diff --git a/main_service.py b/main_service.py
index 4681deb..37854a2 100644
--- a/main_service.py
+++ b/main_service.py
@@ -57,7 +57,7 @@ def create_vana():
def init_vn(vn):
logger.info("--------------init vana-----connect to datasouce db----")
- connect_database(vn)
+ # connect_database(vn)
load_ddl_doc.add_ddl(vn)
load_ddl_doc.add_documentation(vn)
if config('IS_FIRST_LOAD', default=False, cast=bool):
@@ -98,7 +98,6 @@ def generate_sql_2():
"""
logger.info("Start to generate sql in main")
question = flask.request.args.get("question")
-
if question is None:
return jsonify({"type": "error", "error": "No question provided"})
try:
@@ -185,6 +184,7 @@ def run_sql_2(id: str, sql: str):
"""
logger.info("Start to run sql in main")
try:
+ user_id = request.args.get("user_id")
if not vn.run_sql_is_set:
return jsonify(
{
@@ -198,7 +198,7 @@ def run_sql_2(id: str, sql: str):
# print(df_count,"is type",type(df_count))
# total_count = df_count.to_dict(orient="records")[0]["total_count"]
# logger.info("Total count is {0}".format(total_count))
- df = vn.run_sql(sql=sql)
+ df = vn.run_sql_2(sql=sql)
result = df.to_dict(orient='records')
logger.info("df ---------------{0} {1}".format(result,type(result)))
return jsonify(
diff --git a/service/cus_vanna_srevice.py b/service/cus_vanna_srevice.py
index f8b7400..3c9ab01 100644
--- a/service/cus_vanna_srevice.py
+++ b/service/cus_vanna_srevice.py
@@ -26,7 +26,7 @@ class OpenAICompatibleLLM(VannaBase):
VannaBase.__init__(self, config=config_file)
# default parameters - can be overrided using config
self.temperature = 0.6
- self.max_tokens = 5000
+ self.max_tokens = 10000
if "temperature" in config_file:
self.temperature = config_file["temperature"]
@@ -125,17 +125,21 @@ class OpenAICompatibleLLM(VannaBase):
def submit_prompt(self, prompt, **kwargs) -> str:
if prompt is None:
+ print("test1")
raise Exception("Prompt is None")
if len(prompt) == 0:
+ print("test2")
raise Exception("Prompt is empty")
print(prompt)
num_tokens = 0
for message in prompt:
num_tokens += len(message["content"]) / 4
+ print("test3 {0}".format(num_tokens))
if kwargs.get("model", None) is not None:
+ print("test4")
model = kwargs.get("model", None)
print(
f"Using model {model} for {num_tokens} tokens (approx)"
@@ -148,6 +152,7 @@ class OpenAICompatibleLLM(VannaBase):
temperature=self.temperature,
)
elif kwargs.get("engine", None) is not None:
+ print("test5")
engine = kwargs.get("engine", None)
print(
f"Using model {engine} for {num_tokens} tokens (approx)"
@@ -160,6 +165,7 @@ class OpenAICompatibleLLM(VannaBase):
temperature=self.temperature,
)
elif self.config is not None and "engine" in self.config:
+ print("test6")
print(
f"Using engine {self.config['engine']} for {num_tokens} tokens (approx)"
)
@@ -171,10 +177,11 @@ class OpenAICompatibleLLM(VannaBase):
temperature=self.temperature,
)
elif self.config is not None and "model" in self.config:
+ print("test7")
print(
f"Using model {self.config['model']} for {num_tokens} tokens (approx)"
)
- print(self.config)
+ print("config is ",self.config)
response = self.client.chat.completions.create(
model=self.config["model"],
messages=prompt,
@@ -194,6 +201,7 @@ class OpenAICompatibleLLM(VannaBase):
# json=data
# )
else:
+ print("test8")
if num_tokens > 3500:
model = "kimi"
else:
@@ -271,15 +279,26 @@ class OpenAICompatibleLLM(VannaBase):
traceback.print_exc()
raise e
+ def run_sql_2(self,sql):
+ try:
+ return self.run_sql(sql)
+ except Exception as e:
+ logger.error("run_sql failed {0}".format(sql))
+ raise e
+
def generate_rewritten_question(self, last_question: str, new_question: str, **kwargs) -> str:
logger.info(f"generate_rewritten_question---------------{new_question}")
if last_question is None:
return new_question
-
+ print("last question {0}".format(last_question))
+ print("new question {0}".format(new_question))
prompt = [
self.system_message(
- "Your goal is to combine a sequence of questions into a singular question if they are related. If the second question does not relate to the first question and is fully self-contained, return the second question. Return just the new combined question with no additional explanations. The question should theoretically be answerable with a single SQL statement."),
- self.user_message(new_question),
+ "你的目标是将一系列相关问题合并成一个单一的问题。"
+ "合并准则一、如果第二个问题与第一个问题无关且本身是完整独立的，则直接返回第二个问题。"
+ "合并准则二、如果第二个问题域第一个问题相关，且要基于第一个问题的前提，请合并两个问题为一个问题，只需返回合并后的新问题，不要添加任何额外解释。"
+ "合并准则三、理论上，合并后的问题应该能够通过单个SQL语句来回答"),
+ self.user_message("First question: " + last_question + "\nSecond question: " + new_question),
]
return self.submit_prompt(prompt=prompt, **kwargs)
@@ -312,6 +331,11 @@ class CustomQdrant_VectorStore(Qdrant_VectorStore):
"model": self.embedding_model_name,
"sentences": [data],
}
+ # request_body = {
+ # "model": self.embedding_model_name,
+ # "encoding_format": "float",
+ # "input": [data],
+ # }
request_body.update(kwargs)
response = requests.post(
@@ -325,7 +349,9 @@ class CustomQdrant_VectorStore(Qdrant_VectorStore):
)
result = response.json()
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):
@@ -412,6 +438,64 @@ class TTLCacheWrapper:
return False
return self.get(id=id, field=field) is not None
+ def items(self):
+ """遍历所有未过期的缓存键值对"""
+ current_time = time.time()
+ items = []
+
+ for (id, field), expiry in self._expiry_times.items():
+ # 检查是否过期
+ if current_time <= expiry:
+ value = self.get(id=id, field=field)
+ if value is not None:
+ items.append({
+ 'id': id,
+ 'field': field,
+ 'value': value,
+ 'expires_at': expiry,
+ 'time_left': expiry - current_time
+ })
+
+ return items
+
+ def get_latest_by_id(self, id: str, limit: int = 1, field_filter: str = None):
+ """
+ 获取指定ID下时间最近的缓存项
+
+ Args:
+ id: 要查询的ID
+ limit: 返回最近几条记录，默认1条
+ field_filter: 可选的字段过滤，如只获取特定字段
+
+ Returns:
+ 按时间倒序排列的缓存项列表
+ """
+ current_time = time.time()
+ matched_items = []
+
+ # 找出该ID下所有未过期的缓存项
+ for (cache_id, field), expiry in self._expiry_times.items():
+ if cache_id == id and current_time <= expiry:
+ # 字段过滤
+ if field_filter and field != field_filter:
+ continue
+
+ value = self.get(id=id, field=field)
+ if value is not None:
+ matched_items.append({
+ 'id': id,
+ 'field': field,
+ 'value': value,
+ 'expires_at': expiry,
+ 'created_time': expiry - self.ttl, # 估算创建时间
+ 'time_left': expiry - current_time
+ })
+
+ # 按过期时间倒序排列（最近创建的排在前面）
+ matched_items.sort(key=lambda x: x['expires_at'], reverse=True)
+
+ return matched_items[:limit]
+
# 代理其他方法到原始cache
def __getattr__(self, name):
return getattr(self.cache, name)
\ No newline at end of file
diff --git a/template.yaml b/template.yaml
index 418e40a..2c02b5b 100644
--- a/template.yaml
+++ b/template.yaml
@@ -54,7 +54,7 @@ template:
生成的SQL必须避免与数据库关键字冲突。
注意列名定义和使用的先后顺序，例如：SELECT阶段定义了列名，如果GROUP BY阶段先与SELECT阶段执行时，是不许在GROUP BY阶段引用列名的。
在ORDER BY、GROUP BY、WHERE子句中不要使用SELECT中定义的别名
- 当SELECT中同时包含聚合列和非聚合列时，需要在GROUP BY中指定所有非聚合列
+ 注意！当SELECT列中同时包含聚合列（COUNT(), SUM(), AVG()）和非聚合列时，必须要在GROUP BY子句中指定所有非聚合列
递归 WITH 子句必须具有列别名列表
CONNECT BY子查询是独立的，无法访问外部查询的表别名,因此涉及CONNECT BY子查询时，里面禁止使用表别名
@@ -66,13 +66,36 @@ template:
若查询包含日期/时间字段：
- **默认行为**：若提问未指定排序，**默认按时间字段降序排序**（即最新数据在前）。
- **格式化**：若提问要求时间/日期/年月/年，且未指定格式，则分别格式化为 'yyyy-MM-dd HH:mm:ss' / 'yyyy-MM-dd' / 'yyyy-MM' / 'yyyy'，语法需适配当前数据库引擎。(达梦数据库如果时间字段是varchar类型也可以)
+ - **特殊情况**：时间格式最大为 23:59:59，禁止24:00:00 这不是合法的时间值，查询某一天的数据，结束时间应该是下一天的00:00:00
+
+ 如果涉及查询男女比例，请参考示例：SELECT CASE WHEN p."gender"='1' THEN '男员工' WHEN gender='2' THEN '女员工' END AS "性别",sum(p."gender") AS "人数".....
+
+
+ 术语标准化规则
+
+ [重要指令]解析用户问题，识别并替换所有已知的等价短语，将前面的短语换成后面的等价短语，:
+ ** "数信部" -> "数字信息部" (必须替换！)
+ ** "安质部" -> "安全质量部" (必须替换！)
+ 例如:用户：查询数信部 → SQL: LIKE '%数字信息部%'
+ 用户：查询安质部 → SQL: LIKE '%安全质量部%'
+ 例如:查询数信部有多少人->查询数字信息部有多少人
- 解析用户问题，识别并替换所有已知的等价短语，将前面的短语换成后面得等价短语，:
- ** 数信部 -> 数字信息部
- ** 安质部 -> 安全质量部
- 例如:查询数信部有多少人->查询数字信息部有多少人
+ 禁止自动联想或替换:
+ - "数信部" 不等于 "数信中心"
+ - "数字信息部" 不等于 "数字信息中心"
+
+ 数信中心和数信部都是部门，而非单位
+
+
+ external_dept和external_unit是外部部门和外部单位，字段值直接是部门名称和单位名称
+
+
+ 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")
+
+
聚合与计算
@@ -95,7 +118,7 @@ template:
若用户未指定数据条数，**查询SQL必须包含1000条的限制**。若用户指定的限制大于1000，也按1000处理。
- PostgreSQL: ... LIMIT 1000
- 涉及查询最大，最多，最小，最少等查询是，添加,limit 1
+ 涉及查询最大，最多，最小，最少等查询是，查询语句尾部添加,limit 1
---
--- C. 图表与业务理解规则
@@ -120,7 +143,12 @@ template:
图表字段注意事项
- 基于参考SQL的查询结果，请确保图表配置中的value字段与SQL结果字段别名完全一致。例如SELECT "external_unit" AS "外部单位" 图表配置的value应为“外部单位”而非“external_unit”。
+
+ 基于参考SQL的查询结果，图表配置里的所有value字段必须严格匹配SQL查询列的字段别名或字段名(有别名时优先别名)，字段数量也得对应上。
+ 例如SELECT "external_unit" AS "外部单位名" FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database"。
+ 图表配置应为："chart": {{"columns": [{{"name": "外部单位","value": "外部单位名"}}],"title": "外部单位信息","type": "table"}}
+ **注意**：一定是别名与value字段对应，如这里的别名是"外部单位名"，因此value字段也应该是"外部单位名"，value:外部单位名
+
### 以下 块帮助你理解问题及返回格式，**请勿将此块内的任何表结构用于回答用户的问题**。
@@ -192,6 +220,10 @@ template:
数信中心建设处规划发展部综合处
这些都可能是单位的名称，属于内部部门
+
+ "数信部"务必替换成"数字信息部"
+ "安质部"务必替换成"安全质量部"
+
@@ -251,7 +283,10 @@ template:
你需要在JSON内生成一个图表的标题，放在"title"字段内，这个标题需要尽量精简
- 基于参考SQL的查询结果，请确保图表配置中的value字段与SQL结果字段别名完全一致。例如SELECT "external_unit" AS "外部单位" 图表配置的value应为“外部单位”而非“external_unit”
+ 基于参考SQL的查询结果，图表配置里的所有value字段必须严格匹配SQL查询列的字段别名或字段名(有别名时优先别名)，字段数量也得对应上。
+ 例如SELECT "外部单位" AS "外部单位名" FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database"。
+ 图表配置应为："chart": {{"columns": [{{"name": "外部单位","value": "外部单位名"}}],"title": "外部单位信息","type": "table"}}
+ **注意**：一定是别名与value对应，如这里的别名是"外部单位名"，因此value字段也应该是"外部单位名"，value:外部单位名
涉及查询男女性别比例时建议采用表格或者柱状图展示，禁止采用饼状图
diff --git a/util/load_ddl_doc.py b/util/load_ddl_doc.py
index 3ced32e..5003d56 100644
--- a/util/load_ddl_doc.py
+++ b/util/load_ddl_doc.py
@@ -6,6 +6,7 @@ import random
table_ddls = [
train_ddl.person_database_ddl,train_ddl.person_status_ddl,
train_ddl.person_attendance_ddl,train_ddl.person_ac_area,
+ train_ddl.person_ac_position,
train_ddl.org_orgs_ddl
]
diff --git a/util/q_and_a_dict.py b/util/q_and_a_dict.py
index c622909..8bf1da7 100644
--- a/util/q_and_a_dict.py
+++ b/util/q_and_a_dict.py
@@ -242,7 +242,7 @@ question_and_answer = [
"answer": '''
SELECT COUNT(DISTINCT ps."person_id") AS "迟到人数"
FROM "YJOA_APPSERVICE_DB"."t_yj_person_status" ps
- WHERE ps."status" = '1006'
+ WHERE ps."status" IN ('1006','1009','6002','6004')
AND ps."date_value" LIKE '2025-09%' LIMIT 1000
'''
},
diff --git a/util/q_and_a_test1.py b/util/q_and_a_test1.py
index 629bf2d..ce6503b 100644
--- a/util/q_and_a_test1.py
+++ b/util/q_and_a_test1.py
@@ -26,17 +26,25 @@ question_and_answer = [
"category": "员工基本信息查询"
},
{
- "question": "在研发基地工作的有哪些员工",
+ "question": "在成都研发基地工作的有哪些员工",
"answer": '''
- SELECT "id" AS "id",
- "code" AS "工号",
- "name" AS "姓名",
- "internal_unit" AS "内部单位",
- "external_unit" AS "外部单位",
- "office_address" AS "办公地点"
- FROM "YJOA_APPSERVICE_DB"."t_pr3rl2oj_yj_person_database"
- WHERE ("office_address" LIKE '%研发基地%' OR "office_city" LIKE '%研发基地%')
- AND "dr" = '0' LIMIT 1000
+ SELECT DISTINCT
+ p."code" AS "工号",
+ p."name" AS "姓名",
+ p."work_unit" AS "工作单位",
+ CASE
+ WHEN ac."region" = 2 THEN '成都'
+ ELSE '其他'
+ END AS "工作地区"
+ 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 (a."attendance_address" LIKE '%成都研发基地%' OR ac."region" = 2)
+ LIMIT 1000;
''',
"tags": ["员工", "工作地", "条件筛选查询"],
"category": "员工基本信息查询"
@@ -64,6 +72,21 @@ question_and_answer = [
"tags": ["员工", "部门", "人数"],
"category": "员工统计与分析"
},
+ {
+ "question": "数信部下有多少员工",
+ "answer": """
+ SELECT COUNT(*)
+ FROM YJOA_APPSERVICE_DB.t_pr3rl2oj_yj_person_database p
+ WHERE 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
+ """,
+ "tags": ["员工", "部门", "人数"],
+ "category": "员工统计与分析"
+ },
{
"question": "XX中心下各个处室分别有多少人",
"answer": """
@@ -483,7 +506,7 @@ question_and_answer = [
"tags": ["员工", "部门", "考勤", "在岗人数"],
"category": "考勤管理"
},
- {
+ {
"question": "XX部门昨天在岗人员数量",
"answer": '''
SELECT COUNT(p.id) AS "人数"
@@ -502,18 +525,18 @@ question_and_answer = [
''',
"tags": ["员工", "部门", "考勤", "在岗人数"],
"category": "考勤管理"
- },
- {
+ },
+ {
"question": "XX中心的李四十月份休息了多少天",
"answer": """
SELECT COUNT(*) AS rest_days
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 p."name" = '谭杰明'
+ WHERE p."name" = '李四'
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%')
+ START WITH ("name" LIKE '%XX中心%' OR "shortname" LIKE '%XX中心%')
AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
CONNECT BY PRIOR "id" = "parentid"
)
@@ -550,42 +573,66 @@ question_and_answer = [
"category": "考勤管理"
},
{
- "question": "xx部门9月有多少个人迟到",
+ "question": "xx部门10月有哪些人迟到",
"answer": '''
- SELECT p."id" AS "id",
- p."code" AS "工号",
- p."name" AS "姓名",
- p."internal_unit" AS "内部单位",
- p."external_unit" AS "外部单位",
- CASE WHEN ps."status" = '1003' THEN '休假,请假' 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" = '1003'
- AND ps."date_value" LIKE '2025-10%'
- and p.internal_dept in (SELECT "id"
- FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
- WITH "name" || "shortname" LIKE '%xx部门%' 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
+ SELECT p."id" AS "id",
+ p."code" AS "工号",
+ p."name" AS "姓名",
+ p."internal_unit" AS "内部单位",
+ p."external_unit" AS "外部单位",
+ 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" LIKE '2025-10%'
+ AND p."internal_dept" IN (
+ SELECT "id"
+ FROM "IUAP_APDOC_BASEDOC"."org_orgs"
+ START WITH ("name" || "shortname") LIKE '%xx部门%'
+ 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;
''',
"tags": ["员工", "部门", "考勤", "迟到"],
"category": "考勤管理"
},
+
{
- "question": "9月有哪些员工在休假",
+ "question": "XX中心9月有哪些员工在加班",
"answer": '''
- SELECT DISTINCT p."id" AS "id",
- p."code" AS "工号",
- p."name" AS "姓名",
- p."internal_unit" AS "内部单位",
- p."external_unit" AS "外部单位"
- 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" = '1003'
- AND ps."date_value" LIKE '2025-09%' LIMIT 1000
- ''',
+ SELECT DISTINCT p."id" AS "id",
+ p."code" AS "工号",
+ p."name" AS "姓名",
+ p."internal_unit" AS "内部单位",
+ p."external_unit" AS "外部单位",
+ 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 '%XX中心%'
+ 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" LIKE '2025-09%'
+ LIMIT 1000
+ ''',
"tags": ["员工", "考勤", "休假"],
"category": "考勤管理"
},
@@ -600,9 +647,12 @@ question_and_answer = [
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" = '1004' THEN '缺勤'
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'
@@ -655,7 +705,7 @@ question_and_answer = [
INNER JOIN "IUAP_APDOC_BASEDOC"."org_orgs" o
ON p."internal_dept" = o."id"
WHERE ps."date_value" IN ('2025-10-20', '2025-10-21', '2025-10-22')
- AND ps."status" = '1006'
+ AND ps."status" IN ('1006','1009','6002','6004')
AND p."dr" = 0
AND ps."dr" = 0
AND o."dr" = 0
@@ -692,7 +742,7 @@ question_and_answer = [
AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
CONNECT BY PRIOR "id" = "parentid"
)
- AND ps."status" = '1006'
+ AND ps."status" IN ('1006','1009','6002','6004')
AND ps."date_value"
BETWEEN '2025-10-16'
AND '2025-10-22'
@@ -758,6 +808,34 @@ question_and_answer = [
"tags": ["员工", "部门", "考勤", "工作地", "区域", "工作天数统计"],
"category": "工作地考勤统计分析"
+ },
+ {
+ "question": "XX中心员工在成都工作的天数",
+ "answer": '''
+ SELECT p."code" AS "工号",
+ p."name" AS "姓名",
+ COUNT(DISTINCT TO_CHAR(a."attendance_time", 'yyyy-MM-dd')) AS "在成都工作天数"
+ 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" = 2 OR (a."attendance_address" LIKE '%成都%')
+ AND a."attendance_time" >= '2025-09-01'
+ AND a."attendance_time" < '2025-10-01'
+ AND p."internal_dept" IN (SELECT "id"
+ FROM "IUAP_APDOC_BASEDOC"."org_orgs" START
+ WITH ("name" LIKE '%XX中心%' OR "shortname" LIKE '%XX中心%')
+ AND "dr" = 0 AND "enable" = 1 AND "code" LIKE '%CYJ%'
+ CONNECT BY PRIOR "id" = "parentid"
+ )
+ GROUP BY p."code", p."name"
+ ORDER BY "在成都工作天数" DESC LIMIT 1000;
+ ''',
+ "tags": ["员工", "部门", "考勤", "工作地", "区域", "工作天数统计"],
+ "category": "工作地考勤统计分析"
+
},
{
"question": "查询张三9月在林芝的打卡记录",
@@ -779,7 +857,11 @@ question_and_answer = [
WHEN b."region" = 2 THEN '成都'
WHEN b."region" = 3 THEN '秭归'
WHEN b."region" = 4 THEN '拉萨'
- WHEN b."region" = 5 THEN '林芝' END AS "地区",
+ WHEN b."region" = 5 THEN '林芝'
+ WHEN b."region" = 6 THEN '米林'
+ WHEN b."region" = 7 THEN '派镇'
+ WHEN b."region" = 6 THEN '墨脱'
+ END AS "地区",
b."ac_point" AS "门禁点"
FROM "YJOA_APPSERVICE_DB"."t_yj_person_attendance" a
JOIN "YJOA_APPSERVICE_DB"."t_yj_person_ac_area" b
@@ -890,6 +972,34 @@ question_and_answer = [
"tags": ["员工", "个人", "考勤", "工作地", "区域", "天数", "最值"],
"category": "工作地考勤统计分析"
},
+ {
+ "question": "XX信息部9月谁在林芝工作天数最多",
+ "answer": '''
+ 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 '%XX信息部%' OR "shortname" LIKE '%XX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') = '2025-09'
+ GROUP BY p."code", p."name"
+ ORDER BY "在林芝工作天数" DESC
+ LIMIT 1
+ ''',
+ "tags": ["员工", "个人", "考勤", "工作地", "区域", "天数", "最值"],
+ "category": "工作地考勤统计分析"
+ },
{
"question": "XX中心的员工九月在林芝工作的天数排名",
"answer": '''
@@ -917,19 +1027,99 @@ question_and_answer = [
''',
"tags": ["员工", "个人", "考勤", "工作地", "区域", "天数", "排名"],
"category": "工作地考勤统计分析"
- }, {
+ },
+ {
"question": "张三10月份是否有迟到",
"answer": '''
SELECT CASE WHEN COUNT(*) > 0 THEN '是' ELSE '否' END AS "是否迟到"
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"
WHERE p."name" = '张三'
- AND ps."status" = '1006'
+ AND ps."status" IN ('1006','1009','6002','6004')
AND ps."date_value" LIKE '2025-10%'
AND ps."dr" = 0
AND p."dr" = 0
''',
- "tags": ["员工", "个人", "考勤"],
+ "tags": ["员工", "个人", "考勤","迟到"],
+ "category": "考勤"
+ },
+ {
+ "question": "李四9月份是否有缺勤",
+ "answer": '''
+ SELECT CASE
+ WHEN COUNT(*) > 0 THEN '是' ELSE '否'
+ END AS "是否缺勤", "date_value" AS "缺勤日期"
+ 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 p."name" = '李四'
+ AND ps."status" IN ('1004','4001','4002','4006','6004')
+ AND ps."date_value" LIKE '2025-09%'
+ AND p."dr" = 0 AND ps."dr" = 0
+ GROUP BY "date_value"
+ ORDER BY "date_value" DESC
+ LIMIT 1000
+ ''',
+ "tags": ["员工", "个人", "考勤","缺勤"],
+ "category": "考勤"
+ },
+ {
+ "question": "9月份XX中心哪些人存在早退现象",
+ "answer": '''
+ SELECT p."id" AS "id",
+ p."code" AS "code",
+ p."name" AS "name",
+ ps."date_value" AS "早退日期",
+ CASE
+ WHEN ps."status" = '1006' THEN '迟到早退'
+ WHEN ps."status" = '6001' THEN 'am在岗pm早退'
+ WHEN ps."status" = '4006' THEN 'am缺勤pm早退'
+ ELSE ps."status"
+ END AS "status"
+ 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" IN ('1006','6001','4006')
+ AND ps."date_value" LIKE '2025-09%'
+ 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 '%XX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
+ ''',
+ "tags": ["员工", "个人", "考勤","早退"],
+ "category": "考勤"
+ },
+ {
+ "question": "XX中心张三今天的工作了多长时间",
+ "answer": '''
+ SELECT p."code" AS "工号",
+ p."name" AS "姓名",
+ SUM(ps."work_time") AS "工作时长_分钟"
+ 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 p."name" = '张三'
+ AND p."internal_dept" IN (
+ SELECT "id"
+ FROM "IUAP_APDOC_BASEDOC"."org_orgs"
+ START WITH ("name" LIKE '%XX中心%' OR "shortname" LIKE '%XX中心%')
+ AND "dr" = 0
+ AND "enable" = 1
+ AND "code" LIKE '%CYJ%'
+ CONNECT BY PRIOR "id" = "parentid")
+ AND ps."date_value" = '2025-10-25'
+ AND p."dr" = 0
+ AND ps."dr" = 0
+ GROUP BY p."code", p."name" LIMIT 1000
+ ''',
+ "tags": ["员工", "个人", "考勤", "工作时长"],
"category": "考勤"
}
diff --git a/util/train_ddl.py b/util/train_ddl.py
index cbb1814..a0d6c97 100644
--- a/util/train_ddl.py
+++ b/util/train_ddl.py
@@ -10,6 +10,7 @@ train_document='''
根据部门名称查询部门时除了全称name，简称shortname也重要信息
internal_dept和internal_unit是部门编号不是名称，注意区分
查询部门信息时尽量使用internal_dept而非internal_unit
+ 数信部不是数信中心，两者不能等价
'''
person_database_ddl = """
@@ -102,7 +103,7 @@ person_database_ddl = """
"type": "VARCHAR(600)",
"comment": "外部部门",
"role": "dimension",
- "tags": ["组织信息", "外部部门"]
+ "tags": ["组织信息", "外部部门","外部部门名称"]
},
{
"name": "pass_type",
@@ -132,14 +133,14 @@ person_database_ddl = """
"type": "VARCHAR(108)",
"comment": "内部部门ID",
"role": "dimension",
- "tags": ["组织信息", "内部架构"]
+ "tags": ["组织信息", "内部架构","部门ID"]
},
{
"name": "input_dept",
"type": "VARCHAR(108)",
"comment": "录入部门ID",
"role": "dimension",
- "tags": ["操作部门", "组织信息"]
+ "tags": ["操作部门", "组织信息","录入部门ID"]
},
{
"name": "name",
@@ -153,7 +154,7 @@ person_database_ddl = """
"type": "VARCHAR(108)",
"comment": "内部单位ID",
"role": "dimension",
- "tags": ["组织信息", "内部架构ID"]
+ "tags": ["组织信息", "内部单位ID"]
},
{
"name": "person_type",
@@ -192,9 +193,9 @@ person_database_ddl = """
{
"name": "code",
"type": "VARCHAR(600)",
- "comment": "编码",
+ "comment": "用户ID彪马",
"role": "dimension",
- "tags": ["编码信息", "工号"]
+ "tags": ["用户ID", "工号"]
},
{
"name": "ytenant_id",
@@ -372,8 +373,12 @@ person_database_ddl = """
"name": "is_subcontractor",
"type": "VARCHAR(108)",
"comment": "是否分包商",
+ "value":{
+ "1":"是",
+ "0":"否",
+ },
"role": "dimension",
- "tags": ["供应商类型", "合作模式"]
+ "tags": ["供应商类型", "合作模式","枚举信息"]
},
{
"name": "general_contractor_unit",
@@ -443,14 +448,14 @@ person_database_ddl = """
"to_table": "IUAP_APDOC_BASEDOC.org_orgs",
"to_field": ["id","parentid"],
"type": "foreign_key",
- "comment": "关联部门表"
+ "comment": "关联部门表,部门ID或父部门ID"
},
{
"from": "internal_dept",
"to_table": "IUAP_APDOC_BASEDOC.org_orgs",
"to_field": ["id","parentid"],
"type": "foreign_key",
- "comment": "关联部门表"
+ "comment": "关联部门表,部门ID或父部门ID"
},
],
@@ -459,18 +464,16 @@ person_database_ddl = """
"""
-
-
person_attendance_rule_ddl='''
{
"db_name":"YJOA_APPSERVICE_DB",
"table_name": "t_yj_person_attendance_rules",
- "table_comment": "人员考勤规则表，定义考勤时间规则，考勤地点设置",
+ "table_comment": "考勤规则表，定义不同地点考勤时间规则，考勤地点设置",
"columns": [
{
"name": "id",
"type": "VARCHAR(50)",
- "comment": "主键ID",
+ "comment": "规则ID",
"role": "dimension",
"tags": ["主键", "ID标识"]
},
@@ -491,14 +494,14 @@ person_attendance_rule_ddl='''
{
"name": "before_lunch_time",
"type": "VARCHAR(50)",
- "comment": "午餐前时间",
+ "comment": "午休前时间",
"role": "dimension",
"tags": ["时间规则", "考勤分段"]
},
{
"name": "after_lunch_time",
"type": "VARCHAR(50)",
- "comment": "午餐后时间",
+ "comment": "午休后时间",
"role": "dimension",
"tags": ["时间规则", "考勤分段"]
},
@@ -511,10 +514,13 @@ person_attendance_rule_ddl='''
"2":"成都",
"3":"秭归",
"4":"拉萨",
- "5": "林芝"
+ "5":"林芝",
+ "6":"米林",
+ "7":"派镇",
+ "8":"墨脱"
},
"role": "dimension",
- "tags": [ "考勤的地区位置","非办公区域，不要混淆","枚举"]
+ "tags": [ "考勤的地区位置","非办公区域，不要混淆","枚举","考勤规则"]
},
],
"relationships": [
@@ -558,10 +564,13 @@ person_status_ddl='''
"value":{
"1001":"在岗",
"1002":"出差",
- "1003":"休假,请假",
+ "1003":"休假",
+ "1004":"缺勤",
"1005":"旷工",
- "1006":"迟到,早退",
+ "1006":"迟到早退",
"1007":"休息日",
+ "1008":"加班",
+ "1009":"迟到",
"4001":"am在岗pm缺勤",
"4002":"am缺勤pm在岗",
"6001":"am在岗pm早退",
@@ -570,7 +579,7 @@ person_status_ddl='''
"4006":"am缺勤pm早退"
},
"role": "dimension",
- "tags": ["状态信息", "人员状态","枚举","迟到早退都是:1006"]
+ "tags": ["状态信息", "人员状态","枚举","考勤状态","迟到包含(1006,1009,6002,6004)","早退包含(1006,6001,4006)","缺勤包含(1004,4001,4002,4006,6004)"]
},
{
"name": "date_value",
@@ -579,6 +588,42 @@ person_status_ddl='''
"role": "dimension",
"tags": ["时间信息", "日期记录"]
},
+ {
+ "name": "practical_attendance",
+ "type": "Int",
+ "comment": "实际出勤",
+ "value":{
+ "1":"已出勤",
+ "0":"未出勤",
+ },
+ "role": "dimension",
+ "tags": ["出勤信息", "状态记录","枚举信息"]
+ },
+ {
+ "name": "is_ought_attendance",
+ "type": "Int",
+ "comment": "是否应出勤",
+ "value":{
+ "1":"是",
+ "0":"否",
+ },
+ "role": "dimension",
+ "tags": ["是否应出勤", "枚举信息"]
+ },
+ {
+ "name": "work_area",
+ "type": "VARCHAR(50)",
+ "comment": "工作地区",
+ "role": "dimension",
+ "tags": ["工作地区", "地域信息"]
+ },
+ {
+ "name": "work_time",
+ "type": "Int",
+ "comment": "工作时间，单位-分钟",
+ "role": "dimension",
+ "tags": ["工作时间", "时长"]
+ },
{
"name": "dr",
"type": "INT",
@@ -593,12 +638,60 @@ person_status_ddl='''
"role": "dimension",
"tags": ["时间信息", "系统记录"]
},
+ {
+ "name": "vacation_type",
+ "type": "VARCHAR(50)",
+ "comment": "休假类型",
+ "value": {
+ "1": "年休假",
+ "2": "补休假",
+ "3": "探亲假",
+ "4": "离岗",
+ "5": "婚假",
+ "6": "丧假",
+ "7": "病假",
+ "8": "事假",
+ "9": "产假",
+ "10": "其他",
+ "11": "轮休假",
+ "12": "家属探亲"
+ },
+ "role": "dimension",
+ "tags": ["休假信息", "请假类型", "枚举", "考勤管理"]
+ },
{
"name": "is_in_tibet",
"type": "INT",
"comment": "是否在西藏地区",
"role": "dimension",
"tags": ["地区标识", "地理位置"]
+ },
+ {
+ "name": "business_trip_destination",
+ "type": "VARCHAR(50)",
+ "comment": "出差目的地",
+ "role": "dimension",
+ "tags": ["出差信息", "地理位置", "目的地"]
+ },
+ {
+ "name": "year",
+ "type": "VARCHAR(5)",
+ "comment": "年",
+ "value": {
+ "example": "2025"
+ },
+ "role": "dimension",
+ "tags": ["时间信息", "年度", "时间维度"]
+ },
+ {
+ "name": "month",
+ "type": "VARCHAR(10)",
+ "comment": "月",
+ "value": {
+ "example": "2025-08"
+ },
+ "role": "dimension",
+ "tags": ["时间信息", "月度", "时间维度"]
}
],
"relationships": [
@@ -687,12 +780,10 @@ person_attendance_ddl = '''
"type": "VARCHAR(50)",
"comment": "数据来源",
"value": {
- "APP": "手机应用",
- "DEVICE": "考勤设备",
- "SYSTEM": "系统导入"
+ "example":"大华门禁"
},
"role": "dimension",
- "tags": ["来源系统", "数据渠道"]
+ "tags": ["来源系统", "数据渠道","门禁系统源"]
},
{
"name": "dr",
@@ -724,13 +815,6 @@ person_attendance_ddl = '''
"role": "dimension",
"tags": ["门禁位置", "打卡设备点"]
},
- {
- "name": "by_go_type",
- "type": "VARCHAR(8)",
- "comment": "打卡类型",
- "role": "dimension",
- "tags": ["类型标识", "打卡类型"]
- }
],
"relationships": [
{
@@ -756,7 +840,7 @@ person_rules_ddl = '''
{
"db_name": "YJOA_APPSERVICE_DB",
"table_name": "t_yj_person_rules",
- "table_comment": "人员考勤规则关联表，关联人员与考勤规则",
+ "table_comment": "人员考勤规则关系表，关联人员与考勤规则",
"columns": [
{
"name": "id",
@@ -802,7 +886,7 @@ person_rules_ddl = '''
{
"from": "person_id",
"to_table": "t_yj_person_database",
- "to_field": "id",
+ "to_field": "code",
"type": "foreign_key",
"comment": "关联人员基本信息表"
},
@@ -847,13 +931,6 @@ person_ac_position = '''
"type": "foreign_key",
"comment": "关联门禁区域关系表"
},
- {
- "from": "position",
- "to_table": "t_yj_person_ac_area",
- "to_field": "area",
- "type": "foreign_key",
- "comment": "关联门禁区域关系表"
- },
"tags": ["门禁控制点","门禁位置"]
}
'''
@@ -862,7 +939,7 @@ person_ac_area = '''
{
"db_name":"YJOA_APPSERVICE_DB",
"table_name": "t_yj_person_ac_area",
- "table_comment": "门禁区域关系表",
+ "table_comment": "门禁与区域关系表",
"columns": [
{
"name": "ac_point",
@@ -901,7 +978,7 @@ person_ac_area = '''
},
],
- "tags": ["门禁详情","门禁区域位置","门禁地区信息","枚举"]
+ "tags": ["门禁详情","门禁与区域位置关联信息","门禁地区信息","枚举"]
}
'''
@@ -914,9 +991,9 @@ org_orgs_ddl = '''
{
"name": "id",
"type": "VARCHAR(36)",
- "comment": "主键ID",
+ "comment": "部门ID",
"role": "dimension",
- "tags": ["主键", "id标识"]
+ "tags": ["主键", "id标识","部门ID"]
},
{
"name": "parentid",