scripts/amazon_xlsx_to_shoplazza_xlsx.py

1. 添加了两个开关参数 --keep-spu-if-parent-missing：保留父ASIN不在变体列表中的SPU（默认：丢弃整个SPU） --fix-sku-if-title-mismatch：修正标题不一致的SKU而不是丢弃（默认：丢弃标题不一致的SKU） 2. 实现了相关逻辑父ASIN缺失处理：默认：当父ASIN不在变体列表中时，打印警告并丢弃整个SPU 使用 --keep-spu-if-parent-missing：保留SPU，使用第一个变体作为主商品标题不一致处理：默认：当变体标题与主商品不一致时，打印日志并丢弃该SKU 使用 --fix-sku-if-title-mismatch：修正变体标题为主商品标题

scripts/amazon_xlsx_to_shoplazza_xlsx.py
1. 添加了两个开关参数 --keep-spu-if-parent-missing：保留父ASIN不在变体列表中的SPU（默认：丢弃整个SPU） --fix-sku-if-title-mismatch：修正标题不一致的SKU而不是丢弃（默认：丢弃标题不一致的SKU） 2. 实现了相关逻辑父ASIN缺失处理：默认：当父ASIN不在变体列表中时，打印警告并丢弃整个SPU 使用 --keep-spu-if-parent-missing：保留SPU，使用第一个变体作为主商品标题不一致处理：默认：当变体标题与主商品不一致时，打印日志并丢弃该SKU 使用 --fix-sku-if-title-mismatch：修正变体标题为主商品标题
tangwang
1 parent 148ab60d
Showing 1 changed file with 40 additions and 8 deletions Show diff stats
scripts/amazon_xlsx_to_shoplazza_xlsx.py
@@ -420,6 +420,9 @@ def main():
     parser.add_argument("--max-files", type=int, default=None, help="Limit number of xlsx files to read (for testing)")
     parser.add_argument("--max-rows-per-output", type=int, default=40000, help="Max total Excel rows per output file (including模板头部行，默认40000)")
     parser.add_argument("--max-products", type=int, default=None, help="Limit number of SPU groups to output (for testing)")
+    # 默认行为：丢弃不符合要求的数据
+    parser.add_argument("--keep-spu-if-parent-missing", action="store_false", dest="skip_spu_if_parent_missing", default=True, help="Keep SPU even if parent ASIN not found in variants (default: skip entire SPU)")
+    parser.add_argument("--fix-sku-if-title-mismatch", action="store_false", dest="skip_sku_if_title_mismatch", default=True, help="Fix SKU title to match parent instead of skipping (default: skip SKU with mismatched title)")
     args = parser.parse_args()
     if not os.path.isdir(args.input_dir):
@@ -481,24 +484,53 @@ def main():
         if parent_variant:
             variants = [parent_variant] + other_variants
         else:
-            # 如果找不到父ASIN对应的变体，打印错误日志
+            # 如果找不到父ASIN对应的变体
             print(
                 f"WARN: Parent ASIN not found in variants: SPU={spu_id}, "
                 f"variant_count={len(variants)}, first_ASIN={variants[0].get('ASIN') if variants else 'N/A'}",
                 flush=True,
             )
+            # 根据开关决定是否丢弃整个SPU
+            if args.skip_spu_if_parent_missing:
+                print(
+                    f"SKIP entire SPU due to missing parent ASIN: SPU={spu_id}",
+                    flush=True,
+                )
+                continue
-        # 统一变体标题：如果与主商品不一致，改为与主商品一致
+        # 处理变体标题：如果与主商品不一致，根据开关决定修正或丢弃
         main_title = variants[0].get("商品标题") or ""
+        filtered_variants = []
         for v in variants:
             title = v.get("商品标题") or ""
             if main_title and title and title != main_title:
-                print(
-                    f"FIX variant title mismatch: SPU={spu_id}, ASIN={v.get('ASIN')}, "
-                    f"main_title='{main_title}', variant_title='{title}' -> using main_title",
-                    flush=True,
-                )
-                v["商品标题"] = main_title  # 统一为主商品标题
+                if args.skip_sku_if_title_mismatch:
+                    # 丢弃标题不一致的SKU
+                    print(
+                        f"SKIP SKU due to title mismatch: SPU={spu_id}, ASIN={v.get('ASIN')}, "
+                        f"main_title='{main_title}', variant_title='{title}'",
+                        flush=True,
+                    )
+                    continue
+                else:
+                    # 修正标题
+                    print(
+                        f"FIX variant title mismatch: SPU={spu_id}, ASIN={v.get('ASIN')}, "
+                        f"main_title='{main_title}', variant_title='{title}' -> using main_title",
+                        flush=True,
+                    )
+                    v["商品标题"] = main_title  # 统一为主商品标题
+            filtered_variants.append(v)
+        
+        # 如果所有变体都被过滤掉，跳过整个SPU
+        if not filtered_variants:
+            print(
+                f"SKIP entire SPU: all variants filtered out, SPU={spu_id}",
+                flush=True,
+            )
+            continue
+        
+        variants = filtered_variants
         spu_count += 1
         if args.max_products is not None and spu_count > int(args.max_products):