#!/usr/bin/env python3
"""Reproduce source-derived SPHINCS+ SHAKE-128s/128f size arithmetic.

Run with Python 3 (standard library only):
  python3 analyze_sphincs_sizes.py            # refetch pinned upstream sources
  python3 analyze_sphincs_sizes.py --offline  # verify bundled source snapshots

No cryptographic implementation, wallet binary, or benchmark is executed.
"""
import argparse
from fractions import Fraction
import hashlib
import json
from pathlib import Path
import re
import urllib.request

REVISION = '7ec789ace6874d875f4bb84cb61b81155398167e'
SOURCE_CHECK_DATE = '2026-09-21'
SOURCE_RECORDS = [{'path': 'README.md', 'url': 'https://raw.githubusercontent.com/sphincs/sphincsplus/7ec789ace6874d875f4bb84cb61b81155398167e/README.md', 'sha256': '6bc7f7a5bb2d47059d7e6c0115e358ba832523a1f033c56e086988d66a46c754', 'bytes': 3108}, {'path': 'ref/params/params-sphincs-shake-128s.h', 'url': 'https://raw.githubusercontent.com/sphincs/sphincsplus/7ec789ace6874d875f4bb84cb61b81155398167e/ref/params/params-sphincs-shake-128s.h', 'sha256': 'da4f8c1ee5b475a8103d856d80dbec15276eb2f566ce44318dcd286faba05c1d', 'bytes': 2130}, {'path': 'ref/params/params-sphincs-shake-128f.h', 'url': 'https://raw.githubusercontent.com/sphincs/sphincsplus/7ec789ace6874d875f4bb84cb61b81155398167e/ref/params/params-sphincs-shake-128f.h', 'sha256': 'de37a1ad4de21af24c70262a97b74927a0decb8275eae1ad34dcf5ed52daa2e1', 'bytes': 2130}]
ROOT = Path(__file__).resolve().parent


def checked_sources(offline):
    texts = {}
    for record in SOURCE_RECORDS:
        local = ROOT / 'sphincs-size-sources' / record['path']
        if offline:
            data = local.read_bytes()
        else:
            request = urllib.request.Request(record['url'], headers={
                'User-Agent': 'SYNX-editorial-source-check'})
            with urllib.request.urlopen(request, timeout=30) as response:
                data = response.read()
        actual = hashlib.sha256(data).hexdigest()
        assert actual == record['sha256'], ('Source digest changed', record['path'], actual)
        assert len(data) == record['bytes'], ('Source byte length changed', record['path'])
        texts[record['path']] = data.decode('utf-8')
    return texts


def main():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument('--offline', action='store_true')
    args = parser.parse_args()
    texts = checked_sources(args.offline)
    rows = {}
    for line in texts['README.md'].splitlines():
        cells = [cell.strip() for cell in line.strip().strip('|').split('|')]
        if cells[0] in ('SPHINCS+-128s', 'SPHINCS+-128f'):
            assert len(cells) == 11, ('Unexpected table width', cells)
            assert cells[0] not in rows, ('Duplicate parameter row', cells[0])
            rows[cells[0]] = [int(value.replace(',', '')) for value in cells[1:]]
    assert set(rows) == {'SPHINCS+-128s', 'SPHINCS+-128f'}
    expected = {'128s': (32, 64, 7856), '128f': (32, 64, 17088)}
    mapping = {'SPX_N': 0, 'SPX_FULL_HEIGHT': 1, 'SPX_D': 2,
               'SPX_FORS_HEIGHT': 3, 'SPX_FORS_TREES': 4, 'SPX_WOTS_W': 5}
    sizes = {}
    checks = ['All three pinned source SHA-256 digests and byte lengths match.',
              'README contains exactly one 128s row and one 128f row.']
    for variant, expected_sizes in expected.items():
        row = rows['SPHINCS+-' + variant]
        assert tuple(row[-3:]) == expected_sizes, ('Unexpected key/signature sizes', variant)
        header = texts['ref/params/params-sphincs-shake-' + variant + '.h']
        for macro, column in mapping.items():
            match = re.search(r'^#define\s+' + macro + r'\s+(\d+)\s*$', header, re.M)
            assert match is not None, ('Missing parameter constant', variant, macro)
            assert int(match.group(1)) == row[column], ('SHAKE/README mismatch', variant, macro)
        sizes['SPHINCS+-SHAKE-' + variant] = dict(zip(
            ('public_key_bytes', 'secret_key_bytes', 'signature_bytes'), expected_sizes))
        checks.append(variant + ': README sizes match expected values and all six SHAKE-header parameters.')
    small = sizes['SPHINCS+-SHAKE-128s']['signature_bytes']
    fast = sizes['SPHINCS+-SHAKE-128f']['signature_bytes']
    delta = fast - small
    ratio = Fraction(fast, small)
    reduction = Fraction(delta, fast)
    assert delta == 9232
    assert ratio == Fraction(1068, 491)
    assert reduction == Fraction(577, 1068)
    checks.extend(['Signature delta is exactly 9232 bytes.',
                   '128f/128s signature-size ratio is exactly 1068/491.',
                   'Reduction relative to 128f is exactly 577/1068.'])
    conclusion = (
        'We compared the upstream SPHINCS+ parameter table with its SHAKE-specific '
        'headers. Both variants list 32-byte public keys and 64-byte secret keys. '
        'The 128s signature is 7,856 bytes versus 17,088 bytes for 128f: 9,232 fewer '
        'bytes, a 54.03% reduction. This source-derived calculation measures neither '
        'runtime nor compressed storage and does not validate SYNX binaries or '
        'explain the project\'s historical parameter choice.')
    assert 40 <= len(conclusion.split()) <= 80
    result = {
        'analysis': 'SPHINCS+ SHAKE-128s versus SHAKE-128f published byte sizes',
        'source_check_date': SOURCE_CHECK_DATE,
        'source_repository': 'https://github.com/sphincs/sphincsplus',
        'source_revision': REVISION,
        'source_revision_date': '2024-04-15T16:30:51Z',
        'primary_source': 'https://github.com/sphincs/sphincsplus/blob/' + REVISION + '/README.md',
        'sources': SOURCE_RECORDS,
        'method': 'Parse the pinned README parameter rows; match six parameters against the SHAKE-specific headers; calculate exact rational differences. No source code is compiled or executed.',
        'parameter_table': sizes,
        'calculations': {
            'signature_bytes_saved_128s_vs_128f': delta,
            'signature_ratio_128f_over_128s_exact': str(ratio),
            'signature_ratio_128f_over_128s_decimal': round(float(ratio), 10),
            'signature_reduction_128s_relative_to_128f_exact': str(reduction),
            'signature_reduction_percent': round(float(reduction * 100), 10),
            'signature_reduction_percent_display': f'{float(reduction * 100):.2f}%',
            'public_key_byte_difference': 0,
            'secret_key_byte_difference': 0,
            'formulas': {
                'delta': '17088 - 7856 = 9232',
                'ratio': '17088 / 7856 = 1068 / 491',
                'reduction': '(17088 - 7856) / 17088 = 577 / 1068'}},
        'assertions_passed': checks,
        'conclusion': conclusion,
        'limitations': [
            'Raw published key/signature sizes only; no signing or verification runtime measurement.',
            'No compressed-storage, transaction-envelope, block-size, network-throughput or bandwidth benchmark.',
            'No SYNX release binary was executed or independently validated.',
            'No inference about why SYNX historically chose a parameter set.',
            'No claim that SPHINCS+ names establish FIPS-conformant SLH-DSA product validation.']}
    output = ROOT / 'sphincs-size-analysis.json'
    output.write_text(json.dumps(result, indent=2, ensure_ascii=False) + '\n')
    print(json.dumps({'output': str(output), 'revision': REVISION,
                      'delta_bytes': delta, 'reduction_percent': f'{float(reduction * 100):.2f}%',
                      'checks': len(checks)}, indent=2))


if __name__ == '__main__':
    main()