feat: 回填三个Excel关联PDD正式价格 (#226)

2026-08-14 16:36:42 +08:00
parent 619394ac7d
commit 7055519321
4 changed files with 784 additions and 0 deletions
@@ -0,0 +1,396 @@
+package service
+
+import (
+	"database/sql"
+	"encoding/json"
+	"fmt"
+	"math/big"
+	"os"
+	"path/filepath"
+	"regexp"
+	"sort"
+	"strconv"
+	"strings"
+	"time"
+
+	"cmautobuy/admin/model"
+	"cmautobuy/admin/repository"
+	"github.com/xuri/excelize/v2"
+)
+
+const (
+	XLSXPriceBackfillShop = "qwg8fkb044"
+	xlsxPriceSheet        = "在线商品"
+
+	XLSXPriceCandidate         = "candidate"
+	XLSXPricePDDMissing        = "pdd_missing_or_deleted"
+	XLSXPriceAlreadyCollected  = "already_collected"
+	XLSXPriceStatusUnsupported = "status_unsupported"
+	XLSXPriceJSONInvalid       = "pdd_json_invalid"
+	XLSXPriceSourceMismatch    = "spec_source_not_shopee_backfill"
+	XLSXPriceSKUsMissing       = "pdd_skus_missing"
+	XLSXPriceSpecsInvalid      = "pdd_specs_invalid"
+	XLSXPriceAlreadyExists     = "pdd_price_already_exists"
+)
+
+var XLSXPriceBackfillFiles = []string{
+	"shopee_chanpin_1_已处理.xlsx",
+	"shopee_chanpin_2_已处理.xlsx",
+	"shopee_chanpin_3_已处理.xlsx",
+}
+
+var xlsxPricePattern = regexp.MustCompile(`^\s*(\d+(?:\.\d+)?)\s*(?:[~～\-—–]\s*(\d+(?:\.\d+)?))?\s*$`)
+
+// XLSXPDDPriceSource 是三个 Excel 归并后的单个 PDD 正式价格来源。
+type XLSXPDDPriceSource struct {
+	PDDGoodsID         string
+	MaxTWD             string
+	PriceCent          int64
+	ShopeeProductCount int
+}
+
+// XLSXPriceSources 是三个固定工作簿的只读解析汇总。
+type XLSXPriceSources struct {
+	RowCount                       int
+	ShopeeProductCount             int
+	PDDProductCount                int
+	SharedPDDWithDifferentMaxCount int
+	PDDPrices                      map[string]XLSXPDDPriceSource
+}
+
+type xlsxShopeePrice struct {
+	pddGoodsID string
+	maxTWD     *big.Rat
+}
+
+// LoadXLSXPriceSources 只读取白名单中的三个固定文件，并只处理目标店铺。
+func LoadXLSXPriceSources(dir string) (XLSXPriceSources, error) {
+	dir = strings.TrimSpace(dir)
+	if dir == "" {
+		return XLSXPriceSources{}, fmt.Errorf("Excel 目录不能为空")
+	}
+	products := make(map[string]xlsxShopeePrice)
+	result := XLSXPriceSources{PDDPrices: make(map[string]XLSXPDDPriceSource)}
+
+	for _, name := range XLSXPriceBackfillFiles {
+		path := filepath.Join(dir, name)
+		info, err := os.Stat(path)
+		if err != nil {
+			return XLSXPriceSources{}, fmt.Errorf("读取固定 Excel %s 失败: %w", name, err)
+		}
+		if !info.Mode().IsRegular() {
+			return XLSXPriceSources{}, fmt.Errorf("固定 Excel %s 不是普通文件", name)
+		}
+		rows, err := loadXLSXPriceFile(path, products)
+		if err != nil {
+			return XLSXPriceSources{}, err
+		}
+		if rows == 0 {
+			return XLSXPriceSources{}, fmt.Errorf("固定 Excel %s 没有店铺 %s 的数据", name, XLSXPriceBackfillShop)
+		}
+		result.RowCount += rows
+	}
+
+	pricesByPDD := make(map[string]map[string]struct{})
+	maxByPDD := make(map[string]*big.Rat)
+	productCountByPDD := make(map[string]int)
+	for _, product := range products {
+		canonical := product.maxTWD.RatString()
+		if pricesByPDD[product.pddGoodsID] == nil {
+			pricesByPDD[product.pddGoodsID] = make(map[string]struct{})
+		}
+		pricesByPDD[product.pddGoodsID][canonical] = struct{}{}
+		productCountByPDD[product.pddGoodsID]++
+		currentMax, exists := maxByPDD[product.pddGoodsID]
+		if !exists || product.maxTWD.Cmp(currentMax) > 0 {
+			maxByPDD[product.pddGoodsID] = new(big.Rat).Set(product.maxTWD)
+		}
+	}
+	for pddGoodsID, maximum := range maxByPDD {
+		priceCent, err := twdToCNYCent(maximum)
+		if err != nil {
+			return XLSXPriceSources{}, fmt.Errorf("PDD 商品 %s 的价格无法转换: %w", pddGoodsID, err)
+		}
+		result.PDDPrices[pddGoodsID] = XLSXPDDPriceSource{
+			PDDGoodsID: pddGoodsID, MaxTWD: decimalText(maximum), PriceCent: priceCent,
+			ShopeeProductCount: productCountByPDD[pddGoodsID],
+		}
+	}
+	for _, prices := range pricesByPDD {
+		if len(prices) > 1 {
+			result.SharedPDDWithDifferentMaxCount++
+		}
+	}
+	result.ShopeeProductCount = len(products)
+	result.PDDProductCount = len(result.PDDPrices)
+	return result, nil
+}
+
+func loadXLSXPriceFile(path string, products map[string]xlsxShopeePrice) (int, error) {
+	book, err := excelize.OpenFile(path, excelize.Options{RawCellValue: true})
+	if err != nil {
+		return 0, fmt.Errorf("打开固定 Excel %s 失败: %w", filepath.Base(path), err)
+	}
+	defer book.Close()
+	if index, err := book.GetSheetIndex(xlsxPriceSheet); err != nil || index < 0 {
+		return 0, fmt.Errorf("固定 Excel %s 缺少工作表 %q", filepath.Base(path), xlsxPriceSheet)
+	}
+	rows, err := book.Rows(xlsxPriceSheet)
+	if err != nil {
+		return 0, fmt.Errorf("读取固定 Excel %s 失败: %w", filepath.Base(path), err)
+	}
+	defer rows.Close()
+	if !rows.Next() {
+		return 0, fmt.Errorf("固定 Excel %s 是空文件", filepath.Base(path))
+	}
+	headers, err := rows.Columns()
+	if err != nil {
+		return 0, fmt.Errorf("读取固定 Excel %s 表头失败: %w", filepath.Base(path), err)
+	}
+	indexes := make(map[string]int)
+	for index, header := range headers {
+		indexes[strings.TrimSpace(header)] = index
+	}
+	required := []string{"店铺显示名", "商品ID", "货源ID", "价格"}
+	for _, header := range required {
+		if _, exists := indexes[header]; !exists {
+			return 0, fmt.Errorf("固定 Excel %s 缺少列 %q", filepath.Base(path), header)
+		}
+	}
+
+	matchedRows := 0
+	rowNumber := 1
+	for rows.Next() {
+		rowNumber++
+		columns, err := rows.Columns()
+		if err != nil {
+			return 0, fmt.Errorf("读取固定 Excel %s 第 %d 行失败: %w", filepath.Base(path), rowNumber, err)
+		}
+		if xlsxColumn(columns, indexes["店铺显示名"]) != XLSXPriceBackfillShop {
+			continue
+		}
+		matchedRows++
+		shopeeID := xlsxColumn(columns, indexes["商品ID"])
+		pddID := xlsxColumn(columns, indexes["货源ID"])
+		priceRaw := xlsxColumn(columns, indexes["价格"])
+		if shopeeID == "" || pddID == "" || priceRaw == "" {
+			return 0, fmt.Errorf("固定 Excel %s 第 %d 行商品ID、货源ID和价格不能为空", filepath.Base(path), rowNumber)
+		}
+		maxTWD, err := parseXLSXMaxTWD(priceRaw)
+		if err != nil {
+			return 0, fmt.Errorf("固定 Excel %s 第 %d 行价格 %q 无效: %w", filepath.Base(path), rowNumber, priceRaw, err)
+		}
+		current, exists := products[shopeeID]
+		if exists && current.pddGoodsID != pddID {
+			return 0, fmt.Errorf("蝦皮商品 %s 同时关联 PDD %s 和 %s", shopeeID, current.pddGoodsID, pddID)
+		}
+		if !exists || maxTWD.Cmp(current.maxTWD) > 0 {
+			products[shopeeID] = xlsxShopeePrice{pddGoodsID: pddID, maxTWD: maxTWD}
+		}
+	}
+	if err := rows.Error(); err != nil {
+		return 0, fmt.Errorf("遍历固定 Excel %s 失败: %w", filepath.Base(path), err)
+	}
+	return matchedRows, nil
+}
+
+func xlsxColumn(columns []string, index int) string {
+	if index < 0 || index >= len(columns) {
+		return ""
+	}
+	return strings.TrimSpace(columns[index])
+}
+
+func parseXLSXMaxTWD(raw string) (*big.Rat, error) {
+	matches := xlsxPricePattern.FindStringSubmatch(raw)
+	if matches == nil {
+		return nil, fmt.Errorf("只允许单价或用 ~ 连接的价格区间")
+	}
+	first, ok := new(big.Rat).SetString(matches[1])
+	if !ok || first.Sign() <= 0 {
+		return nil, fmt.Errorf("价格必须大于 0")
+	}
+	maximum := first
+	if matches[2] != "" {
+		second, ok := new(big.Rat).SetString(matches[2])
+		if !ok || second.Sign() <= 0 {
+			return nil, fmt.Errorf("价格必须大于 0")
+		}
+		if second.Cmp(maximum) > 0 {
+			maximum = second
+		}
+	}
+	return new(big.Rat).Set(maximum), nil
+}
+
+// twdToCNYCent 按用户确认公式「台币 ÷ 10 × 100」转换，并对正数四舍五入到分。
+func twdToCNYCent(twd *big.Rat) (int64, error) {
+	if twd == nil || twd.Sign() <= 0 {
+		return 0, fmt.Errorf("台币价格必须大于 0")
+	}
+	value := new(big.Rat).Mul(twd, big.NewRat(10, 1))
+	quotient, remainder := new(big.Int), new(big.Int)
+	quotient.QuoRem(value.Num(), value.Denom(), remainder)
+	if new(big.Int).Mul(remainder, big.NewInt(2)).Cmp(value.Denom()) >= 0 {
+		quotient.Add(quotient, big.NewInt(1))
+	}
+	if !quotient.IsInt64() || quotient.Sign() <= 0 {
+		return 0, fmt.Errorf("转换结果超出人民币分的范围")
+	}
+	return quotient.Int64(), nil
+}
+
+func decimalText(value *big.Rat) string {
+	if value.IsInt() {
+		return value.Num().String()
+	}
+	text := value.FloatString(8)
+	return strings.TrimRight(strings.TrimRight(text, "0"), ".")
+}
+
+// XLSXPriceBackfillCandidate 是一件可安全回填正式价格的 PDD 商品。
+type XLSXPriceBackfillCandidate struct {
+	PDDGoodsID   string
+	SourceMaxTWD string
+	PriceCent    int64
+	SKUCount     int
+	Original     repository.PDDPriceBackfillTarget
+	NewSKUsJSON  string
+}
+
+// XLSXPriceBackfillPlan 保存 dry-run 的完整统计和应用所需原值。
+type XLSXPriceBackfillPlan struct {
+	Shop        string
+	GeneratedAt string
+	Sources     XLSXPriceSources
+	Summary     map[string]int
+	Candidates  []XLSXPriceBackfillCandidate
+}
+
+// BuildXLSXPriceBackfillPlan 只在内存中生成计划，不写数据库。
+func BuildXLSXPriceBackfillPlan(sources XLSXPriceSources,
+	targets []repository.PDDPriceBackfillTarget, now time.Time) XLSXPriceBackfillPlan {
+	plan := XLSXPriceBackfillPlan{
+		Shop: XLSXPriceBackfillShop, GeneratedAt: now.UTC().Format(model.TimeLayout),
+		Sources: sources, Summary: make(map[string]int),
+	}
+	targetByID := make(map[string]repository.PDDPriceBackfillTarget, len(targets))
+	for _, target := range targets {
+		targetByID[target.GoodsID] = target
+	}
+	goodsIDs := make([]string, 0, len(sources.PDDPrices))
+	for goodsID := range sources.PDDPrices {
+		goodsIDs = append(goodsIDs, goodsID)
+	}
+	sort.Strings(goodsIDs)
+	for _, goodsID := range goodsIDs {
+		source := sources.PDDPrices[goodsID]
+		target, exists := targetByID[goodsID]
+		if !exists {
+			plan.Summary[XLSXPricePDDMissing]++
+			continue
+		}
+		if target.CollectStatus == string(model.CollectCollected) {
+			plan.Summary[XLSXPriceAlreadyCollected]++
+			continue
+		}
+		if target.CollectStatus != string(model.CollectPending) && target.CollectStatus != string(model.CollectFailed) {
+			plan.Summary[XLSXPriceStatusUnsupported]++
+			continue
+		}
+		newRaw, skuCount, status := fillBackfilledPDDPrices(target.SKUsJSON, source.PriceCent)
+		if status != XLSXPriceCandidate {
+			plan.Summary[status]++
+			continue
+		}
+		plan.Summary[XLSXPriceCandidate]++
+		plan.Candidates = append(plan.Candidates, XLSXPriceBackfillCandidate{
+			PDDGoodsID: goodsID, SourceMaxTWD: source.MaxTWD, PriceCent: source.PriceCent,
+			SKUCount: skuCount, Original: target, NewSKUsJSON: newRaw,
+		})
+	}
+	return plan
+}
+
+func fillBackfilledPDDPrices(raw string, priceCent int64) (string, int, string) {
+	var root map[string]json.RawMessage
+	if err := json.Unmarshal([]byte(raw), &root); err != nil || root == nil {
+		return "", 0, XLSXPriceJSONInvalid
+	}
+	var source string
+	if err := json.Unmarshal(root["spec_source"], &source); err != nil || source != "shopee_backfill" {
+		return "", 0, XLSXPriceSourceMismatch
+	}
+	var skus []map[string]json.RawMessage
+	if err := json.Unmarshal(root["skus"], &skus); err != nil {
+		return "", 0, XLSXPriceJSONInvalid
+	}
+	if len(skus) == 0 {
+		return "", 0, XLSXPriceSKUsMissing
+	}
+	for _, sku := range skus {
+		var options map[string]string
+		if err := json.Unmarshal(sku["options"], &options); err != nil || len(options) == 0 {
+			return "", 0, XLSXPriceSpecsInvalid
+		}
+		for key, value := range options {
+			if strings.TrimSpace(key) == "" || strings.TrimSpace(value) == "" {
+				return "", 0, XLSXPriceSpecsInvalid
+			}
+		}
+		priceRaw, hasPrice := sku["price_cent"]
+		if !hasPrice {
+			return "", 0, XLSXPriceSpecsInvalid
+		}
+		if strings.TrimSpace(string(priceRaw)) != "null" {
+			return "", 0, XLSXPriceAlreadyExists
+		}
+		var availability string
+		if err := json.Unmarshal(sku["availability_status"], &availability); err != nil || availability != "unknown" {
+			return "", 0, XLSXPriceSpecsInvalid
+		}
+		if available, exists := sku["available"]; exists && strings.TrimSpace(string(available)) != "null" {
+			return "", 0, XLSXPriceSpecsInvalid
+		}
+	}
+	encodedPrice := json.RawMessage(strconv.FormatInt(priceCent, 10))
+	for _, sku := range skus {
+		sku["price_cent"] = encodedPrice
+	}
+	encodedSKUs, err := json.Marshal(skus)
+	if err != nil {
+		return "", 0, XLSXPriceJSONInvalid
+	}
+	root["skus"] = encodedSKUs
+	encodedRoot, err := json.Marshal(root)
+	if err != nil {
+		return "", 0, XLSXPriceJSONInvalid
+	}
+	return string(encodedRoot), len(skus), XLSXPriceCandidate
+}
+
+// ApplyXLSXPriceBackfill 在单个事务中应用计划；任一目标变化即整批回滚。
+func ApplyXLSXPriceBackfill(db *sql.DB, plan XLSXPriceBackfillPlan, now time.Time) (int, error) {
+	tx, err := db.Begin()
+	if err != nil {
+		return 0, fmt.Errorf("开始 Excel PDD 正式价格回填事务失败: %w", err)
+	}
+	defer tx.Rollback()
+	at := now.UTC().Format(model.TimeLayout)
+	updated := 0
+	for _, candidate := range plan.Candidates {
+		ok, err := repository.ReplacePDDPricesAndMarkCollected(
+			tx, candidate.Original, candidate.NewSKUsJSON, at, at)
+		if err != nil {
+			return 0, err
+		}
+		if !ok {
+			return 0, fmt.Errorf("PDD 商品 %s 在 dry-run 后发生变化，已回滚整批价格回填", candidate.PDDGoodsID)
+		}
+		updated++
+	}
+	if err := tx.Commit(); err != nil {
+		return 0, fmt.Errorf("提交 Excel PDD 正式价格回填失败: %w", err)
+	}
+	return updated, nil
+}