diff --git a/README.md b/README.md
index 60d0b19..798ebc9 100644
--- a/README.md
+++ b/README.md
@@ -1234,12 +1234,11 @@ curl -X GET "http://127.0.0.1:8000/api/test/pron/start?count=5&lang=fr-FR" \
 
 ### 15. 部署说明
 
-1. 安装依赖: `pip install -r requirements.txt`
-2. 配置数据库连接 (settings.py)
-3. 配置百度翻译API密钥 (BAIDU_APPID, BAIDU_APPKEY)
-4. 启动Redis服务
-5. 运行数据库迁移
-6. 启动服务: `python main.py`
+1. 安装依赖：`uv pip install -r requirements.txt`（或使用 `pip install -r requirements.txt`）
+2. 配置环境：在项目根目录准备 `.env`，写入数据库、Redis、邮件、翻译等密钥（字段参考 `settings.Settings`）
+3. 确保 MySQL 与 Redis 服务已启动并与 `.env` 中的连接信息匹配
+4. 初始化数据库（首次部署）：`aerich upgrade`
+5. 启动服务：`uvicorn main:app --host 0.0.0.0 --port 8000`（开发环境可加 `--reload`）
 
 ---
 
diff --git a/app/api/search_dict/routes.py b/app/api/search_dict/routes.py
index 2e9288b..e27ee99 100644
--- a/app/api/search_dict/routes.py
+++ b/app/api/search_dict/routes.py
@@ -237,44 +237,105 @@ async def search_proverb(proverb_id: int = Form(...), user=Depends(get_current_u
 
 
 @dict_search.post("/search/list/idiom")
-async def search_idiom_list(query_idiom: ProverbSearchRequest, user=Depends(get_current_user)):
+async def search_idiom_list(
+        query_idiom: ProverbSearchRequest,
+        user=Depends(get_current_user)
+):
+    """日语成语检索接口（带语言检测与分类逻辑）"""
+
     if query_idiom.dict_language == "fr":
         raise HTTPException(status_code=400, detail="Dict language Error")
 
+    # 语言检测
     mapping_query, lang, is_kangji = await service.detect_language(text=query_idiom.query)
     query = query_idiom.query
 
-    # ✅ 并发任务列表
-    tasks = [
-        service.suggest_proverb(
-            query=query,
-            lang=lang,
-            model=IdiomJp,
-            search_field="search_text",
-            target_field="text",
-        )
-    ]
+    # 初始化任务列表
+    tasks = []
 
-    if lang == "zh" and is_kangji:
-        # jp_query = all_in_kana(text=query_idiom.query)
+    # ✅ 分类逻辑入口
+    # --- 1️⃣ 日语输入 ---
+    if lang == "jp":
+        if is_kangji:
+            # 有汉字的日语词 → 进行 text + search_text 双字段匹配
+            tasks.append(
+                service.suggest_proverb(
+                    query=query,
+                    lang="jp",
+                    model=IdiomJp,
+                    search_field="text",
+                )
+            )
+            tasks.append(
+                service.suggest_proverb(
+                    query=query,
+                    lang="jp",
+                    model=IdiomJp,
+                    search_field="search_text",
+                    target_field="text",
+                )
+            )
+        else:
+            # 平假名或片假名纯日文
+            tasks.append(
+                service.suggest_proverb(
+                    query=query,
+                    lang="jp",
+                    model=IdiomJp,
+                    search_field="search_text",
+                    target_field="text",
+                )
+            )
+
+    # --- 2️⃣ 中文输入 ---
+    elif lang == "zh":
         tasks.append(
             service.suggest_proverb(
-                query=mapping_query,
+                query=query,
+                lang="zh",
+                model=IdiomJp,
+                search_field="search_text",
+                target_field="text",
+            )
+        )
+        # 若输入为汉字，可能存在对应的日语原型（mapping_query）
+        if is_kangji and mapping_query:
+            tasks.append(
+                service.suggest_proverb(
+                    query=mapping_query,
+                    lang="jp",
+                    model=IdiomJp,
+                    search_field="text",
+                )
+            )
+
+    # --- 3️⃣ 其他语言（默认回退） ---
+    else:
+        tasks.append(
+            service.suggest_proverb(
+                query=query,
                 lang="jp",
                 model=IdiomJp,
-                search_field="text",
+                search_field="search_text",
+                target_field="text",
             )
         )
 
-    # ✅ 并发执行（返回结果顺序与任务顺序一致）
+    # ✅ 并发执行任务
     results = await asyncio.gather(*tasks)
 
-    # ✅ 合并结果
-    result = results[0]
-    if len(results) > 1:
-        result[:0] = results[1]  # 将中文映射查询结果插到最前面
+    # ✅ 结果合并
+    if not results:
+        return {"list": []}
 
-    return {"list": result}
+    merged = []
+    for res in results:
+        merged.extend(res)
+
+    # ✅ 去重（如果你希望返回唯一成语）
+    unique_list = {item["text"]: item for item in merged}.values()
+
+    return {"list": list(unique_list)}
 
 
 @dict_search.post("/search/idiom")