I cannot use stream method using AgentExecutor with memory

[Songjiadong]

Checked other resources

• [X] I added a very descriptive title to this issue.
• [X] I searched the LangChain documentation with the integrated search.
• [X] I used the GitHub search to find a similar question and didn't find it.
• [X] I am sure that this is a bug in LangChain rather than my code.
• [X] The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

agent_executor = AgentExecutor.from_agent_and_tools(
    agent=agent,
    tools=TOOLS,
    memory=chat_memory,
    max_iterations=20,
    handle_parsing_errors=True,
    verbose=verbose
)
agent_executor.stream({"input":"介绍一下自己"})

the function _get_input_output of chat_memory.py has bug. Its output of response has 2propertity,which is output and messages.But this method using restricted condition len(outputs) != 1.

If I dont use memory,the program run normally. Help me,thank you!

Error Message and Stack Trace (if applicable)

Traceback (most recent call last):
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/agents/agent_iterator.py", line 195, in __iter__
    output = self._process_next_step_output(next_step, run_manager)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/agents/agent_iterator.py", line 300, in _process_next_step_output
    return self._return(next_step_output, run_manager=run_manager)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/agents/agent_iterator.py", line 379, in _return
    return self.make_final_outputs(returned_output, run_manager)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/agents/agent_iterator.py", line 142, in make_final_outputs
    self.agent_executor.prep_outputs(
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/chains/base.py", line 455, in prep_outputs
    self.memory.save_context(inputs, outputs)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/memory/summary.py", line 90, in save_context
    super().save_context(inputs, outputs)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/memory/chat_memory.py", line 38, in save_context
    input_str, output_str = self._get_input_output(inputs, outputs)
  File "/usr/local/churchill/lib/python3.10/site-packages/langchain/memory/chat_memory.py", line 30, in _get_input_output
    raise ValueError(f"One output key expected, got {outputs.keys()}")
ValueError: One output key expected, got dict_keys(['output', 'messages'])

Description

agent_executor = AgentExecutor.from_agent_and_tools(
    agent=agent,
    tools=TOOLS,
    memory=chat_memory,
    max_iterations=20,
    handle_parsing_errors=True,
    verbose=verbose
)
agent_executor.stream({"input":"介绍一下自己"})

the function _get_input_output of chat_memory.py has bug. Its output of response has 2propertity,which is output and messages.But this method using restricted condition len(outputs) != 1.

If I dont use memory,the program run normally. Help me,thank you!

System Info

langchain 0.1.9 linux centos 8 python 3.10

[liugddx]

18117