Support multimodal textbox spaces with state inputs and default parameter mapping

This commit is contained in:
Luxferre
2026-09-07 11:38:37 +03:00
parent 9f2a5068f8
commit 491114b2bd
3 changed files with 479 additions and 53 deletions
+201
View File
@@ -1397,3 +1397,204 @@ func TestConfiguredModelName(t *testing.T) {
}
}
func TestMultimodalStateSpaceMockServerCompletion(t *testing.T) {
ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.URL.Path == "/config" {
cfg := GradioConfigResponse{
Version: "5.29.0",
Dependencies: []GradioDependency{
{
ID: 6,
APIName: "chat",
Inputs: []int{12, 16, 20, 21, 22, 23, 24, 25, 26},
Outputs: []int{14, 16},
},
},
Components: []GradioComponent{
{ID: 12, Type: "multimodaltextbox", Props: map[string]interface{}{"label": "Message"}},
{ID: 16, Type: "state"},
{ID: 20, Type: "radio", Props: map[string]interface{}{"label": "Model Type", "value": "Chat"}},
{ID: 21, Type: "checkbox", Props: map[string]interface{}{"label": "Use Internet", "value": false}},
{ID: 22, Type: "slider", Props: map[string]interface{}{"label": "Max Tokens", "value": 32768}},
{ID: 23, Type: "slider", Props: map[string]interface{}{"label": "Temperature", "value": 0.8}},
{ID: 24, Type: "slider", Props: map[string]interface{}{"label": "Top P", "value": 0.95}},
{ID: 25, Type: "checkbox", Props: map[string]interface{}{"label": "stream", "value": true}},
{ID: 26, Type: "textbox", Props: map[string]interface{}{"label": "user", "value": "null"}},
},
}
w.Header().Set("Content-Type", "application/json")
json.NewEncoder(w).Encode(cfg)
return
}
if r.URL.Path == "/gradio_api/info" {
info := GradioAPIInfoResponse{
NamedEndpoints: map[string]GradioEndpointInfo{
"/chat": {
Parameters: []GradioParamInfo{
{ParameterName: "param_0", Label: "Message", Component: "Multimodaltextbox"},
{ParameterName: "param_2", Label: "Model Type", Component: "Radio", ParameterDefault: "Chat"},
{ParameterName: "param_3", Label: "Use Internet", Component: "Checkbox", ParameterDefault: false},
{ParameterName: "param_4", Label: "Max Tokens", Component: "Slider", ParameterDefault: 32768},
{ParameterName: "param_5", Label: "Temperature", Component: "Slider", ParameterDefault: 0.8},
{ParameterName: "param_6", Label: "Top P", Component: "Slider", ParameterDefault: 0.95},
{ParameterName: "param_7", Label: "stream", Component: "Checkbox", ParameterDefault: true},
{ParameterName: "param_8", Label: "user", Component: "Textbox", ParameterDefault: "null"},
},
},
},
}
w.Header().Set("Content-Type", "application/json")
json.NewEncoder(w).Encode(info)
return
}
if r.URL.Path == "/gradio_api/call/chat" {
var body struct {
Data []interface{} `json:"data"`
}
if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
http.Error(w, err.Error(), http.StatusBadRequest)
return
}
if len(body.Data) != 9 {
http.Error(w, fmt.Sprintf("expected 9 inputs, got %d", len(body.Data)), http.StatusBadRequest)
return
}
// Check multimodal dict at index 0
mmDict, ok := body.Data[0].(map[string]interface{})
if !ok {
http.Error(w, "input 0 must be multimodal dict", http.StatusBadRequest)
return
}
txt, _ := mmDict["text"].(string)
evt := "evt_normal"
if strings.Contains(txt, "get_weather") {
evt = "evt_tool"
}
w.Header().Set("Content-Type", "application/json")
json.NewEncoder(w).Encode(GradioJoinResponse{EventID: evt})
return
}
if r.URL.Path == "/gradio_api/call/chat/evt_normal" {
w.Header().Set("Content-Type", "text/event-stream")
flusher, _ := w.(http.Flusher)
fmt.Fprintf(w, "event: complete\ndata: [\"Hello from multimodal space!\", null]\n\n")
flusher.Flush()
return
}
if r.URL.Path == "/gradio_api/call/chat/evt_tool" {
w.Header().Set("Content-Type", "text/event-stream")
flusher, _ := w.(http.Flusher)
fmt.Fprintf(w, "event: complete\ndata: [\"```json\\n[{\\\"name\\\": \\\"get_weather\\\", \\\"arguments\\\": {\\\"location\\\": \\\"Tokyo\\\"}}]\\n```\", null]\n\n")
flusher.Flush()
return
}
http.NotFound(w, r)
}))
defer ts.Close()
gw := NewGradioGateway(ts.URL, "", 10*time.Second)
// 1. Verify Space Inspection
disc, err := InspectSpace(gw.client, ts.URL, DefaultUserAgent)
if err != nil {
t.Fatalf("InspectSpace failed: %v", err)
}
if disc.TotalInputs != 9 {
t.Errorf("expected TotalInputs 9, got %d", disc.TotalInputs)
}
if disc.MessageIndex != 0 {
t.Errorf("expected MessageIndex 0, got %d", disc.MessageIndex)
}
if !disc.MessageIsMultimodal {
t.Errorf("expected MessageIsMultimodal true")
}
if disc.TempIndex != 5 {
t.Errorf("expected TempIndex 5, got %d", disc.TempIndex)
}
if disc.MaxTokensIndex != 4 {
t.Errorf("expected MaxTokensIndex 4, got %d", disc.MaxTokensIndex)
}
if disc.TopPIndex != 6 {
t.Errorf("expected TopPIndex 6, got %d", disc.TopPIndex)
}
if disc.StreamIndex != 7 {
t.Errorf("expected StreamIndex 7, got %d", disc.StreamIndex)
}
// 2. Normal Chat Completion
req1 := ChatCompletionRequest{
Model: "askcyph",
Messages: []ChatMessage{
{Role: "user", Content: "Hello world"},
},
Stream: false,
}
b1, _ := json.Marshal(req1)
httpReq1 := httptest.NewRequest("POST", "/v1/chat/completions", bytes.NewBuffer(b1))
rec1 := httptest.NewRecorder()
err = gw.ExecuteChatCompletion(rec1, httpReq1, req1)
if err != nil {
t.Fatalf("ExecuteChatCompletion failed: %v", err)
}
var resp1 ChatCompletionResponse
if err := json.NewDecoder(rec1.Body).Decode(&resp1); err != nil {
t.Fatalf("failed to decode response: %v", err)
}
if resp1.Choices[0].Message.Content != "Hello from multimodal space!" {
t.Errorf("unexpected content: %v", resp1.Choices[0].Message.Content)
}
// 3. Tool Calling Chat Completion
req2 := ChatCompletionRequest{
Model: "askcyph",
Messages: []ChatMessage{
{Role: "user", Content: "Weather in Tokyo?"},
},
Tools: []Tool{
{
Type: "function",
Function: map[string]interface{}{
"name": "get_weather",
"description": "Get current weather",
"parameters": map[string]interface{}{
"type": "object",
"properties": map[string]interface{}{
"location": map[string]interface{}{"type": "string"},
},
"required": []string{"location"},
},
},
},
},
Stream: false,
}
b2, _ := json.Marshal(req2)
httpReq2 := httptest.NewRequest("POST", "/v1/chat/completions", bytes.NewBuffer(b2))
rec2 := httptest.NewRecorder()
err = gw.ExecuteChatCompletion(rec2, httpReq2, req2)
if err != nil {
t.Fatalf("ExecuteChatCompletion tool calling failed: %v", err)
}
var resp2 ChatCompletionResponse
if err := json.NewDecoder(rec2.Body).Decode(&resp2); err != nil {
t.Fatalf("failed to decode response: %v", err)
}
if resp2.Choices[0].FinishReason != "tool_calls" {
t.Fatalf("expected finish_reason 'tool_calls', got %q", resp2.Choices[0].FinishReason)
}
if len(resp2.Choices[0].Message.ToolCalls) != 1 {
t.Fatalf("expected 1 tool call, got %d", len(resp2.Choices[0].Message.ToolCalls))
}
if resp2.Choices[0].Message.ToolCalls[0].Function.Name != "get_weather" {
t.Errorf("expected get_weather, got %s", resp2.Choices[0].Message.ToolCalls[0].Function.Name)
}
}