Upload 4 files

app.py

@@ -83,27 +83,23 @@ safety_settings_g2 = [
 ]
 @dataclass
 class GeneratedText:
-    """用于存储生成的文本片段"""
     text: str
     finish_reason: Optional[str] = None
 
 
 class ResponseWrapper:
-    """处理非流式响应的包装类"""
     def __init__(self, data: Dict[Any, Any]):
         self._data = data
         self._text = self._extract_text()
         self._finish_reason = self._extract_finish_reason()
 
     def _extract_text(self) -> str:
-        """从响应数据中提取文本"""
         try:
             return self._data['candidates'][0]['content']['parts'][0]['text']
         except (KeyError, IndexError):
             return ""
 
     def _extract_finish_reason(self) -> Optional[str]:
-        """提取完成原因"""
         try:
             return self._data['candidates'][0].get('finishReason')
         except (KeyError, IndexError):
@@ -111,12 +107,10 @@ class ResponseWrapper:
 
     @property
     def text(self) -> str:
-        """获取响应文本"""
         return self._text
 
     @property
     def finish_reason(self) -> Optional[str]:
-        """获取完成原因"""
         return self._finish_reason
 
 class APIKeyManager:
@@ -345,6 +339,7 @@ def chat_completions():
     model = request_data.get('model', 'gemini-2.0-flash-exp')
     temperature = request_data.get('temperature', 1)
     max_tokens = request_data.get('max_tokens', 8192)
+    show_thoughts = request_data.get('show_thoughts', False)
     stream = request_data.get('stream', False)
     hint = "流式" if stream else "非流"
     logger.info(f"\n{model} [{hint}] → ...")
@@ -399,7 +394,7 @@ def chat_completions():
             return handle_api_error(e, attempt)
 
     def generate_stream(response):
-        buffer = b""  
+        buffer = b""
         try:
             for line in response.iter_lines():
                 if not line:
@@ -418,21 +413,25 @@ def chat_completions():
                             if 'content' in candidate:
                                 content = candidate['content']
                                 if 'parts' in content and content['parts']:
-                                    text = content['parts'][0].get('text', '')
-                                    finish_reason = candidate.get('finishReason')
-
-                                    if text:
-                                        data = {
-                                            'choices': [{
-                                                'delta': {
-                                                    'content': text
-                                                },
-                                                'finish_reason': finish_reason,
-                                                'index': 0
-                                            }],
-                                            'object': 'chat.completion.chunk'
-                                        }
-                                        yield f"data: {json.dumps(data)}\n\n"
+                                    parts = content['parts']
+                                    if is_thinking and not show_thoughts:
+                                        parts = [part for part in parts if not part.get('thought')]
+                                    if parts: 
+                                        text = parts[0].get('text', '')
+                                        finish_reason = candidate.get('finishReason')
+
+                                        if text:
+                                            data = {
+                                                'choices': [{
+                                                    'delta': {
+                                                        'content': text
+                                                    },
+                                                    'finish_reason': finish_reason,
+                                                    'index': 0
+                                                }],
+                                                'object': 'chat.completion.chunk'
+                                            }
+                                            yield f"data: {json.dumps(data)}\n\n"
 
                     except json.JSONDecodeError:
                         logger.debug(f"JSONDecodeError, buffer now: {buffer}")
@@ -487,7 +486,7 @@ def chat_completions():
         try:
             text_content = response.text
         except (AttributeError, IndexError, TypeError, ValueError) as e:
-            if "response.candidates" in str(e) or "response.text" in str(e):
+             if "response.candidates" in str(e) or "response.text" in str(e):
                 logger.error(f"用户输入被AI安全过滤器阻止")
                 return jsonify({
                     'error': {
@@ -496,13 +495,28 @@ def chat_completions():
                         'details': str(e)
                     }
                 }), 400
-            else:
+             else:
                 return jsonify({
                     'error': {
-                        'message': 'AI响应处理失败',
-                        'type': 'response_processing_error'
-                    }
+                         'message': 'AI响应处理失败',
+                         'type': 'response_processing_error'
+                      }
                 }), 500
+        try:
+            response_json = json.loads(response.text)
+            if 'candidates' in response_json and response_json['candidates']:
+                 candidate = response_json['candidates'][0]
+                 if 'content' in candidate:
+                    content = candidate['content']
+                    if 'parts' in content and content['parts']:
+                        parts = content['parts']
+                        if is_thinking and not show_thoughts:
+                            parts = [part for part in parts if not part.get('thought')]
+
+                        if parts:
+                            text_content = "".join(part.get('text', '') for part in parts)
+        except json.JSONDecodeError:
+           pass
 
         response_data = {
             'id': 'chatcmpl-xxxxxxxxxxxx',