Does your birth year decide your CPF Retirement Sum?

"""Prep for the CPF Retirement Sums insight (demo).
Reads CPF's two tables (Full and Basic Retirement Sum by date of birth) plus data.gov.sg's copy of the Full Retirement Sum,
checks them against each other, and writes data.json for the page.
Usage: python3 -I prep_cpf.py data_dir out_dir
"""
import csv, json, os, sys
from datetime import date, timedelta
from decimal import Decimal, ROUND_HALF_UP

data_dir, out_dir = sys.argv[1], sys.argv[2]
os.makedirs(out_dir, exist_ok=True)

# Typed from CPF's pages (read 14 Oct 2026): the Enhanced Retirement Sum is set at double the current year's Full Retirement Sum.
ERS = {"2026": 440800, "2027": 456400}
MONTHS = ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"]


def d(s):
    y, m, dd = s.split("-")
    return date(int(y), int(m), int(dd))


def half_up(x, places=0):
    q = Decimal(1).scaleb(-places)
    return Decimal(str(x)).quantize(q, rounding=ROUND_HALF_UP)


def mon(dt):
    return MONTHS[dt.month - 1] + " " + str(dt.year)


frs_rows = [(d(r["dob_on_or_after"]), d(r["turn_55_on_or_after"]), int(r["full_retirement_sum"]))
            for r in csv.DictReader(open(os.path.join(data_dir, "cpf_frs_by_birth_date.csv")))]
brs_by_start = {d(r["dob_on_or_after"]): int(r["basic_retirement_sum"])
                for r in csv.DictReader(open(os.path.join(data_dir, "cpf_brs_by_birth_date.csv")))}
dg = {}
for r in csv.DictReader(open(os.path.join(data_dir, "data_gov_sg_full_retirement_sum.csv"))):
    dd, mm, yy = r["effective_from"].split("/")
    dg[date(int(yy), int(mm), int(dd))] = int(r["full_retirement_sum"])

# ---- checks ----
assert [r[0] for r in frs_rows] == sorted(r[0] for r in frs_rows), "rows must be in date order"
assert all(a[2] < b[2] for a, b in zip(frs_rows, frs_rows[1:])), "the sum must rise at every step"
for dob, t55, amt in frs_rows:
    assert t55 == date(dob.year + 55, dob.month, dob.day), "55th birthday date must be dob + 55 years"
matched = 0
for eff, amt in dg.items():
    row = [r for r in frs_rows if r[1] == eff]
    assert row and row[0][2] == amt, ("data.gov.sg row not matched", eff, amt)
    matched += 1
assert matched == len(dg) == 24
for start, amt in brs_by_start.items():
    f = [r for r in frs_rows if r[0] == start]
    assert f and f[0][2] == 2 * amt, ("Full is not double Basic", start)
assert ERS["2026"] == 2 * [r for r in frs_rows if r[0] == date(1971, 1, 1)][0][2]
assert ERS["2027"] == 2 * [r for r in frs_rows if r[0] == date(1972, 1, 1)][0][2]

# ---- rows ----
rows = []
for i, (dob, t55, amt) in enumerate(frs_rows):
    nxt = frs_rows[i + 1][0] if i + 1 < len(frs_rows) else None
    last = (nxt - timedelta(days=1)) if nxt else None
    if nxt is None or (dob.month == 1 and nxt.month == 1 and nxt.year == dob.year + 1):
        label = str(dob.year)
        t55lab = str(t55.year)
    else:
        label = mon(dob) + " to " + mon(last)
        t55e = date(last.year + 55, last.month, last.day)
        t55lab = mon(t55) + " to " + mon(t55e)
    prev = rows[-1]["frs"] if rows else None
    rows.append({
        "i": i, "from": dob.isoformat(), "to": last.isoformat() if last else None,
        "label": label, "t55": t55lab, "t55year": t55.year, "frs": amt, "brs": brs_by_start.get(dob),
        "rise": (amt - prev) if prev else None,
        "risepct": float(half_up((amt / prev - 1) * 100, 1)) if prev else None,
    })

# ---- the birth-year picker: 1941 to 1972; birth years that straddle two steps are split into Jan-Jun and Jul-Dec ----
def row_for(dt):
    idx = [r["i"] for r in rows if r["from"] <= dt.isoformat()]
    return idx[-1]

years = []
for y in range(1941, 1973):
    a, b = row_for(date(y, 1, 1)), row_for(date(y, 7, 1))
    years.append({"y": y, "h1": a, "h2": b} if a != b else {"y": y, "r": a})

# ---- four decades of births, for the "how fast has it grown" tiles ----
def era_of(r):
    f = r["from"]
    return 0 if f < "1950-07-01" else 1 if f < "1960-07-01" else 2 if f < "1970-01-01" else 3

eras = []
spec = [("Born Jul 1940 to Jun 1950", 0), ("Born Jul 1950 to Jun 1960", 1), ("Born Jul 1960 to Dec 1969", 2), ("Born 1970 to 1972", 3)]
for name, k in spec:
    rr = [r for r in rows if era_of(r) == k]
    first = rr[0]
    nxt = [r for r in rows if era_of(r) == k + 1]
    end = nxt[0] if nxt else rows[-1]
    months = (d(end["from"]).year - d(first["from"]).year) * 12 + d(end["from"]).month - d(first["from"]).month
    eras.append({"k": k, "name": name, "rows": [r["i"] for r in rr], "start_frs": first["frs"], "end_frs": end["frs"],
                 "gain": end["frs"] - first["frs"], "times": float(half_up(end["frs"] / first["frs"], 2)), "months": months,
                 "to_latest": not nxt})

big = max((r for r in rows if r["rise"]), key=lambda r: r["rise"])
assert sum(1 for r in rows if r["rise"] == big["rise"]) == 1
first_frs, latest = rows[0]["frs"], rows[-1]["frs"]
D = {
    "asof": "14 Oct 2026", "n_rows": len(rows), "rows": rows, "years": years, "eras": eras,
    "ers": ERS, "first_frs": first_frs, "latest_frs": latest, "latest_brs": rows[-1]["brs"],
    "times_first": float(half_up(latest / first_frs, 1)), "biggest": {"i": big["i"], "rise": big["rise"], "pct": big["risepct"]},
    "same_sum_row": [r["i"] for r in rows if r["to"] and r["from"] == "1960-07-01"][0],
    "steady_since": [r["i"] for r in rows if r["from"] == "1968-01-01"][0],
    "dg_rows": len(dg), "dg_first": min(dg).isoformat(), "dg_last": max(dg).isoformat(),
}
json.dump(D, open(os.path.join(out_dir, "data.json"), "w"), indent=1)

print("CPF rows", len(rows), "| picker years", len(years), "| data.gov.sg rows matched", matched, "of", len(dg), "(Jul 2003 to Jan 2027)")
print("first", first_frs, "latest", latest, "times", D["times_first"], "| BRS published for", sum(1 for r in rows if r["brs"]), "rows")
print("biggest rise", big["rise"], "at", big["label"], big["risepct"], "%")
for e in eras:
    print(e["name"], e["start_frs"], "->", e["end_frs"], "gain", e["gain"], "times", e["times"], "months", e["months"])
print("split years:", [y["y"] for y in years if "h1" in y][:3], "...", [y["y"] for y in years if "h1" in y][-1], "| single years from", [y["y"] for y in years if "r" in y][0])
for r in rows[-8:]:
    print(r["label"], r["t55"], r["frs"], r["brs"], r["rise"], r["risepct"])