AndyTranscribe/app/Events/RecordingTranscriptionUpdated.php

<?php

namespace App\Events;

use App\Models\Recording;
use Illuminate\Broadcasting\InteractsWithSockets;
use Illuminate\Broadcasting\PrivateChannel;
use Illuminate\Contracts\Broadcasting\ShouldBroadcastNow;
use Illuminate\Foundation\Events\Dispatchable;
use Illuminate\Queue\SerializesModels;

class RecordingTranscriptionUpdated implements ShouldBroadcastNow
{
    use Dispatchable, InteractsWithSockets, SerializesModels;

    /**
     * Reverb's default max payload is 10KB. Keep streamed text well under that.
     */
    public const MAX_DELTA_BYTES = 8_000;

    /**
     * Create a new event instance.
     *
     * @param  array<string, mixed>|null  $whisperDelta
     */
    public function __construct(
        public Recording $recording,
        public ?string $transcriptDelta = null,
        public bool $transcriptReplace = false,
        public ?array $whisperDelta = null,
    ) {}

    /**
     * Get the channels the event should broadcast on.
     *
     * @return array<int, PrivateChannel>
     */
    public function broadcastOn(): array
    {
        return [
            new PrivateChannel('recording.'.$this->recording->id),
            new PrivateChannel('user.'.$this->recording->user_id.'.recordings'),
        ];
    }

    public function broadcastAs(): string
    {
        return 'RecordingTranscriptionUpdated';
    }

    /**
     * @return array<string, mixed>
     */
    public function broadcastWith(): array
    {
        $payload = array_merge(
            $this->recording->transcriptionStatusPayload(),
            [
                'word_count' => $this->recording->word_count,
            ],
        );

        unset($payload['transcript'], $payload['whisper']);

        $delta = $this->transcriptDelta;

        if ($delta !== null && $delta !== '' && strlen($delta) <= self::MAX_DELTA_BYTES) {
            $payload['transcript_delta'] = $delta;
            $payload['transcript_replace'] = $this->transcriptReplace;
        }

        $whisper = $this->compactWhisperDelta($this->whisperDelta);

        if ($whisper !== null) {
            $payload['whisper_delta'] = $whisper;
        }

        return $payload;
    }

    /**
     * @param  array<string, mixed>|null  $whisper
     * @return array<string, mixed>|null
     */
    private function compactWhisperDelta(?array $whisper): ?array
    {
        if ($whisper === null || $whisper === []) {
            return null;
        }

        // Accumulated segment lists belong in HTTP hydrate, not on Reverb.
        unset($whisper['segments']);

        if ($whisper === []) {
            return null;
        }

        foreach (['tokens', 'logprobs', 'words'] as $drop) {
            $encoded = json_encode($whisper);

            if (! is_string($encoded) || strlen($encoded) <= self::MAX_DELTA_BYTES) {
                return $whisper;
            }

            $whisper = $this->dropWhisperField($whisper, $drop);
        }

        $encoded = json_encode($whisper);

        if (! is_string($encoded) || strlen($encoded) > self::MAX_DELTA_BYTES) {
            return null;
        }

        return $whisper;
    }

    /**
     * @param  array<string, mixed>  $whisper
     * @return array<string, mixed>
     */
    private function dropWhisperField(array $whisper, string $field): array
    {
        unset($whisper[$field]);

        if (isset($whisper['segment']) && is_array($whisper['segment'])) {
            unset($whisper['segment'][$field]);
        }

        if (isset($whisper['segments']) && is_array($whisper['segments'])) {
            $whisper['segments'] = array_map(function (mixed $segment) use ($field): mixed {
                if (is_array($segment)) {
                    unset($segment[$field]);
                }

                return $segment;
            }, $whisper['segments']);
        }

        return $whisper;
    }
}