#!/usr/bin/env python3
# verify_ledger_dir_census.py
# Real test: re-derive the live inventory of wizard-work/ledger from the
# filesystem and compare byte-for-byte with the committed artifact. Exits 0
# iff identical. The artifact CSV is excluded from both sides so the check is
# self-consistent across regeneration.
import hashlib, os, csv, sys

HERE = os.path.dirname(os.path.abspath(__file__))          # wizard-work/ledger
REPO_ROOT = os.path.abspath(os.path.join(HERE, "..", ".."))  # troupe-standup root
ARTIFACT = os.path.join(REPO_ROOT, "wizard-work", "ledger", "2026-10-30-ledger-dir-census.csv")
SRC = HERE

def re_derive():
    rows = []
    for dirpath, dirnames, filenames in os.walk(SRC):
        dirnames.sort()
        for fn in sorted(filenames):
            full = os.path.join(dirpath, fn)
            if os.path.abspath(full) == os.path.abspath(ARTIFACT):
                continue
            rel = os.path.relpath(full, REPO_ROOT)
            size = str(os.path.getsize(full))
            h = hashlib.sha256()
            with open(full, "rb") as f:
                for chunk in iter(lambda: f.read(1024 * 1024), b""):
                    h.update(chunk)
            rows.append((rel, size, h.hexdigest()))
    rows.sort(key=lambda r: r[0])
    return rows

def read_committed(path):
    with open(path, newline="", encoding="utf-8") as f:
        r = csv.reader(f)
        header = next(r)
        assert header == ["rel_path", "size_bytes", "sha256"], f"bad header {header}"
        return [tuple(x) for x in r]

committed = read_committed(ARTIFACT)
derived = re_derive()

print(f"artifact: {ARTIFACT}")
print(f"rows committed: {len(committed)}   rows re-derived: {len(derived)}")

if committed == derived:
    print("PASS: committed artifact matches live re-derivation.")
    sys.exit(0)

print("FAIL: mismatch (committed vs re-derived).")
n = min(len(committed), len(derived))
for i in range(n):
    if committed[i] != derived[i]:
        print(f"  first diff at row {i}:")
        print(f"    committed: {committed[i]}")
        print(f"    derived:   {derived[i]}")
        break
else:
    if len(committed) < len(derived):
        print(f"  derived has {len(derived)-len(committed)} extra rows")
    else:
        print(f"  committed has {len(committed)-len(derived)} extra rows")
sys.exit(1)