From 3b7ac4a33626f8ead7819018ade02689d2fc7ffd Mon Sep 17 00:00:00 2001
From: Cxx0822 <1556464090@qq.com>
Date: Tue, 14 Jul 2026 19:14:33 +0800
Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E5=B7=A5=E5=85=B7=E6=96=B9?=
 =?UTF-8?q?=E6=B3=95=20=E5=A2=9E=E5=8A=A0list=E6=95=B0=E6=8D=AE=E5=85=BC?=
 =?UTF-8?q?=E5=AE=B9=E6=80=A7?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 tools/base.py | 40 +++++++++++++++++++++++++++++++++++-----
 1 file changed, 35 insertions(+), 5 deletions(-)

diff --git a/tools/base.py b/tools/base.py
index ec58f20..a972b72 100644
--- a/tools/base.py
+++ b/tools/base.py
@@ -1,3 +1,4 @@
+import numpy as np
 from langchain_core.tools import tool
 import requests
 import pandas as pd
@@ -80,12 +81,21 @@ def filter_data(domain: str, filters: list[dict] | None = None, search: list[str
                 continue
 
             col = df[field]
+
             # 处理多值情况
-            if col.apply(lambda x: isinstance(x, (list, tuple, set))).any():
-                if op == "contains":
+            is_multi = col.apply(lambda x: isinstance(x, (list, tuple, set))).any()
+
+            if is_multi:
+                if op == "eq":
+                    col = col.apply(lambda x: value in x if isinstance(x, (list, tuple, set)) else False)
+                elif op == "ne":
+                    col = col.apply(lambda x: value not in x if isinstance(x, (list, tuple, set)) else True)
+                elif op == "contains":
                     col = col.apply(lambda x: value in x if isinstance(x, (list, tuple, set)) else False)
                 elif op == "not_contains":
                     col = col.apply(lambda x: value not in x if isinstance(x, (list, tuple, set)) else True)
+                else:
+                    continue
                 df = df[col]
                 continue
 
@@ -107,8 +117,17 @@ def filter_data(domain: str, filters: list[dict] | None = None, search: list[str
             for field in cfg.search_fields:
                 if field not in df.columns:
                     continue
-                # 如果包含就表示命中
-                mask |= df[field].astype(str).str.contains(kw, na=False)
+
+                # list 转成可检索的字符串
+                col_str = df[field].apply(
+                    lambda x: " ".join(map(str, x))
+                    if isinstance(x, (list, tuple, set))
+                    else str(x)
+                )
+
+                # 如果包含则命中
+                mask |= col_str.str.contains(kw, na=False)
+
         df = df[mask]
 
     filtered_len = len(df)
@@ -193,8 +212,19 @@ def inspect_data(domain: str, sort_by: str | None = None, ascending: bool = Fals
                 if fld not in row:
                     continue
                 val = row[fld]
-                if pd.isna(val) or val == "":
+
+                # 数组 / Series
+                if isinstance(val, (np.ndarray, pd.Series)) or (isinstance(val, (list, tuple))):
+                    if len(val) == 0:
+                        val = "未填写"
+                    else:
+                        val = "、".join(map(str, val))
+                elif pd.isna(val):
                     val = "未填写"
+                elif val == "":
+                    val = "未填写"
+                else:
+                    val = str(val)
                 lines.append(f"- {cfg.get_chinese_name(fld)}：{val}")
         print(f"[TOOL] inspect_data | rendered directly")
         return "\n".join(lines)