# NOTE: This script was sanitized for the public repository in v6.1.9. # It assumes it is run from the repository root unless JYOTISH_BENCHMARK_ROOT # or JYOTISH_SKILL_SCRIPT is provided. Raw output directories are generated locally # and are intentionally not committed. #!/usr/bin/env python3 import csv import json import math from datetime import datetime, timezone, timedelta from pathlib import Path import swisseph as swe ROOT = Path(__file__).resolve().parents[1] DATA = ROOT / 'data/benchmark_samples.json' OUT = ROOT / 'outputs' SWISS_OUT = OUT / 'swiss_direct' CANON = OUT / 'canonical' PLANETS = { 'Sun': swe.SUN, 'Moon': swe.MOON, 'Mars': swe.MARS, 'Mercury': swe.MERCURY, 'Jupiter': swe.JUPITER, 'Venus': swe.VENUS, 'Saturn': swe.SATURN, 'Rahu': swe.MEAN_NODE, } SIGNS = ['Aries','Taurus','Gemini','Cancer','Leo','Virgo','Libra','Scorpio','Sagittarius','Capricorn','Aquarius','Pisces'] NAKSHATRAS = [ 'Ashwini','Bharani','Krittika','Rohini','Mrigashira','Ardra','Punarvasu','Pushya','Ashlesha', 'Magha','Purva Phalguni','Uttara Phalguni','Hasta','Chitra','Swati','Vishakha','Anuradha','Jyeshtha', 'Mula','Purva Ashadha','Uttara Ashadha','Shravana','Dhanishta','Shatabhisha','Purva Bhadrapada','Uttara Bhadrapada','Revati' ] def normalize(deg): return deg % 360.0 def sign_of(lon): idx = int(normalize(lon) // 30) return SIGNS[idx] def degree_in_sign(lon): return normalize(lon) % 30 def nakshatra_of(lon): unit = 360.0 / 27.0 pos = normalize(lon) / unit idx = int(math.floor(pos)) pada = int(math.floor((pos - idx) * 4)) + 1 if pada > 4: pada = 4 return NAKSHATRAS[idx], pada def julian_day_utc(birth): tz = float(birth['tz']) local = datetime(int(birth['year']), int(birth['month']), int(birth['day']), int(birth['hour']), int(birth['minute'])) utc_dt = local - timedelta(hours=tz) hour = utc_dt.hour + utc_dt.minute / 60 + utc_dt.second / 3600 return swe.julday(utc_dt.year, utc_dt.month, utc_dt.day, hour, swe.GREG_CAL) def calc_swiss(sample): birth = sample['birth'] jd = julian_day_utc(birth) swe.set_sid_mode(swe.SIDM_LAHIRI, 0, 0) flags = swe.FLG_SWIEPH | swe.FLG_SIDEREAL | swe.FLG_SPEED planets = {} for name, pid in PLANETS.items(): res, ret = swe.calc_ut(jd, pid, flags) lon = normalize(res[0]) nak, pada = nakshatra_of(lon) planets[name] = { 'longitude': round(lon, 6), 'sign': sign_of(lon), 'degree_in_sign': round(degree_in_sign(lon), 6), 'nakshatra': nak, 'nakshatra_pada': pada, 'retrograde': bool(res[3] < 0), } rahu_lon = planets['Rahu']['longitude'] ketu_lon = normalize(rahu_lon + 180) nak, pada = nakshatra_of(ketu_lon) planets['Ketu'] = { 'longitude': round(ketu_lon, 6), 'sign': sign_of(ketu_lon), 'degree_in_sign': round(degree_in_sign(ketu_lon), 6), 'nakshatra': nak, 'nakshatra_pada': pada, 'retrograde': planets['Rahu']['retrograde'], } return { 'sample_id': sample['id'], 'engine': 'swiss_direct_lahiri_mean_node', 'julian_day_ut': jd, 'parameters': {'ayanamsa': 'Lahiri', 'node': 'Mean Node', 'flags': int(flags)}, 'planets': planets, } def compare_sample(sample_id, swiss): local_path = CANON / f'{sample_id}.canonical.json' local = json.loads(local_path.read_text()) rows = [] for pname in ['Sun','Moon','Mars','Mercury','Jupiter','Venus','Saturn','Rahu','Ketu']: s = swiss['planets'][pname] l = local['planets'][pname] for field in ['sign','degree_in_sign','nakshatra','nakshatra_pada','retrograde']: sv = s.get(field) lv = l.get(field) status = 'match' delta = '' if field == 'degree_in_sign': try: delta_val = abs(float(sv) - float(lv)) delta = round(delta_val, 6) status = 'match' if delta_val <= 0.1 else 'mismatch' except Exception: status = 'not_comparable' else: status = 'match' if sv == lv else 'mismatch' rows.append({ 'sample_id': sample_id, 'body': pname, 'field': field, 'local_skill': lv, 'swiss_direct': sv, 'delta': delta, 'status': status, }) return rows def write_report(rows): total = len(rows) matches = sum(1 for r in rows if r['status'] == 'match') mismatches = [r for r in rows if r['status'] == 'mismatch'] by_field = {} for r in rows: by_field.setdefault(r['field'], {'total':0, 'match':0, 'mismatch':0}) by_field[r['field']]['total'] += 1 by_field[r['field']][r['status']] = by_field[r['field']].get(r['status'], 0) + 1 lines = [] lines.append('# Jyotish benchmark 第一轮 Swiss direct 对比报告') lines.append('') lines.append('生成时间：2026-06-03') lines.append('') lines.append('## 1. 范围') lines.append('') lines.append('- 对比对象：当前 skill canonical baseline vs 直接调用 Swiss Ephemeris。') lines.append('- 配置：Sidereal Lahiri，Mean Node，行星黄经与 Nakshatra 字段。') lines.append('- 本轮不比较上升、宫位、D9/D10、大运；这些留给下一轮多引擎/参数冻结测试。') lines.append('') lines.append('## 2. 总体结果') lines.append('') lines.append(f'- 字段总数：{total}') lines.append(f'- 匹配：{matches}') lines.append(f'- 不匹配：{len(mismatches)}') lines.append(f'- 匹配率：{matches / total:.2%}' if total else '- 匹配率：N/A') lines.append('') lines.append('## 3. 分字段结果') lines.append('') lines.append('| Field | Total | Match | Mismatch |') lines.append('|---|---:|---:|---:|') for field, stat in sorted(by_field.items()): lines.append(f"| {field} | {stat['total']} | {stat.get('match',0)} | {stat.get('mismatch',0)} |") lines.append('') if mismatches: lines.append('## 4. 不匹配样例') lines.append('') lines.append('| Sample | Body | Field | Local skill | Swiss direct | Delta |') lines.append('|---|---|---|---|---|---:|') for r in mismatches[:80]: lines.append(f"| {r['sample_id']} | {r['body']} | {r['field']} | {r['local_skill']} | {r['swiss_direct']} | {r['delta']} |") lines.append('') lines.append('## 5. 解释') lines.append('') lines.append('- 若 sign/nakshatra 大量一致，说明当前 skill 的核心 Lahiri 行星计算大方向可信。') lines.append('- 若 degree_in_sign 出现系统性差异，优先检查 ayanamsa、True/Mean Node、UTC换算、Swiss flags。') lines.append('- 本轮发现的问题只约束计算层，不直接评价解释和预测能力。') report = OUT / 'jyotish_benchmark_round1_swiss_direct_compare.md' report.write_text('\n'.join(lines)) return report def main(): OUT.mkdir(parents=True, exist_ok=True) SWISS_OUT.mkdir(parents=True, exist_ok=True) samples = json.loads(DATA.read_text()) all_rows = [] for sample in samples: swiss = calc_swiss(sample) (SWISS_OUT / f"{sample['id']}.swiss_direct.json").write_text(json.dumps(swiss, ensure_ascii=False, indent=2)) all_rows.extend(compare_sample(sample['id'], swiss)) csv_path = OUT / 'swiss_direct_comparison_matrix.csv' with csv_path.open('w', newline='') as f: writer = csv.DictWriter(f, fieldnames=['sample_id','body','field','local_skill','swiss_direct','delta','status']) writer.writeheader() writer.writerows(all_rows) report = write_report(all_rows) print(json.dumps({'matrix': str(csv_path), 'report': str(report), 'rows': len(all_rows)}, ensure_ascii=False, indent=2)) if __name__ == '__main__': main()