->segmentSize);

            fseek($this->handle, 1, SEEK_CUR);

            extract(unpack('vnextSegments', fread($this->handle, 2)));

            if($nextSegments < 1)
                break;

            $nexts = [];

            // really shitty weighting system
            for($i = 0; $i < $nextSegments; ++$i) {
                extract(unpack('Voffset/vweight', fread($this->handle, 6)));

                for($j = 0; $j < $weight; ++$j)
                    $nexts[] = $offset;
            }

            $offset = $nexts[array_rand($nexts)];

            fseek($this->handle, $offset, SEEK_SET);
        }

        $this->reset();
        $string = mb_convert_encoding($string, 'utf-8', 'utf-32le');

        return trim($string);
    }
}