数据对比

fields_optimize.py

@@ -0,0 +1,67 @@
+from pymongo import MongoClient
+from bson.objectid import ObjectId
+from util.mogodb_helper import MongoDBInterface  # 调整导入以正确引用文件名
+
+# 参考库配置
+ref_config = {
+    "ip_port": "127.0.0.1:27088",
+    "user": "viewdata",
+    "password": "viewdata",
+    "db": "qfw",
+    "collection": "zktest_data_info_0307"
+}
+
+# 对比库配置
+comp_config = {
+    "ip_port": "127.0.0.1:27088",
+    "user": "viewdata",
+    "password": "viewdata",
+    "db": "qfw",
+    "collection": "bidding"
+}
+
+# 最终存储库配置
+final_config = {
+    "host": "192.168.3.206",
+    "port": 27080,
+    "dbname": "data_quality",
+    "collection_name": "fields_optimize"
+}
+
+# 连接到MongoDB（参考库和对比库使用MongoDBInterface类）
+ref_mdb = MongoDBInterface(ref_config)
+comp_mdb = MongoDBInterface(comp_config)
+final_client = MongoClient(final_config["host"], final_config["port"], socketTimeoutMS=30000)
+final_db = final_client[final_config["dbname"]]
+final_collection = final_db[final_config["collection_name"]]
+
+# 比较字段
+fields_to_compare = ['area', 'city', 'district']
+# 获取参考库集合中的所有文档
+ref_collection_name = ref_config['collection']
+comp_collection_name = comp_config['collection']
+
+for ref_doc in ref_mdb.db[ref_collection_name].find():  # 使用pymongo的find()方法
+    comp_doc = comp_mdb.find_by_id(comp_collection_name, ref_doc['_id'])  # 使用MongoDBInterface方法
+    if not comp_doc:
+        # 如果在对比库中找不到文档，可能需要记录或处理
+        continue
+
+    differences = {}
+    for field in fields_to_compare:
+        if ref_doc.get(field) != comp_doc.get(field):
+            differences[field] = {'ref': ref_doc.get(field), 'comp': comp_doc.get(field)}
+
+    if differences:
+        # 如果有差异，将它们添加到最终的集合中
+        final_collection.insert_one({
+            'id': ref_doc['_id'],  # 保留原始的_id作为id字段
+            'ref_doc': ref_doc,
+            'comp_doc': comp_doc,
+            'differences': differences
+        })
+
+# 断开数据库连接
+ref_mdb.disconnect()
+comp_mdb.disconnect()
+final_client.close()