NIM_E31221299/app/Console/Commands/ScrapTweets.php

<?php

namespace App\Console\Commands;

use Illuminate\Console\Command;
use Illuminate\Support\Str;
use Symfony\Component\Process\Exception\ProcessFailedException;
use Symfony\Component\Process\Process;
use League\Csv\Reader;
use Illuminate\Support\Facades\File;

class ScrapTweets extends Command
{
    protected $signature = 'scrap:tweets
                            {--keyword= : Kata kunci pencarian tweet (tanpa lang:id)}
                            {--limit=100 : Jumlah maksimal tweet}
                            {--output= : Nama file output (opsional, tanpa path)}
                            {--token= : Bearer token Twitter (jika kosong, ambil dari config)}';

    protected $description = 'Scrape tweets via tweet-harvest dan simpan ke public/tweets-data';

    public function handle()
    {
        set_time_limit(0);

        // 1. Ambil & validasi input
        $keyword = trim($this->option('keyword'));
        $limit   = (int) $this->option('limit');
        $token   = $this->option('token') ?: config('services.twitter.token');

        if (! $keyword) {
            $this->error('❌ Keyword wajib diisi. Gunakan opsi --keyword=');
            return 1;
        }

        // 2. Siapkan query & nama file
        $cleanKeyword   = preg_replace('/\s*lang:id\s*/i', '', $keyword);
        $searchQuery    = trim($cleanKeyword) . ' lang:id';
        $slugName       = Str::slug($cleanKeyword);
        $outputFilename = $this->option('output') ?: "{$slugName}.csv";

        // 3. Direktori final: public/tweets-data
        $outputDir = public_path('tweets-data');
        if (! is_dir($outputDir)) {
            mkdir($outputDir, 0755, true);
        }
        $outputPath = $outputDir . DIRECTORY_SEPARATOR . $outputFilename;

        $this->info("🔍 Mulai scraping: \"{$searchQuery}\" dengan limit {$limit}");
        $this->info("💾 Output diharapkan: public/tweets-data/{$outputFilename}");

        // 4. Path ke binary tweet-harvest
        $isWin   = strncasecmp(PHP_OS, 'WIN', 3) === 0;
        $binName = $isWin ? 'tweet-harvest.cmd' : 'tweet-harvest';
        $binary  = base_path('node_modules')
                 . DIRECTORY_SEPARATOR . '.bin'
                 . DIRECTORY_SEPARATOR . $binName;

        if (! file_exists($binary)) {
            $this->error("❌ Binary tweet-harvest tidak ditemukan: {$binary}");
            $this->error("   Jalankan: npm install tweet-harvest di root project");
            return 1;
        }

        // 5. Jalankan proses di workingDir = public/tweets-data sehingga output langsung di folder ini
        $workingDir = $outputDir;
        $outputOption = $outputFilename;  // hanya nama file, tanpa path

        $process = new Process([
            $binary,
            '-o', $outputOption,
            '-s', $searchQuery,
            '--tab', 'LATEST',
            '-l', $limit,
            '--token', $token,
        ], $workingDir);

        // Set environment: pastikan PATH mencakup nodejs & Chromium jika perlu
        $currentPath = getenv('PATH') ?: '';
        if ($isWin) {
            // tambahkan path NodeJS jika diperlukan
            $envPath = 'C:\\Program Files\\nodejs;' . $currentPath;
        } else {
            $envPath = $currentPath;
        }
        $process->setEnv([
            'PATH' => $envPath,
            'TEMP' => sys_get_temp_dir(),
            'TMP'  => sys_get_temp_dir(),
        ]);
        $process->setTimeout(900);

        // Run dan tampilkan output untuk debugging
        $this->info("➡️ Menjalankan tweet-harvest di folder: {$workingDir}");
        $process->run(function ($type, $buffer) {
            $this->line($buffer);
        });

        if (! $process->isSuccessful()) {
            $this->error("❌ Proses tweet-harvest gagal.");
            $this->error($process->getErrorOutput());
            throw new ProcessFailedException($process);
        }

        // 6. Verifikasi file di lokasi utama
if (file_exists($outputPath)) {
    $this->info("✅ File ditemukan di: {$outputPath}");
} else {
    // Coba cari di nested path + file lama (misal .old.csv)
    $nestedDir = $outputDir . DIRECTORY_SEPARATOR . 'tweets-data';
    $fallbackOld = $nestedDir . DIRECTORY_SEPARATOR . pathinfo($outputFilename, PATHINFO_FILENAME) . '.old.csv';

    if (file_exists($fallbackOld)) {
        $this->warn("⚠️ File tidak ditemukan di lokasi utama, tapi ditemukan file lama di: {$fallbackOld}");

        // Salin/move file lama ke lokasi utama
        try {
            File::move($fallbackOld, $outputPath);
            $this->info("➡️ File .old.csv dipindahkan ke: {$outputPath}");

            // Hapus folder nested jika kosong
            @rmdir($nestedDir);
        } catch (\Exception $e) {
            $this->error("❌ Gagal memindahkan file fallback: " . $e->getMessage());
            return 1;
        }
    } else {
        $this->error("❌ File output tidak ditemukan di: {$outputPath}");
        $this->error("   Juga tidak ditemukan fallback di: {$fallbackOld}");
        return 1;
    }
}


        // 7. Hitung jumlah baris CSV (kecuali header)
        try {
            $csv = Reader::createFromPath($outputPath, 'r');
            $csv->setHeaderOffset(0);
            $count = iterator_count($csv->getRecords());
        } catch (\Exception $e) {
            $this->error("❌ Gagal membaca CSV: " . $e->getMessage());
            return 1;
        }

        $this->info("✅ Scraping selesai! Total tweet: {$count}");
        return 0;
    }
}