Files
hunjing/shenshi/internal/server/failover_test.go
T
chenxin 144415247e M4 收口:神识用户面与计量闭环——BYOK/配额三段式/usage 落库/Anthropic+Gemini 适配器/神识阁
- 神识 Go:store 双实现(PG 自动建表+60s 就绪重试 / 内存回退)承载 usage_logs·byok_keys·quota_balances;BYOK 密钥库(AES-GCM 加密、API 脱敏、撤回即失效、主密钥未配置 fail-closed);配额三段式(预扣 max_tokens 钳位 1M→结算实际 token→全败幂等退款,HJ-2004);计量归属(IR user_id→usage_logs,/me/usage 用户维度聚合+配额总览,/admin/usage/sample 对账抽样);Anthropic/Gemini 出站适配器(SSE 解析、system 提升、用量映射);stub 端点补名修复计量空 endpoint
- IR 契约:user_id/byok_key_id 双端同步(Go types.go + Py types.py)
- 灵台:runs.owner_id(迁移 005,server_default 兼容存量)经 kernel 全链透传为 IR user_id;shenshi_client 扩展用户面方法(X-Hunjing-User-Id 代理,Echo 桩带内存实现);/me/byok-keys 登记·列表·撤回与 /me/usage 端点(4 用例)
- 前端神识阁(/models 重建):偏好(生成参数表单)/自带密钥(登记加密保存、脱敏列表、撤回二次确认)/用量(聚合表+配额进度条)三区;me API 库+mock+5 用例;侧栏恢复入口
- 配置与部署:.env.example 增 SHENSHI_DATABASE_URL/SHENSHI_BYOK_MASTER_KEY/SHENSHI_USER_BUDGET_TOKENS;compose 神识容器接自有库(init-shenshi.sql 预建 hunjing_shenshi,与灵台不共库共表)
- 证据:Go 34 用例(BYOK 独占/配额拒绝/结算退款/归属聚合)、后端 532、前端 210、BFF 4 全绿;跨进程 E2E 四场景(灵台 HttpShenshiClient→真 Go 神识:登记脱敏/BYOK 独占不可达整体失败/用量归属聚合/撤回即失效)通过;rewrite-plan M4 标记完成、能力矩阵神识行升已验证
2026-08-27 18:39:17 +08:00

164 lines
5.5 KiB
Go
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
package server
import (
"bufio"
"encoding/json"
"net/http"
"net/http/httptest"
"strings"
"testing"
"time"
"hunjing/shenshi/internal/route"
"hunjing/shenshi/internal/transform"
)
// fakeOpenAI 构造一个 OpenAI 兼容 SSE 假上游:可控延迟/状态码/内容。
func fakeOpenAI(t *testing.T, delay time.Duration, status int, content string) *httptest.Server {
t.Helper()
return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
if status != 0 {
w.WriteHeader(status)
_, _ = w.Write([]byte(`{"error":{"message":"boom"}}`))
return
}
if delay > 0 {
time.Sleep(delay)
}
w.Header().Set("content-type", "text/event-stream")
flusher := w.(http.Flusher)
_, _ = w.Write([]byte("data: {\"choices\":[{\"delta\":{\"content\":\"" + content + "\"}}]}\n\n"))
flusher.Flush()
_, _ = w.Write([]byte("data: {\"choices\":[{\"delta\":{},\"finish_reason\":\"stop\"}],\"usage\":{\"prompt_tokens\":3,\"completion_tokens\":4}}\n\n"))
flusher.Flush()
_, _ = w.Write([]byte("data: [DONE]\n\n"))
}))
}
func registryWith(t *testing.T, firstByte time.Duration, eps ...route.Endpoint) *route.Registry {
t.Helper()
r := route.New()
r.FirstByteTimeout = firstByte
// 清掉默认 stub,仅保留测试端点
r.Reset("chat")
for _, ep := range eps {
r.Register("chat", ep)
}
return r
}
func oaEndpoint(name, baseURL string, priority int) route.Endpoint {
return route.Endpoint{
Name: name, Type: "openai_compatible", Priority: priority,
Adaptor: &transform.OpenAICompatAdaptor{EndpointName: name, BaseURL: baseURL, UpstreamModel: "test-model"},
}
}
func foPostChat(ts *httptest.Server, body string) (*http.Response, string) {
resp, err := ts.Client().Post(ts.URL+"/v1/chat/completions", "application/json", strings.NewReader(body))
if err != nil {
return nil, ""
}
var sb strings.Builder
scanner := bufio.NewScanner(resp.Body)
for scanner.Scan() {
sb.WriteString(scanner.Text())
sb.WriteString("\n")
}
return resp, sb.String()
}
func foChatBody(run string) string {
return `{"ir_version":1,"slot":"chat","messages":[{"role":"user","content":"你好"}],"run_id":"` + run + `","stream":true}`
}
// 场景:首端点首字超时 → 切道到次端点成功(不向客户端暴露失败)。
func TestFailoverOnFirstByteTimeout(t *testing.T) {
slow := fakeOpenAI(t, 500*time.Millisecond, 0, "慢端点")
fast := fakeOpenAI(t, 0, 0, "快端点")
defer slow.Close()
defer fast.Close()
srv := New(Config{}, registryWith(t, 50*time.Millisecond, oaEndpoint("slow", slow.URL, 10), oaEndpoint("fast", fast.URL, 20)))
ts := httptest.NewServer(srv)
defer ts.Close()
resp, body := foPostChat(ts, foChatBody("run_fo"))
if resp == nil || resp.StatusCode != http.StatusOK {
t.Fatalf("状态码异常:%v", resp)
}
if !strings.Contains(body, "快端点") || strings.Contains(body, "慢端点") {
t.Fatalf("应切道到快端点:%s", body)
}
if !strings.Contains(body, `"finish_reason":"stop"`) || !strings.Contains(body, "data: [DONE]") {
t.Fatalf("缺少终帧/DONE:%s", body)
}
}
// 场景:首端点 401 → 凭证级冷却;下一请求直接跳过它走次端点。
func TestCredentialCooldownSkipsEndpoint(t *testing.T) {
bad := fakeOpenAI(t, 0, http.StatusUnauthorized, "")
good := fakeOpenAI(t, 0, 0, "好端点")
defer bad.Close()
defer good.Close()
srv := New(Config{}, registryWith(t, time.Second, oaEndpoint("bad", bad.URL, 10), oaEndpoint("good", good.URL, 20)))
ts := httptest.NewServer(srv)
defer ts.Close()
// 第一次:bad 401 → 冷却 → 切 good
resp, body := foPostChat(ts, foChatBody("run_a"))
if resp == nil || resp.StatusCode != http.StatusOK || !strings.Contains(body, "好端点") {
t.Fatalf("首次应切到好端点:%v %s", resp, body)
}
// 第二次:bad 仍在冷却,直接命中 good(若未跳过会再花一次 401 往返——这里断言成功即可)
resp2, body2 := foPostChat(ts, foChatBody("run_b"))
if resp2 == nil || resp2.StatusCode != http.StatusOK || !strings.Contains(body2, "好端点") {
t.Fatalf("冷却后应直达好端点:%v", resp2)
}
}
// 场景:全部候选失败且未写字节 → HJ-2001 JSON(不是半截 SSE)。
func TestAllCandidatesFailedReturnsJsonError(t *testing.T) {
bad := fakeOpenAI(t, 0, http.StatusBadGateway, "")
defer bad.Close()
srv := New(Config{}, registryWith(t, time.Second, oaEndpoint("bad", bad.URL, 10)))
ts := httptest.NewServer(srv)
defer ts.Close()
resp, body := foPostChat(ts, foChatBody("run_x"))
if resp.StatusCode != http.StatusServiceUnavailable {
t.Fatalf("状态码 = %d", resp.StatusCode)
}
if !strings.Contains(body, "HJ-2001") || strings.Contains(body, "text/event-stream") {
t.Fatalf("应为 JSON 错误而非 SSE:%s", body)
}
}
// 场景:用量聚合端点返回 ok 计数与 token。
func TestUsageAggregate(t *testing.T) {
good := fakeOpenAI(t, 0, 0, "内容")
defer good.Close()
srv := New(Config{}, registryWith(t, time.Second, oaEndpoint("good", good.URL, 10)))
ts := httptest.NewServer(srv)
defer ts.Close()
if resp, _ := foPostChat(ts, foChatBody("run_u")); resp == nil || resp.StatusCode != 200 {
t.Fatal("请求失败")
}
usageResp, err := ts.Client().Get(ts.URL + "/me/usage")
if err != nil {
t.Fatal(err)
}
var payload struct {
Rows []map[string]any `json:"rows"`
}
defer func() { _ = usageResp.Body.Close() }()
if err := json.NewDecoder(usageResp.Body).Decode(&payload); err != nil {
t.Fatal(err)
}
if len(payload.Rows) != 1 || payload.Rows[0]["ok"].(float64) != 1 || payload.Rows[0]["prompt_tokens"].(float64) != 3 {
t.Fatalf("聚合不符:%v", payload.Rows)
}
}