superx-budget-overview/superx_budget/budget.py

""" Budget Parser """

from collections import namedtuple

from .exceptions import BudgetParserError

EXPECTED_TABLE_HEADERS = [
    "Nr.",
    "Projekt",
    "Laufzeit",
    "BA",
    "Budget",
    "Ausgaben",
    "Rest",
]


ExcelRow = namedtuple("ExcelRow", ["row", "data"])

BudgetData = namedtuple(
    "BudgetData",
    [
        "row",
        "number",
        "project_name",
        "period",
        "project",
        "budget",
        "expenses",
        "rest",
    ],
)


def _check_table_header(xl_row):
    fields = [c.strip() for c in xl_row.data[:7]]
    print(fields)
    print(EXPECTED_TABLE_HEADERS)
    if fields != EXPECTED_TABLE_HEADERS:
        raise BudgetParserError(f"unexpected headers: '{xl_row.data}'")


def _skip_empty_lines(rows):
    for xl_row in rows:
        first_cell = xl_row.data[0]
        if first_cell is None:
            continue
        if isinstance(first_cell, str) and first_cell.strip() == "":
            continue
        yield xl_row


def _parse_data_table(rows):
    for xl_row in _skip_empty_lines(rows):
        yield BudgetData(xl_row.row, *xl_row.data[:7])


def parse_budget_data(xls_sheet):
    """ parses the budget data """
    rows = (ExcelRow(i, v) for i, v in enumerate(xls_sheet.values, start=1))
    _check_table_header(next(rows))
    return list(_parse_data_table(rows))