"""Agent 基础设施: spec 定义 + 执行器。

执行语义:
  1. 数据切片为空 / 明确 no_data → 跳过 LLM, 直接返回 stub(省 token 防幻觉)
  2. LLM 调用失败 → fail-open, 报告标记 status=error, 不阻断整体
  3. 解析失败(LLM 没输出合法 JSON) → status=parse_error
"""
from __future__ import annotations

import functools
import logging
from dataclasses import dataclass, field
from pathlib import Path

from src.llm.context_builder import MatchHeader, SliceResult
from src.llm.provider import LLMProvider, LLMResponse

logger = logging.getLogger(__name__)

_PROMPT_DIR = Path(__file__).resolve().parent.parent / "prompts" / "agents"

NO_DATA_SENTINELS = ("无数据", "no data", "no_data")


@functools.lru_cache(maxsize=16)
def load_agent_prompt(name: str, version: str = "v1") -> str:
    """缓存加载 agent prompt 模板。"""
    path = _PROMPT_DIR / f"{name}_{version}.md"
    if not path.exists():
        raise FileNotFoundError(f"agent prompt 不存在: {path}")
    with open(path, encoding="utf-8") as f:
        return f.read()


@dataclass
class AgentSpec:
    """领域专家 agent 定义。"""
    name: str                                    # h2h / form / home_away / injuries / stats
    system_prompt: str                            # system message
    slice_fn: object                              # async (header, before) -> str 切片函数


@dataclass
class AgentReport:
    """专家 agent 统一输出契约。"""
    agent: str
    status: str = "ok"                            # ok | no_data | error | parse_error
    data_sufficiency: str = "medium"              # high | medium | low | none
    analysis: str = ""
    home_edge: float | None = None                # -1.0 ~ 1.0, 正=利主队
    subjective_confidence: float | None = None               # 0.0 ~ 1.0
    key_evidence: list[str] = field(default_factory=list)
    # xg agent 专属
    exp_home_goals: float | None = None
    exp_away_goals: float | None = None
    probable_score: str | None = None
    # 元信息
    model: str = ""
    latency_ms: int | None = None
    prompt_tokens: int | None = None
    completion_tokens: int | None = None

    def to_dict(self) -> dict:
        return {
            "agent": self.agent,
            "status": self.status,
            "data_sufficiency": self.data_sufficiency,
            "analysis": self.analysis,
            "home_edge": self.home_edge,
            "subjective_confidence": self.subjective_confidence,
            "key_evidence": self.key_evidence,
            "exp_home_goals": self.exp_home_goals,
            "exp_away_goals": self.exp_away_goals,
            "probable_score": self.probable_score,
            "model": self.model,
            "latency_ms": self.latency_ms,
            "prompt_tokens": self.prompt_tokens,
            "completion_tokens": self.completion_tokens,
        }


def _is_no_data(slice_text: str) -> bool:
    """兜底: 判断纯字符串切片是否全无数据。

    仅用于 slice_fn 返回 `str`(未升级为 SliceResult)的场景。
    新代码应让切片返回 SliceResult 并显式声明 has_data —— 字符串子串匹配
    依赖具体文案(「无比分数据」「无伤停数据」等变体会漏判),不可靠。
    """
    body = [ln.strip() for ln in slice_text.splitlines() if ln.strip()]
    # 去掉标题行(── 开头)
    content = [ln for ln in body if not ln.startswith("──")]
    if not content:
        return True
    return all(any(s in ln for s in NO_DATA_SENTINELS) for ln in content)


def _slice_has_data(slice_result) -> tuple[str, bool]:
    """把切片返回值统一成 (text, has_data)。

    优先使用 SliceResult.has_data(结构化,可信);若切片函数仍返回 str,
    则回退到文案子串匹配(向后兼容)。
    """
    if isinstance(slice_result, SliceResult):
        return slice_result.text, slice_result.has_data
    text = str(slice_result)
    return text, not _is_no_data(text)


def _stub_no_data(agent: str) -> AgentReport:
    return AgentReport(
        agent=agent,
        status="no_data",
        data_sufficiency="none",
        analysis="该维度无数据,跳过分析。",
    )


def _parse_report(agent: str, parsed: dict, resp: LLMResponse, model: str) -> AgentReport:
    """把 LLM JSON 输出解析为 AgentReport,经过严格校验。"""
    from src.llm.validation import validate_agent_output

    try:
        validated = validate_agent_output(parsed)
    except Exception as e:
        # 校验失败 → 返回 parse_error 而非静默降级
        return AgentReport(
            agent=agent,
            status="parse_error",
            analysis=f"输出校验失败: {e}",
            model=model,
            latency_ms=resp.latency_ms,
            prompt_tokens=resp.prompt_tokens,
            completion_tokens=resp.completion_tokens,
        )

    return AgentReport(
        agent=agent,
        status="ok",
        data_sufficiency=validated.data_sufficiency,
        analysis=validated.analysis,
        home_edge=validated.home_edge,
        subjective_confidence=validated.subjective_confidence,
        key_evidence=validated.key_evidence,
        exp_home_goals=validated.exp_home_goals,
        exp_away_goals=validated.exp_away_goals,
        probable_score=validated.probable_score,
        model=model,
        latency_ms=resp.latency_ms,
        prompt_tokens=resp.prompt_tokens,
        completion_tokens=resp.completion_tokens,
    )


async def run_agent(
    spec: AgentSpec,
    header: MatchHeader,
    provider: LLMProvider,
    *,
    before=None,
    version: str = "v1",
) -> AgentReport:
    """执行单个专家 agent: 切片 → no_data 门控 → 调 LLM → 解析报告。"""
    # 1. 数据切片
    try:
        slice_result = await spec.slice_fn(header, before=before)
    except Exception as e:
        logger.exception("agent %s slice failed", spec.name)
        return AgentReport(agent=spec.name, status="error", analysis=f"数据切片失败: {e}")

    # 2. no_data 门控: 切片显式声明无数据 → 不调 LLM
    slice_text, has_data = _slice_has_data(slice_result)
    if not has_data:
        logger.debug("agent %s: slice is no_data, skipping LLM", spec.name)
        return _stub_no_data(spec.name)

    # 3. 拼 prompt(模板中 {{context}} 为切片占位)
    template = load_agent_prompt(spec.name, version)
    user_prompt = template.replace("{{context}}", slice_text)

    # 4. 调 LLM
    resp = await provider.chat(
        system=spec.system_prompt,
        user=user_prompt,
        json_mode=True,
        temperature=0.2,
        max_tokens=4096,  # 推理模型需要更大余量
    )
    if resp.error:
        logger.warning("agent %s LLM failed: %s", spec.name, resp.error)
        return AgentReport(agent=spec.name, status="error", analysis=f"LLM 调用失败: {resp.error}")

    # 5. 解析
    if not resp.parsed:
        return AgentReport(
            agent=spec.name,
            status="parse_error",
            analysis=f"LLM 输出无法解析为 JSON: {resp.content[:200]}",
        )
    return _parse_report(spec.name, resp.parsed, resp, provider.model)