feat(model): LLM 多供应商适配（Adapter/OpenAI 非流式+SSE 流式/per-role 路由/主备降级/多模态检测/调用记录）
This commit is contained in:
Aska committed
1 parent
51b2bdfba2
commit
d09ccc7570
9 files changed
+1285
No files matched your search
@@ -0,0 +1,256 @@
|
||||
package model
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"errors"
|
||||
"io"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"reflect"
|
||||
"strings"
|
||||
"testing"
|
||||
|
||||
"git.yeij.top/AskaEth/Cyrene_v2/internal/config"
|
||||
)
|
||||
|
||||
const okCompletion = `{
|
||||
"id": "cmpl-1",
|
||||
"model": "gpt-test",
|
||||
"choices": [
|
||||
{"index": 0, "message": {"role": "assistant", "content": "Hello!"}, "finish_reason": "stop"}
|
||||
],
|
||||
"usage": {"prompt_tokens": 5, "completion_tokens": 2, "total_tokens": 7}
|
||||
}`
|
||||
|
||||
// capturedRequest records what an OpenAIChatAdapter sent to a stub server.
|
||||
type capturedRequest struct {
|
||||
method string
|
||||
path string
|
||||
contentType string
|
||||
body map[string]any
|
||||
}
|
||||
|
||||
func newCompletionServer(t *testing.T, status int, contentType, body string, ch chan<- capturedRequest) *httptest.Server {
|
||||
t.Helper()
|
||||
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
var reqBody map[string]any
|
||||
if ch != nil {
|
||||
_ = json.NewDecoder(r.Body).Decode(&reqBody)
|
||||
}
|
||||
if contentType != "" {
|
||||
w.Header().Set("Content-Type", contentType)
|
||||
}
|
||||
w.WriteHeader(status)
|
||||
if body != "" {
|
||||
_, _ = io.WriteString(w, body)
|
||||
}
|
||||
if ch != nil {
|
||||
ch <- capturedRequest{
|
||||
method: r.Method,
|
||||
path: r.URL.Path,
|
||||
contentType: r.Header.Get("Content-Type"),
|
||||
body: reqBody,
|
||||
}
|
||||
}
|
||||
}))
|
||||
t.Cleanup(srv.Close)
|
||||
return srv
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_NonStreaming(t *testing.T) {
|
||||
t.Parallel()
|
||||
ch := make(chan capturedRequest, 1)
|
||||
srv := newCompletionServer(t, http.StatusOK, "application/json", okCompletion, ch)
|
||||
|
||||
recs := make([]CallRecord, 0, 1)
|
||||
temp := 1.0
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL, APIKey: "sk-test", Model: "gpt-test"},
|
||||
WithRecorder(func(cr CallRecord) { recs = append(recs, cr) }))
|
||||
|
||||
ctx := context.Background()
|
||||
resp, err := ad.Chat(ctx, Request{
|
||||
Messages: []Message{TextMessage(RoleUser, "hi")},
|
||||
Model: "gpt-test",
|
||||
Temperature: &temp,
|
||||
MaxTokens: 64,
|
||||
Purpose: "chat",
|
||||
}, Options{})
|
||||
if err != nil {
|
||||
t.Fatalf("Chat error: %v", err)
|
||||
}
|
||||
|
||||
req := <-ch
|
||||
if req.method != http.MethodPost {
|
||||
t.Errorf("method = %s, want POST", req.method)
|
||||
}
|
||||
if req.path != "/chat/completions" {
|
||||
t.Errorf("path = %s, want /chat/completions", req.path)
|
||||
}
|
||||
if req.contentType != "application/json" {
|
||||
t.Errorf("content-type = %s, want application/json", req.contentType)
|
||||
}
|
||||
if req.body["model"] != "gpt-test" {
|
||||
t.Errorf("body.model = %v, want gpt-test", req.body["model"])
|
||||
}
|
||||
if req.body["stream"] != nil && req.body["stream"] != false {
|
||||
t.Errorf("body.stream = %v, want omitted(false)", req.body["stream"])
|
||||
}
|
||||
if req.body["temperature"] != 1.0 {
|
||||
t.Errorf("body.temperature = %v, want 1.0", req.body["temperature"])
|
||||
}
|
||||
if req.body["max_tokens"] != float64(64) {
|
||||
t.Errorf("body.max_tokens = %v, want 64", req.body["max_tokens"])
|
||||
}
|
||||
msgs, _ := req.body["messages"].([]any)
|
||||
if len(msgs) != 1 {
|
||||
t.Fatalf("body.messages = %v, want 1 message", req.body["messages"])
|
||||
}
|
||||
|
||||
if resp.Model != "gpt-test" {
|
||||
t.Errorf("resp.Model = %q, want gpt-test", resp.Model)
|
||||
}
|
||||
if resp.Content != "Hello!" {
|
||||
t.Errorf("resp.Content = %q, want Hello!", resp.Content)
|
||||
}
|
||||
if resp.TotalTokens != 7 || resp.PromptTokens != 5 || resp.CompletionTokens != 2 {
|
||||
t.Errorf("resp usage = (%d,%d,%d), want (5,2,7)", resp.PromptTokens, resp.CompletionTokens, resp.TotalTokens)
|
||||
}
|
||||
if resp.Message.Role != RoleAssistant || len(resp.Message.Content) != 1 || resp.Message.Content[0].Text != "Hello!" {
|
||||
t.Errorf("resp.Message = %+v", resp.Message)
|
||||
}
|
||||
|
||||
if len(recs) != 1 {
|
||||
t.Fatalf("records = %d, want 1", len(recs))
|
||||
}
|
||||
if recs[0].Model != "gpt-test" || recs[0].Purpose != "chat" || recs[0].Tokens != 7 || recs[0].TookMs < 0 {
|
||||
t.Errorf("record = %+v, want model=gpt-test purpose=chat tokens=7 took_ms>=0", recs[0])
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_ImageContent(t *testing.T) {
|
||||
t.Parallel()
|
||||
ch := make(chan capturedRequest, 1)
|
||||
srv := newCompletionServer(t, http.StatusOK, "application/json", okCompletion, ch)
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL, Model: "gpt-test"})
|
||||
|
||||
req := Request{Messages: []Message{ImageMessage(RoleUser, "what is this", "data:image/png;base64,abc")}}
|
||||
if _, err := ad.Chat(context.Background(), req, Options{}); err != nil {
|
||||
t.Fatalf("Chat error: %v", err)
|
||||
}
|
||||
|
||||
got := <-ch
|
||||
msgs, _ := got.body["messages"].([]any)
|
||||
part0, _ := msgs[0].(map[string]any)
|
||||
content, _ := part0["content"].([]any)
|
||||
if len(content) != 2 {
|
||||
t.Fatalf("content parts = %d, want 2 (text+image)", len(content))
|
||||
}
|
||||
found := false
|
||||
for _, c := range content {
|
||||
cp, _ := c.(map[string]any)
|
||||
if cp["type"] != "image_url" {
|
||||
continue
|
||||
}
|
||||
found = true
|
||||
iu, _ := cp["image_url"].(map[string]any)
|
||||
if iu["url"] != "data:image/png;base64,abc" {
|
||||
t.Fatalf("image url = %v, want data:image/png;base64,abc", iu["url"])
|
||||
}
|
||||
}
|
||||
if !found {
|
||||
t.Fatalf("no image_url part found in %v", content)
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_ToolsPlaceholder(t *testing.T) {
|
||||
t.Parallel()
|
||||
ch := make(chan capturedRequest, 1)
|
||||
srv := newCompletionServer(t, http.StatusOK, "application/json", okCompletion, ch)
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL, Model: "gpt-test"})
|
||||
|
||||
req := Request{Messages: []Message{TextMessage(RoleUser, "hi")},
|
||||
Tools: []ToolSpec{{Name: "lookup", Description: "look it up", Parameters: map[string]any{"type": "object"}}}}
|
||||
if _, err := ad.Chat(context.Background(), req, Options{}); err != nil {
|
||||
t.Fatalf("Chat error: %v", err)
|
||||
}
|
||||
|
||||
got := <-ch
|
||||
tools, ok := got.body["tools"].([]any)
|
||||
if !ok || len(tools) != 1 {
|
||||
t.Fatalf("body.tools = %v, want 1 entry", got.body["tools"])
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_ChatStream_SSE(t *testing.T) {
|
||||
t.Parallel()
|
||||
sse := strings.Join([]string{
|
||||
`data: {"model":"gpt-test","choices":[{"index":0,"delta":{"role":"assistant","content":"Hel"}}]}`,
|
||||
``,
|
||||
`data: {"choices":[{"delta":{"content":"lo"}}]}`,
|
||||
``,
|
||||
`data: {"model":"gpt-test","choices":[],"usage":{"prompt_tokens":5,"completion_tokens":2,"total_tokens":7}}`,
|
||||
``,
|
||||
`data: [DONE]`,
|
||||
``,
|
||||
}, "\n")
|
||||
srv := newCompletionServer(t, http.StatusOK, "text/event-stream", sse, nil)
|
||||
|
||||
recs := make([]CallRecord, 0, 1)
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL, Model: "gpt-test"},
|
||||
WithRecorder(func(cr CallRecord) { recs = append(recs, cr) }))
|
||||
|
||||
var deltas []string
|
||||
resp, err := ad.ChatStream(context.Background(),
|
||||
Request{Messages: []Message{TextMessage(RoleUser, "hi")}, Model: "gpt-test", Purpose: "chat"},
|
||||
func(d string) { deltas = append(deltas, d) })
|
||||
if err != nil {
|
||||
t.Fatalf("ChatStream error: %v", err)
|
||||
}
|
||||
if resp.Content != "Hello" {
|
||||
t.Errorf("resp.Content = %q, want Hello", resp.Content)
|
||||
}
|
||||
if resp.Model != "gpt-test" {
|
||||
t.Errorf("resp.Model = %q, want gpt-test", resp.Model)
|
||||
}
|
||||
if resp.TotalTokens != 7 {
|
||||
t.Errorf("resp.TotalTokens = %d, want 7", resp.TotalTokens)
|
||||
}
|
||||
if !reflect.DeepEqual(deltas, []string{"Hel", "lo"}) {
|
||||
t.Errorf("deltas = %v, want [Hel lo]", deltas)
|
||||
}
|
||||
if len(recs) != 1 || recs[0].Tokens != 7 || recs[0].Purpose != "chat" {
|
||||
t.Errorf("records = %+v, want 1 record tokens=7 purpose=chat", recs)
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_UpstreamError(t *testing.T) {
|
||||
t.Parallel()
|
||||
srv := newCompletionServer(t, http.StatusInternalServerError, "application/json", `{"error":"boom"}`, nil)
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL, Model: "m"})
|
||||
|
||||
_, err := ad.Chat(context.Background(), Request{Messages: []Message{TextMessage(RoleUser, "hi")}}, Options{})
|
||||
if err == nil {
|
||||
t.Fatal("want error, got nil")
|
||||
}
|
||||
if !errors.Is(err, ErrUpstream) {
|
||||
t.Fatalf("err = %v, want ErrUpstream", err)
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_NoBaseURL(t *testing.T) {
|
||||
t.Parallel()
|
||||
ad := NewOpenAIChat(config.LLMConfig{})
|
||||
if _, err := ad.Chat(context.Background(), Request{Messages: []Message{TextMessage(RoleUser, "hi")}}, Options{}); !errors.Is(err, ErrUpstream) {
|
||||
t.Fatalf("err = %v, want ErrUpstream", err)
|
||||
}
|
||||
}
|
||||
|
||||
func TestOpenAIChat_Chat_NoModel(t *testing.T) {
|
||||
t.Parallel()
|
||||
srv := newCompletionServer(t, http.StatusOK, "application/json", okCompletion, nil)
|
||||
ad := NewOpenAIChat(config.LLMConfig{BaseURL: srv.URL})
|
||||
if _, err := ad.Chat(context.Background(), Request{Messages: []Message{TextMessage(RoleUser, "hi")}}, Options{}); !errors.Is(err, ErrNoModel) {
|
||||
t.Fatalf("err = %v, want ErrNoModel", err)
|
||||
}
|
||||
}
|
||||
Reference in new issue
Block a user