#!/usr/bin/env python3
"""One command rebuilds everything.

  python3 run.py --offline   rebuild every number and chart from the committed snapshots, then verify them
  python3 run.py             pull fresh keyless data first, then rebuild
  python3 run.py --bq-light  also refresh the two rolling BigQuery queries, about 5 GB (skipped without credentials)
  python3 run.py --bq        also refresh the larger BigQuery snapshots (needs GOOGLE_APPLICATION_CREDENTIALS, free sandbox works)
  python3 run.py --explorer  also rebuild swap aggregates from local Explorer rows

Offline needs Python 3.9+ and nothing else. Node is used only by the model parity test, which is skipped if absent.
"""
import argparse, gzip, hashlib, json, os, pathlib, shutil, subprocess, sys, time, zlib

ROOT = pathlib.Path(__file__).resolve().parent
PY = sys.executable


def step(name, *args, ok_codes=(0,)):
    t0 = time.time()
    res = subprocess.run([PY, *map(str, args)], cwd=ROOT)
    print(f"[{name}] exit {res.returncode}, {time.time() - t0:.1f}s", flush=True)
    if res.returncode not in ok_codes:
        raise SystemExit(f"step failed: {name}")


def content_hash(path):
    """sha256 of a snapshot's decompressed content, or None if the file is damaged."""
    try:
        with gzip.open(path, "rb") as fh:
            return hashlib.sha256(fh.read()).hexdigest()
    except (OSError, EOFError, zlib.error):
        return None


def verify_snapshots():
    """Every committed snapshot must hash to its newest manifest entry."""
    manifest = ROOT / "raw" / "MANIFEST.jsonl"
    latest = {}
    for line in manifest.read_text().splitlines():
        rec = json.loads(line)
        latest[rec["name"]] = rec
    checked = bad = 0
    for name, rec in latest.items():
        path = ROOT / "raw" / "latest" / f"{name}.json.gz"
        if not path.exists():
            continue
        got = content_hash(path)
        checked += 1
        if got != rec["sha256"]:
            bad += 1
            print(f"HASH MISMATCH {name}")
    bq = ROOT / "raw" / "bigquery" / "MANIFEST.jsonl"
    latest_bq = {}
    for line in bq.read_text().splitlines():
        rec = json.loads(line)
        latest_bq[rec["file"]] = rec
    for name, rec in latest_bq.items():
        got = content_hash(ROOT / "raw" / "bigquery" / name)
        checked += 1
        if got != rec["sha256_csv"]:
            bad += 1
            print(f"HASH MISMATCH {name}")
    ex = ROOT / "raw" / "explorer" / "MANIFEST.json"
    if ex.exists():
        for rec in json.loads(ex.read_text())["files"]:
            got = content_hash(ROOT / "raw" / "explorer" / rec["file"])
            checked += 1
            if got != rec["sha256_csv"]:
                bad += 1
                print(f"HASH MISMATCH {rec['file']}")
    print(f"snapshots verified: {checked} ok, {bad} bad")
    if bad:
        raise SystemExit("raw snapshots do not match the manifest")


def main():
    ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
    ap.add_argument("--offline", action="store_true")
    ap.add_argument("--bq", action="store_true")
    ap.add_argument("--bq-light", action="store_true")
    ap.add_argument("--explorer", action="store_true")
    args = ap.parse_args()
    if not args.offline:
        step("public data", "pipeline/fetch_public.py", ok_codes=(0, 1))
        step("chain data", "pipeline/fetch_chain.py", ok_codes=(0, 1))
    if args.bq:
        step("bigquery", "pipeline/run_bq.py", "--only", "01,02,09,10,12,13", "--cap-gb", "120", ok_codes=(0, 1))
    if args.bq_light:
        if os.environ.get("GOOGLE_APPLICATION_CREDENTIALS"):
            step("bigquery rolling", "pipeline/run_bq.py", "--only", "01b,09b", "--cap-gb", "12", ok_codes=(0, 1))
        else:
            print("[bigquery rolling] skipped: no credentials, using committed snapshots")
    if args.explorer:
        step("explorer aggregates", "pipeline/analyze_explorer.py")
    verify_snapshots()
    step("analysis", "pipeline/analyze.py")
    step("model", "pipeline/model.py")
    step("kill criteria", "pipeline/grade.py", *(["--offline"] if args.offline else []))
    step("site", "pipeline/build.py")
    step("claims", "pipeline/verify_claims.py")
    tests = ["tests/test_pipeline.py"]
    if shutil.which("node"):
        tests.append("tests/test_model.py")
    for t in tests:
        step(t, t)
    print("done: open docs/index.html through any static server, e.g. python3 -m http.server -d docs 8000")


if __name__ == "__main__":
    main()