Resolve mismatch between staging and prod db data for financials (#211)

SQL Creation is now done dynamicly by the definition of the enumeration
type.
This commit is contained in:
Philipp Horstenkamp 2023-10-14 17:16:14 +02:00 committed by GitHub
parent 9f7d714403
commit f8c111d7e2
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
8 changed files with 76 additions and 112 deletions

View File

@ -1,9 +1,11 @@
"""Company model."""
from dataclasses import asdict, dataclass
from enum import Enum, StrEnum
from enum import StrEnum
from aenum import MultiValueEnum
from aki_prj23_transparenzregister.utils.enum_types import FinancialKPIEnum
class RelationshipRoleEnum(str, MultiValueEnum):
"""Roles taken by entities in relationships to a Company."""
@ -134,43 +136,6 @@ class CompanyToCompanyRelationship(CompanyRelationship):
name: str
class FinancialKPIEnum(Enum):
"""Financial KPI keys."""
# Umsatz || Erlöse
REVENUE = "revenue"
# Jahresüberschuss || Nettoeinkommen
NET_INCOME = "net_income"
# Ebit
EBIT = "ebit"
# Ebitda
EBITDA = "ebitda"
# Bruttogewinn
GROSS_PROFIT = "gross_profit"
# Betriebsgewinn
OPERATING_PROFIT = "operating_profit"
# Bilanzsumme
ASSETS = "assets"
# Gesamtverbindlichkeiten
LIABILITIES = "liabilities"
# Eigenkapital
EQUITY = "equity"
# Umlaufvermögen
CURRENT_ASSETS = "current_assets"
# Kurzfristige Verbindlichkeiten
CURRENT_LIABILITIES = "current_liabilities"
# Langfristige Verbindlichkeiten
LONG_TERM_DEBT = "long_term_debt"
# Kurzfristige Verbindlichkeiten
SHORT_TERM_DEBT = "short_term_debt"
# Barmittel
CASH_AND_CASH_EQUIVALENTS = "cash_and_cash_equivalents"
# Dividende
DIVIDENDS = "dividends"
# Cash Flow
CASH_FLOW = "cash_flow"
@dataclass
class YearlyResult:
"""Company yearly result."""

View File

@ -6,7 +6,7 @@ from bs4 import BeautifulSoup
from deutschland.bundesanzeiger import Bundesanzeiger as Ba
from aki_prj23_transparenzregister.models.auditor import Auditor
from aki_prj23_transparenzregister.models.company import FinancialKPIEnum
from aki_prj23_transparenzregister.utils.enum_types import FinancialKPIEnum
pd.options.mode.chained_assignment = None # type: ignore

View File

@ -1,5 +1,6 @@
"""Collection of enumeration types for the whole project."""
import enum
from enum import Enum
class SentimentTypeEnum(enum.Enum):
@ -9,3 +10,40 @@ class SentimentTypeEnum(enum.Enum):
sustainability = "sustainability"
environmental_aspects = "environmental_aspects"
perception = "perception"
class FinancialKPIEnum(Enum):
"""Financial KPI keys."""
# Umsatz || Erlöse
REVENUE = "revenue"
# Jahresüberschuss || Nettoeinkommen
NET_INCOME = "net_income"
# Ebit
EBIT = "ebit"
# Ebitda
EBITDA = "ebitda"
# Bruttogewinn
GROSS_PROFIT = "gross_profit"
# Betriebsgewinn
OPERATING_PROFIT = "operating_profit"
# Bilanzsumme
ASSETS = "assets"
# Gesamtverbindlichkeiten
LIABILITIES = "liabilities"
# Eigenkapital
EQUITY = "equity"
# Umlaufvermögen
CURRENT_ASSETS = "current_assets"
# Kurzfristige Verbindlichkeiten
CURRENT_LIABILITIES = "current_liabilities"
# Langfristige Verbindlichkeiten
LONG_TERM_DEBT = "long_term_debt"
# Kurzfristige Verbindlichkeiten
SHORT_TERM_DEBT = "short_term_debt"
# Barmittel
CASH_AND_CASH_EQUIVALENTS = "cash_and_cash_equivalents"
# Dividende
DIVIDENDS = "dividends"
# Cash Flow
CASH_FLOW = "cash_flow"

View File

@ -10,6 +10,7 @@ from aki_prj23_transparenzregister.models.company import (
RelationshipRoleEnum,
)
from aki_prj23_transparenzregister.utils.enum_types import (
FinancialKPIEnum,
SentimentTypeEnum,
)
from aki_prj23_transparenzregister.utils.sql.connector import Base
@ -105,33 +106,17 @@ class Person(Base):
works_for = sa.Column(sa.String(100), nullable=True)
class AnnualFinanceStatement(Base):
"""Finance."""
__tablename__ = "annual_finance_statement"
id = sa.Column(sa.Integer, primary_key=True)
company_id = sa.Column(sa.Integer, sa.ForeignKey("company.id"))
date = sa.Column(sa.Date, nullable=False)
total_volume = sa.Column(sa.Float, default="NaN")
ebit = sa.Column(sa.Float, default="NaN")
ebitda = sa.Column(sa.Float, default="NaN")
ebit_margin = sa.Column(sa.Float, default="NaN")
total_balance = sa.Column(sa.Float, default="NaN")
equity = sa.Column(sa.Float, default="NaN")
debt = sa.Column(sa.Float, default="NaN")
return_on_equity = sa.Column(sa.Float, default="NaN")
capital_turnover_rate = sa.Column(sa.Float, default="NaN")
current_liabilities = sa.Column(sa.Float, default="NaN")
dividends = sa.Column(sa.Float, default="NaN")
net_income = sa.Column(sa.Float, default="NaN")
assets = sa.Column(sa.Float, default="NaN")
long_term_debt = sa.Column(sa.Float, default="NaN")
short_term_debt = sa.Column(sa.Float, default="NaN")
revenue = sa.Column(sa.Float, default="NaN")
cash_flow = sa.Column(sa.Float, default="NaN")
current_assets = sa.Column(sa.Float, default="NaN") # assets vs current assets
# company: Mapped[Company] = relationship(Company)
AnnualFinanceStatement = type(
"AnnualFinanceStatement",
(Base,),
{
"__tablename__": "annual_finance_statement",
"id": sa.Column(sa.Integer, primary_key=True),
"company_id": sa.Column(sa.Integer, sa.ForeignKey("company.id")),
"date": sa.Column(sa.Date, nullable=False),
}
| {_.value: sa.Column(sa.Float, default="NaN") for _ in FinancialKPIEnum},
)
class Sentiment(Base):

View File

@ -56,47 +56,40 @@ def finance_statements() -> list[dict[str, Any]]:
"id": 1,
"company_id": 1,
"date": datetime.date.fromisoformat("2023-01-01"),
"total_volume": 1000.0,
"ebit": 1000.0,
"ebitda": 1000.0,
"ebit_margin": 1000.0,
"total_balance": 1000.0,
"gross_profit": 1000.0,
"equity": 1000.0,
"debt": 1000.0,
"return_on_equity": 1000.0,
"capital_turnover_rate": 1000.0,
"current_liabilities": 1000.0,
"dividends": float("NaN"),
"net_income": float("NaN"),
"assets": 1000.0,
"net_income": 100.0,
"long_term_debt": 1000.0,
"short_term_debt": 1000.0,
"revenue": 1000.0,
"cash_flow": 1000.0,
"current_assets": 1000.0,
"liabilities": 0.0,
"cash_and_cash_equivalents": 1.0,
"dividends": 0.0,
},
{
"id": 2,
"company_id": 1,
"date": datetime.date.fromisoformat("2022-01-01"),
"total_volume": 1100.0,
"revenue": 1100.0,
"net_income": float("NaN"),
"ebit": 1100.0,
"ebitda": 1100.0,
"ebit_margin": 1100.0,
"total_balance": 1100.0,
"gross_profit": 1100.0,
"equity": 1100.0,
"debt": 1100.0,
"return_on_equity": 1100.0,
"capital_turnover_rate": 1100.0,
"current_liabilities": 1100.0,
"dividends": float("NaN"),
"net_income": float("NaN"),
"assets": 1100.0,
"long_term_debt": 1100.0,
"short_term_debt": 1100.0,
"revenue": 1100.0,
"cash_flow": 1100.0,
"current_assets": 1100.0,
"operating_profit": 1.0,
},
]

View File

@ -65,24 +65,22 @@ def test_get_finance_data(full_db: Session) -> None:
"annual_finance_statement_id": {0: 1, 1: 2},
"annual_finance_statement_company_id": {0: 1, 1: 1},
"annual_finance_statement_date": {0: "2023-01-01", 1: "2022-01-01"},
"annual_finance_statement_total_volume": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_revenue": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_net_income": {0: 100.0},
"annual_finance_statement_ebit": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_ebitda": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_ebit_margin": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_total_balance": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_gross_profit": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_operating_profit": {1: 1.0},
"annual_finance_statement_assets": {0: 1000.0, 1: 1100},
"annual_finance_statement_liabilities": {0: 0.0},
"annual_finance_statement_equity": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_debt": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_return_on_equity": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_capital_turnover_rate": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_current_assets": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_current_liabilities": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_dividends": {0: None, 1: None},
"annual_finance_statement_net_income": {0: None, 1: None},
"annual_finance_statement_assets": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_long_term_debt": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_short_term_debt": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_revenue": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_cash_and_cash_equivalents": {0: 1.0},
"annual_finance_statement_dividends": {0: 0.0},
"annual_finance_statement_cash_flow": {0: 1000.0, 1: 1100.0},
"annual_finance_statement_current_assets": {0: 1000.0, 1: 1100.0},
"company_name": {0: "Some Company GmbH", 1: "Some Company GmbH"},
"company_id": {0: 1, 1: 1},
}

View File

@ -3,10 +3,10 @@ from unittest.mock import Mock, patch
import pandas as pd
from aki_prj23_transparenzregister.models.company import FinancialKPIEnum
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:

View File

@ -615,7 +615,6 @@ def test_add_relationships_none(empty_relations: list, full_db: Session) -> None
# noinspection SpellCheckingInspection
@pytest.mark.working_on()
@pytest.mark.parametrize(
"documents",
[
@ -980,7 +979,7 @@ def test_add_annual_report(
year,
{
"financials": {
"ebit": 123,
"revenue": 123,
"ebitda": 235,
"short_term_debt": short_term_debt,
},
@ -999,24 +998,9 @@ def test_add_annual_report(
"id": 3,
"company_id": company_id,
"date": pd.to_datetime(date(year, 1, 1)),
"total_volume": float("NaN"),
"ebit": 123.0,
"revenue": 123.0,
"ebitda": 235.0,
"ebit_margin": float("NaN"),
"total_balance": float("NaN"),
"equity": float("NaN"),
"debt": float("NaN"),
"return_on_equity": float("NaN"),
"capital_turnover_rate": float("NaN"),
"current_liabilities": float("NaN"),
"dividends": float("NaN"),
"net_income": float("NaN"),
"assets": float("NaN"),
"long_term_debt": float("NaN"),
"short_term_debt": short_term_debt,
"revenue": float("NaN"),
"cash_flow": float("NaN"),
"current_assets": float("NaN"),
}
]
)
@ -1025,6 +1009,7 @@ def test_add_annual_report(
pd.testing.assert_frame_equal(
expected_results,
df_prior,
check_like=True,
)