#!/usr/bin/env python3
"""Verify the Hermes profile census artifact (Extract lane).

Checks:
  1. The CSV header is exactly: profile,archives,is_ledger
  2. Every data row has 3 fields, numeric archives/is_ledger
  3. The profile set and counts match a fresh walk of /root/.hermes/profiles
  4. is_ledger flag is correct for every row

Exit code 0 = correct, nonzero = wrong.
"""
import csv
import os
import sys

DEFAULT_CSV = "/root/.hermes/wizard-outputs/ledger/profiles-census.csv"
PROFILES_DIR = "/root/.hermes/profiles"


def expected_rows():
    rows = {}
    for name in os.listdir(PROFILES_DIR):
        pdir = os.path.join(PROFILES_DIR, name)
        if not os.path.isdir(pdir):
            continue
        n = 0
        for root, dirs, files in os.walk(pdir):
            for f in files:
                if f.endswith(".txt"):
                    n += 1
        rows[name] = (n, 1 if name == "ledger" else 0)
    return rows


def main():
    csv_path = sys.argv[1] if len(sys.argv) > 1 else DEFAULT_CSV
    with open(csv_path, newline="") as fh:
        reader = csv.reader(fh)
        rows = list(reader)

    errors = []

    if not rows or rows[0] != ["profile", "archives", "is_ledger"]:
        errors.append("bad header: %r" % (rows[0] if rows else None))

    data = rows[1:]
    if not data:
        errors.append("no data rows")

    expected = expected_rows()
    if len(data) != len(expected):
        errors.append("row count %d != expected %d" % (len(data), len(expected)))

    for r in data:
        if len(r) != 3:
            errors.append("row has %d fields: %r" % (len(r), r))
            continue
        name, a, l = r
        if not (a.isdigit() and l.isdigit()):
            errors.append("non-numeric fields in row: %r" % r)
            continue
        want = expected.get(name)
        if want is None:
            errors.append("unexpected profile %r" % name)
        elif (int(a), int(l)) != want:
            errors.append("mismatch for %r: got (%s,%s) want %s" % (name, a, l, want))

    extra = set(expected) - {r[0] for r in data}
    for name in sorted(extra):
        errors.append("missing profile %r" % name)

    if errors:
        print("FAIL")
        for e in errors:
            print(" -", e)
        sys.exit(1)

    total = sum(int(r[1]) for r in data)
    print("OK: %d profiles, %d total txts, ledger=%s" % (len(data), total, [r for r in data if r[2] == "1"]))
    sys.exit(0)


if __name__ == "__main__":
    main()