from datetime import datetime
from dateutil.relativedelta import relativedelta
import pandas as pd

# ------------------------------------------------------------
# Configuration
# ------------------------------------------------------------

INPUT_FILE = "CalendarDifferenceCases.csv"
OUTPUT_FILE = "PythonResults.csv"

# ------------------------------------------------------------
# Helpers
# ------------------------------------------------------------

def parse_datetime(s: str) -> datetime:
    return datetime.fromisoformat(s)


def decompose(a: datetime, b: datetime):
    """Return normalized calendar difference."""

    if a > b:
        a, b = b, a

    rd = relativedelta(b, a)

    reconstructed = a + rd

    if reconstructed != b:
        raise RuntimeError(
            f"Python reconstruction failed:\n"
            f"From : {a}\n"
            f"To   : {b}\n"
            f"Recon: {reconstructed}\n"
            f"Diff : {rd}"
        )

    return (
        rd.years,
        rd.months,
        rd.days,
        rd.hours,
        rd.minutes,
        rd.seconds,
    )


# ------------------------------------------------------------
# Main
# ------------------------------------------------------------

df = pd.read_csv(INPUT_FILE)

results = []

count = len(df)

for i, (_, row) in enumerate(df.iterrows(), start=1):

    a = parse_datetime(row["from"])
    b = parse_datetime(row["to"])

    y, m, d, hh, mm, ss = decompose(a, b)

    results.append({
        "from": row["from"],
        "to": row["to"],
        "years": y,
        "months": m,
        "days": d,
        "hours": hh,
        "minutes": mm,
        "seconds": ss,
    })

    if i % 100000 == 0:
        print(f"{i:,} / {count:,}")

pd.DataFrame(results).to_csv(
    OUTPUT_FILE,
    index=False
)

print()
print("Finished.")
print(f"Output written to '{OUTPUT_FILE}'")