aki_prj23_transparenzregister/tests/utils/data_extraction/bundesanzeiger_test.py

"""Tests if the bundesanzeiger can be accessed and read."""
from unittest.mock import Mock, patch

import pandas as pd

from aki_prj23_transparenzregister.utils.data_extraction.bundesanzeiger import (
    Bundesanzeiger,
)
from aki_prj23_transparenzregister.utils.enum_types import FinancialKPIEnum


def test_extract_auditor_company_no_hits() -> None:
    input_data = """
    <b>
        Mega GmbH
        Nothing to see here
    </b>
    """
    ba = Bundesanzeiger()
    result = ba.extract_auditor_company(input_data, "Mega GmbH")
    assert result is None


def test_extract_auditor_company_self_referencing() -> None:
    company = "Mega GmbH"
    auditor_company = "Super AG"

    input_data = f"""
    <b>
        {company}
        <br>
        Nothing to see here
    </b>
    <b>
        {auditor_company}
        <br>
        Nothing to see here
    </b>
    """
    ba = Bundesanzeiger()
    result = ba.extract_auditor_company(input_data, company)
    assert result == auditor_company


def test_extract_auditor_company() -> None:
    company_name = "Korrupte Wirtschaftsprüfer GmbH & Co. KG"
    input_data = f"""
        <b>
            {company_name}
            <br>
            Max Mustermann
        </b>
    """
    ba = Bundesanzeiger()
    result = ba.extract_auditor_company(input_data, "Super AG")
    assert result == company_name


def test_extract_kpis() -> None:
    input_data = """
        Die Prj23_Transparenzregister GmbH erwirtschaftete einen Jahresüberschuss 10.000,43 €.
        Des Weiteren sanken die Gesamtverbindlichkeiten 42,00 €
    """
    ba = Bundesanzeiger()
    result = ba.__extract_kpis__(input_data)

    net_income = 10000.43
    liabilities = 42.00

    assert result[FinancialKPIEnum.NET_INCOME.value] == net_income
    assert result[FinancialKPIEnum.LIABILITIES.value] == liabilities


def test_extract_financial_results() -> None:
    input_data = """
    <br>
        Die Prj23_Transparenzregister GmbH erwirtschaftete einen Jahresüberschuss 10.000,43 €.
    </br>
    <h2>Dies ist ein Platzhalter, der ignoriert werden soll</h2>
    <b>Des Weiteren sanken die Gesamtverbindlichkeiten 42,00 €</b>
    """
    ba = Bundesanzeiger()
    result = ba.extract_financial_results(input_data)

    net_income = 10000.43
    liabilities = 42.00

    assert result[FinancialKPIEnum.NET_INCOME.value] == net_income
    assert result[FinancialKPIEnum.LIABILITIES.value] == liabilities


def test_filter_reports() -> None:
    test_data = [
        {"name": "Bedienungsanleitung", "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()
    result = ba.filter_reports(test_df)
    assert len(result) == 1
    assert result.iloc[0].jahr == "1998"


@patch(
    "aki_prj23_transparenzregister.utils.data_extraction.bundesanzeiger.Ba.get_reports"
)
def test_get_information(mock_bundesanzeiger: Mock) -> None:
    mock_bundesanzeiger.return_value = {
        "1": {
            "name": "Bedienungsanleitung",
            "report": "",
            "company": "",
            "raw_report": "",
        },
        "2": {
            "name": "Jahresabschluss 01.01.1998",
            "report": "",
            "company": "PRJ 23 Transparenzregister GmbH",
            "raw_report": "",
        },
    }
    ba = Bundesanzeiger()
    result = ba.get_information("PRJ 23 Transparenzregister GmbH", "Iserlohn")
    assert len(result) == 1


@patch(
    "aki_prj23_transparenzregister.utils.data_extraction.bundesanzeiger.Ba.get_reports"
)
def test_get_information_no_results(mock_bundesanzeiger: Mock) -> None:
    mock_bundesanzeiger.return_value = {}
    ba = Bundesanzeiger()
    result = ba.get_information("PRJ 23 Transparenzregister GmbH", "Iserlohn")
    assert len(result) == 0


def test_extract_tables_from_reports() -> None:
    report = """
        <table>
        </table>
        <div>
            Möge die Macht mir dir sein
            <table class="std_table">
                <tr>
                    <th>Column A</th>
                    <th>Column B</th>
                </tr>
                <tr>
                    <td>42</td>
                    <td>4711</td>
                </tr>
            </table>
        </div>
    """
    ba = Bundesanzeiger()
    result = ba.__extract_tables_from_report__(report)
    assert len(result) == 1


def test_parse_tables_to_kpis() -> None:
    report = """
        <table class="std_table">
            <tr>
                <th>Position</th>
            </tr>
            <tr>
                <td>a) Umlaufvermögen</td>
            </tr>
        </table>
        <table class="std_table">
            <tr>
                <th>Position</th>
                <th>Test</th>
            </tr>
            <tr>
                <td>4711</td>
                <td>4711</td>
            </tr>
        </table>
        <div>
            Möge die Macht mir dir sein
            <table class="std_table">
                <tr>
                    <th>Position</th>
                    <th>2023 in T€</th>
                    <th>1997 in €</th>
                </tr>
                <tr>
                    <td>a) Umlaufvermögen</td>
                    <td>12,13</td>
                    <td>4711</td>
                </tr>
                <tr>
                    <td>+EBIT</td>
                    <td>1123</td>
                    <td>4711</td>
                </tr>
                <tr>
                    <td>To be ignored</td>
                    <td>I've tried so hard and got so far, but in the end it doesn't even matter</td>
                    <td>4711</td>
                </tr>
                <tr>
                    <td>Gewinn</td>
                    <td></td>
                    <td>4711</td>
                </tr>
                <tr>
                    <td>Jahresüberschuss</td>
                    <td>4.130,12</td>
                    <td>4711</td>
                </tr>
            </table>
            <table class="std_table">
            <thead>
                <tr>
                    <th>Position</th>
                    <th>Betrag in</th>
                </tr>
                 <tr>
                    <th>Hallo</th>
                    <th>€</th>
                </tr>
            </thead>
            <tbody>
                <tr>
                    <td>I. Schulden</td>
                    <td>0,12</td>
                </tr>
            </tbody>
        </table>
        </div>
    """
    ba = Bundesanzeiger()
    result = ba.parse_tables_to_kpis(report)
    assert result == {
        "Umlaufvermögen": 12130.0,
        "EBIT": 1123000.0,
        "Jahresüberschuss": 4130120.0,
        "Schulden": 0.12,
    }