Update app.py

app.py

@@ -7,9 +7,9 @@ from flask import Flask, request, jsonify, Response
 # 1. 初始化Flask应用
 app = Flask(__name__)
 
-# 2. gpt-oss.com API的固定配置
+# 2. gpt-oss.com API的固定配置 (移除了Cookie)
 GPT_OSS_API_URL = "https://api.gpt-oss.com/chatkit"
-GPT_OSS_HEADERS = {
+BASE_GPT_OSS_HEADERS = {
     'authority': 'api.gpt-oss.com',
     'accept': 'text/event-stream',
     'content-type': 'application/json',
@@ -19,24 +19,45 @@ GPT_OSS_HEADERS = {
     'x-selected-model': 'gpt-oss-120b',
 }
 
-# 3. 根路由，用于健康检查
+# 3. 根路由，包含新的使用说明
 @app.route('/', methods=['GET'])
 def root():
-    return jsonify({"message": "欢迎使用 GPT-OSS to OpenAI 格式代理API", "status": "ok"})
+    return jsonify({
+        "message": "欢迎使用 GPT-OSS to OpenAI 格式代理API",
+        "status": "ok",
+        "authentication_method": "使用动态API Key作为Session",
+        "api_key_format": "在'Authorization: Bearer'后填入 '你的user_id|你的session令牌'"
+    })
 
-# 辅助函数：创建一个标准的OpenAI流式数据块
+# 辅助函数：创建标准的OpenAI流式数据块
 def create_openai_chunk(content, model="gpt-oss-120b"):
     return {
-        "id": f"chatcmpl-{str(uuid.uuid4())}",
-        "object": "chat.completion.chunk",
-        "created": int(time.time()),
-        "model": model,
+        "id": f"chatcmpl-{str(uuid.uuid4())}", "object": "chat.completion.chunk",
+        "created": int(time.time()), "model": model,
         "choices": [{"index": 0, "delta": {"content": content}, "finish_reason": None}]
     }
 
-# 4. 核心API端点
+# 4. 核心API端点 (包含新的认证逻辑)
 @app.route('/v1/chat/completions', methods=['POST'])
 def chat_completions_proxy():
+    # --- 关键改动：解析API Key作为Session ---
+    auth_header = request.headers.get('Authorization')
+    if not auth_header or not auth_header.startswith('Bearer '):
+        return jsonify({"error": "缺少或格式错误的Authorization Header。请提供 'Bearer 你的组合密钥'。"}), 401
+
+    combined_key = auth_header.split('Bearer ')[1]
+    if '|' not in combined_key:
+        return jsonify({"error": "API Key格式错误。正确格式为 '你的user_id|你的session令牌'。"}), 401
+
+    try:
+        user_id, session_token = combined_key.split('|', 1)
+    except ValueError:
+        return jsonify({"error": "API Key格式解析失败。请确保格式为 'user_id|session_token'。"}), 401
+
+    # --- 动态构建本次请求的Headers ---
+    request_headers = BASE_GPT_OSS_HEADERS.copy()
+    request_headers['cookie'] = f"user_id={user_id}; session={session_token}"
+    
     try:
         openai_request_data = request.json
         user_prompt = next((m['content'] for m in reversed(openai_request_data.get("messages", [])) if m.get('role') == 'user'), None)
@@ -45,10 +66,8 @@ def chat_completions_proxy():
     except Exception as e:
         return jsonify({"error": f"请求格式无效: {e}"}), 400
 
-    request_headers = GPT_OSS_HEADERS.copy()
     request_headers['x-show-reasoning'] = 'true' if stream_requested else 'false'
         
-    gpt_oss_cookies = {'user_id': str(uuid.uuid4())}
     gpt_oss_payload = {
         "op": "threads.create",
         "params": {"input": {"text": user_prompt, "content": [{"type": "input_text", "text": user_prompt}]}}
@@ -57,7 +76,7 @@ def chat_completions_proxy():
     def _internal_proxy_stream():
         try:
             with requests.post(
-                GPT_OSS_API_URL, headers=request_headers, cookies=gpt_oss_cookies,
+                GPT_OSS_API_URL, headers=request_headers,
                 json=gpt_oss_payload, stream=True, timeout=120
             ) as response:
                 response.raise_for_status()
@@ -69,40 +88,34 @@ def chat_completions_proxy():
         except requests.exceptions.RequestException as e:
             raise IOError(f"与后端服务通信失败: {e}")
 
+    # 流式和非流式处理逻辑保持不变
     if stream_requested:
         def stream_formatter():
+            thinking_buffer = []
+            thinking_block_sent = False
             try:
                 for gpt_oss_data in _internal_proxy_stream():
                     event_type = gpt_oss_data.get('type')
                     update_type = gpt_oss_data.get('update', {}).get('type')
                     
-                    # 关键逻辑: 每收到一条思考，就立即格式化并发送
                     if event_type == 'thread.item_updated' and update_type == 'cot.entry_added':
-                        thought = gpt_oss_data['update']['entry']['content']
-                        # 将单条思考包装在它自己的Markdown代码块中
-                        formatted_thought_block = f"```markdown\n[思考中] {thought}\n```"
-                        
-                        # 1. 发送包含代码块的chunk
-                        thought_chunk = create_openai_chunk(formatted_thought_block)
-                        yield f"data: {json.dumps(thought_chunk)}\n\n"
-                        
-                        # 2. 发送一个包含换行符的chunk，以在视觉上分隔不同的思考块
-                        newline_chunk = create_openai_chunk("\n\n")
-                        yield f"data: {json.dumps(newline_chunk)}\n\n"
-                    
-                    # 正常流式传输最终答案
+                        thinking_buffer.append(f"- {gpt_oss_data['update']['entry']['content']}")
+                        continue
+
                     if event_type == 'thread.item_updated' and update_type == 'assistant_message.content_part.text_delta':
-                        delta_content = gpt_oss_data['update'].get('delta', '')
-                        text_chunk = create_openai_chunk(delta_content)
-                        yield f"data: {json.dumps(text_chunk)}\n\n"
+                        if not thinking_block_sent and thinking_buffer:
+                            all_thoughts = "\n".join(thinking_buffer)
+                            formatted_block = f"```markdown\n[思考过程]\n{all_thoughts}\n```\n\n"
+                            yield f"data: {json.dumps(create_openai_chunk(formatted_block))}\n\n"
+                            thinking_block_sent = True
+
+                        yield f"data: {json.dumps(create_openai_chunk(gpt_oss_data['update'].get('delta', '')))}\n\n"
                 
                 yield "data: [DONE]\n\n"
             except IOError as e:
                  yield f"data: {json.dumps({'error': str(e)})}\n\n"
-
         return Response(stream_formatter(), mimetype='text/event-stream')
-    
-    else: # 非流式请求逻辑保持不变
+    else:
         try:
             full_response_content = ""
             for gpt_oss_data in _internal_proxy_stream():
@@ -113,7 +126,7 @@ def chat_completions_proxy():
                 "id": f"chatcmpl-{str(uuid.uuid4())}", "object": "chat.completion", "created": int(time.time()),
                 "model": "gpt-oss-120b",
                 "choices": [{"index": 0, "message": {"role": "assistant", "content": full_response_content.strip()}, "finish_reason": "stop"}],
-                "usage": {"prompt_tokens": None, "completion_tokens": None, "total_tokens": None}
+                "usage": {}
             }
             return jsonify(final_response)
         except IOError as e: