Files
webstatement/app/Jobs/ProcessBillDetailDataJob.php
daengdeni 1b3e0ed30d feat(webstatement): implement batch processing for data jobs
- Tambahkan pemrosesan data secara bertahap (chunking) dengan konstanta `CHUNK_SIZE` untuk mengurangi penggunaan memori dan menangani data dalam jumlah besar.
- Perbarui `ProcessArrangementDataJob`:
  - Tambahkan properti `arrangementBatch` untuk menyimpan batch data sementara.
  - Implementasikan metode `addToBatch` untuk menambahkan data ke batch.
  - Implementasikan metode `saveBatch` untuk menyimpan data batch ke database menggunakan metode bulk `TempArrangement::upsert`.
  - Tambahkan logging untuk melacak progress pemrosesan data per chunk.
  - Reset batch setelah penyimpanan atau ketika terjadi error untuk menghindari pemrosesan ulang data yang gagal.
- Perbarui `ProcessBillDetailDataJob`:
  - Tambahkan properti `billDetailBatch` untuk menyimpan batch data sementara.
  - Implementasikan metode `addToBatch` untuk menambahkan data ke batch.
  - Implementasikan metode `saveBatch` untuk menyimpan data batch ke database menggunakan metode bulk `TempBillDetail::upsert`.
  - Tambahkan logging untuk melacak progress pemrosesan data per chunk.
  - Reset batch setelah penyimpanan atau ketika terjadi error untuk menghindari pemrosesan ulang data yang gagal.
- Perbaiki penghitungan error count dengan menambahkannya saat terjadi error pada pemrosesan batch.
- Tambahkan timestamp (`created_at` dan `updated_at`) pada setiap record dalam batch sebelum disimpan ke database.
- Lakukan cleanup batch secara otomatis setelah pemrosesan selesai.
2025-05-28 09:29:46 +07:00

218 lines
7.1 KiB
PHP

<?php
namespace Modules\Webstatement\Jobs;
use Exception;
use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Facades\Storage;
use Modules\Webstatement\Models\TempBillDetail;
class ProcessBillDetailDataJob implements ShouldQueue
{
use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
private const CSV_DELIMITER = '~';
private const MAX_EXECUTION_TIME = 86400; // 24 hours in seconds
private const FILENAME = 'ST.AA.BILL.DETAILS.csv';
private const DISK_NAME = 'sftpStatement';
private const CHUNK_SIZE = 1000; // Process data in chunks to reduce memory usage
private string $period = '';
private int $processedCount = 0;
private int $errorCount = 0;
private array $billDetailBatch = [];
/**
* Create a new job instance.
*/
public function __construct(string $period = '')
{
$this->period = $period;
}
/**
* Execute the job.
*/
public function handle()
: void
{
try {
$this->initializeJob();
if ($this->period === '') {
Log::warning('No period provided for bill detail data processing');
return;
}
$this->processPeriod();
$this->logJobCompletion();
} catch (Exception $e) {
Log::error('Error in ProcessBillDetailDataJob: ' . $e->getMessage());
throw $e;
}
}
private function initializeJob()
: void
{
set_time_limit(self::MAX_EXECUTION_TIME);
$this->processedCount = 0;
$this->errorCount = 0;
$this->billDetailBatch = [];
}
private function processPeriod()
: void
{
$disk = Storage::disk(self::DISK_NAME);
$filename = "{$this->period}." . self::FILENAME;
$filePath = "{$this->period}/$filename";
if (!$this->validateFile($disk, $filePath)) {
return;
}
$tempFilePath = $this->createTemporaryFile($disk, $filePath, $filename);
$this->processFile($tempFilePath, $filePath);
$this->cleanup($tempFilePath);
}
private function validateFile($disk, string $filePath)
: bool
{
Log::info("Processing bill detail file: $filePath");
if (!$disk->exists($filePath)) {
Log::warning("File not found: $filePath");
return false;
}
return true;
}
private function createTemporaryFile($disk, string $filePath, string $filename)
: string
{
$tempFilePath = storage_path("app/temp_$filename");
file_put_contents($tempFilePath, $disk->get($filePath));
return $tempFilePath;
}
private function processFile(string $tempFilePath, string $filePath)
: void
{
$handle = fopen($tempFilePath, "r");
if ($handle === false) {
Log::error("Unable to open file: $filePath");
return;
}
$headers = (new TempBillDetail())->getFillable();
$rowCount = 0;
$chunkCount = 0;
while (($row = fgetcsv($handle, 0, self::CSV_DELIMITER)) !== false) {
$rowCount++;
$this->processRow($row, $headers, $rowCount, $filePath);
// Process in chunks to avoid memory issues
if (count($this->billDetailBatch) >= self::CHUNK_SIZE) {
$this->saveBatch();
$chunkCount++;
Log::info("Processed chunk $chunkCount ({$this->processedCount} records so far)");
}
}
// Process any remaining records
if (!empty($this->billDetailBatch)) {
$this->saveBatch();
}
fclose($handle);
Log::info("Completed processing $filePath. Processed {$this->processedCount} records with {$this->errorCount} errors.");
}
private function processRow(array $row, array $headers, int $rowCount, string $filePath)
: void
{
if (count($headers) !== count($row)) {
Log::warning("Row $rowCount in $filePath has incorrect column count. Expected: " .
count($headers) . ", Got: " . count($row));
$this->errorCount++;
return;
}
$data = array_combine($headers, $row);
$this->addToBatch($data, $rowCount, $filePath);
}
/**
* Add record to batch instead of saving immediately
*/
private function addToBatch(array $data, int $rowCount, string $filePath)
: void
{
try {
if (isset($data['_id']) && $data['_id'] !== '_id') {
// Add timestamp fields
$now = now();
$data['created_at'] = $now;
$data['updated_at'] = $now;
// Add to batch
$this->billDetailBatch[] = $data;
$this->processedCount++;
}
} catch (Exception $e) {
$this->errorCount++;
Log::error("Error processing Bill Detail at row $rowCount in $filePath: " . $e->getMessage());
}
}
/**
* Save batched records to the database
*/
private function saveBatch()
: void
{
try {
if (!empty($this->billDetailBatch)) {
// Bulk insert/update bill details
TempBillDetail::upsert(
$this->billDetailBatch,
['_id'], // Unique key
array_diff((new TempBillDetail())->getFillable(), ['_id']) // Update columns
);
// Reset batch after processing
$this->billDetailBatch = [];
}
} catch (Exception $e) {
Log::error("Error in saveBatch: " . $e->getMessage());
$this->errorCount += count($this->billDetailBatch);
// Reset batch even if there's an error to prevent reprocessing the same failed records
$this->billDetailBatch = [];
}
}
private function cleanup(string $tempFilePath)
: void
{
if (file_exists($tempFilePath)) {
unlink($tempFilePath);
}
}
private function logJobCompletion()
: void
{
Log::info("Bill Detail data processing completed. " .
"Total processed: {$this->processedCount}, Total errors: {$this->errorCount}");
}
}