rag/app/Helpers/ChunkParser.php

<?php
declare(strict_types=1);

namespace App\Helpers;

use App\Enums\Status;

final class ChunkParser
{
    public string $chunk = '';
    private string $buffer = '';

    private array $results = [];

    public function __construct(private readonly array $tags = ['think', 'content', 'fields'], private readonly string $name = 'inference_rag_')
    {
        foreach ($this->tags as $tag) {
            $this->results[$tag] = null;
        }
    }

    public function append(string $chunk): self
    {
        $this->chunk = $chunk;
        $this->buffer .= $chunk;
        return $this;
    }

    public function parse(): self
    {
        if ($this->buffer === '') {
            return $this;
        }

        foreach ($this->tags as $tag) {
            $positions = $this->positions($this->buffer, $tag);

            if (count($positions) >= 1) {
                $length = strlen('<|' . $this->name . $tag . '|>');
                $content = substr($this->buffer, $positions[0] + $length, ($positions[1] ?? null) - $positions[0] - $length);

                $this->results[$tag] = $content;
            } else {
                $this->results[$tag] = null;
            }
        }

        return $this;
    }

    private function positions(string $buffer, string $tag): array
    {
        $pattern = "/<\\|" . $this->name . $tag . "\\|>/";
        $matches = [];
        preg_match_all($pattern, $buffer, $matches, PREG_OFFSET_CAPTURE);

        return !empty($matches[0]) ? array_column($matches[0], 1) : [];
    }

    public function all(): array
    {
        return $this->results;
    }

    public function toArray(): array
    {
        return ['think' => $this->think, 'content' => $this->content, 'fields' => $this->fields, 'status' => $this->getStatus()];
    }

    private function getStatus(): ?Status
    {
        if ($this->think && !$this->content) {
            return Status::Thinking;
        }

        if ($this->content) {
            return Status::Typing;
        }

        return null;
    }

    public function __get(string $tag)
    {
        if($tag === 'status') {
            return $this->getStatus();
        }

        $string = $this->results[$tag] ?? null;
        if ($tag === 'fields') {
            try {
                return json_decode($string, true);
            } catch (\Throwable) {
                return [];
            }
        }

        return $string;
    }
}