checkpoint(data-extraction): Adapt load to update exisitng entries in order to keep yearly_results

src/aki_prj23_transparenzregister/models/company.py

@@ -76,6 +76,14 @@ class CompanyID:
     district_court: DistrictCourt
     hr_number: str
 
+    def to_dict(self) -> dict:
+        """Transform to dict.
+
+        Returns:
+            dict: Dictionary
+        """
+        return asdict(self)
+
 
 @dataclass
 class Location:

src/aki_prj23_transparenzregister/utils/data_extraction/unternehmensregister/load.py

@@ -5,23 +5,18 @@ import os
 
 from tqdm import tqdm
 
+from aki_prj23_transparenzregister.config.config_providers import JsonFileConfigProvider
 from aki_prj23_transparenzregister.models.company import Company
 from aki_prj23_transparenzregister.utils.mongo.company_mongo_service import (
     CompanyMongoService,
 )
 from aki_prj23_transparenzregister.utils.mongo.connector import (
-    MongoConnection,
     MongoConnector,
 )
 
 if __name__ == "__main__":
-    conn_string = MongoConnection(
-        hostname="localhost",
-        database="transparenzregister",
-        username="username",
-        password="",
-        port=27017,
-    )
+    provider = JsonFileConfigProvider("secrets.json")
+    conn_string = provider.get_mongo_connection_string()
     connector = MongoConnector(conn_string)
     service = CompanyMongoService(connector)
 
@@ -32,4 +27,4 @@ if __name__ == "__main__":
             data = json.loads(file_object.read())
             company: Company = Company(**data)
 
-            service.insert(company)
+            service.migrations_of_base_data(company)

src/aki_prj23_transparenzregister/utils/mongo/company_mongo_service.py

@@ -24,23 +24,37 @@ class CompanyMongoService:
         """_summary_.
 
         Returns:
-            list[Company]: _description_
+            list[Company]: List of retrieved companies
         """
         with self.lock:
             result = self.collection.find()
             return list(result)
 
-    def get_by_id(self, id: str) -> Company | None:
+    def get_by_id(self, id: dict) -> dict | None:
         """_summary_.
 
         Args:
-            id (str): _description_
+            id (CompanyID): CompanyID
 
         Returns:
-            Company | None: _description_
+            dict | None: Company if found
         """
         with self.lock:
-            result = list(self.collection.find({"id": id}))
+            result = list(
+                self.collection.find(
+                    {
+                        "id": {
+                            "$eq": {
+                                "hr_number": id["hr_number"],
+                                "district_court": {
+                                    "name": id["district_court"]["name"],
+                                    "city": id["district_court"]["city"],
+                                },
+                            }
+                        }
+                    }
+                )
+            )
             if len(result) == 1:
                 return result[0]
             return None
@@ -106,3 +120,21 @@ class CompanyMongoService:
             return self.collection.update_one(
                 {"_id": ObjectId(_id)}, {"$set": {"yearly_results": yearly_results}}
             )
+
+    def migrations_of_base_data(self, data: Company) -> InsertOneResult | UpdateResult:
+        """Updates or inserts a document of type company depending on whether an entry with the same id (CompanyID) can be found.
+
+        Args:
+            data (Company): Company related data to persist
+
+        Returns:
+            InsertOneResult | UpdateResult: Result depending on action
+        """
+        entry = self.get_by_id(data.id.to_dict())
+        if entry is None:
+            return self.insert(data)
+        statement = {"$set": dict(data.to_dict().items())}
+        with self.lock:
+            return self.collection.update_one(
+                {"_id": ObjectId(entry["_id"])}, statement
+            )

tests/utils/mongo/company_mongo_service_test.py

@@ -3,7 +3,12 @@ from unittest.mock import Mock
 
 import pytest
 
-from aki_prj23_transparenzregister.models.company import Company, CompanyID, Location
+from aki_prj23_transparenzregister.models.company import (
+    Company,
+    CompanyID,
+    DistrictCourt,
+    Location,
+)
 from aki_prj23_transparenzregister.utils.mongo.company_mongo_service import (
     CompanyMongoService,
 )
@@ -73,21 +78,23 @@ def test_by_id_no_result(mock_mongo_connector: Mock, mock_collection: Mock) -> N
     mock_mongo_connector.database = {"companies": mock_collection}
     service = CompanyMongoService(mock_mongo_connector)
     mock_collection.find.return_value = []
-    assert service.get_by_id("Does not exist") is None
+    id = CompanyID(DistrictCourt("a", "b"), "c").to_dict()
+    assert service.get_by_id(id) is None
 
 
 def test_by_id_result(mock_mongo_connector: Mock, mock_collection: Mock) -> None:
     """Test CompanyMongoService get_by_id with result.
 
     Args:
-        mock_mongo_connector (Mock): Mocked MongoConnector library
-        mock_collection (Mock): Mocked pymongo collection
+    mock_mongo_connector (Mock): Mocked MongoConnector library
+    mock_collection (Mock): Mocked pymongo collection.
     """
     mock_mongo_connector.database = {"companies": mock_collection}
     service = CompanyMongoService(mock_mongo_connector)
     mock_entry = {"id": "Does exist", "vaue": 42}
     mock_collection.find.return_value = [mock_entry]
-    assert service.get_by_id("Does exist") == mock_entry
+    id = CompanyID(DistrictCourt("a", "b"), "c").to_dict()
+    assert service.get_by_id(id) == mock_entry
 
 
 def test_insert(mock_mongo_connector: Mock, mock_collection: Mock) -> None: