Update app.py

app.py

@@ -379,38 +379,6 @@ def billing_usage():
         "total_usage": 0
     })
 
-@app.route('/handsome/v1/dashboard/billing/subscription', methods=['GET'])
-def billing_subscription():
-    if not check_authorization(request):
-        return jsonify({"error": "Unauthorized"}), 401
-
-    total_balance = get_billing_info()
-
-    return jsonify({
-        "object": "billing_subscription",
-        "has_payment_method": False,
-        "canceled": False,
-        "canceled_at": None,
-        "delinquent": None,
-        "access_until": int(datetime(9999, 12, 31).timestamp()),
-        "soft_limit": 0,
-        "hard_limit": total_balance,
-        "system_hard_limit": total_balance,
-        "soft_limit_usd": 0,
-        "hard_limit_usd": total_balance,
-        "system_hard_limit_usd": total_balance,
-        "plan": {
-            "name": "SiliconFlow API",
-            "id": "siliconflow-api"
-        },
-        "account_name": "SiliconFlow User",
-        "po_number": None,
-        "billing_email": None,
-        "tax_ids": [],
-        "billing_address": None,
-        "business_address": None
-    })
-
 @app.route('/handsome/v1/chat/completions', methods=['POST'])
 def handsome_chat_completions():
     if not check_authorization(request):
@@ -463,7 +431,7 @@ def handsome_chat_completions():
             def generate():
                 first_chunk_time = None
                 full_response_content = ""
-                pending_reasoning_lines = []  # Store incomplete reasoning lines
+                accumulated_reasoning = ""  # Accumulate reasoning content
                 for chunk in response.iter_content(chunk_size=1024):
                     if chunk:
                         if first_chunk_time is None:
@@ -497,20 +465,18 @@ def handsome_chat_completions():
                                     "usage"
                                 ]["completion_tokens"]
 
-                            # Improved special handling for deepseek-reasoner in streaming mode
+                            # Improved handling for deepseek-reasoner in streaming mode
                             if model_name == "deepseek-reasoner" and "choices" in response_json and len(response_json["choices"]) > 0:
                                 delta = response_json["choices"][0].get("delta", {})
-                                if "reasoning_content" in delta and delta["reasoning_content"]:
-                                    pending_reasoning_lines.extend(delta["reasoning_content"].splitlines(keepends=True))
+                                if "reasoning_content" in delta:
+                                    accumulated_reasoning += delta["reasoning_content"]
                                 if "content" in delta and delta["content"]:
-                                    # Process any pending reasoning lines before the content
-                                    if pending_reasoning_lines:
-                                        for reasoning_line in pending_reasoning_lines:
-                                            if reasoning_line.endswith("\n"):
-                                                response_content += f"> {reasoning_line}"
-                                            else:
-                                                response_content += f"> {reasoning_line}\n"
-                                        pending_reasoning_lines = []  # Clear pending lines
+                                    # Prepend accumulated reasoning before content
+                                    if accumulated_reasoning:
+                                        reasoning_lines = accumulated_reasoning.splitlines()
+                                        formatted_reasoning = "\n".join(f"> {line}" for line in reasoning_lines)
+                                        response_content += formatted_reasoning + "\n"
+                                        accumulated_reasoning = ""  # Reset
                                     response_content += delta["content"]
                             elif "choices" in response_json and len(response_json["choices"]) > 0:
                                 delta = response_json["choices"][0].get("delta", {})
@@ -534,13 +500,6 @@ def handsome_chat_completions():
                                 f"解析流式响应单行 JSON 失败: {e}, "
                                 f"行内容: {line}"
                             )
-                
-                # Process any remaining reasoning lines after all chunks are received
-                if pending_reasoning_lines:
-                    for reasoning_line in pending_reasoning_lines:
-                        response_content += f"> {reasoning_line}"
-                    if not response_content.endswith("\n"):
-                        response_content += "\n"
 
                 user_content = ""
                 messages = data.get("messages", [])