nabeh/backend/public/test_audio.php

<?php
// Secure token check to prevent unauthorized execution on production
if (($_GET['token'] ?? '') !== 'nabeh_test_audio_1298') {
    http_response_code(403);
    die('Unauthorized access');
}

require_once dirname(__DIR__) . '/app/bootstrap.php';

use App\Services\GeminiService;
use App\Models\ChatbotRule;
use App\Core\Database;

header('Content-Type: text/plain; charset=utf-8');

echo "=== Starting Chatbot Audio Reply Diagnostics ===\n\n";

// 1. Fetch Chatbot Rules and check for Gemini Key
$rules = Database::select("SELECT * FROM chatbot_rules WHERE trigger_type = 'gemini_ai' AND is_active = 1 LIMIT 1");
if (empty($rules)) {
    echo "❌ [Database] No active chatbot rule found with 'gemini_ai' trigger type.\n";
    exit(1);
}

$rule = $rules[0];
$apiKey = $rule['gemini_api_key'] ?: getenv('GEMINI_API_KEY');

if (empty($apiKey)) {
    echo "❌ [Config] Gemini API Key is empty. Please set gemini_api_key in the active chatbot rule or GEMINI_API_KEY in the .env file.\n";
    exit(1);
}

echo "✅ [Database] Found active Gemini rule for Company ID: {$rule['company_id']}\n";
echo "ℹ️ [Config] Gemini API Key starts with: " . substr($apiKey, 0, 6) . "...\n";

// 2. Test Gemini API Audio Output (Puck Voice, multiple models)
echo "\n--- Testing Gemini Audio Response ---\n";
$systemPrompt = "You are a friendly customer service assistant. Speak in a warm, welcoming tone.";
$testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";

$candidateModels = [
    'gemini-3.1-flash-tts-preview',
    'gemini-2.5-flash-preview-tts',
    'gemini-2.0-flash',
    'gemini-2.5-flash',
    'gemini-3.5-flash',
    'gemini-flash-latest'
];

$successfulModel = null;
$audioResponse = null;

foreach ($candidateModels as $model) {
    echo "Testing model: models/{$model} ...\n";
    
    $url = 'https://generativelanguage.googleapis.com/v1beta/models/' . $model . ':generateContent?key=' . $apiKey;
    
    // For TTS models, we do not send systemInstruction because they don't support it (Developer instruction not enabled)
    // Instead we prepend it to the user message parts
    $useSystemInstructionField = !str_contains($model, 'tts');
    
    $contents = [];
    if (!$useSystemInstructionField) {
        $contents[] = [
            'role' => 'user',
            'parts' => [
                ['text' => "System instruction: " . $systemPrompt],
                ['text' => $testMsg]
            ]
        ];
    } else {
        $contents[] = [
            'role' => 'user',
            'parts' => [['text' => $testMsg]]
        ];
    }
    
    $payloadData = [
        'contents' => $contents,
        'generationConfig' => [
            'responseModalities' => ['AUDIO'],
            'speechConfig' => [
                'voiceConfig' => [
                    'prebuiltVoiceConfig' => [
                        'voiceName' => 'Puck'
                    ]
                ]
            ]
        ]
    ];
    
    if ($useSystemInstructionField) {
        $payloadData['systemInstruction'] = [
            'parts' => [['text' => $systemPrompt]]
        ];
    }
    
    $payload = json_encode($payloadData);

    $ch = curl_init($url);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_POST, true);
    curl_setopt($ch, CURLOPT_POSTFIELDS, $payload);
    curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']);
    curl_setopt($ch, CURLOPT_TIMEOUT, 15);
    $response = curl_exec($ch);
    $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);

    if ($httpCode !== 200) {
        $errorData = json_decode($response, true);
        $errMsg = $errorData['error']['message'] ?? 'Unknown error';
        echo "   ❌ Failed (HTTP {$httpCode}: {$errMsg})\n";
        echo "   Response: " . substr($response, 0, 500) . "\n\n";
    } else {
        $data = json_decode($response, true);
        $part = $data['candidates'][0]['content']['parts'][0] ?? null;
        if ($part && isset($part['inlineData'])) {
            $audioResponse = [
                'audio' => $part['inlineData']['data'],
                'mimeType' => $part['inlineData']['mimeType'] ?? 'audio/mp4'
            ];
            $successfulModel = $model;
            echo "   ✅ SUCCESS!\n\n";
            break; // Stop at first successful model
        } else {
            echo "   ❌ Failed (No inlineData in response)\n";
            echo "   Response: " . substr($response, 0, 500) . "\n\n";
        }
    }
}

if ($successfulModel && $audioResponse && !empty($audioResponse['audio'])) {
    echo "\n✅ [Gemini] Successfully generated voice note using model: models/{$successfulModel}\n";
    echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse['mimeType'] . "\n";
    echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse['audio']) . " base64 chars\n";
    
    // 3. Test Audio-to-Audio conversion using the Service
    echo "\n--- Testing Audio-to-Audio (Speech-to-Speech) via GeminiService ---\n";
    $startTime = microtime(true);
    
    $audioResponse2 = GeminiService::generateAudioResponseFromAudio($apiKey, $systemPrompt, $audioResponse['audio'], $audioResponse['mimeType'], 'Puck');
    $elapsedTime2 = round(microtime(true) - $startTime, 2);
    
    if ($audioResponse2 && !empty($audioResponse2['audio'])) {
        echo "✅ [Gemini] Successfully generated Audio-to-Audio response in {$elapsedTime2} seconds!\n";
        echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse2['mimeType'] . "\n";
        echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse2['audio']) . " base64 chars\n";
    } else {
        echo "❌ [Gemini] Audio-to-Audio generation failed.\n";
    }
} else {
    echo "\n❌ [Gemini] All model trials for audio response generation failed.\n";
}

// 4. Check Node.js WhatsApp Gateway Status
echo "\n--- Checking Node.js Gateway Status ---\n";
$gatewayUrl = rtrim(getenv('WHATSAPP_GATEWAY_URL') ?: 'http://localhost:3722', '/');
echo "Gateway URL: {$gatewayUrl}\n";

$ch = curl_init($gatewayUrl . '/health');
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_TIMEOUT, 5);
$healthResponse = curl_exec($ch);
$healthHttpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
$healthError = curl_error($ch);
curl_close($ch);

if ($healthHttpCode === 200) {
    echo "✅ [Gateway] Gateway is ONLINE and healthy.\n";
    echo "ℹ️ [Gateway] Response: {$healthResponse}\n";
} else {
    echo "❌ [Gateway] Gateway is OFFLINE or returning error. HTTP Code: {$healthHttpCode}. Error: {$healthError}\n";
}

echo "\n=== Diagnostics Complete ===\n";