webstatement/app/Jobs/ProcessStmtEntryDataJob.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\StmtEntry;

    class ProcessStmtEntryDataJob 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.STMT.ENTRY.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  $entryBatch     = [];

        /**
         * 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 statement entry data processing');
                    return;
                }

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

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

        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 statement entry 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 StmtEntry())->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->entryBatch) >= self::CHUNK_SIZE) {
                    $this->saveBatch();
                    $chunkCount++;
                    Log::info("Processed chunk $chunkCount ({$this->processedCount} records so far)");
                }
            }

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

            $data = array_combine($headers, $row);
            $this->cleanTransReference($data);
            $this->addToBatch($data, $rowCount, $filePath);
        }

        private function cleanTransReference(array &$data)
        : void
        {
            if (isset($data['trans_reference'])) {
                // Clean trans_reference from \\BNK if present
                $data['trans_reference'] = preg_replace('/\\\\.*$/', '', $data['trans_reference']);
            }
        }

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

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

        /**
         * Save batched records to the database
         */
        private function saveBatch()
        : void
        {
            try {
                if (!empty($this->entryBatch)) {
                    // Process in smaller chunks for better memory management
                    foreach ($this->entryBatch as $entry) {
                        // Extract all stmt_entry_ids from the current chunk
                        $entryIds = array_column($entry, 'stmt_entry_id');

                        // Delete existing records with these IDs to avoid conflicts
                        StmtEntry::whereIn('stmt_entry_id', $entryIds)->delete();

                        // Insert all records in the chunk at once
                        StmtEntry::insert($entry);
                    }

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

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

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