From 2050b49fde947a98264ed5c0d799f658593abcfa Mon Sep 17 00:00:00 2001
From: TrisNol <tristan.nolde@yahoo.de>
Date: Mon, 25 Sep 2023 18:37:39 +0200
Subject: [PATCH] fix(data-extraction): Resolve issue in different
 Bundesanzeiger formats

---
 .../utils/data_extraction/bundesanzeiger.py               | 6 ++++--
 tests/utils/data_extraction/bundesanzeiger_test.py        | 8 ++++++--
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/aki_prj23_transparenzregister/utils/data_extraction/bundesanzeiger.py b/src/aki_prj23_transparenzregister/utils/data_extraction/bundesanzeiger.py
index fd5f1ab..f69e78f 100644
--- a/src/aki_prj23_transparenzregister/utils/data_extraction/bundesanzeiger.py
+++ b/src/aki_prj23_transparenzregister/utils/data_extraction/bundesanzeiger.py
@@ -62,9 +62,11 @@ class Bundesanzeiger:
             pd.DataFrame: Filtered and pruned DataFrame
         """
         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["jahr"] = df_reports.name.apply(
-            lambda name: name.split(" ")[-1].split(".")[-1]
+            lambda name: re.findall(r"\d{2}\.\d{2}.\d{4}", name)[0].split(".")[-1]
         )
         return df_reports.drop(["name", "report", "type"], axis=1)
 
diff --git a/tests/utils/data_extraction/bundesanzeiger_test.py b/tests/utils/data_extraction/bundesanzeiger_test.py
index 8829bbd..94af1ca 100644
--- a/tests/utils/data_extraction/bundesanzeiger_test.py
+++ b/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": "",