多语言索引调整

tangwang
1 parent d7d48f52
Showing 7 changed files with 19 additions and 28 deletions Show diff stats
api/result_formatter.py
config/config.yaml
config/config_loader.py
config/tenant_config_loader.py
indexer/document_transformer.py
search/es_query_builder.py
search/searcher.py
@@ -13,7 +13,7 @@ class ResultFormatter:
     def format_search_results(
         es_hits: List[Dict[str, Any]],
         max_score: float = 1.0,
-        language: str = "zh",
+        language: str = "en",
         sku_filter_dimension: Optional[List[str]] = None
     ) -> List[SpuResult]:
         """
@@ -27,24 +27,17 @@ class ResultFormatter:
             List of SpuResult objects
         """
         results = []
-        lang = (language or "zh").lower()
+        lang = (language or "en").lower()
         if lang not in ("zh", "en"):
             lang = "en"
         def pick_lang_field(src: Dict[str, Any], base: str) -> Optional[str]:
-            """
-            从多语言对象字段中按语言选择一个值：
-              - 新结构: {base: {"zh": "...", "en": "...", ...}}
-              - 兼容旧结构: {base_zh: "...", base_en: "..."}
-            若目标语言缺失则回退到另一种。
-            """
+            """从多语言对象字段中按语言选择一个值：{base: {"zh": "...", "en": "...", ...}}"""
             obj = src.get(base)
-            if isinstance(obj, dict):
-                zh_val = obj.get("zh")
-                en_val = obj.get("en")
-            else:
-                zh_val = src.get(f"{base}_zh")
-                en_val = src.get(f"{base}_en")
+            if not isinstance(obj, dict):
+                return None
+            zh_val = obj.get("zh")
+            en_val = obj.get("en")
             if lang == "zh":
                 return zh_val or en_val
             return en_val or zh_val
@@ -319,12 +312,10 @@ class ResultFormatter:
                             is_selected = (name, value_bucket['key']) in selected_specs
                             # 使用 reverse_nested 的 product_count 统计产品数量（而不是规格条目数量）
-                            # 如果没有 product_count（兼容旧格式），回退到 doc_count
                             product_count_agg = value_bucket.get('product_count', {})
                             if product_count_agg and 'doc_count' in product_count_agg:
                                 count = product_count_agg['doc_count']
                             else:
-                                # 回退到 doc_count（兼容旧格式，但这不是我们想要的计数方式）
                                 count = value_bucket.get('doc_count', 0)
                             value = FacetValue(
@@ -83,7 +83,7 @@ query_config:
   supported_languages:
     - "zh"
     - "en"
-  default_language: "zh"
+  default_language: "en"
   # 功能开关（翻译开关由tenant_config控制）
   enable_text_embedding: true
@@ -153,7 +153,7 @@ spu_config:
 tenant_config:
   # 默认配置（未配置的租户使用此配置）
   default:
-    primary_language: "zh"
+    primary_language: "en"
     translate_to_en: true
     translate_to_zh: false
   # 租户特定配置
@@ -29,7 +29,7 @@ class IndexConfig:
 class QueryConfig:
     """Configuration for query processing."""
     supported_languages: List[str] = field(default_factory=lambda: ["zh", "en"])
-    default_language: str = "zh"
+    default_language: str = "en"
     # Feature flags
     enable_text_embedding: bool = True
@@ -231,7 +231,7 @@ class ConfigLoader:
         query_config = QueryConfig(
             supported_languages=query_config_data.get("supported_languages") or ["zh", "en"],
-            default_language=query_config_data.get("default_language") or "zh",
+            default_language=query_config_data.get("default_language") or "en",
             enable_text_embedding=query_config_data.get("enable_text_embedding", True),
             enable_query_rewrite=query_config_data.get("enable_query_rewrite", True),
             rewrite_dictionary=rewrite_dictionary,
@@ -39,7 +39,7 @@ class TenantConfigLoader:
             # 返回默认配置
             self._config = {
                 "default": {
-                    "primary_language": "zh",
+                    "primary_language": "en",
                     "translate_to_en": True,
                     "translate_to_zh": False
                 },
@@ -66,7 +66,7 @@ class TenantConfigLoader:
         else:
             logger.debug(f"Tenant {tenant_id} not found in config, using default")
             return config.get("default", {
-                "primary_language": "zh",
+                "primary_language": "en",
                 "translate_to_en": True,
                 "translate_to_zh": False
             })
@@ -87,7 +87,7 @@ class SPUDocumentTransformer:
             logger.error(f"SPU {spu_id} has no title, this may cause search issues")
         # 获取租户配置
-        primary_lang = self.tenant_config.get('primary_language', 'zh')
+        primary_lang = self.tenant_config.get('primary_language', 'en')
         # 文本字段处理（使用translator的内部逻辑自动处理多语言翻译）
         self._fill_text_fields(doc, spu_row, primary_lang)
@@ -283,7 +283,7 @@ class SPUDocumentTransformer:
         # - 当商品的类目ID在映射中不存在时，视为“不合法类目”，整条类目相关字段都不写入（当成没有类目）
         # - 仅记录错误日志，不阻塞索引流程
-        primary_lang = self.tenant_config.get('primary_language', 'zh')
+        primary_lang = self.tenant_config.get('primary_language', 'en')
         if pd.notna(spu_row.get('category_path')):
             category_path = str(spu_row['category_path'])
@@ -26,7 +26,7 @@ class ESQueryBuilder:
         source_fields: Optional[List[str]] = None,
         function_score_config: Optional[FunctionScoreConfig] = None,
         enable_multilang_search: bool = True,
-        default_language: str = "zh",
+        default_language: str = "en",
         knn_boost: float = 0.25
     ):
         """
@@ -135,7 +135,7 @@ class Searcher:
         sort_by: Optional[str] = None,
         sort_order: Optional[str] = "desc",
         debug: bool = False,
-        language: str = "zh",
+        language: str = "en",
         sku_filter_dimension: Optional[List[str]] = None,
     ) -> SearchResult:
         """
@@ -275,7 +275,7 @@ class Searcher:
         try:
             # Generate tenant-specific index name
             index_name = get_tenant_index_name(tenant_id)
-            index_name = "search_products"
+            # index_name = "search_products"
             # No longer need to add tenant_id to filters since each tenant has its own index
@@ -556,7 +556,7 @@ class Searcher:
         formatted_results = ResultFormatter.format_search_results(
             es_hits, 
             max_score,
-            language="zh",  # Default language for image search
+            language="en",  # Default language for image search
             sku_filter_dimension=None  # Image search doesn't support SKU filtering
         )