Add feeder_school_info + summer_program_recommendation tools (v1.1)

agent/agentic_engine.py

@@ -248,6 +248,53 @@ TOOLS = [
             },
         },
     },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_feeder_school_info",
+            "description": (
+                "查询某所大学的大陆高中 feeder school 分布（哪些高中送了多少人），"
+                "或查询某所高中的录取档案（该高中历史上送往哪些大学、T10/T15 率）。"
+                "数据来源：meiben 27000+ 条大陆录取记录。"
+            ),
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "school_name": {
+                        "type": "string",
+                        "description": "大学英文名（如 'Stanford University'）",
+                    },
+                    "hs_name": {
+                        "type": "string",
+                        "description": "高中中文名（可选），若提供则返回该高中的录取档案",
+                    },
+                },
+                "required": ["school_name"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_summer_program_recommendation",
+            "description": (
+                "根据学生的专业方向和目标档次，推荐适合的夏校和竞赛项目。"
+                "数据来源：meiben 录取数据中提取的夏校/竞赛给力指数（1-10分）。"
+            ),
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "major": {"type": "string", "description": "专业方向"},
+                    "sat": {"type": "number", "description": "SAT 分数（可选）"},
+                    "target_tier": {
+                        "type": "string",
+                        "description": "目标档次：T10 / T15 / T20",
+                    },
+                },
+                "required": ["major"],
+            },
+        },
+    },
 ]
 
 # ─── 工具执行层 ────────────────────────────────────────────────────────────────
@@ -275,6 +322,10 @@ def _execute_tool(tool_name: str, tool_input: Dict) -> str:
             return _tool_calibrate_school_list(**tool_input)
         elif tool_name == "get_special_case_patterns":
             return _tool_get_special_case_patterns(**tool_input)
+        elif tool_name == "get_feeder_school_info":
+            return _tool_get_feeder_school_info(**tool_input)
+        elif tool_name == "get_summer_program_recommendation":
+            return _tool_get_summer_program_recommendation(**tool_input)
         else:
             return json.dumps({"error": f"Unknown tool: {tool_name}"})
     except Exception as e:
@@ -627,6 +678,169 @@ def _tool_get_special_case_patterns(pattern_type: str, major: str = "") -> str:
     }, ensure_ascii=False)
 
 
+def _tool_get_feeder_school_info(school_name: str, hs_name: str = None) -> str:
+    """
+    查询某所大学的大陆高中 feeder school 分布，或某所高中的录取档案。
+    school_name: 大学英文名（如 'Stanford University'）
+    hs_name: 可选，高中中文名（如 '北京师范大学附属实验中学国际部'），若提供则返回该高中的录取档案
+    """
+    feeder_file = DATA_DIR / "meiben_feeder_kb.json"
+    if not feeder_file.exists():
+        return json.dumps({"error": "meiben_feeder_kb.json not found"})
+
+    with open(feeder_file) as f:
+        kb = json.load(f)
+
+    # 查询高中档案
+    if hs_name:
+        # 模糊匹配高中名
+        hs_profiles = kb.get("high_school_profiles", {})
+        matched = None
+        for name, profile in hs_profiles.items():
+            if hs_name in name or name in hs_name:
+                matched = (name, profile)
+                break
+        if not matched:
+            # 尝试更宽松的匹配
+            hs_lower = hs_name.lower().replace(" ", "")
+            for name, profile in hs_profiles.items():
+                if hs_lower in name.lower().replace(" ", ""):
+                    matched = (name, profile)
+                    break
+        if matched:
+            name, profile = matched
+            return json.dumps({
+                "hs_name": name,
+                "province": profile.get("province"),
+                "total_offers": profile.get("total_offers"),
+                "t10_ivy_offers": profile.get("t10_ivy_offers"),
+                "t15_offers": profile.get("t15_offers"),
+                "t10_rate": profile.get("t10_rate"),
+                "t15_rate": profile.get("t15_rate"),
+                "top_universities": profile.get("top_universities", [])[:12],
+                "years_active": profile.get("years_active"),
+            }, ensure_ascii=False)
+        return json.dumps({"error": f"High school '{hs_name}' not found in database"})
+
+    # 查询大学的 feeder school 分布
+    feeder_kb = kb.get("feeder_school_kb", {})
+    # 模糊匹配大学名
+    matched_uni = None
+    for uni, data in feeder_kb.items():
+        if school_name.lower() in uni.lower() or uni.lower() in school_name.lower():
+            matched_uni = (uni, data)
+            break
+    if not matched_uni:
+        # 尝试缩写匹配
+        abbrev_map = {
+            "stanford": "Stanford University",
+            "harvard": "Harvard University",
+            "mit": "Massachusetts Institute of Technology",
+            "yale": "Yale University",
+            "princeton": "Princeton University",
+            "columbia": "Columbia University",
+            "upenn": "University of Pennsylvania",
+            "penn": "University of Pennsylvania",
+            "duke": "Duke University",
+            "dartmouth": "Dartmouth College",
+            "brown": "Brown University",
+            "cornell": "Cornell University",
+            "washu": "Washington University in St.Louis",
+            "wustl": "Washington University in St.Louis",
+        }
+        for abbrev, full_name in abbrev_map.items():
+            if abbrev in school_name.lower():
+                for uni, data in feeder_kb.items():
+                    if full_name.lower() in uni.lower():
+                        matched_uni = (uni, data)
+                        break
+                if matched_uni:
+                    break
+
+    if not matched_uni:
+        available = list(feeder_kb.keys())[:20]
+        return json.dumps({"error": f"University '{school_name}' not found", "available_universities": available})
+
+    uni_name, data = matched_uni
+    # 同时返回省市偏好
+    province_pref = kb.get("province_preference", {}).get(uni_name, {})
+
+    return json.dumps({
+        "university": uni_name,
+        "total_mainland_offers": data.get("total_offers"),
+        "top_feeder_schools": data.get("top_feeder_schools", [])[:10],
+        "province_distribution": data.get("province_distribution", {}),
+        "year_trend": data.get("year_trend", {}),
+        "province_preference": province_pref,
+        "insight": f"{uni_name} 在大陆的主要来源高中集中在 {', '.join(list(data.get('province_distribution', {}).keys())[:3])} 等省市。"
+    }, ensure_ascii=False)
+
+
+def _tool_get_summer_program_recommendation(major: str, sat: float = None,
+                                             hs_type: str = "国际高中",
+                                             target_tier: str = "T15") -> str:
+    """
+    根据学生背景推荐适合的夏校和竞赛项目。
+    major: 专业方向
+    sat: SAT 分数（可选）
+    hs_type: 高中类型
+    target_tier: 目标档次 T10/T15/T20
+    """
+    feeder_file = DATA_DIR / "meiben_feeder_kb.json"
+    scores_file = DATA_DIR / "summer_program_scores.json"
+
+    programs = []
+
+    # 从 meiben_feeder_kb 的 summer_program_db 获取项目列表
+    if feeder_file.exists():
+        with open(feeder_file) as f:
+            kb = json.load(f)
+        all_programs = kb.get("summer_program_db", [])
+
+        # 按专业方向过滤
+        major_lower = major.lower()
+        major_keywords = {
+            "cs": ["computer", "coding", "programming", "software", "ai", "data"],
+            "bio": ["biology", "biomedical", "life science", "medicine", "health"],
+            "physics": ["physics", "astronomy", "astrophysics"],
+            "econ": ["economics", "business", "finance", "policy"],
+            "env": ["environment", "sustainability", "climate", "earth"],
+            "humanities": ["history", "literature", "philosophy", "writing", "language"],
+            "math": ["math", "statistics", "quantitative"],
+            "engineering": ["engineering", "mechanical", "electrical", "civil"],
+        }
+
+        # 确定专业类别
+        major_cat = "general"
+        for cat, keywords in major_keywords.items():
+            if any(kw in major_lower for kw in keywords):
+                major_cat = cat
+                break
+
+        # 按给力指数排序，过滤出高质量项目
+        min_geili = 7.0 if target_tier == "T10" else 6.0 if target_tier == "T15" else 5.0
+        filtered = [p for p in all_programs if p.get("avg_geili_score", 0) >= min_geili]
+        filtered.sort(key=lambda x: x.get("avg_geili_score", 0), reverse=True)
+
+        # 分类型返回
+        summer_progs = [p for p in filtered if p.get("type") == "summer_program"][:8]
+        competitions = [p for p in filtered if p.get("type") == "competition"][:6]
+
+        programs = {
+            "summer_programs": summer_progs,
+            "competitions": competitions,
+            "total_filtered": len(filtered),
+            "filter_criteria": f"给力指数 >= {min_geili}（{target_tier} 目标）",
+        }
+
+    return json.dumps({
+        "major": major,
+        "target_tier": target_tier,
+        "recommendations": programs,
+        "note": "给力指数 10=顶级（RSI/ISEF），8-9=强力，6-7=良好，5以下=一般",
+    }, ensure_ascii=False)
+
+
 # ─── SSE 工具 ───────��──────────────────────────────────────────────────────────
 
 def _sse(event: str, data: Any) -> str: