feat:南网指标短语替换

Dockerfile

@@ -3,7 +3,7 @@ WORKDIR /app
 COPY . /app
 ENV TZ=Asia/Shanghai \
     LANG=C.UTF-8
-RUN rm -rf  logs .git .idea .venv && apt-get update && apt-get install -y vim curl openssl libssl-dev && pip install -r requirement.txt -i https://mirrors.aliyun.com/pypi/simple/
+RUN rm -rf  logs .git .idea .venv && apt-get update && apt-get install -y vim curl openssl libssl-dev sqlite3 && pip install -r requirement.txt -i https://mirrors.aliyun.com/pypi/simple/
 RUN mkdir -p /app/logs && touch /app/logs/sqlbot.log && rm -rf *.whl
 EXPOSE 8084
 CMD ["python","main_service.py"]

graph_chat/gen_sql_chart_agent.py

@@ -12,9 +12,9 @@ from decouple import config
 from template.template import get_base_template
 from util.utils import extract_nested_json, check_and_get_sql, get_chart_type_from_sql_answer
 from datetime import datetime
-from util import train_ddl,utils
-
+from util import train_ddl, utils
 
+from util.pre_question_answer import merged_dict
 
 gen_history_llm = ChatOpenAI(
     model=config('CHAT_MODEL_NAME', default=''),
@@ -37,6 +37,8 @@ gen_sql_llm = ChatOpenAI(
 '''
 用于生成sql,生成图表的agent，上下文
 '''
+
+
 class SqlAgentState(TypedDict):
     user_question: str
     rewritten_user_question: Optional[str]
@@ -52,9 +54,13 @@ class SqlAgentState(TypedDict):
 
     sql_retry_count: int
     chart_retry_count: int
+
+
 '''
 基于上下文，历史消息，重写用户问题
 '''
+
+
 def _rewrite_user_question(state: SqlAgentState) -> dict:
     logger.info(f"user:{state.get('user_id', '1')} ---------------进入 _rewrite_user_question 节点---------------")
     user_question = state['user_question']
@@ -76,6 +82,17 @@ def _gen_sql(state: SqlAgentState) -> dict:
     logger.info(f"user:{state.get('user_id', '1')} ---------------进入 _gen_sql 节点---------------")
     question = state.get('rewritten_user_question', state['user_question'])
     service = vn
+
+    # 预制问题-答案,后期如模型效果好，可以考虑删除
+    if state.get('user_question', '') in merged_dict.keys():
+        logger.info(f"预制问题-答案-】-{state.get('user_question', '')}-hit")
+        r = merged_dict[state.get('user_question', '')]
+        r['success'] = True
+        return {'gen_sql_result': r, "sql_retry_count": 0, "gen_sql_error": ""}
+    if state.get('rewritten_user_question', '') in merged_dict.keys():
+        r = merged_dict[state.get('rewritten_user_question', '')]
+        r['success'] = True
+        return {'gen_sql_result': r, "sql_retry_count": 0, "gen_sql_error": ""}
     question_sql_list = service.get_similar_question_sql(question)
     if question_sql_list and len(question_sql_list) > 3:
         question_sql_list = question_sql_list[:3]
@@ -97,7 +114,7 @@ def _gen_sql(state: SqlAgentState) -> dict:
         logger.info(f"gensql result: {result}")
         result = orjson.loads(result)
         retry = state.get("sql_retry_count", 0) + 1
-        return {'gen_sql_result': result, "sql_retry_count": retry,"gen_sql_error":""}
+        return {'gen_sql_result': result, "sql_retry_count": retry, "gen_sql_error": ""}
     except Exception as e:
         import traceback
         traceback.print_exc()
@@ -122,7 +139,8 @@ def _gen_chart(state: SqlAgentState) -> dict:
             [{'role': 'system', 'content': sys_char_temp}, {'role': 'user', 'content': user_char_temp}]).text()
         rr = utils.remove_think_tags(rr)
         retry = state.get("chart_retry_count", 0) + 1
-        return {"chart_retry_count": retry, 'gen_chart_result': orjson.loads(extract_nested_json(rr)),"gen_chart_error":""}
+        return {"chart_retry_count": retry, 'gen_chart_result': orjson.loads(extract_nested_json(rr)),
+                "gen_chart_error": ""}
     except Exception as e:
         import traceback
         traceback.print_exc()
@@ -183,4 +201,4 @@ memory = MemorySaver()
 sql_chart_agent = workflow.compile(checkpointer=memory)
 # png_data=sql_chart_agent .get_graph().draw_mermaid_png()
 # with open("D://graph.png", "wb") as f:
-#     f.write(png_data)
+#     f.write(png_data)

main_service.py

@@ -320,8 +320,8 @@ def update_question_feed_back():
         update_user_feedBack(id, '', user_feedback)
         return jsonify({"type": "success"})
     except Exception as e:
-        logger.error(f"查询预制问题失败 failed:{e}")
-        return jsonify({"type": "error", "error": f'查询预制问题失败:{str(e)}'})
+        logger.error(f"保存问题反馈失败 failed:{e}")
+        return jsonify({"type": "error", "error": f'保存问题反馈失败:{str(e)}'})
 
 
 @app.flask_app.route("/yj_sqlbot/api/v0/gen_graph_question", methods=["GET"])
@@ -383,8 +383,9 @@ def gen_graph_question():
         return jsonify(data)
     except Exception as e:
         traceback.print_exc()
-        logger.error(f"查询预制问题失败 failed:{e}")
-        return jsonify({"type": "error", "error": f'查询预制问题失败:{str(e)}'})
+        logger.error(f"sql及图表生成环节异常 failed:{str(e)}")
+        update_conversation(id=id, answer={'type_error': 'error', 'error':f'sql及图表生成环节异常:{str(e)}' })
+        return jsonify({"type": "error", "error": f'sql及图表生成环节异常:{str(e)}'})
 
 
 @app.flask_app.route("/yj_sqlbot/api/v0/run_sql_3", methods=["GET"])
@@ -438,9 +439,9 @@ def run_sql_3():
         summary = rr.get('summary', '')
         run_sql_error = rr.get('run_sql_error', '')
         logger.debug(f"data type  is {type(data)} data is {data} summary is {summary}")
-        if data and not run_sql_error:
+        if not run_sql_error:
             update_conversation(id=id, answer={'data': data,'summary':summary})
-        elif run_sql_error:
+        else:
             update_conversation(id=id,  answer={'type_error':'error','error':"sql执行失败"})
         logger.info(f"run_Sql finish run_sql_error => {run_sql_error}")
         return jsonify(
@@ -452,7 +453,8 @@ def run_sql_3():
         )
     except Exception as e:
         logger.error(f"run sql failed:{e}")
-        return jsonify({"type": "sql_error", "error": str(e)})
+        update_conversation(id=id, answer={'type_error': 'error', 'error': f'run sql环节异常 {str(e)}'})
+        return jsonify({"type": "sql_error", "error":f'run sql环节异常 {str(e)}'})
 
 
 @app.flask_app.route("/yj_sqlbot/api/v0/get_history", methods=["GET"])

service/conversation_service.py

@@ -1,8 +1,8 @@
 import json
 from datetime import datetime
-import pandas as pd
 from db_util.db_main import Conversation, SqliteSqlalchemy
 import logging
+from util.utils import custom_json_serializer
 
 logger = logging.getLogger(__name__)
 
@@ -72,11 +72,27 @@ def update_conversation(id: str, sql=None, question=None, chart_cfg=None, meta=N
                 meta = json.dumps(meta, ensure_ascii=False)
             update_data['meta'] = meta
         if answer is not None:
-            if isinstance(answer, dict):
-                logger.info("answer dict")
-                answer_df = pd.DataFrame(answer)
-                answer = answer_df.to_json(orient='records', date_format='iso', force_ascii=False)
-                # answer = json.dumps(answer, ensure_ascii=False)
+            if isinstance(answer, (dict, list)):
+                # 直接使用JSON序列化，避免DataFrame转换
+                try:
+                    logger.info("json序列化")
+                    answer = json.dumps(answer, ensure_ascii=False, default=custom_json_serializer)
+                    logger.info(f"answer JSON: {answer}")
+                except Exception as e:
+                    logger.error(f"JSON序列化失败,字符串化: {e}")
+                    answer = str(answer)
+            else:
+                logger.info("其他类型直接转换为字符串")
+                answer = str(answer)
+            # if isinstance(answer, dict):
+            #     logger.info("answer dict")
+            #     answer_df = pd.DataFrame(answer)
+            #     # answer_df = answer_df[0]
+            #     answer = answer_df.to_json(orient='records', date_format='iso', force_ascii=False)
+            #     answer = answer.rstrip(']')
+            #     answer = answer.lstrip('[')
+            #     logger.info(f"answer {answer}")
+            #     # answer = json.dumps(answer, ensure_ascii=False)
             update_data['answer'] = answer
         if update_data:
             result = session.query(Conversation).filter(Conversation.id == id).update(update_data)

service/cus_vanna_srevice.py

@@ -492,8 +492,8 @@ class CustomQdrant_VectorStore(Qdrant_VectorStore):
         request_body = {
             "model": self.embedding_model_name,
             "encoding_format": "float",
-            # "input": [data],
-            "sentences":[data]
+            "input": [data],
+            #"sentences":[data]
         }
         #硅基流动的beg3
         # request_body = {
@@ -510,27 +510,23 @@ class CustomQdrant_VectorStore(Qdrant_VectorStore):
         # logger.info(f"request_body:{request_body}")
         request_body.update(kwargs)
         #AI中台的beg3
-        response = requests.post(
-            url=f"{self.embedding_api_base}",
-            json=request_body,
-            headers={"Authorization": f"Bearer {self.embedding_api_key}", 'Content-Type': 'application/json'},
-        )
+
         #硅基流动的
-        # response = requests.post(
-        #     url=f"{self.embedding_api_base}/v1/embeddings",
-        #     json=request_body,
-        #     headers={"Authorization": f"Bearer {self.embedding_api_key}", 'Content-Type': 'application/json'},
-        # )
+        response = requests.post(
+             url=f"{self.embedding_api_base}/v1/embeddings",
+             json=request_body,
+             headers={"Authorization": f"Bearer {self.embedding_api_key}", 'Content-Type': 'application/json'},
+         )
         if response.status_code != 200:
             raise RuntimeError(
                 f"Failed to create the embeddings, detail: {_get_error_string(response)}"
             )
         result = response.json()
         # logger.info("embedding result:{0}".format(result))
-        embeddings = result['embeddings']
-        # embeddings = result['data'][0]['embedding']
-        return embeddings[0]
-        # return embeddings
+        #embeddings = result['embeddings']
+        embeddings = result['data'][0]['embedding']
+        #return embeddings[0]
+        return embeddings
 
 class CustomVanna(CustomQdrant_VectorStore, OpenAICompatibleLLM):
     def __init__(self, llm_config=None, vector_store_config=None):

template.yaml

@@ -75,7 +75,6 @@ template:
         </rule>
         <rule>
           <rule-title>术语标准化规则</rule-title>
-        
           <rule-detail>
             数信中心和数信部都是部门，而非单位
           </rule-detail>
@@ -86,7 +85,15 @@ template:
             internal_dept和internal_unit的字段值是内部部门和内部单位编号，而非名称，涉及内部单位或部门时需要在部门表中递归查询
             例如："internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START WITH ("name" || "shortname") LIKE '%数信中心%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%' CONNECT BY PRIOR "id" = "parentid") 
           </rule-detail>
-
+          <rule-detail>
+            时间范围参考
+              上午 (Morning): 06:00:00 - 11:59:59
+              中午 (Noon): 12:00:00 - 13:59:59
+              下午 (Afternoon): 14:00:00 - 17:59:59
+              傍晚 (Evening): 18:00:00 - 19:59:59
+              晚上 (Night): 20:00:00 - 23:59:59
+              深夜 (Late Night): 00:00:00 - 05:59:59
+          </rule-detail>
         </rule>
         <rule>
           <rule-title>聚合与计算</rule-title>
@@ -565,6 +572,7 @@ template:
           |---|---|---|
           | 数信部 | 数字信息部 | 如果匹配，则替换 |
           | 安质部 | 安全质量部 | 如果匹配，则替换 |
+          | 南网   | 南方电网  | 如果匹配，则替换 |
          
       
       # 输出格式要求

util/pre_question_answer.py

@@ -0,0 +1,224 @@
+q_a = {
+    "数信部上周有哪些人加班": {'sql': '''
+                                      SELECT DISTINCT p."id"                                                     AS "id",
+                                                      p."code"                                                   AS "工号",
+                                                      p."name"                                                   AS "姓名",
+                                                      p."work_unit"                                              AS "工作单位",
+                                                      CASE WHEN ps."status" = '1008' THEN '加班' ELSE '其他' END AS "状态"
+                                      FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                               JOIN "YJOA_APPSERVICE_DB"."t_yj_person_status" ps
+                                                    ON p."code" = ps."person_id"
+                                               JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                      WHERE p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                      WITH ("name" || "shortname" LIKE '%数字信息部%') AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
+                                      CONNECT BY PRIOR "id" = "parentid")
+                                             AND p."dr" = 0
+                                             AND ps."status" = '1008'
+                                             AND ps."date_value" >= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE
+                                               , 'd') - 1 + 7)
+                                             AND ps."date_value" <= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE
+                                               , 'd') - 1) LIMIT 1000
+                                      ''', 'chart-type': 'table'}
+    ,
+    "数信部上周有哪些人迟到": {'sql':'''
+                              SELECT p."id"                   AS "id",
+                                     p."code"                 AS "工号",
+                                     p."name"                 AS "姓名",
+                                     p."work_unit"            AS "工作单位",
+                                     CASE
+                                         WHEN ps."status" = '1006' THEN '迟到早退'
+                                         WHEN ps."status" = '1009' THEN '迟到'
+                                         WHEN ps."status" = '6002' THEN 'am迟到pm在岗'
+                                         WHEN ps."status" = '6004' THEN 'am迟到pm缺勤'
+                                         ELSE ps."status" END AS "状态"
+                              FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                       INNER JOIN "YJOA_APPSERVICE_DB"."t_yj_person_status" ps
+                                                  ON p."code" = ps."person_id"
+                              WHERE ps."status" IN ('1006', '1009', '6002', '6004')
+                                AND ps."date_value" >= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE, 'd') - 1 + 7)
+                                AND ps."date_value" <= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE, 'd') - 1)
+                                AND p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                              WITH ("name" || "shortname" LIKE '%数字信息部%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%')
+                              CONNECT BY PRIOR "id" = "parentid") AND p."dr" = 0 AND ps."dr" = 0 LIMIT 1000
+                              ''','chart-type': 'table'},
+    "数信部本周有哪些人旷工": {'sql':'''
+                              SELECT p."id"                                                          AS "id",
+                                     p."code"                                                        AS "工号",
+                                     p."name"                                                        AS "姓名",
+                                     p."work_unit"                                                   AS "工作单位",
+                                     CASE WHEN ps."status" = '1005' THEN '旷工' ELSE ps."status" END AS "状态"
+                              FROM "YJOA_APPSERVICE_DB"."t_yj_person_status" ps
+                                       INNER JOIN "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                                  ON p."code" = ps."person_id"
+                              WHERE ps."status" = '1005'
+                                AND ps."date_value" >= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE, 'd') - 1)
+                                AND ps."date_value" <= TRUNC(SYSDATE) - (TO_CHAR(SYSDATE, 'd') - 1) + 7
+                                AND ps."dr" = 0
+                                AND p."dr" = 0
+                                AND p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                              WITH ("name" || "shortname" LIKE '%数字信息部%' OR "name" || "shortname" LIKE '%数信中心%') AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
+                              CONNECT BY PRIOR "id" = "parentid")
+                              ORDER BY ps."date_value" DESC LIMIT 1000
+                              ''','chart-type': 'table'},
+    '数信部员工连续在藏天数查询': {'sql':'''
+                                  SELECT pt."year"                     AS "年份",
+                                         p."code"                      AS "人员编号",
+                                         pt."continuous_in_tibet_days" AS "连续在藏天数",
+                                         p."name"                      AS "姓名",
+                                         o."name"                      AS "部门"
+                                  FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                           INNER JOIN "YJOA_APPSERVICE_DB"."t_yj_person_in_tibat" pt
+                                                      ON p."code" = pt."person_id"
+                                           INNER JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                  WHERE p."dr" = 0
+                                    AND pt."dr" = 0
+                                    AND p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                  WITH ("name" || "shortname" LIKE '%数字信息部%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%')
+                                  CONNECT BY PRIOR "id" = "parentid")
+                                  ORDER BY pt."year" DESC, pt."continuous_in_tibet_days" DESC LIMIT 1000
+                                  ''','chart-type': 'table'}
+    ,
+    '数信中心员工累计在藏天数前20名': {'sql':'''
+                                      SELECT p."code"                 AS "人员编号",
+                                             p."name"                 AS "姓名",
+                                             o."name"                 AS "部门",
+                                             SUM(pt."count_in_tibat") AS "累计在藏天数"
+                                      FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                               INNER JOIN "YJOA_APPSERVICE_DB"."t_yj_person_in_tibat" pt
+                                                          ON p."code" = pt."person_id"
+                                               INNER JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                      WHERE p."dr" = 0
+                                        AND pt."dr" = 0
+                                        AND p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                      WITH ("name" || "shortname" LIKE '%数信中心%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%')
+                                      CONNECT BY PRIOR "id" = "parentid")
+                                      GROUP BY p."code", p."name", o."name"
+                                      ORDER BY SUM (pt."count_in_tibat") DESC LIMIT 20
+                                      ''','chart-type': 'bar'},
+    '数信部员工昨日打卡记录查询':{'sql': '''
+                                  SELECT a."person_name"         AS "person_name",
+                                         p."code"                AS "employee_code",
+                                         a."attendance_time"     AS "attendance_time",
+                                         a."attendance_address"  AS "attendance_address",
+                                         CASE
+                                             WHEN a."status" = 1 THEN '在岗'
+                                             WHEN a."status" = 2 THEN '出差'
+                                             WHEN a."status" = 3
+                                                 THEN '休假' END AS "status_desc",
+                                         CASE
+                                             WHEN a."enter_or_exit" = 1 THEN '进'
+                                             WHEN a."enter_or_exit" = 2
+                                                 THEN '出' END   AS "entry_exit",
+                                         o."name"                AS "department_name"
+                                  FROM "YJOA_APPSERVICE_DB"."t_yj_person_attendance" a
+                                           JOIN "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                                ON a."person_id" = p."code"
+                                           JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                  WHERE a."dr" = 0
+                                    AND p."dr" = 0
+                                    AND a."attendance_time" >= TRUNC(SYSDATE - 1)
+                                    AND a."attendance_time" < TRUNC(SYSDATE)
+                                    AND o."id" IN (SELECT o2."id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" o2
+                                  START WITH (o2."name" LIKE '%数字信息部%'
+                                          OR o2."shortname" LIKE '%数字信息部%')
+                                         AND o2."dr" = 0
+                                         AND o2."enable" = 1
+                                         AND o2."code" LIKE '%CYJ%'
+                                  CONNECT BY PRIOR o2."id" = o2."parentid")
+                                  ORDER BY a."attendance_time" DESC LIMIT 1000
+                                  ''','chart-type': 'table'},
+    '数信中心员工昨日考勤查询':{'sql': '''
+                                SELECT p."code"                 AS "工号",
+                                       p."name"                 AS "姓名",
+                                       o."name"                 AS "部门名称",
+                                       ps."date_value"          AS "日期",
+                                       CASE
+                                           WHEN ps."status" = '1001' THEN '在岗'
+                                           WHEN ps."status" = '1002' THEN '出差'
+                                           WHEN ps."status" = '1003' THEN '休假,请假'
+                                           WHEN ps."status" = '1005' THEN '旷工'
+                                           WHEN ps."status" = '1006' THEN '迟到,早退'
+                                           WHEN ps."status" = '1007' THEN '休息日'
+                                           WHEN ps."status" = '1008' THEN '加班'
+                                           WHEN ps."status" = '1009' THEN '迟到'
+                                           WHEN ps."status" = '4001' THEN 'am在岗pm缺勤'
+                                           WHEN ps."status" = '4002' THEN 'am缺勤pm在岗'
+                                           WHEN ps."status" = '6001' THEN 'am在岗pm早退'
+                                           WHEN ps."status" = '6002' THEN 'am迟到pm在岗'
+                                           WHEN ps."status" = '6004' THEN 'am迟到pm缺勤'
+                                           WHEN ps."status" = '4006' THEN 'am缺勤pm早退'
+                                           ELSE ps."status" END AS "人员状态"
+                                FROM "YJOA_APPSERVICE_DB"."t_yj_person_status" ps
+                                         JOIN "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                              ON ps."person_id" = p."code"
+                                         JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                WHERE o."id" IN (SELECT id FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                WITH ("name" LIKE '%数信中心%' OR "shortname" LIKE '%数信中心%') AND "enable" = 1 AND "dr" = 0 AND "code" LIKE '%CYJ%'
+                                CONNECT BY PRIOR "id" = "parentid")
+                                       AND ps."date_value" = TO_CHAR(SYSDATE - 1
+                                         , 'YYYY-MM-DD')
+                                       AND ps."dr" = 0
+                                       AND p."dr" = 0
+                                ORDER BY ps."date_value" DESC LIMIT 1000
+                                ''','chart-type': 'table'},
+    '外部单位人数最多的前五家是哪几个': {'sql':'''
+                                        SELECT o.name AS "外部单位名称", COUNT(p.id) AS "员工人数"
+                                        FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                                 INNER JOIN "YJOA_APPSERVICE_DB"."t_pr3rl2oj_outsourcing_org_new" o
+                                                            ON p.external_unit = o.id
+                                        WHERE p.dr = 0
+                                        GROUP BY o.name
+                                        ORDER BY COUNT(p.id) DESC LIMIT 5
+                                        ''','chart-type': 'column'},
+    '数信部员工10月在林芝工作天数排行': {'sql':'''
+                                        SELECT p."code"                                                   AS "工号",
+                                               p."name"                                                   AS "姓名",
+                                               COUNT(DISTINCT TO_CHAR(a."attendance_time", 'yyyy-MM-dd')) AS "在林芝工作天数"
+                                        FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                                 LEFT JOIN "YJOA_APPSERVICE_DB"."t_yj_person_attendance" a
+                                                           ON p."code" = a."person_id"
+                                                 LEFT JOIN "YJOA_APPSERVICE_DB"."t_yj_person_ac_area" ac
+                                                           ON a."access_control_point" = ac."ac_point"
+                                        WHERE p."dr" = 0
+                                          AND a."dr" = 0
+                                          AND ac."region" = 5
+                                          AND
+                                            p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                        WITH ("name" LIKE '%数字信息部%' OR "shortname" LIKE '%数字信息部%') AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
+                                        CONNECT BY PRIOR "id" = "parentid")
+                                               AND TO_CHAR(a."attendance_time"
+                                                 , 'yyyy-MM') = '2025-10'
+                                        GROUP BY p."code", p."name"
+                                        ORDER BY "在林芝工作天数" DESC LIMIT 1000
+                                        ''','chart-type': 'bar'},
+    '数信中心截止目前满足休假条件的员工有哪些': {'sql':'''
+                                                SELECT pt."year"                     AS "年份",
+                                                       p."code"                      AS "人员编号",
+                                                       pt."continuous_in_tibet_days" AS "连续在藏天数",
+                                                       p."name"                      AS "姓名",
+                                                       o."name"                      AS "部门"
+                                                FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                                                         INNER JOIN "YJOA_APPSERVICE_DB"."t_yj_person_in_tibat" pt
+                                                                    ON p."code" = pt."person_id"
+                                                         INNER JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
+                                                WHERE p."dr" = 0
+                                                  AND pt."dr" = 0
+                                                  AND pt."continuous_in_tibet_days" >= 22
+                                                  AND p."internal_dept" IN
+                                                      (SELECT id FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+                                                WITH ("name" LIKE '%数信中心%' OR "shortname" LIKE '%数信中心%') AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
+                                                CONNECT BY PRIOR "id" = "parentid")
+                                                ORDER BY pt."continuous_in_tibet_days" DESC
+                                                ''','chart-type': 'table'}
+
+}
+q_a_extent = {
+    "数字信息部上周有哪些人加班": q_a['数信部上周有哪些人加班'],
+    "数字信息部上周有哪些人迟到": q_a['数信部上周有哪些人迟到'],
+    "数字信息部本周有哪些人旷工": q_a['数信部本周有哪些人旷工'],
+    "数字信息部员工连续在藏天数查询": q_a['数信部员工连续在藏天数查询'],
+    "数字信息部员工昨日打卡记录查询": q_a['数信部员工昨日打卡记录查询'],
+    "数字信息部10月在林芝工作天数排行": q_a['数信部员工10月在林芝工作天数排行'],
+}
+
+merged_dict = q_a | q_a_extent

util/q_a_next.py

@@ -26,7 +26,7 @@ org_and_attend_q_a = [
     }, {
         "question": "xx中心10月有哪些人迟到",
         "answer": '''
-                  SELECT p."id"                                                               AS "id",
+                  SELECT 
                          p."code"                                                             AS "code",
                          p."name"                                                             AS "name",
                          CASE WHEN ps."status" = '1006' THEN '迟到/早退' ELSE ps."status" END AS "status"
@@ -48,7 +48,7 @@ org_and_attend_q_a = [
     {
         "question": "xx中心10月有哪些人出差",
         "answer": '''
-                  SELECT p."id"                                                          AS "id",
+                  SELECT 
                          p."code"                                                        AS "code",
                          p."name"                                                        AS "name",
                          CASE WHEN ps."status" = '1002' THEN '出差' ELSE ps."status" END AS "status"
@@ -70,7 +70,7 @@ org_and_attend_q_a = [
     {
         "question": "xx中心10月有哪些人旷工",
         "answer": '''
-                  SELECT p."id"                                                          AS "id",
+                  SELECT 
                          p."code"                                                        AS "code",
                          p."name"                                                        AS "name",
                          CASE WHEN ps."status" = '1002' THEN '旷工' ELSE ps."status" END AS "status"
@@ -92,7 +92,7 @@ org_and_attend_q_a = [
     {
         "question": "xx中心10月有哪些人请假",
         "answer": '''
-                  SELECT p."id"                                                          AS "id",
+                  SELECT 
                          p."code"                                                        AS "code",
                          p."name"                                                        AS "name",
                          CASE WHEN ps."status" = '1003' THEN '请假' ELSE ps."status" END AS "status"
@@ -120,7 +120,7 @@ org_and_attend_q_a = [
                          COUNT(ps.id) AS "late_count"
                   FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
                            JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p.internal_dept = o.id
-                           JOIN "YJOA_APPSERVICE_DB"."t_yj_person_status" ps ON p.id = ps.person_id
+                           JOIN "YJOA_APPSERVICE_DB"."t_yj_person_status" ps ON p.code = ps.person_id
                   WHERE p.dr = 0
                     AND ps.dr = 0
                     AND ps.status in ('1006','1009','6002','6004')
@@ -243,7 +243,7 @@ org_and_attend_q_a = [
     {
         "question": "xx中心10月有哪些人加班",
         "answer": '''
-                  SELECT p."id"                                                          AS "id",
+                  SELECT 
                          p."code"                                                        AS "code",
                          p."name"                                                        AS "name",
                          CASE WHEN ps."status" = '1008' THEN '加班' ELSE ps."status" END AS "status"
@@ -262,24 +262,18 @@ org_and_attend_q_a = [
         "tags": ["员工", "部门", "考勤", "加班"],
         "category": "考勤管理"
     },{
-        'question':'xx部有哪些员工',
+        'question':'xx部有哪些正式员工',
         'answer':'''
-                 SELECT p."id"                                                                   AS "id",
+                 SELECT 
                         p."code"                                                                 AS "工号",
                         p."name"                                                                 AS "姓名",
                         o."name"                                                                 AS "单位",
                         CASE WHEN p."gender" = '1' THEN '男' WHEN p."gender" = '2' THEN '女' END AS "性别",
-                        CASE
-                            WHEN p."person_status" = '1' THEN '草稿'
-                            WHEN p."person_status" = '2' THEN '审批中'
-                            WHEN p."person_status" = '3' THEN '制卡中'
-                            WHEN p."person_status" = '4' THEN '已入库'
-                            WHEN p."person_status" = '5' THEN '停用' END                         AS "人员状态"
                  FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
                           JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON p."internal_dept" = o."id"
                  WHERE p."internal_dept" IN (SELECT "id" FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
                  WITH ("name" || "shortname") LIKE '%xx部%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
-                 CONNECT BY PRIOR "id" = "parentid") AND p."dr" = 0
+                 CONNECT BY PRIOR "id" = "parentid") AND p."dr" = 0 and p."person_type"='YG'
                  ORDER BY p."code" ASC LIMIT 1000
                  ''',
         "tags": ["员工", "部门",],

util/q_and_a_test1.py

@@ -224,6 +224,35 @@ question_and_answer = [
         "tags": ["员工", "学历", "部门", "人数统计"],
         "category": "学历分布统计"
     },
+    {
+        "question": "XX部门员工学历或学位排行",
+        "answer": """
+                SELECT p."code" AS "员工编号",p."name" AS "姓名", o."name" AS "部门",
+                       CASE WHEN p."highest_degree"='1' THEN '学士学位'
+                            WHEN p."highest_degree"='2' THEN '硕士学位'
+                            WHEN p."highest_degree"='3' THEN '博士学位'
+                            ELSE '其他'  END AS "学位",
+                       p."graduate_school" AS "毕业院校"
+                FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p
+                INNER JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o ON o."id"=p."internal_dept" 
+                WHERE p."internal_dept" IN (
+                        SELECT "id" 
+                        FROM "IUAP_APDOC_BASEDOC"."org_orgs" 
+                        START WITH ("name" || "shortname" LIKE '%数信中心%' AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%') 
+                        CONNECT BY PRIOR "id" = "parentid") 
+                     AND p.dr=0 
+                     AND o.dr=0
+                     AND p."person_type" = 'YG'
+                ORDER BY 
+                   CASE WHEN p."highest_degree" = '1' THEN 1
+                        WHEN p."highest_degree" = '2' THEN 2
+                        WHEN p."highest_degree" = '3' THEN 3 
+                        ELSE 0 
+                   END DESC LIMIT 1000
+              """,
+        "tags": ["员工", "学历", "部门", "人数统计"],
+        "category": "学历分布统计"
+    },
     {
         "question": "XX部门博士多少人",
         "answer": '''
@@ -807,7 +836,34 @@ question_and_answer = [
         "category": "工作地考勤统计分析"
 
     },
-
+    {
+        "question": "本月XX中心各处室下迟到人次统计",
+        "answer": '''
+                SELECT o1."name" AS "处室名称", COUNT(DISTINCT ps."person_id") AS "迟到人数" 
+                FROM "YJOA_APPSERVICE_DB"."t_yj_person_status" ps 
+                JOIN "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database" p ON ps."person_id" = p."code" 
+                JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o1 ON p."internal_dept" = o1."id"
+                WHERE ps."status" IN ('1006','1009','6002','6004') 
+                      AND ps."date_value" LIKE '2025-11%' 
+                      AND ps."dr" = 0 
+                      AND p."dr" = 0 
+                      AND o1."dr" = 0 
+                      AND o1."enable" = 1 
+                      AND o1."code" LIKE '%CYJ%' 
+                      AND p."internal_dept"IN (
+                            SELECT id 
+                            FROM "IUAP_APDOC_BASEDOC"."org_orgs" 
+                            START WITH ("name" LIKE '%数信中心%' OR "shortname" LIKE '%数信中心%') 
+                                  AND "dr" = 0 
+                                  AND "enable" = 1 
+                                  AND "code" LIKE '%CYJ%' 
+                            CONNECT BY PRIOR "id" = "parentid")
+                GROUP BY o1."name" 
+                ORDER BY "迟到人数" DESC LIMIT 1000
+              ''',
+        "tags": ["员工", "部门", "考勤", "迟到","统计"],
+        "category": "各部门迟到统计分析"
+    },
     {
         "question": "张三9月在林芝上班期间，有多少天早退了？",
         "answer": '''

util/utils.py

@@ -2,6 +2,10 @@ import logging
 from typing import Optional
 import re
 from orjson import orjson
+import pandas as pd
+from pandas import Timestamp
+from datetime import datetime, date
+
 logger = logging.getLogger(__name__)
 keywords = {
     # "gender":{"1":"男","2":"女"},
@@ -107,3 +111,17 @@ def deal_result(data: list) -> list:
         raise Exception(f"sql执行结果处理失败：{str(e)}")
 
 
+def custom_json_serializer(obj):
+    """自定义JSON序列化器"""
+    if isinstance(obj, (Timestamp, pd.Timestamp)):
+        logger.info("format Timestamp")
+        return obj.isoformat()
+    elif isinstance(obj, (datetime, date)):
+        logger.info("format datetime")
+        return obj.isoformat()
+    elif pd.isna(obj):
+        logger.info("format NaN")
+        return None
+    else:
+        logger.error(f"unknown type {type(obj)}")
+        return str(obj)