Spaces:
Paused
Paused
Upload proxy_handler.py
Browse files- proxy_handler.py +66 -82
proxy_handler.py
CHANGED
|
@@ -25,21 +25,17 @@ class ProxyHandler:
|
|
| 25 |
)
|
| 26 |
|
| 27 |
async def aclose(self):
|
| 28 |
-
"""Closes the underlying httpx client."""
|
| 29 |
if not self.client.is_closed:
|
| 30 |
await self.client.aclose()
|
| 31 |
|
| 32 |
-
#
|
| 33 |
def _balance_think_tag(self, txt: str) -> str:
|
| 34 |
opens = len(re.findall(r"<think>", txt))
|
| 35 |
closes = len(re.findall(r"</think>", txt))
|
| 36 |
-
if opens > closes:
|
| 37 |
-
txt += "</think>" * (opens - closes)
|
| 38 |
elif closes > opens:
|
| 39 |
-
for _ in range(closes - opens):
|
| 40 |
-
txt = re.sub(r"</think>(?!.*</think>)", "", txt, 1)
|
| 41 |
return txt
|
| 42 |
-
|
| 43 |
def _clean_thinking(self, s: str) -> str:
|
| 44 |
if not s: return s
|
| 45 |
s = re.sub(r'<details[^>]*>.*?</details>', '', s, flags=re.DOTALL)
|
|
@@ -47,11 +43,9 @@ class ProxyHandler:
|
|
| 47 |
s = re.sub(r'<[^>]+>', '', s)
|
| 48 |
s = re.sub(r'^\s*>\s*', '', s, flags=re.MULTILINE)
|
| 49 |
return s.strip()
|
| 50 |
-
|
| 51 |
def _clean_answer(self, s: str) -> str:
|
| 52 |
if not s: return s
|
| 53 |
return re.sub(r"<details[^>]*>.*?</details>", "", s, flags=re.DOTALL)
|
| 54 |
-
|
| 55 |
def _serialize_msgs(self, msgs) -> list:
|
| 56 |
out = []
|
| 57 |
for m in msgs:
|
|
@@ -60,31 +54,15 @@ class ProxyHandler:
|
|
| 60 |
elif isinstance(m, dict): out.append(m)
|
| 61 |
else: out.append({"role": getattr(m, "role", "user"), "content": getattr(m, "content", str(m))})
|
| 62 |
return out
|
| 63 |
-
|
| 64 |
-
# ---------- upstream ----------
|
| 65 |
async def _prep_upstream(self, req: ChatCompletionRequest) -> Tuple[Dict[str, Any], Dict[str, str], str]:
|
| 66 |
ck = await cookie_manager.get_next_cookie()
|
| 67 |
if not ck: raise HTTPException(503, "No available cookies")
|
| 68 |
-
|
| 69 |
model = settings.UPSTREAM_MODEL if req.model == settings.MODEL_NAME else req.model
|
| 70 |
-
body = {
|
| 71 |
-
|
| 72 |
-
"background_tasks": {"title_generation": True, "tags_generation": True}, "chat_id": str(uuid.uuid4()),
|
| 73 |
-
"features": {"image_generation": False, "code_interpreter": False, "web_search": False, "auto_web_search": False, "enable_thinking": True,},
|
| 74 |
-
"id": str(uuid.uuid4()), "mcp_servers": ["deep-web-search"],
|
| 75 |
-
"model_item": {"id": model, "name": "GLM-4.5", "owned_by": "openai"}, "params": {}, "tool_servers": [],
|
| 76 |
-
"variables": {"{{USER_NAME}}": "User", "{{USER_LOCATION}}": "Unknown", "{{CURRENT_DATETIME}}": time.strftime("%Y-%m-%d %H:%M:%S"),},
|
| 77 |
-
}
|
| 78 |
-
headers = {
|
| 79 |
-
"Content-Type": "application/json", "Authorization": f"Bearer {ck}",
|
| 80 |
-
"User-Agent": ("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36"),
|
| 81 |
-
"Accept": "application/json, text/event-stream", "Accept-Language": "zh-CN",
|
| 82 |
-
"sec-ch-ua": '"Not)A;Brand";v="8", "Chromium";v="138", "Google Chrome";v="138"',
|
| 83 |
-
"sec-ch-ua-mobile": "?0", "sec-ch-ua-platform": '"macOS"', "x-fe-version": "prod-fe-1.0.53",
|
| 84 |
-
"Origin": "https://chat.z.ai", "Referer": "https://chat.z.ai/",
|
| 85 |
-
}
|
| 86 |
return body, headers, ck
|
| 87 |
|
|
|
|
| 88 |
# ---------- stream ----------
|
| 89 |
async def stream_proxy_response(self, req: ChatCompletionRequest) -> AsyncGenerator[str, None]:
|
| 90 |
ck = None
|
|
@@ -97,73 +75,69 @@ class ProxyHandler:
|
|
| 97 |
async with self.client.stream("POST", settings.UPSTREAM_URL, json=body, headers=headers) as resp:
|
| 98 |
if resp.status_code != 200:
|
| 99 |
await cookie_manager.mark_cookie_invalid(ck)
|
|
|
|
| 100 |
err_body = await resp.aread()
|
| 101 |
err_msg = f"Error: {resp.status_code} - {err_body.decode(errors='ignore')}"
|
| 102 |
-
err = {
|
| 103 |
-
"id": comp_id, "object": "chat.completion.chunk", "created": int(time.time()), "model": req.model,
|
| 104 |
-
"choices": [{"index": 0, "delta": {"content": err_msg}, "finish_reason": "stop"}],
|
| 105 |
-
}
|
| 106 |
yield f"data: {json.dumps(err)}\n\n"; yield "data: [DONE]\n\n"; return
|
| 107 |
|
| 108 |
async for raw in resp.aiter_text():
|
| 109 |
-
|
| 110 |
-
for line in raw.split('\n'):
|
| 111 |
line = line.strip()
|
| 112 |
if not line or not line.startswith('data: '): continue
|
| 113 |
|
| 114 |
-
|
| 115 |
-
if
|
| 116 |
if think_open:
|
| 117 |
-
|
| 118 |
-
yield f"data: {json.dumps(
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
yield f"data: {json.dumps(final_chunk)}\n\n"
|
| 122 |
yield "data: [DONE]\n\n"; return
|
| 123 |
|
| 124 |
try:
|
| 125 |
-
parsed = json.loads(
|
|
|
|
| 126 |
except json.JSONDecodeError:
|
| 127 |
continue
|
| 128 |
-
|
| 129 |
-
dat = parsed.get("data", {})
|
| 130 |
-
delta, new_phase = dat.get("delta_content", ""), dat.get("phase")
|
| 131 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 132 |
if new_phase and new_phase != phase_cur:
|
| 133 |
-
if phase_cur == "thinking" and new_phase == "answer"
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
'choices': [{'index': 0, 'delta': {'content': '</think>\n'}, 'finish_reason': None}]
|
| 140 |
-
}
|
| 141 |
-
yield f"data: {json.dumps(close_payload)}\n\n"
|
| 142 |
-
# --- FIX END ---
|
| 143 |
-
think_open = False
|
| 144 |
phase_cur = new_phase
|
| 145 |
|
| 146 |
-
|
| 147 |
-
|
| 148 |
text_to_yield = ""
|
| 149 |
-
if
|
| 150 |
if settings.SHOW_THINK_TAGS:
|
| 151 |
if not think_open:
|
| 152 |
-
|
| 153 |
-
yield f"data: {json.dumps(
|
| 154 |
think_open = True
|
| 155 |
text_to_yield = self._clean_thinking(delta)
|
| 156 |
-
elif phase_cur == "answer":
|
| 157 |
-
text_to_yield = self._clean_answer(delta)
|
| 158 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
if text_to_yield:
|
| 160 |
-
|
| 161 |
-
|
| 162 |
-
|
| 163 |
-
}
|
| 164 |
-
yield f"data: {json.dumps(out_chunk)}\n\n"
|
| 165 |
-
|
| 166 |
except httpx.RequestError as e:
|
|
|
|
| 167 |
if ck: await cookie_manager.mark_cookie_invalid(ck)
|
| 168 |
logger.error(f"Request error: {e}")
|
| 169 |
err_msg = f"Connection error: {e}"
|
|
@@ -174,13 +148,13 @@ class ProxyHandler:
|
|
| 174 |
err = {"id": f"chatcmpl-{uuid.uuid4().hex[:29]}", "choices": [{"delta": {"content": f"Internal error in stream"}, "finish_reason": "stop"}]}
|
| 175 |
yield f"data: {json.dumps(err)}\n\n"; yield "data: [DONE]\n\n"
|
| 176 |
|
|
|
|
| 177 |
# ---------- non-stream ----------
|
| 178 |
async def non_stream_proxy_response(self, req: ChatCompletionRequest) -> ChatCompletionResponse:
|
| 179 |
ck = None
|
| 180 |
try:
|
| 181 |
body, headers, ck = await self._prep_upstream(req)
|
| 182 |
think_buf, answer_buf = [], []
|
| 183 |
-
|
| 184 |
phase_cur = None
|
| 185 |
|
| 186 |
async with self.client.stream("POST", settings.UPSTREAM_URL, json=body, headers=headers) as resp:
|
|
@@ -190,30 +164,38 @@ class ProxyHandler:
|
|
| 190 |
raise HTTPException(resp.status_code, f"Upstream error: {error_detail}")
|
| 191 |
|
| 192 |
async for raw in resp.aiter_text():
|
| 193 |
-
|
| 194 |
-
for line in raw.split('\n'):
|
| 195 |
line = line.strip()
|
| 196 |
if not line or not line.startswith('data: '): continue
|
| 197 |
-
payload = line[6:]
|
| 198 |
-
if payload == '[DONE]': break
|
| 199 |
|
| 200 |
-
|
| 201 |
-
|
| 202 |
|
| 203 |
-
|
| 204 |
-
|
|
|
|
|
|
|
|
|
|
| 205 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 206 |
if new_phase:
|
| 207 |
phase_cur = new_phase
|
| 208 |
|
|
|
|
| 209 |
if not delta: continue
|
| 210 |
|
| 211 |
-
if phase_cur == "thinking":
|
| 212 |
think_buf.append(delta)
|
| 213 |
-
elif phase_cur == "answer":
|
| 214 |
answer_buf.append(delta)
|
| 215 |
-
|
| 216 |
-
|
|
|
|
|
|
|
| 217 |
|
| 218 |
raw_answer = ''.join(answer_buf)
|
| 219 |
ans_text = self._clean_answer(raw_answer)
|
|
@@ -230,6 +212,7 @@ class ProxyHandler:
|
|
| 230 |
choices=[{"index": 0, "message": {"role": "assistant", "content": final_content}, "finish_reason": "stop"}],
|
| 231 |
)
|
| 232 |
except httpx.RequestError as e:
|
|
|
|
| 233 |
if ck: await cookie_manager.mark_cookie_invalid(ck)
|
| 234 |
logger.error(f"Non-stream request error: {e}")
|
| 235 |
raise HTTPException(502, f"Connection error: {e}")
|
|
@@ -237,6 +220,7 @@ class ProxyHandler:
|
|
| 237 |
logger.exception(f"Non-stream unexpected error")
|
| 238 |
raise HTTPException(500, "Internal server error")
|
| 239 |
|
|
|
|
| 240 |
# ---------- FastAPI entry ----------
|
| 241 |
async def handle_chat_completion(self, req: ChatCompletionRequest):
|
| 242 |
stream = bool(req.stream) if req.stream is not None else settings.DEFAULT_STREAM
|
|
|
|
| 25 |
)
|
| 26 |
|
| 27 |
async def aclose(self):
|
|
|
|
| 28 |
if not self.client.is_closed:
|
| 29 |
await self.client.aclose()
|
| 30 |
|
| 31 |
+
# _balance_think_tag, _clean_thinking, _clean_answer, _serialize_msgs, _prep_upstream 保持不變
|
| 32 |
def _balance_think_tag(self, txt: str) -> str:
|
| 33 |
opens = len(re.findall(r"<think>", txt))
|
| 34 |
closes = len(re.findall(r"</think>", txt))
|
| 35 |
+
if opens > closes: txt += "</think>" * (opens - closes)
|
|
|
|
| 36 |
elif closes > opens:
|
| 37 |
+
for _ in range(closes - opens): txt = re.sub(r"</think>(?!.*</think>)", "", txt, 1)
|
|
|
|
| 38 |
return txt
|
|
|
|
| 39 |
def _clean_thinking(self, s: str) -> str:
|
| 40 |
if not s: return s
|
| 41 |
s = re.sub(r'<details[^>]*>.*?</details>', '', s, flags=re.DOTALL)
|
|
|
|
| 43 |
s = re.sub(r'<[^>]+>', '', s)
|
| 44 |
s = re.sub(r'^\s*>\s*', '', s, flags=re.MULTILINE)
|
| 45 |
return s.strip()
|
|
|
|
| 46 |
def _clean_answer(self, s: str) -> str:
|
| 47 |
if not s: return s
|
| 48 |
return re.sub(r"<details[^>]*>.*?</details>", "", s, flags=re.DOTALL)
|
|
|
|
| 49 |
def _serialize_msgs(self, msgs) -> list:
|
| 50 |
out = []
|
| 51 |
for m in msgs:
|
|
|
|
| 54 |
elif isinstance(m, dict): out.append(m)
|
| 55 |
else: out.append({"role": getattr(m, "role", "user"), "content": getattr(m, "content", str(m))})
|
| 56 |
return out
|
|
|
|
|
|
|
| 57 |
async def _prep_upstream(self, req: ChatCompletionRequest) -> Tuple[Dict[str, Any], Dict[str, str], str]:
|
| 58 |
ck = await cookie_manager.get_next_cookie()
|
| 59 |
if not ck: raise HTTPException(503, "No available cookies")
|
|
|
|
| 60 |
model = settings.UPSTREAM_MODEL if req.model == settings.MODEL_NAME else req.model
|
| 61 |
+
body = { "stream": True, "model": model, "messages": self._serialize_msgs(req.messages), "background_tasks": {"title_generation": True, "tags_generation": True}, "chat_id": str(uuid.uuid4()), "features": {"image_generation": False, "code_interpreter": False, "web_search": False, "auto_web_search": False, "enable_thinking": True,}, "id": str(uuid.uuid4()), "mcp_servers": ["deep-web-search"], "model_item": {"id": model, "name": "GLM-4.5", "owned_by": "openai"}, "params": {}, "tool_servers": [], "variables": {"{{USER_NAME}}": "User", "{{USER_LOCATION}}": "Unknown", "{{CURRENT_DATETIME}}": time.strftime("%Y-%m-%d %H:%M:%S"),},}
|
| 62 |
+
headers = { "Content-Type": "application/json", "Authorization": f"Bearer {ck}", "User-Agent": ("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36"), "Accept": "application/json, text/event-stream", "Accept-Language": "zh-CN", "sec-ch-ua": '"Not)A;Brand";v="8", "Chromium";v="138", "Google Chrome";v="138"', "sec-ch-ua-mobile": "?0", "sec-ch-ua-platform": '"macOS"', "x-fe-version": "prod-fe-1.0.53", "Origin": "https://chat.z.ai", "Referer": "https://chat.z.ai/",}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
return body, headers, ck
|
| 64 |
|
| 65 |
+
|
| 66 |
# ---------- stream ----------
|
| 67 |
async def stream_proxy_response(self, req: ChatCompletionRequest) -> AsyncGenerator[str, None]:
|
| 68 |
ck = None
|
|
|
|
| 75 |
async with self.client.stream("POST", settings.UPSTREAM_URL, json=body, headers=headers) as resp:
|
| 76 |
if resp.status_code != 200:
|
| 77 |
await cookie_manager.mark_cookie_invalid(ck)
|
| 78 |
+
# ... (error handling code remains the same)
|
| 79 |
err_body = await resp.aread()
|
| 80 |
err_msg = f"Error: {resp.status_code} - {err_body.decode(errors='ignore')}"
|
| 81 |
+
err = {"id": comp_id, "object": "chat.completion.chunk", "created": int(time.time()), "model": req.model, "choices": [{"index": 0, "delta": {"content": err_msg}, "finish_reason": "stop"}],}
|
|
|
|
|
|
|
|
|
|
| 82 |
yield f"data: {json.dumps(err)}\n\n"; yield "data: [DONE]\n\n"; return
|
| 83 |
|
| 84 |
async for raw in resp.aiter_text():
|
| 85 |
+
for line in raw.strip().split('\n'):
|
|
|
|
| 86 |
line = line.strip()
|
| 87 |
if not line or not line.startswith('data: '): continue
|
| 88 |
|
| 89 |
+
payload_str = line[6:]
|
| 90 |
+
if payload_str == '[DONE]':
|
| 91 |
if think_open:
|
| 92 |
+
close_payload = {'id': comp_id, 'object': 'chat.completion.chunk', 'created': int(time.time()), 'model': req.model, 'choices': [{'index': 0, 'delta': {'content': '</think>'}, 'finish_reason': None}]}
|
| 93 |
+
yield f"data: {json.dumps(close_payload)}\n\n"
|
| 94 |
+
final_payload = {'id': comp_id, 'object': 'chat.completion.chunk', 'created': int(time.time()), 'model': req.model, 'choices': [{'index': 0, 'delta': {}, 'finish_reason': 'stop'}]}
|
| 95 |
+
yield f"data: {json.dumps(final_payload)}\n\n"
|
|
|
|
| 96 |
yield "data: [DONE]\n\n"; return
|
| 97 |
|
| 98 |
try:
|
| 99 |
+
parsed = json.loads(payload_str)
|
| 100 |
+
dat = parsed.get("data", {})
|
| 101 |
except json.JSONDecodeError:
|
| 102 |
continue
|
|
|
|
|
|
|
|
|
|
| 103 |
|
| 104 |
+
# --- FIX START: REVISED LOGIC ---
|
| 105 |
+
delta = dat.get("delta_content", "")
|
| 106 |
+
new_phase = dat.get("phase")
|
| 107 |
+
|
| 108 |
+
# Step 1: Determine the effective phase for THIS chunk
|
| 109 |
+
effective_phase = new_phase or phase_cur
|
| 110 |
+
|
| 111 |
+
# Step 2: Handle phase transition side-effects (closing tags) BEFORE processing content
|
| 112 |
if new_phase and new_phase != phase_cur:
|
| 113 |
+
if phase_cur == "thinking" and new_phase == "answer":
|
| 114 |
+
if think_open and settings.SHOW_THINK_TAGS:
|
| 115 |
+
close_payload = {'id': comp_id, 'object': 'chat.completion.chunk', 'created': int(time.time()), 'model': req.model, 'choices': [{'index': 0, 'delta': {'content': '</think>\n'}, 'finish_reason': None}]}
|
| 116 |
+
yield f"data: {json.dumps(close_payload)}\n\n"
|
| 117 |
+
think_open = False
|
| 118 |
+
# Update phase_cur AFTER handling the transition
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
phase_cur = new_phase
|
| 120 |
|
| 121 |
+
# Step 3: Process the delta content of THIS chunk based on its effective phase
|
|
|
|
| 122 |
text_to_yield = ""
|
| 123 |
+
if effective_phase == "thinking":
|
| 124 |
if settings.SHOW_THINK_TAGS:
|
| 125 |
if not think_open:
|
| 126 |
+
open_payload = {'id': comp_id, 'object': 'chat.completion.chunk', 'created': int(time.time()), 'model': req.model, 'choices': [{'index': 0, 'delta': {'content': '<think>'}, 'finish_reason': None}]}
|
| 127 |
+
yield f"data: {json.dumps(open_payload)}\n\n"
|
| 128 |
think_open = True
|
| 129 |
text_to_yield = self._clean_thinking(delta)
|
|
|
|
|
|
|
| 130 |
|
| 131 |
+
elif effective_phase == "answer":
|
| 132 |
+
text_to_yield = self._clean_answer(delta)
|
| 133 |
+
|
| 134 |
+
# Step 4: Yield the processed content if it exists
|
| 135 |
if text_to_yield:
|
| 136 |
+
chunk_payload = {"id": comp_id, "object": "chat.completion.chunk", "created": int(time.time()), "model": req.model, "choices": [{"index": 0, "delta": {"content": text_to_yield}, "finish_reason": None}],}
|
| 137 |
+
yield f"data: {json.dumps(chunk_payload)}\n\n"
|
| 138 |
+
# --- FIX END ---
|
|
|
|
|
|
|
|
|
|
| 139 |
except httpx.RequestError as e:
|
| 140 |
+
# ... (error handling code remains the same)
|
| 141 |
if ck: await cookie_manager.mark_cookie_invalid(ck)
|
| 142 |
logger.error(f"Request error: {e}")
|
| 143 |
err_msg = f"Connection error: {e}"
|
|
|
|
| 148 |
err = {"id": f"chatcmpl-{uuid.uuid4().hex[:29]}", "choices": [{"delta": {"content": f"Internal error in stream"}, "finish_reason": "stop"}]}
|
| 149 |
yield f"data: {json.dumps(err)}\n\n"; yield "data: [DONE]\n\n"
|
| 150 |
|
| 151 |
+
|
| 152 |
# ---------- non-stream ----------
|
| 153 |
async def non_stream_proxy_response(self, req: ChatCompletionRequest) -> ChatCompletionResponse:
|
| 154 |
ck = None
|
| 155 |
try:
|
| 156 |
body, headers, ck = await self._prep_upstream(req)
|
| 157 |
think_buf, answer_buf = [], []
|
|
|
|
| 158 |
phase_cur = None
|
| 159 |
|
| 160 |
async with self.client.stream("POST", settings.UPSTREAM_URL, json=body, headers=headers) as resp:
|
|
|
|
| 164 |
raise HTTPException(resp.status_code, f"Upstream error: {error_detail}")
|
| 165 |
|
| 166 |
async for raw in resp.aiter_text():
|
| 167 |
+
for line in raw.strip().split('\n'):
|
|
|
|
| 168 |
line = line.strip()
|
| 169 |
if not line or not line.startswith('data: '): continue
|
|
|
|
|
|
|
| 170 |
|
| 171 |
+
payload_str = line[6:]
|
| 172 |
+
if payload_str == '[DONE]': break
|
| 173 |
|
| 174 |
+
try:
|
| 175 |
+
parsed = json.loads(payload_str)
|
| 176 |
+
dat = parsed.get("data", {})
|
| 177 |
+
except json.JSONDecodeError:
|
| 178 |
+
continue
|
| 179 |
|
| 180 |
+
# --- FIX START: REVISED LOGIC ---
|
| 181 |
+
delta = dat.get("delta_content", "")
|
| 182 |
+
new_phase = dat.get("phase")
|
| 183 |
+
|
| 184 |
+
# Step 1: Update the current phase if a new one is provided
|
| 185 |
if new_phase:
|
| 186 |
phase_cur = new_phase
|
| 187 |
|
| 188 |
+
# Step 2: Append the delta to the correct buffer based on the updated phase
|
| 189 |
if not delta: continue
|
| 190 |
|
| 191 |
+
if phase_cur == "thinking":
|
| 192 |
think_buf.append(delta)
|
| 193 |
+
elif phase_cur == "answer":
|
| 194 |
answer_buf.append(delta)
|
| 195 |
+
# --- FIX END ---
|
| 196 |
+
else: # continue for inner loop
|
| 197 |
+
continue
|
| 198 |
+
break # break for outer loop
|
| 199 |
|
| 200 |
raw_answer = ''.join(answer_buf)
|
| 201 |
ans_text = self._clean_answer(raw_answer)
|
|
|
|
| 212 |
choices=[{"index": 0, "message": {"role": "assistant", "content": final_content}, "finish_reason": "stop"}],
|
| 213 |
)
|
| 214 |
except httpx.RequestError as e:
|
| 215 |
+
# ... (error handling code remains the same)
|
| 216 |
if ck: await cookie_manager.mark_cookie_invalid(ck)
|
| 217 |
logger.error(f"Non-stream request error: {e}")
|
| 218 |
raise HTTPException(502, f"Connection error: {e}")
|
|
|
|
| 220 |
logger.exception(f"Non-stream unexpected error")
|
| 221 |
raise HTTPException(500, "Internal server error")
|
| 222 |
|
| 223 |
+
|
| 224 |
# ---------- FastAPI entry ----------
|
| 225 |
async def handle_chat_completion(self, req: ChatCompletionRequest):
|
| 226 |
stream = bool(req.stream) if req.stream is not None else settings.DEFAULT_STREAM
|