CaiWuXiTong/app/Business/CourierFreightImporter.php

<?php

namespace App\Business;

use Illuminate\Database\DatabaseManager;
use InvalidArgumentException;
use RuntimeException;

/**
 * 快递运费导入（.xlsx / .csv）。
 *
 * 源表头：来源 / 成品配件（也接受「成品运费」）/ 交易单号 / 提供成本运费（按表头名匹配，列顺序无关）。
 * 快递方给的文件常见几十万行，xlsx 走 XlsxSheetReader 流式读取，不整表进内存。
 */
class CourierFreightImporter
{
    /** 源表头 → 数据库列 */
    private const COLUMNS = [
        '来源' => 'source',
        '成品配件' => 'part',
        '成品运费' => 'part',
        '交易单号' => 'transaction_no',
        '提供成本运费' => 'freight_amount',
    ];

    private const HEADERS = ['来源', '成品配件', '交易单号', '提供成本运费'];

    /** 表头别名：源文件里这一列有时写「成品配件」、有时写「成品运费」。 */
    private const HEADER_ALIASES = [
        '成品配件' => ['成品配件', '成品运费'],
    ];

    private const CHUNK = 2000;

    public function __construct(
        private readonly DatabaseManager $databases,
        private readonly ?XlsxSheetReader $reader = null,
    ) {}

    /**
     * @return array{file:string,rows:int,inserted:int,skipped:int,amount:string,sources:array<string,int>,seconds:float}
     */
    public function import(string $path, int $companyId = 1, bool $replace = false, string $source = 'courier-freight-import'): array
    {
        if (! is_file($path)) {
            throw new InvalidArgumentException('找不到要导入的文件：' . $path);
        }

        $target = $this->databases->connection();
        $startedAt = microtime(true);
        $now = now();

        if ($replace) {
            $target->table('courier_freight_rows')->where('company_id', $companyId)->delete();
        }

        $inserted = 0;
        $skipped = 0;
        $amount = 0.0;
        $sources = [];
        $buffer = [];

        $flush = function () use (&$buffer, &$inserted, $target): void {
            foreach (array_chunk($buffer, self::CHUNK) as $chunk) {
                $target->table('courier_freight_rows')->insert($chunk);
                $inserted += count($chunk);
            }
            $buffer = [];
        };

        foreach ($this->lines($path) as $line) {
            $sourceValue = $this->text($line['来源'] ?? null, 64);
            $transactionNo = $this->text($line['交易单号'] ?? null, 191);
            if ($sourceValue === null && $transactionNo === null) {
                $skipped++;

                continue;
            }

            $value = $this->number($line['提供成本运费'] ?? null);
            $sources[$sourceValue ?? ''] = ($sources[$sourceValue ?? ''] ?? 0) + 1;
            $amount += $value;

            $buffer[] = [
                'company_id' => $companyId,
                'source' => $sourceValue ?? '',
                'part' => $this->text($line['成品配件'] ?? null, 191),
                'transaction_no' => $transactionNo ?? '',
                'freight_amount' => number_format($value, 4, '.', ''),
                'created_at' => $now,
                'updated_at' => $now,
            ];

            if (count($buffer) >= self::CHUNK * 4) {
                $flush();
            }
        }

        $flush();
        arsort($sources);

        return [
            'file' => basename($path),
            'rows' => $inserted + $skipped,
            'inserted' => $inserted,
            'skipped' => $skipped,
            'amount' => number_format($amount, 2, '.', ''),
            'sources' => $sources,
            'seconds' => round(microtime(true) - $startedAt, 1),
        ];
    }

    /** @return \Generator<int, array<string,string>> */
    private function lines(string $path): \Generator
    {
        if (mb_strtolower(pathinfo($path, PATHINFO_EXTENSION)) === 'csv') {
            $handle = fopen($path, 'rb');
            if ($handle === false) {
                throw new RuntimeException('无法读取文件：' . $path);
            }

            try {
                $header = null;
                while (($cells = fgetcsv($handle)) !== false) {
                    if ($header === null) {
                        $cells[0] = preg_replace('/^\xEF\xBB\xBF/', '', (string) ($cells[0] ?? ''));
                        $header = array_map(static fn ($value): string => trim((string) $value), $cells);
                        $this->assertHeader($header);

                        continue;
                    }

                    $line = [];
                    foreach ($header as $index => $name) {
                        $line[$this->canonicalName($name)] = (string) ($cells[$index] ?? '');
                    }

                    yield $line;
                }
            } finally {
                fclose($handle);
            }

            return;
        }

        if (mb_strtolower(pathinfo($path, PATHINFO_EXTENSION)) !== 'xlsx') {
            throw new InvalidArgumentException('只支持 xlsx、csv。');
        }

        $columns = array_map(
            static fn (string $name): array|string => self::HEADER_ALIASES[$name] ?? $name,
            self::HEADERS,
        );

        yield from ($this->reader ?? app(XlsxSheetReader::class))->rows($path, $columns);
    }

    /** 把表头别名折回规范列名。 */
    private function canonicalName(string $header): string
    {
        foreach (self::HEADER_ALIASES as $canonical => $names) {
            if (in_array($header, $names, true)) {
                return $canonical;
            }
        }

        return $header;
    }

    /** @param list<string> $header */
    private function assertHeader(array $header): void
    {
        foreach (self::HEADERS as $required) {
            $names = self::HEADER_ALIASES[$required] ?? [$required];
            if (array_intersect($names, $header) === []) {
                throw new InvalidArgumentException('文件缺少列：' . $required);
            }
        }
    }

    private function text(mixed $value, int $limit): ?string
    {
        if ($value === null) {
            return null;
        }

        $text = trim((string) (is_scalar($value) ? $value : ''));

        return $text === '' ? null : mb_substr($text, 0, $limit);
    }

    private function number(mixed $value): float
    {
        if ($value === null || $value === '') {
            return 0.0;
        }
        if (is_numeric($value)) {
            return (float) $value;
        }

        $text = str_replace([',', ' ', '¥', '￥'], '', (string) $value);

        return is_numeric($text) ? (float) $text : 0.0;
    }
}