From 16a7e0ce8fe83402b4ed00d8b02ae2475f6cd906 Mon Sep 17 00:00:00 2001 From: Chesars Date: Wed, 26 Nov 2025 17:01:23 -0300 Subject: [PATCH] fix: filter empty SSE lines in BaseModelResponseIterator to prevent extra empty chunks When streaming with stream_options={"include_usage": True}, xAI and other providers using BaseLLMHTTPHandler were returning an extra empty chunk after the usage chunk. This was caused by empty SSE lines (separators between events) being processed as empty GenericStreamingChunks. The fix adds a loop in __next__ and __anext__ to skip empty lines before processing, ensuring only meaningful SSE data events are converted to chunks. Fixes #17136 --- litellm/llms/base_llm/base_model_iterator.py | 89 +++++++++++--------- 1 file changed, 50 insertions(+), 39 deletions(-) diff --git a/litellm/llms/base_llm/base_model_iterator.py b/litellm/llms/base_llm/base_model_iterator.py index 6953b1c5878..62cd503a89e 100644 --- a/litellm/llms/base_llm/base_model_iterator.py +++ b/litellm/llms/base_llm/base_model_iterator.py @@ -125,26 +125,32 @@ class BaseModelResponseIterator: ) def __next__(self): - try: - chunk = self.response_iterator.__next__() - except StopIteration: - raise StopIteration - except ValueError as e: - raise RuntimeError(f"Error receiving chunk from stream: {e}") + while True: + try: + chunk = self.response_iterator.__next__() + except StopIteration: + raise StopIteration + except ValueError as e: + raise RuntimeError(f"Error receiving chunk from stream: {e}") - try: - str_line = chunk - if isinstance(chunk, bytes): # Handle binary data - str_line = chunk.decode("utf-8") # Convert bytes to string - index = str_line.find("data:") - if index != -1: - str_line = str_line[index:] - # chunk is a str at this point - return self._handle_string_chunk(str_line=str_line) - except StopIteration: - raise StopIteration - except ValueError as e: - raise RuntimeError(f"Error parsing chunk: {e},\nReceived chunk: {chunk}") + try: + str_line = chunk + if isinstance(chunk, bytes): # Handle binary data + str_line = chunk.decode("utf-8") # Convert bytes to string + index = str_line.find("data:") + if index != -1: + str_line = str_line[index:] + + # Skip empty lines (common in SSE streams between events) + if not str_line or not str_line.strip(): + continue + + # chunk is a str at this point + return self._handle_string_chunk(str_line=str_line) + except StopIteration: + raise StopIteration + except ValueError as e: + raise RuntimeError(f"Error parsing chunk: {e},\nReceived chunk: {chunk}") # Async iterator def __aiter__(self): @@ -152,30 +158,35 @@ class BaseModelResponseIterator: return self async def __anext__(self): - try: - chunk = await self.async_response_iterator.__anext__() + while True: + try: + chunk = await self.async_response_iterator.__anext__() - except StopAsyncIteration: - raise StopAsyncIteration - except ValueError as e: - raise RuntimeError(f"Error receiving chunk from stream: {e}") + except StopAsyncIteration: + raise StopAsyncIteration + except ValueError as e: + raise RuntimeError(f"Error receiving chunk from stream: {e}") - try: - str_line = chunk - if isinstance(chunk, bytes): # Handle binary data - str_line = chunk.decode("utf-8") # Convert bytes to string - index = str_line.find("data:") - if index != -1: - str_line = str_line[index:] + try: + str_line = chunk + if isinstance(chunk, bytes): # Handle binary data + str_line = chunk.decode("utf-8") # Convert bytes to string + index = str_line.find("data:") + if index != -1: + str_line = str_line[index:] - # chunk is a str at this point - chunk = self._handle_string_chunk(str_line=str_line) + # Skip empty lines (common in SSE streams between events) + if not str_line or not str_line.strip(): + continue - return chunk - except StopAsyncIteration: - raise StopAsyncIteration - except ValueError as e: - raise RuntimeError(f"Error parsing chunk: {e},\nReceived chunk: {chunk}") + # chunk is a str at this point + chunk = self._handle_string_chunk(str_line=str_line) + + return chunk + except StopAsyncIteration: + raise StopAsyncIteration + except ValueError as e: + raise RuntimeError(f"Error parsing chunk: {e},\nReceived chunk: {chunk}") class MockResponseIterator: # for returning ai21 streaming responses