# -*- coding: utf-8 -*-
"""tests/docgen/test_websearch.py —— 联网模版参考验收用例（DC-UNIT-WEB-*）。

2026-09-04 手工测试需求2：库内无匹配模版时联网搜索相似内容结构。
口径：monkeypatch websearch._make_client 注入 MockTransport，全程离线。
执行：bash tests/docgen/run_all.sh unit
"""
from __future__ import annotations

import asyncio
import sys
from pathlib import Path

import httpx
import pytest

REPO = Path(__file__).resolve().parents[2]
sys.path.insert(0, str(REPO))
sys.path.insert(0, str(REPO / "sdk"))

from app.services.docgen import websearch  # noqa: E402

pytestmark = pytest.mark.unit

BING_PAGE = (
    '<html><body><ol id="b_results">'
    '<li class="b_algo"><h2><a href="https://example.com/fanwen/1">范文一</a></h2></li>'
    '<li class="b_algo"><h2><a href="https://example.com/fanwen/2">范文二</a></h2></li>'
    "</ol></body></html>"
)

RESULT_PAGE = (
    "<html><head><title>安全生产通知范文</title></head><body>"
    "<h1>关于安全生产的通知</h1><p>正文……</p>"
    "<h2>一、总体要求</h2><h2>二、重点任务</h2><h3>2.1 隐患排查</h3>"
    "</body></html>"
)


def _install(monkeypatch, handler, *, force_enabled=True):
    """注入 MockTransport 客户端工厂；force_enabled 绕过环境闸。"""
    calls = []

    async def wrapped(request):
        calls.append(str(request.url))
        return await handler(request)

    def factory(timeout):
        return httpx.AsyncClient(transport=httpx.MockTransport(wrapped))

    monkeypatch.setattr(websearch, "_make_client", factory)
    if force_enabled:
        monkeypatch.setattr(websearch, "web_search_enabled", lambda: True)
    return calls


def _run(description):
    return asyncio.run(websearch.search_template_outline(description))


async def _handler_page(request):
    if "bing.com" in request.url.host:
        return httpx.Response(200, text=BING_PAGE)
    return httpx.Response(200, text=RESULT_PAGE)


class TestWebsearch:

    def test_DC_UNIT_WEB_01_outline_extracted(self, monkeypatch):
        calls = _install(monkeypatch, _handler_page)
        result = _run("帮我写一份关于安全生产的通知")
        assert result is not None
        assert result.source_url == "https://example.com/fanwen/1"
        assert result.title == "安全生产通知范文"
        assert result.outline == [
            "关于安全生产的通知",
            "一、总体要求",
            "二、重点任务",
            "2.1 隐患排查",
        ]
        assert len(calls) == 2  # bing 一次 + 抓取页一次

    def test_DC_UNIT_WEB_02_network_failure_none(self, monkeypatch):
        async def handler(request):
            raise httpx.ConnectError("boom", request=request)

        _install(monkeypatch, handler)
        assert _run("通知") is None

    def test_DC_UNIT_WEB_03_oversize_page_none(self, monkeypatch):
        big = "<h1>一</h1>" + "x" * (websearch.MAX_RESULT_BYTES + 10)

        async def handler(request):
            if "bing.com" in request.url.host:
                return httpx.Response(200, text=BING_PAGE)
            return httpx.Response(200, text=big)

        _install(monkeypatch, handler)
        # 截断后标题数 < MIN_HEADINGS → None
        assert _run("通知") is None

    def test_DC_UNIT_WEB_04_too_few_headings_none(self, monkeypatch):
        async def handler(request):
            if "bing.com" in request.url.host:
                return httpx.Response(200, text=BING_PAGE)
            return httpx.Response(200, text="<html><body><p>没有标题的页面</p></body></html>")

        _install(monkeypatch, handler)
        assert _run("通知") is None

    def test_DC_UNIT_WEB_05_gb18030_decodes(self, monkeypatch):
        page = (
            "<html><head><title>通知范文</title></head><body>"
            "<h1>关于开展检查的通知</h1><h2>一、检查范围</h2></body></html>"
        ).encode("gb18030")

        async def handler(request):
            if "bing.com" in request.url.host:
                return httpx.Response(200, text=BING_PAGE)
            return httpx.Response(200, content=page, headers={"content-type": "text/html"})

        _install(monkeypatch, handler)
        result = _run("通知")
        assert result is not None
        assert result.outline == ["关于开展检查的通知", "一、检查范围"]

    def test_DC_UNIT_WEB_06_env_gate_off(self, monkeypatch):
        async def handler(request):
            return httpx.Response(200, text=BING_PAGE)

        calls = _install(monkeypatch, handler, force_enabled=False)
        monkeypatch.setattr(websearch, "web_search_enabled", lambda: False)
        assert _run("通知") is None
        assert not calls  # 关闸后不发任何网络请求