# -*- coding: utf-8 -*-
"""端到端联调用例（Issue #86）—— 跨链路集成场景。

验证数据流（采集 → 落库）与问答流（RAG → LLM 网关）的协同：
基于真实采集到的工艺数据，构建 RAG 知识并完成一次安全问答闭环，
模拟「采集 → 沉淀知识 → 智能问答」的完整业务闭环。
"""
from __future__ import annotations

# 引导加载四个内核模块（必须在测试导入前执行）
import tests.e2e._bootstrap  # noqa: F401

import tempfile
import unittest

from collector.engine import CollectorEngine
from collector.metrics import HealthMetrics
from collector.spool import SpoolStore
from drivers import SimulatorDriver
from point_dict.loader import Point, PointDict

from data_bus import BatchWriter, MemorySink

from rag_kb import RagKnowledgeBase, build_document
from llm_gateway import (
    GatewayResult,
    LLMGateway,
    LocalBackend,
    PromptRegistry,
    RouteTarget,
)


def _collect_samples(tmpdir: str):
    """跑一轮采集，返回（落库样本, 引擎健康度）。"""
    points = [
        Point(device_id="CLF-01", point_id="CLF-01.TEMP", name="1#炉温",
              unit="℃", data_type="float", sample_rate=1000,
              quality_code=True, row_number=2),
        Point(device_id="CLF-01", point_id="CLF-01.PRES", name="1#炉压",
              unit="kPa", data_type="float", sample_rate=1000,
              quality_code=True, row_number=3),
    ]
    pd = PointDict(points)
    spool = SpoolStore(spool_dir=tmpdir, cache_limit_bytes=8 * 1024 * 1024)
    metrics = HealthMetrics()
    engine = CollectorEngine(
        point_dict=pd,
        driver_slots=[("simulator", SimulatorDriver(), [])],
        spool=spool, metrics=metrics, interval_ms=1000, max_pending=100_000,
    )
    engine.collect_once()

    sink = MemorySink()
    writer = BatchWriter(sink=sink, batch_size=10, flush_interval=0.0)
    for s in spool.pending_records():
        writer.push(s)
        spool.ack(s)
    writer.flush()
    return sink.rows, metrics


class CrossPipelineE2ETest(unittest.TestCase):
    """采集 → 落库 → 知识沉淀 → 安全问答 全业务闭环。"""

    def test_collect_then_query_closed_loop(self) -> None:
        """采集数据落库后，结合工艺知识库完成一次安全问答。"""
        tmpdir = tempfile.mkdtemp(prefix="iaop_cross_")
        # 1) 数据流：采集并落库
        rows, metrics = _collect_samples(tmpdir)
        self.assertEqual(len(rows), 2)                 # 2 点位全部落库
        self.assertTrue(metrics.meets_sla())           # 采集健康度达标
        point_ids = {r["point_id"] for r in rows}
        self.assertEqual(point_ids, {"CLF-01.TEMP", "CLF-01.PRES"})

        # 2) 知识沉淀：工艺知识库（模拟由采集数据衍生的工艺规范）
        kb = RagKnowledgeBase()
        kb.add_documents([
            build_document(
                title="1#氯化炉工艺卡",
                text="1#氯化炉（CLF-01）正常炉温 850~920℃，炉压 0.2~0.4MPa。"
                     "CLF-01.TEMP 与 CLF-01.PRES 为关键监控点位。",
                category="process",
            ),
        ])

        # 3) 问答流：基于知识库回答工艺问题
        prompts = PromptRegistry()
        prompts.update(name="qa", version="1.0.0",
                       text="工业 AI 助手回答：{query}")
        gateway = LLMGateway(prompts=prompts, prompt_name="qa",
                             local=LocalBackend(echo_context=True))
        query = "CLF-01 的关键监控点位和正常炉温范围？"
        hits = kb.search(query, top_k=3)
        sources = [h.chunk.title for h in hits]
        self.assertGreater(len(sources), 0)

        result = gateway.ask(query, rag_context=sources, confidence=0.95)
        self.assertIsInstance(result, GatewayResult)
        self.assertEqual(result.route.target, RouteTarget.LOCAL)
        self.assertFalse(result.needs_human)
        # 回答引用了知识库来源
        self.assertIn(sources[0], result.answer)


if __name__ == "__main__":
    unittest.main()