feat: 完善 SYB 蝦皮主数据与软删除 (#203 #204 #205)

2026-08-13 10:39:21 +08:00
parent 3af9bf245a
commit ad1ccc2c60
23 changed files with 864 additions and 90 deletions
@@ -5,7 +5,9 @@
 package repository

 import (
+	"crypto/sha256"
 	"database/sql"
+	"encoding/hex"
 	"errors"
 	"fmt"
 	"strings"
@@ -351,19 +353,7 @@ func UpsertSybOrder(q Execer, o model.SybOrder) (created bool, err error) {

 	now := model.NowISO()
 	if strings.TrimSpace(o.ShopeeGoodsID) != "" {
-		if _, err := q.Exec(`
-			INSERT INTO shopee_products (goods_id, title, source, created_at, updated_at)
-			VALUES (?, ?, 'syb', ?, ?)
-			ON DUPLICATE KEY UPDATE
-				title = CASE
-					WHEN source = 'syb' AND TRIM(VALUES(title)) <> '' THEN VALUES(title)
-					ELSE title
-				END,
-				updated_at = CASE
-					WHEN source = 'syb' AND TRIM(VALUES(title)) <> '' THEN VALUES(updated_at)
-					ELSE updated_at
-				END`,
-			o.ShopeeGoodsID, o.Title, now, now); err != nil {
+		if err := upsertSybShopeeProduct(q, o.ShopeeGoodsID, o.Title, o.ShopName, o.ImageURL, now); err != nil {
 			return false, fmt.Errorf("为顺运宝明细 %s 补建蝦皮商品骨架失败: %w", o.SybID, err)
 		}
 	}
@@ -393,6 +383,173 @@ func UpsertSybOrder(q Execer, o model.SybOrder) (created bool, err error) {
 	return created, nil
 }

+// upsertSybShopeeProduct 用顺运宝观测补建蝦皮商品骨架。
+//
+// 店铺和图片是字段级低优先级数据：只能补空值，或更新原本同样来自 syb 的值；
+// 人工字段和商品目录等权威来源永远不被顺运宝覆盖。空值也不能清除已有内容。
+func upsertSybShopeeProduct(q Execer, goodsID, title, shopName, imageURL, observedAt string) error {
+	_, err := q.Exec(`
+		INSERT INTO shopee_products
+		       (goods_id,title,image_url,shopee_shop_name,
+		        image_source,image_observed_at,image_is_manual,
+		        shop_name_source,shop_name_observed_at,shop_name_is_manual,
+		        source,source_observed_at,created_at,updated_at)
+		VALUES (?, ?, NULLIF(TRIM(?),''), NULLIF(TRIM(?),''),
+		        CASE WHEN TRIM(?)='' THEN NULL ELSE 'syb' END,
+		        CASE WHEN TRIM(?)='' THEN NULL ELSE ? END, 0,
+		        CASE WHEN TRIM(?)='' THEN NULL ELSE 'syb' END,
+		        CASE WHEN TRIM(?)='' THEN NULL ELSE ? END, 0,
+		        'syb', ?, ?, ?)
+		ON DUPLICATE KEY UPDATE
+			title = CASE
+				WHEN source='syb' AND TRIM(VALUES(title))<>'' THEN VALUES(title)
+				ELSE title
+			END,
+			source_observed_at = CASE
+				WHEN source='syb' AND TRIM(VALUES(title))<>'' THEN VALUES(source_observed_at)
+				ELSE source_observed_at
+			END,
+			image_url = CASE
+				WHEN image_is_manual=0 AND TRIM(VALUES(image_url))<>''
+				 AND (image_url IS NULL OR TRIM(image_url)='' OR image_source='syb') THEN VALUES(image_url)
+				ELSE image_url
+			END,
+			image_observed_at = CASE
+				WHEN image_is_manual=0 AND TRIM(VALUES(image_url))<>''
+				 AND (image_url IS NULL OR TRIM(image_url)='' OR image_source='syb') THEN VALUES(image_observed_at)
+				ELSE image_observed_at
+			END,
+			image_source = CASE
+				WHEN image_is_manual=0 AND TRIM(VALUES(image_url))<>''
+				 AND (image_url IS NULL OR TRIM(image_url)='' OR image_source='syb') THEN 'syb'
+				ELSE image_source
+			END,
+			shopee_shop_name = CASE
+				WHEN shop_name_is_manual=0 AND TRIM(VALUES(shopee_shop_name))<>''
+				 AND (shopee_shop_name IS NULL OR TRIM(shopee_shop_name)='' OR shop_name_source='syb') THEN VALUES(shopee_shop_name)
+				ELSE shopee_shop_name
+			END,
+			shop_name_observed_at = CASE
+				WHEN shop_name_is_manual=0 AND TRIM(VALUES(shopee_shop_name))<>''
+				 AND (shopee_shop_name IS NULL OR TRIM(shopee_shop_name)='' OR shop_name_source='syb') THEN VALUES(shop_name_observed_at)
+				ELSE shop_name_observed_at
+			END,
+			shop_name_source = CASE
+				WHEN shop_name_is_manual=0 AND TRIM(VALUES(shopee_shop_name))<>''
+				 AND (shopee_shop_name IS NULL OR TRIM(shopee_shop_name)='' OR shop_name_source='syb') THEN 'syb'
+				ELSE shop_name_source
+			END,
+			updated_at = CASE
+				WHEN source='syb'
+				  OR (image_source='syb' AND TRIM(VALUES(image_url))<>'')
+				  OR (shop_name_source='syb' AND TRIM(VALUES(shopee_shop_name))<>'') THEN VALUES(updated_at)
+				ELSE updated_at
+			END`,
+		goodsID, title, imageURL, shopName,
+		imageURL, imageURL, observedAt,
+		shopName, shopName, observedAt,
+		observedAt, observedAt, observedAt)
+	return err
+}
+
+// backfillSybProductMetadata 把历史货运单中每个商品最新的非空店铺、图片补入商品主表。
+// 重放是安全的：实际覆盖规则仍由 upsertSybShopeeProduct 统一执行。
+func backfillSybProductMetadata(db *sql.DB) error {
+	type metadata struct{ shopName, imageURL, observedAt string }
+	items := map[string]metadata{}
+	readLatest := func(column string, assign func(*metadata, string)) error {
+		query := fmt.Sprintf(`SELECT shopee_goods_id, value_text, updated_at FROM (
+			SELECT shopee_goods_id, %s AS value_text, updated_at,
+			       ROW_NUMBER() OVER (PARTITION BY shopee_goods_id ORDER BY updated_at DESC, syb_id DESC) AS row_no
+			FROM syb_orders
+			WHERE TRIM(COALESCE(shopee_goods_id,''))<>'' AND TRIM(COALESCE(%s,''))<>''
+		) ranked WHERE row_no=1`, column, column)
+		rows, err := db.Query(query)
+		if err != nil {
+			return err
+		}
+		defer rows.Close()
+		for rows.Next() {
+			var goodsID, value, observedAt string
+			if err := rows.Scan(&goodsID, &value, &observedAt); err != nil {
+				return err
+			}
+			item := items[goodsID]
+			assign(&item, value)
+			if observedAt > item.observedAt {
+				item.observedAt = observedAt
+			}
+			items[goodsID] = item
+		}
+		return rows.Err()
+	}
+	if err := readLatest("shop_name", func(item *metadata, value string) { item.shopName = value }); err != nil {
+		return fmt.Errorf("读取历史顺运宝店铺失败: %w", err)
+	}
+	if err := readLatest("image_url", func(item *metadata, value string) { item.imageURL = value }); err != nil {
+		return fmt.Errorf("读取历史顺运宝图片失败: %w", err)
+	}
+	for goodsID, item := range items {
+		if err := upsertSybShopeeProduct(db, goodsID, "", item.shopName, item.imageURL, item.observedAt); err != nil {
+			return fmt.Errorf("回填蝦皮商品 %s 的顺运宝元数据失败: %w", goodsID, err)
+		}
+	}
+	return nil
+}
+
+// UpsertSybShopeeSKU 把格式明确的 SYB 规格作为低优先级蝦皮 SKU 观测写入。
+// 返回 parsed=false 时不写库，避免把猜测结果污染商品主数据。
+func UpsertSybShopeeSKU(q Execer, goodsID, specRaw, observedAt string) (parsed bool, err error) {
+	parsedSpec, ok := spec.ParseShopeeSpec(specRaw)
+	if !ok || strings.TrimSpace(goodsID) == "" {
+		return false, nil
+	}
+	specKey, err := spec.SpecKey(specRaw)
+	if err != nil {
+		return false, nil
+	}
+	sum := sha256.Sum256([]byte(goodsID + "\x00" + specKey))
+	recordID := "syb:" + hex.EncodeToString(sum[:])
+	now := model.NowISO()
+	_, err = UpsertCatalogShopeeSKU(q, CatalogShopeeSKUInput{
+		RecordID: recordID, GoodsID: goodsID, SpecRaw: specRaw, SpecKey: specKey,
+		Color: parsedSpec.Color, Size: parsedSpec.Size, Advice: parsedSpec.Advice,
+		ParseOK: true, Source: "syb", ObservedAt: observedAt, Now: now, UpdatePolicy: "fill_missing",
+	})
+	return true, err
+}
+
+func backfillSybShopeeSKUs(db *sql.DB) error {
+	rows, err := db.Query(`SELECT shopee_goods_id,product_spec,updated_at FROM (
+		SELECT shopee_goods_id,product_spec,updated_at,
+		       ROW_NUMBER() OVER (PARTITION BY shopee_goods_id,product_spec ORDER BY updated_at DESC,syb_id DESC) row_no
+		FROM syb_orders
+		WHERE TRIM(COALESCE(shopee_goods_id,''))<>'' AND TRIM(COALESCE(product_spec,''))<>''
+	) ranked WHERE row_no=1`)
+	if err != nil {
+		return fmt.Errorf("读取历史顺运宝规格失败: %w", err)
+	}
+	defer rows.Close()
+	type observation struct{ goodsID, raw, observedAt string }
+	var observations []observation
+	for rows.Next() {
+		var item observation
+		if err := rows.Scan(&item.goodsID, &item.raw, &item.observedAt); err != nil {
+			return err
+		}
+		observations = append(observations, item)
+	}
+	if err := rows.Err(); err != nil {
+		return err
+	}
+	for _, item := range observations {
+		if _, err := UpsertSybShopeeSKU(db, item.goodsID, item.raw, item.observedAt); err != nil {
+			return fmt.Errorf("回填蝦皮商品 %s 的顺运宝规格失败: %w", item.goodsID, err)
+		}
+	}
+	return nil
+}
+
 // SybSpecObservation 是顺运宝对某个蝦皮商品规格的历史观测汇总。
 // 它不是正式蝦皮 SKU，不得写回 shopee_skus。
 type SybSpecObservation struct {
@@ -482,7 +639,7 @@ func sybOrderFilterClause(filter SybOrderFilter) (string, []any) {

 const sybOrderContextFrom = `
 		  FROM syb_orders so
-		  LEFT JOIN shopee_products sp ON sp.goods_id = so.shopee_goods_id
+		  LEFT JOIN shopee_products sp ON sp.goods_id = so.shopee_goods_id AND sp.deleted_at IS NULL
 		  LEFT JOIN pdd_products pp ON pp.goods_id = sp.pdd_goods_id AND pp.deleted_at IS NULL
 		  LEFT JOIN spec_mappings sm
 		    ON sm.shopee_goods_id = so.shopee_goods_id