Files
2026-07-17 19:43:19 +02:00

217 lines
8.8 KiB
Python

import argparse
import sqlite3
from datetime import date, datetime, timedelta
from decimal import Decimal, InvalidOperation, ROUND_HALF_UP
from pathlib import Path
import openpyxl
GREEN = "FFC2F1C8"
YELLOW = "FFFFFF00"
ORANGE_THEME = 5
MONTH_HEADER_ROWS = (2, 11, 20)
MONTH_START_COLUMNS = (2, 10, 18, 26)
PROJECT_ROOT = Path(__file__).resolve().parents[1]
DEFAULT_DATABASE = PROJECT_ROOT / "instance" / "calendar.sqlite"
REQUIRED_TABLES = {"date_choices", "date_statuses"}
def cell_color(cell):
color = cell.fill.fgColor
if cell.fill.fill_type != "solid":
return None
if color.type == "rgb":
return color.rgb
if color.type == "theme" and color.theme == ORANGE_THEME:
return "orange"
return None
def calendar_date(year, month, week_index, weekday_index):
first = date(year, month, 1)
first_sunday = first - timedelta(days=(first.weekday() + 1) % 7)
result = first_sunday + timedelta(days=week_index * 7 + weekday_index)
return result if result.month == month else None
def amount_cents(value, coordinate):
try:
amount = Decimal(str(value).strip())
if not amount.is_finite() or amount < 0:
raise InvalidOperation
return int((amount * 100).quantize(Decimal("1"), rounding=ROUND_HALF_UP))
except (InvalidOperation, ValueError):
raise ValueError(f"Invalid amount in {coordinate}: {value!r}") from None
def read_workbook(path):
workbook = openpyxl.load_workbook(path, data_only=True, read_only=False)
amounts = {}
statuses = {}
for sheet in workbook.worksheets:
try:
year = int(sheet.title)
except ValueError:
continue
for quarter, header_row in enumerate(MONTH_HEADER_ROWS):
for position, start_column in enumerate(MONTH_START_COLUMNS):
month = quarter * 4 + position + 1
for week_index in range(6):
row = header_row + 2 + week_index
for weekday_index in range(7):
cell = sheet.cell(row=row, column=start_column + weekday_index)
selected_date = calendar_date(year, month, week_index, weekday_index)
if selected_date is None:
continue
color = cell_color(cell)
key = selected_date.isoformat()
if color == GREEN:
# A green blank means worked, but supplies no amount to import.
if cell.value is not None and str(cell.value).strip():
amounts[key] = amount_cents(cell.value, f"{sheet.title}!{cell.coordinate}")
elif color == YELLOW:
statuses[key] = "sick"
elif color == "orange":
statuses[key] = "vacation"
return amounts, statuses
def existing_conflicts(database_path, amounts, statuses):
with sqlite3.connect(database_path) as database:
existing_amounts = dict(database.execute("SELECT choice_date, amount_cents FROM date_choices"))
existing_statuses = dict(database.execute("SELECT status_date, day_status FROM date_statuses"))
amount_conflicts = sum(key in existing_amounts and existing_amounts[key] != value for key, value in amounts.items())
status_conflicts = sum(key in existing_statuses and existing_statuses[key] != value for key, value in statuses.items())
return amount_conflicts, status_conflicts
def inspect_database(database_path):
if not database_path.is_file():
raise RuntimeError(
f"Database does not exist: {database_path}\n"
"Start the app once to initialize it, or pass the correct path with --database."
)
with sqlite3.connect(database_path) as database:
tables = {
row[0]
for row in database.execute("SELECT name FROM sqlite_master WHERE type = 'table'")
}
missing = REQUIRED_TABLES - tables
if missing:
raise RuntimeError(
f"Database has the wrong schema; missing tables: {', '.join(sorted(missing))}"
)
amount_count = database.execute("SELECT COUNT(*) FROM date_choices").fetchone()[0]
status_count = database.execute("SELECT COUNT(*) FROM date_statuses").fetchone()[0]
dates = database.execute(
"""
SELECT MIN(recorded_date), MAX(recorded_date) FROM (
SELECT choice_date AS recorded_date FROM date_choices
UNION ALL
SELECT status_date AS recorded_date FROM date_statuses
)
"""
).fetchone()
return amount_count, status_count, dates[0], dates[1]
def verify_import(database_path, amounts, statuses):
with sqlite3.connect(database_path) as database:
stored_amounts = dict(
database.execute("SELECT choice_date, amount_cents FROM date_choices")
)
stored_statuses = dict(
database.execute("SELECT status_date, day_status FROM date_statuses")
)
bad_amounts = [key for key, value in amounts.items() if stored_amounts.get(key) != value]
bad_statuses = [key for key, value in statuses.items() if stored_statuses.get(key) != value]
if bad_amounts or bad_statuses:
raise RuntimeError(
"Post-import verification failed: "
f"{len(bad_amounts)} amounts and {len(bad_statuses)} statuses do not match."
)
def import_data(database_path, amounts, statuses):
timestamp = datetime.now().strftime("%Y%m%d-%H%M%S-%f")
backup_path = database_path.with_name(f"{database_path.name}.backup-{timestamp}")
with sqlite3.connect(database_path) as source, sqlite3.connect(backup_path) as backup:
source.backup(backup)
with sqlite3.connect(database_path) as database:
database.executemany(
"""
INSERT INTO date_choices (choice_date, amount_cents) VALUES (?, ?)
ON CONFLICT(choice_date) DO UPDATE SET
amount_cents = excluded.amount_cents,
updated_at = CURRENT_TIMESTAMP
""",
amounts.items(),
)
database.executemany(
"""
INSERT INTO date_statuses (status_date, day_status) VALUES (?, ?)
ON CONFLICT(status_date) DO UPDATE SET
day_status = excluded.day_status,
updated_at = CURRENT_TIMESTAMP
""",
statuses.items(),
)
verify_import(database_path, amounts, statuses)
return backup_path
def main():
parser = argparse.ArgumentParser(description="Import Fatima calendar data from Excel.")
parser.add_argument("workbook", type=Path)
parser.add_argument(
"--database",
type=Path,
default=DEFAULT_DATABASE,
help=f"SQLite database used by the app (default: {DEFAULT_DATABASE})",
)
parser.add_argument("--apply", action="store_true", help="Write the data; otherwise perform a dry run.")
args = parser.parse_args()
database_path = args.database.expanduser().resolve()
workbook_path = args.workbook.expanduser().resolve()
if not workbook_path.is_file():
parser.error(f"Workbook does not exist: {workbook_path}")
try:
before_amounts, before_statuses, first_date, last_date = inspect_database(database_path)
except RuntimeError as error:
parser.error(str(error))
print(f"Workbook: {workbook_path}")
print(f"Database: {database_path}")
print(
f"Database before import: {before_amounts} amounts, {before_statuses} statuses"
+ (f", date range {first_date} through {last_date}." if first_date else ", no recorded dates.")
)
amounts, statuses = read_workbook(workbook_path)
if not amounts and not statuses:
parser.error("Workbook produced no importable amounts or statuses; refusing to continue.")
amount_conflicts, status_conflicts = existing_conflicts(database_path, amounts, statuses)
print(f"Found {len(amounts)} amounts, {sum(v == 'vacation' for v in statuses.values())} vacation days, and {sum(v == 'sick' for v in statuses.values())} sick days.")
print(f"Existing differing values to overwrite: {amount_conflicts} amounts, {status_conflicts} statuses.")
if not args.apply:
print("Dry run only; no database changes made.")
return
backup = import_data(database_path, amounts, statuses)
after_amounts, after_statuses, first_date, last_date = inspect_database(database_path)
print(f"Import complete and verified: all {len(amounts)} amounts and {len(statuses)} statuses match.")
print(f"Backup: {backup}")
print(
f"Database after import: {after_amounts} amounts, {after_statuses} statuses, "
f"date range {first_date} through {last_date}."
)
if __name__ == "__main__":
main()