Skip to content

A programming framework for agentic AI

Recommit from my repo - Enhanced speaker name selection during agent name matching. (#2267)

Mark Sze89f3cfd1ad0f
diff --git a/autogen/agentchat/groupchat.py b/autogen/agentchat/groupchat.py
index 05ac06e29..9990f8067 100644
--- a/autogen/agentchat/groupchat.py
+++ b/autogen/agentchat/groupchat.py
@@ -512,6 +512,10 @@ class GroupChat:
def _mentioned_agents(self, message_content: Union[str, List], agents: Optional[List[Agent]]) -> Dict:
"""Counts the number of times each agent is mentioned in the provided message content.
+ Agent names will match under any of the following conditions (all case-sensitive):
+ - Exact name match
+ - If the agent name has underscores it will match with spaces instead (e.g. 'Story_writer' == 'Story writer')
+ - If the agent name has underscores it will match with '\\_' instead of '_' (e.g. 'Story_writer' == 'Story\\_writer')
Args:
message_content (Union[str, List]): The content of the message, either as a single string or a list of strings.
@@ -530,9 +534,17 @@ class GroupChat:
mentions = dict()
for agent in agents:
+ # Finds agent mentions, taking word boundaries into account,
+ # accommodates escaping underscores and underscores as spaces
regex = (
- r"(?<=\W)" + re.escape(agent.name) + r"(?=\W)"
- ) # Finds agent mentions, taking word boundaries into account
+ r"(?<=\W)("
+ + re.escape(agent.name)
+ + r"|"
+ + re.escape(agent.name.replace("_", " "))
+ + r"|"
+ + re.escape(agent.name.replace("_", r"\_"))
+ + r")(?=\W)"
+ )
count = len(re.findall(regex, f" {message_content} ")) # Pad the message to help with matching
if count > 0:
mentions[agent.name] = count
diff --git a/test/agentchat/test_groupchat.py b/test/agentchat/test_groupchat.py
index 8a1e96d07..8a4758d2d 100755
--- a/test/agentchat/test_groupchat.py
+++ b/test/agentchat/test_groupchat.py
@@ -1332,6 +1332,100 @@ def test_select_speaker_message_and_prompt_templates():
)
+def test_speaker_selection_agent_name_match():
+ """
+ In this test a group chat, with auto speaker selection, the speaker name match
+ function is tested against the extended name match regex.
+ """
+
+ user_proxy = autogen.UserProxyAgent(
+ name="User_proxy",
+ system_message="A human admin.",
+ code_execution_config=False,
+ human_input_mode="NEVER",
+ )
+ storywriter = autogen.AssistantAgent(
+ name="Story_writer",
+ system_message="An ideas person.",
+ llm_config=None,
+ )
+ pm = autogen.AssistantAgent(
+ name="Product_manager",
+ system_message="Great in evaluating story ideas.",
+ llm_config=None,
+ )
+
+ all_agents = [user_proxy, storywriter, pm]
+ groupchat = autogen.GroupChat(agents=all_agents, messages=[], max_round=8, speaker_selection_method="auto")
+
+ # Test exact match (unchanged outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="Story_writer")
+ assert result == {"Story_writer": 1}
+
+ # Test match with extra text (unchanged outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents,
+ message_content="' Story_writer.\n\nHere are three story ideas for Grade 3 kids:\n\n1. **The Adventure of the Magic Garden:** A you...",
+ )
+ assert result == {"Story_writer": 1}
+
+ # Test match with escaped underscore (new outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="Story\\_writer")
+ assert result == {"Story_writer": 1}
+
+ # Test match with space (new outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="Story writer")
+ assert result == {"Story_writer": 1}
+
+ # Test match with different casing (unchanged outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="Story_Writer")
+ assert result == {}
+
+ # Test match with invalid name (unchanged outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="NoName_Person")
+ assert result == {}
+
+ # Test match with no name (unchanged outcome)
+ result = groupchat._mentioned_agents(agents=all_agents, message_content="")
+ assert result == {}
+
+ # Test match with multiple agents and exact matches (unchanged outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story_writer will follow the Product_manager."
+ )
+ assert result == {"Story_writer": 1, "Product_manager": 1}
+
+ # Test match with multiple agents and escaped underscores (new outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story\\_writer will follow the Product\\_manager."
+ )
+ assert result == {"Story_writer": 1, "Product_manager": 1}
+
+ # Test match with multiple agents and escaped underscores (new outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story\\_writer will follow the Product\\_manager."
+ )
+ assert result == {"Story_writer": 1, "Product_manager": 1}
+
+ # Test match with multiple agents and spaces (new outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story writer will follow the Product manager."
+ )
+ assert result == {"Story_writer": 1, "Product_manager": 1}
+
+ # Test match with multiple agents and escaped underscores and spaces mixed (new outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story writer will follow the Product\\_manager."
+ )
+ assert result == {"Story_writer": 1, "Product_manager": 1}
+
+ # Test match with multiple agents and incorrect casing (unchanged outcome)
+ result = groupchat._mentioned_agents(
+ agents=all_agents, message_content="Story Writer will follow the product\\_manager."
+ )
+ assert result == {}
+
+
if __name__ == "__main__":
# test_func_call_groupchat()
# test_broadcast()
@@ -1348,5 +1442,6 @@ if __name__ == "__main__":
# test_clear_agents_history()
# test_custom_speaker_selection_overrides_transition_graph()
# test_role_for_select_speaker_messages()
- test_select_speaker_message_and_prompt_templates()
+ # test_select_speaker_message_and_prompt_templates()
+ test_speaker_selection_agent_name_match()
# pass