webstatement/app/Jobs/ProcessDataCaptureDataJob.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\DataCapture;

    class ProcessDataCaptureDataJob 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.DATA.CAPTURE.csv';
        private const DISK_NAME          = 'staging';
        private const CHUNK_SIZE         = 1000; // Process data in chunks to reduce memory usage
        private const CSV_HEADERS        = [
            'id',
            'account_number',
            'sign',
            'amount_lcy',
            'transaction_code',
            'their_reference',
            'narrative',
            'pl_category',
            'customer_id',
            'account_officer',
            'product_category',
            'value_date',
            'currency',
            'amount_fcy',
            'exchange_rate',
            'neg_ref_no',
            'position_type',
            'our_reference',
            'reversal_marker',
            'exposure_date',
            'currency_market',
            'iblc_country',
            'last_version',
            'otor_version',
            'department_code',
            'dealer_desk',
            'bank_sort_cde',
            'cheque_number',
            'accounting_date',
            'contingent_acct',
            'cheq_type',
            'tfs_reference',
            'accounting_company',
            'stmt_no',
            'curr_no',
            'inputter',
            'authoriser',
            'co_code',
            'date_time'
        ];
        private string $period = '';
        private int    $processedCount = 0;
        private int    $errorCount     = 0;
        private array  $captureBatch   = [];

        /**
         * 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 data capture processing');
                    return;
                }

                $this->processPeriod();
                $this->logJobCompletion();
            } catch (Exception $e) {
                Log::error('Error in ProcessDataCaptureDataJob: ' . $e->getMessage());
                throw $e;
            }
        }

        private function initializeJob()
        : void
        {
            set_time_limit(self::MAX_EXECUTION_TIME);
            $this->processedCount = 0;
            $this->errorCount     = 0;
            $this->captureBatch   = [];
        }

        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 data capture 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;
            }

            $rowCount = 0;
            $chunkCount = 0;

            while (($row = fgetcsv($handle, 0, self::CSV_DELIMITER)) !== false) {
                $rowCount++;

                // Skip header row if it exists
                if ($rowCount === 1 && strtolower($row[0]) === 'id') {
                    continue;
                }

                $this->processRow($row, $rowCount, $filePath);

                // Process in chunks to avoid memory issues
                if (count($this->captureBatch) >= self::CHUNK_SIZE) {
                    $this->saveBatch();
                    $chunkCount++;
                    Log::info("Processed chunk $chunkCount ({$this->processedCount} records so far)");
                }
            }

            // Process any remaining records
            if (!empty($this->captureBatch)) {
                $this->saveBatch();
            }

            fclose($handle);
            Log::info("Completed processing $filePath. Processed {$this->processedCount} records with {$this->errorCount} errors.");
        }

        private function processRow(array $row, int $rowCount, string $filePath)
        : void
        {
            // Exclude the last field from CSV
            if (count($row) > 0) {
                //array_pop($row);
                Log::info("Excluded last field from row $rowCount. New column count: " . count($row));
            }

            if (count(self::CSV_HEADERS) !== count($row)) {
                Log::warning("Row $rowCount in $filePath has incorrect column count. Expected: " .
                    count(self::CSV_HEADERS) . ", Got: " . count($row));
                return;
            }

            $data = array_combine(self::CSV_HEADERS, $row);
            $this->addToBatch($data, $rowCount, $filePath);
        }

        /**
         * Add record to batch instead of saving immediately
         */
        private function addToBatch(array $data, int $rowCount, string $filePath)
        : void
        {
            try {
                if (!empty($data['id'])) {
                    // Add timestamp fields
                    $now = now();
                    $data['created_at'] = $now;
                    $data['updated_at'] = $now;

                    // Add to capture batch
                    $this->captureBatch[] = $data;
                    $this->processedCount++;
                }
            } catch (Exception $e) {
                $this->errorCount++;
                Log::error("Error processing Data Capture at row $rowCount in $filePath: " . $e->getMessage());
            }
        }

        /**
         * Save batched records to the database
         */
        private function saveBatch()
        : void
        {
            try {
                if (!empty($this->captureBatch)) {
                    // Bulk insert/update data captures
                    DataCapture::upsert(
                        $this->captureBatch,
                        ['id'], // Unique key
                        array_diff(self::CSV_HEADERS, ['id']) // Update columns
                    );

                    // Reset capture batch after processing
                    $this->captureBatch = [];
                }
            } catch (Exception $e) {
                Log::error("Error in saveBatch: " . $e->getMessage());
                $this->errorCount += count($this->captureBatch);
                // Reset batch even if there's an error to prevent reprocessing the same failed records
                $this->captureBatch = [];
            }
        }

        private function cleanup(string $tempFilePath)
        : void
        {
            if (file_exists($tempFilePath)) {
                unlink($tempFilePath);
            }
        }

        private function logJobCompletion()
        : void
        {
            Log::info("Data capture processing completed. " .
                "Total processed: {$this->processedCount}, Total errors: {$this->errorCount}");
        }
    }