CaiWuXiTong/app/Business/SupplierSummaryInboundImporter.php

940 lines
31 KiB
PHP
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<?php
namespace App\Business;
use App\Models\Business\InventoryWarehouseNameMapping;
use DateTimeImmutable;
use DOMDocument;
use DOMXPath;
use Illuminate\Support\Facades\DB;
use PhpOffice\PhpSpreadsheet\Shared\Date as ExcelDate;
use RuntimeException;
use Throwable;
use XMLReader;
use ZipArchive;
/**
* 供应商汇总表月度增量导入。
*
* 仅追加目标月份的新行,不删除其他月份。内容相同的源明细按出现次序分配
* 稳定标识后全部保留;重复执行时只跳过已经写入的同一源行。
* 单行字段异常会进入失败清单,不阻断其他有效行。
*/
class SupplierSummaryInboundImporter
{
private const TABLE = 'inventory_inbound_detail_rows';
private const DETAIL_SHEET = '明细';
/** @var array<string,list<string>> */
private const REQUIRED_HEADERS = [
'source_no' => ['来源单号'],
'document_no' => ['单据编号(采购单号)', '单据编号(采购单号)', '单据编号'],
'supplier' => ['供应商'],
'item_no' => ['货号'],
'barcode' => ['条码'],
'spec_name' => ['规格名'],
'quantity' => ['库存变动量'],
'unit_price' => ['单价'],
'amount' => ['金额'],
'estimated_unit_price' => ['预估单价'],
'estimated_amount' => ['预估金额'],
'supplier_barcode' => ['供应商条码'],
'remark' => ['备注'],
'warehouse' => ['仓库'],
];
/** @var array<string,list<string>> */
private const OPTIONAL_HEADERS = [
'express_name' => ['快递名称'],
'express_no' => ['快递单号'],
'not_sent' => ['未发出'],
'owner_name' => ['负责人'],
];
/**
* @return array{
* status:string,
* report_month:string,
* source_filename:string,
* success:int,
* duplicates:int,
* failed:int,
* mapped:int,
* total_in_table:int,
* failure_reason_count:int,
* failure_reasons:list<array{row:int,reason:string}>
* }
*/
public function import(string $file, string $reportMonth): array
{
$path = realpath($file);
if ($path === false || ! is_file($path) || ! is_readable($path)) {
throw new RuntimeException('供应商汇总文件不存在或不可读取。');
}
$month = $this->normalizeMonth($reportMonth);
$connection = (string) config('upstream_mirrors.connection', 'business_mirror');
$maps = $this->shortNameMap();
[$reader, $sharedStrings] = $this->streamDetailSheet($path);
try {
$success = 0;
$duplicates = 0;
$mapped = 0;
$failed = 0;
$reasons = [];
$occurrences = [];
$batch = [];
DB::connection($connection)->transaction(function () use (
$connection,
$reader,
$sharedStrings,
$maps,
$month,
&$success,
&$duplicates,
&$mapped,
&$failed,
&$reasons,
&$occurrences,
&$batch,
): void {
$flush = function () use (
$connection,
$month,
&$success,
&$duplicates,
&$batch,
): void {
if ($batch === []) {
return;
}
$keys = array_column($batch, 'row_key');
$existing = DB::connection($connection)
->table(self::TABLE)
->where('report_month', $month)
->whereIn('row_key', $keys)
->pluck('row_key')
->map(static fn (mixed $key): string => (string) $key)
->all();
$existing = array_fill_keys($existing, true);
$insertable = [];
foreach ($batch as $record) {
if (isset($existing[$record['row_key']])) {
$duplicates++;
continue;
}
$insertable[] = $record;
}
if ($insertable !== []) {
$inserted = DB::connection($connection)
->table(self::TABLE)
->insertOrIgnore($insertable);
$success += (int) $inserted;
$duplicates += count($insertable) - (int) $inserted;
}
$batch = [];
};
$headerRow = null;
$columns = [];
$headerCandidates = [];
while ($reader->read()) {
if (
$reader->nodeType !== XMLReader::ELEMENT
|| $reader->name !== 'row'
) {
continue;
}
$rowNumber = (int) $reader->getAttribute('r');
$values = $this->readRow($reader, $sharedStrings);
if ($headerRow === null) {
$headerCandidates[$rowNumber] = $values;
$header = $this->headerColumns($headerCandidates);
if ($header !== null) {
[$headerRow, $columns] = $header;
$headerCandidates = [];
continue;
}
if ($rowNumber >= 20) {
throw new RuntimeException(
'「明细」子表前 20 行未找到供应商汇总标准表头。',
);
}
continue;
}
if ($rowNumber <= $headerRow) {
continue;
}
if ($this->emptyRow($values)) {
continue;
}
try {
$record = $this->record(
$values,
$columns,
$maps,
$month,
$rowNumber,
);
} catch (Throwable $exception) {
$failed++;
$reasons[] = [
'row' => $rowNumber,
'reason' => $exception->getMessage(),
];
continue;
}
$baseKey = (string) $record['row_key'];
$occurrence = ($occurrences[$baseKey] ?? 0) + 1;
$occurrences[$baseKey] = $occurrence;
if ($occurrence > 1) {
$record['row_key'] = hash(
'sha256',
$baseKey . '|occurrence:' . $occurrence,
);
}
$mapped++;
$batch[] = $record;
if (count($batch) >= 500) {
$flush();
}
}
if ($headerRow === null) {
throw new RuntimeException(
'「明细」子表前 20 行未找到供应商汇总标准表头。',
);
}
$flush();
});
$status = $failed === 0
? 'success'
: ($success > 0 ? 'partial' : 'failed');
return [
'status' => $status,
'report_month' => $month,
'source_filename' => basename($path),
'success' => $success,
'duplicates' => $duplicates,
'failed' => $failed,
'mapped' => $mapped,
'total_in_table' => DB::connection($connection)
->table(self::TABLE)
->where('report_month', $month)
->count(),
'failure_reason_count' => count($reasons),
'failure_reasons' => $reasons,
];
} finally {
$reader->close();
}
}
private function normalizeMonth(string $month): string
{
if (preg_match('/^\d{4}-\d{2}$/', $month) !== 1) {
throw new RuntimeException('月份格式应为 YYYY-MM。');
}
return $month . '-01';
}
/**
* @param array<int,array<int,mixed>> $rows
* @return array{0:int,1:array<string,int>}|null
*/
private function headerColumns(array $rows): ?array
{
ksort($rows);
foreach ($rows as $row => $values) {
$positions = [];
foreach ($values as $index => $value) {
$heading = $this->normalizeHeader(
(string) $this->plainValue($value),
);
if ($heading !== '') {
$positions[$heading][] = $index;
}
}
$columns = [];
foreach (self::REQUIRED_HEADERS as $field => $aliases) {
$position = $this->firstHeaderPosition($positions, $aliases);
if ($position !== null) {
$columns[$field] = $position;
}
}
$createdPositions = $positions[$this->normalizeHeader('创建时间')] ?? [];
if (count($createdPositions) >= 2) {
$columns['created_serial'] = $createdPositions[0];
$columns['created_at_raw'] = $createdPositions[1];
}
if (count($columns) < count(self::REQUIRED_HEADERS) + 2) {
continue;
}
foreach (self::OPTIONAL_HEADERS as $field => $aliases) {
$position = $this->firstHeaderPosition($positions, $aliases);
if ($position !== null) {
$columns[$field] = $position;
}
}
foreach (['created_serial', 'created_at_raw'] as $field) {
if (! isset($columns[$field])) {
throw new RuntimeException('「明细」表头缺少第二个「创建时间」列。');
}
}
return [$row, $columns];
}
return null;
}
/** @param array<string,list<int>> $positions @param list<string> $aliases */
private function firstHeaderPosition(array $positions, array $aliases): ?int
{
foreach ($aliases as $alias) {
$heading = $this->normalizeHeader($alias);
if (isset($positions[$heading][0])) {
return $positions[$heading][0];
}
}
return null;
}
private function normalizeHeader(string $heading): string
{
return preg_replace('/[\s\x{3000}]+/u', '', trim($heading)) ?? trim($heading);
}
/**
* @param array<int,mixed> $values
* @param array<string,int> $columns
* @param array{warehouse:array<string,string>,provider:array<string,string>} $maps
* @return array<string,mixed>
*/
private function record(
array $values,
array $columns,
array $maps,
string $month,
int $rowNumber,
): array {
$text = fn (string $field): ?string => $this->text(
$values[$columns[$field] ?? -1] ?? null,
);
$number = function (string $label, string $field) use (
$values,
$columns,
$rowNumber,
): ?float {
$value = $values[$columns[$field] ?? -1] ?? null;
if (is_array($value)) {
$formula = trim((string) ($value['__formula'] ?? ''));
if ($formula !== '') {
return $this->formulaValue(
$label,
$field,
$formula,
$rowNumber,
$values,
$columns,
);
}
$value = $value['__cached'] ?? null;
}
$text = str_replace(
[',', ','],
'',
trim((string) $this->plainValue($value)),
);
if ($text === '') {
return null;
}
if (! is_numeric($text)) {
throw new RuntimeException("字段「{$label}」不是有效数字:" . $text);
}
return (float) $text;
};
$barcode = $text('barcode');
$warehouse = $text('warehouse');
if ($barcode === null) {
throw new RuntimeException('商品编码(条码)为空');
}
if ($warehouse === null) {
throw new RuntimeException('仓库为空');
}
$shortName = $maps['warehouse'][$warehouse]
?? $maps['provider'][$warehouse]
?? null;
if ($shortName === null || trim($shortName) === '') {
throw new RuntimeException('仓库未匹配到唯一仓库简称:' . $warehouse);
}
$createdRaw = $this->moment($values[$columns['created_at_raw']] ?? null);
if ($createdRaw === null) {
throw new RuntimeException('第二个「创建时间」为空或格式无效');
}
$record = [
'report_month' => $month,
'source_no' => $text('source_no'),
'document_no' => $text('document_no'),
'created_serial' => $text('created_serial'),
'supplier' => $text('supplier'),
'item_no' => $text('item_no'),
'barcode' => mb_strtoupper($barcode),
'spec_name' => $text('spec_name'),
'quantity' => $number('库存变动量', 'quantity'),
'unit_price' => $number('单价', 'unit_price'),
'amount' => $number('金额', 'amount'),
'estimated_unit_price' => $number('预估单价', 'estimated_unit_price'),
'estimated_amount' => $number('预估金额', 'estimated_amount'),
'supplier_barcode' => $text('supplier_barcode'),
'remark' => $text('remark'),
'warehouse' => $warehouse,
'warehouse_short_name' => $shortName,
'created_at_raw' => $createdRaw,
'express_name' => $text('express_name'),
'express_no' => $text('express_no'),
'not_sent' => $text('not_sent'),
'owner_name' => $text('owner_name'),
];
unset($record['warehouse_short_name']);
$record['row_key'] = hash(
'sha256',
json_encode($record, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
);
$record['warehouse_short_name'] = $shortName;
$record['imported_at'] = now();
return $record;
}
/**
* 仅接受金额类字段当前行内的乘积公式,避免执行来源文件中的任意公式。
*
* @param array<int,mixed> $values
* @param array<string,int> $columns
*/
private function formulaValue(
string $label,
string $field,
string $formula,
int $rowNumber,
array $values,
array $columns,
): float {
$expectedFields = match ($field) {
'amount' => ['quantity', 'unit_price'],
'estimated_amount' => ['estimated_unit_price', 'quantity'],
default => [],
};
if ($expectedFields === []) {
throw new RuntimeException(
"字段「{$label}」不允许使用公式:" . $formula,
);
}
if (
preg_match(
'/^=?\s*\$?([A-Z]{1,3})\$?(\d+)\s*\*\s*\$?([A-Z]{1,3})\$?(\d+)\s*$/i',
$formula,
$matches,
) !== 1
) {
throw new RuntimeException(
"字段「{$label}」公式无法安全识别:" . $formula,
);
}
if ((int) $matches[2] !== $rowNumber || (int) $matches[4] !== $rowNumber) {
throw new RuntimeException(
"字段「{$label}」公式引用了其他行:" . $formula,
);
}
$actualColumns = [
$this->columnIndex($matches[1]),
$this->columnIndex($matches[3]),
];
$expectedColumns = [
$columns[$expectedFields[0]] + 1,
$columns[$expectedFields[1]] + 1,
];
sort($actualColumns);
sort($expectedColumns);
if ($actualColumns !== $expectedColumns) {
throw new RuntimeException(
"字段「{$label}」公式与取数口径不一致:" . $formula,
);
}
$left = $this->numericReference(
$values[$actualColumns[0] - 1] ?? null,
$label,
);
$right = $this->numericReference(
$values[$actualColumns[1] - 1] ?? null,
$label,
);
return $left * $right;
}
private function numericReference(mixed $value, string $label): float
{
if (is_array($value)) {
if (($value['__formula'] ?? '') !== '') {
throw new RuntimeException(
"字段「{$label}」的公式引用了另一个公式单元格",
);
}
$value = $value['__cached'] ?? null;
}
$text = str_replace(
[',', ','],
'',
trim((string) $this->plainValue($value)),
);
if (! is_numeric($text)) {
throw new RuntimeException(
"字段「{$label}」的公式引用值不是有效数字:" . $text,
);
}
return (float) $text;
}
/** @param array<int,mixed> $values */
private function emptyRow(array $values): bool
{
foreach ($values as $value) {
if (trim((string) $this->plainValue($value)) !== '') {
return false;
}
}
return true;
}
private function text(mixed $value): ?string
{
$text = trim((string) $this->plainValue($value));
return $text === '' ? null : $text;
}
private function plainValue(mixed $value): mixed
{
if (is_array($value)) {
return $value['__cached'] ?? null;
}
return $value;
}
private function moment(mixed $value): ?string
{
$value = $this->plainValue($value);
if ($value === null || trim((string) $value) === '') {
return null;
}
if (is_numeric($value)) {
try {
return ExcelDate::excelToDateTimeObject((float) $value)->format('Y-m-d H:i:s');
} catch (Throwable) {
return null;
}
}
$value = trim((string) $value);
foreach (['Y/m/d H:i:s', 'Y-m-d H:i:s', 'Y/m/d', 'Y-m-d'] as $format) {
$date = DateTimeImmutable::createFromFormat($format, $value);
if ($date instanceof DateTimeImmutable) {
return $date->format('Y-m-d H:i:s');
}
}
$timestamp = strtotime($value);
return $timestamp === false ? null : date('Y-m-d H:i:s', $timestamp);
}
/** @return array{warehouse:array<string,string>,provider:array<string,string>} */
private function shortNameMap(): array
{
$rows = InventoryWarehouseNameMapping::withoutGlobalScopes()
->get(['warehouse_provider', 'warehouse', 'warehouse_short_name']);
$grouped = [];
foreach ($rows as $row) {
$warehouse = trim((string) $row->warehouse);
if ($warehouse !== '') {
$grouped[$warehouse][] = trim((string) $row->warehouse_short_name);
}
}
$byWarehouse = [];
foreach ($grouped as $warehouse => $names) {
$unique = array_values(array_unique($names));
if (count($unique) === 1 && $unique[0] !== '') {
$byWarehouse[$warehouse] = $unique[0];
}
}
$providerGrouped = [];
foreach ($rows as $row) {
$provider = trim((string) $row->warehouse_provider);
if ($provider !== '') {
$providerGrouped[$provider][] = [
'warehouse' => trim((string) $row->warehouse),
'short_name' => trim((string) $row->warehouse_short_name),
];
}
}
$byProvider = [];
foreach ($providerGrouped as $provider => $providerRows) {
$unique = array_values(array_unique(array_column(
$providerRows,
'short_name',
)));
if (count($unique) === 1 && $unique[0] !== '') {
$byProvider[$provider] = $unique[0];
continue;
}
$mainShortNames = [];
foreach ($providerRows as $providerRow) {
if (str_contains($providerRow['warehouse'], '主仓')) {
$mainShortNames[] = $providerRow['short_name'];
}
}
$mainShortNames = array_values(array_unique($mainShortNames));
if (count($mainShortNames) === 1 && $mainShortNames[0] !== '') {
$byProvider[$provider] = $mainShortNames[0];
}
}
return ['warehouse' => $byWarehouse, 'provider' => $byProvider];
}
/** @return array{0:XMLReader,1:list<string>} */
private function streamDetailSheet(string $path): array
{
$zip = new ZipArchive();
if ($zip->open($path) !== true) {
throw new RuntimeException('供应商汇总文件不是有效的 XLSX 工作簿。');
}
try {
$sheetPath = $this->detailSheetPath($zip);
$sharedStrings = $zip->locateName('xl/sharedStrings.xml') !== false
? $this->readSharedStrings($path)
: [];
} finally {
$zip->close();
}
return [
$this->openZipEntry($path, $sheetPath),
$sharedStrings,
];
}
private function detailSheetPath(ZipArchive $zip): string
{
$workbookXml = $zip->getFromName('xl/workbook.xml');
$relationshipsXml = $zip->getFromName('xl/_rels/workbook.xml.rels');
if ($workbookXml === false || $relationshipsXml === false) {
throw new RuntimeException('供应商汇总工作簿缺少工作簿结构文件。');
}
$workbook = new DOMDocument();
$relationships = new DOMDocument();
if (
! $workbook->loadXML($workbookXml, LIBXML_NONET)
|| ! $relationships->loadXML($relationshipsXml, LIBXML_NONET)
) {
throw new RuntimeException('供应商汇总工作簿结构无法读取。');
}
$relationshipId = null;
$xpath = new DOMXPath($workbook);
$xpath->registerNamespace(
'main',
'http://schemas.openxmlformats.org/spreadsheetml/2006/main',
);
$xpath->registerNamespace(
'office',
'http://schemas.openxmlformats.org/officeDocument/2006/relationships',
);
foreach ($xpath->query('/main:workbook/main:sheets/main:sheet') ?: [] as $sheet) {
if ($sheet instanceof \DOMElement
&& $sheet->getAttribute('name') === self::DETAIL_SHEET
) {
$relationshipId = $sheet->getAttributeNS(
'http://schemas.openxmlformats.org/officeDocument/2006/relationships',
'id',
);
break;
}
}
if ($relationshipId === null || $relationshipId === '') {
throw new RuntimeException('供应商汇总表缺少「明细」子表。');
}
$target = null;
foreach ($relationships->getElementsByTagName('Relationship') as $relationship) {
if ($relationship->getAttribute('Id') === $relationshipId) {
$target = $relationship->getAttribute('Target');
break;
}
}
if ($target === null || $target === '') {
throw new RuntimeException('「明细」子表的工作簿关系不存在。');
}
return $this->normalizeZipPath(
str_starts_with($target, '/') ? ltrim($target, '/') : 'xl/' . $target,
);
}
private function normalizeZipPath(string $path): string
{
$segments = [];
foreach (explode('/', str_replace('\\', '/', $path)) as $segment) {
if ($segment === '' || $segment === '.') {
continue;
}
if ($segment === '..') {
array_pop($segments);
continue;
}
$segments[] = $segment;
}
return implode('/', $segments);
}
/** @return list<string> */
private function readSharedStrings(string $path): array
{
$reader = $this->openZipEntry($path, 'xl/sharedStrings.xml');
$strings = [];
$current = null;
$stringDepth = null;
$phoneticDepth = null;
try {
while ($reader->read()) {
if (
$reader->nodeType === XMLReader::ELEMENT
&& $reader->name === 'si'
) {
$current = '';
$stringDepth = $reader->depth;
$phoneticDepth = null;
continue;
}
if ($current === null) {
continue;
}
if (
$reader->nodeType === XMLReader::ELEMENT
&& $reader->name === 'rPh'
) {
$phoneticDepth = $reader->depth;
continue;
}
if (
$reader->nodeType === XMLReader::END_ELEMENT
&& $reader->name === 'rPh'
&& $reader->depth === $phoneticDepth
) {
$phoneticDepth = null;
continue;
}
if (
$reader->nodeType === XMLReader::ELEMENT
&& $reader->name === 't'
&& $phoneticDepth === null
) {
$current .= $reader->readString();
continue;
}
if (
$reader->nodeType === XMLReader::END_ELEMENT
&& $reader->name === 'si'
&& $reader->depth === $stringDepth
) {
$strings[] = $current;
$current = null;
}
}
} finally {
$reader->close();
}
return $strings;
}
private function openZipEntry(string $path, string $entry): XMLReader
{
$reader = new XMLReader();
$flags = LIBXML_NONET | LIBXML_COMPACT;
if (! $reader->open("zip://{$path}#{$entry}", null, $flags)) {
throw new RuntimeException("无法读取工作簿内容:{$entry}");
}
return $reader;
}
/**
* @param list<string> $sharedStrings
* @return array<int,mixed>
*/
private function readRow(XMLReader $reader, array $sharedStrings): array
{
$values = [];
$rowDepth = $reader->depth;
while ($reader->read()) {
if (
$reader->nodeType === XMLReader::END_ELEMENT
&& $reader->name === 'row'
&& $reader->depth === $rowDepth
) {
break;
}
if (
$reader->nodeType !== XMLReader::ELEMENT
|| $reader->name !== 'c'
) {
continue;
}
$reference = $reader->getAttribute('r') ?? '';
if (
preg_match('/^([A-Z]{1,3})\d+$/i', $reference, $matches) !== 1
) {
continue;
}
$values[$this->columnIndex($matches[1]) - 1] = $this->readCell(
$reader,
$sharedStrings,
);
}
return $values;
}
/** @param list<string> $sharedStrings */
private function readCell(XMLReader $reader, array $sharedStrings): mixed
{
if ($reader->isEmptyElement) {
return null;
}
$type = $reader->getAttribute('t') ?? '';
$cellDepth = $reader->depth;
$cached = null;
$formula = null;
$hasFormula = false;
$inlineParts = [];
$phoneticDepth = null;
while ($reader->read()) {
if (
$reader->nodeType === XMLReader::END_ELEMENT
&& $reader->name === 'c'
&& $reader->depth === $cellDepth
) {
break;
}
if ($reader->nodeType === XMLReader::ELEMENT) {
if ($reader->name === 'f') {
$hasFormula = true;
$formula = $reader->readString();
continue;
}
if ($reader->name === 'v') {
$cached = $reader->readString();
continue;
}
if ($reader->name === 'rPh') {
$phoneticDepth = $reader->depth;
continue;
}
if (
$reader->name === 't'
&& $phoneticDepth === null
) {
$inlineParts[] = $reader->readString();
}
}
if (
$reader->nodeType === XMLReader::END_ELEMENT
&& $reader->name === 'rPh'
&& $reader->depth === $phoneticDepth
) {
$phoneticDepth = null;
}
}
$value = match ($type) {
's' => is_numeric($cached)
? ($sharedStrings[(int) $cached] ?? null)
: null,
'inlineStr' => implode('', $inlineParts),
'b' => $cached === null ? null : ($cached === '1' ? '1' : '0'),
default => $cached ?? implode('', $inlineParts),
};
if ($hasFormula) {
return [
'__formula' => (string) $formula,
'__cached' => $value,
];
}
return $value;
}
private function columnIndex(string $column): int
{
$index = 0;
foreach (str_split(strtoupper($column)) as $letter) {
$index = ($index * 26) + (ord($letter) - 64);
}
return $index;
}
}