liumiaomiao 4 dagar sedan
förälder
incheckning
2cd54058f3

+ 5 - 0
app.py

@@ -26,6 +26,7 @@ from tables.fields.bidstarttime import BidstarttimeChecker
 from tables.fields.bidendtime import BidendtimeChecker
 from tables.fields.bidopentime import BidopentimeChecker
 from tables.fields.capital import CapitalChecker
+from tables.fields.publish_org import Publish_orgChecker
 toptype_checker = ToptypeChecker()
 subtype_checker = SubtypeChecker()
 area_checker = AreaChecker()
@@ -49,6 +50,7 @@ purchasing_checker = PurchasingChecker()
 city_checker = CityChecker()
 district_checker = DistrictChecker()
 capital_checker = CapitalChecker()
+publish_org_checker = Publish_orgChecker()
 # 定义检查链
 check_chain = {
     "toptype": {
@@ -113,6 +115,9 @@ check_chain = {
     },
     "capital": {
         "checker": capital_checker,
+    },
+    "publish_org" :{
+        "checker": publish_org_checker,
     }
 }
 

+ 44 - 5
client_mongo_mysql_liantong.py

@@ -102,6 +102,39 @@ def parse_timestamp(timestamp):
         return datetime.fromtimestamp(int(timestamp))
     except (ValueError, TypeError):
         return None
+def insert_bid_statistics (col,conn,query,batch_id):
+    # 使用聚合管道进行多条件统计
+    pipeline = [
+        {"$match": query},
+        {"$facet": {
+            "总量": [{"$count": "count"}],  # 添加总量统计
+            "招标": [{"$match": {"toptype": "招标"}}, {"$count": "count"}],
+            "预告": [{"$match": {"toptype": "预告"}}, {"$count": "count"}],
+            "结果": [{"$match": {"toptype": "结果"}}, {"$count": "count"}],
+            "其它": [{"$match": {"toptype": "其它"}}, {"$count": "count"}],
+            "other": [{"$match": {"toptype": {"$nin": ["招标", "预告", "结果", "其它"]}}}, {"$count": "count"}]
+        }}
+    ]
+
+    result = list(col.aggregate(pipeline))[0]
+
+    # 提取统计结果
+    count_total = result["总量"][0]["count"] if result["总量"] else 0
+    count_zhaobiao = result["招标"][0]["count"] if result["招标"] else 0
+    count_yugao = result["预告"][0]["count"] if result["预告"] else 0
+    count_jieguo = result["结果"][0]["count"] if result["结果"] else 0
+    count_qita = result["其它"][0]["count"] if result["其它"] else 0
+    count_other = result["other"][0]["count"] if result["other"] else 0
+    count_jieguo = count_jieguo + count_qita
+
+    print(f"招标数量: {count_zhaobiao}")
+    print(f"预告数量: {count_yugao}")
+    print(f"结果数量: {count_jieguo}")
+    print(f"其它类型数量: {count_other}")
+    sql_query = """INSERT IGNORE INTO bid_statistics_liantong (zhaobiao_count,yugao_count,jieguo_count,other_count, total_count, batch_id) 
+                   VALUES ( %s, %s, %s, %s, %s,%s)"""
+    params = (count_zhaobiao,count_yugao,count_jieguo,count_other,count_total,batch_id)
+    MysqlUtil.insert_data(conn, sql_query, params)
 
 def batch_load_data():
     """
@@ -117,7 +150,7 @@ def batch_load_data():
     end_date = int(datetime.strptime(f"{today_date} 00:00:00", "%Y-%m-%d %H:%M:%S").timestamp())
     # print("end_date", end_date)
     # 规则查询,根据必要条件 公司名称(用户ID)、版本号
-    rules_id = get_rule("中国联通", "v1.3")
+    rules_id = get_rule("中国联通简版-标讯", "v1.4")
     print(rules_id)
     # 初始化mysql
     conn = MysqlUtil.connect_to_mysql(host='172.20.45.129', port='4000', user='root', password='=PDT49#80Z!RVv52_z',database='quality')
@@ -131,11 +164,13 @@ def batch_load_data():
         "appid": "jyGQ1XQQsEAwNeSENOFR9D",
         "createtime": {"$gte": start_date, "$lte": end_date}
     }
-
+    # mongo_client = MongoClient('mongodb://127.0.0.1:27087/', unicode_decode_error_handler="ignore",directConnection=True)  # 修改为你的连接地址
+    mongo_client = MongoClient('mongodb://172.20.17.61:27080/', unicode_decode_error_handler="ignore",directConnection=True)   # 正式环境
+    client = mongo_client.jyqyfw
+    coll_user = client["usermail"]
+    #存入数据总量,计算缺失率
+    insert_bid_statistics(coll_user,conn,query,today_date)
     while True:
-        client = MongoClient('mongodb://127.0.0.1:27087/', unicode_decode_error_handler="ignore",directConnection=True).jyqyfw  # 修改为你的连接地址
-        # client = MongoClient('mongodb://172.20.17.61:27080/', unicode_decode_error_handler="ignore",directConnection=True).jyqyfw  # 正式环境
-        coll_user = client["usermail"]
         try:
             for item in coll_user.find(query).sort("_id", 1):
                 print("------数据处理开始--------")
@@ -216,6 +251,10 @@ def batch_load_data():
             import traceback
             traceback.print_exc()  # 打印完整堆栈信息
             time.sleep(10)
+        finally:
+            # 确保在循环结束后关闭连接
+            conn.close()  # 关闭MySQL连接
+            mongo_client.close()  # 关闭MongoDB连接
 
 
 if __name__ == '__main__':

+ 161 - 0
client_mongo_mysql_news_liantong.py

@@ -0,0 +1,161 @@
+# coding:utf-8
+import time
+from a2s.tools import json_serialize, json_deserialize
+from a2s.a2s_client import a2s_execute
+from docs.config import ReluMongodb
+from util.mogodb_helper import MongoDBInterface
+from pymongo import MongoClient
+from util.mysql_tool import MysqlUtil
+import json
+from datetime import datetime, timedelta
+from bson import ObjectId
+
+ReluClient = MongoDBInterface(ReluMongodb)
+
+# 评估服务配置
+a2s_ip = "172.20.100.235:9090"
+topic = "quality_bid"
+#本地测试用的主题
+# topic = "test_quality_bid"
+timeout = 300
+
+
+# 开始评估
+def start_quality(data: dict, rules_id: int, a2s_ip, topic, timeout, retry=3):
+    # 本次不使用SSL,所以channel是不安全的
+    row = {"data": data, "rules_id": rules_id}
+    bytes_data = json_serialize(row)
+    for t in range(retry):
+        print("topic",topic)
+        try:
+            resp_data = a2s_execute(a2s_ip, topic, timeout, bytes_data)
+            if resp_data is None:
+                continue
+            result = json_deserialize(resp_data)
+            return result
+        except Exception as e:
+            print(e)
+    return {}
+
+# 获取规则ID
+def get_rule(company, version):
+    rule_id = ReluClient.find_rule_by_company(ReluMongodb["col"], company, version)
+    return rule_id
+
+def insert_batch_data(conn, params):
+    """
+    执行批量插入数据
+    """
+    query = """INSERT IGNORE INTO news_analysis_liantong (mongoid, area, city, title, publish_org, score, error_type, create_time) 
+               VALUES (%s, %s, %s, %s, %s, %s, %s, %s )"""
+    MysqlUtil.insert_data(conn, query, params)
+
+
+def has_non_empty_qa(data):
+    # 获取data字典
+    data_dict = data.get('data', {})
+
+    # 遍历所有键值对
+    for key, value in data_dict.items():
+        # 检查键以'_qa'结尾且值不为空
+        if key.endswith('_qa') and value:  # value不为None、空字典、空列表等
+            return True
+    return False
+
+
+def insert_bid_statistics (col,conn,query,batch_id):
+    # 使用聚合管道进行多条件统计
+    pipeline = [
+        {"$match": query},
+        {"$facet": {
+            "总量": [{"$count": "count"}],  # 添加总量统计
+        }}
+    ]
+
+    result = list(col.aggregate(pipeline))[0]
+
+    # 提取统计结果
+    count_total = result["总量"][0]["count"] if result["总量"] else 0
+
+    sql_query = """INSERT IGNORE INTO bid_statistics_liantong (news_count, batch_id) 
+                   VALUES ( %s, %s)"""
+    params = (count_total,batch_id)
+    MysqlUtil.insert_data(conn, sql_query, params)
+
+def batch_load_data():
+    """
+    批量数据质量检查
+    """
+    # 获取今天的日期(字符串格式)
+    today_date = datetime.now().strftime("%Y-%m-%d")
+    yesterday_date = (datetime.now() - timedelta(days=1)).strftime("%Y-%m-%d")
+    # 获取昨天 00:00:00 的时间戳
+    start_date = int(datetime.strptime(f"{yesterday_date} 00:00:00", "%Y-%m-%d %H:%M:%S").timestamp())
+    # print("start_date",start_date)
+    # 获取今天 00:00:00 的时间戳
+    end_date = int(datetime.strptime(f"{today_date} 00:00:00", "%Y-%m-%d %H:%M:%S").timestamp())
+    # print("end_date", end_date)
+    # 规则查询,根据必要条件 公司名称(用户ID)、版本号
+    rules_id = get_rule("中国联通-新闻", "v1.4.1")
+    print(rules_id)
+    # 初始化mysql
+    conn = MysqlUtil.connect_to_mysql(host='172.20.45.129', port='4000', user='root', password='=PDT49#80Z!RVv52_z',database='quality')
+
+    max_id = ObjectId("0" * 24)
+    # max_id = ObjectId("688363ebf0c6ad8b095e2245")
+    # 查询条件:_id >= max_id, appid匹配,且 createtime 在 [start_date, end_date] 之间
+    query = {
+        "_id": {"$gte": max_id},
+        "createtime": {"$gte": start_date, "$lte": end_date}
+    }
+
+    # mongo_client = MongoClient('mongodb://127.0.0.1:27087/', unicode_decode_error_handler="ignore",directConnection=True)  # 修改为你的连接地址
+    mongo_client = MongoClient('mongodb://172.20.17.61:27080/', unicode_decode_error_handler="ignore",directConnection=True)  # 正式环境
+    client = mongo_client.jyqyfw
+    coll_user = client["usermail_lt_news"]
+    #存入数据总量,计算缺失率
+    insert_bid_statistics(coll_user,conn,query,today_date)
+    while True:
+        try:
+            for item in coll_user.find(query).sort("_id", 1):
+                print("------数据处理开始--------")
+                max_id = item["_id"]
+                item["_id"] = str(item["_id"])
+                print(f"正在处理数据: {max_id}")
+                # 质量检查逻辑
+                result = start_quality(item, rules_id, a2s_ip, topic, timeout)
+                print(result)
+                code = result.get("code")
+                if code != 200:
+                    # 数据出错,跳过
+                    continue
+                #只将有错误的数据存库
+                if has_non_empty_qa(result):
+                    data = result.get("data", {})
+
+                    # 数据插入到 MySQL
+                    area = item.get("area", "")
+                    city = item.get("city", "")
+                    title = item.get("title", "")
+                    publish_org = item.get("publish_org", "")
+                    score = data.get("score", "")
+                    error_type_data = json.dumps(data)
+                    create_time = today_date
+
+                    params = (item["_id"],  area, city, title, publish_org, score, error_type_data,create_time)
+                    insert_batch_data(conn, params)
+
+            print("---- 数据处理完成 ----")
+            break
+        except Exception as e:
+            print(f"错误: {e}")
+            import traceback
+            traceback.print_exc()  # 打印完整堆栈信息
+            time.sleep(10)
+        finally:
+            # 确保在循环结束后关闭连接
+            conn.close()  # 关闭MySQL连接
+            mongo_client.close()  # 关闭MongoDB连接
+if __name__ == '__main__':
+    batch_load_data()
+

+ 29 - 6
client_mongo_mysql_nzj_liantong.py

@@ -101,6 +101,24 @@ def parse_timestamp(timestamp):
         return datetime.fromtimestamp(int(timestamp))
     except (ValueError, TypeError):
         return None
+def insert_bid_statistics (col,conn,query,batch_id):
+    # 使用聚合管道进行多条件统计
+    pipeline = [
+        {"$match": query},
+        {"$facet": {
+            "总量": [{"$count": "count"}],  # 添加总量统计
+        }}
+    ]
+
+    result = list(col.aggregate(pipeline))[0]
+
+    # 提取统计结果
+    count_total = result["总量"][0]["count"] if result["总量"] else 0
+
+    sql_query = """INSERT IGNORE INTO bid_statistics_liantong (nzj_count, batch_id) 
+                   VALUES ( %s, %s)"""
+    params = (count_total,batch_id)
+    MysqlUtil.insert_data(conn, sql_query, params)
 
 def batch_load_data():
     """
@@ -127,14 +145,16 @@ def batch_load_data():
     query = {
         "_id": {"$gte": max_id},
         # "_id": max_id,
-        # "appid": "jyGQ1XQQsEAwNeSENOFR9D",
-        # "createtime": {"$gte": start_date, "$lte": end_date}
+        "createtime": {"$gte": start_date, "$lte": end_date}
     }
 
+    # mongo_client = MongoClient('mongodb://127.0.0.1:27087/', unicode_decode_error_handler="ignore",directConnection=True)   # 修改为你的连接地址
+    mongo_client = MongoClient('mongodb://172.20.17.61:27080/', unicode_decode_error_handler="ignore",directConnection=True)  # 正式环境
+    client = mongo_client.jyqyfw
+    coll_user = client["usermail_lt_nzj"]
+    #存入数据总量,计算缺失率
+    insert_bid_statistics(coll_user,conn,query,today_date)
     while True:
-        # client = MongoClient('mongodb://127.0.0.1:27087/', unicode_decode_error_handler="ignore",directConnection=True).jyqyfw  # 修改为你的连接地址
-        client = MongoClient('mongodb://172.20.17.61:27080/', unicode_decode_error_handler="ignore",directConnection=True).jyqyfw  # 正式环境
-        coll_user = client["usermail_lt_nzj"]
         try:
             for item in coll_user.find(query).sort("_id", 1):
                 print("------数据处理开始--------")
@@ -188,7 +208,10 @@ def batch_load_data():
             import traceback
             traceback.print_exc()  # 打印完整堆栈信息
             time.sleep(10)
-
+        finally:
+            # 确保在循环结束后关闭连接
+            conn.close()  # 关闭MySQL连接
+            mongo_client.close()  # 关闭MongoDB连接
 
 if __name__ == '__main__':
     batch_load_data()

+ 6 - 2
tables/fields/NoField.py

@@ -486,9 +486,13 @@ class NoFieldChecker(object):
         :return:返回true 代表异常
         """
         projectinfo = obj.get("projectinfo", "")
-        if not projectinfo:
-            return True  # 如果没有note字段,视为无
+        infotype = obj.get("infotype", "")
+        if infotype == '政策':
+            if not projectinfo:
+                return True  # 如果没有projectinfo字段,视为无
+            return False
         return False
+
     def check_owner(self,obj, catch_content: CatchContentObject) -> bool:
         """
         联通拟在建类型