content-find-agent-new/scripts/validate_query_prompts_config.py

"""Validate query_prompts config structure (V2-M1D, paired with V2-M2).

M2 will introduce product_documents/配置/query_prompts.v1.json. Until then this
validator is a no-op (status=skip). Once the file exists, it checks each profile
carries the required fields with sane sampling params.
"""

from __future__ import annotations

import json
import sys
from pathlib import Path
from typing import Any

ROOT = Path(__file__).resolve().parents[1]
QUERY_PROMPTS_PATH = Path("product_documents/配置/query_prompts.v1.json")
_REQUIRED = ["system", "user", "temperature", "max_tokens", "evidence_fields", "variants_per_seed"]


def validate_query_prompts_config(config: dict[str, Any]) -> list[dict[str, Any]]:
    findings: list[dict[str, Any]] = []
    profiles = config.get("profiles") or {}
    if not profiles:
        findings.append({"level": "fail", "check_id": "profiles", "message": "no profiles defined"})
    for name, profile in profiles.items():
        for field in _REQUIRED:
            if field not in profile:
                findings.append({"level": "fail", "check_id": "missing_field", "message": f"{name} missing {field}"})
        temp = profile.get("temperature")
        if isinstance(temp, (int, float)) and not 0 <= temp <= 2:
            findings.append({"level": "fail", "check_id": "temperature", "message": f"{name} temperature out of [0,2]: {temp}"})
        if isinstance(profile.get("max_tokens"), int) and profile["max_tokens"] <= 0:
            findings.append({"level": "fail", "check_id": "max_tokens", "message": f"{name} max_tokens must be > 0"})
        if isinstance(profile.get("variants_per_seed"), int) and profile["variants_per_seed"] < 1:
            findings.append({"level": "fail", "check_id": "variants_per_seed", "message": f"{name} variants_per_seed must be >= 1"})
    return findings


def main() -> int:
    path = ROOT / QUERY_PROMPTS_PATH
    if not path.exists():
        print(json.dumps({"status": "skip", "config_path": str(QUERY_PROMPTS_PATH), "reason": "not built until V2-M2"}, ensure_ascii=False, indent=2))
        return 0
    findings = validate_query_prompts_config(json.loads(path.read_text(encoding="utf-8")))
    status = "fail" if any(f["level"] == "fail" for f in findings) else "pass"
    print(json.dumps({"status": status, "config_path": str(QUERY_PROMPTS_PATH), "findings": findings}, ensure_ascii=False, indent=2))
    return 1 if status == "fail" else 0


if __name__ == "__main__":
    sys.exit(main())