Bugfix/yearly results with string as keys (#153)

All entries you found have been updated on prod.

src/aki_prj23_transparenzregister/apps/fix_company_financials.py

@@ -0,0 +1,23 @@
+"""Fix fincancial data of particular companies identified by their ID."""
+from aki_prj23_transparenzregister.apps.enrich_company_financials import work
+from aki_prj23_transparenzregister.config.config_providers import JsonFileConfigProvider
+from aki_prj23_transparenzregister.utils.mongo.company_mongo_service import (
+    CompanyMongoService,
+)
+from aki_prj23_transparenzregister.utils.mongo.connector import MongoConnector
+
+if __name__ == "__main__":
+    config_provider = JsonFileConfigProvider("./secrets.json")
+
+    mongo_connector = MongoConnector(config_provider.get_mongo_connection_string())
+    company_service = CompanyMongoService(mongo_connector)
+
+    entries = [
+        "649f16a4e198338c3b442ab1",
+        "649f16a5e198338c3b442b0a",
+        "649f16a5e198338c3b442ac6",
+    ]
+
+    companies = [company_service.get_by_object_id(entry) for entry in entries]
+    for company in companies:
+        work(company, company_service)

src/aki_prj23_transparenzregister/utils/data_extraction/bundesanzeiger.py

@@ -61,10 +61,15 @@ class Bundesanzeiger:
         Returns:
             pd.DataFrame: Filtered and pruned DataFrame
         """
+        date_regex = r"\d{2}\.\d{2}.\d{4}"
+
         df_reports["type"] = df_reports.name.apply(lambda name: name.split(" ")[0])
-        df_reports = df_reports.loc[df_reports.type == "Jahresabschluss"]
+        df_reports = df_reports.loc[
+            ((df_reports.type == "Jahresabschluss") | (df_reports.type == "Jahres-"))
+            & df_reports.name.str.contains(date_regex, regex=True)
+        ]
         df_reports["jahr"] = df_reports.name.apply(
-            lambda name: name.split(" ")[-1].split(".")[-1]
+            lambda name: re.findall(date_regex, name)[0].split(".")[-1]
         )
         return df_reports.drop(["name", "report", "type"], axis=1)
 

tests/apps/fix_company_financials_test.py

@@ -0,0 +1,6 @@
+"""test aki_prj23_transparenzregister.apps.fix_company_financials."""
+from aki_prj23_transparenzregister.apps import fix_company_financials
+
+
+def test_main() -> None:
+    assert fix_company_financials

tests/utils/data_extraction/bundesanzeiger_test.py

@@ -70,7 +70,11 @@ def test_extracct_financial_results() -> None:
 def test_filter_reports() -> None:
     test_data = [
         {"name": "Bedienungsanleitung", "report": "", "raw_report": ""},
-        {"name": "Jahresabschluss 1998", "report": "", "raw_report": ""},
+        {
+            "name": "Jahresabschluss vom 01.01.1998 bis zum 31.12.1998",
+            "report": "",
+            "raw_report": "",
+        },
     ]
     test_df = pd.DataFrame(test_data)
     ba = Bundesanzeiger()
@@ -91,7 +95,7 @@ def test_get_information(mock_bundesanzeiger: Mock) -> None:
             "raw_report": "",
         },
         "2": {
-            "name": "Jahresabschluss 1998",
+            "name": "Jahresabschluss 01.01.1998",
             "report": "",
             "company": "PRJ 23 Transparenzregister GmbH",
             "raw_report": "",