1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
|
import asyncio
import json
from datetime import datetime
from typing import Optional
import httpx
from litellm.constants import STREAM_SSE_DONE_STRING
from litellm.litellm_core_utils.asyncify import run_async_function
from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
from litellm.litellm_core_utils.thread_pool_executor import executor
from litellm.llms.base_llm.responses.transformation import BaseResponsesAPIConfig
from litellm.types.llms.openai import (
ResponseCompletedEvent,
ResponsesAPIStreamEvents,
ResponsesAPIStreamingResponse,
)
from litellm.utils import CustomStreamWrapper
class BaseResponsesAPIStreamingIterator:
"""
Base class for streaming iterators that process responses from the Responses API.
This class contains shared logic for both synchronous and asynchronous iterators.
"""
def __init__(
self,
response: httpx.Response,
model: str,
responses_api_provider_config: BaseResponsesAPIConfig,
logging_obj: LiteLLMLoggingObj,
):
self.response = response
self.model = model
self.logging_obj = logging_obj
self.finished = False
self.responses_api_provider_config = responses_api_provider_config
self.completed_response: Optional[ResponsesAPIStreamingResponse] = None
self.start_time = datetime.now()
def _process_chunk(self, chunk):
"""Process a single chunk of data from the stream"""
if not chunk:
return None
# Handle SSE format (data: {...})
chunk = CustomStreamWrapper._strip_sse_data_from_chunk(chunk)
if chunk is None:
return None
# Handle "[DONE]" marker
if chunk == STREAM_SSE_DONE_STRING:
self.finished = True
return None
try:
# Parse the JSON chunk
parsed_chunk = json.loads(chunk)
# Format as ResponsesAPIStreamingResponse
if isinstance(parsed_chunk, dict):
openai_responses_api_chunk = (
self.responses_api_provider_config.transform_streaming_response(
model=self.model,
parsed_chunk=parsed_chunk,
logging_obj=self.logging_obj,
)
)
# Store the completed response
if (
openai_responses_api_chunk
and openai_responses_api_chunk.type
== ResponsesAPIStreamEvents.RESPONSE_COMPLETED
):
self.completed_response = openai_responses_api_chunk
self._handle_logging_completed_response()
return openai_responses_api_chunk
return None
except json.JSONDecodeError:
# If we can't parse the chunk, continue
return None
def _handle_logging_completed_response(self):
"""Base implementation - should be overridden by subclasses"""
pass
class ResponsesAPIStreamingIterator(BaseResponsesAPIStreamingIterator):
"""
Async iterator for processing streaming responses from the Responses API.
"""
def __init__(
self,
response: httpx.Response,
model: str,
responses_api_provider_config: BaseResponsesAPIConfig,
logging_obj: LiteLLMLoggingObj,
):
super().__init__(response, model, responses_api_provider_config, logging_obj)
self.stream_iterator = response.aiter_lines()
def __aiter__(self):
return self
async def __anext__(self) -> ResponsesAPIStreamingResponse:
try:
while True:
# Get the next chunk from the stream
try:
chunk = await self.stream_iterator.__anext__()
except StopAsyncIteration:
self.finished = True
raise StopAsyncIteration
result = self._process_chunk(chunk)
if self.finished:
raise StopAsyncIteration
elif result is not None:
return result
# If result is None, continue the loop to get the next chunk
except httpx.HTTPError as e:
# Handle HTTP errors
self.finished = True
raise e
def _handle_logging_completed_response(self):
"""Handle logging for completed responses in async context"""
asyncio.create_task(
self.logging_obj.async_success_handler(
result=self.completed_response,
start_time=self.start_time,
end_time=datetime.now(),
cache_hit=None,
)
)
executor.submit(
self.logging_obj.success_handler,
result=self.completed_response,
cache_hit=None,
start_time=self.start_time,
end_time=datetime.now(),
)
class SyncResponsesAPIStreamingIterator(BaseResponsesAPIStreamingIterator):
"""
Synchronous iterator for processing streaming responses from the Responses API.
"""
def __init__(
self,
response: httpx.Response,
model: str,
responses_api_provider_config: BaseResponsesAPIConfig,
logging_obj: LiteLLMLoggingObj,
):
super().__init__(response, model, responses_api_provider_config, logging_obj)
self.stream_iterator = response.iter_lines()
def __iter__(self):
return self
def __next__(self):
try:
while True:
# Get the next chunk from the stream
try:
chunk = next(self.stream_iterator)
except StopIteration:
self.finished = True
raise StopIteration
result = self._process_chunk(chunk)
if self.finished:
raise StopIteration
elif result is not None:
return result
# If result is None, continue the loop to get the next chunk
except httpx.HTTPError as e:
# Handle HTTP errors
self.finished = True
raise e
def _handle_logging_completed_response(self):
"""Handle logging for completed responses in sync context"""
run_async_function(
async_function=self.logging_obj.async_success_handler,
result=self.completed_response,
start_time=self.start_time,
end_time=datetime.now(),
cache_hit=None,
)
executor.submit(
self.logging_obj.success_handler,
result=self.completed_response,
cache_hit=None,
start_time=self.start_time,
end_time=datetime.now(),
)
class MockResponsesAPIStreamingIterator(BaseResponsesAPIStreamingIterator):
"""
mock iterator - some models like o1-pro do not support streaming, we need to fake a stream
"""
def __init__(
self,
response: httpx.Response,
model: str,
responses_api_provider_config: BaseResponsesAPIConfig,
logging_obj: LiteLLMLoggingObj,
):
self.raw_http_response = response
super().__init__(
response=response,
model=model,
responses_api_provider_config=responses_api_provider_config,
logging_obj=logging_obj,
)
self.is_done = False
def __aiter__(self):
return self
async def __anext__(self) -> ResponsesAPIStreamingResponse:
if self.is_done:
raise StopAsyncIteration
self.is_done = True
transformed_response = (
self.responses_api_provider_config.transform_response_api_response(
model=self.model,
raw_response=self.raw_http_response,
logging_obj=self.logging_obj,
)
)
return ResponseCompletedEvent(
type=ResponsesAPIStreamEvents.RESPONSE_COMPLETED,
response=transformed_response,
)
def __iter__(self):
return self
def __next__(self) -> ResponsesAPIStreamingResponse:
if self.is_done:
raise StopIteration
self.is_done = True
transformed_response = (
self.responses_api_provider_config.transform_response_api_response(
model=self.model,
raw_response=self.raw_http_response,
logging_obj=self.logging_obj,
)
)
return ResponseCompletedEvent(
type=ResponsesAPIStreamEvents.RESPONSE_COMPLETED,
response=transformed_response,
)
|
