"""F4-dev-01 replication script. Reads only data/, writes out/results.json deterministically.
Values are extracted by regex from the release text; matched source lines are printed."""
import json, re, hashlib, os

SRC_ID = "statcan-daily-lfs-2026-09"
SRC = f"data/sources/{SRC_ID}.txt"
README = "data/README.md"
EXPECTED_SHA = "7ea5fec57c87874ff5ee09a8304927a1aaed5a1837db5312fab9f7a2d5f2d0bb"

def wc_l(path):
    with open(path, "rb") as f:
        return f.read().count(b"\n")

raw = open(SRC, "rb").read()
sha = hashlib.sha256(raw).hexdigest()
assert sha == EXPECTED_SHA, f"SHA mismatch: {sha}"
lines = raw.decode("utf-8").split("\n")

# map each line to the most recent '## ' section heading (level-2 only, as required by TASK.md)
section_of, cur = [], None
for ln in lines:
    if ln.startswith("## "):
        cur = ln[3:].strip()
    section_of.append(cur)

def find(pattern):
    hits = []
    for i, ln in enumerate(lines):
        m = re.search(pattern, ln)
        if m:
            hits.append((i + 1, section_of[i], m, ln))
    return hits

def loc(section):
    return f'{SRC_ID} | section "{section}"'

# (a1) national unemployment rate -- first statement in Highlights
ur_hits = find(r"The unemployment rate (?:increased|rose|fell|decreased) \d+\.\d percentage points to (\d+\.\d)%")
ur_nat = [h for h in ur_hits if h[1] in ("Highlights", "Unemployment rate increases to 6.5%")]
ur_vals = {float(h[2].group(1)) for h in ur_nat}
assert len(ur_vals) == 1, ur_vals
ur = ur_vals.pop()
ur_primary = next(h for h in ur_nat if h[1] == "Highlights")

# (a2) national employment change
emp_hits = find(r"^(?:In September, e|E)mployment (declined|increased|rose|fell) by ([\d,]+) \(")
emp_vals = set()
for h in emp_hits:
    sign = -1 if h[2].group(1) in ("declined", "fell") else 1
    emp_vals.add(sign * int(h[2].group(2).replace(",", "")))
assert len(emp_vals) == 1, emp_vals
emp = emp_vals.pop()
emp_primary = next(h for h in emp_hits if h[1] == "Highlights")

print("Matched source lines:")
for tag, hs in (("unemployment_rate", ur_nat), ("employment_change", emp_hits)):
    for (n, sec, m, ln) in hs:
        print(f"  [{tag}] line {n} | section \"{sec}\": {ln}")

# (b) assertion: women +0.3 to 5.3%; men ROSE 0.2 to 5.8%
w = find(r"core-aged women \(25 to 54 years old\), the unemployment rate (rose|fell) (\d+\.\d) percentage points to (\d+\.\d)%")
mn = find(r"unemployment rate for core-aged men (rose|fell) (\d+\.\d) percentage points to (\d+\.\d)%")
assert len(w) == 1 and len(mn) == 1
wd, wpp, wlvl = w[0][2].groups()
md, mpp, mlvl = mn[0][2].groups()
print(f"  [assertion women] line {w[0][0]} | section \"{w[0][1]}\": {w[0][3]}")
print(f"  [assertion men]   line {mn[0][0]} | section \"{mn[0][1]}\": {mn[0][3]}")
women_ok = (wd == "rose" and wpp == "0.3" and wlvl == "5.3")
men_ok = (md == "rose" and mpp == "0.2" and mlvl == "5.8")
verdict = "supported" if (women_ok and men_ok) else "not_supported"
assert w[0][1] == mn[0][1]
reason = (f"The women's element matches (\"rose {wpp} percentage points to {wlvl}%\"), but the release states that "
          f"\"the unemployment rate for core-aged men {md} {mpp} percentage points to {mlvl}% in the month\", "
          f"contradicting the claim that it rose." if not men_ok else
          "Every element of the assertion is stated in the release.")

results = {
    "task_id": "F4-dev-01",
    "unemployment_rate_sep2026": ur,
    "employment_change_sep2026_persons": emp,
    "locators": {
        "unemployment_rate_sep2026": loc(ur_primary[1]),
        "employment_change_sep2026_persons": loc(emp_primary[1]),
    },
    "assertion_verdict": verdict,
    "assertion_locator": loc(w[0][1]),
    "assertion_reason": reason,
    "feasibility": "complete",
    "feasibility_reason": None,
    "data_issues": [
        {"issue": "data/README.md rights note refers to re-set tables ('RESEA Tables 2-3, London Table 3 header') and regression-table rendering that do not exist in this extract (apparently template text from another task).",
         "handling": "Ignored; irrelevant to the LFS extract, whose SHA-256 matches the README value. No tables are reproduced in the extract."},
        {"issue": "Some text (e.g. layoff rate, participation rate, population-aging text box) appears under '## Infographic 2' because chart/infographic titles are rendered as '##' headings; the text-box title is not a '##' heading.",
         "handling": "Not relevant to the extracted figures, which appear under substantive narrative headings; locators use the nearest preceding '##' heading."},
    ],
    "n_input_rows": {"README.md": wc_l(README), f"{SRC_ID}.txt": wc_l(SRC)},
    "n_analysed": 3,
    "random_seed": None,
    "baseline_consistency": {
        "unemployment_rate_statements_agree": True,
        "n_unemployment_rate_statements": len(ur_nat),
        "employment_change_statements_agree": True,
        "n_employment_change_statements": len(emp_hits),
        "alternate_locator_unemployment_rate": loc("Unemployment rate increases to 6.5%"),
        "alternate_locator_employment_change": loc("Employment falls for second consecutive month"),
    },
    "source_sha256": sha,
}
os.makedirs("out", exist_ok=True)
with open("out/results.json", "w") as f:
    json.dump(results, f, indent=2, ensure_ascii=False)
    f.write("\n")
print(f"unemployment_rate_sep2026={ur}; employment_change_sep2026_persons={emp}; assertion_verdict={verdict}")