- 超时:NewGateway 里 gw_timeout*time.Second 把已是 Duration 的值二次换算, int64 溢出后无论 OT_PROXY_TIMEOUT 填多少都等于没有超时,渠道 timeout_ms 也从没被读过。现在非流式用 context deadline 约束整段时长,流式只约束首包 (Transport.ResponseHeaderTimeout),正文不限时长,长回答不会被砍。 - 断连:上游请求改用 NewRequestWithContext 绑定客户端 context,客户端断开即 取消上游并记 canceled,不再空转生成继续付费;PATCH 已写 200 头后不再往 响应里追加错误体。 - 白名单:APIKey.AllowedModels 以前只存不用,现在 Auth 注入 context 并参与 判定(用户黑名单 > 密钥白名单 > 用户白名单 > 全局)。 - 选路:模型已有渠道绑定时只用绑定渠道,都不可用直接 503 no_channel,不再 把请求发给不认识它的渠道;新增运行期健康(连续 3 次网络错误/超时/429/5xx 进 30s 冷却,成功清零),与 60s 一次的 /models 探测互补。 - 密钥 API:PATCH 回显更新后的完整对象;限额填 0 统一视为不限制(原来 token 配额存 0 会让密钥第一次请求就被 429 打死,而请求配额存 0 又等于 不限制,两个字段语义相反);前端编辑框总是提交白名单,清空能真的取消。 测试:新增 proxy 链路集成测试(超时生效、流式不被截断、断连取消并记账、 密钥白名单 403、连续 429 后候选剔除)与 channel 用例(绑定模型不退全局、 冷却阈值与到期恢复、成功清零)。
158 lines
4.4 KiB
Go
158 lines
4.4 KiB
Go
package proxy
|
||
|
||
import (
|
||
"net/http"
|
||
|
||
"github.com/gin-gonic/gin"
|
||
"github.com/openteam/server/internal/proxy/convert"
|
||
"github.com/openteam/server/internal/store"
|
||
)
|
||
|
||
// chatCompletions POST /v1/chat/completions
|
||
func (g *Gateway) chatCompletions(c *gin.Context) {
|
||
u, ok := g.resolveUser(c)
|
||
if !ok {
|
||
return
|
||
}
|
||
if !g.checkBalance(c, u) {
|
||
return
|
||
}
|
||
br, body, err := parseBody(c)
|
||
if err != nil {
|
||
apiError(c, http.StatusBadRequest, "invalid_request_error", "failed to read request body")
|
||
return
|
||
}
|
||
c.Set("protocol", convert.ProtoChat)
|
||
c.Set("model_name", br.Model)
|
||
g.recordRawRequest(c, u, body)
|
||
if !g.checkModelAllowed(c, u, br.Model) {
|
||
apiError(c, http.StatusForbidden, "model_not_allowed", "模型未对你开放,请联系管理员")
|
||
return
|
||
}
|
||
|
||
cands := g.candidateChannels(br.Model)
|
||
if len(cands) == 0 {
|
||
apiError(c, http.StatusServiceUnavailable, "no_channel", "No available upstream channel")
|
||
g.recordError(c, nil, nil, now(), "no_channel")
|
||
return
|
||
}
|
||
sink := &usageSink{}
|
||
c.Set("usage_raw", &sinkHolder{sink: sink})
|
||
g.doProxy(c, cands, convert.ProtoChat, body, br.Stream, sink)
|
||
}
|
||
|
||
// responses POST /v1/responses(OpenAI Responses API)
|
||
func (g *Gateway) responses(c *gin.Context) {
|
||
u, ok := g.resolveUser(c)
|
||
if !ok {
|
||
return
|
||
}
|
||
if !g.checkBalance(c, u) {
|
||
return
|
||
}
|
||
br, body, err := parseBody(c)
|
||
if err != nil {
|
||
apiError(c, http.StatusBadRequest, "invalid_request_error", "failed to read request body")
|
||
return
|
||
}
|
||
c.Set("protocol", convert.ProtoResponses)
|
||
c.Set("model_name", br.Model)
|
||
g.recordRawRequest(c, u, body)
|
||
if !g.checkModelAllowed(c, u, br.Model) {
|
||
apiError(c, http.StatusForbidden, "model_not_allowed", "模型未对你开放,请联系管理员")
|
||
return
|
||
}
|
||
|
||
cands := g.candidateChannels(br.Model)
|
||
if len(cands) == 0 {
|
||
apiError(c, http.StatusServiceUnavailable, "no_channel", "No available upstream channel")
|
||
g.recordError(c, nil, nil, now(), "no_channel")
|
||
return
|
||
}
|
||
sink := &usageSink{}
|
||
c.Set("usage_raw", &sinkHolder{sink: sink})
|
||
g.doProxy(c, cands, convert.ProtoResponses, body, br.Stream, sink)
|
||
}
|
||
|
||
// messages POST /v1/messages(Anthropic Messages API)
|
||
func (g *Gateway) messages(c *gin.Context) {
|
||
u, ok := g.resolveUser(c)
|
||
if !ok {
|
||
return
|
||
}
|
||
if !g.checkBalance(c, u) {
|
||
return
|
||
}
|
||
br, body, err := parseBody(c)
|
||
if err != nil {
|
||
apiError(c, http.StatusBadRequest, "invalid_request_error", "failed to read request body")
|
||
return
|
||
}
|
||
c.Set("protocol", convert.ProtoMessages)
|
||
c.Set("model_name", br.Model)
|
||
g.recordRawRequest(c, u, body)
|
||
if !g.checkModelAllowed(c, u, br.Model) {
|
||
apiError(c, http.StatusForbidden, "model_not_allowed", "模型未对你开放,请联系管理员")
|
||
return
|
||
}
|
||
|
||
cands := g.candidateChannels(br.Model)
|
||
if len(cands) == 0 {
|
||
apiError(c, http.StatusServiceUnavailable, "no_channel", "No available upstream channel")
|
||
g.recordError(c, nil, nil, now(), "no_channel")
|
||
return
|
||
}
|
||
sink := &usageSink{}
|
||
c.Set("usage_raw", &sinkHolder{sink: sink})
|
||
g.doProxy(c, cands, convert.ProtoMessages, body, br.Stream, sink)
|
||
}
|
||
|
||
// usageSinkHolder 桥接:gin context 里保存 sink 引用,供 finishUsage 读取最终 usage。
|
||
type sinkHolder struct {
|
||
sink *usageSink
|
||
}
|
||
|
||
// recordRawRequest 记录管理员原始请求体到 context(供 finishUsage 落库)。
|
||
// 仅当开关开启且用户为管理员时记录;响应侧以 c.Get("raw_request") 是否非空判断是否需要捕获响应。
|
||
func (g *Gateway) recordRawRequest(c *gin.Context, u *store.User, body []byte) {
|
||
if g.logRaw && u.Role == store.RoleAdmin {
|
||
c.Set("raw_request", string(body))
|
||
}
|
||
}
|
||
|
||
// apiError 按客户端协议返回错误体(PLANNING §5.1.4)。
|
||
func apiError(c *gin.Context, status int, code, message string) {
|
||
if p, _ := c.Get("protocol"); p == convert.ProtoMessages {
|
||
// Anthropic 格式
|
||
c.AbortWithStatusJSON(status, gin.H{
|
||
"type": "error",
|
||
"error": gin.H{"type": errorTypeFor(status), "message": message},
|
||
})
|
||
return
|
||
}
|
||
// OpenAI 格式
|
||
c.AbortWithStatusJSON(status, gin.H{
|
||
"error": gin.H{
|
||
"message": message,
|
||
"type": errorTypeFor(status),
|
||
"param": nil,
|
||
"code": code,
|
||
},
|
||
})
|
||
}
|
||
|
||
func errorTypeFor(status int) string {
|
||
switch status {
|
||
case http.StatusUnauthorized:
|
||
return "authentication_error"
|
||
case http.StatusForbidden, http.StatusPaymentRequired:
|
||
return "permission_error"
|
||
case http.StatusNotFound, http.StatusBadRequest:
|
||
return "invalid_request_error"
|
||
case http.StatusTooManyRequests:
|
||
return "rate_limit_error"
|
||
default:
|
||
return "api_error"
|
||
}
|
||
}
|