"""F4-dev-01: extract headline LFS Sept 2026 figures and assess an assertion.
Reads only data/; writes out/results.json deterministically. No randomness (random_seed = null).
Prints every matched source line so the extraction can be audited."""
import json, os, re

ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
SRC_ID = "statcan-daily-lfs-2026-09"
SRC = os.path.join(ROOT, "data", "sources", SRC_ID + ".txt")
README = os.path.join(ROOT, "data", "README.md")


def count_lines(p):  # equivalent to `wc -l`: number of newline characters
    with open(p, "rb") as f:
        return f.read().count(b"\n")


def sectioned_lines(path):
    """Return list of (line_no, current '## ' heading (not '###'), text)."""
    out, sec = [], None
    with open(path, encoding="utf-8") as f:
        for i, line in enumerate(f, 1):
            line = line.rstrip("\n")
            m = re.match(r"^##\s+(?!#)(.*\S)\s*$", line)
            if m:
                sec = m.group(1)
            out.append((i, sec, line))
    return out


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


lines = sectioned_lines(SRC)

# --- national unemployment rate: sentences "The unemployment rate <verb> [x percentage points] to y%"
ur_pat = re.compile(r"\bThe unemployment rate (increased|rose|fell|decreased|declined)"
                    r" (\d+\.\d) percentage points to (\d+\.\d)%")
ur_hits = [(i, sec, m.group(3), t) for i, sec, t in lines for m in ur_pat.finditer(t)]

# --- national employment change: line-initial "Employment <verb> by N" / "In September, employment <verb> by N"
emp_pat = re.compile(r"^(?:Employment|In September, employment) (declined|fell|decreased|increased|rose) by (\d{1,3}(?:,\d{3})+)")
emp_hits = []
for i, sec, t in lines:
    m = emp_pat.search(t)
    if m and sec is not None:
        emp_hits.append((i, sec, m.group(1), m.group(2), t))

print("Matched national unemployment-rate lines:")
for h in ur_hits:
    print(f"  L{h[0]} [## {h[1]}]: {h[3]}")
print("Matched national employment-change lines:")
for h in emp_hits:
    print(f"  L{h[0]} [## {h[1]}]: {h[4]}")

ur_values = sorted({h[2] for h in ur_hits})
assert len(ur_values) == 1, f"inconsistent national unemployment rates: {ur_values}"
emp_values = sorted({(h[2] in ("declined", "fell", "decreased"), h[3]) for h in emp_hits})
assert len(emp_values) == 1, f"inconsistent employment changes: {emp_values}"

ur_primary = next(h for h in ur_hits if h[1] == "Highlights")
emp_primary = next(h for h in emp_hits if h[1] == "Highlights")
unemployment_rate = float(ur_primary[2])
neg, num = emp_values[0]
employment_change = (-1 if neg else 1) * int(num.replace(",", ""))

# --- assertion about core-aged women and men
women_pat = re.compile(r"core-aged women \(25 to 54 years old\), the unemployment rate (rose|fell|increased|decreased) (\d+\.\d) percentage points to (\d+\.\d)%")
men_pat = re.compile(r"the unemployment rate for core-aged men (rose|fell|increased|decreased) (\d+\.\d) percentage points to (\d+\.\d)%")
women = men = None
for i, sec, t in lines:
    m = women_pat.search(t)
    if m:
        women = (i, sec, m.group(1), m.group(2), m.group(3), m.group(0))
    m = men_pat.search(t)
    if m:
        men = (i, sec, m.group(1), m.group(2), m.group(3), m.group(0))
assert women is not None and men is not None, "assertion evidence not found"
print("Assertion evidence:")
print(f"  women: L{women[0]} [## {women[1]}]: {women[5]}")
print(f"  men:   L{men[0]} [## {men[1]}]: {men[5]}")

elements = {
    "women_direction_rose": women[2] in ("rose", "increased"),
    "women_change_0.3pp": women[3] == "0.3",
    "women_level_5.3pct": women[4] == "5.3",
    "men_direction_rose": men[2] in ("rose", "increased"),
    "men_change_0.2pp": men[3] == "0.2",
    "men_level_5.8pct": men[4] == "5.8",
}
for k, v in elements.items():
    print(f"  {k}: {'stated' if v else 'CONTRADICTED'}")
verdict = "supported" if all(elements.values()) else "not_supported"
assertion_locator = " ; ".join(loc(s) for s in sorted({women[1], men[1]}))
if verdict == "not_supported":
    reason = (f"The women's part is stated (\"{women[5]}\"), but the release says \"{men[5]}\", "
              f"which contradicts the asserted rise for core-aged men (the 0.2-point magnitude and 5.8% level match; the direction does not).")
else:
    reason = "All six asserted elements are printed in the release."

results = {
    "task_id": "F4-dev-01",
    "unemployment_rate_sep2026": unemployment_rate,
    "employment_change_sep2026_persons": employment_change,
    "locators": {
        "unemployment_rate_sep2026": loc(ur_primary[1]),
        "employment_change_sep2026_persons": loc(emp_primary[1]),
    },
    "corroborating_locators": {
        "unemployment_rate_sep2026": sorted({loc(h[1]) for h in ur_hits if h[1] != "Highlights"}),
        "employment_change_sep2026_persons": sorted({loc(h[1]) for h in emp_hits if h[1] != "Highlights"}),
    },
    "n_matches_unemployment_rate": len(ur_hits),
    "n_matches_employment_change": len(emp_hits),
    "assertion_verdict": verdict,
    "assertion_locator": assertion_locator,
    "assertion_reason": reason,
    "assertion_elements": elements,
    "feasibility": "complete",
    "feasibility_reason": None,
    "data_issues": [
        {"issue": "data/README.md ('Rights' line) mentions 'RESEA Tables 2-3, London Table 3 header', and the extract header mentions regression tables; neither exists in this extract (apparent boilerplate shared with other bundles).",
         "handling": "Ignored; it has no bearing on the release text or on the extracted numbers."},
        {"issue": "Lines 53-60 (layoff, job-finding and participation paragraphs and the population-aging text box) have no '## ' heading of their own and fall under '## Infographic 2' in the flattened text; the SDG goal list after 'following goals:' is not reproduced.",
         "handling": "Recorded only; none of the requested figures or assertion sentences are in these parts."},
    ],
    "n_input_rows": {
        "README.md": count_lines(README),
        SRC_ID + ".txt": count_lines(SRC),
    },
    "n_analysed": 3,
    "random_seed": None,
}
with open(os.path.join(ROOT, "out", "results.json"), "w") as f:
    json.dump(results, f, indent=2, sort_keys=True)
    f.write("\n")
print(f"unemployment_rate_sep2026={unemployment_rate}; employment_change_sep2026_persons={employment_change}; verdict={verdict}")