Cyrene_v2/plugins/tools/io/search/providers_test.go

package search

import (
	"context"
	"encoding/json"
	"errors"
	"io"
	"net/http"
	"net/http/httptest"
	"testing"
)

// testKey 是夹具里用的假密钥。它只出现在**请求头**里：任何日志/错误/响应里出现它
// 都是缺陷（密钥不泄漏，design/27 §1.5）。
const testKey = "test-api-key-do-not-log"

// mustUnmarshal 解一份 JSON 到 out，失败即 Fatal。
func mustUnmarshal(t *testing.T, raw string, out any) {
	t.Helper()
	if err := json.Unmarshal([]byte(raw), out); err != nil {
		t.Fatalf("unmarshal %q: %v", raw, err)
	}
}

// fixtureServer 造一个「记录请求 + 回固定 JSON」的测试服务器。
func fixtureServer(t *testing.T, fixture string) (*httptest.Server, *captured) {
	t.Helper()
	c := &captured{}
	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		raw, _ := io.ReadAll(r.Body)
		c.set(r, raw)
		w.Header().Set("Content-Type", "application/json")
		_, _ = w.Write([]byte(fixture))
	}))
	t.Cleanup(srv.Close)
	return srv, c
}

// captured 记录最后一次收到的请求（测试同步：handler 先于 Search 返回执行完）。
type captured struct {
	req  *http.Request
	body string
}

func (c *captured) set(r *http.Request, body []byte) {
	// 只留断言需要的部分：URL/Method/Header（不含 Body 读句柄）。
	clone := r.Clone(r.Context())
	c.req = clone
	c.body = string(body)
}

// providerCase 是「一家 provider 的 httptest 夹具」：断言请求形态 + 响应归一化。
type providerCase struct {
	name string
	newP func(Config) Provider
	// method / path 是期望的请求形态。
	method string
	path   string
	// authHeader / authValue 是期望的认证头（authHeader 为空 = 该家不用 key，如 searxng）。
	authHeader string
	authValue  string
	// fixture 是上游返回的 JSON；want* 是归一化后应得到的第一条结果。
	fixture     string
	wantTitle   string
	wantLink    string
	wantContent string
	// checkRequest 做该家特有的断言（请求体字段等）。
	checkRequest func(t *testing.T, c *captured)
}

// searchProviderCaseTable 是六家的夹具表（**禁止真调外部 API** —— 全部打 httptest）。
func searchProviderCaseTable() []providerCase {
	return []providerCase{
		searxngCase(), zhipuCase(), aliyunCase(), bochaCase(), volcengineCase(), baiduCase(),
	}
}

// searxngCase 是 searxng 的夹具（GET + q/format=json，无认证头）。
func searxngCase() providerCase {
	return providerCase{
		name: "searxng", newP: newSearXNG,
		// searxng 的基地址就是端点本身（无固定路径）⇒ 请求落在 "/" 上，
		// 查询串才是它的语义（见 checkRequest）。带路径的端点由 io 包既有测试钉住。
		method: http.MethodGet, path: "/",
		fixture:     `{"results":[{"title":"SearX 结果","url":"https://s.example/1","content":"正文","score":0.5}]}`,
		wantTitle:   "SearX 结果",
		wantLink:    "https://s.example/1",
		wantContent: "正文",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			if got := c.req.URL.Query().Get("q"); got != "昔涟" {
				t.Errorf("searxng q = %q, want 昔涟", got)
			}
			if got := c.req.URL.Query().Get("format"); got != "json" {
				t.Errorf("searxng format = %q, want json", got)
			}
			if c.req.Header.Get("Authorization") != "" {
				t.Errorf("searxng 不该带认证头: %q", c.req.Header.Get("Authorization"))
			}
		},
	}
}

// zhipuCase 是智谱的夹具（POST + Bearer；请求字段 search_engine/search_query/search_intent/content_size）。
func zhipuCase() providerCase {
	return providerCase{
		name: "zhipu", newP: newZhipu,
		method: http.MethodPost, path: zhipuPath, authHeader: "Authorization", authValue: "Bearer " + testKey,
		fixture:     `{"search_result":[{"title":"智谱结果","link":"https://z.example/1","content":"智谱正文"}]}`,
		wantTitle:   "智谱结果",
		wantLink:    "https://z.example/1",
		wantContent: "智谱正文",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			// 只发官方文档确证过的字段：search_engine / search_query / search_intent / content_size，
			// 外加**条件透传**的 count（仅当 Count 落在官方区间 1–50；本夹具 Query 未给 Count ⇒ 不发）。
			var body map[string]any
			mustUnmarshal(t, c.body, &body)
			if body["search_query"] != "昔涟" {
				t.Errorf("zhipu body search_query = %v", body["search_query"])
			}
			if body["search_engine"] != zhipuEngineStd {
				t.Errorf("zhipu body search_engine = %v, want %s", body["search_engine"], zhipuEngineStd)
			}
			if body["search_intent"] != false {
				t.Errorf("zhipu body search_intent = %v, want false（官方标 required）", body["search_intent"])
			}
			if body["content_size"] != zhipuContentSize {
				t.Errorf("zhipu body content_size = %v, want %s", body["content_size"], zhipuContentSize)
			}
		},
	}
}

// aliyunCase 是阿里云 IQS 的夹具（GET + X-API-Key + query 查询串）。
func aliyunCase() providerCase {
	return providerCase{
		name: "aliyun_iqs", newP: newAliyunIQS,
		method: http.MethodGet, path: aliyunIQSPath, authHeader: "X-API-Key", authValue: testKey,
		fixture: `{"pageItems":[{"title":"IQS 结果","link":"https://a.example/1","snippet":"摘要",` +
			`"mainText":"IQS 正文","publishTime":1700000000000,"score":0.9}]}`,
		wantTitle:   "IQS 结果",
		wantLink:    "https://a.example/1",
		wantContent: "IQS 正文",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			if got := c.req.URL.Query().Get("query"); got != "昔涟" {
				t.Errorf("aliyun query = %q, want 昔涟", got)
			}
		},
	}
}

// bochaCase 是博查的夹具（POST + Bearer；summary=true / count 夹在官方区间）。
func bochaCase() providerCase {
	return providerCase{
		name: "bocha", newP: newBocha,
		method: http.MethodPost, path: bochaPath, authHeader: "Authorization", authValue: "Bearer " + testKey,
		fixture: `{"webPages":{"value":[{"name":"博查结果","url":"https://b.example/1",` +
			`"snippet":"摘要","summary":"博查总结","datePublished":"2025-01-01"}]}}`,
		wantTitle:   "博查结果",
		wantLink:    "https://b.example/1",
		wantContent: "博查总结",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			var body map[string]any
			mustUnmarshal(t, c.body, &body)
			if body["query"] != "昔涟" {
				t.Errorf("bocha body query = %v", body["query"])
			}
			if body["summary"] != true {
				t.Errorf("bocha body summary = %v, want true（中文摘要）", body["summary"])
			}
			if body["count"] != float64(bochaDefaultCount) {
				t.Errorf("bocha body count = %v, want %d", body["count"], bochaDefaultCount)
			}
		},
	}
}

// volcengineCase 是火山·豆包搜索的夹具（POST + Bearer；请求字段是大写驼峰）。
func volcengineCase() providerCase {
	return providerCase{
		name: "volcengine", newP: newVolcengine,
		method: http.MethodPost, path: volcenginePath, authHeader: "Authorization", authValue: "Bearer " + testKey,
		fixture: `{"Result":{"WebResults":[{"Title":"火山结果","Url":"https://v.example/1",` +
			`"Snippet":"摘要","Content":"火山正文","PublishTime":1700000000000,"RankScore":0.8}]}}`,
		wantTitle:   "火山结果",
		wantLink:    "https://v.example/1",
		wantContent: "火山正文",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			var body map[string]any
			mustUnmarshal(t, c.body, &body)
			if body["Query"] != "昔涟" {
				t.Errorf("volcengine body Query = %v", body["Query"])
			}
			if body["SearchType"] != volcengineSearchTypeWeb {
				t.Errorf("volcengine body SearchType = %v", body["SearchType"])
			}
		},
	}
}

// baiduCase 是百度千帆的夹具（POST + Bearer；单轮 messages + search_source）。
func baiduCase() providerCase {
	return providerCase{
		name: "baidu_qianfan", newP: newBaiduQianfan,
		method: http.MethodPost, path: baiduQianfanPath, authHeader: "Authorization", authValue: "Bearer " + testKey,
		fixture: `{"references":[{"title":"百度结果","url":"https://q.example/1",` +
			`"snippet":"摘要","content":"百度正文"}]}`,
		wantTitle:   "百度结果",
		wantLink:    "https://q.example/1",
		wantContent: "百度正文",
		checkRequest: func(t *testing.T, c *captured) {
			t.Helper()
			var body map[string]any
			mustUnmarshal(t, c.body, &body)
			if body["search_source"] != baiduSearchSourceV2 {
				t.Errorf("baidu body search_source = %v, want %s", body["search_source"], baiduSearchSourceV2)
			}
			msgs, ok := body["messages"].([]any)
			if !ok || len(msgs) != 1 {
				t.Fatalf("百度搜索仅支持单轮输入，messages = %v", body["messages"])
			}
			first, _ := msgs[0].(map[string]any)
			if first["content"] != "昔涟" || first["role"] != "user" {
				t.Errorf("baidu messages[0] = %v", first)
			}
		},
	}
}

// TestProviders_FixturePerProvider 是验收 ①：六家各一条 httptest 夹具（不打真上游）。
func TestProviders_FixturePerProvider(t *testing.T) {
	t.Parallel()
	for _, tc := range searchProviderCaseTable() {
		tc := tc
		t.Run(tc.name, func(t *testing.T) {
			t.Parallel()
			srv, cap := fixtureServer(t, tc.fixture)
			p := tc.newP(Config{APIKey: testKey, BaseURL: srv.URL, Client: srv.Client()})
			got, err := p.Search(context.Background(), Query{Text: "昔涟"})
			if err != nil {
				t.Fatalf("%s Search: %v", tc.name, err)
			}
			if cap.req == nil {
				t.Fatalf("%s 没发出任何请求", tc.name)
			}
			if cap.req.Method != tc.method {
				t.Errorf("method = %q, want %q", cap.req.Method, tc.method)
			}
			if cap.req.URL.Path != tc.path {
				t.Errorf("path = %q, want %q", cap.req.URL.Path, tc.path)
			}
			if tc.authHeader != "" {
				if gotAuth := cap.req.Header.Get(tc.authHeader); gotAuth != tc.authValue {
					t.Errorf("认证头 %s = %q, want %q", tc.authHeader, gotAuth, tc.authValue)
				}
			}
			if len(got) != 1 {
				t.Fatalf("%s 结果条数 = %d, want 1（归一化失败）: %+v", tc.name, len(got), got)
			}
			r0 := got[0]
			if r0.Title != tc.wantTitle || r0.Link != tc.wantLink || r0.Content != tc.wantContent {
				t.Errorf("%s 归一化结果 = %+v, want title=%q link=%q content=%q",
					tc.name, r0, tc.wantTitle, tc.wantLink, tc.wantContent)
			}
			if tc.checkRequest != nil {
				tc.checkRequest(t, cap)
			}
		})
	}
}

// TestProviders_AllNormalizeToSameResult 是验收 ②：六家输出**归一化成同一 Result**
// （同一结构体、同一语义），且共用同一套后处理（Finalize）。
func TestProviders_AllNormalizeToSameResult(t *testing.T) {
	t.Parallel()
	for _, tc := range searchProviderCaseTable() {
		tc := tc
		t.Run(tc.name, func(t *testing.T) {
			t.Parallel()
			srv, _ := fixtureServer(t, tc.fixture)
			p := tc.newP(Config{APIKey: testKey, BaseURL: srv.URL, Client: srv.Client()})
			got, err := p.Search(context.Background(), Query{Text: "q"})
			if err != nil {
				t.Fatalf("Search: %v", err)
			}
			for i, r := range got {
				if r.Link == "" {
					t.Errorf("[%d] Link 为空 —— 没链接的结果不算结果: %+v", i, r)
				}
				if r.Title == "" {
					t.Errorf("[%d] Title 为空（夹具里给了标题）: %+v", i, r)
				}
			}
			if final := Finalize(got, Options{Limit: 1}); len(final) != 1 {
				t.Fatalf("Finalize 后条数 = %d, want 1", len(final))
			}
		})
	}
}

// TestProviders_EmptyBodyIsEmptyNotError 钉住六家对「200 + 空数组」的处置一致：
// 不是错误，而是 0 条（由上层按「空结果 ≠ 成功」处置）。
func TestProviders_EmptyBodyIsEmptyNotError(t *testing.T) {
	t.Parallel()
	for _, tc := range searchProviderCaseTable() {
		tc := tc
		t.Run(tc.name, func(t *testing.T) {
			t.Parallel()
			srv, _ := fixtureServer(t, tc.emptyFixture())
			p := tc.newP(Config{APIKey: testKey, BaseURL: srv.URL, Client: srv.Client()})
			got, err := p.Search(context.Background(), Query{Text: "q"})
			if err != nil {
				t.Fatalf("空结果不是错误: %v", err)
			}
			if len(got) != 0 {
				t.Fatalf("条数 = %d, want 0", len(got))
			}
		})
	}
}

// emptyFixture 返回该家「0 条结果」的合法应答（候选数组为空）。
func (tc providerCase) emptyFixture() string {
	switch tc.name {
	case "searxng":
		return `{"results":[]}`
	case "zhipu":
		return `{"search_result":[]}`
	case "aliyun_iqs":
		return `{"pageItems":[]}`
	case "bocha":
		return `{"webPages":{"value":[]}}`
	case "volcengine":
		return `{"Result":{"WebResults":[]}}`
	default:
		return `{"references":[]}`
	}
}

// TestExtractResults_MissingFieldsTolerated 钉住「缺字段的容错」：
// 缺 content 回落 snippet；全缺也不 panic、不报错；非对象条目跳过。
func TestExtractResults_MissingFieldsTolerated(t *testing.T) {
	t.Parallel()
	data := []byte(`{"results":[
		{"title":"只有标题","url":"https://x.example/1"},
		{"url":"https://x.example/2","snippet":"只有摘要"},
		"not-an-object",
		{"title":"只有正文","url":"https://x.example/3","content":"正文"}
	]}`)
	got, err := extractResults(data, searxngSpec())
	if err != nil {
		t.Fatalf("缺字段不该报错: %v", err)
	}
	if len(got) != 3 {
		t.Fatalf("条数 = %d, want 3（非对象条目应被跳过）", len(got))
	}
	if got[0].Content != "" || got[0].Title != "只有标题" {
		t.Errorf("缺 content/正文时应原样留空: %+v", got[0])
	}
	if got[1].Content != "只有摘要" {
		t.Errorf("content 缺了必须回落 snippet: %+v", got[1])
	}
	if got[2].Content != "正文" {
		t.Errorf("正文应原样带出: %+v", got[2])
	}
}

// TestExtractResults_EmptyVsMalformed 钉住「空」与「解析不了」必须分开（design/27 §1.4）。
func TestExtractResults_EmptyVsMalformed(t *testing.T) {
	t.Parallel()
	empty, err := extractResults([]byte(`{"results":[]}`), searxngSpec())
	if err != nil {
		t.Fatalf("空数组不是错误: %v", err)
	}
	if len(empty) != 0 {
		t.Fatalf("空数组应归一化成 0 条, got %d", len(empty))
	}
	for _, bad := range []string{`not json`, `{"data":{}}`, `[]`, `{}`} {
		if _, err := extractResults([]byte(bad), searxngSpec()); !errors.Is(err, ErrBadResponse) {
			t.Errorf("输入 %q 应报 ErrBadResponse, got %v", bad, err)
		}
	}
}