feat: sync real case website e2e contract

2026-07-20 13:55:54 +08:00
parent e47aedcc50
commit a509e27ec9
11 changed files with 1228 additions and 2 deletions
@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+"""Create CI-safe capture packets for Muhurta numeric source candidates."""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+from pathlib import Path
+from typing import Any
+
+
+ROOT = Path(__file__).resolve().parents[1]
+TRIAGE = ROOT / "references/oracle/public_worked_example_source_triage_2026_07_20.json"
+OUT = ROOT / "references/oracle/muhurta_numeric_candidate_capture_packet_2026_07_20.json"
+
+
+def sha(obj: Any) -> str:
+    return hashlib.sha256(json.dumps(obj, ensure_ascii=False, sort_keys=True).encode("utf-8")).hexdigest()
+
+
+def next_path(source_id: str) -> str:
+    return f"references/oracle/artifacts/{source_id}_raw_capture_packet.json"
+
+
+def build(date: str) -> dict[str, Any]:
+    triage = json.loads(TRIAGE.read_text(encoding="utf-8"))
+    rows = []
+    for src in triage["sources"]:
+        if src["domain"] != "muhurta_factor_scoring" or not src["numeric_fields_present"]:
+            continue
+        request = {
+            "source_id": src["source_id"],
+            "url": src["url"],
+            "topic": src["topic"],
+            "observed_numeric_fields": src["observed_numeric_fields"],
+        }
+        missing = list(src["missing_for_oracle"])
+        for field in ["raw_capture_hash", "exact_method_settings", "replay_comparison"]:
+            if field not in missing:
+                missing.append(field)
+        rows.append(
+            {
+                "source_id": src["source_id"],
+                "domain": src["domain"],
+                "topic": src["topic"],
+                "url": src["url"],
+                "source_observation_hash": src["observation_hash"],
+                "canonical_request_hash": sha(request),
+                "observed_numeric_fields": src["observed_numeric_fields"],
+                "raw_capture_status": "pending_raw_page_capture",
+                "upgrade_status": "not_oracle_ready",
+                "missing_for_oracle": missing,
+                "next_artifact_path": next_path(src["source_id"]),
+                "claim_boundary": "Numeric-looking public source; not oracle-ready until raw page, exact settings, hash, and local replay comparison are archived.",
+            }
+        )
+    return {
+        "scope": "muhurta_numeric_candidate_capture_packet",
+        "created_at": date,
+        "status": "capture_packet_ready",
+        "claim_status": "source_intake_only",
+        "production_tuning_allowed": False,
+        "truth_matrix_allowed": False,
+        "sources": {"source_triage": str(TRIAGE.relative_to(ROOT))},
+        "summary": {
+            "candidate_count": len(rows),
+            "oracle_ready_count": 0,
+            "pending_raw_capture_count": sum(1 for row in rows if row["raw_capture_status"] == "pending_raw_page_capture"),
+        },
+        "capture_rows": rows,
+        "boundary": "Capture packet staging only; does not calculate or validate Muhurta verdicts.",
+    }
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--date", default="2026-07-20")
+    args = parser.parse_args()
+    print(json.dumps(build(args.date), ensure_ascii=False, indent=2, sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
@@ -0,0 +1,77 @@
+#!/usr/bin/env python3
+"""Create Prashna input contract and numeric oracle candidate queue."""
+from __future__ import annotations
+import argparse, hashlib, json
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+CONTRACT = ROOT / "references/oracle/prashna_input_contract_2026_07_20.json"
+QUEUE = ROOT / "references/oracle/prashna_numeric_oracle_packet_queue_2026_07_20.json"
+
+def h(obj):
+    return hashlib.sha256(json.dumps(obj, ensure_ascii=False, sort_keys=True).encode()).hexdigest()
+
+def build(date: str):
+    contract = {
+        "scope": "prashna_input_contract",
+        "created_at": date,
+        "status": "contract_ready",
+        "claim_status": "ready_contract",
+        "production_tuning_allowed": False,
+        "truth_matrix_allowed": False,
+        "required_fields": [
+            {"field": "question_datetime_local", "format": "YYYY-MM-DDTHH:MM:SS", "boundary": "exact time question is received/accepted"},
+            {"field": "location", "format": "lat/lon + place label", "boundary": "place of querent/astrologer must be explicit"},
+            {"field": "timezone", "format": "IANA or UTC offset", "boundary": "no implicit local machine timezone"},
+            {"field": "ayanamsa", "format": "named sidereal ayanamsa", "boundary": "default must be recorded, e.g. Lahiri"},
+            {"field": "node_mode", "format": "mean|true", "boundary": "Rahu/Ketu mode must be frozen"},
+        ],
+        "optional_fields": ["question_text", "querent_id", "house_focus", "language"],
+        "claim_boundary": "Input contract only; does not validate Prashna predictions or external numeric parity.",
+    }
+    rows = [
+        {
+            "source_id": "vedastro_prasna_marga_ch5_sphuta_example",
+            "domain": "horary_annual_sensitive_points",
+            "technique_family": "sphuta_trisphuta_family",
+            "url": "https://vedastro.org/book/PrasnaMarga/Chapter5",
+            "source_role": "public_numeric_candidate",
+            "numeric_fields_present": True,
+            "expected_values": {
+                "sun": "4s 3° 8' 25\"",
+                "moon": "3s 19° 36' 34\"",
+                "lagna": "3s 27° 22'",
+                "gulika": "3s 14° 10'",
+                "rahu": "3s 8° 16'",
+                "trisphuta": "11s 1° 8' 34\"",
+                "chatusphuta": "2s 15° 18' 34\"",
+                "panchasphuta": "5s 23° 34' 34\"",
+            },
+            "missing_for_oracle": ["complete_prashna_input", "ayanamsa", "node_mode", "timezone", "raw_capture_hash", "local_replay", "pyjhora_or_other_legal_replay"],
+            "upgrade_status": "candidate_not_oracle",
+            "candidate_hash": "",
+            "claim_boundary": "Numeric Sphuta example exists, but full Prashna input/settings are incomplete; use as candidate only.",
+        }
+    ]
+    for row in rows:
+        row["candidate_hash"] = h(row)
+    queue = {
+        "scope": "prashna_numeric_oracle_packet_queue",
+        "created_at": date,
+        "status": "queue_ready",
+        "claim_status": "open_queue",
+        "production_tuning_allowed": False,
+        "truth_matrix_allowed": False,
+        "summary": {"candidate_count": len(rows), "numeric_candidate_count": sum(r["numeric_fields_present"] for r in rows), "oracle_ready_count": 0},
+        "rows": rows,
+        "boundary": "Queue only; no Prashna/Saham/Gulika/Sphuta claim is upgraded until complete input, raw/hash and local/external replay close.",
+    }
+    return {"contract": contract, "queue": queue}
+
+def main():
+    ap=argparse.ArgumentParser(); ap.add_argument("--date", default="2026-07-20"); args=ap.parse_args()
+    data=build(args.date)
+    CONTRACT.write_text(json.dumps(data["contract"], ensure_ascii=False, indent=2, sort_keys=True)+"\n")
+    QUEUE.write_text(json.dumps(data["queue"], ensure_ascii=False, indent=2, sort_keys=True)+"\n")
+    print(json.dumps(data, ensure_ascii=False, indent=2, sort_keys=True))
+if __name__ == "__main__": main()
@@ -0,0 +1,100 @@
+#!/usr/bin/env python3
+"""Build a public-real-case website E2E evaluation contract."""
+
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+from typing import Any
+
+
+CASES = [
+    ("steve_jobs", "Steve Jobs", "1955-02-24", "19:15", "San Francisco, CA, USA", ["career", "wealth", "timing"]),
+    ("albert_einstein", "Albert Einstein", "1879-03-14", "11:30", "Ulm, Germany", ["education", "career", "timing"]),
+    ("barack_obama", "Barack Obama", "1961-08-04", "19:24", "Honolulu, HI, USA", ["career", "migration", "annual"]),
+    ("princess_diana", "Princess Diana", "1961-07-01", "19:45", "Sandringham, England", ["marriage", "family", "timing"]),
+    ("donald_trump", "Donald Trump", "1946-06-14", "10:54", "Queens, NY, USA", ["career", "wealth", "annual"]),
+    ("oprah_winfrey", "Oprah Winfrey", "1954-01-29", "04:30", "Kosciusko, MS, USA", ["career", "wealth", "family"]),
+    ("elon_musk", "Elon Musk", "1971-06-28", "07:30", "Pretoria, South Africa", ["career", "migration", "wealth"]),
+    ("mahatma_gandhi", "Mahatma Gandhi", "1869-10-02", "07:11", "Porbandar, India", ["career", "migration", "timing"]),
+    ("marilyn_monroe", "Marilyn Monroe", "1926-06-01", "09:30", "Los Angeles, CA, USA", ["marriage", "career", "health"]),
+    ("bill_gates", "Bill Gates", "1955-10-28", "22:00", "Seattle, WA, USA", ["career", "wealth", "education"]),
+    ("j_k_rowling", "J. K. Rowling", "1965-07-31", "14:00", "Yate, England", ["career", "wealth", "timing"]),
+    ("nelson_mandela", "Nelson Mandela", "1918-07-18", "14:54", "Mvezo, South Africa", ["career", "timing", "migration"]),
+    ("mother_teresa", "Mother Teresa", "1910-08-26", "14:25", "Skopje, North Macedonia", ["career", "migration", "health"]),
+    ("michael_jackson", "Michael Jackson", "1958-08-29", "19:33", "Gary, IN, USA", ["career", "wealth", "health"]),
+    ("queen_elizabeth_ii", "Queen Elizabeth II", "1926-04-21", "02:40", "London, England", ["career", "family", "annual"]),
+    ("john_f_kennedy", "John F. Kennedy", "1917-05-29", "15:00", "Brookline, MA, USA", ["career", "family", "health"]),
+    ("martin_luther_king_jr", "Martin Luther King Jr.", "1929-01-15", "12:00", "Atlanta, GA, USA", ["career", "timing", "health"]),
+    ("angelina_jolie", "Angelina Jolie", "1975-06-04", "09:09", "Los Angeles, CA, USA", ["marriage", "family", "career"]),
+    ("brad_pitt", "Brad Pitt", "1963-12-18", "06:31", "Shawnee, OK, USA", ["marriage", "career", "wealth"]),
+    ("serena_williams", "Serena Williams", "1981-09-26", "20:28", "Saginaw, MI, USA", ["career", "health", "annual"]),
+]
+
+
+QUESTION_MATRIX = {
+    "career": ["事业主轴是什么？", "哪类阶段更容易爆发？"],
+    "wealth": ["财富来源和风险是什么？"],
+    "marriage": ["婚恋关系中应看哪些印度占星指标？"],
+    "health": ["健康主题只能如何非医疗表达？"],
+    "migration": ["迁移/海外发展应看哪些宫位和 Dasha？"],
+    "family": ["家庭/子女主题应调用哪些分盘和宫位？"],
+    "education": ["学习与教育路径怎么看？"],
+    "timing": ["历史关键阶段能否用 Dasha + Narayana 回看？"],
+    "annual": ["年度运势应如何避免过度承诺？"],
+}
+
+
+def build(date: str) -> dict[str, Any]:
+    cases = []
+    for case_id, subject, date_s, time_s, place, domains in CASES:
+        cases.append(
+            {
+                "case_id": case_id,
+                "subject": subject,
+                "birth": {"date": date_s, "time": time_s, "place": place, "source_policy": "public_record_candidate"},
+                "domains": domains,
+                "prompts": [q for d in domains for q in QUESTION_MATRIX[d]],
+                "expected_runtime_context": [
+                    "birth_input_contract",
+                    "ayanamsa_node_mode",
+                    "D1",
+                    "D9_or_relevant_varga",
+                    "Dasha",
+                    "Narayana_Dasha_for_timing",
+                    "functional_benefic_malefic",
+                    "claim_boundary",
+                    "similar_case_reference_allowed",
+                ],
+            }
+        )
+    return {
+        "scope": "real_case_website_e2e_eval",
+        "created_at": date,
+        "claim_status": "ready_contract",
+        "production_tuning_allowed": False,
+        "truth_matrix_allowed": False,
+        "case_count": len(cases),
+        "cases": cases,
+        "acceptance_rules": [
+            "Website must save/render input contract and runtime context JSON for each prompt.",
+            "Answers may use public cases as explanation references, not prediction proof.",
+            "Precise day/month claims must stay exploratory_unvalidated unless holdout closes.",
+            "Marriage/career/wealth/health/migration/family/education/timing/annual domains must route to matching technique context.",
+            "Any Shadbala/AV/KP conflict must be described as method/source difference, not majority-vote truth.",
+        ],
+        "boundary": "Product E2E quality harness only; not an accuracy benchmark or independent holdout.",
+    }
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--date", default="2026-07-20")
+    args = parser.parse_args()
+    print(json.dumps(build(args.date), ensure_ascii=False, indent=2, sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())