Support Gradio 6 Chatbot format and multi-output spaces like tdecae-personal-chatbot

gr2gw.go

@@ -73,14 +73,14 @@ type ChatMessage struct {
   Name             string      `json:"name,omitempty"`
 }
 
-func (m *ChatMessage) GetContentString() string {
+func extractContentString(val interface{}) string {
-  if m.Content == nil {
+  if val == nil {
     return ""
   }
-  if str, ok := m.Content.(string); ok {
+  if str, ok := val.(string); ok {
     return str
   }
-  if parts, ok := m.Content.([]interface{}); ok {
+  if parts, ok := val.([]interface{}); ok {
     var sb strings.Builder
     for _, p := range parts {
       if str, ok := p.(string); ok {
@@ -88,16 +88,29 @@ func (m *ChatMessage) GetContentString() string {
       } else if itemMap, ok := p.(map[string]interface{}); ok {
         if textVal, ok := itemMap["text"].(string); ok {
           sb.WriteString(textVal)
+        } else if valStr, ok := itemMap["value"].(string); ok {
+          sb.WriteString(valStr)
         }
       }
     }
     return sb.String()
   }
-  b, err := json.Marshal(m.Content)
+  if itemMap, ok := val.(map[string]interface{}); ok {
+    if textVal, ok := itemMap["text"].(string); ok {
+      return textVal
+    } else if valStr, ok := itemMap["value"].(string); ok {
+      return valStr
+    }
+  }
+  b, err := json.Marshal(val)
   if err == nil {
     return string(b)
   }
-  return fmt.Sprintf("%v", m.Content)
+  return fmt.Sprintf("%v", val)
+}
+
+func (m *ChatMessage) GetContentString() string {
+  return extractContentString(m.Content)
 }
 
 type ChatCompletionRequest struct {
@@ -1520,6 +1533,8 @@ type GradioParamInfo struct {
   ParameterName    string      `json:"parameter_name"`
   ParameterDefault interface{} `json:"parameter_default,omitempty"`
   Component        string      `json:"component"`
+  Type             interface{} `json:"type,omitempty"`
+  PythonType       interface{} `json:"python_type,omitempty"`
 }
 
 type GradioEndpointInfo struct {
@@ -1828,20 +1843,32 @@ func InspectSpace(client *http.Client, rawURL, userAgent string) (*SpaceDiscover
       score := 0
       lowerName := strings.ToLower(epName)
 
-      if strings.Contains(lowerName, "chat") {
+      // Penalize non-generation/reset/clear/init endpoints or 0-parameter endpoints
+      if len(epInfo.Parameters) == 0 {
+        score -= 500
+      }
+      if strings.Contains(lowerName, "clear") || strings.Contains(lowerName, "reset") || strings.Contains(lowerName, "init") || strings.Contains(lowerName, "undo") || strings.Contains(lowerName, "delete") {
+        score -= 500
+      }
+
+      if strings.Contains(lowerName, "chat") || strings.Contains(lowerName, "conversation") || strings.Contains(lowerName, "dialogue") {
         score += 100
       }
-      if strings.Contains(lowerName, "predict") || strings.Contains(lowerName, "respond") || strings.Contains(lowerName, "generate") {
+      if strings.Contains(lowerName, "answer") || strings.Contains(lowerName, "respond") || strings.Contains(lowerName, "generate") || strings.Contains(lowerName, "predict") || strings.Contains(lowerName, "completion") {
-        score += 50
+        score += 60
+      }
+      if strings.Contains(lowerName, "ask") || strings.Contains(lowerName, "query") || strings.Contains(lowerName, "question") || strings.Contains(lowerName, "talk") {
+        score += 40
       }
 
       for _, p := range epInfo.Parameters {
         pLower := strings.ToLower(p.ParameterName)
-        if strings.Contains(pLower, "message") || strings.Contains(pLower, "text") || strings.Contains(pLower, "prompt") {
+        pComp := strings.ToLower(p.Component)
+        if strings.Contains(pLower, "message") || strings.Contains(pLower, "text") || strings.Contains(pLower, "prompt") || strings.Contains(pLower, "query") || strings.Contains(pLower, "question") || strings.Contains(pLower, "input") || pComp == "textbox" {
           score += 40
         }
-        if strings.Contains(pLower, "history") || strings.Contains(pLower, "chat") {
+        if strings.Contains(pLower, "history") || strings.Contains(pLower, "chat") || strings.Contains(pLower, "messages") || strings.Contains(pLower, "conversation") || pComp == "chatbot" {
-          score += 20
+          score += 30
         }
       }
 
@@ -1943,6 +1970,7 @@ func InspectSpace(client *http.Client, rawURL, userAgent string) (*SpaceDiscover
     }
     for idx, p := range bestEndpointInfo.Parameters {
       pName := strings.ToLower(p.ParameterName)
+      pComp := strings.ToLower(p.Component)
       if strings.Contains(pName, "system") {
         discovery.SystemIndex = idx
         if p.ParameterDefault != nil && discovery.DefaultSystemPrompt == "" {
@@ -1950,9 +1978,20 @@ func InspectSpace(client *http.Client, rawURL, userAgent string) (*SpaceDiscover
             discovery.DefaultSystemPrompt = strings.TrimSpace(defStr)
           }
         }
-      } else if strings.Contains(pName, "history") || strings.Contains(pName, "chat") {
+      } else if strings.Contains(pName, "history") || strings.Contains(pName, "chat") || strings.Contains(pName, "conversation") || strings.Contains(pName, "messages") || pComp == "chatbot" {
         discovery.HistoryIndex = idx
-      } else if strings.Contains(pName, "message") || (strings.Contains(pName, "prompt") && !strings.Contains(pName, "system")) || strings.Contains(pName, "text") {
+        bType, _ := json.Marshal(p.Type)
+        bPyType, _ := json.Marshal(p.PythonType)
+        pPyType := strings.ToLower(string(bPyType))
+        bTypeStr := strings.ToLower(string(bType))
+        if strings.Contains(pPyType, "list[tuple[") || strings.Contains(pPyType, "list[list[") || strings.Contains(bTypeStr, "tuple") {
+          discovery.HistoryFormat = "pairs"
+        } else if strings.Contains(pPyType, "textmessage") || strings.Contains(pPyType, "dict(text: str") || strings.Contains(bTypeStr, "textmessage") || strings.Contains(bTypeStr, "chatbotdatamessages") {
+          discovery.HistoryFormat = "gradio_messages"
+        } else if strings.HasPrefix(configResp.Version, "5.") || strings.HasPrefix(configResp.Version, "6.") {
+          discovery.HistoryFormat = "gradio_messages"
+        }
+      } else if strings.Contains(pName, "message") || (strings.Contains(pName, "prompt") && !strings.Contains(pName, "system")) || strings.Contains(pName, "text") || strings.Contains(pName, "query") || strings.Contains(pName, "question") || strings.Contains(pName, "input") || pComp == "textbox" {
         discovery.MessageIndex = idx
       } else if strings.Contains(pName, "think_level") || strings.Contains(pName, "thinking_level") {
         discovery.ThinkLevelIndex = idx
@@ -1972,6 +2011,7 @@ func InspectSpace(client *http.Client, rawURL, userAgent string) (*SpaceDiscover
 
   if discovery.FunctionsJSONIndex != -1 || discovery.ThinkLevelIndex != -1 || strings.Contains(cleanURL, "hy3") || strings.Contains(cleanURL, "hunyuan") {
     discovery.IsHunyuan3 = true
+    discovery.HistoryFormat = "messages"
     discovery.Models = append(discovery.Models, "hy3", "hunyuan3", "tencent/Hy3")
     if discovery.PrimaryModel == "gradio-chat" || discovery.PrimaryModel == "" {
       discovery.PrimaryModel = "hy3"
@@ -2301,6 +2341,26 @@ func (g *GradioGateway) BuildGradioPayload(disc *SpaceDiscovery, req ChatComplet
         pairs = append(pairs, []string{u, a})
       }
       data[disc.HistoryIndex] = pairs
+    } else if disc.HistoryFormat == "gradio_messages" {
+      var gMsgs []map[string]interface{}
+      for _, item := range historyArray {
+        role, _ := item["role"].(string)
+        cStr := ""
+        if s, ok := item["content"].(string); ok {
+          cStr = s
+        }
+        gMsg := map[string]interface{}{
+          "role": role,
+          "content": []map[string]string{
+            {"text": cStr, "type": "text"},
+          },
+        }
+        gMsgs = append(gMsgs, gMsg)
+      }
+      if gMsgs == nil {
+        gMsgs = []map[string]interface{}{}
+      }
+      data[disc.HistoryIndex] = gMsgs
     } else {
       data[disc.HistoryIndex] = historyArray
     }
@@ -2393,107 +2453,182 @@ func ParseGradioStreamOutput(rawJSON string) GradioOutputFrame {
       return frame
     }
 
-    // Check if v[0] is an inner slice (e.g. Hy3: [[content, reasoning, tool_calls, history]])
+    // 1. Check if v[0] is an inner slice with len >= 3 (e.g. Hy3: [[content, reasoning, tool_calls, history]])
-    if inner, ok := v[0].([]interface{}); ok {
+    if inner, ok := v[0].([]interface{}); ok && len(inner) >= 3 {
-      if len(inner) >= 3 {
+      s0, _ := inner[0].(string)
-        s0, _ := inner[0].(string)
+      s1, _ := inner[1].(string)
-        s1, _ := inner[1].(string)
+      frame.Content = s0
-        frame.Content = s0
+      frame.Reasoning = s1
-        frame.Reasoning = s1
+      if inner[2] != nil {
-        if inner[2] != nil {
+        b, err := json.Marshal(inner[2])
-          b, err := json.Marshal(inner[2])
-          if err == nil {
-            var tcs []ToolCall
-            if json.Unmarshal(b, &tcs) == nil && len(tcs) > 0 {
-              frame.ToolCalls = tcs
-            }
-          }
-        }
-        frame.OK = true
-        return frame
-      }
-
-      if len(inner) == 2 {
-        s0, ok0 := inner[0].(string)
-        s1, ok1 := inner[1].(string)
-        if ok0 && ok1 {
-          frame.Content = s0
-          frame.Reasoning = s1
-          frame.OK = true
-          return frame
-        }
-
-        // Check if inner is a chat pair: ["user msg", "assistant msg"]
-        if ok1 {
-          frame.Content = s1
-          frame.OK = true
-          return frame
-        }
-      }
-
-      // Check if inner is a list of chat message maps: [{"role":..., "content":...}, ...]
-      // or list of pairs: [["u", "a"], ...]
-      if len(inner) > 0 {
-        lastItem := inner[len(inner)-1]
-        if m, ok := lastItem.(map[string]interface{}); ok {
-          if c, ok := m["content"].(string); ok {
-            frame.Content = c
-            frame.OK = true
-          }
-          if r, ok := m["reasoning_content"].(string); ok {
-            frame.Reasoning = r
-          }
-          if tcsRaw, ok := m["tool_calls"].([]interface{}); ok && len(tcsRaw) > 0 {
-            b, err := json.Marshal(tcsRaw)
-            if err == nil {
-              var tcs []ToolCall
-              if err := json.Unmarshal(b, &tcs); err == nil {
-                frame.ToolCalls = tcs
-              }
-            }
-          }
-          if frame.OK {
-            return frame
-          }
-        } else if pair, ok := lastItem.([]interface{}); ok && len(pair) >= 2 {
-          if aStr, ok := pair[1].(string); ok {
-            frame.Content = aStr
-            frame.OK = true
-            return frame
-          }
-        }
-      }
-    }
-
-    // Check if v[0] is string (standard single output e.g. ["content", null])
-    if s, ok := v[0].(string); ok {
-      frame.Content = s
-      frame.OK = true
-      return frame
-    }
-
-    // Check if v is a flat list of messages: [{"role": "assistant", ...}]
-    lastItem := v[len(v)-1]
-    if m, ok := lastItem.(map[string]interface{}); ok {
-      if c, ok := m["content"].(string); ok {
-        frame.Content = c
-        frame.OK = true
-      }
-      if r, ok := m["reasoning_content"].(string); ok {
-        frame.Reasoning = r
-      }
-      if tcsRaw, ok := m["tool_calls"].([]interface{}); ok && len(tcsRaw) > 0 {
-        b, err := json.Marshal(tcsRaw)
         if err == nil {
           var tcs []ToolCall
-          if err := json.Unmarshal(b, &tcs); err == nil {
+          if json.Unmarshal(b, &tcs) == nil && len(tcs) > 0 {
             frame.ToolCalls = tcs
           }
         }
       }
-      if frame.OK {
+      frame.OK = true
+      return frame
+    }
+
+    // 2. Check if any element of v is a Chatbot message list or Chatbot pair list
+    for _, elem := range v {
+      if msgList, ok := elem.([]interface{}); ok && len(msgList) > 0 {
+        allMaps := true
+        var maps []map[string]interface{}
+        for _, item := range msgList {
+          if m, ok := item.(map[string]interface{}); ok {
+            if _, hasRole := m["role"]; hasRole {
+              maps = append(maps, m)
+              continue
+            }
+          }
+          allMaps = false
+          break
+        }
+        if allMaps && len(maps) > 0 {
+          var targetMsg map[string]interface{}
+          for i := len(maps) - 1; i >= 0; i-- {
+            if r, _ := maps[i]["role"].(string); r == "assistant" {
+              targetMsg = maps[i]
+              break
+            }
+          }
+          if targetMsg == nil {
+            targetMsg = maps[len(maps)-1]
+          }
+
+          cText := extractContentString(targetMsg["content"])
+          frame.Content = cText
+
+          if r, ok := targetMsg["reasoning_content"].(string); ok {
+            frame.Reasoning = r
+          } else if meta, ok := targetMsg["metadata"].(map[string]interface{}); ok {
+            if logStr, ok := meta["log"].(string); ok && logStr != "" {
+              frame.Reasoning = logStr
+            }
+          }
+
+          if tcsRaw, ok := targetMsg["tool_calls"].([]interface{}); ok && len(tcsRaw) > 0 {
+            b, err := json.Marshal(tcsRaw)
+            if err == nil {
+              var tcs []ToolCall
+              if json.Unmarshal(b, &tcs) == nil {
+                frame.ToolCalls = tcs
+              }
+            }
+          }
+
+          if len(frame.ToolCalls) == 0 {
+            tcs, clean, has := DetectToolCalls(cText)
+            if has && len(tcs) > 0 {
+              frame.ToolCalls = tcs
+              frame.Content = clean
+            }
+          }
+
+          frame.OK = true
+          return frame
+        }
+
+        // Check if elements are pairs [user, assistant]
+        allPairs := true
+        var pairs [][]interface{}
+        for _, item := range msgList {
+          if p, ok := item.([]interface{}); ok && len(p) == 2 {
+            pairs = append(pairs, p)
+            continue
+          }
+          allPairs = false
+          break
+        }
+        if allPairs && len(pairs) > 0 {
+          lastPair := pairs[len(pairs)-1]
+          cText := extractContentString(lastPair[1])
+          frame.Content = cText
+          tcs, clean, has := DetectToolCalls(cText)
+          if has && len(tcs) > 0 {
+            frame.ToolCalls = tcs
+            frame.Content = clean
+          }
+          frame.OK = true
+          return frame
+        }
+      }
+    }
+
+    // 3. Check if v itself is a list of message maps: [{"role": "assistant", ...}]
+    allMaps := true
+    var maps []map[string]interface{}
+    for _, item := range v {
+      if m, ok := item.(map[string]interface{}); ok {
+        if _, hasRole := m["role"]; hasRole {
+          maps = append(maps, m)
+          continue
+        }
+      }
+      allMaps = false
+      break
+    }
+    if allMaps && len(maps) > 0 {
+      var targetMsg map[string]interface{}
+      for i := len(maps) - 1; i >= 0; i-- {
+        if r, _ := maps[i]["role"].(string); r == "assistant" {
+          targetMsg = maps[i]
+          break
+        }
+      }
+      if targetMsg == nil {
+        targetMsg = maps[len(maps)-1]
+      }
+      cText := extractContentString(targetMsg["content"])
+      frame.Content = cText
+      if r, ok := targetMsg["reasoning_content"].(string); ok {
+        frame.Reasoning = r
+      }
+      if tcsRaw, ok := targetMsg["tool_calls"].([]interface{}); ok && len(tcsRaw) > 0 {
+        b, err := json.Marshal(tcsRaw)
+        if err == nil {
+          var tcs []ToolCall
+          if json.Unmarshal(b, &tcs) == nil {
+            frame.ToolCalls = tcs
+          }
+        }
+      }
+      if len(frame.ToolCalls) == 0 {
+        tcs, clean, has := DetectToolCalls(cText)
+        if has && len(tcs) > 0 {
+          frame.ToolCalls = tcs
+          frame.Content = clean
+        }
+      }
+      frame.OK = true
+      return frame
+    }
+
+    // 4. Check if v[0] is an inner pair [content, reasoning]
+    if inner, ok := v[0].([]interface{}); ok && len(inner) == 2 {
+      s0, ok0 := inner[0].(string)
+      s1, ok1 := inner[1].(string)
+      if ok0 && ok1 {
+        frame.Content = s0
+        frame.Reasoning = s1
+        frame.OK = true
         return frame
       }
+      if ok1 {
+        frame.Content = s1
+        frame.OK = true
+        return frame
+      }
+    }
+
+    // 5. Check if v[0] is a non-empty string or single output
+    if s, ok := v[0].(string); ok && (s != "" || len(v) == 1) {
+      frame.Content = s
+      frame.OK = true
+      return frame
     }
 
   case map[string]interface{}:

gr2gw_test.go

@@ -298,6 +298,13 @@ func TestParseGradioStreamOutput(t *testing.T) {
   if !frame7.OK || frame7.Content != "Mocked answer" || frame7.Reasoning != "Mocked reasoning" || len(frame7.ToolCalls) != 0 {
     t.Errorf("unexpected frame7: %+v", frame7)
   }
+
+  // 8. Gradio 6 Chatbot multi-output format with TextMessage list
+  g6Raw := `["", [{"role": "user", "content": [{"text": "weather in Tokyo?", "type": "text"}]}, {"role": "assistant", "content": [{"text": "<tool_call>\n{\"name\": \"get_weather\", \"arguments\": {\"city\": \"Tokyo\"}}\n</tool_call>", "type": "text"}]}]]`
+  frame8 := ParseGradioStreamOutput(g6Raw)
+  if !frame8.OK || len(frame8.ToolCalls) != 1 || frame8.ToolCalls[0].Function.Name != "get_weather" {
+    t.Errorf("unexpected frame8: %+v", frame8)
+  }
 }
 
 func TestHunyuan3BuildPayload(t *testing.T) {