Spaces:
Runtime error
Runtime error
Commit ·
b93f165
1
Parent(s): fe067bc
修复流式输出错乱问题
Browse files- app/server/chat.py +29 -10
app/server/chat.py
CHANGED
|
@@ -952,10 +952,13 @@ def _create_real_streaming_response(
|
|
| 952 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
| 953 |
has_started = True
|
| 954 |
|
| 955 |
-
|
|
|
|
|
|
|
|
|
|
| 956 |
if not last_chunk_was_thought and not full_thoughts:
|
| 957 |
yield f"data: {orjson.dumps({'id': completion_id, 'object': 'chat.completion.chunk', 'created': created_time, 'model': model_name, 'choices': [{'index': 0, 'delta': {'content': '<think>'}, 'finish_reason': None}]}).decode('utf-8')}\n\n"
|
| 958 |
-
|
| 959 |
data = {
|
| 960 |
"id": completion_id,
|
| 961 |
"object": "chat.completion.chunk",
|
|
@@ -966,14 +969,19 @@ def _create_real_streaming_response(
|
|
| 966 |
],
|
| 967 |
}
|
| 968 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
|
|
|
| 969 |
last_chunk_was_thought = True
|
| 970 |
|
| 971 |
-
|
|
|
|
|
|
|
|
|
|
| 972 |
if last_chunk_was_thought:
|
| 973 |
yield f"data: {orjson.dumps({'id': completion_id, 'object': 'chat.completion.chunk', 'created': created_time, 'model': model_name, 'choices': [{'index': 0, 'delta': {'content': '</think>\n'}, 'finish_reason': None}]}).decode('utf-8')}\n\n"
|
| 974 |
last_chunk_was_thought = False
|
| 975 |
-
|
| 976 |
-
|
|
|
|
| 977 |
data = {
|
| 978 |
"id": completion_id,
|
| 979 |
"object": "chat.completion.chunk",
|
|
@@ -988,6 +996,7 @@ def _create_real_streaming_response(
|
|
| 988 |
],
|
| 989 |
}
|
| 990 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
|
|
|
| 991 |
except Exception as e:
|
| 992 |
logger.exception(f"Error during OpenAI streaming: {e}")
|
| 993 |
yield f"data: {orjson.dumps({'error': {'message': 'Streaming error occurred.', 'type': 'server_error', 'param': None, 'code': None}}).decode('utf-8')}\n\n"
|
|
@@ -1147,19 +1156,29 @@ def _create_responses_real_streaming_response(
|
|
| 1147 |
try:
|
| 1148 |
async for chunk in generator:
|
| 1149 |
all_outputs.append(chunk)
|
| 1150 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1151 |
if not last_chunk_was_thought and not full_thoughts:
|
| 1152 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': '<think>'}).decode('utf-8')}\n\n"
|
| 1153 |
-
full_thoughts += t_delta
|
| 1154 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': t_delta}).decode('utf-8')}\n\n"
|
|
|
|
| 1155 |
last_chunk_was_thought = True
|
| 1156 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1157 |
if last_chunk_was_thought:
|
| 1158 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': '</think>\n'}).decode('utf-8')}\n\n"
|
| 1159 |
last_chunk_was_thought = False
|
| 1160 |
-
|
| 1161 |
-
|
|
|
|
| 1162 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': visible_delta}).decode('utf-8')}\n\n"
|
|
|
|
| 1163 |
except Exception as e:
|
| 1164 |
logger.exception(f"Error during Responses API streaming: {e}")
|
| 1165 |
yield f"data: {orjson.dumps({**base_event, 'type': 'error', 'error': {'message': 'Streaming error.'}}).decode('utf-8')}\n\n"
|
|
|
|
| 952 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
| 953 |
has_started = True
|
| 954 |
|
| 955 |
+
# Robust thought delta calculation
|
| 956 |
+
new_thoughts = chunk.thoughts or ""
|
| 957 |
+
if len(new_thoughts) > len(full_thoughts):
|
| 958 |
+
t_delta = new_thoughts[len(full_thoughts) :]
|
| 959 |
if not last_chunk_was_thought and not full_thoughts:
|
| 960 |
yield f"data: {orjson.dumps({'id': completion_id, 'object': 'chat.completion.chunk', 'created': created_time, 'model': model_name, 'choices': [{'index': 0, 'delta': {'content': '<think>'}, 'finish_reason': None}]}).decode('utf-8')}\n\n"
|
| 961 |
+
|
| 962 |
data = {
|
| 963 |
"id": completion_id,
|
| 964 |
"object": "chat.completion.chunk",
|
|
|
|
| 969 |
],
|
| 970 |
}
|
| 971 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
| 972 |
+
full_thoughts = new_thoughts
|
| 973 |
last_chunk_was_thought = True
|
| 974 |
|
| 975 |
+
# Robust text delta calculation
|
| 976 |
+
new_text = chunk.text or ""
|
| 977 |
+
if len(new_text) > len(full_text):
|
| 978 |
+
text_delta = new_text[len(full_text) :]
|
| 979 |
if last_chunk_was_thought:
|
| 980 |
yield f"data: {orjson.dumps({'id': completion_id, 'object': 'chat.completion.chunk', 'created': created_time, 'model': model_name, 'choices': [{'index': 0, 'delta': {'content': '</think>\n'}, 'finish_reason': None}]}).decode('utf-8')}\n\n"
|
| 981 |
last_chunk_was_thought = False
|
| 982 |
+
|
| 983 |
+
visible_delta = suppressor.process(text_delta)
|
| 984 |
+
if visible_delta:
|
| 985 |
data = {
|
| 986 |
"id": completion_id,
|
| 987 |
"object": "chat.completion.chunk",
|
|
|
|
| 996 |
],
|
| 997 |
}
|
| 998 |
yield f"data: {orjson.dumps(data).decode('utf-8')}\n\n"
|
| 999 |
+
full_text = new_text
|
| 1000 |
except Exception as e:
|
| 1001 |
logger.exception(f"Error during OpenAI streaming: {e}")
|
| 1002 |
yield f"data: {orjson.dumps({'error': {'message': 'Streaming error occurred.', 'type': 'server_error', 'param': None, 'code': None}}).decode('utf-8')}\n\n"
|
|
|
|
| 1156 |
try:
|
| 1157 |
async for chunk in generator:
|
| 1158 |
all_outputs.append(chunk)
|
| 1159 |
+
|
| 1160 |
+
# Robust thought delta calculation
|
| 1161 |
+
new_thoughts = chunk.thoughts or ""
|
| 1162 |
+
if len(new_thoughts) > len(full_thoughts):
|
| 1163 |
+
t_delta = new_thoughts[len(full_thoughts) :]
|
| 1164 |
if not last_chunk_was_thought and not full_thoughts:
|
| 1165 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': '<think>'}).decode('utf-8')}\n\n"
|
|
|
|
| 1166 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': t_delta}).decode('utf-8')}\n\n"
|
| 1167 |
+
full_thoughts = new_thoughts
|
| 1168 |
last_chunk_was_thought = True
|
| 1169 |
+
|
| 1170 |
+
# Robust text delta calculation
|
| 1171 |
+
new_text = chunk.text or ""
|
| 1172 |
+
if len(new_text) > len(full_text):
|
| 1173 |
+
text_delta = new_text[len(full_text) :]
|
| 1174 |
if last_chunk_was_thought:
|
| 1175 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': '</think>\n'}).decode('utf-8')}\n\n"
|
| 1176 |
last_chunk_was_thought = False
|
| 1177 |
+
|
| 1178 |
+
visible_delta = suppressor.process(text_delta)
|
| 1179 |
+
if visible_delta:
|
| 1180 |
yield f"data: {orjson.dumps({**base_event, 'type': 'response.output_text.delta', 'output_index': 0, 'delta': visible_delta}).decode('utf-8')}\n\n"
|
| 1181 |
+
full_text = new_text
|
| 1182 |
except Exception as e:
|
| 1183 |
logger.exception(f"Error during Responses API streaming: {e}")
|
| 1184 |
yield f"data: {orjson.dumps({**base_event, 'type': 'error', 'error': {'message': 'Streaming error.'}}).decode('utf-8')}\n\n"
|